Donor Intervention, Economic Growth and Poverty Reduction: the Case of Sierra Leone

DONOR INTERVENTION, ECONOMIC GROWTH AND POVERTY REDUCTION: THE CASE OF SIERRA LEONE

A thesis submitted to the University of Manchester for the Degree of Doctor of Philosophy In the Faculty of Humanities

2012

PHILIP MICHAEL KARGBO

Institute for Development Policy and Management (IDPM)

School of Environment and Development

UNIVERSITY OF MANCHESTER

Table of Contents

Table of Contents ...... 2 List of Figures ...... 5 List of Tables ...... 6 List of Appendices ...... 7 Acronyms and Abbreviations ...... 8 Abstract ...... 10 Acknowledgements ...... 14

CHAPTER 1: DONOR INTERVENTION , ECONOMIC GROWTH AND POVERTY REDUCTION : INTRODUCTION AND RESEARCH BACKGROUND ...... 15 1.1 Introduction ...... 15 1.2 Background and Motivation for the study ...... 17 1.3. Research Objectives ...... 26 1.4: Research Scope ...... 26 1.5: Thesis Chapter Outline ...... 27

CHAPTER 2: FOREIGN AID AND ECONOMIC DEVELOPMENT IN SIERRA LEONE: AN OVERVIEW ...... 29 2.1 Introduction ...... 29 2.2 Sierra Leone: Political Background ...... 30 2.3 Economic Performance in Post-independent Sierra Leone ...... 36 2.4 Foreign aid management and flows in Sierra Leone ...... 43 2.5 Foreign aid and Development Outcomes: An Overview ...... 49 2.6 Conclusion ...... 50

CHAPTER 3: THE EFFECTIVENESS OF FOREIGN AID IN DEVELOPING COUNTRIES : A SURVEY OF THE EVIDENCE ...... 53 3.1: The Literature on the Relationship between Foreign Aid and Economic Growth ...... 53 3.1.1: The Aid-Growth Relationship: What does Theory Suggest? ...... 53 3.1.2: The Empirical Evidence...... 59 3.1.2.1 Aid –Growth Relationship: The Cross-Country Empirical Evidence ...... 59 3.1.2.2 The Country Literature: Empirical Evidence ...... 66 3.2: The Empirical Evidence on the Relationship between Aid and poverty Reduction ..... 70 3.2.1: Foreign Aid and Pro-poor Growth ...... 70 3.2.2: Foreign Aid and Welfare of the Poor ...... 73 3.3: Recipient policy and political institutions and aid effectiveness ...... 76 3.4: Review of the Literature on Aid Disaggregation ...... 82 3.5: Summary of the Evidence and gaps in the Literature ...... 90

CHAPTER 4: CONCEPTUAL FRAMEWORK AND OVERALL RESEARCH STRATEGY ...... 93 4.1 Conceptual Framework ...... 93 4.1.1: Introduction ...... 93 4.1.2 An Overview of the Conceptual Framework ...... 93

4.1.3 The Research Questions and Hypotheses ...... 96 4.2. The Overall Research Strategy ...... 103

CHAPTER 5: IMPACT OF FOREIGN AID ON ECONOMIC GROWTH: EMPIRICAL ANALYSIS ...... 108 5.1 Introduction ...... 108 5.2 The Aid-Growth Relationship...... 108 5.2.1. The Empirical Model and Data ...... 108 5.2.2 Data Description and Construction ...... 112 5.2.3 The Estimation Procedure ...... 115 5.2.3.1 Unit Roots Tests ...... 121 5.2.4 The Empirical Results ...... 122 5.2.4.1 The ARDL Estimation Results ...... 122 5.2.4.2 The Johansen ML Estimation Results...... 128 5.2.4.3 Discussions and conclusion on the Impact of Foreign Aid on Economic Growth ...... 131 5.3 Further Analysis of the Impact of Aid on Economic Growth: Aid Disaggregation .... 134 5.3.1 Grants versus Loans ...... 134 5.3.2 Impact of Technical Cooperation Assistance on Economic Growth ...... 137 5.3.3 Bilateral Aid versus Multilateral Aid ...... 142 5.4 Overall Discussion and Conclusion ...... 150

CHAPTER 6: IMPACT OF FOREIGN AID ON PRO-POOR GROWTH IN SIERRA LEONE: EMPIRICAL ANALYSIS ...... 153 6.1: Methodological Framework and Data ...... 153 6.1.1: The Empirical Model and Data ...... 153 6.1.2: Data Description and Construction ...... 161 6.1.3 The Overall Estimation Procedure ...... 164 6.1.3.1: The ARDL Approach to Cointegration ...... 164 6.1.3.2: The Johansen Maximum Likelihood Approach to Cointegration ...... 166 6.1.4: Unit Root Tests ...... 166 6.2: The Aid-Pro-Poor Growth Empirical Results ...... 167 6.2.1: Impact of aid on Pro-Poor Growth: The ARDL Results ...... 168 6.2.2: Impact of Foreign Aid on Pro-Poor Growth: The Johansen Results ...... 173 6.2.3: Discussion of the Aid-Pro-Poor Growth Results ...... 175 6.3: Aid Modality and Pro-poor Growth...... 176 6.3.1: Impact of Food Aid on Pro-Poor Growth ...... 176 6.3.2: Technical assistance and pro-poor growth ...... 182 6.3.3: Impact of Grants and Loans and Pro-Poor Growth ...... 186 6.3.4: Conclusion on the aid modality results ...... 189 6.4: General Discussions and Conclusion ...... 189

CHAPTER 7: IMPACT OF FOREIGN AID ON WELFARE : AFRICA AND SIERRA LEONE ...... 193 7.0 Introduction ...... 193 7.1. Methodology ...... 193 7.1.0 Introduction...... 193 7.1.1: The Empirical Model ...... 194 7.1.2: Data Description and Sources ...... 198

7.1.3: Estimation technique ...... 201 7.1.4: Diagnostics and Inference ...... 204 7.2. The Empirical Results ...... 205 7.2.1: Impact of aid on Welfare (on HDI and IMR) ...... 205 7.2.1.1: Foreign Aid and Human development ...... 205 7.2.1.2 Aid and Infant Mortality Rate ...... 208 7.2.1.4 Discussion ...... 211 7.2.2: Aid Structure and Human Development in Sierra Leone ...... 218 7.2.2.1 Grants versus Loans ...... 219 7.2.2.2: Technical Assistance versus Non-Technical Assistance ...... 220 7.2.2.4 Discussion ...... 222 7.2.3 Foreign Aid, Politics and Human Development in Sierra Leone ...... 228 7.2.3.1 The model and Empirical Results ...... 228 7.2.3.2 Discussion ...... 235 7.3: Conclusion ...... 236

CHAPTER 8: CONCLUSION AND POLICY IMPLICATIONS ...... 238 8.1 Introduction ...... 238 8.2 Foreign Aid and Economic Growth – Summary, Conclusion and Policy Implications ...... 240 8.3 Foreign Aid and Pro-poor growth – Summary, Conclusion and Policy Implications . 243 8.4 Foreign Aid and Aggregate Welfare – Summary, Conclusion and Policy Implications ...... 246 8.5 Politics, Foreign Aid and Human Welfare- Summary, Conclusion and Policy Implications...... 249 8.6 Limitations and Directions for Future Research ...... 250 References ...... 253 Appendix ...... 269

Word Count: 86,378

List of Figures Figure 2.1: GDP per capita growth: Sierra Leone versus developmental groups…………… 37 Figure 2.2: Average GDP per capita growth rates (1970-2007) ……………………………. 38 Figure 2.3: Sierra Leone: Sectoral Contribution to GDP……………………………………. 40 Figure 2.4: Trends in human development index (HDI)…………………………………….. 41 Figure 2.5: Trends in infant mortality rate (IMR)…………………………………………… 42 Figure 2.6: Average foreign aid disbursed (1970-2007)…………………………………….. 45 Figure 2.7: Sierra Leone: Average disaggregated aid flows (1970-2007)…………………… 46 Figure 2.8: Comparative Analysis of foreign aid disbursed to Africa, Sub-Saharan Africa and Sierra Leone from 1970-2007 ………………………………………………………….. 48 Figure 3.1: Solow Model with Foreign Aid………………………………………………….. 57 Figure 4.1: A Conceptual Framework showing the Schematic relation of Donor Intervention, Economic Growth and Poverty Reduction …………………………………… 94 Figure 4.2: The Overall Research Strategy…………………………………………………. 106 Figure 5.1: CUSUM and CUSUMSQ Plots for the base model……………………………. 125-126 Figure 6.1: CUSUM and CUSUMSQ Plots for the Base Model on Aid-Pro-Poor Growth .... 169-170

List of Tables

Table 2.1: Simple correlations between foreign aid and economic growth for Sierra Leone 49 (1970-2007)…………………………………………………………………… Table 2.2: Simple correlations between foreign aid and Human Welfare for Sierra Leone 50 (1980-2007)…………………………………………………………………… Table 5.1: ADF Unit Roots test results - including constant but without trend: 122 1970-2007………………………………………………………… Table 5.2: Long-run Impact of Foreign Aid on Economic Growth: The ARDL Results……... 123 Table 5.2.1: Short-run Impact of Foreign Aid on Economic Growth: The ARDL Results... 124 Table 5.3: Long-run Impact of Foreign Aid on Economic Growth: The Johansen Results… 129 Table 5.4: Long-run Impact of Grants on Economic Growth: The ARDL Results………… 136 Table 5.5: Long-run Impact of Technical Cooperation Assistance on Economic Growth: The 139 ARDL Results……………………………………………………………………… Table 5.6: Long-run Impact of Technical Cooperation Assistance on Economic Growth: 141 The Johansen Cointegration Results…………………………………………….. Table 5.7: Long-run Impact of Bilateral and Multilateral Aid on Economic Growth: The 145 ARDL Results……………………………………………………………………. Table 5.8: Long-run Impact of Bilateral and Multilateral Aid on Economic Growth: The 147 Johansen Cointegration Results………………………………………………… Table 6.1: ADF Unit Roots test for the Pro-poor Growth Model Variables – including 167 constant but without trend: 1970-2007…………………………………………… Table 6.2: ARDL Long run Estimates for Impact of Aid on Pro-poor Growth ……………. 168 Table 6.2.1: ARDL Error Correction representation (Short-run Estimates) of the impact of 171 aid on Pro-poor Growth………………………………………………………. Table 6.3: Long run Estimates for Impact of Aid on Pro-poor Growth using Johansen 174 Approach……………………………………………………………………….. Table 6.4: Long run Estimates for Impact of Food Aid on Pro-poor Growth using ARDL 180 Approach…………………………………………………………………………. Table 6.5: Long run Estimates for Impact of Technical Assistance on Pro-poor Growth using 184 ARDL Approach…………………………………………………………………. Table 6.6: Long run Estimates for Impact of Grants and loans on Pro-poor Growth using 187 ARDL Approach…………………………………………………………………. Table 7.1: Foreign Aid and Human Development…………………………………………… 206 Table 7.2: Foreign Aid and Infant Mortality Rate…………………………………………… 209 Table 7.3: Structure of Aid and Human Development……………………………………… 219 Table 7.4: Politics and aid effectiveness ……………………………………………………. 231

List of Appendices Appendix 1.1: History of Foreign Aid………………………………………………………. 269 Appendix 1.2: Historical Human Development Ranking for Sierra Leone (1990-2007)…… 269 Appendix 2.1: Post-Colonial Regimes in Sierra Leone…………………………………….. 270 Appendix 2.2: Per Capita GDP Growth: Sierra Leone versus Underdeveloped Countries…………. 271 Appendix 2.3: Human Development Index: Sierra Leone versus Africa…………………………… 271 Appendix 2.4: Infants Mortality Rate: Sierra Leone versus Africa………………………………… 271 Appendix 2.5: Sierra Leone: Sectoral Contribution to GDP (1970-2007) (in %)…………………… 271 Appendix 4.1: Appendix 4.1: List of Fieldwork Interviews………………………………… 272-73 Appendix 4.2: Summary of the Overall Research Design…………………………………… 275-77 Appendix 5.1: Principal Component Analysis for the generation of the Policy Index………………………………………………………………………… 278 Appendix 5.2A: The Impact of Aid on Economic Growth: Summary statistics………… ….. 278 Appendix 5.2B : Plots of the log level variables as used in the model……………………….. 278-79 Appendix 5.3: Long-run Impact of aid on economic growth using the ARDL Approach– Robustness Specifications ……………………………………………………. 280 Appendix 5.4: Short-run Impact of Aid on Economic Growth using the ARDL Approach - Robustness Check Specifications …………………………………………... 280 Appendix 5.5: Long-run Impact of Foreign Aid on Economic Growth using Johansen Approach— Robustness Specifications………………………………………. 281 Appendix 5.6: ADF Unit Roots Test for Aid Disaggregates with Intercept but without trend: 1970-2007……………………………………………………………………... 281 Appendix 5.7: Short-run Impact of aid Structure on economic growth –Grant versus Loan using ARDL Approach……………………………………………………….. 282 Appendix 5.8: Short-run Impact of Technical Assistance on economic growth using ARDL Approach - Robustness Check……………………………………………….. 282 Appendix 5.9: Short-run Impact of Foreign Aid Structure on Economic Growth - Bilateral aid versus Multilateral Aid using ARDL Approach………………………….. 283 Appendix 5.7.1: CUSUM and CUSUMSQ Plots for Model on Impact of Grants and Loans on Economic Growth…………………………………………………………. 284 Appendix 5.8.1: CUSUM and CUSUMSQ Plots for Model on Impact of Technical Assistance on Economic Growth……………………………………………. 285 Appendix 5.9.1: CUSUM and CUSUMSQ Plots for Model on Impact of Bilateral and Multilateral Assistance on Economic Growth ……………………………… 286 Appendix 6.1: Principal Component Analysis for the generation of the Trade Policy Index ………………………………………………………………………. 286 Appendix 6.2: Error Correction representation (Short-run Estimates) of the impact of Food aid on Pro-poor Growth Using ARDL Approach…………………………… 287 Appendix 6.3: Error Correction representation (Short-run Estimates) of the impact of Technical Cooperation Assistance on Pro-poor Growth Using ARDL Approach… 287 Appendix 6.4: Error Correction representation (Short-run Estimates) of the impact of Grants on Pro-poor Growth Using ARDL Approach…………………………………. 287 Appendix 6.2.1: CUSUM and CUSUMSQ Plots for Model on Food Aid and Pro-poor Growth…………………………………………………………………………. 288 Appendix 6.3.1: CUSUM and CUSUMSQ Plots for Model on Technical Assistance and Pro-Poor Growth………………………………………………………………. 289 Appendix 6.4.1: CUSUM and CUSUMSQ Plots for Model on Grants versus Loans on pro- poor growth …………………………………………………………………… 290 Appendix 7.1: Construction of the pro-poor expenditure (PPE) index………………………. 290-91 Appendix 7.2: AID and Pro-poor Expenditure……………………………………………….. 291 Appendix 7.3: Foreign Aid and Human Development……………………………………….. 292 Appendix 7.4: Foreign Aid and Infant Mortality Rate……………………………………….. 293 Appendix 7.5: Structure of Aid and Human Development…………………………………… 294 Appendix 7.6: Politics and aid effectiveness (SSA Sample)…………………………………. 295

Acronyms and Abbreviations

ADF – African Development Fund ADF – Augmented Dickey Fuller AfDB – African Development Bank AFRC – Armed Forces Revolutionary Council AIC – Akaike Information Criterion APC – All Peoples Congress AR – Arellano - Bond ARDL – Autoregressive Distributed Lag BA – Bilateral Aid BBC – British Broadcasting Corporation CUSUM - cumulative sum of recursive residuals CUSUMSQ - cumulative sum of squares of the recursive residuals CPI – Consumer Price Index DACO – Development Assistance Coordinating Office DFID – British Department for International Development DCPS- Domestic credit to the private sector DSL – Dummy for Sierra Leone EU – European Union EDA – Effective Development Assistance ECM – Error Correction Mechanism ECOMOG - Economic Community of West African States Monitoring Group FA – Food Aid FDI – Foreign Direct Investment GDP – Gross Domestic Product GTZ – Deutsche Gesellschaft fur Internationale Zusammenarbeit GEX – Government Expenditure as share of GDP GMM – Generalised Methods of Moments HIPC – Highly Indebted Poor Countries HDI – Human Development Index HQC - Hannan-Quinn Criterion ICOR – Incremental Capital-Output Ratio ICRG – International Crisis and Risk Guide IDA – International Development Assistance IFIs – International Financial Institutions ILO – International Labour Organisation IMF – International Monetary Fund IMR – Infant Mortality Rate INF – Inflation Rate IQ1 – Property Rights IQU – Quality of Governance MDGs – Millennium Development Goals MA – Multilateral Aid MDBS – Multi-Donor Budget Support ML – Maximum Likelihood MODEP – Ministry of Development and Economic Planning MOFED – Ministry of Finance and Economic Development MNEs – Multi-National Enterprises NaCSA – National Commission for Social Action

NFA – Non-Food Aid NGO – Non-Governmental Organisations NTCA – Non-Technical Assistance grants ODA – Official Development Assistance OECD – Organisation for Economic Cooperation and Development OLS – Ordinary Least Squares PCA – Principal Components Analysis PETS – Public Expenditure Tracking Survey PFM – Public Financial Management PRSP – Poverty Reduction Strategy Paper PPE – Pro-Poor Expenditure RAGDP – Real Agricultural Gross Domestic Product RGDP – Real Gross Domestic Product RUF – Revolutionary United Front SAPs – Structural Adjustment Programmes SBC - Schwarz Bayesian Criterion SIC – Schwartz Information Criterion SIPRI – Stockholm International Peace Research Institute SLPP – Sierra Leone Peoples Party 2SLS – Two Stage Least Squares SNA – System of National Accounting SPSS – Statistical Package for Social Scientists SSA – Sub-Saharan Africa TCA – Technical Assistance Grants TOPEN – Trade Openness UK – United Kingdom UN – United Nations UNCTAD – United National Conference on Trade and Development UNDP – United Nations Development Programme US – United States USAID – United States Agency for International Development VAR – Vector Autoregression VECM – Vector Error Correction Model WDI – World Development Indicators WFP – World Food Programme

Abstract

In capital-scarce low income economies, the lack of attractiveness to private foreign investment implies that the only readily available source of external financing for economic development has to come from foreign aid which normally comes with an altruistic motive. However, despite long history of aid-giving to low income countries and especially Sub- Saharan Africa, evidence of effectiveness of such assistance has remained debatable, particularly with the dominance of cross-country studies in such enquiry. With yet no existing country study for Sierra Leone, a typical aid dependent country, this research investigates the relationship between donor intervention (in their aid disbursement) and the development outcomes of economic growth and poverty reduction in the country. In conducting such an enquiry, the study proposed three objectives. The first examines the relationship between aid and economic growth. The second objective investigates the relationship between aid and poverty reduction considering two variants of poverty reduction: improvement of pro-poor growth and aggregate human welfare. The final objective assesses the effect of domestic politics on aid’s effectiveness in improving human welfare.

Arising from a pluralistic analytical framework involving a triangulation of econometric estimation approaches complemented with qualitative enquiry, the study finds that aid to Sierra Leone is significant in promoting economic growth in the country. In terms of the impact on poverty, the results show that foreign aid to Sierra Leone has significantly improved long-run pro-poor growth in the country, but this impact could not be confirmed in the short-run. With respect to the other strand of poverty, the study finds that though aid may have not improved human well-being in Africa, it is found to significantly improve human development in Sierra Leone, though the evidence could not support its reduction of infant mortality rate as a second indicator of human well-being. Finally, for the investigation of the link between aid, politics and human development in Sierra Leone, the study finds that though aid is significant in directly improving human development in the country, yet pro- democratic politics (as against autocratic regimes) can also be good a policy option for aid‘s impact on human development in the country. Accounting for disaggregation bias of foreign aid, the study finds that whilst grants seem to consistently improve economic growth, pro- poor growth and human welfare, the study could not find strong evidence to suggest that technical assistance and loans likewise improve economic development the country. The impact of food aid on pro-poor growth is found to be moderate in conformity with the study’s hypothesis.

Concluding from the analysis, it is evident in the case of Sierra Leone that the supplemental theories largely hold that foreign aid is vital in the promotion of a country’s economic development. Hence, the intervention of donors in the economy of Sierra Leone has not seemed to be in vain, but has rather proved to be largely useful. It implies that Sierra Leone’s persistent poverty characterisation amidst notable donor presence and participation in the country’s economy has little to do with the fact that foreign aid has not been effective in promoting the country’s economic development, but it may however be that the magnitude of the effect may not have been that high to completely eradicate poverty. The study’s identification of the most effective types of aid as well the realisation of political stability and democracy for enhanced effectiveness of aid in the country could be crucial if the economic significance of foreign aid is to be improved in Sierra Leone.

Declaration

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this thesis or any other university or other institute of learning .

Philip Michael Kargbo, April, 2012

I. Copyrights of this thesis rests with the author (including any appendices and/or schedules to this thesis) who owns certain copyright or related rights in it (the “Copyright”) and he has given the University of Manchester certain rights to use such Copyright, including for administrative purposes.

II. Copies of this thesis, either in full or in extracts, and whether in hard copy or electronic copy, may be made only in accordance with the Copyright, Design and Patent Act 1998 (as amended) and regulations issued under it, or where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any copies made.

III. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “intellectual property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in the thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions.

IV. Further information on the conditions under which disclosure publication, and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy which can be accessed on http://www.campus.manchester.ac.uk/medialibraries/policies/intellectualproperty.pdf , in any relevant Thesis restriction declarations deposited in the University Library, the University Library’s regulations available on its website which can be accessed on http://www.manchester.ac.uk/library/aboutus/regulations and in the University’s policy on the presentation of Theses.

V. Further information on the conditions under which disclosures and exploitation may take place is available from the Head of School of Environment and Development.

Dedication

This thesis is dedicated to my late wife Sylvia Kargbo, who inspired me for further studies; and to my lovely kids Philippa and Roland.

Acknowledgements

It is important to recognise that this piece of work has not been done entirely without the direct and indirect support of some individuals; and hence the need to so acknowledge them in the document.

First and foremost, I’m most grateful to the Almighty God whose guidance has seen me through these difficult years of undertaking this project. Thank you Lord.

To my Supervisors, Professor Kunal Sen and Dr. Admos Chimhowu, your constructive criticisms, comments and support have all but made the completion of this study a possibility. Your research expertise has not only enhanced my ability to complete this piece of work, but the skills I have gained from that would be certainly useful in my research career. I am most grateful.

Thanks to the Commonwealth Scholarship Commission for granting me a scholarship to undertake this research; and to the British Council for their excellent management of the award in addition to their moral support during perilous times.

My deepest appreciation goes to my family who had provided the necessary support during this period of my research. To my late wife, Sylvia, even after you longed for this study, you would not be there to provide the needed company as you had to respond the call of eternity. You’ll always be part of me. To my son, Roland, who gave me the needed company of a father and son during the final year of my study, I love you boy. To my daughter, Philippa, we have been distant apart, yet your timely calls were no less than anything I needed.

To my parents, Mrs. Adama Koroma and Dr. Roland Kargbo, I am most thankful. Especially to my mum, who opted to care for my son whilst I undertook my study and who has ever been there to pray for my success, I am more than grateful. To my extended family that had been very supportive of me, I am thankful.

Lastly, I express my gratitude to my numerous academic and close friends, whose names I cannot all mention here, but who will always remain in my memory. To Dr. BIB Kargbo and Santigie Kargbo, your academic companionship even when you have been miles away has been appreciated. To Sahr Davowah and Gbassay, your friendship has been unimaginable. To colleagues at IDPM and Department of Economics, I thank you all.

CHAPTER 1: DONOR INTERVENTION , ECONOMIC GROWTH AND POVERTY REDUCTION : INTRODUCTION AND RESEARCH BACKGROUND 1.1 Introduction

The urge to aid the development of low income countries has only continued to gain momentum. Several reasons may be associated with this, but crucial amongst them is the need to uplift the living standards of its people who have remained to endure a continued impoverishment. This uncharacteristic status of well-being in low-income countries largely bears its origin from slow economic growth and weak development outcomes in such economies, which means any thoughts of low-income countries catching up with developed countries can only be a slim possibility in the distant future. The cause of this slow economic progress in developing countries, and in particular Africa, has two schools of thought which tend to propose opposing views: one suggesting the cause emanates from external forces (Rodney, 1972; Chabal, 1994; Harvey, 2003), and the other arguing it is largely internal (Ayittey, 2004). Irrespective of which argument (from these schools of thought) may be more convincing, what remains obvious is that the persistence of poverty, the apparent consequences of global inequality together with the increasing urge for globalisation, implies the slow and weak development trend in developing countries becomes a concern to not only the developing country authorities, but perhaps equally so to the international community. Hence, the promotion of economic development to low-income countries becomes a collective issue, for which the proposition and commitment to global development targets such as the millennium development goals is evidence.

Whilst there may be several factors that contribute to economic development, a crucial one particularly applicable to the economic development of developing countries is capital availability. Key theorists of economic development recognise the importance of capital for economic growth (Harrod, 1939; Domar, 1947; Solow, 1956; Arrow, 1962; Romer, 1986; Lucas, 1988). Since developing countries lack capital needed to sufficiently stimulate rapid economic growth, several pathways have been followed to raise capital levels, stimulate investment and foster economic growth in developing countries. Whilst other forms of external capital (such as foreign direct investment) are vital for economic development, that which low-income countries can readily access is foreign aid.

Foreign aid is a form of foreign capital disbursed in the form of grants or concessionary loans. It can be from official bilateral or multilateral sources. The main motive is usually to foster economic development in such aid recipient countries (Thorbecke, 2000; Hjertholm and White, 2000; Sachs, 2005; Radelet, 2006; Tarp, 2006), although others like Alesina and Dollar (2000), Anderson et al (2005), Radelet (2006) and Headey (2007) suggest that there may be other non-charitable reasons. The OECD 1 defines foreign aid as grants and loans to developing countries and territories, which are undertaken by the official sector of the donor country, with the promotion of economic development and welfare in the recipient country as a core objective and at concessional financial terms (Temple, 2010: 4425). Whilst grants do not have a repayment obligation from the recipient country, loans do have to be repaid, but with interest rate below the market rate and with grant element of at least 25% to qualify concessional loans as aid (ibid). Aid may also be given to recipient countries in the forms of technical assistance, food aid, programme aid and project aid which by their nomenclature tells the direct purpose of the assistance.

Though other motives of aid may exist, the interest usually from aid donors is an altruistic one, where the objective of granting aid is associated with the promotion of economic development and improvement of human welfare in the recipient country (see next section for an elaboration). In some situations, donors may wish to ascertain if aid allocations have met the intended purpose while recipient country citizens may also want to know if aid disbursed to their country has been well used for its intended purpose. The need for accountability becomes obvious. One way to show accountability of aid is through assessing how effective it has been in achieving its intended purpose. Since the ultimate target of foreign aid is developmental, the need to assess the effectiveness of aid in terms of its impact on economic growth and poverty reduction is at the core of contemporary poverty discourse. This is what this present study is about.

The study assesses the effectiveness of foreign aid in Sierra Leone, a typical aid dependent country where not much work exploring those links exists. While much work has been done on the impact of aid on economic growth, there have been increasing calls to look at effectiveness beyond the economic growth criterion (Feeny, 2005; Bigsten et al., 2006;

1 See direct OECD definition at www.oecd.org 16

Temple, 2010). This study looks at not only the impact of aid on economic growth, but also on the impact on poverty in the recipient country. We also assess aid effectiveness beyond the use of a single technique of estimation as has been the case in much of the country studies thus far and rather employ a combination of techniques of analysis to establish the effectiveness of aid in the country of study.

This chapter (which is divided into five sections) sets the stage on the theme of foreign aid effectiveness by next providing a background on the origin of foreign aid and the debate on its effectiveness. This background is extended to accommodate the motivation for undertaking this study, involving a presentation of the importance of investigating the relationship between foreign aid, economic growth and poverty reduction in Sierra Leone and highlighting the significance and originality of this study. The chapter next details the objectives of the study and highlights the research scope. Lastly, it outlines the structure of the thesis.

1.2 Background and Motivation for the study

Contemporary foreign aid experiences date back to the late 1940s with aid support for the reconstruction of war-torn Europe. It eventually spread to low income countries as part of strategies to aid their development (Riddell, 2007). Though aid had actually started prior to this period (e.g. US overseas aid – 1812 Act for the relief of citizens of Venezuela and the commencement of US food aid in 1896; aid to British colonies - 1929 Colonial Act), it was more popular in the 1940s when it was institutionalised within international organisations and international forums as development aid. The International Labour organisation (ILO) for instance, advocated for aid funds to be provided in order to raise the living standards of the poor countries (Riddell 2007). Some like Hjertholm and White (2000), Easterly (2009) and Temple (2010) provide a detailed review of the history of foreign aid to recipient countries with regards the context of the assistance, development orthodoxy and the changing focus of donor assistance.

Evident in the history of foreign aid are a couple of observations. Firstly, though there has been some political motive for aid, the development motive seemed to have dominated. Indeed the political motive of aid is evident in the nature of the US Marshal Plan and as well

17 evident in the nature of aid-giving in the aftermath of the cold war. In as much as the Marshall Plan had its development objective of economic reconstruction, the establishment of a strong alliance by the US with (Western) Europe largely as an attempt of consolidating its superiority over the Soviet Union as world super power (Kunz, 1997; Hjertholm and White, 2000; Cox and Kennedy-Pipe, 2000) shows some evidence of political motive in aid-giving. In fact, Hjertholm and White (2000) argue that the political and economic interest of donors have contributed to the changing focus of foreign assistance over the years of foreign aid and has consequently distorted the development objective of aid. Heady (2007) showed that the developmental impact of bilateral aid could only be evident in the post-cold war era suggesting that aid from bilateral donors has been largely political during the era of the cold war.

However, whilst political motives of aid do exist, the development objective of aid involving economic growth and poverty reduction clearly dominates. In Appendix 1.1, we present Temple’s (2010) schematic representation of the history foreign aid, and in this table, the dominance of the development objective of aid is clearly evident. Though the development orthodoxy of donor that dictate aid disbursement may have varied in specificity (central planning, Washington Concensus, promotion of market friendly institutions and investment), yet it all centred on the economic development of the recipient county.

The second observation deduced in this history of foreign aid is that though the purpose is largely developmental, yet the developmental focus of aid does vary over the years. Easterly (2009) describes this changing focus of foreign aid as the recycling of ideas on aid policies and focus, returning to ideas that have been suggested before. In the early period of aid- giving, the major focus of foreign assistance was community development. In the 1960s, bilateral institutions took prominence as major aid donors, with aid focus being on financing the productive sectors (for instance the green revolution) and infrastructures of the recipient countries (Hjertholm and White, 2000; Temple 2010). In the 1970s, the focus of aid was on poverty reduction with particular emphasis on agriculture and basic needs and in low income countries (Hjertholm and White, 2000; Temple, 2010). During this period, whilst multilateral donors funded projects, bilateral donors rather disbursed technical assistance and budget support to recipient economies. The 1980s witnessed the period of global debt crisis, fading of central planning and the rise of NGOs. The effect was the dominance of the Washington Consensus as development orthodoxy. Recipient countries were hence subjected to market- 18 based adjustments, rolling back the state in some elements of economic development. In the1990s, aid focus again turned to poverty reduction (Hulme, 2009; Temple, 2010), but this time alongside the promotion of good governance and the investment climate as crucial ingredients for economic development and the effectiveness of aid. Aid moved towards sector support and the financing of the HIPC initiative. In the 2000s, recognising the economic hardship in low income countries, anti-globalisation campaigns gained momentum alongside high media attention of world poverty (Easterly, 2009; Temple, 2010). This led to increased donor focus on poverty reduction and especially the enhancement of human welfare in the new millennium. International development targets such as the Millennium Development Goals, though set in the mid-1990s, yet were commissioned in the millennium with the aim of significantly reducing income poverty and improving human welfare by 2015 (Hulme, 2009).

This contemporary poverty orientation of foreign aid has been well seen in the case of Africa for which there have been increasing calls for more aid effort so as to achieve the targets of the MDGs by 2015. With much of sub-Saharan Africa lagging behind other developing regions in progress towards meeting the MDGs targets by 2015, the focus of aid has been on promoting economic growth and reducing poverty in this region. Easterly (2009) described the increasing aid effort to Africa as an escalation of the ‘big push’ approach by the West to ‘save’ Africa, which has existed since the early days of aid-giving. Highlighting this recent escalation of international aid to Africa, Easterly (2009) noted five of such cases. First was the G8 summit in Gleneagles 2005, which called for the doubling of Africa’s aid from US$25billion to US$50billion a year. The second was the G8 summit in Germany 2007, which further emphasised the need for more aid to the region. The third was the G8 summit in Japan 2008, where the rich countries promised to donate additional US$25billion a year unto 2010 for Africa especially. The fourth was the commitment by Japan in 2008 to double its aid to Africa. Finally, evidence of notable celebrity involvement in advocating for increased aid to Africa (independently involving Geldof, Bono, Angelina Jolie, and Bill Gates) has yet been another proof of escalating aid effort of the West to save Africa.

In spite of donor effort in providing aid to developing countries including Sub-Saharan African countries, the evidence of its effectiveness has remained in debate. Its impact has been in dispute, with some concluding that aid works while others suggest it does not. Those who argue that aid does not work, base their argument on the evidence of prevalent poverty 19 and slow economic growth in aid recipient countries in spite of increased aid inflows. Further, they argue that the self-interest motive of aid users, especially the recipient authorities is the key factor towards the failure of aid. Chapter 3 of the thesis provides a review of this empirical literature and highlights some of these arguments. However, what appears clear is that aid to developing countries, and indeed its escalation to Africa since the 1970s, has not quite established its development impact. Development performance in Africa remains dismal amidst prolonged and increased donor involvement and support in the continent.

In Sierra Leone, the focus of this study, poor economic performance has been characteristic of the country since the 1980s. Donors have intervened since independence period of 1961 in large numbers and with varied and modified policies and increased levels of aid support in an attempt to promote economic stability, growth and poverty reduction; yet poverty statistics reveal that poverty is still eminent in the country. For instance, the UNDP’s Human Development Reports since the 1990s have consistently placed Sierra Leone at the bottom end of the human development index ranking (Appendix 1.2). Chapter 2 of this thesis further provides some detailed analysis describing the trends in foreign aid and the unimpressive economic performance of the country. It is therefore expected that a research that probes into the role of donor intervention on the country’s poverty situation will help provide information as to whether aid has been effective in supporting economic development in Sierra Leone.

The need to conduct this study is particularly driven by the fact that Sierra Leone remains a country where empirical studies on development policy have been quite limited. To date, no comprehensive study has been conducted on the impact of foreign aid in Sierra Leone. Aid effectiveness findings beyond donor self-evaluation reports have virtually been absent. By conducting this research, we have generated a basis for further empirical studies. In addition to its usefulness as a baseline for further studies on aid effectiveness in the country, the need for this study is also inspired by the fact that whilst Sierra Leone remains one of the poorest countries in the world despite long history of partnership and notable donor participation in the country’s economy (shown in chapter 2), no known country-wide study on the impact of such support has been conducted for the country. It is hoped that the conduct of this study will inform policy on the effectiveness of aid resources in the country as well as establishing ways in which aid could be made more effective to minimise poverty and stimulate economic performance. 20

This study contributes to the aid effectiveness research in the academic literature in two ways: Firstly, it contributes by providing further evidence for an interesting country case study characterised by a mixture of aid dependency, slow economic growth, abject poverty and prolonged civil conflict. It is evident following the review of the aid effectiveness literature that there exist mixed/inconclusive findings on the effectiveness of foreign aid in both the cross-country and country literature in terms of its impact on economic growth and poverty reduction. Some researchers have shown that donor aid has had a positive contribution to economic development, yet others have not. Hence, by providing further evidence on a unique country case study to a literature that remains inconclusive on its findings, will throw more insights on the impact of foreign aid. The fact that no empirical study exists on the impact of aid to Sierra Leone implies evidence gathered from this study itself will contribute to that already existent for other countries.

Secondly, this study will generate findings to fill the knowledge gaps in aid effectiveness research. This study’s investigation of the importance of domestic politics on aid effectiveness at the country level of analysis is an important area of contribution. Previous research on domestic politics for instance (e.g. Oloka-Onyango and Barya, 1997; Commack, 2007; Brock et al., 2002; Hickey, 2003) have shown the importance of domestic politics in the development process, but have not investigated the importance of such politics on aid effectiveness. The few that attempt to investigate the link between politics and aid effectiveness (Boone, 1996; Arvin and Barillas, 2002; Kosac, 2003; and McGillivray and Noorbakhsh, 2007) have only done so at the cross country level of analysis, and not specifically exploring the relationship at the country level. Thus, by providing some investigation in this area for Sierra Leone, this study will contribute to the literature and scenarios to policy makers on the role of politics in aid and development policy formulation.

In Sierra Leone particularly, an investigation of the role of politics in the aid business deserves some attention. Firstly, there has been an increasing drive towards democratic institutions or democratisation in the country, particularly since the end of the country’s civil war for which the then politically less competitive one party rulership had been identified as one of the reasons for the civil conflict in the country (Zack-Williams, 1999). This has been further reinforced by the increasing support by donors for democratic reforms and institutions following the end of the country’s civil conflict. Fieldwork interviews with some the 21 country’s leading donors (for instance DFID) emphasised the promotion of democracy as one of its reasons for granting aid to Sierra Leone. This is also in tune with the USAID’s Millennium Challenge Account (for which Sierra Leone is a beneficiary), which also emphasises the importance of democratic governance as crucial to granting aid. Despite the emphasis of democracy as an important condition for aid allocation and development in general, the review of the literature on political institutions and economic growth by Przeworski and Limongi (1993) conclude that in as much as politics is importance for the process of development, there is no clear-cut conclusion on whether it is democracy or autocracy that contributes more to economic development. Yet still, Kosac (2003) argues that the effect of aid on human development is only evident under good democracies. The study by McGillivray and Noorbakhsh (2007) on the other hand could not confirm the finding of Kosac (2003). This study therefore contributes to the literature and attempts to enter the argument on aid, politics and human development by extending the analysis for Sierra Leone, a country that has experienced notable episodes of both democratic and autocratic regimes.

Further, though using a cross-country dataset, this study examines the impact of aid on welfare at the country level of analysis. This is an original approach as most studies that examine this relationship have only limited their analysis at the cross-country level.

Thirdly, this research contributes to the aid effectiveness literature by disaggregating aid into its types and sources for all the development outcomes of economic growth, pro-poor growth and human development for which the impact of aid is examined. Mavrotas (2007) emphasised that a major short-coming of the aid effectiveness literature is its neglect of the heterogonous character in which aid is disbursed to recipient economies. This has often been referred to as the ‘aggregation bias’ in aid effectiveness. According to this argument, the use of a single figure for aid as commonly used in aid effectiveness studies, cannot capture this aid heterogeneity, hence leading to aggregation bias in the empirical results often reported (Cassen, 1994; White, 1998; Mavrotas, 2002, 2005, 2007; Mavrotas and Ouattara, 2006a, 2006b, 2007; Ouattara and Strobl, 2008). These proponents of aid disaggregation often present two main reasons to justify their argument. Firstly, government may attach different utility to each category of aid (particularly within an endogenous fiscal response framework) and hence aggregation bias may result from the use of a single figure for aid. Secondly, there is the likelihood of a changing composition of aid received over time, and hence ignoring this possibility is likely to affect the econometric results. In addition to the general aggregation 22 bias argument, the need to disaggregate aid has become even more important in the context of the increasing call for the scaling up of aid to developing countries, especially in Africa. This has also been echoed by Maxwell (2002) and Mavrotas (2007) who suggest that increasing aid without considering the modality in which it is delivered will be a missed opportunity. In fact, Mavrotas and Nunnenkamp (2007) argue that within the context of doubling aid efforts, it would be insufficient to double aid without consideration of the quality of aid in the context of making significant effort in exploring the various routes and mechanisms though which the various types of aid operate. Observed within the limited aid heterogeneity literature is that much of recent research on aid disaggregation increasingly focus on the fiscal response effect (e.g. Cassimon and Campenhout, 2007; Mavrotas, 2005; Mavrotas and Ouattara, 2006a; Mavrotas and Ouattara, 2006b; Mavrotas and Ouattara, 2007), which though important, yet the need for further examination of aid disaggregation on ultimate aid targets deserves attention. This is particularly so because aid, whether in the form of grants, loans, or technical assistance has its ultimate objective of economic development; and hence ignoring the development objectives of aid in the evaluation of aid modalities is insufficient. Even further observed in this literature is that whilst disaggregation is reported to be increasingly neglected in the aid literature, what is even more ignored is disaggregation impact on aid objectives other than economic growth and recipient budget (e.g. on pro-poor growth and human welfare which are both correlated with poverty reduction). This is important because as is previously stated, the reduction of poverty is seen as the main development target of recent aid efforts. This study hence expands its analysis of aid disaggregation to capture the impact on the development objective of not only economic growth, but also pro-poor growth and human welfare as correlates of poverty reduction.

In Sierra Leone particularly, the assessment of the various aid delivery modes and sources deserves special attention. Firstly, there has been growing argument/obsession against technical assistance in developing countries generally, and this has been evident in daily development policy discourse in Sierra Leone particularly. Fieldwork interviews with civil society groups, local media elites, public audit institutions including parliament, and as well as public sector officials clearly highlighted the growing public obsession with too much aid emphasis on technical assistance and for which much of this is devoted to the procurement of foreign expatriate largely from the donor country itself. It is therefore relevant to provide an empirical study that proves the effectiveness (or ineffectiveness) of technical assistance on the outcomes of development in the country. Secondly, with Sierra Leone being a typically 23 poor country for which debt burden cannot be a distant possibility, it is important to analyse whether aid loans in its concessional form (but which may as well add to the country’s debt burden) have had a significantly positive impact on the country’s development effort. Finally, bilateral aid has been largely argued to have geopolitical and strategic interest motive and therefore ineffective on development targets of the recipient country (Ram, 2003; Headey, 2007; Javid and Qayyum, 2011). With Sierra Leone being a largely poor and small country with perhaps relatively less visible outright political attributes that should interest bilateral donors, it would be interesting to assess whether it can as well be the case that bilateral aid can have a development motive as opposed to its commonly argued geopolitical motive. Hence, the need to recognise the importance of aid disaggregation generally and particularly in the case of Sierra Leone as well as assessing its impact is worth it.

Finally, the study provides evidence base using the case of Sierra Leone that improves reliability on existing individual country findings on the impact of aid on economic growth and poverty reduction through a variety of ways: First, by providing more robust results at the country level through the use of the method of triangulation. This generally involves the use of more than one set of insight into an investigation (Downward and Mearman, 2007). The aim of triangulation is to overcome shortfalls in any one method of analysis, and in effect arriving at more valid conclusions on the enquiry. Whilst robust results through the use of more than one estimation method may have been evident in some of the cross-country aid literature, it is uncommon in the country studies where this research is positioned. The research carried out both within-method triangulation (where we used two quantitative methods of estimation) and between-methods triangulation (where we added some qualitative analyses to the quantitative methods) in the study of the relationship between donor intervention, economic growth and poverty reduction in the case of Sierra Leone.

Triangulation (through the addition of qualitative methods) has been especially useful in this study by helping to provide some explanation for the findings of the quantitative analysis. Hence, the study uses the quantitative analysis to capture the impact of aid, and further draws on qualitative information to offer explanations to findings obtained from the quantitative techniques. According to Bigsten et al. (2006: 3), ‘a main reason for doing evaluations is to conclude by proposing improvements to the aid process, so aid becomes more effective in the future’. Thus, researching into donors’, recipient authorities’ and other stakeholders’ perception of aid effectiveness in addition to establishing the impact on economic growth and 24 poverty reduction will in some way contribute to addressing the challenge of opening up what is often referred to as the ‘black box’ in aid evaluation of its ultimate development impact.

Secondly, reliability of aid effectiveness findings is also enhanced by this study through the use of more standard techniques of data analysis. For instance, the technique of autoregressive distributed lag approach to estimating time-series data as mainly employed in this research, which though has the superior advantages over other time series techniques (as described in the methodology chapter) has only been more recently used in the country aid effectiveness analysis (i.e. only by Lloyd et al., 2001; Gounder, 2001; and Feeny, 2005; Javid and Qayum, 2011). Our final grounds through which we find our research to improve on reliability of aid effectiveness results is the fact that we are employing a country study as opposed to a cross-country analysis. The vast majority of studies on the impact of aid and the conditioning factors to its effectiveness have mostly been cross-country. Country studies are seen as having the merit of portraying the true picture of the relationship between aid and growth or poverty as opposed to the masking effect of cross-country studies (Murthey et al., 1994; Solow, 2001; Kenny and Williams, 2001; Mavrotas, 2002; Feeny, 2005; Bigsten et al., 2006; and Temple, 2010). These critics generally present two failings in cross-country studies. Firstly, cross-country studies assume that the process and components of the economy is the same across countries, an assumption which is unrealistic. Secondly, cross- country studies fail to recognise the complex causal nature of the social world. In the context of these failings, this would imply that findings on the growth performance and its causal factors at the cross-country level may be largely inadequate to rely upon. Thus, backed with the fact that country studies are generally far less researched in the aid effectiveness literature as compared to cross-country studies, our study of the impact of donor interventions on economic growth and poverty reduction in Sierra Leone as a country study is worth undertaking.

Hence, it is expected that the research findings following our investigation will contribute to the academic literature on aid effectiveness as well as provide some policy considerations that will guide donor interventions as well as government behaviour and management of donor supported development policies and aid flows.

1.3. Research Objectives

The aim of this study is to assess the relationship and impact of donor aid on economic growth and poverty reduction in Sierra Leone.

The specific objectives of this research therefore are:

1. To examine the impact of aid on economic growth 2. To determine the impact of aid on poverty reduction 3. To assess the effect of domestic politics on aid’s effectiveness in improving human welfare

1.4: Research Scope

The study employs a macro-level scope of analysis in its investigation of the relationship between donor intervention, economic growth and poverty reduction in the case of Sierra Leone. Three reasons are advanced for adopting a country macro study. The first reason has to do partly with the fact that most of the data available for analysing the outcome variables (economic growth, pro-poor growth and aggregate human welfare) as well as the determinants of these outcome variables are largely available at the macro level for the country. The second reason is related to the fact that foreign aid has been largely argued to be fungible (Pack and Pack, 1993; Feyzioglu et al., 1998; Dreher et al., 2008), and which implies micro analysis at the project and sectoral level may produce unrealistic findings. The third reason has to do partly with the need to recommend policy at the macro level which is applicable at the general country level. Hence, the need for generalisation of the findings at the country level is enhanced with a macro-level study rather than micro-level study of individual projects financed by foreign aid.

The study period covers 1970-2007, a period largely sufficient for time series analysis and for which data is readily available on the variables of the various models employed in the study. Based on this time line, the research was able to examine the long-run and short-run impact of foreign aid on economic growth and pro-poor growth in Sierra Leone and further investigate the contribution of foreign aid on aggregate human welfare for Africa and Sierra Leone. However, for the analysis of the impact of aid on human welfare, the period of

26 analysis covers 1980-2007 as UNDP data for human development prior to 1980 has not been available as at the time of the study.

The research also involved in-depth interviews with policy officials, donor agencies, civil society and the media, political elites and public sector audit institutions including the Anti- Corruption Commission, parliamentary oversight committees, government internal audit and Auditor General’s office. However, for the scope of this thesis, only some excerpts from these interviews are utilized in the analysis.

1.5: Thesis Chapter Outline

The Thesis is divided into eight chapters. Chapter 1 provides an introduction to the thesis, highlighting the background and justification for the study as well as its purpose and objectives. The next chapter (chapter 2) provides an overview of the study country’s political economy as well as foreign aid flows and economic performance. This chapter describes the country context for this study in which the intervention of donors in the context of the effectiveness of their assistance to the country is investigated. It provides a descriptive analysis of the country’s economic standing relative to other developmental groups and also over time and further illustrates the country’s attractiveness to foreign aid.

Chapter 3 provides a review of the literature on aid effectiveness covering both theoretical and empirical literature. Specifically, literature is reviewed on the impact of foreign aid on economic growth and poverty reduction and as well on the (political) factors to aid effectiveness. Empirical research that has been done on both cross-country and country econometric analysis is carefully reviewed with such providing a basis for the conceptual framework and model specification for the conduct of this assessment.

Chapter 4 describes the conceptual framework and overall research strategy of the study. In the first part, the conceptual framework is described which follows the theory and empirical arguments on the impact of foreign aid on economic growth and poverty reduction as well as the role of politics on aid effectiveness. The framework highlights the explanation of the research questions and hypotheses to be tested as well as the relationship between them. It further defines the concepts and provides a theoretical basis for the variables employed to

27 proxy the concepts to be investigated. The second part of this chapter follows from the conceptual framework and describes the overall methodology of our investigation. It provides a description of the overall research strategy.

The next chapter (Chapter 5) provides the first empirical analysis, presenting the detailed methodology and empirical results for the examination of the impact of foreign aid on economic growth. Results of the analysis of the impact of both total aid and disaggregated aid on economic growth are presented here. The chapter ends by discussing the results and proving a conclusion of the findings of the impact of foreign aid on economic growth in the case of Sierra Leone.

Chapter 6 then gives the empirical analysis of the impact of donor intervention on poverty reduction in the case of Sierra Leone, considering pro-poor growth as the first variant of poverty reduction. Here, the model, estimation techniques and empirical estimates for investigating the impact of foreign aid on pro-poor growth are presented and discussed. The analysis further involves the investigation of the impact of disaggregated aid (and in particular food aid) on pro-poor growth in Sierra Leone.

Chapter 7 then presents the second variant of poverty reduction: improvement in human welfare. Here, the empirical analysis of the impact of foreign aid on human welfare in Africa as well as Sierra Leone is presented and discussed. The chapter further assesses the role of politics (democratisation) on the impact of aid on human welfare, in particular on human development for Sierra Leone relative to Africa.

The final chapter (chapter 8) comprises the conclusion and policy implications of the study, providing policy and further research considerations on the basis of the findings.

CHAPTER 2: FOREIGN AID AND ECONOMIC DEVELOPMENT IN SIERRA LEONE: AN OVERVIEW

2.1 Introduction

Whilst there may be many vital issues in the consideration of Sierra Leone’s dependence on foreign aid since its 1961 independence from Britain, it is obvious that such analysis will be grossly incomplete without consideration of its role on economic output and poverty reduction. These two attributes are the crucial components of any analysis of economic development. Sierra Leone’s economy has been known for its underperformance. Being a largely capital starved economy with low savings and investment ratios, it is not surprising that Sierra Leone is dependent on foreign capital. It is largely on the basis of the necessity to spring from such economic paralysis that international donors have been actively and consistently engaged in the economic development of Sierra Leone. Foreign aid being provided by these donors is by all means critical to provide the needed capital and investment to stimulate the country’s economic growth and ultimately reduce the impoverishment which has characterised this resource rich country for decades.

In this chapter of the thesis, we provide the contextual background analysis of foreign aid and development outcomes in Sierra Leone. Attention is first given to the political background of the study country, Sierra Leone. This involves an analytical review of political regimes, and their nature and practice of governance. We then provide an overview of economic performance in the country over the years comparing economic growth with the rest of Africa, sub-Saharan Africa (SSA) and other developmental groups. This will provide a comparative view of the performance of Sierra Leone’s economy over the years as well as its performance relative to the rest of Africa and the world. It will also highlight the key sectors in the economy that drive overall economic growth in the country. An overview of the country’s performance in terms of human well-being is also captured in this section. In the sections that succeed, we focus our analysis on the background of foreign aid management and correlation with economic growth and poverty reduction in Sierra Leone. Such an overview will provide the background for the empirical analysis of the impact of foreign aid on economic growth and poverty reduction in the country. We conclude in the final section.

2.2 Sierra Leone: Political Background

Sierra Leone lies along the West coast of Africa and neighbours Liberia on the south-east and Guinea in the north and east. It has a size of 71,740 square kilometres and an estimated population of 5.85 Million as at 2007 (World Bank, 2009). The World Bank estimates also show that Sierra Leone’s population is largely rural with 2007 figures suggesting that 62.6% of the population lives in rural settlement while the remaining 37.4% is urban. The country has 12 provincial districts with the northern region, the largest of the country’s four regions hosting five of these districts.

Sierra Leone became a crown colony in 1808 and gained independence from Britain in April 27, 1961. The post-colonial political regimes in Sierra Leone have been characterised by a mix of periods of democratic rule and some level of autocratic rule with the latter largely been made possible by military junta regimes. Appendix 2.1 is a table updated from Sesay (1995:168) that summarises this political heritage of post-colonial Sierra Leone.

From the table of Appendix 2.1, it is seen that Sierra Leones’ political regimes since independence have taken the form of both democratic and autocratic (authoritarian) regimes. Considering the period of 1961-2007, there has been less number of more democratic regimes than otherwise. The multiparty democratic regimes have comprised the regimes of Sir Milton Margai, Sir Albert Margai and Ahmed Tejan Kabbah. In total, these regimes have served 18 years of the country’s 47 years of post-independence. The remainder of the country’s regimes have comprised one-party form of political governance served by Siaka Stevens and Joseph Saidu Momoh (total of 23 years) and military junta regimes involving those of Andrew Juxon-Smith, Valentine Strasser, Maaada Bio and Johnny Paul Koroma (which in total served 6 years). Both military junta and the one-party regimes are arguably autocratic regimes. Sesay (1995) describe the one-party system of political governance by both President Stevens and President Momoh as authoritarian. In effect, power was largely restricted to the executive instead of the parliament (Baland et al., 2010). There were no presidential elections. Even as Stevens retired from power, he manipulated the transfer of power to his successor (Luke and Riley, 1989). Even though there were parliamentary elections during the one-party regime of Siaka Stevens, yet these were violent and manipulated elections with restricted level of political competition (Luke, 1985). It should be noted however that, in Sierra Leone just as it may be in some other developing countries, no post-independence regime unto 2007 could be

30 described as purely democratic or purely autocratic. The description of regimes as shown in appendix 2.1 shows that the form of political regimes has never been perfectly democratic, but neither could any regime be described as strongly authoritarian. Some periods of autocracy had some democratic principles and some periods of democracy experienced some autocratic forms of governance. This is evident in Marshall and Jaggers’ (2008) POLITY IV scoring estimation of the level of democracy and autocracy in the country’s polity regimes. Even to date, when some modern forms of democracy is operational, the state can be best described as in the process of democratisation as some of the institutions required to attain a strong democracy such as the existence of an electoral system that guarantees effective representation (Ndulo, 2003) are still not completely developed. This is evident in the reported fraud in some polling stations in the 2007 presidential and parliamentary elections (Robinson, 2008; DFID 2008).

A common attribute of these regimes, just as is typical of most African regimes is the prevalence of patrimonial practices in politics. Baland et al. (2010: 4612) define patrimonialism also referred to as neo-patrimonialism, or personal rule or clientelism as

“A style of governance when politicians control power through a system of personal relationships where policies/favours are distributed in exchange for political support”

In their precise chapter on governance and development, Baland et al. (2010) traced the origin of bad governance in Sierra Leone and linked it with prevalence of patrimonial politics in the country’s political regimes, particularly that by President Siaka Stevens. This political practice/bad governance, the article argues, has been the most crucial impediment to the country’s economic development and poverty reduction. Baland et al. (2010) suggest that since independence, the sequence of political regimes in the country has had no interest or incentive in providing the basic ingredients that can make a nation prosper; and this they suggest emanates from the political strategy of patrimonialism that had been employed by post-independent politicians in Sierra Leone. Patrimonialism can be generally disastrous for economic policy and development; and Baland et al. (2010) highlight four of these ways. Firstly, the highly inefficient manner in which patrimonial practices take place ensures that the required public goods are generally not attractive to be used by the patrons (for service delivery) in order to gain political support and hence can be undersupplied under patrimonialism. Patrons will rather use private goods to be targeted to supporters but withheld from opponents. Secondly, property rights are insecure under patrimonialism, as this form of

31 political practice creates insecurity and uncertainty in such a way that patrons ensure people become reliant on them for their future success or failure. As property rights can be withdrawn as and when patrons feel, this adversely affects investment. Thirdly, patrimonialism can be detrimental through the creation by political regimes of distortion in market prices in order to create rents which can be politically allocated. Finally, the detrimental impact of patrimonialism is evident in its undermining of the coherence of the bureaucracy though the frequent shuffling of bureaucrats (so as to avoid their conspiracy against rulers) as well as encouraging bureaucrats to be corrupt. In essence, patrimonial practices, which have been evident since independence have been largely associated with bad governance and inefficient and ineffective public service delivery in Sierra Leone.

In as much as patrimonialism may have been more evident in the Siaka Steven’s rule (as is argued by Baland et al., 2010), this study argues that this has been a common attribute across the country’s political regimes. Sesay’s (1995) review of political regimes seen in Appendix 2.1 shows that evidence of patrimonial practices existed in the regimes preceding that by Siaka Stevens. Emerging from a British colonial rule to independence in April 27, 1961, it could be expected that Sierra Leone’s first president at independence would operate under a political rule that is largely democratic with multiparty political competition. However, that does not guarantee the absence of patronage as the then government of Sir Milton Margai longed to build and consolidate patron-client networks to support its political continuity. In this verge, the role of local governance and chieftaincy became crucial as such are closest to and considerably regarded by most rural electorates. Following the early death of Sir Milton Margai in 1964, it was no surprise that his achievement backed by the patron networks he built could guarantee an election of his brother, Sir Albert Margai and secure continuity of the ruling party in governance. Neopatrimonial politics under such scenario could not be astonishing, and therefore, Albert Margai’s political rule, though with some notable reforms, was largely characterised by party politics, tribalism, corruption and largely authoritarian in his last days of power. This provided the necessary impetus for unpopularity of the ruling party, a fact proved in the 1967 elections when the opposition party of President Siaka Stevens claimed a contentious victory. But as institutions were already very fragile following highly contested elections under notable country divisions engineered by patrimonial politics (and with the larger proportion of the military officials coming from South-Eastern strong base of the then ruling SLPP party), it merely provided an invitation for military intervention.

However, a counter coup ensured that the elected government of Siaka Stevens assumed power in 1968.

Steven’s reign of power has been characterised by relative political stability, but with economic decline, largely undemocratic polity and surge in neo-patrimonial politics. It is therefore no surprise that under such features, corruption of state resources can ensue. This is the regime that Baland et al.’s (2010) review of bad governance in Sierra Leone largely associate with high levels of patrimonialism. Building on the grounds of a divided country, Stevens introduced a republic and formed a one-party system of governance under the All People’s Congress (APC) party (Baland et al., 2010). The formation of the one-party form of governance implied political competition was eliminated at the presidential level and only restricted to parliamentary elections. This did not only ensure that Stevens ruled as the longest serving head of state (17years), but further made sure that he built and strengthened patron-client networks to support his uncontested leadership and violent parliamentary elections. Stevens failed to introduce economic reforms as he himself admitted that politics was his priority against economic development (Luke and Riley, 1989). His failure to implement the contractionary IMF suggested economic reforms was partly as the need to avoid the political and social costs of such austerity measures. Luke and Riley (1989:137) put it in a direct way:

“To have acted vigorously in these policy areas would have undermined the patronage bases of the regime and associated public-office corruption, and angered the normally quiescent urban population-especially the soldiers, police, and civil servants ”.

Stevens used the social networks of his political party (which came into power with strong support from the north - mainly from the Temnes and Limbas) to build his patrimonial regime, and as well as largely changing political institutions (as previously emphasised) shortly upon assuming power in 1967 (Baland et al., 2010).

Though President Joseph Momoh, who peacefully took over power from President Stevens, succeeded in amending the country’s constitution with the major amendment being the reintroduction of multiparty elections (passed in 1991), yet the then prevailing political instability, economic hardship, discriminatory reforms in the diamond mining, the growing powerfulness of his ‘Ecutay’ cabal in the government, with power above that of the cabinet

33 and parliament (Zack-Williams, 1999), characterised his political governance. Thus patrimonial politics was also a feature of the Momoh regime.

Even with military regimes, patrimonial politics was also evident. In particular, the political association of the military NPRC leaders and the SLPP party suggests some level of patrimonial politics at play. Maada Bio, who toppled his NPRC compatriot Valentine Strasser and passed over power to the democratic civilian regime of Tejan Kabbah of the SPPP party, has himself become the flag bearer as presidential candidate for the 2012 presidential elections under this same party (The Patriotic Vanguard, 2011). Further, Tejan Kabbah was coincidentally a key adviser to the military regime, which implies the military regime of the NPRC may have been subject to political manipulation and association with the SLPP party. Hence, even in the military, patrimonialism has been evident.

Despite being largely democratic, the current government of Ernest Bai Koroma, following fieldwork interviews has also been largely criticised for re-instituting patronage practices particularly in the appointment of government officials. Hence, patrimonialism has been a common feature displayed by political regimes in the country’s post-independent history, and not only in the one-party regimes of President Stevens and President Momoh as is emphasised by Baland et al. (2010).

A final point worth noting in Sierra Leone’s political history is that a substantial number of years in the country’s post-independence period had been subject to political instability partly as a result of coup d’états, but mostly due to the country’s civil conflict which lasted from 1991-2002. Political regimes during this period of rule ranging from President Momoh, Captain Valentine Strasser, Maada Bio and particularly Tejan Kabbah have had their governance significantly interrupted. Though he ensured a peaceful end to the 11 year conflict in 2002, President Kabbah’s leadership in the first term of office had encountered some notable instability. In 1997, he was overthrown from power by some factions of military (headed by Major Johnny Paul Koroma) who eventually cooperated with the rebels in ruling the country. President Kabbah fought back and regained power in 1998 with the help of ECOMOG and British forces. In January 1999, his seat of power in Freetown was also temporary destabilised by the fighting Revolutionary United Front (RUF) and the Armed Forces Ruling Council (AFRC) rebels. Few months later, he succeeded again with the help of ECOMOG in securing the city and much of the country. Before the end of that same year major peace talks with the fighting factions began which eventually resulted in the

34 disarmament of the fighting forces and the formal declaration of the end of the war in 2002. It would be no surprise therefore that economic growth and human welfare can be adversely affected by such political instability. Further, because political and economic institutions are affected during this period, the potential effectiveness of aid could as well be adversely affected.

In essence, Sierra Leone’s post-independence political regimes have been both in forms of democratic and authoritarian regimes; and have been subject to lengthy periods of political instability which occurred though the country’s 11 year civil conflict and military coups d’états. The result is that the lack of strong institutions exacerbated by the moderation of political competition in either one-party, military or junta regimes may have contributed to slow economic development in the country. Further, the analysis showed that patrimonial politics characterised by patronage, tribalism and corruption has been a common feature of the country’s post-independence political regimes. However, with democratisation and the increasing realisation of Sierra Leone as a nation state alongside the increased sense of nationalism as well as the increased citizens’ preference for development rather than patrimonial exchanges, the intensity of patrimonialism may have minimised (Baland et al 2010), particularly since the end of the civil conflict. But, whilst democratisation may have ensured much improvement in political practices, the realisation of the benefit of economic improvement by the vastly poor Sierra Leoneans has remained largely unattained despite the immense natural resource endowment of a country possessing manageable agricultural lands and rich in mineral resources such as diamonds, gold, rutile, bauxite and perhaps most importantly iron ore and recently oil –as per recent oil discovery.

Hence, this study argues that politics mattered in the country’s development. In particular, lengthy periods of political instability together with evidence of patrimonial politics may have significantly deterred the development of the country. However, it will be important to see whether cumulatively, the combined form of polity rule in the country (taking into cognisance multiparty democracy, one-party governance and military junta rules) has not been detrimental to the country’s economic development and aid effectiveness. If at all it has not, this could be explained by the favourable incentives of democracy cushioning the detrimental effect of military rule/decree and one-party rule (with limited political competition). It could as well imply that the one party period of political rule has not been that detrimental to the economy’s growth to the extent that combined with periods of multiparty democracy, the cumulative effect of politics has not been detrimental to the

35 development process. However, autocracy should adversely affect human development and aid effectiveness. Having reviewed Sierra Leone’s post-independence political regimes and their implication on the country’s economic development, in the following, we present an overview of the economic performance of the country since independence to provide some descriptive evidence of how the country had performed relative to regional and economic groups as well as over time. This will help to put into context our investigation of the relationship between foreign aid, economic growth and poverty reduction in Sierra Leone.

2.3 Economic Performance in Post-independent Sierra Leone Following from the overview of the political background of Sierra Leone since independence in 1961, this section provides an overview of the country’s post independent economic performance, but however commencing from 1970, the period that matches the commencement of the quantitative/empirical analysis as used in this research. The analysis of this overview takes the format of comparative trend analyses from five-yearly averaged data as well as overall period averages compared with developmental groupings comprising of Africa, sub-Saharan Africa (SSA), low-income countries, developing countries (low-income + middle income countries), and Highly Indebted Poor Countries (HIPC).

Appendix 2.2 and Figure 2.1 below compare trends in GDP per capita growth for Sierra Leone against the developmental groups and regions. The analysis shows that the trend in GDP per capita growth for Sierra Leone has been fluctuating since the 1970s just as it does for the other developmental groups. Generally, there has been an increase in Sierra Leone’s GDP per capita from 1.85% during 1970-74 to 3.61% during 2005-07. Improvements in GDP per capita growth were seen during 1970-74, 1980-84 and in the post-war periods of 2000-04 and 2005-07. Despite the overall increase in trends, there were lengthy periods of unimpressive GDP per capita growth for the country. Real GDP per capita growth was in the negative region during 1975-79 and from 1985-1999. The highest fall in growth of GDP per capita was evident during 1995-99 with a decline of 6.28%. This period coincided with the peak of the civil war especially in the urban areas halting business activities, manufacturing and services whose operations are more dominant in the country’s urban settings. This period also witnessed political instability in terms of coup d’état as shown in Appendix 2.1. Hence,

36 it is no surprise that with intense fighting during this period, economic activities were seriously destabilised resulting in a decline in economic growth of this magnitude.

Compared to the rest of the developmental groups specified in Appendix 2.2 and Figure 2.1 below, it is evident that the overall trend has been an increase in GDP per capita growth not only for Sierra Leone but for all the developmental groups. Even sub-Saharan Africa, despite its usual criticism for underperforming, showed an overall trend of growth in GDP per capita from 2.87% during 1970-74 to 3.53% during 2005-07. However SSA’s growth was in the decline from 1975-79 right through to 1990-95. It should however be noted that this trend has been fluctuating over the years.

Figure 2.1: GDP per capita growth: Sierra Leone versus developmental groups. Source: Data from World Bank’s World development indicators. Periods shown on the horizontal axis are 5-year averages. The chart compares the performance of Sierra Leone with regards to economic growth relative to HIPCs, low income countries, developing countries, and Sub- Saharan Africa for the period 1970-2007.

What appears evident from this comparative economic performance is that since the commissioning of the International development goals: namely the MDGs in 2000, GDP per capita growth for Sierra Leone just as with the rest of these developmental sub-groups has been impressive. From 2000 onwards, growth has been positive for all these groups including Sierra Leone. For HIPC countries and indeed most other developing countries, this period also coincided with the debt relief and the poverty oriented developmental assistance through PRSPs and MDGs. Evident even further, is that the growth of Sierra Leone during this period

37 has even seemed to outperform these developmental groups. During 2000-04 particularly, while Sierra Leone showed an average GDP per capita growth of 9.28%, the other developmental groups showed a comparative lower average growth for the same period. This may be probably due to the emergence of post-conflict reconstruction and economic activity for the country after recessional period in 1999 following the intensified civil war all over the country. Agricultural activities, mining, manufacturing and trading resumed and at notable momentum during this period, and hence economic growth is expected to boom as private investment and particularly FDIs flows could have significantly increased. Thus, for the case of Sierra Leone, in addition to increased commitment of development partners through MDGs and PRSPs, the resumption of economic activities following the end of intensive rebel activities all over the country and indeed the pronouncement of the end of the war in 2002 may have all contributed to the impressive economic growth during 2000-04 to the extent the country outperforms other developmental sub-groups during this period.

Figure 2.2: Average GDP per capita growth rates (1970-2007) – expressed in percentages on the vertical axis. Source: Data from World Bank’s world development indicators.

Figure 2.2 above follows from the aforementioned analysis of trends in per capita GDP growth for Sierra Leone and developmental groups and here presents a chart of overall average per capita GDP growth from 1970-2007. The performance of developing countries as was initially pointed out in the previous analysis is even clearly evident here with an overall average GDP per capita growth of 2.87%. HIPC countries showed the least overall average GDP per capita growth with as low as 0.33% from 1970-2007. The underperformance of Sierra Leone is also worth noting with an overall average GDP per capita growth of 0.36%

38 during 1970-2007. Thus, the periods of decline in GDP growth per capita from 1985-89 right through the intensive conflict periods from 1991-99 may have had some damaging effect on the overall economic performance of the country. At an overall average GDP per capita growth of 0.36%, the only sub-group that Sierra Leone outperforms is the HIPC which showed an overall average of 0.33% for the entire period of 1970-2007.

In national accounting, three component sectors of an economy usually constitute the key sectors that contribute to a country’s economic output (i.e. GDP). These sectors are normally agriculture, industry and services; alternatively referred to as primary, secondary and tertiary sectors respectively. The agricultural sector usually comprises crop production, animal husbandry, hunting, forestry and fishing. The industry sector comprises manufacturing and mining and quarrying; whilst the services sector usually comprises trade, hotels and restaurants, transport, communication, financing and insurance, real estate and business services. Analysing the sectoral contribution of these sectors to the Sierra Leone economy, it is evident that on average for the entire period of 1970-2007, agriculture contributes the highest proportion to the country’s economic output (39.8%) as against 31.5% from services sector and 22.9% from industry sector (Appendix 2.5)

In figure 2.3 below, a chart is constructed showing a trend in the relative contribution of these sectors on total economic output over time. The chart shows that the contributions of these sectors on GDP all appear to fluctuate over the period of the study. The contribution of agriculture, besides two periods (1980 and 1990) of swaggering, has however seemed to be on the increase unto 1995 before it started falling. In the early periods before 1985, it could be seen that the services sector was the most contributing sector to the economy outperforming both agriculture and industry. After this period, the relative contribution of services sector declined sharply while agriculture emerged since then as the most contributing sector to the economy. Whilst agriculture may have later emerged as the most contributing sector to the economy, it is important to note that from the chart it is vivid that since the 1995-99 period, the contribution of the services sector had continued to be on the rise whilst that for agriculture rather declined consistently unto 20005-07 period.

Figure 2.3: Sierra Leone: Sectoral Contribution to GDP. The chart shows the contribution of each of the three crucial sectors (agriculture, industry and services) to GDP – expressed in percentages. Source: Data from World Bank’s World development indicators. Year shown on the horizontal axis are in 5-year averages.

The country’s poor performance in terms of economic growth is also mirrored in its performance in terms of poverty reduction, both trend-wise and relative to the African content as a group. Using two measure of aggregate human welfare: the human development index (HDI) and infant mortality rate (IMR), it is evident that Sierra Leone has performed quite poorly in terms of improved levels of human welfare. In Appendix 2.3 and figure 2.4 below, five-year averages are presented for the period 1980-2007 of the performance of the country across this trend, as well as relative to Africa and Sub-Saharan Africa with respect to the UNDP’s estimated human development index.

In terms of the magnitude of the human development index, it is observed that Sierra Leone’s score on the index is considerably lower than the average for SSA and Africa. The overall average score for the period 1980-2007 for Sierra Leone stands at 0.25 while that for SSA is 0.42 and that for Africa as 0.44. Hence, it is clearly observed that the level of human development in Sierra Leone is way lower than that on average for SSA and Africa even though both happen to be the region and continent with the most dismal welfare statistics in the world.

Similarly, trend-wise, the trend in human development for the country does not seem impressive; thus being consistent with the weak performance in average levels. However, though not quite notable, yet the trend is not as weak as the levels performance show relative

40 to the rest of Africa. The trend from 1980-84 to 1985-89 for Sierra Leone is relatively better than that for SSA and Africa as a continent (figure 2.4). However, whilst Sierra Leone’s HDI trend level seems to slightly decline and stagnate from then unto 1995-99, that for SSA and Africa has rather increased for the same period. This however, may have been largely due to the adverse impact of the civil war on human well-being in the country. This is justified by the increase in the HDI for the country in the aftermath of major hostilities evident unto 2005-07 at a rate slightly higher than that for SSA and Africa. The slope of the increasing trend tends to be higher than that for SSA and Africa, meaning that the rate of increase of human development for the country is slightly higher than that for Africa during this period.

Figure 2.4: Trends in human development index (HDI). The HDI is computed by the UNDP and comprises a simple average of per capita income, life expectancy and literacy rates. The higher the score of the index, the better the human welfare levels and by implication the lower the poverty levels.

Hence, on the basis of trend in the post-war period, Sierra Leone’s performance on the HDI has been somewhat inspiring. This in itself just indicates how damaging the civil conflict must have been on welfare. Thus, though relative to the rest of the world and even Africa, the country remains with damningly low HDI levels, but in terms of its improvement over time there has been some progress particularly since the end of the civil conflict.

Sierra Leone’s weak poverty status with respect to low HDI, is also reflected in the country’s high infant mortality rate which seems to be the highest in the world (Appendix 2.4 and Figure 2.5). For the period 1980-2007, statistics show that on average in Sierra Leone, 177 infants die out of every 1000 births compared the average of 97 for SSA and 92 for Africa.

Hence, the country has one of the highest infant mortality rates and lowest human development index in the world, to the extent that even with the particularly underperforming African continent, the statistics show that on average Africa is even a far better performer in terms of welfare statistics.

In terms of trend as well, though there is seemingly an improvement as the trend shows a falling IMR, yet the rate at which it is declining is quite weak thus indicating an unimpressive reduction of IMR over time. In fact, though there is an overall falling trend, between 1985-89 and 1990-94, there was a rise in infant mortality rate in the country.

However, the weak fall in IMR over time for Sierra Leone as is shown by the trend has not been different when compared to Africa and SSA. The fall in IMR over time is quite weak for the continent to the extent it’s almost showing a stagnation in the IMR over the years. The decline is not quite conspicuous for the region when seen from the chart (figure 2.5). Thus, though in terms of average, the number of infant deaths per 1000 births for Africa and SSA may have been considerably lower when compared to Sierra Leone, yet in terms of the improvement in this IMR over time, the continent, just like the country seems to perform poorly with weak and hardly noticeable declines in IMR over time.

Figure 2.5: Trends in infant mortality rate (IMR). This is measured as the number of infant death per 1000 live births and is sourced from the World Bank world development indicators and the UNDP’s human development reports (various). Figures are in five year averages from 1980 to 2007. 1980 as a five year average implies an average from 1980-84, and so forth. The lower the infant mortality rate the better is the welfare level.

Hence, whilst trends in HDI may have shown some improvement for the country over the study period and especially in the post-war era, yet in terms of IMR, trends as well as levels have just been equally unimpressive.

2.4 Foreign aid management and flows in Sierra Leone

The management and administration of foreign aid in Sierra Leone has been largely the responsibility of the country’s Ministry of Finance and Economic Development in coordination with resident donor agencies. Following increased aid flows in the post-war era, there was the need to better coordinate aid flows into the country and temporarily relieve the already busy Ministry of Finance of the coordination aspect of aid management. Hence, with the country’s vice president being prominent within the then ruling SLPP government, this coordinating role was passed to the office of the vice president. A department under this office, namely, the Development Assistance Coordinating Office (DACO) was set up with the responsibility of coordinating development assistance from the country’s official donors. Of recent, (2010), the role of aid coordination has been restored back to the Ministry of Finance, with DACO remaining as a unit within the ministry. A 2008 DACO report on development assistance to the country for the year 2006 shows that the country’s major donors comprise the UK Department for International Development (DFID), the European Commission (EU), the World Bank, the African Development Bank (AfDB), United Nations agencies, USAID, the International Monetary Fund (IMF), Irish Aid and China. Of these donors, five (DFID, World Bank, AfDB, EU, and UNDP 2) form part of the Multi-Donor Budget Support (MDBS) partners providing direct budget support to the government through pooled basket funding.

The overall target of foreign aid since the end of the war has been primarily poverty reduction. After the war, aid has been provided to support post-war recovery and the implementation of poverty reduction programmes. The attention of aid shifts with shift in national priorities defined in the PRSP. Poverty reduction is the overall goal for aid-giving towards which maintaining peace, promoting good governance, and enhancing growth are preconditions (IMF, 2007). As part of this research, interviews were conducted with resident

2 The UNDP has however pulled out of the country’s MDBS partners as at the period of fieldwork interviews in 2010. 43 donor agencies of which six of the country’s main donors were interviewed. One of the issues covered in the interviews was the purpose of donor intervention in the country. An interview with a representative of the AfDB revealed that Sierra Leone mostly benefits from two of its funding windows: The African Development Fund (ADF) window, which is the concessionary window, provides more of grants to the country for growth related investment and poverty reduction ventures. The second window the country benefits from is the fragile state facility window, which as the name implies is given to fragile state economies (of which Sierra Leone qualifies as at the time of the interviews in 2010) for the purpose of economic recovery, poverty reduction and peace consolidation. Another interview with a representative of the World Bank also stressed the reduction of poverty as the primary objective of their intervention in the country, in addition to the fact that Sierra Leone is a contributing member of the World Bank. Even DFID, the country’s leading donor, confirmed that though political reasons (such as democracy and historical colonial association) for granting aid to Sierra Leone cannot be ruled out, DFID’s aid allocation is mostly based on a resource allocation model which itself depends on recipient country needs (comprising population levels and income levels). Hence, granting aid to Sierra Leone is largely based on the low income level of the country and hence with the need to raise national income to improve standards of living and welfare of its citizens. Again, the representative of the IMF confirmed that their intervention in Sierra Leone is to support balance of payment problems and hence economic stabilisation, which itself is a precondition for economic growth. Hence, it is evident that all these donors indicate their intervention in the country as being largely focused on economic growth and poverty reduction.

Evidence on the target of aid 3 during and before the war is not quite defined, but may not be quite different from the purpose of poverty reduction (as suggested by the country’s main resident agencies and as is seen in the post-war era PRSPs and MDGs) of which growth and peace are preconditions. Sierra Leone being characterised by high poverty levels before and during the war, it is not surprising that donor aid target has a poverty reducing motive. In fact, the target of aid in Sierra Leone could not be different from the historical target of aid to recipient countries and Africa in particular since its inception. This is particularly so because Sierra Leone has been a largely aid dependent country as will be shown in this section. The purpose of donor assistance to Africa has had the main motive of promoting economic

3 Note Aid as used throughout this chapter is Official Development Assistance (ODA) as % of GDP 44 growth either through investment or via the restructuring of the public sector in the recipient countries and the reduction of poverty and inequality (Easterly, 2009). Thus as the target of foreign aid to the country has been largely developmental, an assessment of the effectiveness of such aid to Sierra Leone over the years cannot be conducted without assessing its impact on economic growth and poverty reduction which are the main measures of economic development.

Sierra Leone has been a highly aid dependent country with volumes of aid disbursed higher than even the average for the most aid dependent sub-region – SSA. Figure 2.6 shows foreign aid disbursed to Sierra Leone for the period 1970-2007 compared with SSA and Africa the most aided sub-region and continent.

Figure 2.6: Average foreign aid disbursed 1970-2007. This implies an overall average of official development assistance as % of GDP for the entire period of 1970-2007. Data from World Bank (2009) African Development Indicators

From the chart, evidence of high aid intensity in Sierra Leone is clearly observed. At an average disbursed aid of 14.2% of GDP, Sierra Leone gets about three times the SSA average and nearly four times the African average for the period 1970-2007. This implies despite the fact that Africa happens to be the most aided continent, yet Sierra Leone, a constituent state of this continent is by far a higher recipient of foreign aid. Disaggregating this aid by source and type, figure 2.7 shows that on average for the period 1970-2007, bilateral donors (7.5% of GDP) tend to marginally disburse more aid to the country compared to multilateral donors (6.6% of GDP). In terms of aid being disbursed as either grants or loans, the chart shows that

45 for the period 1970-2007, aid in the form of grants tends to be the most disbursed type on average (10.3% of GDP) compared to loans (3.8% of GDP). Aid in the form of technical assistance constitutes just around 2.6% of GDP on average for the period 1970-2007, whilst that disbursed to the country that has nothing to directly do with technical assistance amount to 11.6% of GDP. Hence, the study hopes to quantitatively prove (in the econometric analysis) whether the higher the aid type disbursed on average, the better the effectiveness in terms of fostering economic growth and reducing poverty.

Figure 2.7: Sierra Leone: Average disaggregated aid flows (1970-2007). Source: Data from OECD. All values are as % of GDP. TCA denotes technical assistance; NTCA denotes non-technical assistance; MA denotes multilateral assistance and BA denotes bilateral assistance.

However, whilst the above analysis may only paint a picture in levels terms (since only period averages are reported), it is important to see how aid flows into the country change with time. Hence, in Figure 2.8, comparative analysis of total aid (from all donors) disbursed to Sierra Leone compared to Africa and SSA over five year averages for the period 1970- 2007 is presented.

Trend-wise, the flow of aid to Sierra Leone tends to mimic that for the entire African continent and SSA region. From the period 1970-74 to 1980-84, a steady increase of aid can be observed for the country which is also similar for Africa and SSA. However, the trend for Sierra Leone during this period tends to be slightly steeper. From 1980-84 to 1990-94, foreign aid flows into the country continued to increase and again this trend is similarly seen for Africa and SSA. Yet again, the trend for Sierra Leone shows that for this period the

46 increased inflows of aid to the country tend to increase at a higher rate compared to Africa and SSA as the chart emerges to be steeper. The sharp trend in this period is not surprising. This was the period when the country was seriously suffering from IFI imposed adjustment policies with notable restrictive expenditures on the economy (Luke and Riley, 1989; Sesay, 1995); and as GDP computation is largely expenditure based, it was clear that GDP during this period was unimpressive. This difficult period marked an era of increased donor dependency; and with the country’s cooperative participation in the adjustment effort, total donor assistance disbursed in absolute financial terms increased by 22% (World Development Indicators - online). Hence, it is expected that with increased ODA flows combined with unimpressive GDP, aid as percent of GDP will consequently increase.

During 1990-94, increased aid could not have also been a surprise as though it coincided with the early period of the civil conflict in the country which started from 1991, yet it also coincided with the change of governance from a predominantly unpopular government/party (by then) that had been in power for 23years to a young and seemingly innocent military regime (Zack-Williams, 1999). Donors responded in support to the change of government by disbursing more aid to the country with the hope that it could be better managed once there had been a change of the corrupt government (Zack-Williams, 1999). In effect, there is expected to be an increase in aid flows to the country during this period.

However, from 1990-94 to 1995-99, that was a period associated with significant political instability as had been earlier explained. This instability constituted a military government in 1992, civilian democratic government in 1996, military junta in 1997, civilian democratic government again in 1998, rebels entering the capital city in 1999, and finally the democratic government retaking control of the city and governance. Therefore, the intensity of the civil conflict backed by the double change of heads of state during this period implied donors saw the country as a fragile state to be confidently disbursing development assistance and therefore, the observed sharp decline in aid during this period could be expected. The decline in aid flows during this period was also observed for Africa and SSA, which could be associated with general fall in aid following the end of the cold war (Hjertholm and White, 2000). However, the decline is steeper for Sierra Leone compared to Africa and SSA, which just reflects the added consequence of the country’s political instability during this period.

From 1995-99 to 2000-04, whilst an observed stagnation of aid to Africa and SSA could be seen, in the case of Sierra Leone, foreign aid flows into the country significantly increased with a sharp increase in the trend during this period. For Sierra Leone, this period coincided with the end of the war and the commencement of the country’s economic reconstruction. Hence, donors responded by throwing in considerable amount of aid to the country to support its reconstruction and development efforts. From 2000-04 to 2005-07, though aid flows were still high on average, even higher than that for Africa and SSA, yet in terms of trends, it declined whilst Africa and SSA showed a slight increase during this same period. But as the immediate post-conflict period gradually fades away, it is expected that aid intensity will consequently reduce, at least minimally. It should however be noted that, even though a decline was evident during this period, yet aid flows to the country did not reduce to pre-war levels.

Figure 2.8: Comparative Analysis of foreign aid disbursed to Africa, Sub-Saharan Africa and Sierra Leone from 1970-2007 (five year averages). Source: derived from World Bank’s (2009) African Development Indicators. 1970 as shown on the horizontal axis implies average for 1970-74, and so forth except 2005 which is 2005-2007

Based on this analysis so far, the following key points are worth noting. Firstly, the target of aid to Sierra Leone has been largely developmental, in particular improving living standards and economic restoration in the post-war era. Secondly, there is some evidence of donor coordination (with the establishment of the Multi-Donor Budget Support partners) in pursuant of the recommendations of the Paris Declaration and Accra Declaration. Thirdly, the descriptive analysis has shown that Sierra Leone is largely aid dependent and the level of aid intensity in the country is by far much higher than the average for sub-Saharan African, the World’s most aided region. Fourth, overall, there has been an increasing trend in aid effort to

Sierra Leone from 1970-2007, though it dropped during the peak of the country’s civil war (1995-1999). This overall increase in aid trends for Sierra Leone may reflect the effect of the ‘big push’ for increased aid to help the development of African countries, but also as result of post-conflict recovery support. It will be worth further having an overview of the correlation between foreign aid and the development outcomes including economic growth and poverty reduction to see whether increases in aid intensity is consistent with these development outcomes. This analysis is done in the next section.

2.5 Foreign aid and Development Outcomes: An Overview

Having analysed the individual trends in economic growth, poverty and foreign aid flows to Sierra Leone for the period 1970-2007, this section provides an overview of the relationship between aid and these development outcomes by presenting simple correlation analysis. This is presented in tables 2.1 and 2.2 below.

In Table 2.1, we present an analysis of the relationship between aid and economic growth using simple correlations. It is evident from this table that there is a positive correlation between foreign aid and economic growth, through the strength of the correlation is only moderate. Foreign aid has a positive correlation with economic growth measured by GDP growth and GDP per capita growth both at around 0.5. This corroborates the propositions of the supplemental theorists who posit that capital (part of which comprise of foreign aid) has a positive relationship with economic growth. However, as this only shows simple correlations, it is important to conduct regression analysis as is done in the main empirical chapters to ascertain the relationship between foreign aid and economic growth for Sierra Leone.

Table 2.1: Simple correlations between foreign aid and economic growth for Sierra Leone (1970-2007) Foreign Aid Real GDP growth Real GDP per Capita growth Foreign Aid 1 Real GDP growth 0.47 1 Real GDP per Capita 0.48 0.99 1 growth Source: Author’s computation with data from World Bank (2009) African Development Indicators (Online)

Whilst an overview of donor effort in disbursing aid to Sierra Leone may have tended to be largely correlated with improved economic growth in the country, it is also important to

49 further attest the existence of correlation between such donor assistance with well-being of the poor in the country. Hence, in Table 2.2 below, the study computes the simple correlations between aid effort in the country with the well-being indicators comprising human development index and infant mortality rate. The correlation results generally show high levels of correlation between foreign aid and both of human development and infant mortality. This confirms assertions from theoretical literature suggesting that capital (and indeed foreign aid) positively relates with raising livings standards in developing countries.

Table 2.2: Simple correlations between foreign aid and Human Welfare for Sierra Leone (1980-2007) Foreign Aid HDI IMR Foreign Aid 1 HDI 0.72 1 IMR -0.70 -0.82 1 Source: Author’s computation with data from UNDP’s Human Development Reports (various) and the World Bank’s World Development Indicators (on line).

It is clear from this discussion so far that there is some correlation between foreign aid and the development outcomes of economic growth and human welfare consistent with intuition. Thus, it is would not be surprising to find the econometric analysis in the empirical chapters to reveals that foreign aid has impacted on the economic growth and poverty reduction in consonance with intuition. Further, it is observed that though on overall, welfare indicators tend to correlate with foreign aid, yet the HDI tends to show a slightly higher correlation coefficient than does the IMR. This leads us to identify some key points in the conclusion that follows.

2.6 Conclusion

Concluding from this background analysis, it is obvious that Sierra Leone’s economic performance has been remarkably weak compared to the rest of Africa and SSA as a whole, even though this continent happens to be the least performing in the world. In spite of being a resource rich country with the potential for a good standard of living, the country’s inhabitants have rather had to do with one of the world’s worst human welfare and economic growth. Hence, in absolute terms, Sierra Leone’s economic growth for 1970-2007 has averaged much lower than the average for SSA, low-income countries and developing countries in their groups for this same period. With an average per capita GDP growth rate of

0.36% over this period, it only stands above the 0.33% average for HIPC countries. At such a rate, it is difficult to see how standards of living can be guaranteed for its inhabitants. This is reflected in the equivalently low levels of human welfare in the country to the extent that Sierra Leone has been consistently placed at the bottom rank of the UNDP’s human development index for several years as Appendix 1.2 shows. The background analysis showed that relative to Africa and SSA, Sierra Leone’s HDI and IMR are far much worse to the extent it appears to project the world’s weakest continent and sub-region as a performing region.

Trend-wise, however, the performance of developing countries and other groups including Africa, SSA and low-income countries do not appear to be different or much impressive than that of Sierra Leone. Trends in economic growth, HDI and IMR are somehow similar with that for the country, for the same period of analysis.

When aid intensity is compared with these developmental outcomes (economic growth and welfare), the analysis showed that aid tends to be correlated with these development outcomes. However, as this may only show an analysis of simple correlations, it remains to be further seen whether in fact there is any significant relationship between aid and these development outcomes for the country during this period of study. Whilst the country’s development outcomes may be quite low on average compared to other developmental groups, its indifferent performance over time with other groups of underdeveloped countries shows that the country is not performing that badly over time. It is just that the pace of development may be slower to enable the country to catch up. And hence, it would not be surprising to find aid to be statistically significant in determining growth and welfare in the country, but probably finding the magnitude of the effect to be small to make notable economic significance. Even if foreign aid was found to be significantly related to these development outcomes, their low or slow performance may also imply that aid alone may not be sufficient to move the country from its current low and slow development level to that of high performing developing countries.

However, the evidence of a positive correlation between aid and development outcomes for Sierra Leone would imply unsurprising results in the econometric analysis should we find foreign aid to significantly contribute to fostering economic growth and reducing poverty in the country as this study hypothesise. In all of this, it is important to establish quantitatively 51 through empirical/econometric analysis whether aid significantly improves economic growth and poverty reduction in the country. This is what this research does in the empirical chapters of this thesis. In the next chapter, we focus on the review of the literature on the relationship between foreign aid, economic growth and poverty reduction.

CHAPTER 3: THE EFFECTIVENESS OF FOREIGN AID IN DEVELOPING COUNTRIES : A SURVEY OF THE EVIDENCE

This chapter is a review of the literature, both theoretical and empirical, on foreign aid effectiveness. The review is structured in conformity with the objectives of the thesis. Hence, the first section of the literature reviews the theoretical propositions and empirical findings on the relationship between foreign aid and economic growth, while the second section reviews the evidence on the small but increasingly appealing literature on the relationship between foreign aid and poverty reduction. The third section reviews the literature on the role of recipient country political system on the effectiveness of foreign aid. Section four reviews the literature on aid disaggregation, while section five provides a summary of the survey of the evidence.

3.1: The Literature on the Relationship between Foreign Aid and Economic Growth

Research on the effectiveness of foreign aid started with an assessment of the impact of aid on investment and eventually focused on the effect on economic growth of the recipient countries. Even today, though foreign aid is found to target poverty reduction as a primary objective, yet its impact on economic growth remains recognised as many development economists and donors are still convinced that poverty reduction could well be attained once economic growth is assumed. This section of the literature reviews the aid-growth theories, explaining how aid is theoretically postulated to affect economic growth. It then provides a review of the empirical evidence at both the cross-country and country analysis on the aid- growth relationship.

3.1.1: The Aid-Growth Relationship: What does Theory Suggest?

White (1974) categorises the theories of foreign aid into two sets: those based on economic theories and those associated with political theory. Economic theories of foreign aid have their basis on the resource transfer side of aid. White (1974) sees these theories as being of two sets: the supplemental theories and the displacement theories. This work’s review of the theories of foreign aid adopts White’s (1974) classification of economic theories of foreign

53 aid (supplemental versus displacement theories), and further reviews the economic growth theories upon which the contemporary literature on aid effectiveness have largely based their empirical models. In essence, the first part of the review brings out the argument between supplemental and displacement theories of foreign aid, highlighting their propositions and the basis of their argument. Then, we review the growth theories involving the Harrod-Domar, neoclassical and endogenous (new growth) theories, for which their postulation of the effect of foreign aid is largely in conformity with the supplemental argument of foreign aid.

Theory suggests that foreign aid promotes economic growth by supplementing limited domestic savings as well as the foreign exchange constraints of recipient developing countries. The work by Chenery and Strout (1966) which itself has its basis on the Harrod- Domar model of economic growth, has been important in this respect. The three elements of the Harrod-Domar model are income (growth), investment (savings) and capital-output ratio (ICOR) related in the form: g = I/ICOR

The incremental capital-out ratio (ICOR) represents ratio of additional investment to additional output; ‘g’ is the growth rate of the economy; and ‘I’ represents investment (which is equated to savings). Hence, with the ICOR remaining constant, the rate of economic growth will be directly determined by the rate of investment. With investment assumed to be equal to savings, this implies that a poor country with low savings will have low investment and therefore low growth. It is thus expected that a supplementation of domestic savings by foreign aid will resort to an increase in investment, and hence economic growth (White, 1974: 114). With the given Harrod-Domar growth equation, the supplemental attribute of aid ensures that when aid is received by a country say, ’a’, that increases the growth rate to g=(s+a)k; where s = savings, and k = output-capital ratio (Griffin1970: 101). If g* is a targeted growth rate of a recipient country, assuming the output-capital ratio, k is constant and considering capital accumulation, c=s+a, it follows that the savings gap or amount of aid required to achieve the targeted growth rate is the difference between s and c (ibid). Chenery and Strout (1966) based the first step of this two-gap analysis in the case where resource limits on skills and savings are important, and described this scenario as ‘investment limited growth’ (ibid: 683). In the second step, they considered the possibility for attaining self-

54 sustaining growth when balance of payments limit is effective and hence describe this situation as ‘trade limited growth’ (ibid: 683).

Calculation of the savings gap is made possible from the Harrod-Domar equations. A savings gap occurs, when the quantum of domestic savings available is less than the amount of investment required to attain the target growth rate, and hence this gap can be filled with foreign aid (ibid). Similarly, a foreign exchange gap arises when the net receipts of the county’s exports fall short of the foreign exchange requirements; the resulting gap being filled by foreign aid (Jhingan, 2004: 472). Hence, the dual-gap analysis makes an important contribution to development theory by emphasising not only the traditional domestic savings gap but also the foreign exchange gap which itself stresses the importance of imports and foreign exchange in the growth process. Thirlwall (2003: 552) reckons that “On the one hand it [dual gap analysis] embraces the traditional view of foreign assistance as merely a boost to domestic saving; on the other hand it takes the more modern view that many of the goods necessary for growth cannot be produced by the developing countries themselves and must therefore be imported with the aid of foreign assistance”.

However, such supplemental theories have in the early literature been seriously challenged by Griffin (1970), Griffin and Enos (1970) and Bauer (1971), all in the context of displacement theorists, as they argue that foreign aid rather displaces the domestic savings of recipient countries through its utilisation in unproductive consumption. Displacement theorists rather contend that aid eases the pressure on the recipient to generate resources for development. If savings is the key resource, the recipient who receives aid, will not make much effort to generate taxes to fund development works. Displacement theorist argue that governments receiving aid use that to finance prestigious projects that are unproductive; whereas domestic savings generated in the private sector would lead to a pattern of investment governed by market forces, thus yielding a higher rate of return. Griffin (1970) for instance, dismisses the assumptions of the supplemental theory models on the premise that increased investment is neither necessary nor sufficient for the attainment of high rate of growth in developing countries. He dismisses the argument that once foreign aid is tied to a particular investment it cannot be switched to consumption. This argument is supported with empirical evidence of which he shows that

S/Y = α + β(A/Y) + ℮ 55

Estimating the model, he arrived at the following result: S/Y = 11.2 – 0.73(A/Y), R² = 0.71 (0.11) (standard error)

From the high coefficient of the aid variable (-73%), Griffin (1970) shows a negative and statistically significant relationship between aid and savings and concluded that aid actually displaces domestic savings. This is further supported with the observation that aid is fungible and hence misappropriation from its intended targets becomes a possibility. Fungibility of aid, according to Griffin (1970: 103) essentially means ‘foreign capital finances not the marginal investment project but the marginal expenditure project, and investment on the margin is just as likely to be (perhaps more likely) to be on consumption goods as on capital goods’. Thus, with large inflows of aid relative to the total investment programme of the aid recipient country, there is the possibility of a decline in the marginal productivity of capital and real interest rate. This fall in real rate of interest will further lead to a fall in domestic savings – thus increasing the likelihood that aid will supplement current consumption more than capital accumulation (ibid).

However, the theories favouring foreign aid as augmenting the recipient country’s domestic savings and hence economic growth have prevailed eventually with the emergence of the neoclassical growth theory and the new endogenous growth theories that both postulate capital (including foreign aid) to be vital for the growth process. The neoclassical model, which is largely inspired by the Solow model of long-run growth (see Solow, 1956), was constructed as an alternative to Harrod-Domar views and hence was devoid of the latter’s critical assumptions of fixed proportions in production. This implies Solow’s neoclassical growth model assumes a continuous production function relating output to the inputs of capital and labour which (as opposed to the Harrod-Domar model) are substitutable and exhibit diminishing returns to the individual factors of production (Jhingan, 2004). The assumption of diminishing returns as is proposed by Solow implies for each unit of additional capital invested into the economy, it produces smaller returns unto the point when no more profits are accrued from additional capital. In terms of foreign aid specifically, the neoclassical model implies that aid contributes to growth via higher investment levels. While the marginal propensity to save in developing countries may be high, the problem is a low average propensity to save because income obtained is spent on purchasing basic services. These low average savings thwart capital accumulation and hence economic growth.

Therefore, foreign aid is to augment this low savings in developing economies in order to stimulate economic growth (Saunders, 2008).

(1+n)k Output Per Capita (A) (1+ δ)k(t) +Sy’

(1+ δ)k(t) +Sy

K’ K’’ Capita Per Capita

Figure 3.1: Solow Model with Foreign Aid Source: Saunders (2008) adapted from Ray (1998)

Figure 3.1 above illustrates how foreign aid increases economic growth, as increased savings through aid flows (from K’ to K’’) shifts the output function from (1+ δ)k(t) + Sy(t) upwards to (1+ δ)k(t) + Sy(t)’. Thus, increasing capital per worker will increase productivity consequently leading to increase in economic growth.

From the shortcomings of the neoclassical model of economic growth was constructed the endogenous growth theory with Arrow (1962), Romer (1986) and Lucas (1988) being the key contributors. The new growth theory acknowledges the importance of endogeneity of capital in the growth process. Another distinguishing attribute between the neoclassical and endogenous growth theories was that whilst the neoclassical growth theory assumes diminishing returns in production, the new growth theory rather assumes constant returns. This implies the new endogenous growth model does not acknowledge diminishing return in production (as is postulated in the neoclassical theory) and hence allows for unbounded long- term growth (Saunders, 2008). This theory also emphasized the importance of human capital to the growth process. Lucas for instance, assumes that investment in education leads to

57 production of human capital which is the crucial determinant of the growth process. Issues of research and development and learning by doing or by investment became important in the new growth theory (Jhingan 2004). The implication of this theory for developing countries is that such countries stand to benefit more from trade with developed countries by drawing upon new knowledge in research and development and new technologies and hence was the need to encourage trade openness. The new growth theory in particular recognises the importance of public policy to economic growth (Jhingan, 2004) and this has been evident in the inclusion of policy variables in empirical aid-growth regressions. Further, the assumption of increasing returns to capital of the new growth model implies that foreign aid will improve growth well into the long-run.

However, these aid-growth theories have not gone without criticisms beyond the challenge by the displacement theorists. For instance, the aid-growth theories based on the Harrod-Domar models have been strongly critiqued in terms of their usefulness to reliably explain the role of capital, and foreign aid in particular in promoting economic growth in developing countries. The closed economy assumption of this model has been strongly singled out for criticism on the basis that international trade plays an important role for economic development of developing countries; and hence that the model was not originally developed for addressing the development problems of nowadays developing countries, but rather for the developed world for which their economic structure differs from that of developing countries (White, 1974; Easterly, 2001; Jhingan, 2004).

The associated two-gap model by Chenery and Strout (1966) though improved on the closed economy assumption of the Harrod-Domar model by assuming an open economy with trade gap being crucial in the role of aid on economic growth, yet it is also flawed with limitations. The restrictive assumptions underpinning the theory are viewed as unrealistic to achieving target growth rates in the developing countries. Further, dual gap analysis is criticised for being a highly aggregative approach considering all types of capital investments as homogenous; which is argued to be unrealistic considering the fact that capital requirements in developing countries are meant for specific needs, and that aid is received for different sectors and projects (White, 1974; Jhingan, 2004).

Likewise, the neoclassical model, though is recognised as a major improvement to the Harrod-Domar model of economic growth as Solow abandoned the unrealistic assumption of 58 fixed proportions in production typical of the Harrod-Domar model, yet it is as well flawed with limitations. First that, whilst recognizing the improvement in the neoclassical model, it as well maintained some of the main features of the Harrod-Domar model such as homogeneity of capital and proportional savings function (Jhingan 2004: 264). In real fact, capital, even in the form of foreign aid comes in heterogeneous forms. Further, the neoclassical growth theory by Solow was criticised for its non-recognition of endogeneity of technical progress and rather treating it as exogenous in the growth process. The new growth theory, though built on the limitation of the neoclassical model, yet still faced criticism especially for its too much emphasis on the role of human capital and neglect of the role of institutions (Jhingan, 2004).

3.1.2: The Empirical Evidence

Following the aforementioned aid-growth theories, a considerable number of empirical studies have been conducted to ascertain the theoretical construct of the aid-growth relationship at both cross-country as well as the country level. However, what seems established from their review is that evidence is inconclusive and mixed, with the findings of some studies confirming the relationship and others not.

3.1.2.1 Aid –Growth Relationship: The Cross-Country Empirical Evidence

The cross-country literature on the relationship between foreign aid and economic growth is by far the most researched compared to the empirical country studies. From the early periods of quantitative aid evaluation to date, cross-country econometric studies remain dominant in the aid effectiveness literature. Whilst in general the purpose of such studies have been to investigate foreign aid’s contribution to economic growth in the developing world, the evidence so far has been inconclusive just as there have been different forms the debate on aid effectiveness had taken. Some studies have proved that aid has a significant role in promoting economic growth, whilst others have provided evidence that it does not. Further, some have argued that differences in the results have been associated with methodological differences in the estimation of the relationship; and yet others base their argument on the existence of an aggregation bias in assessing the impact of aid (reviewed in a separate section – section 3.4) or that the impact of aid is contingent on the recipient country characteristics

(reviewed in section 3.3). In this thesis’s review of the cross-country aid-growth literature, these analytical points are taken into consideration.

Cross-country studies that project a positive relationship between foreign aid and economic growth Those who have found meaningful contribution of aid on growth across the cross-country regressions in the early periods of the literature have included studies by researchers such as Papanek (1973) and Levy (1988). In the more recent literature, a much considerable number of studies (such as Clemens et al., 2004; Loxley and Sackey, 2008; Karras, 2006; Feeny, 2007) do establish the existence of a positively significant impact of foreign aid on economic growth. Papanek (1973) was one of the first researchers to do an empirical analysis on the impact of foreign aid on economic growth, as earlier researchers on aid effectiveness had focused on assessing the impact of foreign aid (or foreign assistance as it was then used) on either savings or investment in tune with the Harrod–Domar models. Papanek’s cross-section analysis shows a positive and significant impact of foreign aid on economic growth. This impact becomes evident following the treatment of foreign aid, foreign investment, other foreign inflows and savings as separate independent variables. The study distinctly makes a significant contribution to aid-growth evaluation studies, as it became the first to disaggregate the foreign inflows into aid, foreign investment and other foreign inflows as opposed to previous studies that has amalgamated all these components into one variable of foreign capital flows. Hence, the findings of previous studies on the actual impact of foreign aid on growth and investment may have been overestimated. Further, this study uses a larger sample size of countries compared to previous studies, which is required of cross country data in order to avoid the problems of inconsistent estimates. Just as with Papanek’s findings, the studies by Levy (1988), Karras (2006), Feeny (2007), Clemens et al. (2004), and Loxley and Sackey (2008) using either or both cross- section and panel data all show a significantly positive impact of foreign aid on economic growth with studies by Levy (1988) and Loxley and Sackey (2008) using Africa as their sample.

These studies however, though have used quantitative methods that have the advantage of establishing the statistical impact of foreign aid on economic growth, yet have got their flaws. The study by Papanek (1973) for instance, made critical innovation to estimating the relationship, but however used limited number of variables as determinants of growth, claiming the focus is on the effect between foreign resources and growth as well as the 60 relationship between savings and foreign resources (ibid: 121). It can be argued that, this may bias the results with the coefficients becoming significant even though they may not have. Further, the regional results in their analysis should only be considered with caution (even as he himself admits) as the regional samples are so small that results obtained from such may be biased and unusable. Hence, the advantage of increase in sample size in this study is only applicable in the full sample analysis and not in the regional analysis. Also, there is no assurance in the form of diagnostic tests that the model used is free from the problem of misspecification. Yet still, as is typical of earlier studies, issues of endogeneity of foreign aid and other explanatory variables are not addressed. The study by Clemens et al. (2004) is as well limited in the sense that it focuses largely on short-run economic growth, when in contemporary times the importance of long-run growth (as is emphasised in the endogenous growth theories) is increasingly acknowledged. Karras’ (2006) study though was able to identify the advantage of the time effect of panel data, yet was also flawed with methodological problems related to limited variable use in his model (see discussion of this shortfall in the subsequent sub-section on effect of methodological differences). Thus it can be seen that most of these studies are associated with one or more limitations that imply a cautious use of their findings for policy and planning purposes.

Cross-Country growth regressions that show an insignificant impact of foreign aid on economic growth Studies that have found an insignificant relationship between aid and growth have also been notable. For instance, studies by (Mosley, 1980; Mosley et al., 1987; Rajan and Subramanian, 2008; Saunders, 2008) could not establish a positively significant impact of foreign aid on economic growth, with even some suggestions of an inverse relationship. With methodological improvements in the estimations, Mosley’s (1980) study found that as contrary to Papanek’s (1973) findings of a positive and significant relationship, the relation between aid and growth is not significant for his sample of developing countries. His decomposition of developing countries into sub-groups rather yielded different results. Foreign aid to the poorest countries yielded a positive relationship. The relationship is also found to be positive and significant for UK aided countries (especially Anglophone Africa) as opposed to France and Scandinavian aided countries where the relationship is negative though insignificant (ibid). Finally, Mosley (1980) found that the difference in the impact of foreign aid across countries arises from differences in the behaviour of the public sector in the aid recipient countries, and in particular with regards their attitude towards tax effort. 61

The analytical framework of the study by Mosley et al. (1987) incorporated indirect effect of aid on growth through two channels: through government expenditure patterns (fungibility), and through changes in relative prices (with the resultant effect on the private sector). They therefore explained the insignificant relationship between foreign aid and growth by suggesting that aid must have been spent on non-productive government expenditure which implies aid is fungible; and also that the resultant negative price effects impact on the private sector. It is no surprise then that their findings and recommendations influenced donor emphasis on conditionality of aid based on recipient countries meeting the criteria of significant estimated rates of return on investments, the estimated impact of aid on the private sector, and on the proportion of aid allocated to non-productive (recurrent) expenditures.

Likewise, the study by Rajan and Subramanian (2008) does not only provide evidence to dispute the non-existence of a positively significant relationship between foreign aid and economic growth, but is special in that it is one of the key few cross-country studies in the aid effectiveness literature that is done on a comprehensive basis. This study has been comprehensive in the sense that it captures the impact of aid on growth accounting for example: methodological differences, aid variable disaggregation, and conditional factors to aid effectiveness. Their overall findings reveal little evidence of a robust positive correlation between aid and growth, and even after correcting for the bias of conventional estimation procedures (OLS) against finding a positive impact of aid. A further strength of the Rajan and Subramanian (2008) study is seen in its novel instrumentation strategy to correcting for the problem of endogeneity of the aid variable where they construct instruments for aid with the main identification assumption that non-economically motivated aid is unlikely to be driven by economic outcomes. While this instrumentation strategy is useful for cross-country studies, it is inapplicable to country studies, which is typical of our research. The variables on colonial relationship and measures of influence (used in the construction of the instrumental variable) are only comparable across countries but cannot be useful when studying a single country whose colonial relationship or influence may not change with time.

Therefore, even studies that empirically dismiss the proposal of aid being significant in promoting economic growth across countries, some of which are short in their analysis and their methodologies inapplicable in the context of a country study like that proposed by this thesis. 62

Variation in results due to methodological differences The methodology adopted for examining the aid-growth relationship (in terms of data analysis, sample, and accounting for endogeneity) seems to influence the cross-country results. Studies (e.g. Levy, 1988; Karras, 2006) that compare the effect across both cross- section and panel data, prove the advantage of time effect in the panel data estimation in terms of robustness of the results. Karras for instance argues for panel regressions on the basis of his findings that: ‘The results show not only that the use of time-series data substantially clarifies the issue, enabling us to arrive at sharper estimates of the growth effects of foreign aid; but also that ignoring the time dimension of the series and relying on cross-sectional data can mask the true relationship and leave the researcher with weak and misleading insignificant results’ (ibid: 16). Some studies (e.g. Mosley, 1980; Dalgaard and Hansen, 2001; Easterly et al., 2001) establish that results do vary with the use of different samples or model specifications; though others (e.g. Karras, 2006) do not. For instance, many studies that tend to restrict the cross-country sample to Africa or Sub-Saharan Africa tend to doubt the evidence of a positive effect of aid in the continent or region. Even studies that have done quantitative meta-analysis of the cross-country regressions on the effect of aid to Africa have found aid to be insignificant in spurring growth in the continent (Doucouliagos and Paldam, 2008; Mekasha and Tarp, 2011). These meta studies both show that aid-growth results are associated with regional differences, particularly the inclusion of Asian economies in the cross-country regression sample, as it tends to lead to a more positive and significant impact. When Africa is singled out, the impact of aid emerge to be insignificant. This does not however imply that other studies have not found a positive and significant effect of aid in the continent as some have (e.g. Levy, 1988; Gomanee et al., 2005; Loxley and Sackey, 2008).

A comparison of these two sets of quantitative meta-studies by Doucouliagos and Paldam (2008) and Mekasha and Tarp (2011) apparently brings to light the impact of methodological difference on the aid-growth results. Even though they may agree on the insignificance of the results for Africa, these studies largely disagree on many of the aid–growth results for the entire sample of developing countries and in other forms of assessing the relationship due to the use of different estimation methodologies. Dalgaard and Hansen (2001) as well as Easterly et al. (2001), both disprove the results of the widely cited Burnside and Dollar 63

(2000) study on aid, policies and growth to be non-robust largely on the basis of sample differences as well as changes to the specifications. The implication is that the cross-country results are debatable and largely dependent on sample and methodological differences.

The literature also show that the aid variable measure employed (aid/capita or aid/GDP, and ODA or EDA) to estimate the aid effect also matters (Gomanee et al., 2005; Karras, 2006) and the use of these aid measures in different estimations tend to produce different results. However, the difference in the results in the use of ODA or EDA is not supported by World Bank (1998), who rather suggests the results do not significantly differ.

In the literature also, there are several studies that do not account for endogeneity of the aid variable, and generally, results from estimations of such studies do seem to wrongly estimate the aid-growth relationship and hence produce inconsistent estimates. Boumahdi and Thomas (2008) suggest that the presence of endogenous regressors in a model implies the consistency of parameter estimates becomes questionable since the fundamental assumption of exogeneity of the regressors no longer holds. However, whilst the sophistication of the methodology used may guarantee more realistic results, it does not guarantee the results to be necessarily positive. For instance, Mosley’s (1980) study, being the first to correct for endogeneity of the explanatory variables obtained results that show aid as not having a positively contributory impact on economic growth. However, accounting for endogeneity of the explanatory variables, particularly the aid variable, points to more realistic results. Thus, the more sophisticated is the methodology used, the more realistic are the findings; and hence this work will account for these methodological shortfalls in most of the studies.

Conclusion From this review of the cross-country literature, it is inferred that a considerable number of studies particularly in the recent literature support Papanek’s (1973) findings of a positive and statistically significant relationship between foreign aid and economic growth. Even in the very poorest of the continents, Africa, where the impact of aid has seemed to be most debated, there are studies that still find a positive and statistically significant relationship, even though economic growth rates and poverty appear pertinent of the continent. But as Gomanee et al. (2005) argue, probably the problem of low growth rates in Africa may be associated with problems not related to foreign aid. However, that does not imply the aid effectiveness debate is closed, as there are still studies reporting a negative or insignificantly 64 positive relationship between foreign aid and economic growth. Hence, in the aid-growth cross-country literature, in as much as there is increasing evidence of a positive and significant relationship between aid and growth, there are still studies that could not establish such results, thus leaving the evidence still mixed.

The cloudy nature of the evidence is further made possible by the increasing use of different methodologies to estimating the relationship, when in fact methodological differences have been shown to affect the results. Hence, the debate remains. The review has shown that methodological differences and aid disaggregation (see section 3.4) influence the findings of the cross-country regressions. Studies contrasting the results from cross-section and panel estimations have mostly favoured panel data as much more advantageous in incorporating the time effect and hence showing more robust results as compared to cross-section regressions. With the effect of the use different aid variable measures, the results seem different. Whilst some have generally found that the use of different aid variables does not alter the significance of the estimates, yet other studies have proven the effect of such changes on the aid estimates. The quantitative meta-analysis of Doucouliagos and Paldam (2008) however supports the assertion that changes in the aid measure do change the estimates of aid impacts. In terms of the disaggregated aid variables, results have mostly found different aid components to have varying effects on economic growth.

Hence, the aid literature on the impact of foreign aid on economic growth (as has been pursued by the cross-country regressions) still remains contentious as is largely confirmed by studies conducting a meta-analysis of the literature findings, particularly that by Doucouliagos and Paldam (2008), that has mostly pointed at significant portion of the aid- growth estimates in the literature proving an insignificant relationship between aid and economic growth. Further, because of the methodological limitation of cross-country studies particularly in terms of their masking of the true picture of country situations, such studies remain fragile for determining the effectiveness of aid and hence for the application of policy recommendation derived from them. This therefore makes relevant the need for country studies that account for this cross-country limitations, and similarly use regression analysis that recognises the time effect and as well being able to establish the impact of a relationship particularly at macro level.

3.1.2.2 The Country Literature: Empirical Evidence

The country literature on the impact of foreign aid on economic growth has basically involved regression analysis, specifically employing time series econometrics techniques. This section of the literature on the aid-growth relationship is one of most relevance to this study as it intends primarily doing a country study to analyse the impact of foreign aid on economic growth using Sierra Leone as a case study. Generally, country studies are seen as having the advantage of portraying the real picture of the relationship as opposed to the masking effect of cross-section studies. Country studies on aid-growth regressions have been by far less researched compared to the literature available with cross-country regressions. This limited research at the country level partly have to do with problems of availability of substantially long time-series data of the required variables that could obtain adequate degrees of freedom to produce reliable and robust results. However, a few country studies have been conducted initially with somewhat crude econometric methods, but eventually with standard and much reliable methods, following vast improvement in econometric modelling and analysis.

The findings of the impact of foreign aid on economic growth as analysed using time series regressions in the country studies have been mixed and inconclusive just as is evident in the cross-country empirical literature. Some researchers (Murthey et al., 1994; Lloyd et al., 2001; Gounder, 2001; Mavrotas, 2002; Bhattarai 2009) have provided evidence to support the supplemental theories that aid promotes economic growth in aid recipient countries; while others (Islam, 1992; Mbaku, 1993; Feeny 2005; M’amanja and Morrissey 2005) have not. The literature also shows that while researchers seem to investigate the relationship between foreign aid and economic growth, yet they do argue for some accompanying growth factors with which they compare aid’s contribution. The earlier country studies in particular sought to not only estimate the impact of aid on growth, but also compare foreign aid to domestic resources in terms of which matters more for economic growth in their distinguished country case studies. Studies by Islam (1992), Mbaku (1993) and Murthy et al. (1994) were typical examples. Murthy et al. (1994) found using time-series regression for Cameroon that both foreign aid and domestic resources (proxied by savings-GDP ratio) were significant determinants of economic growth. On the other hand, Islam (1992) and Mbaku (1993) using time-series regression for Bangladesh and Cameroon respectively could not establish the existence of a positively significant impact of foreign aid on economic growth, and with both

66 establishing that domestic savings was a more important determinant of economic growth (as this was found to positively and significantly impact on growth) than foreign capital.

Some of the later studies that emerged utilised much standard estimations of the aid-growth relationships with the evolvement of new standard econometric techniques and somewhat longer period time-series. In particular, the studies by Lloyd et al. (2001), Gounder (2001), Mavrotas (2002), Feeny (2005), M’amanja and Morrissey (2005), Bhattarai (2009) and Javid and Qayyum (2011) have been in consonance with these improvements. Some of these studies (Lloyd et al., 2001; Gounder, 2001; Mavrotas, 2002; Bhattarai, 2009) provide evidence that support foreign aid to have a significant direct contribution to economic growth and some (Feeny, 2005; M’amanja and Morrissey, 2005; and Javid and Qayyum, 2011) do not.

Some of these studies (e.g. by M’amanja and Morrissey 2005) have compared the impact of aid on growth with the impact of the forms of investment on economic growth. This study for Kenya found that while aid did not have a significantly positive impact on the country’s economic growth; investment on the other hand showed a significantly positive impact. In particular, the study proved private investment (as compared to foreign aid) to be important for economic growth in the country as this was typical in both the short–run and long-run relationships. Therefore, the study argues that private investment as opposed to foreign aid is more important for economic growth in Kenya with the implication that donor and government attention be diverted to private sector investment to ensure faster growth for the country.

Some of the studies (Islam 1992; Gounder, 2001; Mavrotas 2002; Feeny, 2005) additionally argue that aid composition/disaggregation matters in establishing the impact of aid on economic growth. However, detail of this review is done in a separate section of the literature on aid disaggregation (section 3.4).

Analyses of the country literature show that studies are depicted with several limitations/weaknesses to allow holistic acceptability of their findings for policy implementation and other related uses. Weaknesses have ranged from the use of short-time series, crude methodologies, and weak proxies. The use of short time series was more prominent with the early literature; with the effect of obtaining lesser degrees of freedom and 67 consequently less robust results. The studies by Islam (1992) (with 17 years series), Mbaku (1993) (with 20 years), and Murthy et al. (1994) (20years series) are especially evident with quite short time series data. Much longer time series estimation (like that which this thesis undertakes) will provide better reliable estimates. This set of early studies showed several violations of the assumptions of the OLS estimator used, such as collinearity, misspecification and normality problems making the estimated coefficients biased and inefficient to be used for policy interpretation and adoption. The methodological weaknesses of these early studies were so evident that Bring (1994) was prompted to stage a strong critique particularly for Mbaku’s (1993) study but for also that of Islam (1992). In his applied economic letter, ‘how not to find a relationship between foreign aid and economic growth’, Bring (1994) argued that methodologies of these studies were below standard and violated the assumptions of OLS to the extent that findings emerging from them should not be relied upon.

The eventual studies that used fairly longer time series and standard econometric time series techniques are as well plagued with weaknesses that pose a challenge to their acceptability. For instance, most of the existing country studies have been overwhelmed with problems of using inappropriate proxies thus leading to the wrong results and interpretations. The study by Lloyd et al. (2001) on Ghana was an example that used growth of private consumption to represent economic growth; which is simply not adequate a proxy for economic growth. Government consumption in low-income countries like Ghana is likely to be much substantial than private consumption; even if government consumption was not as big enough as private consumption, yet it remains quite substantial to be ignored and hence use growth in private consumption solely as representing growth in the economy. Growth in GDP (which constitutes both private consumption and government consumption) is the most reliable and closest of a proxy for economic growth. Hence their study is useful in establishing the impact of aid on growth in private consumption, and rather may not contribute to the literature on the impact of foreign aid on economic growth. Thus, issues of comparisons of their results with others remain questionable. Likewise, the study by M’amanja and Morrissey (2005) used ‘loans’ as proxy for foreign aid; which is hardly acceptable. Loans are just one component of foreign aid; grants as well constitute a significant portion of foreign aid received by a country. In fact, in lower income countries, donors tend to provide more of grants than loans in their aid disbursement, cognisant of the debt burden that may be associated with lower income countries should they receive more of aid loans. Hence, even though there has only 68 been limited number of country studies to have investigated the impact of aid on growth using time series approach, yet most of these studies further have several limitations that render their findings acceptable only with caution, and hence further research in yet wider case studies like ours on the impact of foreign aid on economic growth and poverty reduction therefore remains in place.

Concluding from the review of the country evidence, it is realised that earlier research was inundated with primitive econometric techniques in the estimation of aid-growth relationship at the country level. Recent research however, has employed standard time series techniques making results emerging from their use somewhat more reliable; though with weak proxies for a number of these studies. Coincidentally, recent time series research on the impact of foreign aid on economic growth at the country level seems to show varying conclusions (some establishing a positively significant impact while others do not). This confirms the continuity in the debate on aid effectiveness in terms of its impact on growth, as even the recent of studies employing standard econometric techniques could not unanimously establish that aid has a positively significant impact on economic growth. Heterogeneity of circumstances under which foreign aid may be effective at promoting economic growth across countries has largely explained the differences in the aid-growth results of the individual country studies. Further, some of the country studies conducted so far have been found with several weaknesses that caution the acceptability of their findings. Weaknesses related to short time series, primitive methodologies especially with early studies, and weak proxies. The need for further research at the country level of the aid-growth relationship for which this study provides evidence for another country case study is hence justified.

On overall, the review of the aid-growth evidence for both cross-country and country regressions shows that the evidence is not only inconclusive from both the cross-country and country regressions, but also confronted with several limitations, and means a cautious use of their findings for policy and planning purposes. Therefore, the need for further investigations with these limitations in mind cannot be overemphasised.

3.2: The Empirical Evidence on the Relationship between Aid and poverty Reduction

The literature on aid effectiveness has vastly focused on the impact of foreign aid on economic growth as is previously reviewed by this thesis. But aid effectiveness itself is not only about the impact of aid on economic growth; it is equally about the impact of aid on poverty reduction. In fact, arguably, considering the focus on world poverty reduction as is prioritised by international development organisations, donors and recipients alike, the poverty criterion of aid effectiveness is of most importance. Because one of the key objectives of this thesis is to evaluate the impact of foreign aid on poverty reduction in our country case study, reviewing this section of the literature is without doubt appropriate; as such a review will not only reveal the evidence so far, but also helps identify areas where this thesis will contribute in this attractive area of the literature and perhaps most importantly serve as a basis for comparing our research findings with the evidence to date. Two strands of the literature on the impact of foreign aid on poverty reduction exist so far and which are mostly in tune with the thesis framework for assessing the relationship: the literature on aid having an impact on pro-poor growth as a form of growth more relevant for the poor; and the literature on aid improving human development (or social welfare) indicators as correlates of income poverty reduction. The evidence on these strands of the literature is therefore reviewed in the following.

3.2.1: Foreign Aid and Pro-poor Growth

Two forms of aid’s impact on pro-poor growth exist in the literature: the literature arguing that foreign aid impacting on economic growth and under good policies impacts on poverty reduction as well, a position typical of World Bank and other researchers such as Gomanee and Morrissey (2002) and Verschoor and Kalwij (2006); and the literature considering growth in the areas and sectors where the poor are most found as pro-poor growth (Feeny, 2007, Feeny and Ouattara, 2009; Akpokodje and Omojimite, 2008; and Olakojo and Folawewo, 2011).

The set of researchers that have provided some investigation of aid impacting on the poor via overall economic growth (The World Bank researchers; Gomanee and Morrissey, 2002; Verschoor and Kalwij, 2006) do so by estimating the income elasticity of poverty. Their key

70 argument is that overall economic growth benefits the poor and so increases in economic growth emanating from aid flows reduces poverty. The work by Dollar and Kraay (2000) is one which argues in this vein. The study shows, following their estimations, that the average incomes of the poorest fifth of society grow proportionately with overall average incomes, as the elasticity of income of the poor (located in the bottom fifth quantile) with respect to average incomes appears around 1; implying that increase in average incomes of the poor is responsive to the increase in overall average incomes, and this the study showed to be robust across specifications and estimators. Similar to this method of investigating the pro-poor nature of economic growth stimulated by foreign aid, researchers such as Gomanee and Morrissey (2002) and Verschoor and Kalwij (2006) have estimated the effect of foreign aid on pro-poor growth by estimating through regressions the income growth elasticity of poverty. Their findings support that of Dollar and Kraay (2000) that foreign aid significantly reduces poverty reduction via economic growth.

While many development researchers and development economists may have mulled over this supposed discovery in development research, the article by Lubker et al. (2002) is one that provides a strong critique particularly of the study by Dollar and Kraay (2000). Whilst they acknowledge robustness of the finding that growth benefits the poor, they however could not confirm robustness of the other finding by Dollar and Kraay that growth arising from IFI policy packages is good for the poor. Lubker et al. (2002) argue that there is a flaw in the statistical modelling used to sustain this finding, evidence of poor quality of data used to test the model, and that Dollar and Kraay’s interpretation of the statistical inference drawn from the model testing was non-standard. Hence, they argue that such a finding is not robust.

Beyond methodological limitation of this study, it could be seen that the finding that overall average income increase is proportional to average income increase of the poor is distribution neutral; meaning that there is virtually no need to advocate the redistribution of income in favour of the poor as they equally benefit from growth in overall average income, which is simply unrealistic. A further critique of the Dollar and Kraay finding could be further premised on their use of averages in justifying their finding. For that which is true on average is not necessarily true for all components of the average (Lubker et al., 2002: 556). This could be interpreted as meaning that increases in income of some members of the poor may not be responsive at least at unit elasticity to increases in overall economic growth. Further, with increasing evidence of widening income inequality levels between the rich and the poor, 71 even if growth is distribution neutral as is claimed by Dollar and Kraay (2000), that will adversely affect the poor.

Unlike the technique of the World Bank (through Dollar and Kraay, 2000) and the aforementioned researchers in investigating the pro-poor impact of foreign aid, another set of researchers (e.g. Feeny and Ouattara, 2007; Fenny 2007; Akpokodje and Omojimite, 2008; Olakojo and Folawewo, 2011) have sought to examine this effect by disaggregating economic output (GDP) into its various components and estimating the impact of aid on these different components; but focusing on the area or sector where the poor are most associated with. For aid to impact on pro-poor growth, it must have a significant impact on the rural or agricultural sector growth where the poor are believed to mostly live. Hence, these studies generally identify agriculture as the sector where the poor are found with evidence of aid’s impact on agricultural growth being mixed. The studies by Feeny and Ouattara (2009); and Akpokodje and Omojimite (2008) find aid to improve growth in the agricultural sector as agricultural GDP significantly improves with increases in aid flows. Those by Feeny (2007) and Olakojo and Folawewo (2011) find that aid does not improve growth in the agricultural sector.

Feeny and Ouattara’s (2009) study for instance, evaluates the impact of foreign aid into the various disaggregates of GDP growth following the product method and expenditure methods of national accounting using a panel of cross country data. Their results reveal that the impact of aid on per capita agricultural GDP is positive and statistically significant, though the impact is not quite strong. Likewise, the study by Akpokodje and Omojimite (2008) though not intended to investigate the impact of foreign aid on poverty, but carried out a study on the impact of foreign aid on agricultural growth in Nigeria. The results show similar to the findings of Feeny and Ouattara (2009) that agricultural growth is significantly influenced by foreign aid. The results further reveal that while there was a positive relationship between foreign aid and agricultural growth, that between aid and per capita income was rather negative.

The study by Feeny (2007) on the other hand provides evidence that foreign aid has had no impact either positive or negative on the rural sector of Melanesia (a group of South-Pacific countries) over the study period, thus suggesting that foreign aid had not impacted on agricultural growth rate and hence on pro-poor growth in this region. In a similar vein, 72

Olakojo and Folawewo (2011) used an error correction model for the period 1977-2008 to show that for the case of Nigeria, agricultural growth (and by implication pro-poor growth) does not respond to aid flows. Thus for the same country Nigeria, the results for the impact of foreign aid on agricultural growth and hence pro-poor growth emerge to be different. The study by Akpokodje and Omojimite (2008) finds a positive impact of aid on agricultural growth where as Olakojo and Folawewo (2011) could not confirm such a relationship for the same country. Further, there seems to be opposing findings between the paper by Feeny (2007) and that by Akpokodje and Omojimite (2008) in terms of aid’s impact on general economic growth and agricultural growth. Whereas Feeny finds aid to significantly impact on general economic growth but not so on agricultural growth; Akpokodje and Omojimite on the other hand find aid to positively impact on agricultural growth but not so on general per capital income. The difference in the findings between these two studies may be associated with the differences in country circumstances; and hence the need not to generalise findings cannot be overemphasized. Further, the fact that results even differ for the same country of study only further confirms the inconclusiveness of the evidence; and thus warrants the need for further investigation.

3.2.2: Foreign Aid and Welfare of the Poor

A small but growing volume of the literature on aid effectiveness has looked at the impact of foreign aid on the welfare of the poor, which according to Morrissey (2007) is highly positively correlated with income poverty reduction. These studies mainly point at aid financing public social expenditures like education, health, and water and sanitation which are welfare sectors from which the poor are most likely to benefit. Thus, in examining this aid effect on welfare, some researchers have compared the effect either via an indirect medium or directly on the poverty or human development indicators. Indirect effects have been either via pro-poor government expenditure or via an equilibrium effect in which aid’s impact on one indicator influences the impact on the other.

The findings on the direct effect of aid on human development outcomes (as welfare correlates) are mixed, with some researchers (Gomanee et al., 2005a; and Fielding et al., 2006) finding aid to directly impact on human development indicators and others (Mosley et al, 2004; Williamson, 2008; and Asiama and Quartey, 2009)) not confirming that effect. On

73 the direct impact of aid on welfare, Gomanee et al. (2005a) found aid to directly impact on welfare though could not confirm that through the indirect means of pro-poor public expenditure. Researchers, such as Fielding et al (2006), also show that aid has a positive and statistically significant effect on human development indicators. On child mortality, aid is found to have a negative and statistically significant direct impact, implying that increases in aid is associated with reduction in child mortality. However, the study by Asiama and Quartey (2009) for sub-Saharan Africa could not establish the direct impact of aid on human development neither could the studies by Mosley et al. (2004) and Williamson (2008) confirm that. Likewise, the studies by McGillivray and Noorbakhsh (2007) and Kosac (2002), though not concerned with the direct effect of aid on human development as central in their studies, yet their estimation of this direct relationship could not show strong evidence in support of a significantly positive relationship between aid and human development in their respective cross-country samples.

Others, who argue that aid’s impact on human welfare is indirect, mostly examine this indirect effect through pro-poor government expenditures. These expenditures constitute spending on social welfare services which are mostly viewed to impact more on the poor. For this, most of the literature (e.g. Gomanee and Morrissey, 2002; Mosley et al., 2004; Wolf, 2007) find that aid significantly improves the welfare of the poor through pro-poor government expenditure. Contrary to this evidence, the studies by Gomanee et al. (2005a) and Dreher et al. (2008) find no evidence that aid improves human development/welfare outcomes through pro-poor government expenditure.

Further, the literature on the impact of foreign aid on welfare has argued that the effect is better found and much representative of the poor when the poor are disintegrated on quantile basis. Typical studies by Gomanee et al. (2005b) and Fielding et al. (2006) have investigated the impact on human development via this quantile regression technique; with Gomanee et al. (2005b) specifically employing the technique of quantile regressions by Koenker and Bassett (1978;1982). Quantile regressions generally estimate the impact of a set of covariates upon specific quantiles of the response variable, thus yielding more comprehensive relationship between variables (Koenker, 2005). In this case, quantile regressions involve estimating the impact of foreign aid (and its control variables) upon a welfare distribution expressed in the form of quantiles. The advantage usually is that whereas in least squares regression, foreign aid as an explanatory variables only captures the impact on the mean welfare indicator, in 74 quantile regressions, foreign aid further captures the impact on the quantile distribution of the welfare indicator, thus allowing comparison of how some quantiles of the welfare indicator may be affected by foreign aid. Thus, these studies argue that the effects of aid on improving welfare will depend on whether the effect is being observed at the lowest or highest level of welfare. The results of these studies generally point to the finding that aid has a positive effect on human development indicators and varies less across the quantiles. A merit of these studies is the fact that though they use cross-section analysis, yet they distinguish the effect of aid across the various welfare distributions (i.e. five quantiles), in this way capturing the true effect of aid on the poor who are usually located in the lower quantiles. It is however noted that though Fielding’s study considers various areas of human development, this still does not coincide with the requirement of the HDI as per capita income is left out. Further, though these studies are quite relevant for particularly locating the impact of aid on welfare of the group of the poor, yet the evidence show that disaggregation into quantiles does not matter considerably as the studies show aid has a positively significant impact on human development indicators irrespective of the quantile.

Generally, the studies on the impact of foreign aid on human welfare though have made some important inroads on evaluating aid effectiveness in terms of the poverty criterion, yet have their limitations just as those that had investigated the impact of aid on pro-poor growth. First, it is seen that their use of cross-country regressions means that inferences for specific countries cannot be drawn from their evidence. Secondly, there are wide variations in the use of the poverty indicators; whilst some may be close proxies (such as HDI), the majority are merely partial indicators such as the individual indicators in different social sectors. This leads to difficulty in explaining the poverty variables.

Thirdly, most of these studies (e.g. Gomanee and Morrissey, 2002; Gomanee et al., 2005a, 2005b) did not adequately account for the problem of endogeneity which is a common feature of the aid variable. For instance, Gomanee and Morrissey (2002) corrected for endogeneity by only including one period lagged aid levels instead of the standard instrumental variable technique. However, there is the difficulty in constructing a viable instrumental variable for poverty reduction particularly with the problems of non-availability of poverty data and hence might be a reason for the observed rare use of instrumental variables in solving for the problem of endogeneity as is seen in the literature.

Fourth, most of the studies used very few control variables in estimating the relationship with the consequence of omitted variable bias. There are many variables that should affect poverty which were virtually not included in the regression specifications. However, a reason for this could be the fact that there is no well specified theory on the determinant of poverty reduction as there are for the determinants of economic growth. Another limitation is found in studies that used pro-poor public expenditure (PPE) as a medium through which aid affects poverty, and yet did not account for the content of aid in PPE. The only studies to have accounted for this are those by Gomanee et al. (2005a; 2005b) where they eliminated the aid content of PPE through the method of generated regressors and in so doing were able to capture the true effect of PPE impacting on human welfare.

What is however very clear about the findings emanating from these studies on the impact of foreign aid on the welfare of the poor is that they remain inconclusive. With the aforementioned limitations together with the inconclusiveness of the findings on the impact of foreign aid on welfare, the need for further research remains in place and hence our study using a cross-country analysis for Sierra Leone is as well worth it.

Overall, the findings on aid effectiveness in terms of the poverty criterion are just as inconclusive as those from the economic growth criterion of aid effectiveness; hence the need to conduct further research to contribute to evidence on this grey area.

3.3: Recipient policy and political institutions and aid effectiveness

Whilst it is important to establish the impact of aid on development outcomes, the factors to its (in)effectiveness also warrant investigating because findings from which would be essential for policy advice in making foreign aid much effective. Hence, the research intends to further review the role of domestic political systems (among the many recipient characteristics) in making aid effective in terms of improving human well-being. Research on the importance of recipient characteristics on aid effectiveness has largely focused on recipient macroeconomic policies and the quality of their institutions as important factors for aid effectiveness; and this has its origin from the World Bank (1998) Assessing Aid report, with its important background paper by Burnside and Dollar(1997; 2000).

Burnside and Dollar (2000) find that the contributory effect of aid in fostering economic growth in recipient countries is contingent on sound macroeconomic policy condition in such countries. The Burnside and Dollar study however suffered from several criticisms. Some critiques have dammed the study as lacking robustness (e.g. Hansen and Tarp, 2000a; Dalgaard and Hansen, 2001; Brumm, 2003; and Easterly et al., 2004). Hansen and Tarp (2000a) for instance show that the result turns out to be sensitive to data and model specification. A small extension (about 2%) of the sample negates the results. Likewise, Dalgaard and Hansen’s (2001) re-estimation of the Burnside-Dollar regression found the results being sensitive to the choice of instruments for solving the problem of endogeneity of the aid variable. Also applying the Burnside-Dollar methodology, Easterly et al (2004) conclude that the Burnside-Dollar results are not robust as they are sensitive to changes in the sample. Other critics contend that policy is not the only important conditioning factor to aid effectiveness (Guillaumont and Chauvet, 2001; Easterly et al., 2004; Dalgaard et al., 2004; Loxley and Sackey, 2008). Guillaumont and Chauvet (2001) for instance, dismiss that assumption of solitary conditioning criterion and argue that aid effectiveness also depends on exogenous (mostly external) environmental factors (terms of trade trend and real value of exports instability, climatic shocks etc).

In their following paper, and recognising the critique of policy not being the only factor to aid effectiveness, Burnside and Dollar (2004) investigated the possibility of institutional quality being a condition for aid effectiveness; and of which they confirm to be a crucial factor just as they found for macroeconomic policy. Other studies by Dollar and Levin (2005) and Baliamoune-Lutz and Mavrotas (2008) also provide evidence that institutional quality is an important factor for aid effectiveness. Studies by Sacks et al (2004) and Chandar and Caprio (2007) on the other hand could not find evidence to support the importance of institutional quality for aid effectiveness.

In all of these recipient characteristics, the role of politics on the effectiveness of aid appears to have been less researched. This section of the literature thus reviews the relationship between recipient politics and the effectiveness of foreign aid in terms of its impact on economic growth and poverty reduction. The importance of reviewing this section of the literature lies in the fact that firstly, the review identifies the limitations of research in this area of aid effectiveness and secondly, it further provides a basis for comparing the existing

77 evidence with the study’s findings on the role of politics for aid effectiveness in the case of Sierra Leone.

By the beginning of the new millennium, the international community had come to accept and broadly agree that not only does governance matter in aid delivery and effectiveness, but that the technocratic processes inherent in governance were political. Thus, the Monterey consensus of 2002 recognised the importance of politics as an integral part in the development process as well as its relevance in governance. Hence, de Haan and Everest- Phillips (2007: 2) narrate that in the Monterey consensus, ‘both domestic and international politics were integral to the governance problem and also to its solution’. The article concludes on the relevance of the Monterey Consensus in the recognition of the importance of politics in development that “Monterrey created a ‘Faustian’ bargain between rich and poor. The developed world would give more money in return for political elites in poor countries delivering the better politics and governance essential for delivering the MDGs” (de Haan and Everest-Phillips, 2007: 3).

Despite the importance of domestic politics in recent aid modalities, research as to how important it is for aid effectiveness has been quite limited. Few studies (Boone, 1996; Kosac, 2002; Arvin and Barillas, 2002; McGillivray and Noorbakhsh, 2007; and Angeles and Neanidis, 2009) do establish quantitatively the importance of the form of political rule for aid effectiveness. Boone (1996) examined the relationship between the effectiveness of foreign aid and political regimes of recipient countries. The article hypothesizes that politicians will use aid resources for transfers and not to reduce internal distortions as the aid goal requires. In order to capture the impact of aid on development outcomes under certain conditions; and in this case under the conditions of political liberties, the article employed an interaction term within a panel data analysis to examine whether certain political regimes use aid more effectively than others. The study’s main finding is that aid is effective when it is conditional on policy and/or political reforms, and can also be effective where aid is non-fungible. The paper concludes that short-term aid (instead of long-term aid) targeted at supporting new liberal regimes and encouraging greater political and social liberties may be more effective in promoting sustainable development and reducing poverty.

Likewise, on a study of the impact of aid on the quality of life (proxied by human development index), Kosac (2003) finds, using cross-country panel analysis that aid is not 78 effective in all cases, but finds it to highly significantly improve the quality of life when a recipient country is under democracy. Aid to autocracies on the other hand is found to have an insignificant or even harmful effect on the quality of life. The article uses two alternative measures of democracy (POLITY IV and Freedom house measures) and in both situations, aid is found to be effective under democracy. Therefore, the article argues that the political environment matters in making aid effective in significantly improving growth in the quality of life. However, the study has only been done and generalised at the cross-country level.

On the other hand, and using income per capita as proxy for poverty, the study by Arvin and Barillas (2002) also used cross country analysis on an annual basis to investigate the causal relationship between aid and poverty, but rather showed, for the full sample of countries studied, that under the condition of democracy, aid does not have a significant impact on poverty neither is the level of aid affected by the level of poverty. The results however differ when samples are split into sub-groups/regional groups. Under the condition of democracy, aid is found to reduce poverty in East Asia and Pacific, but has a detrimental impact on low- income countries. In a similar vein, McGillivray and Noorbakhsh (2007) though had the main purpose of their study on investigating the link between aid, conflict and human development, yet they made some attempt to examine the link between aid, democracy and human development as a measure of human welfare and therefore the quality of life. In this investigation (using cross-section analysis for 94 developing countries), they could not find strong support that good democracies improve the impact of aid in promoting human development in aid recipient countries.

The study by Angeles and Neanidis (2009) examines the impact of aid on growth contingent upon the power and attitudes of local elites, who the study describes to be characterised by broad political and economic power and with minimal concern for the rest of the population, arguably because they belong to a different ethnic cohort. The study proxies ‘local elites’ by the proportion of European settlements in the recipient country during colonial times. Using this measure, the study finds the presence and behaviour of local elites to be an important factor for the impact of foreign aid on economic growth. Specifically, they showed that the extensive power and behaviour of local elites has an adverse effect on the effectiveness of aid on economic growth. Whilst methodological applications may have been strong as do the robustness of the findings, yet, this study has notable limitations that caution the acceptability of the findings. Firstly, they employ a panel dataset of countries involving both developed 79 and developing countries. The implication is that some of the developed countries as well as the strong performing developing countries in the sample may have been receiving less or no aid at all for considerable periods in the time-series. Hence, combining such countries in a single sample may introduce some bias in the results. Secondly, their use of European settlements in colonial times seems a remote proxy for local elites in aid recipient countries, especially in regions like West Africa (for which some of its countries form part of the study sample), where evidence of European settlement is quite rare for the past four decades or so. Local elites in most low-income recipient countries are likely to comprise indigenous politicians, the highly educated and owners of large private firms. The use of a more representative and closer proxy for capturing local elites (which requires further research) in low-income recipient countries (who are the larger recipients of development aid) could better capture the role of local elites on the effectiveness of aid. Less than that, an explorative study should provide some insight in the analysis of this relationship.

Hence, the evidence on the role of recipient political systems on the effectiveness of foreign aid has not only been limited in terms of the number of existing studies as well as being restricted to only cross-country context, but also that the evidence is mixed, which warrants further research in this important recipient characteristics; thus the relevance of our study.

Other studies which have provided some discursive analysis on the importance of domestic politics have only been few and even those studies have mostly analysed the importance of politics for development process and outcomes and hardly so on its importance for aid effectiveness. Such explorative discursive studies have included those by Oloka-Onyango and Barya (1997), de Haan and Everest-Phillips (2007), Cammack (2007), and Magassa and Meyer (2008). The argument of most of these papers points to the fact that the nature of politics and political will matter for economic development of the recipient countries and not on whether politics matter for aid effectiveness. De Haan and Everest-Phillips (2007) for instance argue following experiences in most successful African Countries (where political leaders do not stick to power beyond their elected terms and at the expense of development) that ‘State legitimacy matters for economic growth’ (ibid: 9). Cammack (2007) argues that the nature of politics in Africa is such that patrimonial politics driven by both supply side and demand side factors appear to be more influential in the choice of political leaders than does policies and development. She reports that in Africa, tribal and regional lines are common factors determining the voting direction and campaign strategies. It is common then that 80 region, tribe and ethnicity, and religion can play the much greater role in elections decision compared to policies and development which may even be ignored by both the candidates and voters. Thus, Cammack’s (2007) article provides an explanation (from a neopatrimonial analysis) as to why there has not been any ‘political will’ from recipient governments as has been often accused of them by donors. However, whilst Cammack’s argument may be relevant for understanding how neopatrimonial politics influence governance and the development process, the role of such politics in reducing or enhancing aid effectiveness remains unexplained.

Other commentators (such as Oloka-Onyango and Barya, 1997; and Magassa and Meyer, 2008) argue that for politics to positively influence development and/or donor interventions, there must be a strengthening of the role of political actors particularly the increasingly vibrant civil society, which is an actor group whose importance in political development and aid effectiveness has been subject to increasing recognition from donor, political economists and development practitioners. These commentators argue that once these civil society groups are capacitated and well recognised within the development process, then accountability of the state and donors in terms of the delivery and use of development resources will be enhanced. While this discourse on the implications of the nature of domestic politics and institutions for the development process is well highlighted by these researchers and commentators, yet how such neopatrimonial politics and institutions affect the effectiveness of aid in significantly contributing to economic growth and poverty reduction is not purposeful explained or researched; which makes it an area for further research.

In conclusion, whilst the literature on the factors to aid effectiveness has largely focused on the role of recipient policies and institutional quality to making aid effective, that on the role of politics has been comparatively limited. Besides the works by Boone (1996), Arvin and Barillas (2002), Kosac (2003) and McGillivray and Noorbakhsh, 2007) that provide some level of analysis on this area, there have been virtually no detail research on the role of recipient country politics on the impact of aid on development outcomes. Even the works by the aforementioned researchers have only been conducted on a cross-country analysis, whose findings have limited use for realistic application of policy at the country level. Hence, it seems obvious that the need to capture the effectiveness of aid under the political regime remains an area for further research particularly in the country context. 81

3.4: Review of the Literature on Aid Disaggregation

A detailed literature review on the disaggregation of foreign aid and its impact on economic growth and poverty reduction in aid recipient country is virtually non-existent; which merely reflects the focus of research on the impact of foreign aid in its aggregate form. But as there have been increasing calls to account for aggregation bias in the assessment of aid impact (Cassen, 1994; Mavrotas, 2002, 2007; Mavrotas and Ouattara, 2006a, 2006b, 2007; Mavrotas and Nunnenkamp, 2007; Feeny, 2007), the importance of reviewing this evidence becomes more important. In this thesis, the need to review the literature on aid disaggregation is additionally relevant because greater portions of this research constitute an assessment of the impact of aid in its disaggregated forms on both economic growth and poverty reduction.

Generally, research on aid heterogeneity seems to focus on two forms of aid disaggregation: disaggregation based on aid targets (and categories of aid considered to be pro-poor) and disaggregation based on mode and source of aid delivery. Research on disaggregated aid targets has overtly looked at the effect of aid disbursed to recipient economic and social sectors such as aid to the agricultural sector, aid to education, aid to health sector, and aid to infrastructural development (including water and sanitation). The review of the evidence on this form of aid disaggregation is only briefly covered in this thesis (partly because some of which is already reviewed in section 3.2); but the works by Fielding et al (2006), Wolf (2007), Asiedu and Nandwa (2007), Dreher et al. (2008), and Williamson (2008) provide some evidence in this respect.

Generally, the evidence from these studies is not entirely in agreement and hence seems to vary. Aid to education has been largely found to have a positive impact (Fielding et al, 2006; Wolf, 2007; Dreher et al., 2008). In fact, Wolf (2007) shows that aid allocated to both health and education sectors improves the welfare indicators in those sectors, but could not confirm such results in the case of aid allocated to water and sanitation. The study by Asiedu and Nandwa (2007) disaggregates aid into pro-poor areas and examines the impact of these aid areas on economic growth. In particular, the study examines the impact of aid allocated to the education sector on economic growth. The study disaggregates aid to education into primary, secondary and tertiary education aid and employed the dynamic panel system GMM estimator to find that the effect of aid varies across the different levels of education, and for low-income and middle income countries. In particular, they find that aid to primary schools

82 has a positively significant impact on economic growth in low income countries, but could not find similar evidence for post-primary education aid. Aid to the higher education (tertiary) schools is rather found to be significant in contributing to economic growth in middle income counties, but aid to primary and secondary schools rather have adverse effects on growth in such countries. Hence, the study concludes that the effect of aid on growth depends on the level of development of the recipient country (low income and middle income) and the level of education at which aid is being targeted (primary, secondary and tertiary). The study by Dreher et al. (2008), in a similar sectoral aid assessment, finds aid allocated to the education sector to promote primary school education; and this result is consistent across model specifications, estimation methods, and even after instrumenting to control for endogeneity of aid.

The study by Williamson (2008) though did not compare between total aid and sectorally allocated aid, yet it exclusively used aid to the health sector to investigate the impact on human welfare (health indicators); and found after accounting for endogeneity that aid to the health sector does not significantly improve overall health in aid recipient countries. Hence, it does not seem to support the positive finding by Wolf (2007) that aid to the health sector contributes to human welfare. Thus, these studies attempting to find the effect of disaggregated sectoral aid show results that are not entirely consistent, but largely show that when aid is disaggregated into sectors, the effectiveness of aid is stronger than when the analysis is limited to only total aid.

The other form of aid disaggregation which emphasises the mode and source of aid delivery has looked at aid in the form of grant or loans, project aid or programme aid/budget support, technical versus non-technical assistance, food aid, and bilateral and multilateral assistance. Because this research’s investigation of aid heterogeneity focuses largely on the mode and source of aid delivery, much emphasis is put on this section of the review. Research both theoretical and empirical on these various forms of aid delivery is also seen to be dichotomised into two forms: one that seems to examine the fiscal response effect of the various modes of aid delivery (in the context of aid heterogeneity); and the other form which examines the impact of heterogeneous modes of aid delivery on the ultimate development outcomes of economic growth and to a very little extent on poverty reduction.

Studies on the fiscal response effect of heterogeneous aid modalities generally argue that the aid types differ in their effect on government fiscal behaviour. In particular, the effect on public expenditures and investment, tax revenue, and government borrowing differ between the various aid modalities and hence the use of a single figure of aid to examine the fiscal response effect of foreign aid leads to aggregation bias. There has been increasing research interest in this research area. Studies by Mavrotas (2005), Mavrotas and Ouattara (2006a, 2006b, 2007), and Cassimon and Campenhout (2007) have all provided some evidence in this respect. From a fiscal response utility model, the study by Mavrotas (2005) estimated the impact of disaggregated aid (project aid, programme aid, food aid and technical assistance) on government fiscal behaviour in Uganda using time-series data and the non-linear three staged least square estimation approach and finds that different types of aid have different effect on key fiscal variables. The government of Uganda responds accordingly if the types of aid come from technical assistance, project aid, programme aid or food aid. Arguing from a similar fiscal response framework, Mavrotas and Ouattara (2006a and 2006b) show that different aid modalities have different effects on government fiscal behaviour. Mavrotas and Ouattara (2006a) used theoretical modelling to argue that project aid, programme aid, technical assistance and food aid have different effects and that if the government has differing interest in each of these categories of aid, failure to disaggregate aid in a fiscal response analysis will introduce aggregation bias in the results.

Likewise, in their second paper, Mavrotas and Ouattara (2006b) empirically tested the fiscal response framework in the context of heterogeneous aid types using time series data for Cote’ D’Ivoire and found in their analysis that using a single figure of aid revealed that foreign aid is fully consumed; but disaggregating this aid variable into project, programme, technical assistance and food aid, results showed that the government responds differently to these different categories of aid. Employing the non-linear three staged least square estimator, they find that the direct and total accumulated impact of these categories of aid on public investment and consumption do vary; although the direct impact on tax revenue emerges to be the same - a negative impact. Specifically, the results show that project aid and programme aid behave differently with regards their impact on investment and consumption. Whereas programme aid positively impacts on public investment but negatively on public consumption, project aid on the other hand negatively impacts on public investment but positively affects public consumption. Technical assistance and food aid have similar impacts- they both positively relate with public investment, but negatively with government 84 consumption. This therefore emphasises the need for disaggregating aid to more precisely determine the effect of each category on government budget or fiscal behaviour.

In a further empirical test of the fiscal response model in the context of aid disaggregation, Mavrotas and Ouattara (2007) used a panel of 106 aid recipient countries to compare the budgetary effect of project aid and financial programme aid, and find that both project and financial programme aid have a positive and statistically significant impact on public expenditure, but that whilst project aid is associated with increases in capital expenditures, financial programme aid is rather associated in increased government consumption. In terms of the relative impact on revenue, their study finds neither project aid nor financial programme aid to be associated with reduction in tax effort. In another fiscal response model (though on a different category of aid), Cassimon and Campenhout (2007) examine the effect of debt relief as a form of aid modality on fiscal variables involving current expenditure, government investment, taxation and domestic borrowing. Their study, using a panel of 28 HIPCs, finds that the effect of debt relief on public investment changes with time. In particular, it initially reduces public investment, but the effect becomes positive after two years. They also find that an increase in debt relief does not reduce domestic revenue collection, and relatively performs better than grants and loans particularly in the longer-run where increases in debt relief is associated with increases in domestic revenue collection.

Studies on the effect of heterogeneous aid delivery modes on economic development generally argue likewise that not all aid types have the same direct effect on economic growth and poverty reduction. Whilst the various modes of aid delivery (grants, loans, project aid, programme aid/budget support, food aid, and technical assistance) and the sources of aid (bilateral and multilateral) are generally examined, the arguments on aid effectiveness in the context of aid heterogeneity have largely focused on grants versus loans, project versus programme aid (or budget support) and bilateral versus multilateral assistance.

In terms of aid grants and aid loans, most of the studies (Gounder, 2001; Feeny, 2007; Loxley and Sackey, 2008) show that grants are more effective in promoting economic growth in recipient countries compared to aid loans. The study by Gounder (2001) used the ARDL model to show that in Fiji, aid in the forms of grants, instead of loans, have a significantly positive impact in promoting economic growth in the country. Feeny (2007) also finds using a sample of Melanesian countries that it is aid grants rather than aid loans that have a 85 significantly positive impact in promoting economic growth in the region. Loxley and Sackey (2008) find in a panel of 40 African countries that the impact of pure aid grants on growth is about twice the impact of total net official development assistance; thus suggesting the disbursement of more grants to the continent in order to achieve much of the MDGs. The study however did not compare between loans and grants, but only assesses the importance of aid grants to achieving faster economic growth in the continent. Likewise, though did not compare between aid grants and aid loans, M’Amanja and Morrissey (2005) showed for the case of Kenya that aid loans do not significantly improve the country’s economic growth. Even though aid in the form of grants is mostly found to positively promote economic growth, some do find aid loans to significantly contribute to economic growth and even outperform aid grants. For instance, Islam’s (1992) study for Bangladesh showed that only foreign aid in the form of loans (as opposed to grants) appears to have some significant and positive contribution to economic growth in the country. The study by Feeny (2005) on Papua New Guinea, however, could not find either of grants or loans to have a significantly positive impact in promoting economic growth; a result that mimics his finding that total aid does not promote economic growth in the country.

In terms of the argument between project aid and budget support (or programme aid), empirical evidence has largely favoured project aid as being more directly effective on economic growth. For instance, Feeny’s (2005) study on Papua New Guinea on aid disaggregated into project and budget support revealed that only project aid shows some moderate impact (at the 10% significance level). Likewise, the study by Ouattara and Strobl (2008) which employed a panel data series using OLS, 2SLS and GMM estimators show that project aid positively impacts on economic growth but with diminishing returns, while financial programme aid tends to have a negative impact. Some studies however show that neither project nor programme aid have a significantly positive impact on growth. For instance, Mavrotas (2002) estimated the impact of aid disaggregation on economic growth in India using time series data and the Johansen ML approach, and finds that both project and programme aid have a negative impact in influencing economic growth in the country. Programme aid is only positively significant when lagged at least 2 periods. Similarly, Islam (1992) could not find evidence to show that project aid has a significant and positive contribution to economic growth in Bangladesh.

In terms of the argument on the effectiveness of technical assistance, the evidence is mixed. The study by Gounder (2001) on Fiji shows aid in form of technical assistance to have a positive and significant impact on economic growth. However, the statistical significance is only moderate. Likewise, Feeny’s (2007) study on a group of Melanesian countries showed that technical assistance has a positive and significant impact in promoting economic growth in the region. However, the study by Ouattara and Strobl (2008) could not confirm the findings by Gounder (2001) and Feeny (2007). Ouattara and Strobl (2008) could not find evidence to show that technical assistance significantly fosters economic growth in the sample of countries examined.

The evidence on the impact of food aid on economic growth has likewise been mixed. Whilst Islam’s (1992) study found such aid type to be positive and significantly related to economic growth in Bangladesh, Ouattara and Strobl’s (2008) panel analysis could not find food aid to significantly improve economic growth.

The evidence on the debate between bilateral and multilateral aid sources has equally been mixed. The study by Gounder (2001) further showed that bilateral assistance has a positive and significant impact on economic growth while multilateral aid does not. In Javid and Qayyum’s (2011) comparison of the effectiveness of bilateral and multilateral aid sources on economic growth in Pakistan, they showed that both aid types perform better under the condition of good macroeconomic policies. However, even though the direct effect on economic growth may not have been quite strong, yet bilateral aid tends to perform better than multilateral aid at least in the short-run. The study however tends to emphasise its discussion on the negative but insignificant long-run relationship between bilateral aid and economic growth, even though multilateral aid despite being positive could not be significant in determining economic growth in Pakistan in both the long-run and short-run. Heady (2007) however, found using OLS and GMM estimators that the effect of multilateral aid on economic growth is much stronger than that for bilateral aid. Specifically, the study finds that multilateral aid has a significantly large effect in fostering economic growth, whereas the positive and significant effect of bilateral aid can only be evident in the post-cold war era. In the cold war era, bilateral aid has no significant effect in fostering economic growth; and hence the study concludes that bilateral aid has political motive which makes it ineffective on economic development of the recipient countries.

Some of these studies on the choice between aid modalities, particularly the theoretical ones have also argued that the relative effect of different aid modalities on growth and/or poverty reduction depends on the conditions under which such aid types are delivered. In particular, it depends on the recipient country characteristics and the degree of alignment of donor interest with that of the recipient government. Specifically, studies by Cordella and Dell’Ariccia (2007), Cordella and Ulku (2004) and Hefeker (2006) have provided some theoretical backing in this respect. Cordella and Dell’Ariccia (2007) carried out a theoretical appraisal of budget support and project aid types to determine which type is more effective than the other. They showed that when the priorities of donor are aligned with those of the recipient, budget support can be preferable to project aid; which is also true if the assistance is small relative to the recipient’s own resources. Project assistance is however preferable in the reverse situation. The study by Hefeker (2006) used a theoretical model to argue that the choice between project aid and budget support in terms of their effectiveness in promoting economic growth and poverty reduction depends on the interest of the government and financial institutions (as donors). If the interest of the government is enriching the elites (i.e. politically self-interested) rather than targeting poverty reduction, the study argues that more donor project assistance directed towards the poor (in the case that donors’ interest is poverty reduction) will lead to government policy readjustment in the context that government will see the increased donor project assistance to the poor as an opportunity for it to allocate more of its discretionary payments to the rich. Hefeker argues therefore that only a closer alignment of the preferences of government with those of the donors (in this case the international financial institutions) can avoid the possibility of more money being channelled to the rich and hence target developmental needs. Thus, Hefeker (2006) argues that conditional project aid or budget support cannot be alternatives to government ownership of policies. Hence, official project aid can as well crowd out funds that government could have allocated to the poor, and in essence cannot only lead to an increase in the income of the poor, but to that of the rich as well if government’s preference is political self-interest.

The study by Cordella and Ulku (2004) not only conducted a theoretical analysis but further empirically tested their model by employing panel data for a sample of 70 developing countries using OLS and system GMM estimators. Their key argument is that the choice between grants or concessionary loans in terms of their impact on economic growth depends on the prevailing conditions of the recipient country. Under the aid recipient conditions of varying quality of the policy environment, per capital income levels, and the degree of 88 indebtedness, they argue that the choice between grants or aid loans is contingent on the position of the recipient country with regards these conditions. Using an analytical framework originating from Krugman (1988), Cordella and Ulku (2004) theoretically showed that the degree of loan concessionality is positively related with economic growth if the recipient country is poor, has bad policies and highly indebted. Hence, aid grant (with 100% concessionality) is most preferable to countries characterised by low levels of income, weak macroeconomic policies and highly indebted. Countries with stronger policies, higher levels of per capita income and less indebted can attract higher loans and hence higher volumes of total aid as such could be more effective under such conditions. Their empirical analysis to test the model showed consistent results, even after employing a number of robustness checks. In their conclusion, they argue that it is not an issue of choosing ‘all grants’ or ‘all loans’ as being the most desirable outcome, but that the optimal mix between these aid categories depends on the characteristics of the aid recipient.

However, whilst these theoretical models have been able to identify the necessary conditions under which aid categories are more effective on the recipient country’s economic development, they may only serve as stepping stone for further research on their findings. Unless such analytical frameworks are empirically tested (with the exception of that by Cordella and Ulku, 2004) their findings may still be preliminary. Further, the assumptions underlying most of these theoretical models seem too simplistic to the extent they appear unrealistic. For instance, the analytical framework by Cordella and Dell’Ariccia assumes aid to be delivered to only one country and also that the amount of aid provided by the donor is fixed. The assumption of aid being fixed is also typical of the assumption of the analytical frameworks by Hefeker (2006) and Cordella and Ulku (2004). Further, whilst these studies have been relevant for identifying the types of aid that are more effective under certain conditions, the focus of research had largely been on economic growth, which may only be one amongst the developmental targets of foreign aid.

From this review of the literature on aid heterogeneity, the following points are noted:

• The Disaggregation of aid in its assessment is vital since different categories of aid have different impacts, whether on fiscal policies or on ultimate development goals particularly that of economic growth. Hence, aid modalities and aid targets have different effects.

• The choice between aid categories in terms of their effectiveness on economic development depends on recipient characteristics (especially for grants and loans) and/or on degree of alignment of donor interest with that of the recipient government (especially for project and budget support, but which is also true for the budgetary effect of project and programme aid). • The empirical evidence on the impact of heterogeneous aid types is mixed. Whilst some studies find one type of aid to be effective, some other study will find the same type of aid ineffective. • Much of the predictions from theoretical models have yet to be tested empirically especially for the debate on the relative effectiveness of project aid versus budget support (or programme aid) on economic development. • Generally, research on aid heterogeneity has not been as elaborate as that of total aid effect, but is increasingly attracting research attention in recent times; although much of this recent attention seems to be drawn towards the fiscal effects of heterogeneous aid categories, rather than on the ultimate development goals which are of course the ultimate target of aid. • Much of the research on the development impact of aid disaggregates seems to focus more on the effect on economic growth and largely at the cross country level of analysis. Research at the country level is limited just as is research on aid heterogeneity and its effect on poverty reduction. In cognisance of the limitations of cross country evidence (as has been previously argued) and the importance of poverty reduction in the recent donor aid targeting, the need for extended research on poverty reduction and at the country level of research cannot be overemphasised. This is where this study attempts to contribute to the literature on aid disaggregation; by extending the evidence on the effect of aid disaggregation on not only economic growth, but also on pro-poor growth and aggregate human welfare in the context of improving the well-being of the poor.

3.5: Summary of the Evidence and gaps in the Literature

The findings of the literature on the impact of foreign aid have been mixed across the respective development goal of aid – promoting economic growth and poverty reduction. The empirical evidence to support the theory that aid promotes growth has remained mixed.

Though it has been thought that increasing evidence since the late 1990s is showing that aid promotes growth, yet there remain a notable number of studies that show that it is far from being the case thus creating continuity for research into this literature. Whilst generally the aid effectiveness literature has vastly been on the impact of foreign aid on economic growth, aid effectiveness itself further involves assessing the impact of aid on poverty reduction, which remains an important (or even more important) goal of development aid as compared to the impact on economic growth.

The literature on the poverty criterion of aid effectiveness is relatively small (compared to that for the economic growth criterion), but is increasingly attracting the attention of development/economics researchers in recent times particularly with the shift in emphasis of aid target towards reducing world poverty. No matter the measure of poverty used, what seems established in these studies is that the evidence remains as mixed in the poverty criterion just as in the economic growth criterion of aid effectiveness. These aid-poverty studies have also got their limitations which warrant further investigation into this area of research like what this study does. Such limitations have generally included: limitation of the studies to only cross-country context which broadly assume homogeneity of countries; wide variation of the use of poverty indicators with many not being complete or close proxies; some studies not adequately accounting for the problem of endogeneity of the aid variable which is usually a common occurrence; and limited use of control variables in the poverty regressions. These limitations together with the inconclusiveness of the evidence make relevant the conduct of this research.

In the third part of the review, the evidence on the importance of politics for aid effectiveness is as well found to be less researched despite its increasing recognition in recent developmental agendas. In the final part of the review, the findings on aid heterogeneity are mixed, and research focused largely on the impact on government fiscal behaviour and economic growth, with the effect on poverty being ignored. Following this survey of the evidence, the gaps in the literature are highlighted: • Research on the importance of domestic politics/elite behaviour for aid effectiveness remains a gap. That which largely exist in the literature are studies on the importance of domestic politics on the development process

• A common weakness in the literature has been methodological limitations in a good number of the aid effectiveness studies conducted, with the robustness of their findings being most times questioned. This therefore calls for more robust findings particularly in the country aid effectiveness literature, where the use of a single technique of estimation has been the norm. The employment of methodological triangulation to strengthen the robustness of the country aid effectiveness findings becomes relevant. • Studies on the effectiveness of foreign aid on poverty and specifically on welfare of the poor has largely been restricted to cross-country context without extending to individual country analysis. • Research on disaggregation of foreign aid, though is attracting research interest in recent times, yet seems to largely focus on only the economic growth criterion of aid effectiveness. The need to disaggregate aid in order to identify the categories that are more effective for reducing poverty in recipient countries arguably deserves even more attention. • There has also been a call from the OECD (Bigsten et al., 2006) and other researchers (e.g. Murthey et al., 1994; Solow, 2001; Mavrotas, 2002; and Feeny, 2005; Temple, 2010) for a focus on the conduct of more country studies as opposed to cross-country studies because of the masking effect typical of cross-country aid-growth regressions.

Recognising these gaps in the aid effectiveness literature, together with the need to conduct a comprehensive aid effectiveness study for Sierra Leone, a country where one is yet to be done, makes this research a useful one to contribute to knowledge; and hence a conceptual framework for analysing aid effectiveness in terms of the impact of donor interventions on the development goals of economic growth and poverty reduction and the crucial condition of the quality of political regime for achieving these goals in the case of Sierra Leone, is presented in the succeeding chapter.

CHAPTER 4: CONCEPTUAL FRAMEWORK AND OVERALL RESEARCH STRATEGY 4.1 Conceptual Framework 4.1.1: Introduction

This section describes the conceptual framework as is shown in figure 4.1 below, upon which this study is based. The term, foreign aid as used in this research (and as is precisely defined in the introductory chapter) refers to official development assistance which is foreign capital in the forms of grants and/or concessional loans from bilateral and multilateral sources. The framework follows from the theoretical suppositions and research arguments on the impact of foreign aid on economic growth and poverty reduction as ultimate goals of development aid, as well as the quality of politics as a key condition upon which aid effectiveness is presumed to depend upon.

4.1.2 An Overview of the Conceptual Framework

A conceptual framework is defined as ‘‘a visual or written product, one that explains, either graphically or in narrative, the main things to be studied -the key factors, concepts, or variables – and the presumed relationships among them” (Miles and Huberman, 1994: 18 ). In defining the functions of the conceptual framework for doctoral research in educational leadership, Johnson and Farmer (2007) similarly find the function of their framework as clarifying the structure of the concept to be studied, defining its purpose, as well as delineating the areas of investigation of the thesis. In yet a similar vein, Maxwell (2005a) finds the function of any conceptual framework as aiding the researcher evaluate and refine the research goals, formulate research questions, choose appropriate methodology and identify likely validity to his/her conclusion. In acknowledging these functions, this study builds on these in the construction of a conceptual framework of aid effectiveness as shown in figure 4.1.

Fig. 4.1: Conceptual Framework

Donor Interventions (Aid flows and Policies )

Aid Effectiveness Domestic Politics

Welfare levels Reducing Contributing to Poverty Levels Economic Growth

Pro -poor growth

Ultimate goals of development aid

Figure 4.1: A Conceptual Framework showing the Schematic relation of Donor Intervention, Economic Growth and Poverty Reduction. Source: Author’s construct.

Donor interventions have been in the form of aid flows as well as policy advice to the recipient country, with the latter likely being a conditionality for access to financial aid. These aid flows and intervention policies are intended to enhance development in the recipient country. Therefore, aid effectiveness is meant to be aid having a positively significant impact on economic growth and poverty reduction as ultimate goals of development aid. During the initial periods of research on aid effectiveness, this concept had meant, aid having a significant effect in supplementing investment (and/or savings as implied). This was particularly so following models by Harrod-Domar and Chenery and Strout (1966). Eventually, aid effectiveness was premised on aid having a significant effect in fostering economic growth (an impact which can be direct, or indirect through the channel of investment). This latter consideration of aid effectiveness, as meaning aid having an impact on economic growth has dominated the aid effectiveness literature, until the late 1990s when aid effectiveness has incorporated poverty reduction in addition to economic growth.

In fact, aid’s impact on poverty reduction is now largely considered as the primary target of development aid, particularly with poverty reduction being prioritised in global development targets (such as the MDGs) and national development plans (such as the PRSPs). In providing a justification for analysing aid effectiveness in terms of the quality of life rather than just through economic growth, Kosac (2003) argues that the economic growth variable (growth in per capita income) says nothing definite about standard of living, as it says nothing about the distribution of this income. Kosac based his argument on Sen’s (1999: 14) proposition that: “An adequate conception of development must go much beyond the accumulation of wealth and the growth of gross national product and other income-related variables. Without ignoring the importance of economic growth, we must look well beyond it”. With this, Kosac (2003) argues that development of a country should be measured by the actual quality of life of its citizens rather than the income each citizen should receive if the total productive income were to be divided equally among its population, which is never realistic. Hence, the article argues that per capita measure is a measure of ‘potential rather than actual services and income available to the population (Kosac, 2003: 2). Thus, aid effectiveness as conceptualised in this study framework means aid having an impact on economic growth and poverty reduction.

Even though economic growth and poverty reduction are both considered to be the ultimate goals of development aid, development economists and aid donors in general, have in more recent times argued that economic growth is not an end in itself as the goal of development aid, but rather consider poverty reduction as the ultimate target of development aid. Therefore, economic growth should be the means to influence poverty reduction. This implies that foreign aid is expected to impact on poverty reduction through economic growth. Yet others have argued that economic growth is just one channel for influencing poverty levels, as aid can have direct effect on welfare indicators which are found to be strongly correlated with poverty (Gomanee and Morrissey, 2002; Gomanee et al., 2005a, 2005b; Morrissey, 2007). Hence, with the use of welfare indicators as proxy for poverty reduction, foreign aid can directly reduce poverty levels, not just through the channel of economic growth. Yet still, others emphasise pro-poor growth as a means of poverty reduction. Implying that only growth that is pro-poor can reduce poverty levels, as general economic growth may or may not impact on poverty reduction (as is shown by a dotted arrow in the framework). Pro-poor growth is meant as growth with strong poverty elasticity or even so as growth in those areas 95 where the poor are mostly found. However, others have insisted that economic growth in general can influence poverty reduction. Therefore, this research considers economic growth and poverty reduction as two separate research objectives in the study. Poverty reduction is hence considered at two parts, pro-poor growth (which is proxied by growth in agricultural GDP) and improvement in welfare levels of the poor/public (proxied by improvements in HDI and infant mortality rate).

Whilst aid effectiveness involves aid having an impact on economic growth and poverty reduction, this study further considers as crucial in the evaluation of aid impact, the assessment of key obstacles to aid effectiveness. Put in another way, the critical factors upon which aid effectiveness is contingent upon as is recently explored in the literature. This particularly follows from the World Bank’s (1998) ‘assessing aid’ report with its key background article by Burnside and Dollar (1997; 2000), for which favourable recipient country policy is seen as crucial to aid being effective. Whilst there may have been several recipient characteristics explored in the literature to be vital for aid effectiveness, the most commonly researched ones have been recipient policy and institutional quality/good governance. But as these factors have been extensively researched, this study rather focuses on the quality of politics in our conceptual framework in an attempt to investigate the likelihood of this factor being detrimental/important to aid effectiveness at the country level of focus. This study’s interest in investigating the importance politics for aid effectiveness arises not only because it is relatively less researched, but also because politics has been increasingly recognised as an important factor for the development process in developing countries as is emphasised in more recent development strategies (De Haan and Everest- Phillips, 2007).

4.1.3 The Research Questions and Hypotheses

Following the above overview of the conceptual framework, the thesis research questions and the accompanying hypotheses to be proven are highlighted in this section. There has been a persistence of poverty and dismal economic performance amidst long and notable presence of donors in Sierra Leone. This thesis therefore asks the question: ‘Has donor aid been effective in Sierra Leone? To address this overall research question, the following specific questions are asked:

4.1.3.1 Does foreign aid significantly foster economic growth in Sierra Leone?

Answering this specific research question is crucial in assessing the impact of foreign aid in Sierra Leone. The theoretical relationship between foreign aid and economic growth has been reviewed in the literature (chapter 3), and hence this conceptual framework of how foreign aid is related to economic growth recognises that. The study’s empirical model (to be derived in the next chapter) for assessing the impact of foreign aid on economic growth is derived from a capital-based model, which recognises capital to be crucial for the growth process. Key growth models such as the Harrod-Domar, neoclassical and the endogenous growth models all emphasise capital as an important determinant of economic growth. The ‘supplemental arguments’ for the impact of foreign aid on growth are very much in tune with the demands of these models. These arguments (particularly by Chenery and Strout, 1966) suggest that, generally, as developing countries are faced with the problem of low average savings to stimulate growth, foreign aid can fill that savings gap as well as a foreign exchange gap to raise output. Thus, foreign aid to developing countries is expected to positively contribute to economic growth by supplementing low savings and relaxing exchange rate constraints; which implies the hypothesis to be tested is that foreign aid to Sierra Leone has had a positively significant impact on the country’s economic growth over the years.

4.1.3.2 Is poverty significantly reduced by donor aid in the case of Sierra Leone?

In assessing the impact of foreign aid on poverty, the conceptual framework is such that two variants of poverty reduction are employed: pro-poor growth and aggregate welfare levels. Thus, for this research question, two hypotheses are to be tested:

1. Foreign aid significantly improves pro-poor growth in Sierra Leone

Specifically, growth in agricultural GDP is used to measure pro-poor growth and hence serving as proxy for poverty reduction. A significant impact of foreign aid on growth in agricultural GDP will correspondingly translate to a significant impact of foreign aid on poverty reduction in the country. This proxy is so chosen because in Sierra Leone, rural poverty stands at around 80% of households as revealed in the country’s poverty reduction strategy paper (GoSL, 2005: 28). The country’s poverty profile as reported in this strategy paper also reveals that the poor are more likely to work in agriculture particularly in food

97 production; which implies that the country’s poor are mostly those engaged in agriculture. Therefore, significant growth in the agricultural sector should translate to improvement in the standard of living of the poor. Put in another way, growth in agricultural output is treated as pro-poor growth.

In addition to the fact that the agricultural sector employs the vast majority of the rural poor, expenditure on income realised by growth in this sector is argued as the significant driver of growth in the overall economy thus benefiting even the non-farm poor. Mellor (2000: 18) posits ‘it is the expenditure of this increased (primary sector) income on locally produced, labour intensive, non-tradable goods and services that drives the employment creation, that in turn, explains the poverty reduction’. Further, Mellor (1999) argues that in spite of the suggestion that overall economic growth reduces absolute poverty, it is in fact the direct and indirect effect of growth in agricultural sector (as a proportion of the overall economic growth) that accounts for practically all of the decrease in poverty. Though may not have investigated the impact of aid on agricultural output, but the importance of increasing agricultural output as a means to reducing poverty is well supported by several other studies in the literature (e.g. Aluwalia, 1978; Ravallion and Datt, 1996; Timer, 1997). In fact, Mellor (1999) further argues that as opposed to overall economic growth, agricultural growth (whether through its direct and indirect effects) does not manifest the unfavourable distributional effect of income; and that it is only growth in agricultural sector that significantly reduces poverty. All of these studies provide strong evidence that the structure of growth matters for poverty reduction; and in particular that increase in agricultural output per head of the rural population significantly decreases poverty. With this recognition of the importance of increasing agricultural output, this thesis therefore posits that the impact of foreign aid on reducing poverty is well seen by an assessment of the extent to which foreign aid promotes agricultural output in the country. This implies growth in agricultural output is used as proxy for poverty reduction.

2. Foreign aid is effective in improving human development in Sierra Leone

The second variant through which aid impacts on poverty as used in this thesis is via its impact on the well-being of the poor. Morrissey (2007) describes this approach to assessing poverty reduction as ‘public goods approach’ to aid; with well-being interpreted as increasing access of the poor to public social services particularly those related to health, education and 98 sanitation, in addition to reducing income poverty (ibid: 5). Typical poverty measurements are those related to income poverty- the head count index, i.e. the proportion of people living on less than $1 a day corrected for PPP or the proportion of the population below the national poverty line. However, proponents of welfare indicators of poverty have always argued against this income poverty measure. First, that despite the fact that poverty measure of headcount index has the advantage of being internationally comparable, yet the reliability of its measure cannot be guaranteed (Reddy and Pogge, 2002). Second, that it only depicts the opportunities available to the person but not the use the person makes of this income. They argue that someone may have less income but may have free access to health, education and other social services which does not make him worse than the person who has more income but lacks these social services; and hence that it is the quality of life the person leads that matters most rather than the income he possesses (Anand and Sen, 1992; Kosac, 2003). Therefore, improving the well-being of the poor through raising human welfare is itself a measure of poverty. In this vein, this thesis further assesses the impact of foreign aid on poverty by assessing the impact on human welfare. The choice of human welfare indicators as measure of poverty is also justified by the scarcity of income poverty data for panel data analysis (a reason, which is also advanced by Gomanee et al., 2005a).

In the assessment of this relationship, this thesis employs aggregate welfare indicators comprising of the Human Development Index (HDI) and Infant Mortality Rate (IMR) as used by Gomanee and Morrissey (2002); Gomanee et al. (2005a; 2005b); and Verschoor and Kalwij (2006). These aggregate welfare indicators (HDI and IMR) are reported by these researchers to be correlated with income poverty indicators. The IMR, which is a non-income measure, captures the material hardship aspect of poverty. Gomanee et al (2005a: 357), for instance, favours the use of the infant mortality rate as a measure of aggregate human welfare on the premise that not only is data availability for this indicator, but that it may be ‘reasonable indicator of the welfare of the poor’. The HDI, which is an index of longevity, education and real per capita income ($PPP) has the advantage that it incorporates income which is an important component of poverty as well as accounting for other welfare components. Kosac (2003) uses growth in HDI as a measure of quality of life; and describes this indicator as best measure of quality of life – ‘more sophisticated and consistent’ – and has the further advantage that it incorporates measure of wealth (per capita income) in addition to the social welfare indicators of education and health (ibid: 2). The strength of the HDI is further supported by the UNDP (2002:153-156), who sees it as having the general 99 merit of being aggregative in its measure of human welfare and also has data available (being calculated on a consistent basis for every five years since 1980) for a large sample of countries. Thus, the choice of these welfare indicators is based on the fact that they are close substitutes for income poverty measures which is lacking in data and does not capture non- income measures, a disadvantage seemed to be readily corrected by HDI and infant mortality rate. Hence, improvement in aggregate welfare tends to improve the well-being of the poor just as much as declines in income poverty (Morrissey, 2007).

In consonance with most of the literature (e.g. Gomanee and Morrissey, 2002; Mosley, 2004; Gomanee et al., 2005a, Verschoor and Kalwij, 2006; Morrissey, 2007), the conceptual framework of how aid should impact on welfare/well-being is argued to occur via three means: a. Foreign aid impacting directly on aggregate welfare indicators. This is so because there are direct donor-managed projects targeting service provision in the welfare services of health, education and sanitation. Such donor-managed aid projects can generate income-earning opportunities or improve welfare directly (Gomanee et al 2005a). b. Foreign aid indirectly impacting on aggregate welfare via economic growth. This implies aid improves these social indicators for a given level of per capita income (as Mosley et al., 2004 and Gomanee et al., 2005a suggest) c. Aid impacting on aggregate welfare through indirect means of financing government expenditures in the social welfare indicators (i.e. Pro-Poor Public Expenditure - PPE). Social sector expenditures such as education, health and sanitation remain the strongest contenders. This implies aid is expected to significantly contribute to pro-poor social expenditure. According to (Wolf, 2007: 1), increased aid inflows and debt relief are expected to augment government expenditure towards the improvement of education, health and infrastructure which is considered the ‘classical task’ of government for which water and sanitation is considered an important part of infrastructure. Whilst spending on public services does not guarantee that the poor are better off, because there can be distributional problems (Castro- Leal et al., 1999; Morrison, 2002); there is yet evidence that increased spending on social services is likely to provide some benefit to the poor generally even if they are the least to benefit (Lanjouw and Ravallion, 1999). It is further argued that these social sectors are the areas of public spending that is most likely to improve welfare, particularly health and education (Gupta et al., 2002). 100

Thus, this study addresses the research question of whether aid reduces poverty levels in Sierra Leone by looking at the responsiveness of agricultural growth as well as human welfare indicators from increases in foreign aid.

4.1.3.3 Is the system and quality of politics detrimental to the effectiveness of foreign aid in promoting human development in the case of Sierra Leone?

Whilst aid effectiveness is assessed on the impact of foreign aid on economic growth and poverty reduction, it is also important to examine the factors that matter for aid (in)effectiveness. It is on the basis of this investigation that policy considerations for improving aid effectiveness can most likely arise. This study investigates the influence of domestic politics (as a recipient characteristic) on the effectiveness of foreign aid in the Sierra Leone context.

The following hypothesis is posited to address this research question:

The System and Quality of Domestic politics matter in making aid effective in Sierra Leone

Following the review of the literature that the nature of domestic politics matters for the development process and outcomes, domestic politics in the context of democratisation has thus been examined whether is it not detrimental to the effectiveness of foreign aid in promoting human welfare as a proxy for poverty reduction. Further, a comparison of the effectiveness of foreign aid on human welfare under the political systems of institutionalised democracy and autocracy has also been explored in the study. The importance of domestic politics for the development process is exemplified in the suggestion by Devarajan and others (2001) that policy changes are driven primarily by domestic political economy; hence justifying the relevance of domestic politics for development policy.

As this domestic politics matter for the development process, this research posits that autocratic form of political leadership, largely displaying patrimonial practices as well as some level of authoritative leadership, leads to inefficient management and use of not only domestic resources, but also foreign aid that is intended for the development process. This argument is supported by Angeles and Neanidis (2009) that the potential for misuse of aid is

101 higher when the elite within a recipient country is characterised by extensive economic and political power with minimal concern for social groups besides itself. Therefore, the study hypothesises that whilst democratic forms of rulership and the tendency towards democratisation may be good for the effectiveness of aid, autocratic polity on the other hand could be detrimental to the impact of aid on human welfare. Whilst a discourse analysis of neopatrimonial politics on aid effectiveness is paramount for confirming this hypothesis, however, for the scope of this thesis, our investigation is limited to quantitatively examining the effectiveness of aid on human welfare under the political systems of democracy and autocracy in order to test the importance of the quality of politics and nature of political rule for aid effectiveness in Sierra Leone.

4.1.3.4: Does aid disaggregation matter in the examination of the impact of aid in Sierra Leone? In addition to examining the developmental impact of total aid, the study further disaggregates foreign aid into its various modes of delivery to determine which categories are better effective in promoting economic growth, pro-poor growth and aggregate human welfare in the case of Sierra Leone. The aim is to show whether disaggregating aid uncovers possible aid types that may not be significant for improving economic growth and reducing poverty in the country, even if total aid is found to be effective. By identifying aid types that perform better than others, donors and government will be able to decide on which modality to prioritise should there be scaling down of aid flows into the country since aid-giving is not indefinite and with proportions of the various disaggregates of aid varying with time and donor-type. Even with the recent indication by the OECD and other development partners for the scaling up of aid for Africa (see Gupta et al., 2006), the types and sources of aid that should be scaled up deserve much importance if aid effectiveness is to be maintained and strengthened in the continent and Sierra Leone in particular.

In addition to the above, there are further reasons for analysing the disaggregated form of total aid with respect to avoiding the aggregation bias argued in the literature (Cassen, 1994; Mavrotas, 2002; Mavrotas and Ouattara 2003, 2006a, 2006b, 2007; Feeny, 2007). Mavrotas and Ouattara (2006a) argue that aid is heterogeneous, with each of its components/disaggregates exerting different macroeconomic impacts on the aid-recipient country; and hence that the use of a single figure of foreign aid as has been typical in the vast majority of the aid effectiveness literature fails to capture this heterogeneity. The implication 102 is that this leads to aggregation bias. Further, in situations where the aid-recipient government attaches different utility to the various categories of aid (as is argued by Mavrotas and Ouattara, 2006a), there is the likelihood of this affecting the impact of each of these categories on its intended purpose and consequently on economic development. Using a single figure of foreign aid in such scenarios will therefore lead to aggregation bias in the results and conclusions arrived at. For the scope of this study, six types of aid modality are considered: Grants, Loans, technical assistance, food aid, multilateral and bilateral aid.

With respect to examining the impacts of these heterogeneous forms of aid, the research makes four propositions:

1. In a low-income country like Sierra Leone, grants rather than loans are more important for improving economic growth and reducing poverty. 2. Technical assistance has not had a significant impact in promoting economic growth and reducing poverty in Sierra Leone. 3. In a low-income country with less significant political importance (as is the case with Sierra Leone), bilateral assistance can have a favourable development impact. 4. In cognisance of its role and form, food aid does have a positive and moderately significant impact in fostering pro-poor growth in the case of Sierra Leone.

The reasoning behind each of these propositions is presented in the relevant sections on aid disaggregation in the empirical chapters that follow. In the next section, we present the overall research strategy for investigating the relationship between donor intervention, economic growth and poverty reduction in the case of Sierra Leone.

4.2. The Overall Research Strategy

This second section of the chapter presents the overall research strategy employed for investigating aid impact in Sierra Leone, including a justification for its choice and an overview of the techniques of enquiry and analysis. Detailed presentation of the methodology for analysing each of the research objectives is done in the relevant empirical chapters that follow. The overall research strategy consists of a mixed methods approach, combining quantitative and qualitative data and methods to address the research questions. A country

103 level analysis is employed and is particularly relevant because there exists structural differences in each country and hence avoids universalisation of aid policy applications to developing countries in general.

Triangulation simply implies ‘combining more than one set of insight in an investigation’ (Downward and Mearman, 2007:80). Its use in social science research is not new as it dates back to the early ideas of Campbell and Fiske (1959) who advocated the combination of a collection of methods/operations to reduce the effect of the peculiar biases of each one method (Blaikie, 1991:116). Likewise, its use in economic research, though mostly limited to policy economics and econometric analysis (Downward and Mearman 2007:80), has attracted attention particularly since the publication by Lawson (1997). Its particular use in social and economic research has been justified for the purpose of providing explanations to answers obtained from quantitative methods. In some cases where quantitative techniques could not be possible due mostly to unavailability of data, it has been found to be valuable. As Onwuegbuzie and Leech (2005:287) argue, as long as the study’s research objective is both exploratory and confirmatory, then mixed methodological data analysis techniques could be used. This study’s use of triangulation recognises these justifications and proceeds with its use in investigating the relationship between donor intervention, economic growth and poverty reduction in Sierra Leone.

This study employs both within-methods and between-methods triangulation approaches in its examination of aid impact in the case of Sierra Leone. The within-methods triangulation approach had involved the use of quantitative techniques involving a combination of two time series regression approaches so that any short comings in any one of these approaches is overcome by obtaining findings from both, thus making conclusions much more reliable. The problems of robustness and non-usability of results in one approach is hence nuanced or validated by the results obtained from the other.

The between-method triangulation is also employed to bring in some qualitative analysis involving interviews with experts in the relevant public sector and civil society bodies to provide explanations to the ‘why’ question of aid (in)effectiveness. Hence, the study’s use of triangulation captures the counterfactual through the quantitative techniques, and further includes qualitative techniques to provide explanations to the findings obtained from the use of the quantitative techniques. 104

On overall, data collection methods and sources comprise information mainly from secondary sources complemented by primary sources. More broadly, secondary sources, which were collected using a desk survey, comprise of donor and institutional online statistical databases; documentary information on donor policies and reports; and reports on aid use and management from the public sector, public audit institutions, media and civil society. The primary data on the other hand was sourced from in-depth interviews with the relevant public sector departments, public audit institutions, resident donor agencies, media elites and civil society bodies making use of their experiences, views and perceptions on the use, management and effectiveness of foreign aid in Sierra Leone.

The type of data and the method of data analysis employed in this study are highlighted in the schematic diagram in figure 4.2. As used in this study, the quantitative data generally comprise time series data on real GDP, poverty indicators, governance indicators, macroeconomic data, public expenditures, and foreign aid inflows which have been collected in order to estimate the impact of foreign aid on development outcomes. Data has as well been collected on some of these variables for a sample of African countries as part of the data required for analysing the welfare impact of foreign aid in Africa and the case of Sierra Leone. Such quantitative data was collected from the online databases of the World Bank, the International Monetary Fund, the OECD, the International Crisis Group, UNESCO, Earth Trends, UN Common Database, the UNDP and POLITY IV.

As can be observed from figure 4.2 and as previously emphasised, the research involved a mixed methods approach in both the methods of data collection and data types, but also on the method of data analysis. The method of analysing the quantitative data generally involved the use of econometric analysis. However, the type of econometric approach employed varies in the respective objectives of the research. For analysing the impact of aid on economic growth as well as the impact on pro-poor growth, a combination of time series cointegration approaches were employed. A Dynamic panel data approach, specifically involving a Generalised Method of Moments (GMM) estimator was used in analysing the impact of aid on aggregate human welfare in Africa and Sierra Leone, as well as in analysing the relationship between aid, politics and welfare. A detail of relevant empirical model, the estimation approach and justification for the choice of each methodology is found in each of the respective empirical chapters. 105

Figure 4.2: The Overall Research Strategy Source: Author’s Construct

The qualitative data which generally includes experiences, views and perceptions of experts and civil society representatives have been obtained from a total of thirty-nine (39) in-depth interviews. A list of the interviews conducted is shown in Appendix 4.1. These interviews were conducted during a five-month long fieldwork to the country of study, Sierra Leone lasting from 28 th March 2010 to 27 th August 2010. Qualitative secondary data involving documentary evidence also form part of this qualitative information and much of this was obtained during the fieldwork. These interviews and reports have been analysed in this study to complement the findings of the qualitative analysis. Quotations from the interviews and reports have been utilised to provide evidence and explanations to the findings of the quantitative analysis. Pseudo names of the interviewees referenced in the work have been used for reasons of confidentiality. Assurance of confidentiality during the interviewing process enabled the researcher obtain detailed and confidential information from interviews, without which could have biased the findings. Appendix 4.2 shows a table that summarises the overall research design.

106

In the empirical chapters that follow, we first present the methodology of investigation for each research question (involving the derivation and specification of the model, variable definition and sources and estimation approach) before presenting and discussing the results for every research question or objective.

107

CHAPTER 5: IMPACT OF FOREIGN AID ON ECONOMIC GROWTH: EMPIRICAL ANALYSIS

5.1 Introduction

This chapter provides a description of the methodology and presentation of the empirical results following the examination of the impact of foreign aid on economic growth in Sierra Leone covering the period 1970-2007. This aid-growth relationship is examined using econometric techniques involving a triangulation of time-series regression approaches. The aim of combining methods is to triangulate findings in order to draw much valid conclusions. Hence, in this chapter, we first present the model, estimation methodology and empirical results on the examination of the relationship between foreign aid and economic growth for both estimation approaches employed. This is followed by a further analysis of the aid- growth relationship by looking at the importance of aid disaggregation on economic growth – focusing on grants, loans, technical assistance and bilateral aid.

5.2 The Aid-Growth Relationship

In this section of the chapter, we present the empirical model, the estimation techniques and the empirical results on the impact of foreign aid on economic growth in Sierra Leone.

5.2.1. The Empirical Model and Data

In the description of the empirical model for the examination of the aid–growth relationship, we first present the main model, which is the base model and from which main inference is drawn upon about the impact of foreign aid on economic growth. We then present further specifications where we change variables at a time in an attempt to check the robustness of our findings on the impact of foreign aid on economic growth.

5.2.1.1 The Aid-Growth Base Model Following from the review of the economic growth theories typical of the Harrod-Domar, neoclassical and endogenous growth theories (in Chapter 3), it is apparent that all postulate capital to be an important determinant of economic growth. Hence, the empirical model specification for estimating the impact of foreign aid on economic growth specifies capital

108 and other key determinants of economic growth as commonly suggested in the growth literature. Therefore, in deriving our empirical model for estimating the aid-growth relationship for Sierra Leone, we posit that:

Y= f(X, Z) …………….(5.1)

Y denotes output (i.e. Real GDP), X is a vector of capital sources, and Z is a vector of other growth determining variables as found in the empirical literature and which are crucial for technological productivity. The endogenous growth model in particular generally emphasises the importance of capital (both physical and human) and policy for promoting economic growth. On this basis, the above theoretical model motivates the general empirical growth model for the time-series growth regression which is specified as follows:

RGDP t = α + βXt + γZt + t …………….. (5.2)

Where RGDP is Real Gross Domestic Product being a proxy for economic output, and X and

Z are as previously defined. t is the error term, while subscript t, denotes time. Critical capital sources for economic growth of developing countries comprise foreign aid and private investment. These capital sources are physical capital sources as data for human capital for the country under study is virtually unavailable.

Hence, X = f(Aid, PI) …………..(5.3)

Where ‘Aid’ denotes foreign aid, and ‘PI; denotes private investment. This assumes that PI which constitutes private domestic capital sources (such as domestic credit to the private sector) and foreign direct investment is a critical source of capital in addition to foreign aid that can augment economic growth. Herzer and Morrissey (2011), in their cross-country study of foreign aid effectiveness presented capital as the single most important factor that influences domestic output. Of this capita stock, they specified foreign aid and private investment in addition to domestic taxes in the production function as the critical components of capital that determine domestic output.

As commonly found in the literature, other growth determinants, Z = f(Policy, IQ1) ……(5.4)

109

Where ‘Policy’ denotes macroeconomic policy index, accounting for fiscal, monetary and trade policies. Fiscal policy is proxied by government final consumption expenditure (Easterly and Rebelo, 1993). Monetary policy is proxied by inflation (Fischer, 1993; Burnside and Dollar, 1997; 2000). Trade policy is proxied by trade openness measure, which is (imports + exports)/GDP (Burnside and Dollar, 1997; Feeny, 2005; Javid and Qayyum, 2011). These policy measures have been widely argued to influence growth of the economy.

The variable, IQ1 denotes property rights score, whose component variables follow Knack and Keefer’s (1995) component variables in their construction of an index of property rights/ institutional quality. The literature on property rights seem to be in agreement that secure property rights positively contribute to economic growth through the promotion of investment (Przeworski and Limongi, 1993; Knack and Keefer, 1995). As economic theory and empirical evidence generally find investment to be a critical determinant of economic growth, the protection of property rights which itself provides insurance for investment, will ultimately contribute to economic growth. In addition, North and Thomas (1973) and North (1990) argue that secure property rights are important for economic growth.

Thus, substituting (5.3) and (5.4) in (5.2), gives our detailed empirical growth model as:

RGDP t = α + β(Aid, PI) t + γ(Policy, IQ1) t + t ……(5.5)

Simplifying, this gives us the empirical model for estimation as:

RGDP t = β0 + βaAid t + βiPI t + βpPolicy t + βiq IQ1 t + t …….(5.6)

To capture economic growth using RGDP, we use log of RGDP (as has been used by Kargbo and Adamu, 2010; Adhikary, 2011; and Herzer and Morrissey, 2011), as log difference of RGDP implies economic growth. Correspondingly, all the regressors are expressed in logarithms with the exception of the policy index which has some negative observations. Thus, the model as used in the empirical analysis is specified as:

LRGDP t = β0 + βaLAID t + βiLPI t + βpPOLICY t + βiq LIQ1 t + t …….(5.7)

110

A positive and statistically significant coefficient of the aid variable is interpreted as aid having an impact in promoting economic growth in the country.

5.2.1.2 Extensions of the Model for Robustness Check The above aid-growth base model is extended with further specifications to capture the robustness of the impact of aid on economic growth across specifications.

5.2.1.2.1 With Addition of Political Crisis Variable In the first re-specification, we add the influence of the political crises (CRISIS) to the base model to see whether the effect of aid on growth significantly changes. This model is specified as:

LRGDP t = β0 + βaLAid t + βiLPI t + βpPolicy t + βiq LIQ1 t + βcCRISIS t + t …….(5.8)

CRISIS is a variable of political instability. The literature on political instability and economic growth recognises that political instability lowers the availability of factors of production and consequently lowers economic growth (Alesina et al., 1996; Alesina and Perotti, 1996). In the study country, Sierra Leone, there has been a lengthy and destructive civil war which makes relevant the inclusion of political crisis variable in the aid-growth model.

5.2.1.2.2 With use of Different measures of political and institutional quality Here, the study attempts to test the robustness of the impact of aid on economic growth by altering the measures of political and institutional quality in the base model. This implies the property rights measure is replaced in the model by quality of governance and quality of domestic politics at a time. The measures of political and institutional quality as used in the literature have been subject to some scepticism. Criticisms have ranged from the fact that they are quite persistent over time as the quality of institution and polity change very minimally over time (Burnside and Dollar 1997; Herzer and Morrissey, 2011); to problems of data availability especially in the earlier periods in the series for which unavailable data is only represented by the earliest available observation as is done in Burnside and Dollar (1997; 2000), a technique we also use for IQ1 and IQU in this study. Thus, the study uses three of these measures: property rights (IQ1) as used in the base model, governance (IQU) and polity quality (POLIT).

111

5.2.1.2.2.1 With the use of Governance measure (IQU) When we use the quality of governance measure, the specification becomes:

LRGDP t = β0 + βaLAid t + βiLPI t + βpPolicy t + βiq LIQU t + t …….(5.9)

Where IQU is governance quality variable which is recognised in the literature (Mauro, 1995; Jones, 1998; Knack, 1999; Burnside and Dollar, 2004; Feeny, 2005; Maxwell, 2005; and Baliamoune-Lutz and Mavrotas, 2008) as an important determinant of economic growth.

5.2.1.2.2.2 With the addition of political Regime Variable With the addition of the political regime indicator, the extended model is specified as:

LRGDP t = β0 + βaLAid t + βiLPI t + βpPolicy t + βpo Polit t + βcCRISIS t + t …….(5.10)

Where ‘Polit’ is the variable representing the quality of the system of political regime. Though the literature on the systems of the political regime (democratic versus authoritarian regime) may be controversial with regards to their relative importance for economic growth, it seems however unanimous in generally finding politics to be an important determinant of economic growth. Przeworski and Limongi (1993) in their critical review of the literature on political regimes and economic growth, indicate in their opinion that politics generally matters for economic growth, though however remain indifferent on the debate of democracy versus autocracy in promoting economic growth since the findings on the latter are inconclusive.

5.2.2 Data Description and Construction

All data series collected for this study are mostly either from the World Bank’s World development indicators (WDI) and African Development Indicators databases, Polity IV or the ICRG database. The series is annual and covers from 1970 to 2007, a period that reflects a good availability of data for Sierra Leone in all the variables of the model.

Economic growth is measured by the annual log difference in real GDP and sourced from the World Bank’s WDI database.

112

Foreign aid as used in this research comprises official development assistance as a share of GDP sourced from the World Bank’s African Development Indicators database.

The data on private investment is sourced from the WDI database and represents private investment as share of GDP.

The variable, IQ1 is property rights score, whose construction follows Knack and Keefer’s (1995) construction of an index of property rights. Hence, we construct a property right score constituting the component of corruption, rule of law, bureaucratic quality, expropriation risk, and government repudiation of contracts taken from IRIS3 database of the ICRG. The components: corruption, rule of law and bureaucratic quality have a rating of 0-6 with higher values indicating better ratings (i.e. less risk). The other two components: expropriation risk and repudiation of contracts are rated 0-10 with higher values indicating better rating (i.e. less risk). To arrive at a comparative rating of these five components that form the property rights measure, we follow Knack and Keefer (1995) by converting corruption, rule of law and bureaucratic quality to 10-point scales (multiplying them by 5/3). Once, all the five variables are of uniform scale of 0-10, a simple average of these five indices is then taken to arrive at the composite score of property rights denoted as IQ1. These components of property rights largely match those used as an index of institutional quality in the literature (e.g. Burnside and Dollar, 2004; Sacks et al., 2004; Dollar and Levin, 2005; Chandar and Caprio, 2007; and Baliamoune-Lutz and Mavrotas, 2008). Hence, in as much as we refer to it as a measure of property rights, for comparisons with other studies, it corresponds to its use as a measure of institutional quality.

The variable, IQU is Governance Quality score provided by the IRIS3 of the ICRG and represents an average of the three component scores of governance quality {i.e. (a) corruption in government; (b) the rule of law; (c) bureaucratic quality}. Its construction follows Knack’s (1999) construction of an index of governance quality. The rating range from a minimum of 0 (representing the highest potential risk) to a maximum score of 6 (representing lowest potential risk). This implies the higher the rating the lower the political risk and hence the better the governance quality.

113

The variable ‘Polit’ is a measure of the quality of political regime and is sourced from the POLITY IV database. It is the combined and revised polity2 score as reported in the POLITY IV database and we use it in this study as a proxy to capture the influence of the system of the political regime (or domestic politics) on economic growth in the country. This polity indicator combines democracy and autocracy as systems of political regime in a single measure as the difference of the score of democracy from the score of autocracy. Specifically, it is the democracy score less autocracy score. The polity2 measure was specifically modified to enhance its use in time-series analysis.

CRISIS/WAR is a variable of political instability and is captured as a dummy variable for the impact of the civil conflict in Sierra Leone which covered the period 1991-2001. Therefore the CRISIS dummy takes the value of 1 for periods of the civil war and 0 otherwise.

Construction of the policy index The index of macroeconomic policy as used in this study is constructed using the technique of principal component analysis with SPSS 15.0. The construction of the index is advantageous not only because it helps conserve the number of degrees of freedom following the reduction in the number of variables to be run in the model, but also that it helps avoid the potentially high correlation among the macroeconomic variables. Sricharoen and Buchenrieder (2005:2) accords that “PCA is an indicator reduction procedure to analyse observed variables that would result in a relatively small number of interpretable components (group of variables), which account for most of the variance in a set of observed variables.” The variables that formed the index include government consumption expenditure as share of GDP (GEX) as proxy for fiscal policy, inflation rate (INF) as proxy for monetary policy, and trade openness (TOPEN) as proxy for trade policy.

The PCA technique involves a data reduction process in which the variables are scored following their running with the SPSS package. The eigenvalues are also calculated for each component. In our construction of the ‘policy’ index, the eigenvalues show that 58% of the variance is explained by the first principal component, with the second and third accounting for the remaining 42%; which implies that the first principal component alone explains the variation of the dependent variable better than any other combination of the three variables used. Hence, the first principal component is considered as an appropriate measure of the

114 macroeconomic policy index. In the construction of this macroeconomic policy variable, the scores obtained are 0.439 for trade openness, -0.440 for inflation and 0.436 for government expenditure (Appendix 5.1). Hence, the individual contribution of the components of the policy index is shown as follows: Policy= -0.440 INF +0.439 TOPEN +0.436GEX …….(5.11)

5.2.3 The Estimation Procedure

The time series analysis of the impact of foreign aid on economic growth in Sierra Leone follows the technique of cointegration, which is employed to estimate the long-run impact of aid on economic growth; accompanied by an Error Correction Model representation which provides estimates for the short-run and the adjustment term once cointegration is found to exist. The time series period covers from 1970 to 2007 which is a relatively long series with the advantage of obtaining adequate degrees of freedom as well as having the capacity to incorporate more growth determinants into the model and hence yielding less unbiased estimates that could have been made biased as a result of omitted variables. We introduce a triangulation approach involving a combination of different techniques of cointegration to establish the impact of foreign aid on economic growth.

As most macroeconomic variables are found to be non-stationary or integrated of order 1 over time (Hendry and Donik, 2001), econometric practice had previously involved the differencing of any non-stationary variables before doing the necessary estimations. The cost to such a practice had involved the loss of long-run information on the variables. Therefore, to deal with the problem of non-stationarity, later studies have increasingly used the standard technique of cointegration and error correction mechanism (ECM) to estimate time-series relationships. Generally, cointegration establishes the existence of long-run relationship among the variables. As our study involves the use of economic variables which are largely seen to be non-stationary, in order to obtain long-run information, there is the need to establish the existence of cointegrating relationship between economic growth and its determinants including foreign aid which is our variable of interest.

There are three main methods of cointegration that are commonly used in the econometrics literature. The Engel- Granger two-step procedure, the Johansen Likelihood approach and the

115 more recent Autoregressive Distributed Lag (ARDL) bounds test approach to cointegration modified by Pesaran and Shin (1999). The Engel and Granger (1987) approach which involves a two step procedure is limited to a bivariate model and hence cannot be applicable in this study for which the empirical models constitute more than two variables. Thus, the study employs the other two approaches (the Autoregressive Distributed Lag (ARDL) bounds test approach and the Johansen Maximum Likelihood technique to cointegration) to estimate the aid-growth relationship, with both approaches being applicable in a multivariate regression situation. Though this study uses and triangulates findings from both approaches, it considers the ARDL approach as the main technique of estimation; which implies should there be any contradicting results between these two techniques, priority is given to those estimates from the ARDL approach. Prioritisation of this approach is based on the unique power of its estimates being found to be more efficient and reliable in small samples than those from counterpart estimators such as the Johansen technique (Inder, 1993; Banerjee et al., 1993).

The choice of the ARDL methodology to approaching this study’s investigation is premised on several considerations. First, as opposed to the Johansen Likelihood approach to cointegration analysis, the ARDL approach avoids the problem of order of integration. The cointegration approach by Johansen (1991) and Johansen-Juselius (1990) requires that variables be of the same order of integration (i.e. I(1)). Hence, the ARDL approach is found to have the flexibility advantage in that it can be applied irrespective of whether the variables are of different order of integration (Pesaran and Pesaran, 1997). Secondly, Inder (1993) showed that estimates from ARDL approaches are much reliable than their counterparts even if the dynamic structure is over-specified; and also that sizes of the t-tests from an estimator that uses an ARDL approach are much more reliable. Thirdly, Banerjee et al. (1993) show that the ARDL approach to cointegration is especially attractive when carrying out cointegration in small samples, and that it is yet more efficient than other VAR methods. This is also confirmed by Pesaran and Shin (1999) and Pesaran et al. (2001) who show that the ARDL model outperforms alternative approaches like the Phillip and Hansen’s Fully Modified OLS when the sample size is small. Finally, Pesaran and Shin (1999) show that an appropriate modification of the orders of the ARDL model is adequate to simultaneously correct for residual serial correlation and the problem of endogenous regressors, thus giving the ARDL an advantage over other approaches to cointegration. This is also justified by Harris and Sollis (2003), and Constant and Yue (2010). The inclusion of dynamics is shown 116 by Inder (1993) and Pesaran and Pesaran (1997) to help correct for endogeneity bias. As foreign aid has been largely argued to be endogenous (Boone, 1996; Franco-Rodriguez et al., 1998; Burnside and Dollar, 2000; Feeny, 2005), this makes the use of this methodology all but appropriate for the estimation of the aid-growth relationship.

The modified approach by Pesaran and Shin (1999) uses the error correction version of the ARDL model and takes the following form:

m−1 m−1 ’ yt = α0 + α1t + ∑ φiyt-i + ∑ πi xt-i + d 1yt-1 + d 2xt-1 + ηt ……….(5.12) i=1 i=0

Where d 1 and d 2 are parameters of the long-run relationship variables. Φ and π are matrices of parameters yt is a vector of endogenous variables,

and x t is a vector of explanatory variables α is a vector of constants and t is a deterministic trend m = max (q, s+1), Embedding (5.7) into the error correction form of the ARDL model, the conditional VECM becomes:

p LRGDP t = β0 +δ1LRGDP t-1+ δ2LAID t-1+ δ3LPI t-1+ δ4LIQI t-1+ δ5Policy t-1 + ∑ φiLRGDP t- i=1

q q q q I + ∑ ωjLAID t-j+ ∑ lLPI t-l+∑ γmLIQI t-m+ ∑ ηpPolicy t-p+εt …………(5.13) j=1 l=1 m=1 p=1

Where δi are the long-run multipliers, β0 is the drift and εt are white noise errors. p and q are the appropriate ARDL model orders. This becomes the base equation.

The ARDL approach uses the F-test to establish the existence of a cointegrating relationship. The test however has a non-standard distribution, which implies the critical values differ from those in the standard distribution. Pesaran and Pesaran (1997) generate separate critical values that tabulate two sets of values. The first value (upper critical bound) of the F-test assumes that all the variables are I(1) and the second (the lower critical bound) that they are I(0). If the calculated F-statistic appears above the upper value of this band, the null hypothesis is rejected suggesting the existence of cointegration between the variables irrespective of whether they are I(1) or I(0). If the F-statistic falls below the lower critical bound, the null hypothesis of no cointegration cannot be rejected; while a value within the

117 bounds (i.e. within the lower critical bound and the upper critical bound) implies an inconclusive test. In this study, critical values are used for all three sets of level of significance i.e. at the 1%, 5% and 10% levels of significance. However, the critical values used by Pesaran and Pesaran (1997) are generated for samples from 500 observations. As this paper uses a sample of 38 years, we therefore use the critical values generated by Narayan (2004) which are available for samples of 30 observations to 80 observations.

Considering the aid-growth base model, the hypothesis is specified as follows:

Ho: δ1= δ2= δ3= δ4= δ5=0 Against HA: δ1≠ δ2≠ δ3≠ δ4≠ δ5≠ 0

Once cointegration is established, in the second step, the conditional ARDL (p, q 1, q 2, q 3, q 4)

long-run model for LRGDP t can be estimated. This involves selecting the optimal orders of

ARDL model in the variables (i.e. p, q 1, q 2, q 3, q 4) which is done using either Akaike information criteria (AIC) or the Schwartz/Bayesian Information criteria (SIC or SBC). The model selection criteria according to Shrestha and Chowdhury (2005) are a function of the residual sums of squares and are equivalent asymptotically. In this study, the SIC is used to select the orders of the ARDL specifications, which in the context of this study, has a comparative advantage over the AIC. In a comparison of the AIC and SIC in Monte Carlo experiments, Pesaran and Shin (1999) showed that though the ARDL-AIC and ARDL-SIC have quite similar small-sample properties, the ARDL-SIC performs slightly better in the majority of the experiments. This they suggest may be due to the fact that the Schwartz criterion is a consistent model-selection criterion whereas the Akaike is not. Hence, the SIC can be described as being more parsimonious with the lag length selection and is a consistent model selection criterion (Pesaran and Shin, 1999).

In the third and final step, we obtain the short-run dynamic parameters by estimating an error correction model associated with the long-run estimates. This is specified as:

p q q q LRGDP t = + ∑ φiLRGDP t-i+ ∑ ωjLAID t-j+ ∑ lLPI t-l++ ∑ γmLIQ1 t-m i=0 j=1 l=1 m=1

q ∑ ηnPOLICY t-n +ζecm t-1+εt ……(5.14) n=1 Where φ, ω, , γ and η are the short-run dynamic coefficients of the model’s convergence to equilibrium, and ζ is the speed of adjustment

118

The adjustment term, which indicates the speed of adjustment to disequilibrium, is also estimated at this stage.

To ascertain the goodness of fit and/or model adequacy, diagnostic and stability tests are further conducted in the ARDL approach. The diagnostic tests which are automatically derived by the Microfit software upon estimation of the ARDL model, examine serial correlation (or autocorrelation), functional form of the model, normality of the residuals, and heteroscedasticity associated with the models. This is so because the ARDL is OLS and so the need to satisfy the classical assumptions of least squares is obvious if the model is to be considered adequate and the estimates considered reliable for inference. The stability test of the regression parameters is conducted using the technique of stability testing by Brown et al. (1975), namely the cumulative sum of recursive residuals (CUSUM) and the cumulative sum of squares of the recursive residuals (CUSUMSQ). It tests for structural stability of the parameters within the 5% critical bounds.

The second methodology used in our enquiry is the Johansen ML approach to cointegration. In 1988, Johansen developed a technique of cointegration suited to the case of multivariate models and with the possibility of simultaneity in the variable relationships. The approach has several advantages that warrant its use in this study to estimate the aid-growth relationship. First, variables of the study’s base model for examining the impact of foreign aid on economic growth are all I(1) which is a requirement for the application of the Johansen ML approach to obtain the long-run estimates of the aid-growth relationship. De Beof (2000) and Villavicencio and Bara (2008) for instance, argue that using the Johansen approach in a model with a mixture of I(1) and I(0) variables produces biased results. Secondly, the Johansen approach accounts for the possibility of endogenous regressors in the model for which foreign aid has been argued (Boone, 1996; Franco-Rodriguez et al., 1998; Burnside and Dollar, 2000; and Feeny, 2005) to be endogenous. Third, using the Johansen procedure allows for comparison with some portion of the existent country aid-growth literature for which this technique has also been used (e.g. M’amanja and Morrissey, 2005; Bhattarai, 2009). Finally, our use of the Johansen Approach arises from the need to complement the findings of the ARDL approach with the aim of triangulating the findings and strengthening reliability of results. 119

The Johansen cointegration analysis in Microfit crucially depends on whether the VECM contains intercepts and/or time trends, and also whether the intercept or trend coefficients are restricted (Pesaran and Pesaran, 1997). In this study, the option where allowance is made for an intercept in the cointegrating vector but no trend is chosen. A visual observation of the plots of the variables (see Appendix 5.2B) does not show an obvious trend in the variables and neither do we expect a trend in the cointegration of economic growth and its determinants including foreign aid, with time. Therefore, a trend is not included in the analysis.

The Johansen ML cointegration test uses two sets of statistics to test for the presence of cointegration: the Maximal Eigenvalue statistic and the Trace statistic. Microfit further computes model selection criteria for AIC, SIC and HQC for different values of r, the rank of the long-run matrix. For the sake of consistency with the ARDL and for its advantage of being a parsimonious model selection criterion, this study uses the SIC here to further confirm the choice of the r number of cointegrating relationships. Hence, where there is a conflict in the choice of r between the Trace and Maximal Eigenvalue tests, the study complements with the r chosen by SIC. But before the test for cointegration is made, the Johansen technique requires the specification of the order of the var, which implies there is the requirement to test for the order of the var of the model before doing the cointegration tests. Using SIC for similar reasons as mentioned previously, all the models in this study choose 1 as the order of the var. Further, Pesaran and Pesaran (1997) recommend that a confirmation of the non-existence of serial correlation is necessary at this stage before conducting the cointegration test. In all the specification run in this study, there is no evidence of serial correlation and hence had the cause to proceed with use of the chosen lag order of 1 in the cointegration tests.

For both techniques of cointegration analysis, the study employs Microfit 4.0 regression package to conduct the cointegration tests, diagnostic test and run the regressions; as this econometric software is more suited to running time-series regressions.

120

5.2.3.1 Unit Roots Tests Before cointegration analyses are deemed to be carried out, unit roots test must be first conducted. Unit root test is a test conducted to ascertain the stationarity of the variables. Generally, cointegration requires that the variables to be estimated are integrated of order 1. This is a particular requirement for the Johansen Method. It is only when all the variables indicate the presence of a unit root will there exist a long-run relationship in the case of the Johansen technique. In the ARDL approach, Pesaran and Shin (1999) argue that the approach is such that it does not require pre-testing for stationarity before estimation of the model. However, as the F-Test for cointegration is such that the critical values used to ascertain the existence of a cointegrating relationship lie within the bounds of I(0) and I(1), it implies the variables used in the estimation must be either I(0) or I(1); and hence needs a unit root test to ensure none of the variables go beyond I(1) or else the ARDL technique cannot be used to carry out the estimation. Ouattara (2004) argues that the approach collapses in the presence of I(2) variables. Further, it is necessary that the dependent variable be integrated of order 1 (as is also argued by Afzal et al., 2010:45) in as much as the regressors can remain a mixture of I(0) and I(1). Earlier regressions run with an I(0) dependent variable had coefficients of the ECM term beyond the theoretical bounds of 0 to -1. It is thought that the stationary nature of the dependent variable may have been the cause. In response, this study had to resort to the use of transformed variables in their natural logarithms, which emerged to have resolved the problem.

Therefore, both cointegration techniques require the need to conduct unit root tests. Several approaches are available in the literature that are used to conduct unit roots test, but this study however employs the Augmented Dickey Fuller (ADF) approach for two main reasons: First, though the ADF has been subject to criticisms for its low power when applied to near unit root processes (Elbadawi and Soto, 1997), the ADF has been widely used to test for stationarity in time series data. Thus, for comparison of findings, it is essential to use the most widely used technique. Secondly, the ADF has proved to perform satisfactorily when there are fewer observations (Elbadawi and Soto, 1997); which cannot be very different from this study with a fairly sufficient time series of 38 years, but which cannot be said to be very long. Therefore, we adopt the ADF approach and present unit root test result for the variables used across the models (Table 5.1). The test results show that all the variables used in the base model are I(1) which allows the use of both the ARDL and Johansen approaches to cointegration in the estimation of the base model.

121

Table 5.1: ADF Unit Roots test results - including constant but without trend: 1970-2007 Levels 1st Difference Variable SIC Lag ADF t-stat. SIC Lag ADF t-stat. I(d) LRGDP 2 -1.790406 0 -4.472844 I(1) LAID 0 -1.736667 0 -7.227074 I(1) LPI 0 -2.445824 0 -6.924393 I(1) LIQ1 0 -1.835226 0 -6.090586 I(1) POLICY 1 -2.174771 0 -4.762549 I(1) Additional Variables CRISIS/WAR 0 -1.555815 0 -5.830952 I(1) POLIT 0 -2.175246 0 -5.804381 I(1) LIQU 0 -1.267710 0 -6.167158 I(1) INF 0 -2.950797 I(0) LTOPEN 0 -3.286945 I(0) LGEX 0 -2.110433 0 -8.711782 I(1)

Critical Value 1% 2 -3.632900 0 -3.626784 5% -2.948404 -2.945842 Critical Value 1% 1 -3.626784 5% -2.945842 Critical Value 1% 0 -3.621023 5% -2.943427

5.2.4 The Empirical Results

5.2.4.1 The ARDL Estimation Results

In this section, the study applies the ARDL cointegration approach to present the cointegration test results, the long-run estimates, and the short-run estimates for the impact of foreign aid on economic growth in Sierra Leone and the tests of model adequacy.

Cointegration test As the test for stationarity of the regression variables using the ADF unit roots test (Table 5.1) showed that all the variables in the base regression (Table 8- Model A below) are I(1), with none being I(2), it implies the study can proceed with the use of the ARDL technique to cointegration which is applicable when the regressors used in the model are either I(0) or I(1) or a mixture of both. The cointegration test (Table 5.2) for the existence of a long-run relationship between economic growth and its regressors including the foreign aid variable of interest reveals the existence of cointegration at the 5% level of significance when the F- statistic for cointegration is compared with the special F-test critical values. With the confirmation of long-run relationship between economic growth and its determinant in Sierra Leone, the study can proceed with the estimation of the long-run and short-run dynamics to determine the impact of foreign aid on economic growth in the country.

122

Table 5.2: Long-Run Impact of Foreign Aid on Economic Growth: The ARDL Results Model B: Disaggregated Policy DEPENDENT VARIABLE IS LOG REAL GDP MODEL A: Base Model Variables Foreign Aid 0.379** 0.351** (2.193) (2.276) Private Investment 0.293** 0.261** (2.600) (2.388) Property Rights 1.861* 2.292* (1.673) (1.797) Macro Policy 0.134 - (1.003) Government Expenditure - 0.658 (1.142) Inflation - 0.002 (1.243) Trade Openness - 0.578 (1.370) CONSTANT 16.536*** 12.219** (7.890) (2.617)

F-TEST FOR COINTEGRATION 4.773** 4.265** SERIAL CORRELATION 3.75 4.88** FUNCTIONAL FORM 0.01 0.01 NORMALITY 0.02 0.57 HETEROSCEDASTICITY 2.16 5.27** Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T-statistic in parenthesis

The Long-Run Empirical Estimates The ARDL estimates presented in Table 5.2 show foreign aid to have a positive and significant impact on economic growth in Sierra Leone. For any 1% increase in aid/GDP ratio, economic output will increase by 0.4%. The control variables all emerged with the expected signs with intuition. Private investment is found to have a significant impact in promoting economic growth in the country. For any 1% increase in private investment in the country, economic output will increase by 0.3%. The quality of institutions (otherwise referred to as property rights) shows some indication of determining economic growth in the country as it moderately impacts on economic growth at the 10% level of significance. Macroeconomic policy index (comprising fiscal, trade and monetary policy) has not proved to be impressive for economic growth in Sierra Leone. Even when attempt is made (model B in Table 5.2) to disaggregate the policy variables to its individual components (note however that this model fails the test for serial correlation and heteroscedasticity), the individual policy variables also emerge to be insignificant in determining long-run economic growth in the country.

The Short-run Estimates The short-run estimates are also of relevance in the estimation of the aid-growth relationship. According to De Boef (2000: 82), “the short-run change is necessary to maintain the long-run

123 relationship”. Krause (1997) in fact chose not to present the long-run estimates of the relationship between economic expectation and economic conditions, but only reported the short-run estimates because such are unaffected by the limitations in the use of long-run estimates. However, in the case of foreign aid, there is expected to be a long-run and short- run relationship with economic performance as aid is mostly given to boost investment, but also to augment recurrent expenditure and short-term shocks in macroeconomic stability. Table 5.2.1 below presents the dynamic short-run estimates of the aid-growth relationship.

Table 5.2.1 Short-Run Impact of Foreign Aid on Economic Growth: The ARDL Results Dependent variable is LOG REAL GDP (dLRGDP) Regressors MODEL A: Base Model Model B: With Disaggregated Policy variable First lag of Real GDP -0.458** -0.578*** (2.620) (3.403) Foreign Aid 0.065*** 0.058*** (4.417) (3.687) Private Investment 0.051** 0.043** (2.371) (2.180) Property Rights 0.321*** 0.378*** (4.488) (5.487) Macro Policy 0.023* - (1.703) Government Expenditure - 0.108* (0.750) Inflation - 0.003 (1.272) Trade Openness - 0.095** (2.12) CONSTANT 2.850 2.013 (1.585) (1.167) Ecm(-1) -0.172* -0.165** (1.960) (2.065) R2 0.65 0.73 Adjusted R 2 0.58 0.65 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T-statistic in parenthesis

The short run estimates (Table 5.2.1) are very much in consonance with the long-run estimates as foreign aid is found to impact on economic growth at a high level of significance in the short-run. However, the proportion by which aid contributes to economic growth is comparatively lower in the short-run than in the long-run. In the short-run, the estimates show that for a 1% increase in aid, economic output will increase by an approximate proportion of only 0.1%. However, in the short-run, all the control variables emerge to significantly contribute to economic growth in the country. Private investment significantly determines economic growth at the 5% level of significance, while property rights impact at the 1% level. Macroeconomic policy index, though at only 10% level, emerges to contribute to economic growth in the short-run. This may not be surprising as the usual aim of

124 macroeconomic policy is largely to stabilise the economy following shocks from inflation and balance of payment deficit. In the disaggregated policy variable model (Model B of Table 5.2.1), the short run estimates show that significance of the macro policy index is largely attributed to trade policy and government expenditure.

The error correction mechanism tells the degree to which the equilibrium behaviour drives the short-run dynamics (De Boef, 2000: 82). Thus, the ECM term is of importance in cointegration analysis. The coefficient of the ECM term which signifies the speed of adjustment of the model to equilibrium in the event of shocks, shows that 17.2% of disequilibrium errors are corrected. The ECM term is also found to be negative and significant; further confirming the existence of a long-run relationship between foreign aid and economic growth in Sierra Leone.

Diagnostic test The diagnostic tests indicate the adequacy of the model in terms of the reliability of its estimates for inference. The model passes all the diagnostic tests for serial correlation, functional form, normality and heteroscedasticity associated with the model at the 5% level of significance.

Plots of the stability tests (CUSUM and CUSUMSQ) for the base model are given in figure 5.1 below.

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

125

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Figure 5.1: CUSUM and CUSUMSQ Plots for the base model

CUSUM and CUSUMSQ plotted against the critical bounds at the 5% level of significance (figure 5.1) show that the coefficients of the model are stable over time. Therefore, the model estimates are sufficiently reliable enough for inference. In Model B where the components of the policy index are individually added into the model, the diagnostic test are not quite strong as though the model passes the tests for functional form of the model and normality of the residuals, yet it fails the tests for serial correlation and heteroscedasticity; the latter probably due to the fact there is a mix of I(1) and I(0) variables in the estimated ARDL model since the trade openness and inflation variables are I(0) (see Shrestha and Chowdhury, 2005; Kargbo and Adamu, 2010: 53). Hence, this study’s analysis and inference are focused more on the base model where the composite policy index is used.

5.2.4.1.1 Further Robustness Specifications Further robustness check regressions (as specified previously) were estimated to ascertain the impact of foreign aid on economic growth in Sierra Leone (Appendix 5.3 & 5.4). All of these models clearly show the existence of cointegration between economic growth and its determinants including foreign aid, thus enabling the conduct of long-run estimates. The model estimates are very much consistent with the main estimation model as aid is found in to have a positive and significant impact on economic growth. Specifications for the long-run estimates include Models B, C and D in Appendix 5.3. In model C, a variable for political crisis is added to the base model which specifically is a dummy variable capturing the impact of the 11 civil war of 1991-2001 in Sierra Leone. Here as well, foreign aid remains to have a significant impact on economic growth. In the other two models (Model B and D), where

126 attempt is made to alter the indicators of political and institutional quality, the impact of foreign aid remains robust. In the regression with use of governance indicator (Model B), foreign aid is found to have a significant impact on economic growth at the 5% level of significance. In the regression where polity quality indicator is used (Model D), aid is found to even have a highly significant impact on growth just as private investment.

The short-run estimates of these robustness check models (Appendix 5.4) just mirrors the long-run estimates and further confirm the robustness to aid’s impact on economic growth in both the long and short-run. Only in model B where the measure of governance is used to replace property rights that foreign aid did not emerge to be significant in the short-run. The ECM terms for these specifications all emerge to be negative and significant as expected to further confirm the existence of cointegration in these models and also to ensure that any disequilibrium in the model is corrected for at a certain proportion.

The models pass the diagnostic tests for functional form, serial correlation, normality and heteroscedasticity associated with the models. Only Model C, where the political crisis dummy is added to the base model that failed the test for serial correlation. But as the estimates of the ARDL model are robust in the presence of serial correlation (Pesaran and Shin 1999:372; Laurenceson and Chai 2003:30), it implies even the estimates from that model are largely reliable to serve as robustness check for the impact of aid on economic growth as is estimated by the main model. Laurenceson and Chai (2003) argue that the presence of autocorrelation does not affect the parameters and that the model selected is optimal based on SIC. Pesaran and Shin (1999:372) suggest that in situations of serial correlation, the ARDL specification needs to be augmented with an adequate number of lagged changes in the regressors before estimation and inference are made. In Pesaran and Pesaran (1997), they suggest that for annual data the maximum number of lags for the ARDL specification should be 2; which is adopted in this estimation. In this augmented specification, Pesaran and Shin (1999) showed that results can be directly applicable to the OLS estimators of the short-run and long-run parameters. Hence, they conclude that a suitable choice of the orders of the ARDL model is vital for valid inference. But once that is carried out, OLS methods for estimation of the long-run parameters and computation of the standard errors are valid. As a whole, the ARDL estimate of the relationship between foreign aid and economic growth finds aid to be an important determinant of economic growth in Sierra Leone, and this result is found to be robust across specifications. 127

5.2.4.2 The Johansen ML Estimation Results

From the Johansen cointegration test results, the hypothesis that there exists no cointegrating relationship is rejected by both the Maximal and Trace statistic at the 90% confidence interval. This therefore implies there is at least one cointegrating relationship. The test that there is more than one cointegrating relationship is however rejected by both test statistics, thus confirming the existence of only one cointegrating relationship among the variables. Therefore, one cointegrating relationship is chosen in our estimation of the long-run relationship. Once cointegration is confirmed, the long-run relationship is then estimated.

Table 5.3 below presents the long-run estimates of the aid-growth relationship for Sierra Leone using the Johansen ML approach. Two models of the same variables are presented. In Model A, the study uses the base model and adds an intercept in the cointegrating vector (The ‘restricted intercept, but no trend’ option in Microfit). Normalising with the dependent variable, LRGDP, the long-run estimates show that foreign aid has a highly significant positive impact on economic growth in the country with 1% increase in foreign aid leading to a 0.2% increase in economic growth. Private investment also emerges to have a highly significant and positive impact on economic growth. Property rights have a moderately significant impact in determining economic growth in the country. Macroeconomic Policy, though with a positive sign, does not appear to significantly impact on economic growth.

128

Table 5.3: Long-Run Impact of Foreign Aid on Economic Growth: The Johansen Results DEPENDENT VARIABLE IS LOG OF REAL GDP MODEL A: with MODEL A.1: with Restricted Intercept Unrestricted Intercept Foreign Aid 0.191*** 0.189*** (0.0484) (0.0476) Private Investment 0.236*** 0.240*** (0.0509) (0.050) Property Rights 0.575* 0.544* (0.3329) (0.322) Macro Policy 0.0085 -0.0022 (0.039) (0.0377) CONSTANT 18.938*** 5.374*** (0.5659) (1.487)

Var Order (SIC) 1 1 Cointegration Test R (Maximal Eigen) 1* 1* R (Trace ) 1* 1* R (used in Regression) 1 1 Short-run Results ECM -0.292*** -0.283*** (0.0745) (0.078) Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. Standard errors in Parenthesis

In Model A.1, the study uses the same variables but rather employs an unrestricted intercept (the ‘unrestricted intercept and no trend’ option in Microfit). The results do not significantly change from those with the restricted intercept. Foreign aid still emerges to have to highly significant and positive impact on economic growth with 1% increase in aid related to a 0.2% increase in economic output. Private investment as well shows to have a highly significant influence in promoting economic growth; and property rights have a moderate impact. Macroeconomic policy just as in the original regression (Model A) did not also emerge to significantly influence economic growth in the country.

The ECM term which indicates the speed of adjustment of the model to equilibrium is negative and highly significant, further confirming the existence of cointegration among the variables of the model. The coefficient of the ECM shows that should there be shocks that will cause disequilibrium in the economy, 29.2% of the errors emerging from such disequilibrium will be corrected.

129

5.2.4.2.1 Further Robustness Specifications To ascertain our finding following the Johansen cointegration procedure, that foreign aid is an important determinant of economic growth in Sierra Leone in the long-run, the study further checks for robustness of such results by altering the base specification just as is done in the ARDL estimations to see whether with a change of specification, foreign aid remains significant on economic growth as was found with the base model.

Appendix 5.5- Models B, C and D present the results of these robustness check specifications which are the same as those also specified in the ARDL model. In all of these models, at least one cointegrating relationship is found to exist in the model of economic growth and its determinants. In specifications where there exists more than one cointegrating vector as in model D, the study reports only the estimates for vector 1 which corresponds to the ECM1 in the short-run adjustment. This is so because though the test shows that there may be two cointegrating relationships and for which the SBC also chooses, the ECM results could only confirm cointegration (through a negative and significant ECM) in ECM1, and hence this vector could only be found to be the one with the valid estimates. In model C where the political crisis variable is added to the base model, foreign aid still remains to have a significant impact on economic growth in the long-run at the 5% level of significance. In Models B and D where replace the property rights score is replaced with other indicators of political and institutional quality, foreign aid remains an important determinant of economic growth. In Model B, where the property rights score is substituted with the measure of governance, foreign aid is also found to be effective but only at the 10% level of significance. In model D, where the polity quality measure is used, aid showed to have a positive and highly significant impact on economic growth.

In all of these specifications, the ECM term emerges to be negative and significant in as expected. This therefore further confirms the existence of a long-run relationship among the variables in the various specifications; and that some proportion of any errors arising from disequilibrium of the model is corrected.

Therefore, the Johansen estimates only further show that the study’s finding that aid is an important determinant of long-run economic growth in Sierra Leone is robust across approaches and specifications.

130

5.2.4.3 Discussions and conclusion on the Impact of Foreign Aid on Economic Growth The aforementioned analysis of the aid-growth relationship in Sierra Leone has shown that using a triangulation of approaches and specifications, foreign aid is found to significantly promote economic growth in Sierra Leone within the period 1970-2007. Both the estimates of the ARDL bounds test and the Johansen approaches to cointegration are in agreement that foreign aid is a positive and significant determinant of economic growth in the country. Changes to the aid-growth specification in both approaches also confirm that foreign aid significantly fosters economic growth in Sierra Leone. The agreement of this finding across approaches and specifications only provides support for reliability and validity of the findings and conclusions reached on the impact of foreign aid on economic growth.

The study’s finding that foreign aid has a positive and significant impact on growth is consistent with the greater portion of the aid-growth literature. The finding is supportive of the supplemental aid-growth argument which postulates that aid contributes to economic growth in recipient countries. As evident in the Chenery and Strout (1966) framework, foreign aid flows into a country is expected to positively contribute to economic growth of that country. Chenery and Strout (1966) see foreign aid to fill the savings and trade/foreign exchange gap typical in less developed countries like Sierra Leone, and in effect should contribute to fostering economic growth in such countries. This finding therefore, does not provide support for the displacement theorists (like Griffin 1970; Griffin and Enos 1970; and Bauer, 1971) who rather argue that foreign aid negates economic growth.

Moving away from the theoretical literature, the empirical literature has been inconclusive on the impact of aid on economic growth. However, this finding, through the ARDL and Johansen estimation methods, is supportive to the portion of the empirical literature that shows that foreign aid has a significant and positive effect on economic growth. Emerging from a time series analysis, this result supports the empirical findings of Murthey et al. (1994); Lloyd et al. (2001); Gounder (2001); Mavrotas (2002); and Bhattarai (2009) who also used time series analysis in different country studies to show that foreign aid has a significant and positive impact in determining economic growth. The study by Lloyd et al (2001) for instance used the ARDL approach to cointegration (the main approach also employed in this study) to investigate the impact of aid on economic growth at the country level for Ghana. In consistence with their finding that aid positively impacts on economic growth, the findings of

131 this study just further confirm their conclusions but with more estimation approaches and specifications to provide robustness check, and in another case study in a country (Sierra Leone) where government expenditures are vastly aid-dependent. The study by Gounder (2001) also used the ARDL model by Pesaran and Shin (1999) for the period 1968-1996 to show that foreign aid to Fiji has a significant impact in promoting economic growth in the country. Murthey et al. (1994) and Bhattarai (2009) both used the Johansen ML approach and found aid to significantly contribute to economic growth in Cameroon and Nepal respectively. However, this study’s finding could not provide support for the studies by Islam (1992), Mbaku (1993), Fenny (2005) and Javid and Qayyum (2011) who could not find aid to significantly contribute to economic growth in Bangladesh, Cameroon, Papua New Guinea and Pakistan respectively. Fenny (2005) used the ARDL methodology but could not find foreign aid to significantly contribute to economic growth in Papua New Guinea. Likewise, Javid and Qayyum (2011) used the same ARDL approach to examine the impact of aid on economic growth for Pakistan, but could not establish a significant impact of aid, unless when aid is interacted with macroeconomic policy index.

Commentators may hence be interested to know why foreign aid could be effective in directly contributing to growth in Sierra Leone, as against the findings from some other case studies. Whilst it may remain an area for further investigation, some explanation may be suggested. First, whilst methodological differences cannot entirely be ruled out for deferring aid effectiveness results, yet the studies by Feeny (2005) and Javid and Qayyum (2011) also used the ARDL approach but with different aid effects.

One possibility may be the differing purposes of foreign aid to recipient countries. Aid to Sierra Leone may be less politically oriented compared to aid to Pakistan. Whilst the promotion of democracy may be a crucial reason for donors granting aid to Sierra Leone, yet poverty reduction and indeed economic growth (a conduit to poverty reduction) may be a more crucial reason for granting aid to the country (as fieldwork interviews with the country’s leading aid donors revealed). In fact, Sierra Leone received significant proportions of aid even in non-democratic periods. As the introductory section of this paper reveals Sierra Leone to be one of the poorest in the world, the granting of aid to such a country may hence be more growth and poverty oriented than otherwise. This is further reflected in the donor disbursement of more grants than loans to the country (as shown in chapter 2). With Sierra Leone being a largely poor and hence capital starved economy, it is expected that foreign aid 132 disbursed to such a country will supplement savings and consequently boost economic growth.

Further, as Sierra Leone is a non-industrialised country, intermediate inputs required for production in the scarcely available domestic factories and industries have to be imported from overseas. Similarly, as a largely mineral producing country, nearly all of the equipment required for mining have to be imported. With the earnings from exports very much likely to be lower than the import requirement of the country (particularly so since exported products from the country are virtually all in their primary form), a foreign exchange gap becomes inevitable. What this implies is that the large inflows of foreign assistance from donors fill this foreign exchange gap (as is explained by Chenery and Strout, 1966) and hence ultimately promote economic growth.

Finally, whilst the period of the civil conflict may have adversely affected the performance of the economy, it may also be that in the case of Sierra Leone, the aid enhancing environment has not been that appalling to ensure aid is ineffective on growth in the country. Sierra Leone has obviously had a history of corruption in public places (see comments in Appendix 2.1) and which implies the misappropriation of donor assistance may not be an exception. However, past and present political regimes have been largely associated with some informal donors. China and Libya for instance have been close friends to some past and present political regimes in the country (Neville, 2002; Kamara, 2008; Acemoglu and Robinson, 2012), with some proportion of their aid being unofficial. As such informal aid may be given on personal grounds and does not enter the official conduits, the implication is that such aid could be used in securing neo-patrimonial networks by politicians without necessarily affecting the development effectiveness of official aid. Hence, while there could be some evidence of official aid being misused by the bureaucracy, aid corruption by politicians may be largely skewed towards non-official aid whose purpose may be personal.

In conclusion, in terms of the economic growth criterion, the study provides evidence to show that foreign aid to Sierra Leone is positively associated with economic growth. Hence, if poverty remains evident in the country amidst aid inflows, it is not that foreign aid has not been effective in promoting economic growth. It may be that either growth may not have been pro-poor or that aid has not directly reduced poverty in the country. Thus, if the purpose of donor aid to Sierra Leone is to promote economic growth, then the study’s finding that aid 133 fosters economic growth is a motivation for donors to continue to give aid to Sierra Leone as it yields the desired results. However, it should also be noted that even though growth has been found to respond to aid disbursement, the economic significance in terms of the magnitude of the response has not been that high. Elasticity of 0.4 (as the ARDL aid estimate shows) or even 0.2 (as the Johansen aid estimate shows) would only imply that the response is relatively weak despite being statistically significant. Hence, for enhanced growth effort in the country, other factors such as private investment and the quality of institutions, which have also been found to induce growth response, should be strengthened to complement aid effort.

5.3 Further Analysis of the Impact of Aid on Economic Growth: Aid Disaggregation

In addition to examining the relationship between total foreign aid and economic growth for the entire study period of 1970-2007; in this section the study carries out further examination of the aid-growth relationship, by looking at the disaggregates of foreign aid to see whether the modality of the aid delivered differs in their relative impact on economic growth in the country. Though total foreign aid to Sierra Leone is found to contribute to economic growth, it is vital to further examine whether all disaggregated aid types have been effective in promoting economic growth in the country. The reasons for further examining aid in its disaggregated forms are already discussed in the study’s conceptual framework (Chapter 4). For the scope of the analysis on economic growth, five types of aid modality are considered: Grants, Loans, technical assistance, multilateral and bilateral aid. The unit roots test result for the aid disaggregates is shown in Appendix 5.6.

5.3.1 Grants versus Loans

In this section, grant and loans are compared in terms of their relative impact on economic growth. The impact of grants especially is worth examining. There has been an increasing shift to grants by donors for poorest countries of which Sierra Leone is one. According to a 2005 article by an anonymous author, the World Bank’s IDA for poorest countries is to increasingly be in the form of grants. In a meeting in April 2005, the IDA countries agreed to focus aid delivery in the form of grants in account of vulnerability to debt in their financial

134 support and hence the increasing direction of support towards grants for such countries. Thus, Sierra Leone being an HIPC member and implying being highly indebted, it is expected that grants and to some extent highly concessional loans could only be disbursed to such a country to promote economic growth with less debt burden. Therefore, it is important to examine whether in fact aid in this form of grants is effective in terms of contributing to economic growth. Further, the African Development Bank (AfDB) for instance (following fieldwork interviews) classify Sierra Leone as a fragile state and hence is a recipient of a special window of assistance in the form of grants that fragile African states receive. The Bank therefore does not provide loans to the country under such classification. It is hence essential to provide empirical evidence to support or dispute the disbursement of this aid type by the AfDB to the country. Another reason for assessing growth effectiveness of grants is based on the assertion that in a country with weak institutions (like Sierra Leone), the disbursement of aid in the form of grants (for which the recipient government is not tied to the requirement to repay) may provide an incentive for misuse. Cohen et al. (2007) for instance argues that since grants do not need to be repaid, the need for effecting financial management is not implied; which could be translated as grants being subjected to financial mismanagement with the ultimate effect of adversely affecting its goal of promoting growth. Thus, though the study assesses and compares grants and loans, the examination of the effectiveness of grants in this country context remains of utmost priority.

When the relative impact of grants and loans on economic growth is examined, the extended growth model is specified as:

LRGDP t = β0 + βgLGrant t + βlLoan t + βiLPI t + βpPolicy t + βiq LIQI t + t …….(5.15)

Where ‘Grant’ denotes grants as share of GDP (in natural log), and ‘Loan’ denotes loans as share of GDP. Note that since there are some negative observations of the loan variable, natural logarithm cannot be used on it and hence is used in its non-log form. As the ‘Loan’ emerges to be an I(0) variable, it implies the study cannot use the Johansen approach here to triangulate with the results of the ARDL approach. This is so because the Johansen approach can only estimate long-run relationships when the variables are all I(1).

135

5.3.1.1 Impact of Grants on Economic Growth Using ARDL Cointegration Approach

Table 5.4 below presents the cointegration test results and long-run estimates of the relative impact of grants and loans on economic growth using the ARDL Approach. Model A in Table 5.4 is the extend form of the aid-growth base model.

Table 5.4: Long-run Impact of Grants on Economic Growth: The ARDL Results DEPENDENT VARIABLE IS LOG REAL GDP MODEL A: Grants 0.287*** (2.799) Loans 0.055* (1.663) Private Investment 0.224*** (3.078) Property Rights 2.490* (1.994) Macro Policy 0.110 (1.166) Constant 15.809*** (7.315)

F-TEST FOR COINTEGRATION 6.224*** SERIAL CORRELATION 0.82 FUNCTIONAL FORM 0.28 NORMALITY 0.67 HETEROSCEDASTICITY 0.01 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T-statistic in parenthesis

Test for cointegration The test for the existence of cointegration between the regressors and economic growth (as reported in Table 5.4) reveals the existence of a long-run relationship at the 1% level of significance. Hence, this study’s estimation of the long-run parameters is in place.

The Empirical Estimates The long-run results (Table 5.4) show that both loans and grants have a positive and significant impact on economic growth. However, comparatively, grants tend to perform better than loans in terms of their long-run impact on economic growth in the country. Whereas loan is only significant at the 10% level of significance, grant rather is significant at the 1% level of significance; which makes the latter better than the former in terms of their impact on economic growth. Thus, grant outperforms loans in terms of their relative impact on economic growth. The magnitude of the coefficient on the grant variable suggests that a 1% increase in grants/GDP ratio relates to 0.3% increase in economic output.

136

The model is found to pass all of the vital diagnostic tests for serial correlation, functional form of the model, normality of the residuals and heteroscedasticity associated with the model. The CUSUM and CUSUMSQ plots for stability of the model (Appendix 5.7.1) reveals that on both plots the model fits within the 5% critical bounds and thus the parameter estimates are stable over time.

However, as opposed to the long-run estimates, in the short-run, grants did not prove to be significant in promoting economic growth in the country. Rather, ‘loans’ remains to be significant and at even 5% level compared to its moderate impact at 10% in the long-run estimates (Appendix 5.7). Thus in the short-run, loans outperform grants in terms of their relative impact on economic growth in the country. The ECM term is significant and with the expected negative sign, confirming the existence of a long-run relationship between the regressors and economic growth in the country. It also signifies that any previous errors emerging from a disequilibrium in the model will be corrected in the current period. The magnitude of the coefficient of the term suggests that 20.7% of the errors will be corrected.

5.3.1.2: Discussion of the impact of grants and loan

The study’s finding that grants are highly significant and outperform loans in terms of their relative long-run contribution to economic growth is consistent with the evidence in the cross-country studies by Feeny (2007) and Loxley and Sackey (2008). Feeny (2007) finds using a cross-country study of small island states in the Melanesia region that grants have a better impact on economic growth in the region than loans. Loxley and Sackey (2008) find grants to be even more effective on economic growth than total aid. As grants have been found to be important for economic growth in Sierra Leone, it implies the increasing drive of donors towards the giving of aid grants to low-income countries is in the right direction.

5.3.2 Impact of Technical Cooperation Assistance on Economic Growth

From previous results, total aid is found to have a positively significant impact on economic growth in Sierra Leone; yet there have been reports of dissatisfaction from the citizens (from research interviews conducted as part of this study) about too much aid diversion towards technical assistance. Donors interviewed on the other hand respond or tend to argue their case

137 for their increased focused on technical cooperation aid type on the premise that the country lacks the absorptive and specifically the human capacity to implement aid projects and execute technically demanding assignments; and hence employing foreign expatriates should suffice. Therefore, the study attempts to examine the importance of technical assistance for economic growth in the country employing a combination of both ARDL and Johansen estimation approaches.

In the context of examining the effectiveness of technical assistance, the disaggregated aid variable becomes technical cooperation assistance (TCA) and non-technical assistance (NTCA). The empirical model derived from the aid-growth base model then becomes:

LRGDP t = β0 + βtLTCA t + βnt LNTCA t + βiLPI t + βpPolicy t + βiq LIQI t + t …(5.16)

Where TCA denotes technical cooperation assistance and NTCA denotes non-technical cooperation assistance – both expressed as shares of GDP before being logged. As both LTCA and LNTCA variables are I(1), this makes possible the study’s estimation using both ARDL and Johansen techniques of cointegration in an attempt to triangulate the findings.

5.3.2.1: Impact of Technical Cooperation Assistance on Economic Growth Using ARDL Cointegration Table 5.5 below presents the results of the cointegration test and long-run estimates of the specified model using the ARDL approach to cointegration.

138

Table 5.5: Long-run Impact of Technical Cooperation Assistance on Economic Growth: The ARDL Results DEPENDENT VARIABLE IS LOG REAL GDP MODEL A: Tech. Coop Assistance 0.336 (1.488) Non- Tech Coop Assistance 0.135 (1.305) Private Investment 0.253*** (3.050) Property Rights 1.682* (2.029) Macro Policy 0.159 (1.268) Constant 17.206*** (11.950)

F-TEST FOR COINTEGRATION 5.198*** SERIAL CORRELATION 1.23 FUNCTIONAL FORM 0.65 NORMALITY 0.30 HETEROSCEDASTICITY 2.41 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance – inclusive of F-Test for cointegration. T- statistic in parenthesis

Test for Cointegration The test for cointegration reveals the existence of long-run relationship (at 1% level of significance) between economic growth and the regressors; and hence the study is in place to run long-run estimates to examine the impact of technical assistance on economic growth in Sierra Leone.

The Empirical Estimates The long–run estimates as presented in Table 5.5 above show that foreign aid in the form of technical cooperation assistance is not significant in promotion growth in Sierra Leone. The control variable, private investment maintains its importance for growth in the country and it emerges to be highly significant in determining economic growth. Property rights as well shows some evidence of being important for growth, but macroeconomic policy (just as in the aid-growth base model) does not.

The short-run estimates (Appendix 5.8) of the impact of technical assistance merely mimic those from the long-run analysis - that technical assistance is insignificant in determining economic growth in the country. Here macroeconomic policy as a control variable emerges to be highly significant compared to its insignificance in the long-run regression. The ECM term is negative and significant as expected, thus further confirming the existence of a long-run

139 relationship between economic growth and technical assistance. The magnitude of the ECM coefficient suggests that in the event of a disequilibrium, 21% of the errors arising from any disequilibrium in the previous period will be corrected in the current period.

This model is found to largely pass the conditions of model adequacy and hence makes its estimates largely reliable and valid for inference. The model passes the diagnostic test for serial correlation, functional form, normality and heteroscedasticity, which ensures the estimates are reliable for interpretation. Further, the CUSUM and CUSUMSQ tests show that the model is stable over time as it lies within the 5% critical bounds in both plots (Appendix 5.8.1).

5.3.2.2: Impact of Technical Cooperation Assistance on Economic Growth Using Johansen Cointegration Approach In this section, the study uses the Johansen approach to examine the impact of technical assistance on economic growth in Sierra Leone using the same base model as in the ARDL estimation. Table 5.6 below presents the results of the cointegration test and long-run estimates. Though the maximal eigenvalue test could not confirm the existence of cointegration, the Trace statistic test rather confirms the existence of cointegration among the variables. It shows the existence of one cointegrating relationship at the 90% confidence interval. Therefore r=1 is used in the estimation of the long-run relationship.

140

Table 5.6: Impact of Technical Cooperation Assistance on Economic Growth: The Johansen Results DEPENDENT VARIABLE IS LOG REAL GDP (LRGDP) MODEL A: Technical Cooperation Assistance -0.098 (0.136) Non-Technical Cooperation Assistance 0.218** (0.084) Private Investment 0.242*** (0.049) Property Rights 0.620** (0.304) Macro Policy -0.040 (0.044) CONSTANT 18.933*** (0.511) Short Run Results ECM(-1) -0.250*** (0.081)

Var Order (SBC) 1 Cointegration Test R (Maximal Eigen) 0 R (Trace ) 1* R (used in Regression) 1 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. Standard errors in parenthesis. Note that the signs of the long-run estimates have been appropriately adjusted in the table presentation to allow easy comparison with the ARDL estimates.

The long-run estimates reveal that technical assistance is insignificant in determining economic growth in the country; complementing the results of the ARDL approach. Private investment and property rights retain their significance in contributing to economic growth in Sierra Leone.

The adjustment term is negative as expected and highly significant to further confirm the existence of cointegration among the variables. The coefficient of the ECM term shows that in the event of a disequilibrium, 25% of errors arising from such previous shock will be corrected in the current period.

5.3.2.3: Discussion of the impact of technical assistance

Compared with the limited empirical studies on the impact of technical assistance on economic growth, this study’s finding supports the evidence by Ouattara and Strobl (2008), but disagrees with the study by Feeny (2007). Using a panel data analysis for three sets of specifications common in the cross-country aid effectiveness literature, Ouattara and Strobl (2008) could not find technical assistance to foster economic growth; a result which conforms

141 to our finding that technical assistance does not emerge to foster economic growth in Sierra Leone. On the other hand, whilst this study finds technical assistance inflows not to significantly contribute to promoting economic growth in the country, the study by Feeny (2007) from a cross-country analysis of the Melanesian countries finds that technical assistance is significant in promoting economic growth in this region both in the short and long-run.

That technical cooperation assistance is not significant in determining foreign aid in Sierra Leone is a finding worth noting. In the case of Sierra Leone, this is not a result that may surprise the several sceptics of donor technical assistance. Fieldwork interviews conducted with civil society and national elites as part of this study (precisely discussed in chapter 7) clearly indicated the reservations they had about too much donor prioritisation and conditioning of aid to the provision of technical assistance largely in the form of procurement of expatriates. These expatriates are mostly reported in the interviews (including admissions by public officials) to be mainly suggested by the donors themselves or even imposed as part of the aid package against the wish of national authorities or policy experts. This according to the interviews, is kind of waste of donor funds meant for the country on foreign expatriates; a situation the interviews suggest is a ‘chase’ of the aid funds by the donors themselves.

5.3.3 Bilateral Aid versus Multilateral Aid

In another form of aid disaggregation, the study examines the aid-growth relationship by analysing the disaggregates of aid in the form of bilateral and multilateral assistance to determine which source is more effective for influencing economic growth in Sierra Leone. In addition to the need to avoid aggregation bias, two further reasons are advanced for investigating the effectiveness of bilateral and multilateral assistance.

First, bilateral aid especially, is largely considered to be tied, in which sense it may burden the government and give it less freedom to appropriate its aid resources to sectors it considers to be growth generating; and hence the effectiveness of bilateral assistance for growth could be questionable. Commentators, such as Randel et al. (2004) and Radelet (2006) suggest that donor interest which may be in the form of tied aid, adversely affects economic development of the aid-recipient country. Randel et al. (2004) suggest that aid may be given with donor

142 interest in mind instead of the interest of the aid-recipient country, and in effect may adversely affect economic growth in the recipient country since net gains in aid giving goes to the donor instead of the recipient country. Radelet (2006) argues that when portions of bilateral donor aid are tied, it becomes more costly and less effective; as this implies donors will force the recipient country to spend portion of the aid money on the donor’s goods and charge a non-competitive price which becomes more costly for aid recipients. Such tied aid maybe in the forms of compulsory purchases of trade imports, compulsory procurement of technical assistance from the donor country, compulsory export to donor country and so forth. This may be done against world market prices and hence may deny the country of real gains from such resources and may consequently adversely affect the country’s growth. Therefore, in as much as government generally will maximise its utility for attracting more aid irrespective of the source (whether bilateral or multilateral), however, in the context of which source may provide less repayment burden, the type of aid matters. If bilateral aid heavily burdens the government with repayment difficulties due to its tied nature, it may be expected to adversely affect growth or perform less than the supposedly less tied multilateral aid.

On the other hand, multilateral donors like the IFIs had introduced structural adjustment programmes (SAPs) which have been criticised for distorting economic progress (Koeberle, 2003; Riddell, 2007). If this is the case, multilateral aid would not be expected to be quite impressive for economic growth especially given the extensive and intensive conditionalities that come along with them. Yet, donors argue that such conditions are pro-growth conditionalities; implying multilateral aid should improve growth. With all these arguments and counterarguments, it is relevant that the evidence in a typical aid-recipient country like Sierra Leone be sought to contribute in addressing such disagreements.

Yet, another reason for investigating the importance of this aid modality on economic growth is premised on the argument by some commentators (Ram, 2003; Heady, 2005; and Javid and Qayyum, 2011) that bilateral aid is political and in effect gives the government more leverage to defy repayment of aid loans; and hence the tendency for government to divert its use from productive sectors to non-productive uses may be high. While casting the blame of aid ineffectiveness on the donors side, Heady (2007) suggests that a major reason for aid’s ineffectiveness in his cross-country study is due to the large degree to which allocations are biased towards geopolitically important countries wherein aid is used to achieve in what he 143 refers to as ‘non-development outcome, such as geopolitical allegiance’ (ibid: 5). The article argues that in such a situation, the selfish donor is unlikely to demand much accountability from the recipient government with respect to their use of aid resources for fear of weakening political allegiance. And with Myrdal (1968) considering the system of bilateral aid as having a strong tendency towards misallocation in the context of its geopolitical attribute, it implies bilateral aid is ineffective in its geopolitical purpose. Whilst this may be true in clearly geopolitical countries, it is relevant to confirm this argument in an extremely poor country like Sierra Leone to see if bilateral aid is not only geopolitical, but can as well respond to development needs of the aid-recipient country.

Hence, in the succeeding analysis, this study examines the relative impact of bilateral and multilateral aid on economic growth in Sierra Leone, first, using the ARDL approach to cointegration, and second, using the Johansen Approach to cointegration.

In this context, the disaggregated total aid variable becomes bilateral aid (BA) and multilateral aid (MA). The empirical model arising from the aid-growth base model then becomes:

LRGDP t = β0 + βtLBA t + βnt LMA t + βiLPI t + βpPolicy t + βiq LIQI t + t …….(5.17)

Where BA denotes bilateral aid and MA denotes multilateral aid as shares of GDP– both used in logs. As all the variables in this specification are I(1) as shown in Appendix 5.6, it allows the use of both the ARDL and Johansen approaches to carry out the estimations.

5.3.3.1: Impact of Bilateral and Multilateral Aid on Economic Growth Using ARDL Cointegration Approach In this section, the ARDL approach to cointegration is used to test for the existence of long- run relationship among the variables of the model and present the long and short-run estimates of the impact of bilateral and multilateral aid on economic growth in Sierra Leone.

Model A in Table 5.7 below presents the results for the cointegration test and the long-run estimates. The test for cointegration reveals that there exists a long-run relationship between economic growth and bilateral and multilateral aid in Sierra Leone. The F-statistics for cointegration, when compared with the critical values by Narayan (2004) reveals that there is 144 a high level of cointegration between economic growth and the regressors used in the model including bilateral and multilateral aid at the 1% level of significance. Therefore, the study can proceed with the estimation of the long-run relationship between economic growth and bilateral and multilateral aid for Sierra Leone.

Table 5.7: Long-run Impact of Bilateral and Multilateral Aid on Economic Growth: The ARDL Results DEPENDENT VARIABLE IS LOG REAL GDP MODEL A: MODEL B Multilateral Aid 0.070 0.250 (0.696) (1.554) Bilateral Aid 0.315* 0.269* (1.920) (1.705) Private Investment 0.293*** 0.363*** (2.803) (2.952) Governance 1.664* (1.661) Property Rights 1.722* (1.794) Macro Policy 0.130 0.243 (1.082) (1.322) Constant 17.00*** 17.926*** (9.900) (13.725)

F-TEST FOR COINTEGRATION 5.181*** 5.910*** SERIAL CORRELATION 4.81** 1.78 FUNCTIONAL FORM 0.002 0.04 NORMALITY 0.01 4.61 HETEROSCEDASTICITY 4.29** 1.12 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance – inclusive of F-Test for cointegration. T- statistic in parenthesis

The regression estimates as presented in Model A of Table 5.7 above show some evidence of bilateral aid being significantly important for fostering long-run economic growth in Sierra Leone, though only moderately important (at 10% level of significance). Multilateral aid on the other hand is insignificant in determining long-run economic growth in the country. The control variable, private investment is found to have a highly significant impact in promoting growth. Property right, though moderately, also shows some evidence of promoting growth in the country. Macroeconomic policy on the other hand does not prove to have a long-run impact on economic growth.

In model B of Table 5.7, the study re-estimates the empirical model by replacing the property rights index with the governance index to check the robustness of the estimates to changes in specification. The results are very much in consonance with those obtained from the main model estimates: that bilateral aid is (moderately) significant in determining economic growth while multilateral aid is not.

145

In the short-run (Appendix 5.9), neither bilateral nor multilateral aid provides evidence of being positively significant in determining economic growth in the country. The evidence from the main model (Model A) and the robustness check model (Model B) does not show evidence for the existence of the short-run impact of bilateral and multilateral aid. In both models, the control variables of private investment, property rights, and macroeconomic policy prove to be significant in the short-run in determining economic growth. The ECM term in both models is significant and negative as expected to further confirm the existence of cointegration among the variables. In the main model (Model A), the coefficient of the ECM term shows that in the event of a disequilibrium, 18.7% of the previous errors will the corrected in the current period; 19.9% of these errors will be corrected in the robustness check model (Model B).

The main model passes some of the tests for model adequacy. In particular, it passes the tests for functional form of the model and normality of the residuals at the 5% level of significance. It however fails the tests for serial correlation and heteroscedasticity. In model B, all the above four diagnostic tests are passed thus compensating for failure of some tests by the main model. The effect does not seem to matter however, as the estimates remain consistent with both models, showing bilateral aid to have a moderate impact in promoting long-run growth, whilst multilateral aid does not. The CUSUM and CUSUMSQ plots of the main model are presented in Appendix 5.9.1 and shows the model parameters to be stable over time as the plots fit within the 5% critical bounds.

5.3.3.2: Impact of Bilateral and Multilateral Aid on Economic Growth Using Johansen Cointegration approach In this section, the results of the relative impact of bilateral and multilateral aid on economic growth in Sierra Leone are presented using the Johansen cointegration approach. The same models as in the ARDL approach are used, with the intent of triangulating the findings obtained from both approaches. Model A in Table 5.8 below presents the cointegration test results and the empirical estimates from the main model.

The test for cointegration shows that there is a long-run relationship among the variables including bilateral and multilateral aid in the case of Sierra Leone (Table 5.8). Though the maximal eigenvalue test could not confirm cointegration among the variables, yet the trace 146 statistic test does confirm the existence of one cointegrating relationship among the variables at the 95% confidence interval. Hence, the study uses r=1 in the estimation of the long-run estimates for the impact of bilateral and multilateral aid on economic growth in Sierra Leone.

Table 5.8: Long-run Impact of Bilateral and Multilateral Aid on Economic Growth: The Johansen Cointegration Results DEPENDENT VARIABLE IS LOG REAL GDP (LRGDP) MODEL A MODEL B Multilateral Aid -0.053 -0.045 (0.075) (0.077) Bilateral Aid 0.194** 0.169*** (0.070) (0.046) Private Investment 0.303*** 0.289*** (0.054) (0.031) Governance -0.055 (0.326) Property Rights -0.047 (0.293) Macro Policy -0.005 -0.014 (0.028) (0.050) CONSTANT 19.955*** 19.975*** (0.413) (0.351) Short Results ECM -0.225** -0.228** (0.091) (0.101)

Var Order (SBC) 1 1 Cointegration Test

R (Maximal Eigen) 0 2** R (Trace ) 1** 1** R (used in Regression) 1 1 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. Standard errors in parenthesis. Note that the signs of the long-run estimates have been appropriately adjusted in the table presentation to allow easy comparison with the ARDL estimates.

The long-run estimates show that bilateral aid has a significant impact in improving growth in Sierra Leone. On the other hand, multilateral assistance does not emerge to significantly improve growth in the country. Private investment remains highly significant in determining growth in the country.

To examine whether this finding is robust against a change of the specification, another model (Model B in table 5.8 above) is run wherein the property rights score is replaced with governance quality score. Here as well, the results just confirm those from the main model that bilateral assistance is significant in determining economic growth in the country, and multilateral assistance is not.

147

In both models, the ECM term is negative and significant to further ascertain the existence of cointegration among the variables, and ensuring that errors arising from a shock in the previous period disequilibrium are corrected in the current period.

5.3.3.3: Discussion on the relative impact of bilateral and multilateral aid

Following the regression analysis using both the ARDL and Johansen approaches to cointegration, the results are consistent across both approaches and show that bilateral assistance has been more important for economic growth in Sierra Leone than multilateral assistance. Whilst there is some evidence of bilateral aid being significant in contributing to economic growth in the country, multilateral assistance does not. Even with a change of specification, these results are robust, that aid from bilateral donors is a better source of aid for promoting economic growth than multilateral assistance in the case of Sierra Leone.

The insignificance of multilateral aid seems to support the finding of Javid and Qayyum 2011) who also found that multilateral aid is insignificant to long and short-run economic growth in Pakistan. The finding is also consistent with the earlier findings by Gounder (2001; 2002) that multilateral aid is insignificant in promoting economic growth in both Fiji and Solomon Island. However, while Javid and Qayyum (2011) also find bilateral aid to be insignificant to long-run growth (though significant in the short-run), in the case of Sierra Leone the study finds bilateral aid to have a positive and significant impact on economic growth in the long-run though could not be confirmed in the short-run. The results do not also seem to support the argument that because bilateral aid is largely geopolitical (as has been argued by Ram, 2003) it’s therefore ineffective in its development target (Heady, 2007; Javid and Qayyum, 2011). This study’s finding that bilateral aid significantly impacts on long-run economic growth in Sierra Leone contrasts such propositions.

Thus, despite the supposed bias of bilateral aid to geopolitical interest, in the case of Sierra Leone, the study argues for finding a significant impact of bilateral aid on the basis that such aid may have been given to Sierra Leone largely based on country needs instead of geopolitical purposes (as fieldwork interviews with donors also revealed). This is especially so because Sierra Leone remains one of the poorest countries in the world, and hence the developmental purpose of aid may have been applicable in the case of Sierra Leone. Therefore, though bilateral aid is said to be given largely with geopolitical motives in mind,

148 yet in the case of Sierra Leone, country needs may have been the crucial purpose for bilateral aid. Fieldwork interview with DFID representative reveals that DFID as a bilateral donor, largely allocate aid to Sierra Leone in response to the country’s development needs; and this has been similarly stated by other resident aid donors interviewed during fieldwork (e.g. AfDB, The World Bank, EU, IMF and the UNDP).

However, in as much as the study could not find strong evidence to support the geopolitical and tied aid argument of bilateral aid, yet, it does not completely dispute the tied aid argument of bilateral aid being an impediment to its effectiveness on growth. From the results, the impact of bilateral aid, though significant, is found to be moderate in the ARDL results (which is our main approach), but truly significant with the Johansen ML results. As the estimates of the ARDL are found to be more reliable in small samples (Inder, 1993; Banerjee et al., 1993) making such estimates more dependable, it may as well be possible that the moderate impact of bilateral aid on economic growth could have been much more significant had bilateral aid not been largely tied. This study argues that the adverse effect of tied aid is partially true if technical assistance which seems to be imposed or conditioned on aid given by donors particularly bilateral donors (as is largely reported following fieldwork interviews in the case of Sierra Leone) is insignificant in improving economic growth. Following the estimates in the previous aid modality section, technical assistance was found to be insignificant in determining economic growth in the country.

Hence, in as much as the study provides evidence to counter the non-development impact argument of bilateral aid in the case of sierra Leone, it also argues that the tied aid attribute typical of bilateral assistance may have as well adversely affected the level to which bilateral aid could have significantly promoted growth in the country. This is so because technical assistance which is a large portion of bilateral aid in Sierra Leone (and which is usually tied) is found to be insignificant in determining economic growth in the country. The implication of this finding is therefore that the non-developmental impact explanation of bilateral aid that is widely argued in the literature should only be considered with caution as it may be country dependent and only partially evident. As the finding shows that bilateral assistance has some level of importance for economic growth in Sierra Leone, it implies the country should continue to encourage bilateral assistance and utilize it for developmental purposes and not take advantage of its geopolitical purposes to misappropriate it.

149

5.4 Overall Discussion and Conclusion

Foreign aid remains an important source of public expenditure in most developing countries including Sierra Leone. Its impact on development and particularly economic growth has been debatable following the empirical literature. In a country where a quantitative study on aid effectiveness was yet to be conducted, this study only serves to fill this gap by providing the empirical evidence as a basis for policy reformulation and further research. This chapter has provided the evidence on the aid-growth relationship for Sierra Leone. Using the bounds test (otherwise referred to as the ARDL) approach to cointegration by Pesaran and Shin (1999) and the Johansen Maximum Likelihood Approach to cointegration for the period 1970-2007, this study obtained estimates for the impact on economic growth of foreign aid and the aid structure.

The study finds that aid to Sierra Leone is significant in promoting economic growth in the country. A further analysis of the structure of aid revealed that grants perform better than loans in contributing to long-run economic growth, though the opposite is true in the short- run. Technical assistance fails to prove its significance in promoting economic growth in the country. While bilateral aid is found to be significant in the long-run, multilateral aid on the other hand fails to prove its effectiveness for promoting growth in the country both in the long and short-run.

The evidence in the case of Sierra Leone has only provided support for the supplemental theories (as discussed in the theoretical literature) that foreign aid is vital in the promotion of a country’s economic development. Hence, donor intervention in the economy of Sierra Leone has not seemed to be in vain, but has rather proved to be largely useful. It implies that Sierra Leone’s persistent poverty characterisation amidst notable donor presence and participation in the country’s economy has little to do with the fact that foreign aid has not been effective in promoting the country’s economic growth. It may however be that the magnitude of the effect has not been that high to completely eradicate poverty. The aid supporting institutions and policies require much strengthening to increase the magnitude of the impact. In particular, efforts towards a politically stable state and the promotion of democratisation should be pursued as such are crucial elements of political institutions which are vital for aid’s impact in promoting economic growth in the country. Institutions which may be very much relevant in guiding the effectiveness of aid during this conflict periods

150 may be weak and such weakness may not be completely eliminated even in the post-war periods. It is also evident that the promotion of private investment in Sierra Leone is almost as important as the disbursement of foreign aid with respect to promoting economic growth in the country.

The finding that grants are highly significant and generally outperform loans in terms of their impact in promoting long-run economic growth implies the scepticism from donors (e.g. the AfDB) of the country’s fragility to responsibly and effectively manage loans has some truth at least in the case of Sierra Leone. That loans are only moderately significant in the long-run in promoting growth in the country implies that the burden of repayment may have had an adverse effect in reducing the extent to which loans could improve growth in the country and in effect, if aid-loans have to be disbursed to the country it must be highly concessionary to the extent it becomes a near-grant disbursement. It does not necessary imply that loans giving to Sierra Leone should not be encouraged, as there is further evidence that in the short-run, it emerged to be more important than grants in terms of their relative impact on growth. Thus, in the event that the country is in dire need to minimise inflation and fund its budget deficit which may be short-term policy interventions, accessing concessionary loans may be a necessity. In which case, the IMF’s balance of payments support loans could be useful.

That technical assistance is insignificant in determining economic growth in the country implies the considerable donor diversion of aid towards technical assistance (mostly on donor recommended expatriates) should be revisited as it does not show to translate to economic growth both in the short and long-run. The finding that bilateral assistance is somehow effective on economic growth, while multilateral assistance is not, implies that despite its criticism for being non-developmental, in the case of Sierra Leone, bilateral assistance should be encouraged. Thus, if aid’s impact on economic growth is to be promoted in a situation of limited resources, aid in the forms of grants and bilateral assistance should be prioritised (and to some extent, loans) compared to technical assistance and multilateral aid.

On a whole, the econometric analysis showed donor intervention to be useful for promoting economic growth in the country. A point of caution to donors, however, is that following the results of this analysis, the weak evidence for technical assistance, multilateral aid and loans calls for a reconsideration into the manner in which such aid types are disbursed and utilized to achieve their intended purpose. However, the fact that aid has been generally effective on 151 the country’s economic growth should encourage donors to continue in their effort to provide aid to the capital-starved country. Whilst the study may be attributable to the case of Sierra Leone, the applicability of its findings may not be only limited to the country, but generally to any typical aid-dependent poor country that had experienced prolonged political instability. Hence, as the analysis finds aid to significantly contribute to economic growth, it implies for further research into the country’s weak poverty standing amidst increased donor aid effort, there is need to further probe aid’s impact on pro-poor growth as well as its direct impact on human development.

152

CHAPTER 6: IMPACT OF FOREIGN AID ON PRO- POOR GROWTH IN SIERRA LEONE: EMPIRICAL ANALYSIS

As described in the conceptual framework in chapter 4, this study examines the impact of foreign aid on poverty on two dimensions of poverty: pro-poor growth and human welfare. This chapter presents the empirical analyses of the impact of foreign aid on agricultural growth as proxy for pro-poor growth. The justification for using agricultural growth as proxy for pro-poor growth is already explained in the study’s conceptual framework (Chapter 4). Hence, in this chapter, we present the methodology for analysing the impact of foreign aid on pro-poor growth followed by an analysis of the empirical findings. We make further specifications of the models, with the aid variable being expressed in its various forms (i.e. food aid versus non-food aid, grants versus loans and technical assistance versus non- technical assistance) in order to capture which component of aid is most relevant for poverty reduction in terms of agricultural productivity and growth. We then conclude in the final section.

6.1: Methodological Framework and Data

To examine the impact of foreign aid on pro-poor growth in Sierra Leone, we employ a triangulation approach involving a combination of time series cointegration approaches. The purpose of employing triangulation is to ensure much valid results, following possible weakness in any one of the techniques of analysis. This section describes the methodology employed and the empirical results obtained.

6.1.1: The Empirical Model and Data

The theoretical framework follows from a capital-based model just as is done in the aid- growth analysis in chapter 5, and hence posits agricultural output as a function of capital sources and other determinants of agricultural productivity as commonly found in the empirical literature. Formally, this is expressed as:

Y=f(K, Z) ……..(6.1)

153

Where Y is Agricultural output, K is capital, and Z is a vector of other determinants of agricultural output common in the empirical literature.

The above theoretical model motivates our empirical specification derived as follows:

RAGDP t = α + βXt + γZt + t …………….. (6.2)

Where RAGDP is Real Agricultural Gross Domestic Product being a measure of agricultural output, X is a vector of capital sources, and Z is a vector of other agricultural growth determining variables as found in the empirical literature which are crucial for pro-poor

growth, t is the error term, while subscript t, denotes time.

In conformity with the above reduced form model, our empirical model follows from those employed by Odhiambo et al. (2004), Feeny (2007) and Akpokodje and Omojimite (2008).

In addition to the typical neoclassical production function determinants of growth involving capital, labour and land, Odhiambo et al. (2004) in their examination of the determinants of agricultural performance in Kenya suggest trade policy, activities of the government (i.e. government involvement), human capital, and climate as further determinants. Feeny (2007) suggests in his specification for the impact of foreign aid on agricultural GDP growth as proxy for pro-poor growth that aid, macroeconomic policy variables, political instability, real exchange rate depreciation, price shocks, environmental shocks and rural population growth as the determinants of agricultural GDP growth in the Panel of Melanesian states. Likewise, in their estimation of the impact of aid on agricultural growth in Nigeria, Akpokodje and Omojimite (2008) suggest aid, savings, exports and inflation as the determinants of agricultural growth.

Our model for estimation is therefore a modification and adaptation from these empirical models in the sense that it incorporates only relevant variables due to limitation of the length of the time series and the need to have sufficient degrees of freedom.

Assuming crucial capital sources for pro-poor growth to be foreign aid (‘Aid’) and private investment (PI), and other pro-poor growth determinants to be macroeconomic policy 154

(‘Policy’) and political crisis (‘CRISIS’), our empirical base model for estimation is specified as:

LRAGDP t = β0 + β1LAID t + β2 LPI t + β3 POLICY t + β4CRISIS t + t ………………. (6.3)

Where LRAGDP is log of real agricultural component of GDP as proxy for pro-poor growth; AID is the ratio of ODA to GDP; PI is ratio of private investment to GDP. Following M’Amanja and Morrissey (2005); Bhattarai (2009), Gomanee et al. (2005), these sources of capital (AID and PI) are logged in the estimation. POLICY is macroeconomic policy index (comprising of inflation, government expenditure and trade openness), and CRISIS denotes political instability.

Foreign aid is hypothesised to have a significant impact in promoting pro-poor growth. This follows from the aid-growth theory as reviewed in Chapter 2 of this thesis. According to Chenery and Strout (1966), foreign aid is given to augment low savings and investment typical of developing countries as well as augment foreign exchange gap with the ultimate effect of promoting economic growth. As agricultural growth is a large share of economic growth in developing countries, it implies foreign aid should foster agricultural growth in developing countries of which Sierra Leone is one. In fact, bearing in mind that a substantial portion of aid to poor countries is allocated to the agricultural sector on the grounds that it employs the vast majority of the poor, it is expected that foreign aid should impact on agricultural growth in such countries.

Our study included investment (private investment) in the model, which as emphasised by Gomanee et al. (2005) and Feeny (2007), is often excluded in the empirical models by most aid effectiveness studies because of the risk of double counting it may introduce as most investments (especially public) in developing countries are largely funded by foreign aid. Yet it would be also not fair to leave out investment that may not be directly due to aid. Hence, we add private investment to the model as a determinant of agricultural growth. As private investment itself is another form of capital, which is generally prioritised in a production function, it is expected that this form of capital should foster economic growth and hence agricultural growth. However, as agriculture in many developing countries especially in the case of Sierra Leone is largely subsistent, the impact of private capital for agricultural growth may be insignificant as such agricultural practice may not be attractive to

155 the profit oriented investors. Components of private investment may include foreign direct investment as well as domestic credit to the private sector. Hence, it may be also important to assess whether these components individually determine agricultural growth in Sierra Leone.

‘POLICY’ is a vector of policy variables that may be important for growth in the agricultural/rural sector or poverty reduction. Following the Burnside and Dollar (2000) study, macroeconomic policy variables constitutes proxies from monetary policy, trade policy and fiscal policy. Trade policy in particular is crucial for agricultural development as export of agricultural produce remains vital for growth of the sector just as imports of inputs of production and competitive food imports. Trade liberalisation is a crucial attribute of any trade policy. Krueger (1998: 1520) simply defines trade liberalisation as “the action of making a trade regime less restrictive.” Trade restriction by implication, thus generally involves the imposition of high import tariffs and/or export taxes. Oyejide (1986) suggests three ways through which trade protection in the context of import-substitution industrialisation strategy (involving the setting of high import tariffs) can affect the agricultural sector. First, that import tariffs also tax exports and hence hurts the agricultural sector whose growth largely depends on performance of its exports. Second, that a trade policy that protects industry raises the costs of imported agricultural inputs such as machinery and agro-chemical inputs. Third, that protection affects the real exchange rate in a way that it hurts the agricultural sector and diverts resources (including the crucial agricultural resource of labour) away from agriculture, thus damaging productivity of the sector. Trade protection, in the context of imposition of high rate of import tariffs, lowers the real exchange rate below the equilibrium real rate. This leads to a reduction in the domestic prices of traded agricultural goods as against those of the industrial sector thus translating to a reduction in profitability of the agricultural sector compared to the industrial sector – which may warrant the movement of labour away from the agricultural sector. Hence, this goes with the suggestion that agricultural sector is expected to deliver the most significant gains from trade liberalisation (World Bank, 2002); whilst also implying that trade protection can severely hurt the sector.

Analytical frameworks on how trade liberalisation/openness impacts on the rural poor has generally been seen when its lowers imports prices for poor consumers and producers and increases export prices for poor producers; when trade liberalisation that broadly encompasses domestic market liberalisation creates or destroys the markets where the poor 156 participate; when it affects the taxes paid by the poor and government revenue; whether transitional unemployment associated with structural adjustment of trade liberalisation is concentrated on the poor; when it transfers technology and innovation which is particularly beneficial to low skilled farmers in developing countries; and when trade liberalisation improves overall economic growth (Winters, 2002; Berg and Krueger, 2006). Hence, in as much as openness is expected to largely improve agricultural productivity and poverty reduction in general, yet it can as well have an adverse impact on the poor. Trade liberalisation through loss of markets can adversely affect the poor, but can also have a positive effect if new markets (e.g. marketing boards) where the poor participate are created by trade reforms (Winters, 2002).

Though the evidence may appear mixed, yet there seems to be an increasing consensus among researchers that trade policies remain vital for agricultural productivity and with a reasonable number of studies on that already carried out (Odhiambo et al., 2004; Ram, 1985; Tybout, 1992; Havrylyshyn, 1990; Winters et al., 2004). According to Tybout (1992:189), opening up of the economy through trade generally improves the allocation of factors of production across sectors thereby stimulating a one-time increase in the value of domestic production as is espoused in trade models that presume perfect competition. Ram (1985), in his cross country study of the importance of export for economic growth, found that exports performance is vital for economic growth. In a similar vein, and in his review of the literature, Havrylyshyn (1990) concludes that more open trade policies are associated with greater efficiency and productivity gains. Winters (2002) presents evidence of a positive effect of trade liberalisation in Zimbabwe through the creation of competition and the rise in prices and farm incomes (though the opposite effect was evident in the case of Zambia). Also, Winters et al. (2004) report of strong evidence in favour of the positive impact of trade liberalisation on productivity; and further dismisses concerns of an adverse effect on employment and wages of the poor as well as reduced government spending on the poor arising from declining fiscal revenues associated with trade liberalisation, though admit of the existence of these limitation in specific instances.

In terms of the impact of fiscal policies on pro-poor growth, according to Odhiambo et al. (2004), government activities in the general economy and the agricultural sector in particular affect the performance of the sector. Government intervention with regards to consumption expenditure and investments influences agricultural outputs. We therefore consider 157 government consumption expenditure share of GDP (a proxy for fiscal policy following Easterly and Rebelo, 1993) to be an important indicator in the construction of the pro-poor macroeconomic policy index. Government consumption according to Odhiambo et al. (2004) can have a direct and indirect impact on agricultural incomes. Expenditures in the general public services, defence and security, and social security indirectly affect factor productivity and may or may not be positively related to economic productivity and growth. Government spending that can directly affect agricultural productivity include those that are complementary to private investment such as expenditures on health, education, roads, other transport and communication infrastructure, and a range of wider economic services. In the agricultural sector specifically, Odhiambo et al. (2004) continue that expenditures on research, extension and veterinary services, feeder roads, and credit provision are likely to directly affect agricultural performance. These expenditures may positively influence growth in the agricultural sector, especially where the private sector is not willing to invest in agricultural production for which the returns may either not be immediate or be less profitable.

Inflation is also an important policy variable that adversely affects agricultural production (Ruttan, 1979; Johnson, 1980). Increases in food prices may or may not benefit farmers. In countries where a significant majority of farm production is on subsistence basis, it may emerge that rise in food prices may not quite benefit farmers who hardly produce for the market. Rather, increases in prices may adversely affect their production, since inputs of production (fertilizers, chemicals, animal feed, young stock of livestock etc) are mostly imported and purchased, thus constraining accessibility and hence adversely affecting production. This argument is supported by Ruttan (1979) who suggests that inflation dampens productivity growth. Likewise, Johnson (1980) suggests the probability of fluctuating inflation and price uncertainty in having some small negative effects on agricultural output and productivity.

CRISIS is a dummy for periods of the civil war. It is not uncommon that for a stable and productive agricultural sector, political stability both at the central level and in the rural areas should be in place. Feeny (2007) considers political crises as a crucial variable for agricultural output and growth. In the case of Sierra Leone, it is expected that the 11 year civil war should adversely affect agricultural production and hence pro-poor growth. This is so because the rural areas where agricultural production mostly occurs were heavily hit by the 158 conflict and subject to long periods under combat and seizure, thus preventing the smooth flow of capital sources to aid productivity in those areas. Hence, political crisis is expected to significantly negate pro-poor growth in the country.

6.1.1.1: Extended specification For robustness check, the study runs further regressions where we alter the base model specification to examine whether the impact of foreign aid on pro-poor growth remains robust against a change of the specification. Various specifications are considered:

With the use of Trade Policy Index When we use trade policy index in the regression, the model is specified as:

LRAGDP t = β0 + β1LAID t + β2 LPI t + β3 TPOLI t + β4CRISIS t + t ………………. (6.4)

Where TPOLI is trade policy index constructed using principal component analysis comprising of trade openness and real exchange rate as component variables. This variable replaces the macroeconomic policy index in the base model. Following Odhiambo et al. (2004), trade policy is seen to affect productivity through two main avenues. The first is through increased outward trade or openness. Trade openness is postulated to impact on economic growth through “ specialisation and intensification effects, greater economies of scale associated with larger markets, greater capacity utilisation and rapid technological change” (Odhiambo et al., 2004:38). The second avenue through which trade policy is found to affect growth and productivity is through foreign exchange market (ibid: 39). Exchange rate could have either a positive or negative effect on the agricultural sector. A depreciation of the exchange rate (i.e. a fall in the exchange rate) according to Feeny (2007) could stimulate demand for agricultural production and exports thus enhancing agricultural growth. If on the other hand much of the inputs for agricultural production are imported, then devaluation could harm the agricultural sector and hence agricultural growth by making these inputs more expensive. Trade policy is therefore crucial for agricultural development, as exports of agricultural produce remain vital for growth of the sector just as imports of inputs of production and competitive food imports. The PCA construction of the trade policy index is shown in the succeeding section.

159

With the use of overall Trade openness measure In this robustness check specification, we only use the trade openness measure of the entire economy to replace the macro policy index in the base model to test the robustness of the impact of foreign aid on agricultural growth. This has the further advantage of investigating the impact of overall economic trade on agricultural growth in the country. This model is specified as:

LRAGDP t = β0 + β1LAID t + β2 LPI t + β3 LTOPEN t + β4CRISIS t + t ………………. (6.5)

Where LTOPEN is log of the trade openness measure.

With the use of agricultural trade measure In this specification, we specifically employ the agricultural trade instead of the overall trade openness measure. This further has the advantage of verifying the importance of agricultural trade on agricultural growth in the country. This specification is provided in the following:

LRAGDP t = β0 + β1LAID t + β2 LPI t + β3 LATOPEN t + β4CRISIS t + t ………………. (6.6)

Where LATOPEN is agricultural trade used in natural logarithm in the estimation

With the use of domestic credit to the private sector as proxy for private capital sources In this specification, we use domestic credit to the private sector to replace private investment as a source of private capital in order to check whether impact of aid remains significantly unchanged with a change of the base specification. This model is specified as:

LRAGDP t = β0 + β1LAID t + β2 LDCPS t + β3 POLICY t + β4CRISIS t + t ………………. (6.7)

Where DCPS denotes domestic credit to the private sector as a share of GDP, used in natural logarithm in the estimation. It is expected that credit to the private sector should promote agricultural growth though the lending of microfinance to agricultural firms and farms. Banks and microfinance institutions are expected to provide small loans and microfinance to SMEs operating in the agricultural production and marketing. Should this be the case, credit to the private sector should foster agricultural growth. However, as indicated earlier, because of the unattractive nature of the subsistent type agriculture in developing countries and in Sierra Leone especially, it is no guarantee that credit to the private sector will be passed to the largely unprofitable agricultural practice dominant in the country. And if this is the situation, it may emerge that credit to the private sector does not foster agricultural growth in the country.

160

With the use of foreign direct investment as proxy for private capital sources In this final robustness check specification, we use foreign direct investment as proxy for private capital source. This is specified as:

LRAGDP t = β0 + β1LAID t + β2 FDIt + β3 TPOLI t + β4CRISIS t + t ………………. (6.8)

FDI denotes flows of net foreign direct investment as a percent of GDP. This assumes that in typically poor country like Sierra Leone, investment from domestic sources is marginal, which is evident with the dominance of multinational companies in the private sector of the country. Though we had argued earlier for a possibility of private capital not being important or agricultural growth in a predominantly subsistent agricultural system, in Sierra Leone, the impact of FDIs may be different. Foreign direct investment, just as with overall economic growth, is expected to positively affect agricultural growth. This is especially so because the measure of agricultural GDP as computed from the SNA 93, comprises not only crop production and animal husbandry, but further consists of forestry and fishing. Fishing comprises both artisanal and industrial fishing. With Sierra Leone being endowed with large deposits and species of fishery in its large territorial waters, its industrial fishing has been largely private capital driven and hence the flow of FDIs in such may as well imply that this source of private capital may have a positive impact in influencing overall agricultural growth in the country.

6.1.2: Data Description and Construction

The data covers the period 1970-2007 for Sierra Leone, the country case study; and are sourced from the World Bank’s World/African Development Indicators online database.

LRAGDP denotes log of real agricultural GDP as proxy for agricultural growth and hence pro-poor growth (since log difference of agricultural GDP corresponds to agricultural growth – as is used by Kargbo and Adamu 2010; Adhikary 2011, Herzer and Morrissey 2011 in their measure of economic growth). In consonance with the system of national accounting (SNA 1993), Agricultural GDP comprises forestry, hunting and fishing as well as cultivation of crops and production of livestock. Specifically, agricultural value added is used as proxy for agricultural GDP since it is really the value added that constitutes the GDP in its estimation using the systems of national accounting. This agric value added data is sourced from the World Bank’s World/African Development Indicators database. However, as at the time of

161 the study, real agricultural GDP data was not available for the country under study (Sierra Leone), and hence conversion was made from the nominal agricultural GDP using the GDP deflator to arrive at the real agric GDP estimates. The current agric value added (as proxy for current agricultural GDP) and GDP deflator variables were both sourced from the World Banks’ World/African Development Indicators database.

The variable of interest, foreign aid (AID), is defined as net official development assistance as a share of GDP and is sourced from the World Bank’s World/African Development Indicators database.

Private investment (PI) is defined as private investment as share of GDP and is sourced from the world/African development indicators database of the World Bank.

‘Crisis’ (or WAR) is a proxy for political instability and is a dummy variable assuming a value of 1 for periods of political instability (specifically corresponding to the periods of the civil conflict in Sierra Leone) and 0 otherwise.

Domestic credit to the private sector (DCPS) is defined as net domestic credit to the private sector as share of GDP (i.e. its ratio to GDP) and is sourced from the World Bank’s World/African Development indicators.

The variables that constitute the macroeconomic policy index include government expenditure share of GDP (GEX), inflation rate (INF) and trade openness (TOPEN). GEX is government consumption expenditure as share of GDP and is sourced from the World Bank’s World Development Indicators. Inflation rate is annual average inflation rate obtained from the World Bank’s World Developments Indicators database. Trade openness (TOPEN) is the sum of imports and exports as share of GDP (i.e. (imports +exports)/GDP) and is sourced from the World Bank’s World Development indicators.

Agricultural trade is computed as sum of agricultural imports and exports and expressed as share of nominal GDP (i.e. (agric imports + agric exports)/GDP). The agricultural exports and imports, and nominal GDP data are sourced from the World Bank’s World Development Indicators database.

162

Foreign direct investment (FDI) represents net foreign direct investment as percent of GDP and is obtained from the World Bank’s World Development indicators.

RER as a component variable of the trade policy index, denotes real exchange rate and is derived by deflating the nominal exchange rate with the price ratio. Following Alam and Quazi (2003), the computation is detailed as follows:

RER = e(P*/P d)……………(6.9) Where e = the nominal exchange rate (Sierra Leone Currency per US$), P* = US CPI, and P d = Sierra Leone CPI

The nominal exchange rate and the price indices were obtained from the World Bank’s World Development indicators database.

The Macroeconomic Policy index and its Construction

Indicators included in the composite macroeconomic policy index in the pro-poor growth model constitute annual average Inflation rate, trade openness and government consumption expenditure as proportion of GDP; and the construction of this policy index is already discussed in the previous chapter (chapter 5) and shown in Appendix 5.1.

Trade Policy Index and its construction

As discussed earlier, we consider two variables in the construction of our trade policy index: the exchange rate and trade openness variables. The principal component analysis technique was employed to provide a composite trade policy index comprising the real exchange rate and trade openness variables. The variables were first standardised in SPSS as they carry different scales and units of measurement. The results of this PCA show that 73% of the standardised variance is explained by the first principal component with the second component accounting for the remaining percent (Appendix 6.1). This implies we consider the first principal component as an appropriate measure of trade policy as it better explains the variation of the dependent variable better than any other linear combination of the indicators used. The factor scores which provide the weight of the trade openness and exchange rate indicators emerged to be same at 0.585. This score was used to weigh both

163 indicators to arrive at the trade policy index denoted as TPOLI automatically computed by SPSS.

6.1.3 The Overall Estimation Procedure

The examination of the impact of foreign aid on pro-poor growth in Sierra Leone follows the econometric technique of cointegration, which is employed to estimate the long run effect of aid on pro-poor growth; accompanied by an Error Correction Model representation which provides estimates for the short-run and the adjustment term once cointegration is found to exist. The time series period covers from 1970 to 2007. We introduce a triangulation approach involving a combination of two different techniques of cointegration to establish the impact of foreign aid on pro-poor growth in Sierra Leone. In this respect, we employ the autoregressive distributed lag (ARDL) bounds test approach to cointegration by Pesaran and Shin (1999) and the Johansen Maximum likelihood technique of cointegration to estimate the aid-pro-growth relationship. The following briefly describes the cointegration technique of each estimation approach.

6.1.3.1: The ARDL Approach to Cointegration

The ARDL bounds test procedure for examining the impact of foreign aid on pro-poor growth follows that modified by Pesaran and Shin (1999) and Pesaran et al (2001). A detail of the technique and the rational for its use in this estimation is discussed in the previous chapter (chapter 5). In tune with the ARDL approach, model (6.3) is specified in a VECM as:

p LRAGDP t = β0 + δ1LRAGDP t-1+ δ2LAID t-1+ δ3LPIt-1+ δ4POLICY t-1+∑ φiLRAGDP t-i + i=1

q q q ∑ ωjLAID t-j+ ∑ lLPIt-l+ ∑ γmPOLICY t-m + ψCRISIS t + εt ……(6.10) j=1 l=1 m=1

Where δi are the long-run multipliers, β0 is the drift and εt are white noise errors. Following these specifications, the test for cointegration is carried out using the F-test. The F. statistic is generated in Microfit 4.0 and compared against the critical values generated by Narayan (2004) which is particularly suited to small samples compared to those generated by Pesaran and Pesaran (1997) which were originally derived with a sample of 500 observations. The critical values generated by Narayan (2004) are applicable to samples in the range 30-80

164 observations within which our sample size fits. An F-stat value higher than the upper bound confirms the existence of cointegration; while an F-stat lower than the lower bound confirms the non-existence of cointegration. When the F-stat falls within the lower and upper bounds, then the test becomes inconclusive. The hypothesis for testing the existence of cointegration among the agricultural growth determinants is such that the:

Null hypothesis for instance in vecm1 of the base model, Ho is:

Ho: δ1= δ2= δ3= δ4= 0

Is tested against the alternative hypothesis H A

HA: δ1≠ δ2≠ δ3≠ δ4≠ 0 Once cointegration is established, the next step involves the estimation of the long run and short run relationships. The long-run model to be estimated is as follows:

q3 p q1 q2 LRAGDP t = β0 + ∑ δ1LRAGDP t-1+ ∑ δ2LAID t-1+ ∑ δ3LPIt-1+ ∑ δ4POLICY t-1 + i=1 i=0 i=0 i=0

ψCRISIS t +εt ….(6.11) The SIC selects the appropriate orders of the ARDL model as such information criteria is described as more parsimonious with the lag length selection and is a consistent model selection criteria; and hence more reliable in small samples (Peasaran and Shin 1999). The specifications above represent only the long-run equilibrium state of agricultural growth determinants. However, for policy purposes, the short-run adjustment of agricultural growth to changes in its determinants is necessary. To account for the speed of adjustment, a dynamic error correction model is further specified. However, while this is only for the sake of simplicity, with the ARDL approach, both long-run and short-run estimates are captured in a single equation specification of the ARDL model.

The specification for the short-run dynamic parameters estimated by the error correction model is specified as follows:

p q q q LRAGDP t = + ∑ φiLRAGDP t-i+ ∑ ωjLAID t-j+ ∑ lLPIt-l+ ∑ γmPOLICY t-m+ i=0 j=1 l=1 m=1

ψCRISIS t+ζecm t-1+εt ……(6.12) Where φ, ω, , and γ are the short-run dynamic coefficients of the model’s convergence to equilibrium, and ζ is the speed of adjustment

165

The ECM term in the error correction representation also confirms the existence of cointegration/long-run relationship among the variables when this term is negative and statistically significant.

This long run and short run estimation is of most importance in the technique, as it is on the basis of this that inference can be drawn as to whether foreign aid has an impact on agricultural growth and hence pro-poor growth. Further models that this study runs to check for robustness of the results follow similar ARDL specification as above.

6.1.3.2: The Johansen Maximum Likelihood Approach to Cointegration

In addition to the ARDL cointegration approach, we use another technique of cointegration, the Johansen Maximum Likelihood approach to examine the aid-pro-poor growth relationship for Sierra Leone with the aim of triangulating findings obtained from the ARDL approach. We employ Microfit 4.0 by Pesaran and Pesaran (1997) just as with the ARDL analysis, to estimate the long-run relationship between foreign aid and economic growth. A description of this approach, its merits, and specification is already presented in the previous chapter.

6.1.4: Unit Root Tests

As justified in the previous chapter, both approaches to estimating the long-run aid-growth relationship requires the testing for the presence of unit roots in the variables used in the regressions. With the ARDL Approach, there is the need to verify that none of the variables are integrated of I(2) and also that the dependent variable is I(1). For the Johansen approach, there is the need to verify with unit roots tests that all the variables used in the regression are I(1) before a conduct of long-run estimation can be acceptable. Hence in this section, we conduct unit root tests using the Augmented Dickey Fuller (ADF) technique with an intercept but no trend.

The results of the unit root tests for all the variables used in the aid-growth regressions ran in this chapter are presented in Table 6.1 below. The ADF unit root test results show that all the variables employed in any regression run here are either all I(1) or a mixture of both I(0) and I(1). With this results, we are at best to use the ARDL approach to cointegration whose

166 technique produces robust long-run and short-run estimates whether the regressors used in the regressions are either all I(0) or I(1) or a mixture of both. As the variables in the base model are all I(1), we can as well do cointegration analysis using the Johansen approach for this model allowing us to triangulate findings obtained from the ARDL approach. Further, two of the models we additionally run to check for robustness have all their variables integrated of order one, thus allowing us to do Johansen cointegration estimation.

Table 6.1: ADF Unit Roots test – including constant but without trend: 1970-2007 Levels 1st Difference Variable SIC Lag ADF t-stat. SIC Lag ADF t-stat. I(d) LRAGDP 0 -1.650035 0 -6.385132 I(1) LAID 0 -1.736667 0 -7.227074 I(1) LPI 0 -2.445824 0 -6.924393 I(1) POLICY 1 -2.174771 0 -4.762549 I(1) CRISIS/WAR 0 -1.555815 0 -5.830952 I(1) Additional Variables TPOLI 0 -2.461589 0 -6.596917 I(1) LDCPS 0 -1.864518 0 -7.604095 I(1) FDI 0 -5.754089 0 I(0) LATOPEN 0 -3.013694 0 I(0) LTOPEN 0 -3.286945 0 I(0)

Critical Value 1% 2 -3.632900 0 -3.626784 5% -2.948404 -2.945842 Critical Value 1% 1 -3.626784 5% -2.945842 Critical Value 1% 0 -3.621023 5% -2.943427

In the next part of the chapter, we present the empirical results for both estimation procedures employed for analysing the relationship between foreign aid and pro-poor growth.

6.2: The Aid-Pro-Poor Growth Empirical Results

In this section, we present the empirical estimates for the impact of foreign aid on pro-poor growth in Sierra Leone. We first present the cointegration tests, the diagnostic test, and the long-run and short-run estimates for the impact of foreign aid on pro-poor growth for the main model using the ARDL approach. We then present the results of the further regressions done to check for the robustness of the aid-pro-poor growth estimates to changes in the specification using the ARDL approach. Following this, we present the cointegration test results and the long-run estimates of the aid –pro-poor growth relationship using the Johansen ML approach; and further test for robustness of the findings with two further regressions

167 where we slightly alter the base specification. We then discuss the overall aid-growth results and conclude.

6.2.1: Impact of aid on Pro-Poor Growth: The ARDL Results

Diagnostic and Cointegration Tests The test for cointegration shows that there exists a long-run relationship between foreign aid and pro-poor growth in Sierra Leone. A comparison of the estimated cointegration F-statistic with the F-test critical values generated by Narayan (2004) confirms the existence of cointegration in the model though at the moderate 10% level of significance (Model 1 in Table 6.2). With the confirmation of a cointegrating relationship, we can proceed with the conduct of the long-run estimation.

Table 6.2: ARDL Long run Estimates for Impact of Aid on Pro-poor Growth Dependent Variable is Growth in Agricultural GDP (LRAGDP) Model 1: Model 2 Model 3 Model 4: Model 5: Model 6: Foreign Aid 0.254*** 0.251*** 0.245*** 0.266*** 0.207*** 0.269*** (5.294) (4.978) (5.535) (4.940) (3.781) (6.704) Private Investment -0.055 -0.063 -0.032 -0.082 - - (0.708) (0.790) (0.416) (-0.993) Foreign Direct - - - - - 0.016** Investment (2.132) Domestic Credit to the - - - - -0.231 - Private Sector (1.462) Macro Policy -0.018 - - - -0.012 -0.071* (0.424) (0.293) (1.929) Trade Policy - -0.008 - - - (0.178) Trade Openness - - -0.178 - - - (0.989) Agricultural Trade - - - -0.151 - - (0.874) Constant 19.414*** 19.443*** 20.081*** 19.792*** 19.772*** 19.235*** (95.852) (97.122) (31.195) (52.702) (56.298) (204.920) Crisis -0.424*** -0.443*** -0.402*** -0.515*** -0.456*** -0.352*** (2.810) (2.938) (2.873) (2.971) (3.471) (4.115) Cointegration test (F- 3.605* 3.697* 3.639* 4.143** 3.251 + 3.494* Test) Serial Correlation 2.06 2.28 0.43 0.47 1.55 0.04 Functional Form 0.12 0.30 0.06 0.21 0.001 0.01 Normality 1.36 1.23 1.43 1.62 1.00 1.08 Heteroscedasticity 0.93 0.83 1.69 1.74 0.96 2.85 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T-stats in bracket. + Inconclusive when compared with critical values by Narayan (2004), but cointegrated following the negatively significant ECM.

168

The Long-Run Empirical Estimates The long-run estimates of the relationship between foreign aid and pro-poor growth is presented in Model 1 in table 6.2 above. The results show that foreign aid does have a highly significant impact on pro-poor-growth in Sierra Leone. And the direction of this relationship is positive, signifying that foreign aid fosters pro-poor growth in Sierra Leone. The ARDL estimates reveal that for every 1% increase in aid/GDP, pro-poor growth will be improved by 0.25%. In this base specification, private investment and macroeconomic policy as control variables prove to be insignificant in their contribution to pro-poor growth in the country. The political instability variable reveals a highly adverse impact on pro-poor growth in the country, which is expected as the brutal civil conflict had seriously thwarted rural agricultural activities in the country. For every year there is political crisis of the type experienced by the country during the 11 year civil conflict, pro-poor growth will be reduced by 42%. Hence the significance of having a stable political situation is crucial if agricultural growth and hence pro-poor growth is to be improved in the country.

This model is seen to be adequate enough to rely on its estimates and the conclusions reached. It passed all the diagnostic tests for serial correlation, functional form, normality and heteroscedasticity. The model also passed the Brown et al (1975) test for stability of the parameter estimates over time as the CUSUM and CUSUMSQ plots shown in figure 6.1 below reveal the model to fit within the 5% critical bounds.

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

169

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Figure 6.1: CUSUM and CUSUMSQ Plots for the Base Model on Aid-Pro-Poor Growth

Hence, we find the estimates from this model to be reasonably reliable for valid conclusions with regards the relationship between foreign aid and pro-poor growth in Sierra Leone.

Short-run estimates In the short-run however, the findings of the long-run estimation do not appear to be consistent with those of the short-run emerging from the error correction representation (Model 1 in Table 6.2.1 below). In the short-run, foreign aid does not emerge to foster pro- poor growth. It in fact emerges to moderately negate pro-poor growth in the country. Even the one year lag foreign aid appears to negate pro-poor growth in the short-run. However, consistent with the long-run estimates is the results that private investment and macroeconomic policy as control variables both emerged to be insignificant in influencing pro-poor growth in the country. Further, the impact of political crisis, just as in the long-run, also shows in the short-run to highly adversely affect pro-poor growth. This further stresses the importance of a stable political atmosphere for an enhanced agricultural sector performance.

170

Table 6.2.1: ARDL Error Correction representation (Short-run Estimates) of the impact of aid on Pro- poor Growth Dependent Variable is Growth in Agricultural GDP (dLRAGDP) Regressor Model 1: Model 2 Model 3 Model 4: Model 5: Model 6: First lag of Real Agric GDP - - - -0.266* - - (1.813) Foreign Aid -0.082* -0.085* -0.091** -0.100** -0.079* -0.019 (1.763) (1.887) (2.102) (2.242) (1.760) (0.361) First Lag of Foreign Aid -0.097** -0.092** -0.110** -0.133*** -0.102** -0.096** (2.189) (2.157) (2.576) (2.875) (2.427) (2.368) Private Investment 0.021 -0.023 -0.012 -0.027 - - (0.726) (0.798) (0.420) (0.998) Foreign Direct Investment - - - - - 0.007** (2.144) Domestic Credit to the Private - - - - -0.091 - Sector (1.539) Macro Policy -0.007 - - - -0.005 -0.031* (0.403) (0.283) (1.654) Trade Policy - -0.003 - - - - (0.176) Trade Openness - - 0.047 - - - (0.677) Agricultural Trade - - - -0.050 - - (0.871) Constant 7.355*** 7.045*** 7.682*** 6.573*** 7.739*** 8.337*** (3.622) (3.818) (3.922) (3.659) (3.900) (4.265) Crisis -0.160*** -0.161*** -0.154*** -0.171*** -0.178*** -0.153*** (3.097) (3.081) (3.063) (3.420) (3.847) (4.192) Ecm(-1) -0.379*** -0.362*** -0.383*** -0.332*** -0.391*** -0.433*** (3.578) (3.783) (3.986) (3.602) (3.807) (4.248) R2 0.57 0.57 0.62 0.62 0.60 0.63 Adjusted R 2 0.47 0.47 0.50 0.51 0.50 0.53 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T-stats in bracket

The error correction mechanism (ECM) term is negative and highly significant as expected which further confirms the existence of cointegration. The coefficient of this term shows that should there be a shock in the system, 37.9% of the previous disequilibrium errors will be corrected in the current period.

6.2.1.1. Robustness check Regression Results For the purpose of robustness check of our finding on the impact of foreign aid on pro-poor growth, we estimate further regressions which we specified previously to verify the consistency of our finding across slight changes in the specification of the empirical model. Models 2-6 in Table 6.2 present the long–run estimates of these robustness check specifications using the ARDL cointegration analysis. In model 2, we replace the macroeconomic policy variable with the trade policy index; in model 3, we use trade openness as a policy variable only; in model 4, we rather use agricultural trade variable

171 instead of overall economic trade; in model 5, we use domestic credit to the private sector to proxy private capital flows; and in the model 6, we use foreign direct investment to proxy private capital flows. In all of these specifications, cointegration was found to exist among the variables thus warranting furtherance with the conduct of the long-run estimations to verify the impact of aid on pro-poor growth across the specifications. In Model 5, a comparison of the cointegration F-statistic with the critical values by Narayan (2004) suggests an inconclusive test as the F-stat lies between the upper and lower bounds of the critical values. However, when this statistic is compared with the critical values by Pesaran and Pesaran (1997), the existence of cointegration at 10% level of significance is confirmed. Further, the ECM term as presented in Model 5 of Table 6.2.1 further confirms the existence of cointegration as this term comes out be negative and highly significant. Hence, all these robustness check specification confirm the presence of cointegration thus allowing us to proceed with the long-run estimation.

In all of those situations, the impact of foreign aid came out to be consistent with the main model (Model 1in Table 6.2) that aid has a highly significant long-run impact in determining pro-poor growth in Sierra Leone. The adverse impact of political instability on agricultural growth in the country is quite evident as well in these specifications. Tests for the adequacy and reliability of these specifications for the robustness check showed that estimates can be relied upon. They all pass the relevant diagnostic tests for serial correlation, functional form of the model, normality of the residuals and heteroscedasticity associated with the model.

In the short-run (Models 2-6 in Table 6.2.1), the impact of aid on pro-poor growth is consistent with that from the main model (Model 1 in Table 6.2.1) that foreign aid does not promote pro-poor growth in the short-run, and that aid in fact tends to moderately negate pro- poor growth in the short-run. The only exception is in Model 6, where, though the direction of the aid-pro-poor growth relationship is negative, it does not emerge to be significant. That which is consistent across all these specifications however, is that foreign aid does not positively affect pro-poor growth in short-run as is evident in the long-run. The significantly adverse impact of political crisis on pro-poor growth is consistent in all these models and so is the estimate of the ECM term which emerged to be negative and highly significant to further confirm the existence of long-run relationship among the variables.

172

6.2.2: Impact of Foreign Aid on Pro-Poor Growth: The Johansen Results

In this section, we estimate the aid-pro-poor growth relationship using the Johansen ML approach with the same base model as we used in the ARDL estimation. In addition, we run two further specifications to test for robustness of the aid-pro-poor growth relationship. We limit ourselves to two robustness check specification here because these are the only two further specifications (as compared to the five in the ARDL) for which all the variables in the regressions are I(1), which is a requirement for long-estimation using the Johansen ML approach.

VAR Order and Cointegration Tests

In Model 1 of Table 6.3 below, the var order of the base regression reveals that the SBC chooses Var order 1, which is consistent with recommendation by Ahmed (2003), and Dutta and Ahmed (2004) that for annual data, a lag length of 1 is the standard practice commonly employed in cointegration estimation. With this var order, the cointegration results for both the trace statistic and the maximal eigenvalue statistic show that there is one cointegrating relationship among the variables of the model at the 95% confidence interval. Therefore, we use r=1 in the estimation of the long-run relationship between pro-poor growth and its determinants including foreign aid, the variable of interest. With the existence of cointegration confirmed, in the next sub-section we estimate the long-run relationship and analyse the results.

173

Table 6.3: Long run Estimates for Impact of Aid on Pro-poor Growth using Johansen Approach Dependent Variable Log of Agricultural GDP (LRAGDP) Model 1: Model 2 Model 5: Foreign Aid 0.281*** 0.296*** 0.139 (0.0549) (0.054) (0.129) Private Investment -0.2504* -0.280** - (0.1381) (0.134) Domestic Credit to the Private - - -0.740 Sector (0.533) Macro Policy 0.0202 - 0.124 (0.058) (0.112) Trade Policy - -0.009 - (0.050) Constant 19.751*** 19.735*** 20.737*** (0.315) (0.282) (1.132) Crisis -0.176*** -0.174*** -0.132*** (0.043) (0.046) (0.044) VAR ORDER (SBC) 1 1 1 Cointegration test (F-Test) R (MAXIMAL EIGEN VALUE) 1** 1** 1** R (TRACE STAT) 1** 1** 2** R USED IN REGRESSION 1 1 1 ECM -0.231 -0.227 -0.159 (0.051)*** (0.055)*** (0.048)*** Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. Standard errors in parenthesis

The Empirical Estimates

Model 1 of Table 6.3 presents the long-run results for the impact of foreign aid on pro-poor growth using the Johansen ML approach to cointegration from the base specification. The results show that foreign aid has a highly significant impact in fostering pro-poor growth in Sierra Leone. For every 1% increase in foreign aid, pro-poor growth increases by 0.28%. The control variables, private investment and macroeconomic policy do not tend to foster pro- poor growth in the country. In fact, private investment tends to negate pro-poor growth though only moderately. Political instability negatively impacts on the agricultural growth at a high level of significance. Hence, the Johansen long-run estimates are largely consistent with our ARDL estimates of the impact of foreign aid on pro-poor growth in Sierra Leone.

The adjustment term estimated in the error correction representation of the model emerges to be negative and highly significant to further confirm the existence of cointegration among the variables of the model. The magnitude of its coefficient shows that, should there be a shock in the system, 23.1% of disequilibrium errors in the previous period will be corrected in the current period.

174

6.2.2.1: Further Robustness Regressions

Here, we present the empirical results of the further specifications we run to confirm the robustness of the aid-pro-poor growth relationship following our estimation of the main model. Models 2 and 3 in Table 6.3 above, present these estimates. In both models, the existence of at least one cointegrating relationship among the variables is confirmed by both the Trace statistic and the Maximal Eigenvalue statistic thus allowing us to estimate the long- run relationship. In model 2, we replace the macroeconomic variable with the trade policy index; and with a var order of 1, both the Trace statistic and the Maximal Eigenvalue statistic reveal the existence of one cointegrating relationship at the 95% confidence level. Therefore, we use r=1 in the long-run estimation of the model. In model 3, where we use domestic credit to the private sector to proxy private capital sources, with a var order of 1 chosen by SBC, the Trace statistic reveals one cointegrating relationship among the variables at the 95% confidence interval whilst the Maximal Eigenvalue statistic reveals the existence of two cointegrating relationships at the 95% confidence level. With this variation in the two Johansen cointegration test statistics, we use r=1 which is also chosen by the SBC.

In model 2, the long-run estimates reveal that foreign aid remains highly significant in determining pro-poor growth in the country, just as with the main model. In model 3, however, this long-run impact of aid is not confirmed. In both models, the ECM term obtained from the error correction representation of the model emerges to be negative and highly significant to further confirm the existence of cointegration among the variables used in each specification.

6.2.3: Discussion of the Aid-Pro-Poor Growth Results

Our finding that foreign aid has a significant impact in promoting pro-poor growth supports the studies by Akpokodje and Omojimite (2008) and Feeny and Ouattara (2009) that find foreign aid to significantly contribute to promoting agricultural growth. Akpokodje and Omojimite (2008) find in their study that aid promotes agricultural growth in Nigeria, and likewise Feeny and Ouattara (2009) show aid to contribute to agricultural growth in a cross country study. The significance of aid in promoting agricultural growth was however not strong in the Feeny and Ouattara (2009) study as the significance was only found to occur at the 10% level of significance. The finding however contradicts the study by Feeny (2007) in

175 his cross-country study of the impact of foreign aid to Melanesian countries. Feeny (2007) found that though foreign aid was significant in promoting general economic growth, it was however insignificant in promoting pro-poor growth in the group of South Pacific countries; a finding that does not entirely mirror our country study of the impact of foreign aid on pro- poor growth in Sierra Leone. This contradiction of findings just further throws light on the need to conduct individual country studies as generalising findings across countries may not bring out the country differences; hence the essence of our country study.

6.3: Aid Modality and Pro-poor Growth

The aforementioned analysis revealed that foreign aid has a significant impact in fostering pro-poor growth in Sierra Leone. While this only represents total aid, it is relevant to avoid aggregation bias and test the disaggregates of aid that are relevant to poverty reduction and examine whether they have a significant impact on pro-poor growth in the country. The argument against aggregation bias and further reasons to disaggregate total aid into its several categories, have already been explained in the conceptual framework and in the previous chapter. The study thus sought to analyse whether the aid structure varied for aid’s impact on pro-poor growth. We used three aid structure types: food aid versus non-food aid, grant versus loan, and technical assistance versus non-technical assistance to examine which aid modality is better off in terms of their relative impact on pro-poor growth over the study period in Sierra Leone.

As some of these disaggregates appear to be I(0) (see Appendix 5.6), we cannot use the Johansen approach to estimate their impact on pro-poor growth. Hence our analysis is entirely based on the ARDL approach which is applicable whether the regressors are I(0) or I(1) or a mixture of both.

6.3.1: Impact of Food Aid on Pro-Poor Growth

Food aid, as a form of foreign aid has got varied forms or purposes. It can be in the form of food for work, as supplementary feeding initiatives and emergency relief, and in its largest form, it is provided to recipient governments to be sold to support their budgets (Maxwell and Singer, 1979). Barrett and Maxwell (2005) reclassify them as project, emergency and

176 programme food aid respectively. Programme food aid (or budget support food aid), according to Barrett and Maxwell (2005) involves direct government-to-government transaction typically for sale in domestic markets of the recipient country. This form of food aid normally goes with policy conditionalities involving putting the proceeds from the sale of food aid into counterpart funds (which may as well involve nutritional and agricultural sector support), and changing macroeconomic, trade or agricultural policies. Project food aid is usually provided to recipient governments, a multilateral development agency (e.g. WFP) or NGOs operating in the recipient country for use in development projects. Aid in the forms of food for work and for school feeding initiatives and supplementary feeding centres for children and mother are all forms of project food aid (Barrett and Maxwell, 2005). These two types of food aid (programme and project aid) are sometimes referred to as developmental food aid (Barrett and Maxwell, 2005: 14). Emergency food aid is the form of food aid usually disbursed in response to disasters.

As all of these forms of food aid are not directly related to agricultural growth, which we use to proxy pro-poor growth, it implies the impact of food aid in agricultural growth is not as straight forward as the impact of other types of aid. With respect to these forms of food aid, the impact of ‘food for work’, if work is on agricultural production or feeder road construction, is expected to positively impact on agricultural growth. Barrett and Maxwell (2005) suggest that if food aid in the form of ‘food for work’ is effectively strategised, it has the potential to be a considerable source for irrigation development and rural roads construction; which we argue should ultimately foster agricultural production and marketing. In fact, there have been reports that food aid as an input and in the form of food for work to support agricultural production has actually improved local production (Cathie, 1982: 74-75). Food aid in the form of supplementary feeding programme, we argue does not foster agricultural growth. If at all there is any relation, supplementary feeding programmes may negate agricultural growth as farmers are not motivated to produce since demand for similar domestically cultivated feeding products may not be enhanced by the availability of freely imported feeding products.

The impact of food aid in the form of budget support may have both a positive and negative impact on agricultural growth. If part of the proceeds from the food aid sales is used to fund agricultural budget, then this form of food aid is well expected to positively impact on agricultural growth. On the other hand, food aid, if sold in the open domestic market may 177 well have a disincentive effect through food prices as is widely argued in the literature (Schultz, 1960; Maxwell and Singer, 1979; Maxwell, 1986a; 1986b; 1986c; 1991; Isenman and Singer, 1993). According to the disincentive argument, the sale of food aid in the domestic market will provide competition for domestically produced food produce thus serving as a demotivation for agricultural production. In addition to the price effect argument for food aid, there is also a policy effect argument for food aid, which also falls under the rubric of the disincentive argument of food aid. Isenman and Singer (1993) suggest a significant disincentive effect on the overall agricultural policies of the recipient country as a result of inflows of food aid. They argue that there will be a total neglect of the agricultural sector in favour of other economic sectors in as much as food aid continues to flow into the recipient country. The recipient government is not encouraged to formulate and implement policies geared towards local agricultural production as there is less pressure of food unavailability locally, following the inflows of food aid. Further, by attracting labour to food for work site, Maxwell and Singer (1979) suggest that food aid may deny the agricultural sector the labour that should be otherwise available for farm production. This implies food aid will adversely affect agricultural growth. However, in the context of output, food aid is argued to lift the constraint on growth and self-reliance by providing the real resources required to expand investment and dampen inflation (Maxwell and Singer, 1979); though much of this evidence has only been narrative and inconclusive (OECD, 2006). Further, in an attempt to dismiss the disincentive effect of food aid through the price mechanism, the proponents of food aid argue that the price effect of food aid may not well be reflected on the farmers but rather on the traders (Dantwala, 1963; Dandekar, 1965; and Maxwell and Singer, 1979). We find this counterargument not strong enough however, because even assuming the price effect to directly affect only the traders of farm produce, this will eventually trickle down to the farmers as non-profit demotivation to the traders will imply less demand for the farm produce from the farmers and hence consequently demotivates farmers to produce more.

Following the aforementioned pros and cons of food aid, in the case of Sierra Leone, we posit that in the context of agricultural growth, food aid should have a moderately positive effect. Government influence on agricultural sector in Sierra Leone, just as in any other developing country appears important and so food aid in the form of budget support may as well be substantially used to support agricultural budget, which itself is growth enhancing. A further reason for expecting a positive impact of food aid on agricultural growth follows its use as food for work on agricultural production or feeder roads which enhance marketing of 178 agricultural produce. In spite of this pro-food aid justification, there are adverse effects which combined makes the positive impact of food aid on agricultural growth limited to a moderate positive impact. The attraction of labour from agricultural farms to non-agriculturally based food for work sites limits the growth impact that food aid may have had on agricultural growth. This cannot be ruled out in the case of Sierra Leone, where recent programmes to employ youths have partly involved food and cash for work (Walton, 2010) and with the agricultural sector devoid of the required work force to drive significant growth in the sector. Further, the adverse price effect of food aid supplies on local agricultural produce market also provides a limiting factor to the expected significant impact of food aid on agricultural growth in Serra Leone. Food aid sales may provide price competition for locally produced food crops and livestock whose proceeds farmers may use to provide for other needs and to expand next season production. However, in the case of Sierra Leone, in as much as this cannot be completely disputed, the largely subsistent nature of agriculture in the country implies the inflation dampening effect of food aid may better hold an argument thus making the food aid adverse price effect less significant as food production is largely for own consumption rather than for sales and profit.

In effect, instead of strongly fostering agricultural growth, we posit in this study that following the few possible adverse effects, food aid should have a positively significant impact in promoting agricultural growth, but only moderately. We hence, provide in the following section, the empirical test for the impact of food aid on pro-poor growth in Sierra Leone.

We use the aid-pro-poor growth specification where we used trade policy and disaggregate the aid variable into food aid (FA) and aid that is not due to Food aid (NFA). We use this model because it produced cointegrated estimates and acceptable diagnostic tests. The model is specified as:

LRAGDP t = β0 + β1LFA t + β2LNFA t +β3 LPIt + β4 TPOLI t + β5CRISIS t + t ……………(6.13)

Where FA is food aid flows (sourced from the OECD) and NFA denotes non-food aid and is total net ODA less food aid. Both variables are first converted to shares of GDP before being logged.

179

Unit roots and Cointegration Test The unit root test (Appendix 5.6) of both LFA and LNFA variables reveal that LFA is an I(0) variable while LNFA is I(1). This implies we are at best to use the ARDL bounds test approach to do the long-run estimation, a technique which is applicable when all the variables of the model are I(1), I(0) or a mixture of both. Since the unit roots test further shows that none of the variables used in the model are I(2), we can use the ARDL approach to estimate the long-run relationship between food aid and pro-poor growth in the country.

Model 1 in Table 6.4 below provides the cointegration test results and the long-run estimates using the ARDL approach to cointegration. The cointegration test reveals that there exists a long-run relationship between food aid and pro-poor growth for Sierra Leone at the 10% level of significance when the cointegration F-statistic is compared against the critical values generated by Narayan (2004). Therefore, with the confirmation of cointegration, we estimate the long-run relationship between food aid and pro-poor growth in Sierra Leone.

Table 6.4: Long run Estimates for Impact of Food Aid on Pro-poor Growth using ARDL Approach Dependent Variable is Growth in log of Real Agricultural GDP (LRAGDP) Model 1: Model 2: Food Aid 0.094* 0.081* (1.939) (1.723) Non-Food Aid 0.189*** 0.203*** (3.974) (4.353) Private Investment -0.032 -0.035 (0.465) (0.492) Trade Policy 0.016 - (0.410) Agricultural Trade - -0.033 (0.226) Constant 19.566*** 19.594*** (114.651) (66.592) Crisis -0.384*** -0.386*** (3.065) (3.125)

Cointegration test (F-Test) 3.714* 3.468* Serial Correlation 1.98 2.12 Functional Form 0.03 0.0001 Normality 1.33 1.16 Heteroscedasticity 0.22 0.21 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance.

The Empirical Estimates The long-run estimates as revealed in Model 1 of Table 6.4 above show that food aid moderately fosters pro-poor growth in Sierra Leone as its impact occurs at the 10% level of significance. The results further reveal that aid in the form of non-food aid has a highly

180 significant impact on pro-poor growth. This implies that the high level of impact of total aid in fostering pro-poor growth in Sierra Leone is more highly attributable to aid that is non- food aid.

In the short-run (Model 1 of Appendix 6.2), neither food aid nor non-food aid has a significant impact in promoting pro-poor growth. This result reflects the earlier results that total aid does not foster pro-poor growth in the country in the short-run. The ECM term is negative as expected and highly significant to further confirm the existence of long-run relationship among the variables of the model. The coefficient of the ECM reveals that should there be a shock in the system, 42.1% of the disequilibrium errors will be corrected.

These results are largely reliable for inference and for conclusion to be considered valid. The model passes a number of diagnostic tests including the test for serial correlation, functional form of the model, normality of the residual, and heteroscedasticity associated with the model. The test of stability of the parameter estimates as shown by the CUSUM and CUSUMSQ plots in Appendix 6.2.1 shows that the estimates are stable over time within the 5% critical bounds

Robustness Check In addition to the above results, we run a further specification where we use the agricultural trade openness variable to account for policy influence in an attempt to test the robustness of our food-aid-pro-poor growth estimates across a change of specification. This specification (as shown in Model 2 of Table 6.4) is found to show the existence of cointegration among the variables and hence we can do long-run analysis.

The long–run estimates shown in Model 2 of Table 6.4 reveals that food aid moderately promotes pro-poor growth in the country; and that non-food aid significantly fosters pro-poor growth at the high level of significance. The result just mimics those obtained from the main model, and hence the finding that food aid only moderately fosters pro-poor growth is robust across specifications. The short-run result of this robustness check model (Model 2 of Appendix 6.2) is also found to be consistent with the finding of the main model that in the short-run food aid is not significant in determining pro-poor growth just as non-food aid or total is. The ECM term in this specification is negative as expected and highly significant to

181 further confirm the existence of a long-run relationship among the variables and shows that 42.9% of any disequilibrium errors are corrected.

Hence, in as much as total aid impacts on pro-poor growth in the long-run, that impact is largely as a results on non-food aid, as food aid only fosters pro-poor growth at the moderate level of significance. However, it is important to note that in as much as the impact could only be moderate, food aid itself does not emerge to adversely affect agricultural growth in the country.

6.3.2: Technical assistance and pro-poor growth

In another attempt to capture the importance of the structure of aid on pro-poor growth, we disaggregate foreign aid into technical assistance and non-technical assistance. The aim here is to investigate whether technical assistance has a significant impact in fostering pro-poor growth in the case of Sierra Leone.

Technical assistance is intuitively expected to positively impact on agricultural growth, if it adequately and efficiently finances agricultural research, training of farmers and extension services generally. However, in practical terms technical assistance may as well be insignificant for agricultural growth as Bachman (1965:1083) argues that rapid transfer of technology is difficult to adapt in agriculture compared to non-farm enterprises; and this has its bearing in the level of variations in specific efficient production requirements in agriculture. Further, Bachman (1965) argues that technological transfer in agriculture requires more adaptation to the physical environment in which they operate compared to the non-farm sector, in effect making the positive impact of technical assistance to agriculture limited.

Technical assistance in the forms of imparting technological transfer and technical capacity is also made difficult in the case of agriculture in developing countries because of the very low educational levels of farmers, with the majority of them being illiterate. This is also true for the case of Sierra Leone where rural literacy and hence agrarian community literacy is at its very low level (World Bank, 2009 estimates the country’s adult literacy rate for 2004 at 35%, and could be lower for rural settlers), so that technological transfer and capacity building associated with any technical assistance programme may have a very minimal impact if any on agricultural growth.

182

Yet, Bachman (1965:1084) also argues that the lack of research emphasis in US technical assistance programmes makes it unsuccessful in positively impacting on agricultural growth in recipient countries. He posits that “research by the social scientists is needed to provide sharper focus on both the strategic technical changes and the critical social and economic institutions necessary for rapid increase in agricultural production” (ibid:1084). Likewise, it is also analogous that research by the physical scientists is required to develop local technology that can adapt to local physical environment. Hence technical assistance targeted towards funding of agricultural research should enhance agricultural growth.

Further, the low education and technical capacity of famers and farms imply there is a large absorptive capacity gap that any substantial and efficiently disbursed technical assistance can significantly translate into agricultural growth and hence pro-poor growth. Thus, we argue that under very low technical ability of farmers and low levels of technology in agricultural production, with an adequately well designed and targeted technical assistance programme, such aid modality should significantly foster agricultural growth; otherwise its impact becomes insignificant.

In our investigation of the impact of technical assistance on pro-poor growth in Sierra Leone, we use the model with trade policy index as this model emerged to show model adequacy with respect to the diagnostic and cointegration tests when the total aid variable is disaggregated into technical and non-technical assistance. The empirical model is specified below:

LRAGDP t = β0 + β1LTCA t + β2LNTCA t +β3 LPIt + β4 TPOLI t + β5CRISIS t + t ………. (6.14)

TCA is technical cooperation assistance as share of GDP and NTCA is total aid less technical cooperation assistance as a share of GDP. Both variables are logged in their use in the estimation.

Unit Roots and Cointegration Tests The unit roots test for the aid disaggregates as shown in Appendix 5.6, reveals that both LTCA and LNTCA are I(1) making all the variables in the above specified model being I(1). With this stationarity test result, it is applicable to use the ARDL approach to cointegration to

183 estimate our long-run relationship between technical assistance and pro-poor growth. In Model 1 of Table 6.5 which displays the result, the cointegration test shows that there exists a long–run relationship among the variables.

Table 6.5: Long run Estimates for Impact of Technical Assistance on Pro-poor Growth Dependent Variable is Growth in log of Real Agricultural GDP (LRAGDP) Model 1 Model 2 Model 3: Technical Cooperation Assistance 0.018 0.117 -0.011 (0.132) (0.871) (0.087) Non-Technical Cooperation Assistance 0.190** 0.151** 0.221*** (2.438) (2.127) (2.870) Private Investment -0.071 -0.057 -0.084 (0.888) (0.763) (1.135) Trade Policy 0.003 - - (0.073) Trade Openness - 0.113 - (0.499) Agricultural Trade - - -0.211 (1.275) Constant 19.721*** 19.174*** 20.096*** (96.423) (23.534) (54.691) Crisis -0.501*** -0.379*** -0.502*** (2.818) (2.774) (3.033)

Cointegration test (F-Test) 4.100** 3.892* 5.658*** Serial Correlation 2.77 1.80 3.41 Functional Form 1.36 0.26 0.14 Normality 1.89 0.54 2.23 Heteroscedasticity 3.05 0.95 1.30 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. T statistic in bracket

The Empirical Estimates The long-run estimates (Model 1 in Table 6.5) show that technical assistance is insignificant in determining pro-poor growth in Sierra Leone. Non-technical assistance aid rather has a positive and significant impact on pro-poor growth. This implies the significant impact of total aid on pro-poor growth has been largely due to aid in the form of non-technical assistance.

The model is found to be largely adequate for inference as it passes the vital diagnostic tests for serial correlation, functional form, normality and heteroscedasticity at the 5% level of significance (Model 1 in Table 6.5). It also reveals with the CUSUM and CUSUMSQ plots (Appendix 6.3.1) that the model fits within the critical bounds to indicate its stability over time. Therefore, the findings of this model will be largely relied upon for much valid

184 conclusions of the impact of technical assistance on agricultural growth in Sierra Leone; which emerges to be insignificant on such growth.

Short-run estimates also reflect the findings of the long-run analysis that technical assistance is not significant in determining pro-poor growth in Sierra Leone (Model 1 in Appendix 6.3). Further, aid in the form of non-technical assistance emerges to negate pro-poor growth which reflects our finding that total aid tends to be negatively related with pro-poor growth in the short-run. Consistent however, is that the short-run estimates mimic the results from the long- run estimates that technical assistance has not promoted pro-poor growth in the country. The ECM term is negative and highly significant, further confirming the existence of cointegration between technical assistance and agricultural growth in the country. The coefficient of the ECM term which signifies the speed of adjustment of the model to shocks reveals that should such shocks occur, about 32.9% of the disequilibrium errors will be corrected in the current period.

Robustness Check In an attempt to test the robustness of impact of technical assistance on pro-poor growth, we further run two specifications: one in which trade openness is used to replace the policy variable (model 2 in Table 6.5), and the second where we use the agric trade ratio instead (Model 3 in Table 6.5). In both specifications, the cointegration tests reveal the existence of long-run relationship between pro-poor growth and its determinants. In model 2, cointegration is found to exist at the 10% level of significance, where as in Model 3, cointegration occurs at the 1% level of significance when the cointegration F-statistic is compared with the critical values by Narayan (2004). In both specifications, long-run estimates are consistent with those from the main model (model 1 in Table 6.5) that technical assistance is not significant in fostering pro-poor growth in Sierra Leone. Rather, aid in the form non-technical assistance proves to be significant in promoting long-run pro-poor growth in the country.

In the short-run, the results are also found to be consistent with the long-run estimates and the short-run estimates of the main model that technical assistance in not important in fostering pro-poor growth in Sierra Leone (Appendix 6.3). The ECM terms in both specifications are negative and significant as expected to further confirm the existence of long-run relationship between pro-poor growth and its regressors in both models. The coefficient of the ECM terms 185 which denotes the speed of adjustment of the models to disequilibrium reveals that in Model 2, should there be a shock in the system, 36.5% of the disequilibrium errors will be corrected in the current period, and in Model 3, 34.2% of the disequilibrium errors will be corrected.

These models pass the vital diagnostic tests for serial correlation, normality, functional form and heteroscedasticity. Thus, both models are adequate to reliably provide robustness check to the finding of the main model which emerges to be largely consistent.

6.3.3: Impact of Grants and Loans and Pro-Poor Growth

We next examine the relative impact of grant versus loans on pro-poor growth. We use the model with trade openness and disaggregate the aid model into grants and loans in order to assess the impact of grants on pro-poor growth in Sierra Leone. We use this model because of its adequacy to diagnostic tests and test for cointegration. Below is the specified model.

LRAGDP t = β0 + β1LGRANT t + β2LOAN t +β3 LPIt + β4 LTOPEN t + β5CRISIS t + t ……. (6.15)

LGRANT is grant as a share of GDP and logged in the regression. Loan is loan as share of GDP; this variable is not logged here because it has some negative observations.

Unit Roots and Cointegration Tests Unit roots test results shown in Appendix 5.6 show that LGRANT is I(1) while LOAN is I(0), which implies we are at best to use the ARDL approach to cointegration. The cointegration test results are presented in Model 1 in Table 6.6 below. The cointegration test results reveal that the test is inconclusive as it lies between the upper and lowers bounds of the critical F- values. However, the significance and negative sign of the ECM (Model 1 in Appendix 6.4) confirms the existence of a long-run relationship between pro-poor growth and grants and loans.

186

Table 6.6: Long run Estimates for Impact of Grants and loans on Pro-poor Growth using ARDL Approach Dependent Variable is Growth in log of Real Agricultural GDP (LRAGDP) Model 1 Model 2 Grants 0.228*** 0.210*** (3.981) (3.713) Loans -0.003 - (0.129) Lloan - 0.252 (1.601) Private Investment -0.047 -0.094 (0.403) (0.780) Trade Openness 0.039 -0.106 (0.162) (0.421) Constant 19.386*** 19.668*** (23.196) (22.579) Crisis -0.375 -0.687** (1.363) (2.264)

Cointegration test (F-Test) 2.540++ 3.460* Serial Correlation 0.96 0.20 Functional Form 0.002 0.11 Normality 1.88 0.23 Heteroscedasticity 0.60 1.43 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance. ++ Inconclusive but confirmed to be cointegrated with a significantly negative ECM term

The empirical estimates The long-run estimates reveal that grants perform better than loans in terms of promoting pro- poor growth in Sierra Leone. Grants have a highly positive impact on pro-poor growth, whilst loans on the other hand have an insignificant impact on pro-poor growth. For every 1% increase in grants disbursed to Sierra Leone, agricultural growth improves by 0.2%. Thus, the highly significant impact of total aid on pro-poor growth may have been due to the highly performing grants and not loans. The implication is that for aid to maintain its significant impact in determining agricultural growth, more of grants and less of loans should be disbursed to the agricultural sector. The control variables, private investment and trade openness are found to be insignificant in promoting agricultural and hence pro-poor growth in the country.

With the short-run estimates (Model 1 in Appendix 6.4), both grants and loans were not important for promoting pro-poor growth. In fact, loans tend to moderately negate agricultural growth in the short-run. Thus, as opposed to the findings of the long-run analysis, in the short-run, aid in the form of grants does not tend to promote agricultural growth just as loans; which just reflect the short-run finding that total aid does not positively promote agricultural growth in the country. The ECM term has the expected negative sign and is

187 significant, further confirming the existence of long-run relationship between the covariates in this model and agricultural growth, but also indicating that any previous disequilibrium in the model will be corrected in the current period. The magnitude of the ECM term shows that 31.3% of disequilibrium errors are corrected in the current period should there be a shock in the system.

The model passes the test of goodness of fit to ensure the estimates obtained are reliable enough for inference. It passes the crucial diagnostic tests (Model 1 Table 6.6 above) for serial correlation, functional form, normality and heteroscedasticity. It also passes the Brown et al (1975) test for stability of the model over time as the CUSUM and CUSUMSQ plots show the model to fit within the 5% critical bounds (Appendix 6.4.1). Hence, we find the estimates emerging from this model largely reliable for inference.

Robustness Check Robustness check for the relative impact of grants and loans on pro-poor growth was further made with a specification where we use log of loan. Here, as the loans variable has some negative observations, we add to the loans observations a constant value that is higher than the most negative loan observation in the data. That is log of loans is equivalent to 3.5 plus the loan value, to see whether using a natural log of loans as we did with grants will affect our finding.

LLoan= log (3.5 + Loan) …..(6.16)

The result of this regression is shown in Model 2 of Table 6.6 above. The cointegration test result shows that there exists a long-run relationship between pro-poor growth and the regressors of this model at the 10% level of significance. The long-run estimates emerge to be consistent with those obtained from our main model (Model 1in Table 6.6) that grants have a highly significant impact in fostering pro-poor growth, while loans do not. The estimates show that for every 1% increase in grants disbursed to the country, pro-poor growth will improve by 0.2%. Hence, grants outperform loans with respect to their relative impact in promoting pro-poor growth in the country.

The short-run estimates of this robustness check specification (Model 2 in Appendix 6.4) is consistent with our long-run findings; that grants have a significant impact in promoting pro- 188 poor growth whereas loans do not. However, this result is not consistent with our short-run estimates of our main model (Model 1 in Appendix 6.4) which had shown that both grants and loans don’t positively impact on pro-poor growth in the short-run. The ECM term of this specification is negative and significant to further confirm the existence of a long-run relationship. The coefficient of this ECM term reveals that 27.6% of any disequilibrium errors in the previous period will be corrected in the current period.

We find the model to be reasonable reliable to infer from its estimates. The model passes the diagnostic tests for serial correlation, functional form, normality, and heteroscedasticity. These results are hence found to be robust in that the significant long-run impact of total aid on pro-poor growth in Sierra Leone may have been largely as a result of the significant effect of grants on pro-poor growth. If aid in form of grants has been the main modality of disbursement to the agricultural sector compared to loans, then it’s no surprise to find total aid to positively impact on agricultural growth in the country.

6.3.4: Conclusion on the aid modality results

In conclusion, the empirical results on aid modality and pro-poor growth reveal that in the long-run, grants have a highly significant impact in fostering pro-poor growth in the country, and food aid has a moderately significant impact in promoting pro-poor growth. On the other hand, technical assistance and loans do not come out to foster pro-poor growth in the country. Hence for aid to maintain a significant impact in promoting pro-poor growth in the country, it should be delivered largely in the form of grants.

6.4: General Discussions and Conclusion

In Sierra Leone, just as in many other least developed countries, the vast majority of the poor obtain their livelihood through agricultural production and marketing. Hence, in this study, we consider growth in agricultural GDP to proxy pro-poor growth. And as the ultimate aim of development assistance is to reduce poverty that appears imminent in most developing countries, we find it vital to examine the impact of such development aid on pro-poor growth. Using a triangulation of approaches to cointegration involving the ARDL technique of cointegration by Pesaran and Shin (1999) and the Maximum Likelihood approach by

189

Johansen (1988), we investigated the impact of aid on pro-poor growth for the case of Sierra Leone, where aid flows have particularly been disbursed to reduce poverty in a seemingly poverty stricken country. We also extended our investigation to further probe whether the structure of aid matters for aid’s impact of pro-poor growth using the ARDL approach to cointegration.

The results show that generally, total aid flows to Sierra Leone from 1970-2007 have had a significant long-run impact in promoting pro-poor growth in the country. Even when the specifications were altered in an attempt to establish whether our findings were robust, the results emerged to be consistent with the estimates from the original model that foreign aid has fostered pro-poor growth in the country. These results were also found to be largely consistent with the use of both estimation approaches. In the short-run however, the impact of foreign aid in promoting pro-poor growth could not be confirmed. Further examination of the importance of the aid structure on pro-poor growth revealed that though aggregate total aid does impact on pro-poor growth in the long-run, yet aid disbursed in the forms of loans was found to be insignificant in fostering pro-poor growth as opposed to aid disbursed in the form of grants. Against the backdrop of a growing donor prioritisation of technical assistance, this was not found to positively promote pro-poor growth in the case of Sierra Leone. Food aid, though not highly significant in determining pro-poor growth in the country, emerged to moderately promote agricultural growth and hence pro-poor growth in the country.

These findings appear to provide support for some sections of the literature, but also dispute some other sections. In agreement with intuitional expectation, foreign aid is found to have a positive impact on pro-poor growth in Sierra Leone; a result which supports the finding by Akpokodje and Omojimite (2008) for Nigeria, and Feeny and Ouattara (2009) in a cross- country study; where they find foreign aid to significantly contribute to promoting agricultural growth as proxy for pro-poor growth. Our finding however disputes that from Feeny (2007) in a cross-country study of Melanesian countries where his study found that though foreign aid generally promotes overall economic growth, yet it emerged to be insignificant in promoting pro-poor growth. As we could not find evidence in the literature, the other results we present on the structure of aid in this study could not be compared with the literature, which points at the novelty of these areas of research in the aid effectiveness research.

190

The implications of these findings are worth noting for policy makers in the country of study, but also for donors and researchers with interest in developing countries in general. That foreign aid does promote pro-poor growth in Sierra Leone implies that aid strategies and policy conditions designed to enhance aid’s impact on pro-poor growth appear to be in place. Hence aid disbursement strategies, and policies and institutions geared towards enhancing aid effectiveness should be largely continued or strengthened as those have emerged to be vital for aid impact on long-run pro-poor growth in the country. The short-run insignificance of aid on pro-poor growth should however be of concern as this immediate effect can hurt farmers who clearly form the majority of the poor.

Further, the result that grants matter more for aid’s impact on pro-poor growth than loans only provides support for the stance of donors and the international community that a fragile state of a typical developing country does not guarantee its ability to effectively handle loans, is true at least in the case of Sierra Leone where loans are not found to foster pro-poor growth. In fact, it also implies that total aid’s significant impact on pro-poor growth may have been largely due to the impressive performance of grants which emerged to be highly significant in promoting pro-poor growth in the country. It may be that the repayment requirement and burden associated with loans may not have been beneficial to the agricultural sector which employs the vast majority of the rural poor in the country.

That aid in the form of technical assistance does not positively contribute to pro-poor growth in a least developed country like Sierra Leone implies that donors should revisit their growing tendency to prioritise technical assistance as that has not benefited pro-poor growth in the country. Probably, it may not be necessary to halt technical assistance, but revisit the way it is provided in the country. Fieldwork interviews conducted as part of this study revealed that donors spend too much of aid money on expatriates and recruited local consultants, which sort of serve as a disincentive for the excessive number of civil servants and national policy experts who largely operate on merely less than subsistent salaries. The resulting effect of this disincentive may be an inefficient human capital that fails to contribute to pro-poor growth in the country. It may also point to the fact that technical assistance may not have funded the promotion of research in agricultural productivity and economics as has been argued by Bachman (1965).

191

The finding that food aid has a moderately significant impact in fostering agricultural growth and hence pro-poor growth in the country is worth noting. It implies, the disincentive effect argument of food aid on agricultural growth, at least in the case of Sierra Leone, is not entirely true. This implies food aid may have been delivered in the country amidst food insecurity and inadequacy and hence has not deterred agricultural growth due to any competitive attribute it may have thought to introduce. It further implies some proportion of the proceeds from government sale of food aid may have been utilized to finance agricultural budgets. In the short-run, food aid is not significant in determining pro-poor growth, a result that reflects the non-significance of total aid on pro-poor growth in the country.

Overall, the findings show that foreign aid significantly promotes overall economic growth and hence pro-poor growth in the long-run. And for aid to better impact in promoting pro- poor growth, grants instead of loans, and less of technical assistance should be disbursed. In the short-run however, the impact of foreign aid has not been significant.

192

CHAPTER 7: IMPACT OF FOREIGN AID ON WELFARE : AFRICA AND SIERRA LEONE 7.0 Introduction

In the previous chapter (chapter 6), we investigated the impact of foreign aid on poverty using pro-poor growth as proxy for poverty reduction. In this chapter, we further look at another strand of poverty reduction to assess the impact of foreign aid on poverty in Sierra Leone. In consonance with the conceptual framework discussed in chapter 4, this second strand of poverty is aggregate welfare. We therefore investigate the impact of foreign aid on welfare in Africa, but more so for Sierra Leone. The indicators of aggregate welfare as already discussed in the conceptual framework constitute human development index (HDI) and infant mortality rate (IMR). Particular focus is paid on human development as this indicator is more inclusive of the crucial welfare elements of health and education as well as national income. Gomanee et al. (2005a: 364) conclude “There is no strong reason to favour HDI or infant mortality as an indicator of welfare, although the former has the merit of a somewhat broader base.” Because the HDI is much broader in its inclusion of much welfare related measures and also because human development is the most commonly employed indicator of human welfare in the literature, we restrict our further analysis (involving aid disaggregation and the importance of politics) in the aid-welfare relationship to the human development measure. In this chapter therefore, we present the analytical methodology and the results of our investigation of the aid-welfare relationship and the importance of politics in this relationship before we conclude in the final section.

7.1. Methodology 7.1.0 Introduction

In our methodological approach for examining the impact of foreign aid on welfare in Africa and Sierra Leone, we first present the model for our empirical analysis and then describe the data and their sources. We then present the technique of estimation for addressing the research question of assessing the impact of foreign aid on human well-being in Africa and Sierra Leone in particular.

193

In analysing the impact of foreign aid on welfare, an econometric analysis is employed. This implies that the hypothesis that foreign aid improves human welfare is tested using econometric analysis. However, because aggregate welfare data are not sufficiently available on an annual basis for Sierra Leone, a time series regression for Sierra Leone has not been possible. Therefore, a cross country panel regression of African countries covering the period 1980-2007 on the impact of foreign aid on welfare of the poor is estimated with Sierra Leone being singled out to consider the impact of foreign aid on human welfare in the country. Singling out the country effect in the cross-country regression involves taking Sierra Leone as a dummy in the regression and interacting this country dummy with the aid variable to capture the impact of foreign aid on the indicator of human welfare for Sierra Leone. That is, the Sierra Leone dummy is interacted with the aid variable to form an aid*DSL interaction term. The significance of the coefficient of this variable is then tested to establish the impact of foreign aid on the human welfare indicator in Sierra Leone using cross-country analysis. The use of this method may not have been the first in development/economics research but to our knowledge has been the first in the aid effectiveness literature. Rodrik and Subramanian (2004) used this method (of cross-country analysis) to investigate the triggers to India’s impressive economic performance. Likewise, Davoodi and Grigorian (2007) used this method to investigate institutional quality as a determinant of tax collection and potential for Armenia. This study will hence apply this method of interactive country dummy by interacting the Sierra Leone country dummy with the aid variable to establish the impact of aid on aggregate welfare in Sierra Leone, the country under study.

In a further form of the analysis in examining the impact of aid on well being for Sierra Leone, the study additionally use sub-Saharan Africa (SSA) sample of countries (in addition to the full African sample) to examine whether the impact of aid on human welfare in Sierra Leone remains significantly unchanged by changing the sample of countries. This is especially so because Sierra Leone being a very poor country, using a sample of sub-Saharan Africa for its interaction would seem more realistic as Sierra Leone shares much similar economic structural and poverty characteristics with sub-Saharan African countries.

7.1.1: The Empirical Model

The empirical model follows from an adaptation of that by Gomanee et al. (2005a; 2005b) in their estimation of the impact of foreign aid on aggregate welfare for a panel of developing

194 countries. The choice of their analytical framework is mainly based on the fact that their model is not only parsimonious but includes more realistic poverty/welfare oriented variables compared to other empirical models used in the literature. Generally, there remains no theoretical relationship on the determinants of human welfare; and this is attested by Asiama and Quartey (2009) who admitted that theory on aid and welfare is still in its early stages. However, in accordance with the empirical studies on aid and welfare (and particularly those by Gomanee et al., 2005a; 2005b), the determinants of aggregate welfare levels are found to be initial income per capita (GDPPC_1 it ), Pro-poor public expenditure (PPE), military

expenditures (G m,it ) and foreign aid (AID it ). The basic model of the relationship between foreign aid and human welfare is therefore derived as:

Wit = b 0 + b 1GDPPC_1 it + b 2PPE it + G m,it + b 3AID it + e it ; - - - - (7.1)

Where W is the welfare measure; GDPPC_1 is the Initial income measure; PPE it is the pro- poor government expenditure (whose construction is described in Appendix 7.1), G m,it

represents military expenditures, AID it is the foreign aid measure and e it is the error term This model shows that aid impacts on welfare levels directly by establishing the significance of the coefficient of the aid variable.

However, the literature (Gomanee and Morrissey, 2002; Mosley et al., 2004; Gomanee et al., 2005b; and Wolf, 2007) further argues that aid may as well impact on welfare through the indirect means of pro-poor public expenditure. The argument is that large proportions of government expenditures on welfare are largely financed by foreign aid. According to Wolf (2007), in developing countries, particularly low-income countries, a large part of the public expenditure on social services (education, health, and water and sanitation) is funded by foreign aid. In estimating the poverty impact of aid, Gomanee and Morrissey (2002) further suggested that one should focus on factors that are conducive to growth even if the objective is to alleviate poverty rather than growth (quoting Dollar and Kraay, 2001); and also concentrating on public expenditure (measure of poverty expenditure) as a transmission mechanism. The latter involves the provision of public goods such as education, health and social services. Therefore, it is important to assess whether aid significantly contributes to pro-poor government expenditures via recipient government budget. And if aid significantly contributes to pro-poor government expenditure, which in turn significantly contributes to welfare improvement, it implies aid indirectly fosters welfare levels. But perhaps more

195 important in assessing the impact of aid on pro-poor public expenditure has to do with the need to avoid double counting. As foreign aid is already one of the determinants of welfare in the above regression model, it is important to check whether foreign aid is in fact not also a significant portion of pro-poor public expenditure, which is another determinant added in the model. Gomanee et al. (2005a) allows for this problem of double counting by generating the

∧ value for PPE (now denoted as PPE it ) such that it includes only PPE that is not funded by aid – through the method of generated regressors. However, if donor support on poverty (including welfare) is largely via direct project support and NGOs rather than budget support, it may be expected that aid does not significantly contribute to pro-poor government expenditure and hence its contribution to welfare may be limited to its direct impact. It is common to find that not all of donor aid support to these sectors is normally transmitted through government budgets – some go through direct donor-managed projects (with separate project implementation units that rather use donor systems instead of country systems) and via NGOs.

Hence, in our analysis, we initially estimated the impact of foreign aid on pro-poor government expenditure using the sample of African countries and also Sub-Saharan countries. Following Gomanee et al. (2005a), the impact of aid on PPE is related in the following form:

PPE it = a 0 + a 1GDPPC it + a 2AID it + a 3Grit + u it ………. (7.2)

Where PPE is pro-public government expenditure expressed as a proportion of GDP; and G r is other sources of government revenue. The regression results presented in Appendix 7.2 show that aid does not significantly impact on pro-poor public expenditure in the panel samples for both the full African sample and the SSA sample. With this result, we justified this study’s inclusion of both the aid variable and the PPE index as determinants of welfare levels in the estimation, since the issue of double counting has not been crucial. Hence, this study’s PPE measure is not a generated regressor PPE.

Further considering the fact that other factors such as governance/institutional quality (Wolf, 2007; McGillivray and Noorbakhsh, 2007) and military expenditures (Gomanee et al., 2005a)

196 may affect human welfare, these variables are added to the specification to give our empirical equation for estimation which becomes:

Wit = d 0 + d 1GDPPC_1 + d 2PPE it + d 3Gm;it + d 4AID it + d 5IQU + d 6DSL + d 7AID*DSL + e it ….(7.3) Where IQU is the measure of governance, whose construction is described in chapter 5 of the thesis. Gm is military expenditure as a proportion of GDP; GDPPC_1 is initial GDP. PPE is the unweighted pro-poor public expenditure and ‘W’ is the welfare measure which can be either HDI or IMR. DSL denotes Sierra Leone dummy, which is further interacted with the aid variable (as AID*DSL) to form an interactive country dummy that captures the impact of aid on welfare with Sierra Leone as the country of interest.

Our addition of the measure of governance to the original empirical model of Gomanee et al. (2005a) is relevant as the measure of institutional quality/governance has been suggested by Wolf (2007) to be important in improving human welfare. Wolf (2007) suggests that the efficiency of services is very much determined by the different uses to which funds are put such as financing unproductive expenditures and construction of flashy physical structures; and that the allocation of resources in turn is dependent upon the quality of governance; implying that increases in public expenditure are likely to result to increases in public service delivery provided the right institutions are in place to ensure resources are used efficiently. Hence, improving on the quality of institutions can improve effective public service delivery which consequently improves human welfare; thus justifying the addition of the quality of governance variable in our welfare model specification.

Aid is expected to improve aggregate human welfare as is explained in the conceptual framework (chapter 4) by targeting expenditures in those areas in which the poor are most likely to benefit, including social expenditures. According to Gomanee et al. (2005b: 303) “….aid may be more effective in increasing welfare in poorer countries if the marginal effectiveness of aid in improving welfare is greatest where welfare is lowest”. According to Verschoor and Kalwij (2006), aid independently increases the (absolute) income elasticity of poverty – i.e. partly through government budgets. This is so because aid can fund government social expenditure budgets and also when aid financed budgets fund NGOs engaged in activities that target poverty reduction. Such aid expenditures can add to the welfare of the poor even if it does not contribute to growth. But perhaps more importantly, foreign aid can directly fund social expenditures without necessarily going through government budgets. In 197 such a way, aid can directly improve human aggregate welfare which is associated with improved well-being of the poor, particularly in a country like Sierra Leone where the vast majority of the population are poor. Hence, if aid is found to be related with higher levels of human welfare, we translate it as aid being associated with improved well-being of the poor, and hence lower poverty levels. A significant impact of aid in improving aggregate welfare would hence imply that the well-being of the poor would have been worse off if there were no foreign aid.

Robustness Check In order to check whether the estimates we obtain from examining the impact of foreign aid on human welfare are robust, we further run the models with the addition of year dummies; in effect, altering the model specification. According to Bond et al. (2001: 16), the addition of time dummies is equivalent to transforming the variables into deviations from time means. They hence argue that any random pattern in the time means is consistent with a constant mean of the transformed series for each country. Likewise, in their panel analysis of the impact of globalisation on welfare expenditures in OECD countries, Garret and Mitchell (2001) included year dummies in their regression, which they suggest to control for temporal time effects. Further, Roodman (2006) suggests that it is usual to include time dummies in GMM estimations. In our analysis therefore, year dummies are included into the regressions (but not reported) to control for year specific effects as is justified by Hjelm (2002) and these regressions serve as check for robustness of our main aid-welfare estimates, which are run without year dummies since the study operates on a small sample size.

7.1.2: Data Description and Sources

The estimation period covers 1980-2007 and each period comprises five year averages with the exception of that for 2005 which covers the average of 2005 to 2007. The use of five year averages to constitute a period (year) in the panel is common in the cross country aid effectiveness regressions, which is usually done in order to avoid the problem of modelling cyclical dynamics in aggregate welfare levels. In this study, the need to use period averages is further necessitated by the existence of annually missing observations in most of welfare and public expenditure related variables in the panel. This means time periods in the model include 1980, 1985, 1990, 1995, 2000, and 2005. The data covers a sample of 31 African countries which are found to have data for the variables required for estimating the 198 regressions. A sample of African countries was selected because this happens to be the world’s poorest continent with the majority of its member countries including Sierra Leone among the World Bank classified low income countries. Such sample of countries was also chosen to interact the Sierra Leone country dummy not only because it happens to be the parent continent for Sierra Leone, but because its member countries share similar economic structure with the country of interest, Sierra Leone.

The human development index, HDI is obtained from UNDP human development reports and represents an index of income per capita, longevity and education. This measure of the quality of human development is used to proxy the well-being of the poor and hence poverty levels. This measure particular suits the measure of poverty levels as Anand and Sen (1992) argue that this measure of the quality of life is better than the income poverty measure because it additionally includes non-income attributes of well-being that better explain the quality of life a person leads. Further, Gomanee and Morrissey (2002), Gomanee et al. (2005a; 2005b), and Verschoor and Kalwij (2006) find this measure to be correlated with the income poverty measure. Specifically, UNDP uses the GDP per capita to proxy income levels; life expectancy at birth to proxy longevity, and literacy rate to proxy education.

The infant mortality rate (IMR) is another measure of human welfare as used by Boone (1996), Gomanee and Morrissey (2002), Ishfaq (2004), Gomanee et al. (2005a; 2005b), Verschoor and Kalwij (2006), and Asiama and Quartey (2009) in their aid-welfare analysis. This measure is also argued to be a reasonable indicator of poverty on the basis that most of the infant deaths tend to concentrate within families that live below the poverty line (Ishfaq 2004). Data on IMR is sourced from the World Bank’s human development indicators database and the UNDP’s Human Development Reports. The infant mortality rate as defined by these World Bank and the UNDP databases represents the number of deaths in 1000 of children born under the age of 1.

Data on public sector expenditure on health (P h) and public sector expenditure on education

(P e) both as component of the PPE index are obtained from various sources. Government expenditure on education is measured as share of GDP and is sourced from the World Bank’s World Development Indicators (online), the United Nations Common database and the World Resource Institute’s Earth Trends database. Government expenditure on health is measured as

199 share of GDP and is sourced from the World Bank’s World Development Indicators (online), UC Atlas online database and the World Resource Institute’s Earth Trends database. GDPPC_1 denotes initial per capita income and is measured as the per capita GDP lagged one period. Per capita GDP is obtained from the World Bank’s World Development Indicators database.

AID denotes foreign aid and is measured as official development assistance as a share of GDP obtained from the World Bank’s World Development Indicators.

The variable, IQU is a measure of the quality of governance, whose derivation follows Knack’s (1999) construction of an index of the quality of governance. It constitutes the component of corruption, rule of law, and bureaucratic quality taken from IRIS3 database of the ICRG. A detail of its computation is already discussed in chapter 5 of this thesis.

DSL is the Sierra Leone country dummy constructed by allotting 1 to Sierra Leone and 0 for other countries in the panel sample.

Military expenditure, G mit denotes expenditures on military as share of GDP and is sourced from The World Bank’s WDI database, the SIPRI and the US Arms control and disarmament agency.

Government revenue, TGR, denotes total government domestic revenue that excludes grants. It specifically comprises tax revenue and domestic government revenue from fees, royalties and fines. It is measured as a share of GDP and is sourced from the IMF’s international financial statistics database and the World Banks’ World Development Indicators database.

GRANTS, is official development assistance in the form of grants measured as share of GDP and sourced from the OECD database.

LOANS, is official development assistance in the form of loans measured as share of GDP and sourced from the OECD database.

TCA, denotes official development assistance in the form of technical cooperation grants measured as share of GDP. This variable is obtained from the OECD database 200

NTCA, denotes official development assistance in the form of non-technical cooperation grant as share of GDP. It is derived as total aid (i.e. total ODA as share of GDP) less TCA.

Further variables used in the regressions include interaction terms constructed with the Sierra Leone country dummy (DSL) to specifically single out the impact of foreign aid on welfare levels in Sierra Leone. These interaction terms include: The aid*country dummy (ADSL) interaction term, the grant*country dummy (GrantDSL) and loan*country dummy (LoanDSL) interaction terms, the TCA*country dummy (TCADSL) and the NTCA * country dummy (NTCADSL) interaction terms. ADSL interaction term captures the aid disbursed to Sierra Leone and its test of significance captures the impact of aid to Sierra Leone in the cross-country panel data analysis. GrantDSL measures the grants disbursed to Sierra Leone and its significance shows the impact of grants on well-being of the poor in Sierra Leone. The same is true for LoanDSL, TCADSL and NTCADSL interaction terms which respectively capture the impact of loans, technical cooperation assistance and non-technical cooperation assistance on human development in Sierra Leone.

7.1.3: Estimation technique

For estimating the panel of 31 African countries over the period 1980-2007 (five-year period averages), the study employs the robust one-step system Generalised Methods of Moments (GMM) by Arellano and Bover (1995) and Blundell and Bond (1998). Generally, the technique of GMM has several advantages over cross-section regressions and other estimation methods for dynamic panel data models: Firstly, estimates will no longer be biased by the omission of any time-invariant variables (i.e. unobserved country-specific or fixed effects). Hence, GMM is useful for its treatment of unobserved individual heterogeneity that may be correlated with the explanatory variables. Secondly, the use of instrumental variables allows the parameters to be estimated consistently in models which include endogenous right hand side variables (Arellano and Bond, 1991, Arellano and Bover, 1995; Blundell and Bond, 1998; Windmeijer, 2006; Soto, 2010). “Regressors are endogenous when they are correlated with current (and possibly past) shocks…”.(Windmeijer, 2006: 9). Static panel estimators such as pooled OLS, fixed effects and Random effects models usually have problems such as heteroscedasticity, serial correlation, but most importantly endogeneity of some explanatory

201 variables. Problems of heteroscedasticity in static panels can be corrected by using robust standard errors. However, serial correlation and endogeneity of an explanatory variable cannot be convincingly corrected in such static models. Some researchers merely use the lagged value of the endogenous explanatory variable to correct for endogeneity, but this has often been criticised as not sufficient. Instrumental variables in 2SLS have also been employed to correct for endogeneity, but the difficulty of finding an appropriate instrument makes this technique less popular. With GMM, these limitations are all captured in the technique. Thirdly, the use of instruments potentially allows consistent estimation even in the presence of measurement errors (Arellano and Bond, 1991; Bond et al., 2001).

The GMM estimator in its improved form as system GMM by Arellano and Bover (1995) and Blundell and Bond (1998) is further found to be superior an estimator for dynamic panel data models. System GMM has the advantage over the earlier differenced form GMM (by Arellano and Bond, 1991) by having superior finite sample properties. In addition to having suitably lagged levels as instruments for the standard sets of equations in first-differences as is inherent in difference GMM, System GMM is such that it further allows the use of suitably lagged first-differences of the series as instruments for the equations in levels (Arellano and Bover, 1995). By combining these two sets of instruments, the system GMM ensures it has stronger instruments, which results in reduction of finite sample biases. Hence, system GMM is asymptotically more efficient. Further, simulation results reported in Blundell and Bond(1998) show that first difference GMM estimations may be subject to downward finite sample bias (due to weak instruments), particularly when the number of available time periods is small. System GMM on the other hand has superior finite sample properties and hence best suited to estimating autoregressive models with persistent panel data. This is also supported by Bond et al. (2001) who carried out some Monte Carlo simulations to show that the system GMM is asymptotically more efficient than difference GMM and other alternative estimators of dynamic panel data models as long as the assumption of no autocorrelation between the differenced dependent variable and the unobserved country-specific effects (fixed effects) holds.

In our analysis, we however use the one-step instead of the supposedly more efficient two- step System GMM estimator. This is largely due to the very small size of our sample of 31 countries. Moreover, Soto (2010) finds using Monte Carlo simulations that in cross-country regressions using very small country size as 35, the one step system estimator is the most 202 efficient. And considering our sample size of 31 countries, we therefore employ the use of one-step system GMM. Further, Bond et al. (2001:18) argue that “For the special case of spherical disturbances, the one-step and two-step GMM estimators are asymptotically equivalent for the first-differenced estimator. Otherwise the two-step estimator is more efficient…” They however point out that, in finite samples, the choice between the two estimators is not clear cut as Monte Carlo studies have shown that the efficiency gain is typically small, and that the two-step GMM estimator has the disadvantage of converging to its asymptotic distribution relatively slowly. They maintain that in small samples, the asymptotic standard errors associated with the two-step GMM estimators can be acutely biased downwards and thus form an unreliable guide for inference. Cognizant of this, this study prefers to report the results of the one-step System GMM, with robust standard errors for which Blundell and Bond (1998) find to be not only asymptotically robust to heteroscedasticity but also more reliable for finite sample inference. We further use the small- sample corrections to the covariance matrix estimates by adding in the syntax the ‘small’ option. The effect is that STATA reports t-statistics instead of the z-statistics for the estimates of the individual parameters and the F-test instead of the Wald chi squared test for overall fit or joint significance of the regressors.

We use STATA 11.0 package to run our regression estimations and derive the descriptive statistics. The STATA command ‘xtabond2’ written by Roodman (2006) is used instead of the ‘xtabond’ or ‘xtdpd’ in our regressions because of its several advantages: Firstly, the xtabond2 command allows the use of lag operators within the instrument matrix while xtabond does not. Secondly, xtabond2 command enables flexibility in the use of differing lag limits in the variables used as instruments where as the xtabond and xtdpd commands do not; and hence xtabond2 gives the user more control over the instrument matrix. Thirdly, Roodman (2003) indicates that for one-step, robust estimation, xtabond2 command reports the Hansen J test of overidentifying restrictions, which is actually the standard Sargan test for the two-step estimator. He argues that unlike the Sargan test for the one-step estimator, the Hansen J is robust to heteroscedasticity and autocorrelation within panels (ibid). By using the ‘robust’ option in the syntax in the one-step GMM, Roodman (2006:35) explains that “xtabond2’s robust is equivalent to cluster(id) in most other estimation commands, where id is the panel identifier variable, requesting standard errors that are robust to heteroscedasticity and arbitrary patterns of autocorrelation within individuals.” This also implies the t-statistics reported in brackets in the regressions are heteroscedasticity corrected. Hence in our 203 regression tables, we report only the Hansen test for overidentifying restrictions which is more meaningful since we are using the one-step System GMM instead of the two step system GMM. We use lagged levels at time t-2 and earlier as instruments for the equations in first difference, which is consistent with the recommendation by Bond et al. (2001) and Roodman (2006). Because our sample is small with the potential of having more instruments than the number of groups, we employed the ‘collapse’ command (as recommended by Roodman, 2006) to further reduce the number of instruments to a manageable level.

7.1.4: Diagnostics and Inference

It is however a requirement in system GMM that the instruments for the level equation (i.e. the lagged differences of the endogenous variables), are valid only if they are orthogonal to the fixed effect. The dynamic System GMM model specifications we present in this study are only valid if the estimator is consistent and the instruments are valid. The GMM system estimator is consistent if there is no second-order serial correlation. The Hansen test checks for the validity of the instruments. In the test for overidentifying restrictions, we report only the Hansen test (and not the Sagan Test) as this is suggested to be robust, though can be weakened by too many instruments, whereas the Sagan test is reported as not being robust but cannot be weakened by too many instruments. Our use of the ‘collapse’ command has however reduced the potential problem of having too many instruments in the small sample countries in the panel, and hence ensures our use of the Hansen test. Further, since we are running robust estimators, it implies the Hansen test-statistics is best used since it is the appropriately applicable test in the robust estimation. The Hansen test for overidentifying restriction is a test for exogeneity of the GMM instruments. Hence, the null hypothesis here is that the instruments as a group are exogenous, and hence acceptance of this hypothesis implies our model is valid. The Arellano-Bond test for autocorrelation has a null hypothesis of no autocorrelation, with the AR(1) testing for no first order autocorrelation and the AR(2) testing for no second order autocorrelation. By default, STATA reports these diagnostic tests along with the regression estimates.

204

7.2. The Empirical Results

In this section, we present, interpret and discuss the regression results on the impact of total foreign aid as well as disaggregated aid on welfare indicators as proxy for poverty levels. Further investigation is made on the importance of politics for aid’s impact on human development. In the first sub-section, we analyse the impact of total foreign aid on well being of the poor proxied by human development and infant mortality rate in Africa, Sub-Saharan Africa and Sierra Leone. We then check for robustness of the impact of aid, with a change of the specification where year dummies are added to the regressions. This is followed by the analysis of the impact of foreign aid in its disaggregated forms on human development as a measure for the well-being of the poor in Sierra Leone. In the third sub-section, we examine the importance of politics (in the context of the improvement in democracy and improvement in the quality of polity regime) for aid’s influence in improving human development in the Sierra Leone context. In each sub-section, we discuss the results in the context of the literature and implication for aid effectiveness in respect of the poverty reduction criterion for Sierra Leone.

7.2.1: Impact of aid on Welfare (on HDI and IMR)

Tables 7.1 and 7.2 below present the system GMM regression results for the impact of foreign aid on well-being using a 31 African country sample and 27 Sub-Saharan African sample. Both indicators of well-being: human development and infant mortality are considered, to examine the impact that foreign aid has had on well-being in Africa, sub- Saharan Africa and Sierra Leone for the period 1980-2007.

7.2.1.1: Foreign Aid and Human development

Before the estimates of a regression are deemed useful for interpretation, it is required to conduct diagnostic tests to ascertain the models used in the regression are valid. To test for the validity of our model, we present in Table 7.1 the test for autocorrelation using the Arellano-Bond AR(2) and the test for validity of the instruments using the Hansen test for overidentifying restrictions. These two tests are done for each regression we run as they constitute the crucial diagnostic tests in system GMM analysis. The result of the AR(2) test reveals that the hypothesis of no autocorrelation is accepted in all the regressions presented in the table, thus ensuring that there is no problem of serial correlation in our regressions. This

205 also implies that our estimator is consistent, which is an important requirement for the acceptability of the estimation results. Also, the Hansen test for overidentifying restriction reveals that the hypothesis for exogeneity of the instruments is accepted, implying that the GMM instruments used in these regressions are valid. Together, both tests confirm that our regression models are all valid and that conclusions arrived from these estimates can be largely relied upon for inference.

Table 7.1: Foreign Aid and Human Development Sample Full African Sample Sub Saharan Africa Dep. Variable/Model HDI HDI HDI HDI (Model 1) (Model2) (Model 3) (Model4) Regressors HDI(-1) 0.320*** 0.307*** 0.275*** 0.259*** (4.01) (3.93) (3.23) (3.09) AID -0.004 -0.024 -0.044 -0.069 (0.06) (0.41) (0.64) (1.12) AID*DSL - 0.352*** - 0.390*** (3.86) (4.00) PPE 0.071* 0.068* 0.096*** 0.096*** (1.86) (1.77) (2.87) (2.82) ME -0.079* -0.090* -0.106** -0.120** (1.83) (1.95) (2.08) (2.21) GDPPC(-1) 0.158* 0.140* 0.108 0.089 (1.79) (1.80) (1.29) (1.15) IQU 0.075 0.100 0.108 0.140 (0.77) (1.03) (1.05) (1.34) DSL -0.382*** -1.414*** -0.324*** -1.466*** (3.67) (6.26) (3.05) (6.50) Constant -1.847** -1.701*** -1.509** -1.357** (2.66) (2.78) (2.33) (2.26) No. Of Obs. 135 135 116 116 No. of Instruments. 16 20 16 20 F-Stat 535.08 884.11 194.79 476.69 AR(2) 0.894 0.897 0.919 0.609 Hansen Test for (OIR) 0.185 0.383 0.137 0.445

Note :- All variables are in logs except the country dummy. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%. OIR denotes Overidentifying Restrictions.

Table 7.1 also presents the GMM estimates for the relationship between foreign aid and human development in Africa, Sub-Saharan Africa, and further specifically capturing the impact for Sierra Leone. The GMM estimation of the relationship between foreign aid and human development in Africa reveals that aid does not significantly improve human development and hence, well-being in the continent. The coefficient of the aid variable in Model (1) of table 7.1 above emerges to be statistically insignificant, thus implying for the sample of African countries over the period 1980-2007, foreign aid does not improve human development. The most obvious determinants of human development in the continent are pro-

206 poor public expenditure (which comprises of government expenditures on health and education) and initial per capita income with both emerging to be a significant determinants of human development in the continent. Military expenditure, as expected, is negatively related with human development and tends to moderately have an adverse effect on aggregate well-being in the continent. Governance, on the other hand does not emerge to be a significant determinant. The Sierra Leone country dummy emerged to be significantly negative which may as well be interpreted that relative to the rest of the continent the level of human development in the country has been significantly unimpressive.

In model 3 of table 7.1 above, we further estimate another model with the use of sub-Saharan Africa sample of countries to examine whether aid is rather effective in this sub-region. The results are very much consistent with the estimates using the full African sample for the dynamic panel data analysis. Foreign aid remains insignificant in improving human development in sub-Saharan Africa, just as with the full African sample. Rather, the control variable, Pro-poor Public Expenditure only emerged to be significant in improving human development in SSA, while military expenditure tends to adversely affect the improvement of human development in the sub-region; further confirming the results of the full African Sample.

In model 2 of Table 7.1 above, an attempt is made to capture the impact of aid on human development for Sierra Leone relative to the African continent by adding an interaction term of foreign aid and Sierra Leone country dummy (i.e. AID*DSL). The coefficient of this interaction term emerged to be highly significant implying that relative to the African continent, foreign aid has had a significant contribution to improving human development in Sierra Leone. The aid estimate reveals that in Sierra Leone, for every 1% increase in aid/GDP ratio, human development will improve by 0.328% (i.e. -0.024+0.352). This implies that though aid may have been insignificant in fostering improvement in human development in Africa, yet in the case of Sierra Leone aid is an important determinant of human development. Even when we re-run this model (Model 4 of Table 7.1) with the use of the sub-Saharan African sample to interact aid and the Sierra Leone country dummy, our finding that aid is a significant determinant of human development improvement in Sierra Leone remains unchanged. The coefficient of the AID*DSL interaction term is highly significant, showing that in Sierra Leone relative to the rest of sub-Saharan Africa, for every 1% increase

207 in aid/GDP ratio, human development will improve by approximately 0.321% (i.e. - 0.069+0.390), consistent with the estimate of the regression using the full African sample.

In all of the regressions reported, the lagged value of the human development index is statistically significant which confirms our suppositions that improvement in well-being has a dynamic nature. This also provides support for the suggestion by Addison et al. (2008) who advocates for the inclusion of dynamics into poverty analysis as one front in which to improve research in poverty analysis and hence the effectiveness of poverty reduction policies. They argue that “There is now a wide acceptance that static analyses have limited explanatory power and may conceal the processes that are central to the persistence of poverty and/or its elimination” (ibid: 2). One of the ways they suggested information about poverty dynamics can be captured is via the use of panel data methods. This method they argue captures the changes of poverty situation over time. Hence, our finding of the significance of the lagged dependent human well-being variable provides support for their suggestion.

7.2.1.2 Aid and Infant Mortality Rate

After examining the impact of aid on human development, attention is then turned to the second indicator of well-being. Here, we examine the impact of foreign aid on infant mortality rate in Africa, Sub-Saharan Africa and Sierra Leone. Table 7.2 below presents the results of the diagnostic tests and regression estimates for examining this relationship.

As mentioned in the previous analysis, the crucial diagnostic tests for validity of system GMM estimates are the test for serial correlation (with AR(2) test) and the Hansen test for overidentifying restrictions. The diagnostic tests for the full African sample (Table 7.2-Model 1) show that the model passes the tests for serial correlation as well as the Hansen test for overidentifying restrictions. This implies that our GMM estimator is consistent (as the model is free from serial correlation), and also that the instruments we use are valid hence ensuring that our estimates for the determinants of infant mortality as an indicator of well-being in Africa are largely reliable enough for inference. Likewise, in the remaining three models estimated as presented in the table (Table 7.2), both diagnostic tests are passed to ensure consistency of our GMM estimator as well as validity of our models. Hence, the regression estimates from these models can be said to be valid for inference.

208

Table 7.2: Foreign Aid and Infant Mortality Rate Sample Full African Sample Sub-Sahara Africa Dep. Variable/Model IMR IMR IMR IMR (Model 1) (Model 2) (Model 3) (Model 4) Regressors IMR(-1) 0.951*** 0.948*** 0.826*** 0.821*** (11.62) (11.06) (7.56) (7.16) AID 0.051 0.046 -0.028 -0.029 (1.01) (0.92) (0.50) (0.51)

AID*DSL - -0.012 - -0.005 (0.21) (0.048) PPE -0.013 -0.014 -0.007 -0.009 (0.35) (0.38) (0.29) (0.37) ME -0.026 -0.027 -0.011 -0.013 (0.92) (0.91) (0.52) (0.57) GDPPC(-1) 0.048 0.042 -0.048 -0.048 (0.84) (0.75) (0.68) (0.71) IQU -0.115** -0.112** -0.039 -0.038 (2.12) (2.01) (0.63) (0.60) DSL -0.016 0.031 0.172 0.190 (0.22) (0.17) (1.55) (0.93) Constant -0.171 -0.104 1.198 1.229 (0.34) (0.21) (1.44) (1.51) No. Of Obs. 153 153 133 133 No. of Instruments. 16 20 16 20 F-Stat 502.64 964.89 488.55 866.96 AR(2) 0.353 0.372 0.428 0.446 Hansen Test for (OIR) 0.073 0.185 0.107 0.341

Note :- All variables are in logs except the country dummy. T statistic in parenthesis. *** Significant at 1%, ** significant at 5% and * significant at the 10%.

The estimates for the full African sample generally show that foreign aid does not reduce infant mortality rate in the continent, with the results even suggesting that increases in aid in the continent tend to be associated with an increase in infant mortality rate though the estimate itself is statistically insignificant. Besides the lagged IMR variable (i.e. IMR(-1)), the only significant determinant of infant mortality rate in the model is governance, which shows that it is an important determinant in reducing infant mortality in Africa.

In Model 3 where we examine the impact of foreign aid on infant mortality rate in sub- Saharan Africa, the results again show that foreign aid does not significantly reduce infant mortality rate in the sub-region. The sign of the aid estimate though in the right direction with intuition, yet remains insignificant. Hence, the results are largely consistent in showing that foreign aid does not reduce infant mortality rate in both Africa and Sub-Saharan Africa. However, as opposed to the full African Sample where governance is found to significantly reduces infant mortality rate in the continent, in the case of SSA, governance remains

209 insignificant (though with an encouraging negative sign) in reducing infant mortality in the sub-region.

After previously finding foreign aid to be a significant determinant of human development in Sierra Leone (despite showing to be non-responsive for the full African sample and sub- Sahara Africa), we consider its importance in reducing infant mortality rate in the country. In model 2 of table 7.2, we present the IMR regression estimates for Africa for which the AID*DSL interaction term is added, and in Model 4, we use the Sub-Saharan Africa sample and add the AID*DSL interaction term. In both regressions, the AID*DSL interaction term is insignificant, thus showing that in Sierra Leone, foreign aid has not improved infant mortality rate in the country just as it does for the entire African Continent and Sub-Saharan Africa. Hence, though foreign aid may have been an important determinant of human development in Sierra Leone (as previously shown), yet it has not significantly reduced infant mortality rate in the country. Though the direction of the AID*DSL interaction term is negative showing that increases in aid to the country may be associated with reduction in infant mortality rate, yet the estimate is statistically insignificant.

7.2.1.3 Robustness Check In an attempt to check for the robustness of the estimates against a change of the model, we re-estimated the above regressions by adding year dummies to the models. Appendix 7.3 and 7.4 present these regressions. Appendix 7.3 presents the estimates for the impact of aid on human development with the addition of year dummies, whilst Appendix 7.4 shows the regression results for the impact on infant mortality rate with added year dummies. In both analyses, all the models we run pass the diagnostic tests to be described as valid and hence can largely rely on these estimates for inference. The test for autocorrelation (AR (2)) reveals that the hypothesis of no autocorrelation cannot be rejected for all the regressions we run here. Similarly, the Hansen test for overidentifying restrictions reveals that in all the models we run, the hypothesis of exogeneity of the instruments used cannot be rejected, thus ensuring that these models are valid.

The regression estimates tend to be largely consistent (especially for the aid estimates) with those obtained without time dummies in the model as presented in the main analysis. The impact of aid in improving human development in Africa and SSA does not emerge to be significant (Model 1 and 3 in Appendix 7.3) just as we had in the main models when no year

210 dummies were added. In terms of the control variables, the dynamic term (the lagged HDI variable) and the pro-poor public expenditure maintain their significance in improving human development in Africa and SSA. However, when year dummies are added, military expenditure and initial income variable rather turn insignificant. Governance (i.e. IQU), rather becomes significant in improving human development in Africa and Sub-Saharan Africa when year dummies are added. In the case of Sierra Leone (Model 2 and 4 in Appendix 7.3), foreign aid has a highly significant impact in improving human development in the country, as is shown by the significance of the coefficient of the AID*DSL interaction term in both the full African and sub-Saharan African samples; thus being consistent with our main estimation. The marginal effectiveness of aid in terms of its impact of human development for Sierra Leone (as is shown by the magnitude of the coefficient of the AID*DSL interaction term) however, tends to be a little lower than the values in the regressions without time dummies (i.e. in our main analysis). Consistent however, is that even when we add time dummies to the main regressions, the results remain unchanged that foreign aid, in the case of Sierra Leone is an important determinant of the improvement of human development in the country. Hence, our findings on the impact of foreign aid on human development in Sierra Leone appear to be robust to a change of the panel sample and model specification with or without year dummies.

The same story of consistency of the aid estimates is also evident when we analyse the impact of foreign aid on infant mortality rate in Africa, SSA and Sierra Leone (Appendix 7.4). Here also, the estimates of the AID*DSL interaction term are insignificant , implying that foreign aid in the case of Sierra Leone has not had a significant impact in improving infant mortality rate in the country, just as it is generally for Africa and sub-Saharan Africa. Thus in Sierra Leone, irrespective of changes to the model specification (by the addition of year dummies) and the panel sample used, foreign aid does not seem to significantly reduce infant mortality rate though it shows to significantly improve human development levels in the country. In sub-Saharan Africa, and even the whole of Africa, foreign aid does not generally improve the well-being of the poor in both its indicators employed in this study.

7.2.1.4 Discussion

The analysis of the impact of foreign aid on well being reveals that aid is insignificant in improving welfare of the poor in Africa. For both indicators of welfare involving human

211 development index and infant mortality rate, the system GMM regression estimates reveal that foreign aid does not improve human development neither does it reduce infant mortality rate in the continent. Even when we limited our sample to sub-Saharan Africa, aid still does not emerge to improve welfare in the region. Further limiting the analysis to the country level context for Sierra Leone, we find that relative to Africa or even Sub-Saharan Africa, foreign aid rather emerges to be positively significant in improving human development in the country. However, in as much as aid may significantly improve human development in Sierra Leone, it does not however, significantly reduce infant mortality rate in Sierra Leone.

The finding that aid does not holistically improve welfare in Africa and sub-Saharan Africa only provides more evidence on the mixed cross country findings about the impact of aid on welfare. Using a dynamic panel analysis, our results support the finding by Kosack (2003) that directly, foreign aid does not significantly improve human development in aid recipient countries. His study on a panel of aid-recipient countries shows that aid only improves human development contingent on good democracies; and hence that aid does not directly/independently promote the improvement of human development. Likewise, McGillivray and Noorbakhsh (2007) could not find aid to improve human development in a cross-section analysis of aid recipient countries, and even finds a significantly negative relationship between foreign aid and human development. These results of the insignificant impact of aid on improving well being in Africa also provide support to the study by Boone (1996) who could not find evidence to show that aid is associated with low levels of infant mortality rate. The non-significant direct impact of aid on welfare at the cross country level also supports the findings by Mosley et al. (2004) and Williamson (2008) who could not as well find evidence to show that aid directly impacts on welfare. Further, Asiama and Quartey (2009) using the same methodology of system GMM for a panel of sub-Saharan countries, find that bilateral aid as proxy for official aid does not significantly improve welfare in the region, thus being in conformity with our findings. They provided evidence to show that aid does not improve human development neither does it reduce infant mortality rate in sub- Saharan Africa.

Our finding in terms of the impact of aid on welfare in Africa does however dispute those by Gomanee et al. (2005a), Fielding et al (2006) and Gillanders (2010) who rather find foreign aid to directly improve welfare at the cross country level. Gillanders (2010), though using a different proxy in growth rate of life expectancy for human development, found that aid 212 improves human development in sub-Saharan Africa. Gomanee et al. (2005a) finds aid to directly improve well-being on both its indicators of human development and infant mortality rate in a cross-country study of aid-recipient countries. Likewise, Fielding et al (2006) show that aid significantly improves human development indicators.

Our study’s finding of an insignificant direct impact of foreign aid in improving well-being in the African continent may be associated with reasons provided by aid sceptics such as Griffin 1970; Griffin and Enos (1970); Mosley et al. (1987); and Boone (1996) for aid’s insignificance in improving development outcomes in aid-recipient countries. Their main argument is that foreign aid may be used to augment consumption of those who are relatively well-off in developing countries and hence that aid is not targeted on the poor or poverty oriented programmes and projects in such aid-recipient countries.

Perhaps the most interesting finding here is that though foreign aid may not have significantly improved wellbeing in Africa, in Sierra Leone it does directly improve human development though it does not significantly reduce infant mortality rate in the country. Thus, whilst our cross-country analysis for Africa and Sub-Saharan Africa may not have supported the findings by Gomanee et al. (2005a), Fielding et al. (2006) and Gillanders (2010) that aid directly improves well-being, our estimation for the case of Sierra Leone does support their evidence in the case of the impact of aid on human development. It does not however support their finding that aid reduces infant mortality rate, which also is in consonance with the findings by Ishfaq (2004), who finds that foreign aid does not reduce infant mortality rate in the case of Pakistan. The question then is why should aid improve human development in Sierra Leone, but fails to do so for the entire continent or even sub-Saharan Africa? And secondly, why should aid improve human development but fails to improve infant mortality rate in the country? Several reasons may be suggested for the positively significant impact of foreign aid in improving human development in Sierra Leone.

First, Sierra Leone has a lower rate of HDI compared to the rest of Africa as is shown by the negatively significant country dummy relative to the African sample and compared to the average HDI for Africa (i.e. 0.25 average for Sierra Leone against 0.44 average for African Panel and 0.42 average for SSA sample). Gomanee et al. (2005a and 2005b) find that the impact of aid in improving human development is more significant in countries with lower HDI value. By further splitting their sample to a set of low-income countries, they find that 213 the marginal effectiveness of aid on HDI tends to be higher in countries with low HDI. This may be due to the fact that such low HDI countries have higher capacity to accommodate more resources for developmental purposes. Gomanee et al. (2005b: 308) suggest one explanation for the observed higher impact of aid on human development on the more lower income countries as probably due to the fact “the lower the welfare, the more room for improvement to be brought by aid and pro-poor spending hence the larger the impact.” This implies these social sectors in low income countries have a high absorptive capacity for resources and for which foreign aid may well supplement and hence become effective. In tune with this finding, as Sierra Leone generally has a low HDI even lower than that of Sub- Saharan Africa, this may explain our finding that aid tends to be more effective on human development in Sierra Leone (with relatively lower HDI) compared to the rest of Africa and Sub-Saharan Africa.

A second reason for the significant impact of aid in promoting human development in Sierra Leone may be associated with the historical practice of donors directing much aid on project or sector based activities many of which may be poverty oriented. Sierra Leone has been well known to be one of the poorest countries in the world as has been typical in the UNDP’s HDI ranking since its inception. Hence, donors have targeted much of their aid on poverty- reducing activities which they have seemed to finance directly either through project/sector support or via NGOs. Instead of going through government budgets (as is the case with direct budget support), rather, projects aid directly funds projects in government ministries and is managed by separate project implementation units largely supervised and monitored by the donors themselves. The advantage of this is that aid may directly target the poor thus avoiding the possible fungibility problems it might face if it were to go through government budgets.

Further, by directly funding the agricultural sector, project aid to agricultural sector can provide the largely poor farmers with much returns, which might as well improve human development by availing much income to increasingly finance expenditures on health and education. McGillivray and Noorbakhsh (2007), while suggesting reasons for hypothesising a positive relationship between aid and human development, suggest that aid induced growth may lead to higher wages due to increased labour demands following increases in economic activity and also leads to increased private and public expenditures on health and education. This is evident in our earlier finding (chapter 6) that aid promotes pro-poor growth in Sierra 214

Leone, which is growth in those areas/sectors where the poor are mostly employed. Specifically, our study showed that higher aid inflow is associated with higher agricultural growth. Hence, by increasing the income of the poor farmers, aid may as well improve human development.

Further, NGO assistance in the country seems to have focused more on poverty reduction. It is evident that since the start of the civil conflict in the early 1990s to the post-war period extending to 2007, there has been a proliferation of NGOs in Sierra Leone largely targeting the poor in the areas of welfare and human rights. Many of these NGOs have accessed funding from official donors with notable reference to USAID and German based GTZ who seem to disburse much of their aid either through direct project support or via international and local NGOs and CBOs. Masud and Yontcheva (2005) showed that though official aid (proxied by only bilateral aid) does not significantly reduce well-being of the poor in terms of infant mortality rate, NGO aid on the other hand does.

Thirdly, aid may have directly improved human development in Sierra Leone owing to the stringent donor monitoring of their project support which is largely geared towards poverty reduction initiatives. Direct donor funded projects are reported (following interviews with donor representatives and public officials) to be closely monitored by the donors themselves to ensure the objectives the projects are designed for are largely met. And since a significant portion of these projects are poverty targeted, this will accordingly ensure poverty reduction is achieved to a larger extent. Further, donor conditionality on re-direction of government expenditure priorities towards poverty reduction cannot be ruled out for the observed impact of aid on human development. Since the commencement of the MDGs and PRSP concepts, donors have conditioned debt relief (following the HIPC initiative) on the diversion of savings from such relief towards poverty oriented expenditures. Osei and Quartey (2001) argue that debt relief (aid) directed towards welfare expenditures and welfare oriented project support, particularly if government budget on welfare sectors of health and education are already low, may as well enhance aid’s impact in significantly improving human development. They argue that “…if all savings [from debt relief] is spent on social sectors, it may improve the welfare of the poor…..” (Osei and Quartey, 2001: 10). If this holds in the case of Sierra Leone, it may well imply human development will be improved following increased aid (debt relief) flows in the country.

215

Yet, it may be paradoxical that aid is significant in improving development outcomes in Sierra Leone amidst evidence of public sector corruption. We however argue that the possibility of this occurrence in the case of Sierra Leone might as well be associated with the source of funds that the government does misallocate. It’s no secret that Sierra Leone has a long history for misuse of public resources. Public Expenditure Tracking Survey (PETS) reports and service delivery reports have all shown evidence of misuse of public resources in the country. It is therefore indisputable that foreign aid (which constitutes part of public sector resources) may have been subject to misuse in the country. However, for aid to improve human development amidst evidence of public sector corruption, may as well imply that, it may turn out that a considerable proportion of public resources misused may have mostly come from domestically generated revenue, bank financing of the public sector and from unofficial politically affiliated aid. Political elites can easily use these sources of funds in their patrimonial expenditures; since expenditures on such funds are not monitored by the influential donors in the countries and hence elites are at leverage to misappropriate such resources. And since the much donor monitored project support and debt relief (aid) may be largely poverty oriented, it may well be that human development will as well be improved by aid flows amidst the existent corruption in the public sector. It may however be that the marginal effectiveness of aid on human development in the country could have been much higher in a situation of much controlled public sector corruption. It may therefore not be surprising that in the case of Sierra Leone, though foreign aid is found to improve human development in the country, yet the level of human development remains comparatively low compared to other developing countries.

Hence, if all these reasons hold, we may as well find foreign aid to improve human development in Sierra Leone as the regression estimates have shown. But that aid does not reduce infant mortality rate despite improving human development in the country is yet another finding worth exploring. The limited evidence on the impact of aid on welfare so far seems to point to the impact being evident more on improvement of human development than on reduction of infant mortality. Boone (1996) for instance could not find evidence to show that aid is associated with low levels of infant mortality rate, which is in agreement with our finding for Sierra Leone and even the African continent. Likewise, Masud and Yontcheva (2005), though used bilateral aid as proxy for official aid, find that aid does not reduce infant mortality rate in aid recipient countries; a finding also confirmed by Asiama and Quartey (2009) who used the similar proxy for official aid to show that aid does not reduce infant 216 mortality rate in sub-Saharan Africa. Likewise, Ishfaq (2004) could not find foreign aid to significantly reduce infant mortality rate in Pakistan.

In their examination of the impact of aid on both welfare measures of human development and infant mortality at the cross country level, Gomanee et al. (2005a) also find that though the impact of aid in reducing IMR was significant, yet the effect was generally weaker compared to that on human development. Hence, they argue that the results from the IMR estimations are generally weaker than those from HDI. These results are also observed in Gomanee et al. (2005b) where the study shows that there is only weak tendency for aid to reduce IMR in those countries with higher IMR. Assuming this to hold, and with Sierra Leone being a country with a quite high IMR, higher than the average for Africa and SSA (i.e. 177 average for Sierra Leone compared to 92 average for Africa and 97 for SSA sample), it goes without saying that it may not be obvious to expect aid to significantly reduce IMR in the country as has been shown in our GMM estimates for the extended analysis on Sierra Leone.

A second possibility for expecting aid to significantly improve human development but not reduce infant mortality may be associated with aid being effective in increasing the income of the poor (as has been shown by the aid-pro-poor growth analysis in chapter 6), but does not significantly improve health sector indicators which are associated with reduction in infant mortality rate. As the HDI measure constitutes an income measure in addition to health and education components, aid that increases income may as well improve human development levels, but may not necessarily reduce infant mortality rate. The ineffectiveness of public resources (including foreign aid) on the health sector in Sierra Leone is clearly evident in the several Public Expenditure Tracking Survey (PETS) reports in the country. Thompson (2007) and Robinson (2008) cited PETS reports in the post-conflict era as showing that only 5% of medical supplies disbursed from the central medical stores (which can be largely aid-funded) actually reach the primary health units across the country for which they were intended. This implies very little of the targeted medical supplies reach the very poor rural households in the country. Therefore, it may be expected that though foreign aid may emerge to improve human development in Sierra Leone, yet it does not show likewise in the reduction of infant mortality in the country.

217

7.2.2: Aid Structure and Human Development in Sierra Leone

After having estimated the relationship between total foreign aid and well-being in the context of human development in the case of Africa, and Sierra Leone in particular, in this section, we attempt to further examine this relationship when total aid is disaggregated for the case of Sierra Leone. The justification for disaggregating aid is generally discussed in the previous analytical chapters (5 & 6) of this thesis. Hence, though total aid may have shown to have a direct association with higher levels of human development in Sierra Leone, it is vital to assess whether in fact it is all modes of aid delivery that respond to improvements in human development in the country. In this respect, we consider two sets of aid structure disaggregation: grants versus loans and technical assistance versus non-technical assistance.

Table 7.3 below presents the results of this analysis. The results of the diagnostic tests reveal that the models we run here are all valid and hence largely reliable for inference. The test for autocorrelation (AR (2)) shows that in all the models, our system GMM estimator is consistent, as the hypothesis of no autocorrelation is not rejected in all the models. Similarly, the Hansen test for overidentifying restrictions reveals that the hypothesis that the instruments used in each model are valid, cannot be rejected. Thus, in all, our GMM specifications are all valid enough to accept the estimates obtained from our regressions.

218

Table 7.3: Structure of Aid and Human Development Dependent Variable is HDI Sample Full African Sample Sub-Saharan Africa Model M1 M2 M3 M4 Regressor HDI(-1) 0.277*** 0.292*** 0.236** 0.222** (3.41) (3.39) (2.65) (2.40)

GRANT -0.04 - -0.035 - (0.10) (0.83) LOAN -0.007* - -0.009** - (1.70) (2.10) TCA - -0.094 - -0.146* (1.08) (1.80) NTCA - 0.020 - 0.006 (0.66) (0.18) GRANT*DSL 0.384*** - 0.391*** - (7.36) (7.72) LOAN*DSL -0.002 - 0.006 - (0.11) (0.36) TCA*DSL - 0.080 - 0.086 (1.44) (1.54) NTCA*DSL - 0.232*** - 0.253*** (3.59) (4.10) ME -0.079* -0.080* -0.107* -0.111* (1.72) (1.75) (1.94) (1.99) PPE 0.051 0.073* 0.074** 0.104** (1.22) (1.95) (2.15) (2.72) GDPPC(-1) 0.174*** 0.108* 0.137** 0.054 (2.82) (1.67) (2.32) (0.83) IQU 0.099 0.150 0.124 0.217 (1.05) (1.43) (1.24) (1.94)* DSL -1.403*** -1.068*** -1.415*** -1.072*** (7.59) (6.94) (8.12) (8.31) Constant -1.993*** -1.571*** -1.756*** -1.297** (4.09) (3.39) (3.85) (2.71) No. of Obs. 134 133 115 114 No. of Instruments 25 25 25 25 F-Stat 803.63 394.75 599.17 236.47 AR(2) 0.766 0.658 0.993 0.958 Hansen Test for OIR 0.974 0.555 0.990 0.469 Note :- All variables are in logs except the country dummy and loan variable. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%.

7.2.2.1 Grants versus Loans

In our first analysis, we disaggregate the total aid variable into grants and loans. In model 1 of Table 7.3 above, we find that grants tend to be highly significant in improving human development in the case of Sierra Leone. The GRANT*DSL interaction term emerged to be positive and statistically significant to imply that increases in grants as a form of aid modality is associated with improvement in human development in the case of Sierra Leone. The combined coefficient of the grant variable and GRANT*DSL term indicates that for every 1 % increase in grants/GDP ratio, human development in Sierra Leone will improve by 0.34%. 219

Loans on the other hand (as shown by the LOAN*DSL term) did not prove to foster human development in the country, and even tend to be negatively related, though insignificant. Hence, our finding that total aid significantly improves human development in the case of Sierra Leone is largely due to the use of grants but not loans. Thus, though total aid significantly improves human development in the country, yet aid in the form of loans does not. In model 3 of table 7.3 above, we use the sub-Saharan Africa sample and add the AID*DSL interaction term to the model. Our re-estimation only further confirms the finding that grants, and not loans are the important form of aid that improves human development and hence well- being of the poor in Sierra Leone.

7.2.2.2: Technical Assistance versus Non-Technical Assistance

In our second aid disaggregation analysis, we disaggregate foreign aid into technical assistance and non-technical assistance aid (Model 2 and 4 in Table 7.3 above). The results of the system GMM estimation reveal that technical assistance does not improve human development in Sierra Leone. In fact, it tends to be negatively related though insignificant in improving human development in the country. The significance of total aid on human development in the country as has been found earlier, is thus only due to aid disbursed in the form of non-technical assistance.

7.2.2.3 Robustness Check To examine whether the estimates of the aid disaggregates are robust to changes in the model specification, we run the previous models with the addition of year dummies (though not reported) (Appendix 7.5) using both the full African sample (Models 1 and 2) and a sample of sub-Saharan African countries (Models 3 and 4). The results of the diagnostic tests reveal that our models are all valid and hence reliable enough for acceptability of conclusions reached with regards the importance of aid disaggregation on well-being in case of Sierra Leone relative to Africa and Sub Saharan Africa. The test for serial correlation shows that our system GMM estimator is consistent, as hypothesis of no serial correlation is not rejected across all the models estimated. Similarly, the Hansen test for overidentifying restriction shows that the instruments we employ in all the models are valid, as hypothesis that the instruments are valid is not rejected across all the models. Therefore, our models are largely valid to accept their estimates for the purpose of inference.

220

The results of the disaggregation estimates (as robustness check) on human development in Sierra Leone are quite consistent with the original regression analysis previously presented.

The regression estimates for examining the relative importance of grant and loans on human development in Sierra Leone using both the full sample of African countries and SSA sample only confirms our original finding in spite of adding year dummies to the regression that grants rather than loans is the crucial aid type that improves human development in Sierra Leone. The disaggregated aid estimates show that grants have a highly significant impact in improving human development in the country, while aid in the form of loans does not. In model 1 of Appendix 7.5, grants have a highly significant impact whereas loans do not. In model 3 of Appendix 7.5 where we use the SSA sample, the significant impact of grants is also evident. Loans again could not significantly improve human development in the country. Worth noting however, is that when year dummies are added, the magnitude by which grants improve human development tends to be smaller than when year dummies are not included in the regression.

In the robustness check for the impact of technical assistance on human development in Sierra Leone (models 2 and 4 in Appendix 7.5), the estimates are also quite consistent with our original models with no year dummies. Aid in the form of technical assistance does not improve human development; with its direction even tending to be negative though insignificant. It is only aid used in the form of non-technical assistance that significantly improves human well-being when the analysis is extended in the case of Sierra Leone. In model 2, we add year dummies to the specification for the full African sample, and in model 4, we add year dummies to the model for the SSA sample. In both regressions, the results are quite consistent – that foreign aid in the form of technical assistance does not significantly improve human development in the case of Sierra Leone relative to Africa or sub-Saharan Africa. Hence, though total aid significantly improves human development in Sierra Leone, in its form of technical assistance, it does not. Hence, the aid disaggregation results are very much robust to the use of year dummies and change of the panel sample – that though total aid may have shown to be significant for improving human development in Sierra Leone, yet in its forms of loans and technical assistance, aid is not a significant determinant of human well-being improvement in the country.

221

7.2.2.4 Discussion

The regression results for disaggregated aid have shown that though total foreign aid seems to significantly improve human development in the case of Sierra Leone, it is not all types of aid that improve human development in the country. Aid in the form of Grant improves human development, but increases in loans and technical assistance aid types do not emerge to be associated with improvement in levels of human development in Sierra Leone. This merely confirms the aggregation bias of foreign aid and hence the importance of aid disaggregation in investigating the types of aid that can improve the welfare of the poor.

The literature on disaggregated aid and human development is far more limited than it is with the growth criterion of aid effectiveness. To our knowledge, only the study by Asiama and Quartey (2009) attempted a form of disaggregation of aid on human development for a panel analysis for sub-Saharan Africa. However, there are several questions on the reliability and applicability of their aid disaggregation results. Firstly, the study used bilateral aid as proxy for official aid and so disaggregation of bilateral aid does not sum up to total aid. In addition, the study is not clear as to whether the sum of sector aid and programme aid in fact amounts to total aid. Secondly, it is the sum of project aid and programme aid that sums up to total aid, and as project aid may not necessary be equal to sector aid, it implies these disaggregated aid types do not exactly disaggregate total aid. Thirdly, the disaggregated aid types are used in different estimations at a time and not within the same regression estimation, and it is therefore no surprise that they find possibly misleading results that show that though aggregated bilateral aid may be insignificant in improving human development in sub- Saharan Africa, yet its disaggregated forms of sector aid and programme aid are both significant in improving human development in the region. And fourthly, their study limited the disaggregation of aid to only project and programme aid as important disaggregates for welfare. Other aid disaggregation types such as grants and loans are also important for human welfare. With all these limitation of the disaggregation analysis by Asiama and Quartey (2009), we virtually cannot find much reliable findings for the estimation of disaggregated aid on human welfare. Masud and Yontcheva (2005) merely used bilateral aid (just as is done by Asiama and Quartey 2009) as proxy for official aid so cannot be classified as disaggregated aid estimations. As a result, our aid disaggregation estimates involving grants versus loans and technical assistance versus non-technical assistance on human development cannot be compared with the literature.

222

Irrespective of the unavailability of findings in the aid literature to compare the impact of our aid disaggregates on human development, there is still the need to provide possible reasons for our findings that aid in the form of grants promotes human development in the case of Sierra Leone, whilst loans and technical assistance do not. We generally do not find it surprising as to why grants should improve human development, but loans fail to do so in the case of Sierra Leone; and further that technical assistance, the form of aid modality that donors more recently seem to prioritise in many aid recipient countries including Sierra Leone fail to improve human development in the country. As support towards the reduction of poverty has a lot to do with charity and peaceful global co-existence, aid usually given towards such ventures are hardly attached with strings. Hence, it is no surprise that grants which usually have less or no conditions and repayment burden attached to it, foster human development in a largely poor country like Sierra Leone. Because of the long standing poverty that has characterised the country since independence, donors have had continued support towards poverty reduction in the country; and with this being clearly prioritised since the country’s civil war and post-war periods and as well as having its basis from the global MDG targets and PRSP undertaking in the late 1990s, it is therefore expected that the largely grant mode of aid disbursed by donors towards poverty reduction can be associated with higher levels of human development in the country. Loans on the other hand, because of their requirement for repayment and with interest, are usually associated with stringent donor conditionality and repayment burden such that their impact on poverty reduction cannot be guaranteed.

Generally, the use of loans to reduce poverty in aid recipient countries may be done either through microfinance schemes to economically active poor households or via the disbursement of loans for education and health. In Sierra Leone, some donors have sought to use loans to alleviate poverty via supporting government established microcredit programmes. For instance, in the post-war period, the National Commission for Social Action (NaCSA), a government parastatal set up with the responsibility of implementing poverty reduction programmes by the government and its development partners, had a specific programme, the Social Action and Poverty Alleviation programme that had as one its core goal, the provision of microcredit to economically active poor rural households with the ultimate aim of improving the income and livelihoods of such cohort of people and in effect reducing poverty in the country (Standard Times, 2008). This microcredit programme has 223 been funded by the African Development Bank (AfDB), a major donor to the country. Support of other donors beyond the AfDB has not been evident towards this poverty oriented scheme. It also remains to be seen whether in fact such microfinance is given to poor households or to those who can be assessed to be capable of repaying. There has been increasing discourse on the lenders’ preferred recipients of microcredit schemes. Many have argued that because lenders are cognizant of the need to recoup their capital and interest in the loans they disburse, their credit worthiness criterion in the selection of their beneficiaries implies, the real poor people are exempt from receiving loans since their repayment capability cannot be guaranteed. In such a situation, it is no surprise that if loans are disbursed via the giving of microcredit, it would not emerge to significantly reduce poverty. Hence, the often assumed role of microfinance in reducing poverty may have changed with the profit oriented intention of lenders who prefer giving loans/microcredit to the relatively better off businesses or individuals who have the ability to repay the loans with their added interests.

Further, the patrimonial practices of several African states including Sierra Leone, do not guarantee the non-interference of ruling governments in using the microcredit scheme as political campaign tool to gain and buy votes from recipients. Once the recipients/electorates are of the understanding that this loans/microcredit they receive is from government and for political purposes, the possibility for them repaying becomes slim. In effect, the donors may either terminate funding in response or that the scheme may no longer be viable to expand and reach other poor households. An interview with a representative of one of the country’s leading donors, Sisco, reveals that at some point in the 2007 presidential elections in Sierra Leone, the ruling governing party used the aid money meant for microcredit to poor fishing community to buy votes and in effect the loan was not repaid. The donors responded by terminating the project. This donor microcredit financing, according to him, was a loan to the government of Sierra Leone. This was also supported by fieldwork interview with Alagba, a leading media practitioner, who on his response to a question about the existence of neopatrimonial practices in the distribution of development resources including aid resources in the country simply puts it:

“The aspect of the PRSP that has political undertones was the microcredit aspect, wherein people were given credit; and during elections they were whispered to: don’t worry you will

224 not pay. Imagine the person who gives you the money is whispering to you about the issue of repayment ”

In effect, collection of loaned money is not ensured and therefore the poor beneficiaries may not even invest the loans on productive uses that will enhance their capital and build up their income to improve their well-being. The consequence is not only on the poor beneficiary, but also on the government who will have to repay the loan leading to entrenched indebtedness. No doubt Mann (2007: 1) quoted the Kenyan Nobel Laureate, Wangari Maathai speaking at the World Social Forum seminar in Kenya that :

“The debt burden continues to make it impossible for many governments to give services to the people .” Maathai continued that “ Indeed, the loans were used to oppress the people, strengthen the ruling and co-operating elites, and exploit resources at the expense of the health of the environment and the welfare of the people ” (Mann 2007:1).

Maathai justified her argument by suggesting that eight million people in Africa die each year from treatable diseases because they can’t access the required medicines, whilst their governments continue to service debt; and Joel Vilando, a member of the Philippines Parliament concurs to Maathai’s argument by pointing that “ debt payment gets priority over the provision of social services ” (Mann 2007:1-2)

The other avenue through which loans may reduce poverty as mentioned earlier is via direct support to education and health sectors, which are the welfare related sectors. However, because of the nature of loans disbursement in requiring repayment and with interest, it is difficult to see such aid type being disbursed to education and health sectors, which may only receive grants. In effect, it is hardly strange to find that grants rather than loans improve human development and hence the well being of the poor.

On the possible reasons for the failure of technical assistance in improving human development in Sierra Leone, one explanation could be that though technical assistance programmes may be targeted to transfer knowledge to aid-recipient countries, its form of administration is such that it does not significantly transfer knowledge. One of the key purposes of aid is to transfer knowledge to government bureaucrats in aid-recipient countries. Whilst this is a good policy, the fear is that knowledge is sometimes not actually transferred 225 to yield the intended benefit of technical cooperation aid. It is therefore hardly strange in such scenarios to find that technical assistance does not improve human development in aid recipient countries. This cannot be different for Sierra Leone. Begovic (2008) argues that the problem with technical assistance is usually that donors, especially the IFIs, become hostages of the recipient government to the extent that they (the donors) care less whether knowledge is transferred or not as the programme must go on. In addition, technical assistance in Sierra Leone has seemed to be largely spent on funding expatriates who despite their knowledge endowment, transfer less than expected knowledge and expertise to the local policy officials in the country. This was commonly pointed out by civil society groups, media bodies, and bureaucrats interviewed as part of this study. One leading media practitioner, Alagba, puts it

“Two-thirds of the donor funds go back to the donors themselves in the form of remunerations, in the form of salaries, vehicles or other amenities. Huge, huge salaries compared to the lower salaries than the indigenes claim. So if you have $30,000, $20,000 to be spent in those things and then only $10,000 to the project itself. So I wonder if you can call that effectiveness of foreign aid .”

Yet, a senior Parliamentarian of the Public Accounts Committee also recognises the issue of donor payback in the aid business in the country. While responding to a question about the relative effectiveness of direct donor-supported project aid as compared to general budget support, he pointed out that the problem with project aid is that donors chase back their monies mostly through recruitment of expatriates as consultant in aid projects in the country; and that from his assessment almost half of those funds go back to donors, which to him is not effective. Such assessment is also backed by another editor of a senior independent newspaper in the Country, Jonan, who laments that:

“Most of these agencies of donors send their own representatives to monitor the funding implementation but the disappointing aspect of donor funding is that half of their money go back to their staff .”

Whilst there are reports of donor gains in the aid business in the form of technical assistance towards expatriates, it is even more glaring with bilateral donors. An assessment of some memorandum of understanding agreements between the government of Sierra Leone and some bilateral donors reveal that in one 2009 project agreement for technical assistance, it is 226 stated in the agreement that the consultant should only be procured from the donor country with 80% of the cost of the project being payment for salaries and other fees to the expatriate and the remaining 20% of the fund being meant to purchase a vehicle and computer for the consultant. In effect 100% of the aid money goes back to the donor. In another US$30million project (loan) agreement to develop water supply to six communities in Sierra Leone, an assessment of the project document revealed that all the money is virtually going back to the donor in the form of consultancy fee for the consultant who must be a national of the donor country and the remaining of the funds being the cost of bringing equipments from the donor country itself. It is even reported from a senior official of Sierra Leone’s Ministry of Finance and Economic Development (MOFED) that usually what happens is that if a foreign expert does not validate any work done by Sierra Leonean consultants and experts, it is not valued by donors; and that most donors recommend consultants (who are usually donor country expatriates) to the country’s aid administrators.

It is therefore no surprise that technical assistance aid as has been mostly reported to be less important for development purposes in the country has also not shown in the regression estimations to be significant for improving human development in the country.

Representatives from Donor agencies interviewed however responded to this ‘finger- pointing’ by arguing that local expertise are usually not available and that even when donors advertise locally for technical assistance, they are either not available or not of the required standard. For instance, a representative from one of the country’s major donors, Joe, while responding to the issue of too much donor monies being spent on expatriates, puts it:

“Many a times people quarrel on this, even in government. …They say these donors bring monies and chase it with expatriates, but the fact of the matter is you do not have the technical expertise [locally] ”.

He identified three cases where they could not find local expertise when they advised for technical assistance. He however, admitted that the expenses on expatriates are considerably high even compared to monies the donor agency gives to the country as budget support. Thus, in as much as the intended purpose of technical assistance may not be bad for developmental purposes, the extent to which it is prioritised as well as its role in the transfer

227 of knowledge to local policy experts should be reviewed in order to make technical assistance more effective on development outcomes.

Hence, following our regressions, though increases in aggregate foreign aid tend to be associated with improvement in human development in Sierra Leone, the regressions show that increases in levels of human development does not respond to all types of disaggregated aid as though grants emerge to improve human development, aid in the forms of loans and technical assistance do not. Hence, for improving human well-being as a measure of poverty reduction in Sierra Leone, donors should disburse more of grants and less of loans and technical assistance to the country.

7.2.3 Foreign Aid, Politics and Human Development in Sierra Leone 7.2.3.1 The model and Empirical Results

After probing the effectiveness of foreign aid in both its aggregate and disaggregate forms in terms of its impact on well being in Africa, Sub-Saharan Africa and Sierra Leone, in this section, the study further investigates the importance of politics in influencing the impact of aid on human development in the case of Sierra Leone. Donors are usually sceptical about disbursing aid to fragile political regimes for the fear that such aid may be used to fund patrimonial practices; and hence not only being misappropriated but may also fund political corruption. Therefore, it is essential to investigate whether democratisation as a good political practice (and therefore likely to reduce political corruption) is not detrimental to aid’s impact in improving human development in Sierra Leone.

Two indicators of politics are used here - the quality of democracy and the quality of the political regime. These two variables are individually added to the aid-human development base model specification. The governance variable is replaced by either indicators of politics. This is so because governance and politics are intuitively related factors/attributes. In the POLITY IV measure of the quality of polity regime and level of democracy, some of the components of these variables happen to as well be direct or indirect components of governance as measured by the ICRG (e.g. quality of the bureaucracy and rule of law). To avoid problems of double counting therefore, both indicators of politics and governance

228 cannot be used together in the same regression estimation. As the purpose of this section is to assess the impact of politics on human well-being as well as its impact in influencing foreign aid’s impact on well-being, we therefore replace the governance variable in the base model with the indicator of politics. This is contrary to the specification of McGillivray and Noorbakhsh (2007) who added the measures of democracy and governance in the same regression. As justified previously, we deviate from that specification and rather replace the measure of governance with the measure of politics when we are interested in analysing the impact of politics on human development as well as its importance for aid effectiveness in Sierra Leone. Following the base model in equation (7.3), our modified specification for analysing the impact of politics on aid’s effect in improving human development is specified as:

Wit = d 0 + d 1GDPPC_1 + d 2PPE it + d 3Gm;it + d 4AID it + d 5POLIT + d 6DSL +

d7AID*POLIT*DSL + e it ….(7.4)

Wit = d 0 + d 1GDPPC_1 + d 2PPE it + d 3Gm;it + d 4AID it + d 5DEM + d 6DSL + d7AID*DEM*DSL + e it ….(7.5)

In equation 7.4, we use polity quality (denoted as POLIT) as a measure of politics sourced from the POLITY IV database (Marshall and Jaggers, 2008). It is the revised polity2 score in the POLITY IV database, specially constructed to enhance its use in time series analysis. This indicator is the combined score resulting from the subtraction of the autocracy score from the democracy score, and hence we interpret it as the extent to which a country’s political rule tends towards democratic institutions (i.e. democratisation). We therefore use it in this study as a measure of the quality of politics. This combined score ranges from a scale of +10 (strongly democratic) to -10 (strongly autocratic). The interaction term AID*POLIT*DSL is a three-way interaction comprising an interaction of foreign aid, polity quality and Sierra Leone country dummy and captures the impact of foreign aid under pro-democratic politics in Sierra Leone. In other words, it captures the importance of politics for aid’s impact in improving human development in Sierra Leone. When the interaction term is positive and significant, it implies that aid is effective under the conditions specified. The system GMM regression estimate for this specification is presented in model 2 of Table 7.4 below

In equation 7.5 above, we use the measure of Democracy (denoted as DEM) as another measure of politics sourced from the POLITY IV database (Marshall and Jaggers, 2008). This indicator is the simple score for institutionalised democracy comprising three 229 interdependent components. These components include the existence of institutions and procedures via which the citizens can express their choices about alternative policies and leaders (competitive elections), the presence of institutionalised limitations on the use of executive powers (constraints on the chief executive); and the assurance of civil liberties to all citizens in their political participation (openness of politics and freedom of expression). Other elements of plural democracies such as rule of law, system of accountability and press freedom are further considered as means to these general principles outlined (Marshall and Jaggers, 2002 in Teorell et al., 2009). This democracy score is scaled from 0-10, with 0 signifying low institutionalised democracy and 10 implying high institutionalised democracy. The AID*DEM*DSL interaction term comprises an interaction of foreign aid, institutionalised democracy and the Sierra Leone country dummy, and it captures the impact of aid on human development in Sierra Leone under the condition of democracy. In other words, it captures the importance of democracy on aid’s impact in improving human development in the case of Sierra Leone. The system GMM regression estimates for this specification are presented in model 4 of Table 7.4 below.

Table 7.4 below present the results of the diagnostic tests and regression estimates using the full African sample. The diagnostic tests show that in all the four models presented in the table, the regression estimates are largely reliable enough for inference. The crucial system GMM diagnostic tests involving the test for serial correlation (the AR (2) test) and the model validity test (the Hansen test for overidentifying restrictions) are both passed by the models. Hence, the System GMM estimator we employ is a consistent estimator since autocorrelation is not detected in the regressions; and the models we specify are valid since the Hansen test for validity of the GMM instruments we use shows that the hypothesis that the instruments used are valid is not rejected. Therefore, we find the regression estimates interpretable to draw valid conclusion on our findings of the relevance of politics on aid’s impact in improving human development in the case of Sierra Leone.

230

Table 7.4: Politics and aid effectiveness (full African Sample) Dependent Variable is HDI Model Model 1 Model2 Model3 Model4 Model5 Model6 Regressors HDI(-1) 0.199* 0.203* 0.204* 0.204* 0.191 0.194 (1.66) (1.69) (1.72) (1.71) (1.52) (1.55) AID -0.083 -0.084 -0.081 -0.077 -0.084 -0.085 (1.22) (1.27) (1.20) (1.16) (1.23) (1.27) POLIT 0.004 0.004 - - - - (1.24) (1.23) POLIT*DSL 0.023*** - - - - - (4.89) AID*POLIT*DSL - 0.009*** - - - - (6.83) DEM - - 0.009 0.009 - - (1.16) (1.15) DEM*DSL - - 0.063*** - - - (7.63) 0.020*** AID*DEM*DSL - - - (7.67) - -

AUT - - - - -0.012 -0.012 (1.48) (1.45)

-0.020** AUT*DSL - - - - (2.59) -

AID*AUT*DSL ------0.01*** (4.48)

PPE 0.083** 0.081** 0.077** 0.075** 0.085*** 0.083** (2.63) (2.57) (2.29) (2.23) (2.78) (2.71) ME -0.076* -0.078* -0.076* -0.077* -0.072* -0.075* (1.77) (1.79) (1.81) (1.84) (1.73) (1.76) GDPPC(-1) 0.106 0.106 0.106 0.112 0.104 0.104* (1.39) (1.42) (1.36) (1.44) (1.33) (1.36) DSL -0.301*** -0.301*** -0.458*** -0.471*** -0.293*** -0.268*** (4.13) (4.32) (6.00) (6.42) (3.96) (3.73) Constant -1.387** -1.372** -1.401* -1.443** -1.346** -1.335* (2.14) (2.17) (2.13) (2.22) (2.03) (2.06) No. Of Obs. 135 135 135 135 135 135 No. of Instruments 17 20 17 18 17 20 F-Stat 129.70 120.75 670.55 873.90 130.27 103.29 AR(2) 0.614 0.557 0.751 0.756 0.583 0.582 Hansen Test for (OIR) 0.251 0.577 0.322 0.373 0.284 0.590 Note :- All variables are in logs except ‘Polity regime’, democracy and autocracy variables and the country dummy. T statistic in parenthesis. *** significant at 1% ** significant at 5% and * significant at the 10%. AUT denotes autocracy score and AID*AUT*DSL denotes the interaction term of foreign aid, autocracy and the Sierra Leone dummy.

Before analysing the importance of politics for aid’s impact in improving human development in the case Sierra Leone, in models 1 and 3, attempt is first made to assess the significant impact of politics in advancing human development levels in the case of Sierra Leone. In Model 1 specifically, the impact of the quality of the polity regime is captured through the addition of a POLIT*DSL interaction term, whilst Model 3 captures the impact 231 of democracy in improving well being in Sierra Leone through the DEM*DSL interaction term.

In model 1, the GMM regression estimates show that improvement in human development in Sierra Leone for the period under study does respond to higher quality of the polity regime. The POLIT*DSL interaction emerges to be positive and significant, implying that improvement in the quality of the polity regime significantly relates with an improvement of well-being in Sierra Leone. Likewise, institutionalised democracy is found to be important for improving well-being of the poor in Sierra Leone. The GMM regression estimates as presented in Model 3, shows that DEM*DSL interaction term is significant at the 1% level of significance, indicating that institutionalised democracy as a form of political rule can significantly improve human development in Sierra Leone. Hence, both the quality of the polity regime and institutionalised democracy as measures of politics tend to directly improve human development in the country.

More important for the purpose of this study, in Models 2 and 4 of Table 7.4 above, attempt is made to capture the importance of politics on aid effectiveness in Sierra Leone in terms of improving the well-being of the poor in the country. In model 2, an AID*POLIT*DSL interaction term is added to capture the impact of aid in improving human development in Sierra Leone under conditions of improving quality of the polity regime. The results show that foreign aid can be effective in advancing human development in Sierra Leone when the quality of politics is improved. The AID*POLIT*DSL interaction term emerged to be significant at the 1% level of significance. Hence, improvement in the quality of the polity regime is important for aid’s effectiveness in promoting human development in the country. It does not however imply that aid cannot be effective in improving human development if the polity regime is not improved; as previous analysis showed that aid does have a direct positive impact in improving human development. This result only ensures that improving the quality of political regime (towards a pro-democratic regime) is not bad a policy in making foreign aid foster human development in Sierra Leone.

Likewise, model 4 considers the exclusive measure of institutionalised democracy in assessing the importance of politics in aid’s impact on human development. This is captured with the addition of the AID*DEM*DSL interaction term. The regression estimates show that AID*DEM*DSL interaction is also significant and positive, implying that improving the 232 level of democracy in Sierra Leone is not bad a policy for the significant performance of foreign aid in improving human development in the country. An interesting finding here is that though both measures of politics emerge to be significant for aid’s impact on human development, yet the pure institutionalised democracy measure tends to perform better than the polity quality measure. The AID*DEM*DSL interaction term has a greater marginal effectiveness (as is shown by the magnitude of the coefficient) than the AID*POLIT*DSL interaction term. This implies that though the tendency to move away from autocratic to democratic rule (as is measured by the quality of polity regime) may be a good policy for making aid effective in terms of improving human welfare, yet the improvement in the quality of institutionalised democracy is a better policy option to aid’s impact in improving well-being of the poor in Sierra Leone.

In models 5 and 6 of Table 7.4 above, the study further attempts to test whether the corollary is true that in as much as the improvement of democracy is not a bad policy option for directly improving human development and for aid’s impact on human development in the case of Sierra Leone, autocracy on the other hand should have a detrimental effect. In model 5, the regression to assess the impact of autocratic form of political rule showed that in the case of Sierra Leone, autocracy has proved to have a detrimental effect on improvement of human development and hence well-being in the country. The AUT*DSL interaction term which captures the impact of autocracy on human development in the case of Sierra Leone emerges to be negative and significant at the 5% level of significance to imply that autocracy has an adverse effect on the improvement of human development in the case of Sierra Leone. In model 6, the GMM regression to examine the detrimental effect of autocracy on the effectiveness of foreign aid in improving human development in the case of Sierra Leone confirmed that indeed autocracy has a detrimental effect. The AID*AUT*DSL interaction term emerges to be negative and significant to imply that autocracy as a form of political rule can be damaging on the effect of aid in improving human development in the country. This result further confirms why the institutionalised democracy measure has a higher marginal effectiveness than the polity quality measure in terms of their relative influence on the impact of aid on human development as previously shown. As polity quality is a combined measure of autocracy and democracy, it implies the relative effect of polity quality on the effect of aid on human development could be expected to be lesser than that of the pure institutionalised democracy measure.

233

7.2.3.1 Robustness To check for robustness of the finding on the importance of politics on aid effectiveness in Sierra Leone with respect to the poverty criterion of aid effectiveness, the study used the same models in the previous analysis on a sample of Sub-Saharan African countries instead of full African sample. The results are shown in Appendix 7.6. In model 2, the study assesses the importance of the quality of politics on aid’s impact on human development in Sierra Leone through the AID*POLIT*DSL interaction term. The results show that even using the sample of SSA countries, the results are very much consistent to our previous use of the full African sample. The AID*POLIT*DSL interaction term is significant at the 1% level and with the marginal effect (coefficient of the AID*POLIT*DSL term) being almost identical. This implies that though foreign aid is itself significant in directly improving human development in the case of Sierra Leone, yet politics geared towards the promotion of better quality of the political regime is no bad policy measure for aid’s impact in improving human development in the country.

Similarly, in Model 4, we assess the importance of democracy as our indicator of politics in terms of its importance for promoting aid’s impact in fostering improved human development in Sierra Leone. The result of this regression is also consistent with the use of the full African sample as discussed previously. The AID*DEM*DSL term which captures this effect emerges to be highly significant and with a marginal effect almost identical with the one obtained when the full African sample was used. This implies, the improvement of democracy in the country is not bad a political reform measure for aid’s effectiveness in improving human development in Sierra Leone. The finding that democracy is better than the quality of politics for aid’s impact on human development as was observed in the main analysis is also consistent with the use of the SSA sample. The AID*DEM*DSL term has a higher marginal effect (magnitude of the coefficient) than he AID*POLIT*DSL interaction term. Even when year dummies were added to the models as further robustness check, the results are very much consistent (These regression estimates are not reported but available upon request from the author).

In a similar vein, the estimation to investigate whether autocracy has a detrimental effect on the effect of aid on human development in the case of Sierra Leone also proved to be consistent with the previous analysis using the full African Sample. Model 6 of Appendix 7.6 shows that the AID*AUT*DSL interaction term is negative and significant to imply that 234 autocracy as a form of political rule can have a detrimental impact in improving human well- being in Sierra Leone.

Hence, the findings that political reforms towards more democratic form of politics are not detrimental to aid effectiveness in terms of improving human development and that pure democracy is a relatively better indicator than the combined polity quality in terms of their influence on aid effectiveness in Sierra Leone, are robust to changes of the panel sample and specification. Therefore, an improvement in the quality of politics (or democratisation) in Sierra Leone is not bad a policy for aid to have an impact in improving human development in the country.

7.2.3.2 Discussion

The GMM regression analysis of the importance of politics on aid’s impact in promoting human development in Sierra Leone revealed that though foreign aid and democratic politics independently improve human development in the country, yet reform measures aimed at enhancing the quality of democratic politics in the country are not detrimental to making foreign aid improve human development in Sierra Leone. Our findings for Sierra Leone appear to support the cross-country findings that had existed in the limited literature on aid, politics and well-being. In particular, the study provides some evidence to support the findings by Kosack (2003) that foreign aid is significant in improving human development under good democracies. It however does not support his conclusion that aid’s impact in improving human development is only contingent under good democracies. In as much as our study shows that conditions of good democracy are not detrimental to aid’s impact in improving human development, we find that aid can directly improve human development irrespective of the quality of political rule or strength of the democracy. McGillivray and Noorbakhsh (2007) though laid much emphasis on the link between aid, conflict and human development, they made some attempt to investigate the link between aid, democracy and human development and rather find that good democracies do not improve the impact of aid in promoting human development in aid-recipient countries.

Hence, whilst our analysis for the link between aid, democracies and human development for Sierra Leone may have not provided support for the cross-section analysis by McGillivray and Noorbakhsh (2007), it partly provides support for the panel analysis by Kosack (2003)

235 that the promotion of democratisation is a good policy action for aid’s impact in improving human development. We however differently provide evidence, that irrespective of the democracy condition, aid remains an important determinant for the improvement of human development in the case of Sierra Leone. Even when we additionally use the combined measure of the quality of the polity regime as a measure of the extent to which a political regime moves from an autocratic towards a more democratic form of political governance, yet politics remains a non-detrimental factor for aid’s impact on human development. This may be explained that the existence of more democratic form of political governance encourages the inflow of increased donor aid and increases aid spending as well as government budgetary spending on social sectors, which ultimately will resort to improved human welfare. With respect to our finding on the direct impact of aid on human development and the link between aid, politics and human development in the case of Sierra Leone, we conclude that, for the case of Sierra Leone, though the promotion of good quality politics is important for aid’s impact in improving human development, aid itself is significant in improving human development in the country irrespective of the quality of the political regime or the level of democracy.

7.3: Conclusion

Poverty reduction has emerged to be the widely recognised target of development aid particularly following the commissioning of the MDGs by the international community. The importance of poverty in development assistance is further enhanced by the decision of donors that HIPC countries formulate a PRSP as a requirement for debt relief and/or the further disbursement of more aid from international donors. As a correlate of income poverty, the improvement of human welfare (or well-being) expressed in the forms of improvement of human development and the reduction of infant mortality rate has increasingly emerged as the development outcomes upon which to measure the effectiveness of development aid. Hence, in this thesis, we attempt to contribute to this limited but increasingly popular section of the aid effectiveness literature by using the more reliable estimation technique of system GMM to examine this relationship for Africa, sub-Saharan Africa and in more detail for Sierra Leone.

In this chapter, we have investigated the direct impact of foreign aid on human development and infant mortality rate as indicators of human well-being in the case of Africa, SSA and

236

Sierra Leone. We further disaggregated foreign aid into grants and loans, as well as technical assistance and non-technical assistance to further examine which types of aid better improve human welfare in the case of Sierra Leone. We lastly looked at the link between aid, politics and human development for the case of Sierra Leone with the motive of finding out whether democratisation process in Sierra Leone is not detrimental to the impact of aid on human development in the country.

The results of our estimations reveal some interesting findings. We find that foreign aid does not significantly improve human well-being in Africa and sub-Saharan Africa for the period 1980-2007. Specifically, aid is not found to be significant in improving human development and reducing infant mortality rate in Africa and Sub-Saharan Africa. However, though aid may have not significantly improved human well-being in Africa, it is found to significantly improve human development in Sierra Leone, though it does not significantly reduce infant mortality rate in the country. In our estimation of the impact of disaggregated aid on human development in Sierra Leone, we find that grants significantly improve human development in the country while loans and technician assistance could not. Finally, for our investigation of the link between aid, politics and human development in Sierra Leone, we find that though aid is significant in directly improving human development in the country, yet pro- democratic politics is also good a policy option for aid’s impact on human development in the country. The corollary is however true, that more autocratic regimes have a detrimental effect on the role of aid in improving well-being in country. In this respect, we argue that as opposed to the findings from the literature (Kosack, 2003) that aid is only significant in improving human development under conditions of good democracy, in the case of Sierra Leone, in as much as the promotion of democratic politics is not bad a policy option for foreign aid to improve human development, yet aid can directly improve human development irrespective of the quality of the politics in the country.

On overall, we find our estimates to be robust to the change of the sample from a full African sample to that of sub-Saharan Africa, as well as robust to the changes of the specification through the addition/deletion of year dummies.

237

CHAPTER 8: CONCLUSION AND POLICY IMPLICATIONS 8.1 Introduction

The ‘big push’ for foreign assistance to low-income countries has only continued to gain momentum, and particularly for the lagged continent, Africa, where economic performance and poverty reduction remain unimpressive. Part of the big-push has been facilitated with the setting and financing of global development targets, mainly the Millennium Development Goals as at 2015. Much research has shown that in as much as most regions of the world are on course to achieve these targets, Africa has rather remained off course. Hence, more resources, particularly foreign aid are being scaled up for Africa to aid its development. With more foreign assistance to such underdeveloped continent, it is believed that investment and therefore economic growth alongside human welfare will be significantly improved.

However, in as much as aid is postulated to improve growth and welfare in recipient countries, empirical evidence on this line has been mixed and hence the effectiveness of foreign aid has remained debatable. A review of the aid effectiveness literature reveals that whilst a considerable number of studies have shown that aid works in terms of promoting economic growth and in others in terms of reducing poverty, yet some studies have argued and empirically proven that aid does not achieve its development purpose. Yet, others have shown that the effectiveness of aid is contingent on recipient country characteristics. Much of the research on the effectiveness of aid has however vastly focused on cross-country analysis which have been criticised for not (sufficiently) accounting for country heterogeneity (Murthy et al., 1994; Solow, 2001; Mavrotas, 2002; Feeny, 2005; Bigsten et al., 2006; and Temple, 2010). Most of these critics have called for a focus on country studies to better analyse the effectiveness of foreign aid, considering the distinct country circumstances and as well as to enable better comprehensive analysis of aid effectiveness. Once analytical, and theoretically informed, country case studies are viewed to be more illuminating than cross- country studies (Temple, 2010).

Hence, this study uses a country analysis for the case of Sierra Leone to examine the impact of foreign aid on economic growth and poverty reduction as well as the impact of recipient country characteristics on the effectiveness of aid. The Sierra Leone case study is especially unique in the sense that the country has been among the top aided countries in the world, thus making it inevitably an aid-dependent economy. At the same time, this country has been

238 characterised by dismal economic performance and poverty levels to the extent it remains one of the world’s poorest countries lying within the bottommost end of the UNDP’s HDI as well as with one of the highest infant mortality rates in the world. Hence, it seems a paradox that the country remains highly and continuously aided and yet with dismal development outcomes. In spite of this seeming paradox, there remains no country study on the effectiveness of aid in terms of its impact on economic growth and poverty reduction in the country.

Therefore, the purpose of this study is to examine the relationship between the intervention of donors (in terms of foreign aid disbursed), economic growth and poverty reduction in Sierra Leone for the period 1970-2007. The study further examines the importance of the quality of politics for the effectiveness of aid on human welfare (as a correlate of poverty reduction) in the country. Hence, the study specifies three objectives. The first objective examines the impact of foreign aid on economic growth in Sierra Leone. Two specific research questions are explored with respect to this objective. The first is whether foreign aid significantly fosters economic growth in Sierra Leone. The second research question with respect to the aid-growth objective is whether aid disaggregation maters in examining the impact of aid on economic growth in Sierra Leone. In answering these specific research questions, the methodology involved a triangulation approach comprising the application of two advanced time series cointegration techniques with the advantage of obtaining more robust results and valid conclusions on the relationship between aid and growth at the country level for Sierra Leone.

The second objective of the study involved an investigation of the impact of foreign aid on poverty reduction in the case of Sierra Leone. Here, two strands of poverty dimensions are examined. First, the study investigates the impact of foreign aid on pro-poor growth as a measure of poverty reduction. The technique of investigation involves the quantitative use of triangulated times series cointegration approaches as previously employed in the aid-growth objective. Here as well, the study further examines aid aggregation bias in the impact of aid on pro-poor growth in the country. In particular, the role of food aid as well as aid in the forms of grants and loans on pro-poor growth is given prominence. The second strand of poverty reduction is that related to well-being of the poor. Hence, the study investigates the role of aid on human well-being as proxy for poverty reduction for Africa, Sub-Saharan Africa and Sierra Leone in particular. In this investigation, a dynamic panel data analysis 239 approach is employed, with the specific effect on Sierra Leone captured by using an interaction of a Sierra Lone country dummy with the variable of interest. Again, the importance of aid disaggregation in assessing the impact of aid on human well-being in the case of Sierra Leone is further assessed. In a further form of triangulation approach, interviews from fieldwork are brought in to explain the findings.

The final objective of the study looks at the investigation of the role of politics on aid effectiveness at the country analytical level for Sierra Leone. The investigation is limited to the role of politics on the impact of aid on human welfare in the case of Sierra Leone relative to a sample of African countries. To address this objective, dynamic panel data analysis, with an interactive country dummy interacted with the variables of interest is employed (to specifically capture the impact on Sierra Leone) just as is done in the analysis of the aid- welfare objective.

Overall, the general strategy employed for analysing the relationship between donor intervention, economic growth and poverty reduction in the case of Sierra Leone had involved the use of pluralistic techniques of analysis. The aim is to triangulate the analysis to obtain more robust findings on aid effectiveness in Sierra Leone and for which conclusions reached can be largely relied upon.

The remainder of this chapter provides the summary of the findings as well as the conclusion and policy implications for each of the objectives of the research. In doing this, the second objective on the impact of foreign aid on poverty reduction is further split into the two strands of poverty reduction. The final subsection highlights the limitation of the study and suggests areas for future research.

8.2 Foreign Aid and Economic Growth – Summary, Conclusion and Policy Implications

The study’s examination of the relationship between foreign aid and economic growth in Sierra Leone using a triangulation of two time series cointegration estimation techniques over the period 1970-2007 reveals that foreign aid is a significant determinant of economic growth in the country. This result is found to be robust against specification changes and across the

240 two estimation methodologies employed. Both the ARDL bounds test and Johansen cointegration approaches show that foreign aid is an important determinant of economic growth in the country. This is also evident in both long-run and short-run analyses. However, the economic significance of aid on growth is higher in the long-run than it has in the short- run. The long-run ARDL estimates show that for every 10% increase in foreign aid, economic output will rise by 4%; whilst the short-run estimates show that a 10% increase in foreign aid is associated with a 0.7% rise in economic output.

In accordance with the theoretical review of the literature, our finding of a significant impact of aid in fostering economic growth in Sierra Leone provides support for the supplemental theories of foreign aid and as explained by the two-gap model of Chenery and Strout (1966), which posits that through the supplementation of limited savings in recipient countries and via the filling of foreign exchange gap needed to support the import requirement for productive purposes, foreign aid should promote economic growth in poor recipient countries. Empirically, this research finding provides support in just another country case study in favour of the existing studies that prove a positively significant relationship between foreign aid and economic growth. In particular, it supports the findings by Murthey et al. (1994), Gounder (2001), Lloyd et al. (2001), Mavrotas (2002), and Bhattarai (2009) who used different country case studies to show that aid promotes economic growth. On the other hand, the finding disputes those by Islam (1992), Mbaku (1993), Feeny (2005) and Javid and Qayyum (2011) who also used different country case studies but rather could not find strong evidence to show that aid promotes economic growth.

Whilst methodological and sample differences cannot be ruled out for generally finding differences in the results for the impact of foreign aid on economic growth, in the case of Sierra Leone, we argue that our finding of a positively significant impact of aid on growth is justified in the proposition of the supplemental theory of aid as previously explained. Perhaps in particular, the differences may arise on the main purpose for which aid is given to a recipient country. With Sierra Leone being known for one of the poorest nations in the world, we argue that aid given to Sierra Leone is largely on the basis of the country’s poverty standing and hence if the purpose of aid is for economic development and poverty reduction, then it could well be that its purpose could be met since it is largely targeted on developmental objectives.

241

The second research question answered within this objective is that related to the importance of disaggregation in the assessment of the impact of aid on growth. By disaggregating total aid into its various forms and sources, we find results that confirm the existence of aggregation bias and hence suggest that in as much as total aid contributes to growth, it is not all types of aid that influence economic growth in the country. In particular, the findings suggest that in the case of Sierra Leone, multilateral aid and technical assistance do not significantly foster economic growth. The impact of loans on economic growth is only found to be moderate for the long-run results, but truly significant in the short-run. These results have also been found to be robust to specification changes and across the use of different methods of estimation. It is further observed that when foreign aid was disaggregated into the various aid types and sources, the estimation results seem to show that the aid types and sources that appear to be relatively larger as per disbursement tends to be more effective.

These results on the relationship between foreign aid and economic growth in Sierra Leone have important policy implications for donors, recipient authorities and researchers at large, as well as for the improved effectiveness of aid. First, the finding that foreign aid is significant in fostering economic growth in Sierra Leone implies that the country’s persistent poverty characterisation amidst notable donor presence and participation in the economy has little to do with the fact that aid has not been effective in fostering the country's economic growth, but it may be that the magnitude of the effect has not been that high to completely eradicate poverty. Higher elasticity/responsiveness of growth to increasing aid effort must be evident to guarantee notable economic impact of aid on growth in recipient countries including Sierra Leone. However, compared with other determinants of growth in Sierra Leone, foreign aid has the highest elasticity. Therefore, it remains the most important determinant of growth in Sierra Leone alongside private investment and private property rights. Thus, if economic growth is the primary objective of donor intervention in Sierra Leone, then donors should continue in their effort to disburse more aid to Sierra Leone as it tends to achieve the intended aid target. The results may also imply that whilst foreign aid remains vital for the country’s development, depending on aid alone will not boost long-term growth to levels that the country will speedily catch up with the rest of the world. Hence, there is need to encourage private investment and secure property rights (which have also been found to be growth enhancing) alongside increased aid effort in order to raise growth levels to enable the country to catch up with the rest of the world.

242

Second, the study’s finding that aggregation bias exists in our examination of the impact of foreign aid on growth in the country implies there is need to disaggregate aid in order to identify the most effective types and as well as those that are less effective in order to improve on the net effectiveness of aid especially on its economic significance in the country. The finding that aid in the form of technical assistance does not seem to work in the case of Sierra Leone suggests the need to look into the ways in which technical assistance is delivered as well as a critical review of its expected role of transferring knowledge. This is even more necessary because of the frequently mentioned limitation of absorptive capacity in recipient countries. Hence, the knowledge transfer objective of technical assistance in order to improve the capacity of national policy expert to be able to handle high aid intensity in the country should be prioritised as opposed to the current donor prioritisation of exorbitant expenditures of technical assistance towards the employment of expatriates from donor home countries who largely do not tend to transfer their technical knowledge as expected. In effect, donors are chasing back their money. Also, that bilateral loans emerge to be more effective as compared to multilateral loans only signals a caution to the critics of bilateral aid that if such aid is disbursed with poverty reduction and economic development as its primary objective (as it seems to be in the case of Sierra Leone), it can as well be effective. Finally, our finding that the aid types and sources that are of higher magnitude (upon disbursement) tend to be more effective than aid types that are less disbursed implies that the higher the amount of aid disbursed relative to GDP, the higher the effectiveness; and which could as well imply there is the need for more aid since higher aid intensity tends to be associated with higher effectiveness.

8.3 Foreign Aid and Pro-poor growth – Summary, Conclusion and Policy Implications

Part of the objective of investigating the impact of foreign aid on poverty reduction in the case of Sierra Leone, involved the exploration of the role of aid in promoting pro-poor growth in the country. Using agricultural growth as proxy for pro-poor growth as is commonly argued in the literature (because it involves income growth in the sector where the poor are mostly employed), we find results just as with the aid-growth analysis that tend to support the supplemental argument of foreign aid. The results generally show that foreign aid significantly fosters long-run pro-poor growth in the country and with foreign aid in fact

243 emerging as the single most important factor in promoting agricultural growth which proxy pro-poor growth. However, in the short-run, the impact of aid on agricultural growth could not be confirmed by the analysis.

As opposed to the finding with economic growth, private investment does not seem to relate to pro-poor growth, thus making foreign aid the single most important long-run determinant of pro-poor growth in the country amongst the explanatory variables used in our parsimonious model. Here again, the importance of political stability is crucial for pro-poor growth, as the regression results consistently show that political instability as denoted by the civil conflict in the country is a crucial adverse factor to pro-poor growth in the country. These results are found to be robust against changes in the regression specification as well as across the use of different estimation methods. The study’s triangulation of estimation methods involving the ARDL and Johansen approaches both show consistent results with regards the impact of foreign aid on pro-poor growth in the country. These results tend to support the studies by Akpokodje and Omojimite (2008) for Nigeria and Feeny and Ouattara (2009) in a cross country study- that foreign aid promotes pro-poor growth.

In a further examination of the relationship between foreign aid and pro-poor growth by disaggregating total aid, the existence of aggregation bias, just as was evident in the aid- growth analysis, has also been evident here. In as much as total aid is found to significantly foster pro-poor growth in Sierra Leone, not all types of aid could confirm likewise. When total aid is disaggregated into grants and loans, the latter could not prove to be significant for determining pro-poor growth in the country. Also, technical assistance as a form of donor assistance could not prove to be vital for pro-poor growth in the country. Food aid, though not quite significant, yet proved to be moderately significant for pro-poor growth in tune with the study’s proposition.

These findings have important policy implication for donors, national policy officials and researchers at large. First, that aid promotes pro-poor growth in the country implies that the strategy that had been used to utilize aid to the agricultural sector appears to be in place and so is the aid enhancing environment in the sector. Hence, the aid disbursement and utilization strategy, and policy and institutional environment that enhance the effectiveness of aid on pro-poor growth should be continued as such prove to support the long-run effectiveness of aid in being pro-poor. 244

However, that aid in the short-run does not foster pro-poor growth is worth considering as ineffectiveness of aid in the short-run may be costly to the rural poor and farmers who are quite vulnerable in situations of short-term shocks in the economy and during natural disasters.

Second, the finding that aid in the form of grants rather than loans is pro-poor provides support for the stance seemed to have been taken by the donor community that the disbursement of loans in a typical fragile state is non-effective and does not yield the desired purpose than does grants. This arguments seems to be true at least in the case of Sierra Leone, a fragile poor state where the results have shown that loans do not foster pro-poor growth as opposed to grants. This may imply that the repayment requirement and debt burden associated with loans may be detrimental to the performance of the agricultural sector which largely employs the poor. This is also justified by Cordella and Ulku (2004) who provide evidence and argue that the provision of more aid (but with less concessional aid package) to low income countries has a detrimental impact on not only their immediate economic development, but also in the future through the accumulation of unsustainable debt burden. It may as well imply the results of the significant impact of foreign aid on pro-poor growth may have been largely as the results of the disbursement of grants which also appears to be in larger quantum compared to the quantum of loans disbursed to the country.

Third, that foreign aid in the form of technical assistance does not improve pro-poor growth in a least developing country like Sierra Leone implies there is the need to revisit the way technical assistance is disbursed and managed in the country as has been suggested in the previous objective.

Fourth, that food aid moderately improves agricultural growth and hence pro-poor growth implies the traditional disincentive argument against food aid is not entirely true at least in the case of Sierra Leone. That food aid has not shown to have a detrimental impact on pro- poor growth implies this form of aid may have been disbursed in the country amidst periods of food insecurity and inadequacy to the extent it turns out to rather promote agricultural growth and hence pro-poor growth. It may have also been used as budget support partly to the agricultural sector thus consequently improving growth. But also that its long-run impact could only turn out to be moderate and neither is it evident in the short-run implies the disincentive argument may not have been completely overruled. In essence, the volumes and 245 management of food aid in the country has not been that damaging to the agricultural sector as has been often argued; it in fact has at least a moderate effect in improving pro-poor growth in the country. Hence, donors and the country authorities should encourage the inflow of food aid to the country at manageable levels that cannot be damaging to the farmers in the agricultural sector, but also at levels that could support food security in the country in the context of availability and affordability of a variety of food stuff to the citizenry.

Finally, as the results, both for the examination of the impact of aid on pro-poor growth and even overall economic growth have shown that the impact of foreign aid tends to be more evident in the long-run than in the short-run, aid sceptics should be more cautious in quickly concluding that aid is ineffective when their analysis could not have attempted to capture the long-run impact of foreign aid. This research consideration is also in conformity with the argument by Temple (2010: 4448) that “Although the attempt to quantify the effect of aid be central to this literature, it is hard to draw conclusions when the distinction between short-run and long-run effects is so rarely made”. Hence, research on the effectiveness of foreign aid should attempt to conduct both long-run and short-un investigation of the impact of aid before arriving at conclusion with regards its effectiveness.

8.4 Foreign Aid and Aggregate Welfare – Summary, Conclusion and Policy Implications

In the poverty reduction objective of foreign aid, a larger portion of it has to do with improvement of human welfare as a correlate of poverty reduction. The formulation of international development targets in the late 1990s and their commissioning in the beginning of the millennium just indicate the prioritisation of poverty reduction in development assistance. And as much of the international development targets have to do with human welfare improvement by 2015, it justifies the importance of assessing the role of aid on human welfare as an increasingly recognised target of development aid.

Using GMM panel regression analysis for Africa, Sub-Saharan Africa and specifically, Sierra Leone, the study’s investigation of the impact of foreign aid on aggregate human welfare yield some interesting findings. For the period 1980-2007, the study finds that foreign aid does not significantly improve human welfare in Africa and Sub-Saharan Africa. Both indicators of human welfare employed, comprising of the human development index and the

246 infant mortality rate do not turn out to significantly respond to aid intensity. This implies we could not find evidence to show that foreign aid improves human development and neither does it significantly reduce infant mortality rate in Africa.

However, using the same sample of Africa and Sub-Saharan Africa and further restricting the analysis to the case of Sierra Leone, some of the findings emerge to change. Though the study still could not find evidence to show that foreign aid significantly reduce infant mortality rate in Sierra Leone, yet it does significantly emerge to improve human development in the country for the same period of analysis. Thus, whilst the insignificant impact of foreign aid on infant mortality is consistent for Africa and Sierra Leone, the impact of foreign aid on human development rather turns out to be positive and significant in the case of Sierra Leone as opposed to Africa as a continent. In a further investigation for aggregation bias, the study disaggregated total foreign aid and explored which types to aid are most effective in improving human development in the country context. The results show that as was the case with economic growth and pro-poor growth, grants as a form of aid outperforms loans. Whilst grants emerge to significantly improve human development for Sierra Leone, we could not find similar evidence for concessional loans. Technical assistance, again, could not emerge to significantly improve human development in the country, just as was found in the economic growth and pro-poor growth analysis.

These findings have important policy implications. That aid is effective in improving human development in Sierra Leone suggests that increased foreign aid flows to the country should be encouraged as that is crucial for reducing the high poverty levels prevalent in the country. It also implies that the comparatively low human development levels in Sierra Leone could have been worse than what statistics show had it not been for donor intervention in the country. As this result tends to support a common conclusion from the studies by Gomanee et al (2005a; 2005b) that the lower the income of the country, the better the effectiveness of aid in terms of improving human development, this only provides support for increased aid flows to low-income countries as aid tends to be more effective in such economies. As Gomanee et al (2005a) explains, this may be probably due to the supplementation of their limited income to support the poor or poverty reducing projects/activities. The capacity for more aid to support poverty reduction in low-income countries is more obvious the lower the development status of the country; a situation which may have been evident in the case of

247

Sierra Leone as average HDI and IMR values are comparatively lower than those for Africa and Sub-Saharan Africa.

However, that foreign aid does not significantly reduce infant mortality rate in Africa, SSA and Sierra Leone in particular implies that if aid is to reduce infant mortality rate, national authorities together with the country’s development partners should restructure the aid management and targeting strategy that had existed in the country and the continent in general since the 1980s to a form that can directly target the reduction of infant mortality. In the case of Sierra Leone, it means the recent policy by the government and donors to provide significant flows of aid (through the 2010 free health care programme) to directly target the reduction of infant mortality rate and rescue maternal mortality is in the right path provided it is effectively monitored. And if successful, such a programme intervention could be replicated to other low-income countries with seemingly high infant mortality rates.

Finally, the finding that not all types of aid improve human development tells the existence of aggregation bias in the aid effectiveness discourse. That foreign aid in the forms of loans and technical assistance could not establish a significant impact in improving human development in Sierra Leone implies the volume and nature of disbursement and management of such aid types should be revisited if aid is to improve on its economic effectiveness in Sierra Leone. It implies the positive significance of aid on human development for Sierra Leone as shown by the study may have been largely the results of aid disbursed in the forms of grants and non- technical assistance (both of which are coincidentally observed to be higher in volume and intensity). Hence, whilst the delivery of more grants and non-technical assistance aid should be encouraged, the disbursement of loans and technical assistance should be revisited for higher economic significance of overall aid to the country; and particularly if the target of development aid is to improve human well-being which has remained in deplorable levels in the country. It may not be the right policy option for donors to cease the disbursement of loans and technical assistance to low-income countries like Sierra Leone, but the manner in which such aid types are managed could be revisited just as there is the need to find appropriate strategies to cushion the possible adverse effects loans may have on the poor. Technical assistance is in general vital for development and particularly long-term human development, but if its intended purpose of transferring knowledge and improving the capacity of recipient country policy officials is not seemingly met possibly due to donor prioritisation of technical assistance towards financing expatriates from donor home 248 countries, it may be worthwhile to consider other options for which technical assistance could be better utilized to enhance its intended purpose and consequently improve the effectiveness of foreign aid.

8.5 Politics, Foreign Aid and Human Welfare- Summary, Conclusion and Policy Implications

Country recipient characteristics have become increasingly recognised as important for making aid effective in recipient countries, particularly since the 1990s when aid effectiveness research seemed to have realised this. The importance of recipient country politics as a factor to development and aid effectiveness has particularly been recognised by the international community. But whilst this is realised, not much research has focused attention on it, particularly at the country level. Using a cross country analysis, but focusing on a single country context for Sierra Leone, the study examines the role of politics on the impact of aid on human development.

The study’s investigation of the link between aid, politics and human development in Sierra Leone, finds that though aid is significant in directly improving human development in the country (as has been shown in the previous analysis), yet tending towards pro-democratic politics is also good a policy option for aid’s impact in improving human development in the country. The corollary is however true, that more autocratic regimes have a detrimental effect on the role of aid in improving well-being in country. In this respect, we argue that as opposed to the findings from the literature (Kosack, 2003) that aid is only significant in improving human development under conditions of good democracy, in the case of Sierra Leone, in as much as the promotion of democratic politics is not bad a policy for foreign aid to improve human development, yet aid can directly improve human development irrespective of the quality of the politics in the country. Hence, whilst aid is not dependent on good politics to be effective, yet good politics in the context of more democratic rule is not detrimental to the impact of aid on human development in Sierra Leone.

This finding has an important policy implication. That the promotion of democratisation in the country is not detrimental to the impact of aid on human development, implies that the country together with its development partners should proceed with the democratisation

249 process in Sierra Leone as it is not only important for directly raising human development levels, but also important for the continued impact of aid in raising human development levels in the country. This is particularly true because the investigation also showed that more autocratic regimes have an adverse effect on the effectiveness of aid in improving human welfare. Therefore, the move away from autocratic forms of rule towards democratisation is vital for the impact of foreign aid in significantly improving human development in Sierra Leone.

8.6 Limitations and Directions for Future Research

It is essential to indicate that this research suffers from some possible limitations. Firstly, findings on the impact of aid on economic growth as well as on pro-poor growth are limited to just one country, Sierra Leone. And though this is a macro level study, but since each country may not necessarily be the same, generalisation of policy implications across countries may not be quite applicable. Secondly, the research’s use of country interactions amidst panel regression was largely as a result of insufficient annual data for the required estimation period for the country under investigation. And as panel analysis can accommodate period averages, it was appropriate to employ this analytical framework. Whilst this could not be an inappropriate technique of analysis (as Rodrik and Subramanian (2004) for India and Davoodi and Grigorian (2007) for Armenia, have both applied this technique of analysis), the use of time-series analysis once sufficient data is available on annual basis is the ideal choice. Thirdly, though income is incorporated as part of the measure of HDI, yet the indicators used in this research to typically proxy poverty reduction (i.e. aggregate welfare), though have been widely argued to be the relevant measures, have not specifically included the income poverty measure which is more commonly used as measure of poverty. The non-availability of sufficient data on this variable on a time series basis could not allow its use in this research. Ideally, including this measure together with the aggregate welfare indicators comprising the HDI and IMR would have provided more robust findings and stronger conclusions on the impact of foreign aid on poverty reduction.

Fourth, the research has only captured the importance of politics on aid’s impact on human development in SL, but measures of politics have only involved the use of formal politics in the context of institutionalised democracy, autocracy and polity quality. However,

250 patrimonial politics which is yet to be quantitative measured has not been singled out in the analysis since no quantitative measure is available. But, an exploration of qualitative analysis to capture the impact of patrimonial politics which is a form of politics common in most African countries including Sierra Leone (as has been indicated in the background chapter) is equally important. This form of politics though of high intensity in autocratic regimes, is evident even in democratic regimes. However, the scope and volume required of this thesis could not allow the researcher to include such analysis in this thesis. Finally, the scope of this research did not attempt to capture the long-run impact of aid on welfare. This research’s examination of the impact of aid on economic growth and pro-poor growth in Sierra Leone provide evidence to show that the impact of development aid is more obvious in the long-run than in the short-run. As the panel data analysis for the impact of aid on aggregate welfare in Africa and Sub-Saharan Africa could not provide evidence that foreign aid improves welfare in this continent and sub-region, it would have been relevant to further explore the long-run impact of aid before concluding on the ineffectiveness of aid on human welfare in Africa as found by this research.

Cognisant of these limitations, further research could be conducted to strengthen the findings and conclusions reached on the effectiveness of foreign aid as examined by this research. In particular, future research may look at extending the analysis to further country studies of similar economic structure particularly in Sub-Saharan Africa in order to generalise the findings and policy application across the sub-region. Also, whilst the research’s use of country dummy interactions with the variables of interest in a panel dataset could not be an inappropriate technique of analysis, the use of time-series analysis once sufficient data is available on annual basis is ideally preferable and hence recommended for future research. Future research could also look at opportunities for further extending the analysis to include other indicators of poverty especially income poverty indicators once sufficient data is available for this indicator. This will provide more robust findings and hence stronger conclusions on the impact of aid on poverty reduction. Further, as the scope of this research did not attempt to capture the long-run impact of aid on welfare, it is recommended that future research attempts to adopt the analysis employed by Rajan and Subramanian (2008) by using five and/or ten-year lagged values of foreign aid to capture the long-run impact of aid. Finally, as patrimonial politics is a form of political practice common in most African countries including Sierra Leone, it would it be worth it for future research to look into the impact of patrimonial politics on the effectiveness of foreign aid even at an explorative level. 251

Despite these limitations, since this is the first study to investigate the relationship between donor intervention, economic growth and poverty reduction in Sierra Leone and at a comprehensive country analysis, the research is expected to contribute to the literature on aid effectiveness in Sierra Leone and in aid recipient countries in general.

252

References

Acemoglu, D. and Robinson, J. (2012). Why Nations Fail: The origins of Power, Prosperity and Poverty . The blog: Fancy Schools, (27 th February). Available at: www.whynationsfail,com/blog

Addison, T., Hulme, D. and Kanbur, R. (2008). Poverty Dynamics: Measurement and Understanding from an Interdisciplinary Perspective, BWPI Working Paper 19, University of Manchester.

Adhikary, B.K. (2011). FDI, Trade Openness, Capital Formation, and Economic Growth in Bangladesh: A Linkage Analysis, International Journal of Business and Management , 6(1): 16-28.

Afzal, M, Farooq, M.S., Ahmad, H.K., Begum, I. and Quddus, M.A. (2010). Relationship between School Education and Economic Growth In Pakistan ARDL Bounds Testing Approach to Cointegration, Pakistan Economic and Social Review, 48(1): 39-60.

Ahmed, N. (2003). Trade Liberalisation and Endogenous Growth of Manufacturing Industries in Bangladesh: An Empirical Investigation, Applied Economics , 35: 305-314.

Aiken, L.S. and West, S.G. (1991). Multiple Regression: Testing and Interpreting Interactions, Newbury Park: Sage Publications.

Akpokodje, A and Omojimite, B.U. (2008). The effects of aid flows on Nigeria’s Agricultural Growth, Pakistan Journal of Social Sciences , 5(6): 514-520.

Alam, M.I. and Quazi, R.M. (2003). Determinants of Capital Flight: An Econometric Case Study of Bangladesh, International Review of Applied Economics , 17(1): 85-103.

Alesina, A. and Dollar, D. (2000). Who gives foreign aid to whom and why? Journal of Economic Growth , 5(1): 33-63.

Alesina, A., Ozler, S, Roubini, N. and Swagel, P. (1996). Political Instability and Economic Growth, Journal of Economic Growth 1: 193-215.

Alesina, A. and Perotti, R. (1996). Income Distribution, Political Instability, and Investment, European Economic Review 40: 1203-1228.

Ahluwalia, M. S. (1978). Rural poverty and agricultural performance in India, Journal of Development Studies , 14: 298-323.

Anand, S., and Sen, A. (1992). Human development index: Methodology and measurement, Background Paper for Human Development Report 1993, New York: Oxford University Press.

Andersen T.B, Hansen H, Markussen T. (2005). US politics and World Bank IDA-lending. The Journal of Development Studies 42(5): 772–794.

Andersen, O. W. and Therkildsen, O. (2007). Harmonisation and Alignment: The Double-edged Swords of Budget Support and Decentralised Aid Administration. DIIS Working Paper No. 2007/4. Copenhagen: Danish Institute for International Studies.

Ang, J.B. (2008). Financial Development and Economic Growth in Malaysia ; Oxon, UK: Routledge.

Angeles, L. and Neanidis, K. (2009). Aid effectiveness: the role of the local elite, Journal of Development Economics , 90: 120–134.

Anonymous (2005). Shift to Grants for the Poorest, Finance & Development 42(3): 3.

Arellano, M. and Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations, Review of Economic Studies 58(2): 277–97.

253

Arellano, M. and Bover, O. (1995). Another look at the instrumental variables estimation of error-components models, Journal of Econometrics, 68: 29–51.

Arrow, K.J. (1962). The Economic Implications of Learning by Doing, Review of Economic Studies , 80: 155- 173.

Arvin, B.M and Barillas, F. (2002). Foreign aid, poverty reduction, and democracy; Applied Economics , 34(17): 2151-2156.

Asiama, J.P. and Quartey, P. (2009). Foreign Aid and the Human Development Indicators in Sub-Saharan Africa, Journal of Developing Societies, 25: 57.

Ayittey, G.B.N. (2004). Africa Unchained: The Blueprint of Africa’s Future, Palgrave Macmillan.

Bachman, K.L. (1965). Agricultural Economics and Technical Aid in Foreign Development, Journal of Farm Economics , 47(5): 1079-1090.

Bahmani-Oskooee, M. and Rhees, H. (1997). Response of Domestic production to Depreciation in Korea: An Application of Johansen’s Cointegration Methodology, International Economic Journal , 14: 61-74

Baliamoune-Lutz, M. and Mavrotas, G. (2008). Aid Effectiveness: Looking at the Aid-Social Capital-Growth Nexus, UNU-WIDER Research Paper No. 2008/75

Baland, J., Moene, K.O. and Robinson, J.A. (2010), Governance and Development, In D. Rodrik, and M.R. Rosenzweig (2010) Handbook of Development Economics, Vol.5 . London: Elsevier

Banerjee, A., Dolado, J., Galbraith, J. and Hendry, D. (1993). Co-integration, Error- Correction, and the Econometric Analysis on Non-Stationary Data , Oxford: Oxford University Press.

Barrett, C.B, and Maxwell, D.G. (2005). Food Aid after Fifty Years: Recasting its Role , London: Routledge.

Barro, R.J. (1991). Economic Growth in a Cross-Section of Countries, Quarterly Journal of Economics, 106,(2): 407-44.

Bauer, P. (1971). Dissent on Development , Canberra: Weidenfeld & Nicolson.

Begovic, B. (2008). Does Foreign Aid Work?, Centre for International Private Enterprise, Washington: Development Institute. Available at: http://developmentinstitute.org

Berg, A. and Krueger, A.O. (2006). Trade, Growth and Poverty: A Selective Survey. In A. Mody and C. Pattillo (Ed.) Macroeconomic Policies and Poverty Reduction , London: Routledge. Pp.71

Bhattarai, B.P. (2009). Foreign Aid and Growth in Nepal: An Empirical Analysis, Journal of Developing Areas , 42(2): 283-302

Bigsten, A., Gunning, J.W. and Tarp, F. (2006). The Effectiveness of Foreign Aid: Overview and an Evaluation Proposal, Available at: http://www.econ.ku.dk/ftarp/miscellaneous/Docs/evaluation_proposal_february%202006.pdf

Blaikie, N.W.H. (1991). A Critique of the Use of Triangulation in Social Research, Quality and Quantity , 25: 115-136.

Blundell, R. and Bond, S. (1998). Initial Conditions and Moment Restrictions in Dynamic Panel Data Models, Journal of Econometrics, 87(1): 115–43.

Bobba M. and Powell, A. (2007). Aid Effectiveness: Politics Matter, Inter-American Development Bank Research Department working paper No. 601 .

Bond, S., Hoeffler, A. and Temple, J. (2001) GMM Estimation of Empirical Growth Models, CEPR Discussion Paper No. 3048 .

254

Boone, P. (1996). Politics and the effectiveness of foreign aid, European Economic Review, 40: 289-329.

Booth, D. (2003). Introduction and Overview, Development Policy Review: Special Edition on PRSP. 21(2): 131-159.

Boumahdi, R. and Thomas, A. (2008). Endogenous Regressors and Correlated Effects in Marquez et al. (ed.). The Econometric of Panel Data: Advanced Studies in Theoretical and Applied Econometrics , Vo. 46, Part 1, P.89-112. Berlin: Springer.

Bräutigam, D., and Knack, S. (2004). Foreign Aid, Institutions, and Governance in Sub-Saharan Africa, Economic Development and Cultural Change , 52 (2): 255-85.

Bring, J. (1994) How not to find the relationship between foreign aid and economic growth, Applied Economics Letters , 1: 32–33.

Brock, K., McGee, R. and Ssewakiryanga, R. (2002). Poverty Knowledge and Policy Processes: a case Study of Ugandan National Poverty reduction Policy, IDS Research Paper No. 53 . Brighton: IDS.

Brown, R. L., Durbin, J. and Evans. J. (1975). Techniques for Testing the Constancy of Regression Relations over Time, Journal of the Royal Statistical Society , 37: 149-163.

Brumm, H. J. (2003). Aid, policies and growth: Bauer was right, Cato Journal, 23:167-74.

Burnside, C., Dollar, D. (1997). Aid, policies, and growth, World Bank Policy Research Working Paper 1777. Washington DC: World Bank.

Burnside, C., Dollar, D. (2000). Aid, policies and growth, American Economic Review, 90(4):847-68.

Burnside, C., and Dollar, D. (2004). Aid, policies and growth: revisiting the evidence, World Bank Policy Research Working Paper No. 3251 , World Bank, Washington.

Cammack, D. (2007) The Logic of African Neopatrimonialism: What Role for Donors?, Development Policy Review , 25 (5): 599-614

Cassen, R. (1994). Does Aid Work? Oxford: Oxford University Press.

Cassimon, D., and B. Van Campenhout (2007). Aid Effectiveness, Debt Relief and Public Finance Response: Evidence from a Panel of HIPC Countries. Review of World Economics/Weltwirtschaftliches Archiv 143 (4): 742–763.

Castro-Leal, F., Dayton, J., Demery, L. and Mehra, K. (1999). Public social spending in Africa: do the poor benefit?, World Bank Research Observer , 14: 49–72.

Cathie, J. (1982) The Political Economy of Food Aid . Cambridge: Gower

Chabal, P.(1994). Power in Africa: An essay in political interpretation, London: Macmillan.

Chandar, A. and Caprio, J. (2007). Aid Effectiveness and Growth in the Developing World: Does Institutional Quality Matter?, BA Honours in Economics Thesis; University of Massachusetts.

Chenery, H.B., (1964). Objectives and Criteria for Foreign Assistance. In G. Ranis (ed) The United States and the Developing Economies, New York: W.W. Norton and Company Inc, Pp. 80-91.

Chenery, H., Strout. S. (1966). Foreign assistance and economic development, American Economic Review, 66: 679-753.

Cherp, A., George, C. and Kirkpatrick, C. (2004). A Methodology for Assessing National Sustainable Development Strategies, Impact Assessment Research Centre Working Paper Series Paper No. 6.

255

Chu, K., and others (1995). Unproductive Public Expenditures: A pragmatic approach to policy analysis, IMF pamphlet series No. 48 , Washington: IMF

Clemens, M., Radelet, S., and Bhavnani, R. (2004). Counting chickens when they hatch: The short-term effect of aid on growth. Centre for Global Development. Working Paper No. 44, Washington D.C.: Centre for Global Development.

CMI (2005), Should Corrupt Countries Receive Budget Support?, CMI Brief, 4(4) .

Cohen, D; Jacquet, P., and Reisen, H. (2007). Forgive Debt, but Keep Lending, Policy Insights N0.44. OECD Development Centre.

Collier, P. and Dehn, J. (2001). Aid, Shocks, and Growth, Policy Research Working Paper 2688 , Washington: World Bank.

Collier, P. and Dollar, D. (2002). Aid allocation and poverty reduction, European Economic Review , 46: 1475– 1500.

Constant, N. and Yue, Y. (2010). An Econometric Estimation of Import Demand Function for Cote D’Ivoire, International Journal of Business and Management , 5(2): 77-84

Cordella, T. and Dell’Ariccia, G. (2007) Budget Support versus Project Aid: A Theoretical Appraisal. The Economic Journal , 117: 1260–1279

Cordella, T. and Ulku, H. (2004). Grants versus Loans; IMF Working Paper, WP/04/161 ; Washington D.C.: IMF

Corden, M. (1971). The Effects of Trade on the Rate of Growth. In J. Bhagwati, et. al. eds., Trade, Balance of Payments, and Growth: Papers in Honour of Charles P. Kindleberger .Amsterdam: North-Holland

Cox, M. and Kenndy-Pipe, C. (2005). The Tragedy of American Diplomacy? Rethinking the Marshall Plan. Journal of Cold War Studies , 7(1): 97-134

Cramer, C, Stein, H. and Weeks, J. (2006). Ownership and Donorship: Analytical Issues and a Tanzanian Case Study, Journal of Contemporary African Studies, 24(3): 415-436.

Dabla-Norris, E., Matovu, J., and Wade, P. (2004). Debt relief, demand for education and poverty. In T. Addison, H. Hansen, & F. Tarp (Eds.), Debt relief for poor countries (pp. 241–266). Basingstoke: Palgrave Macmillan in association with UNU-WIDER.

Dalgaard, C.J., Hansen, H. (2001). On aid, growth and good policies, Journal of Development Studies 37(6):17- 41.

Dalgaard, C.J., Hansen, H., Tarp, F. (2004). On the empirics of foreign aid and growth, Economic Journal , 114(496): F191-F216.

Dandekar, V. M. (1965). The Demand for Food and Conditions Governing Food Aid during Development, WFP Studies, No. 1 (Rome:).

Dantwala, M. L. (1963). International planning to combat the scourge of hunger throughout the world, Annals of Collective Economy, 34(1).

Davoodi, H.R. and Grigorian, D.A. (2007). Tax Potential vs. Tax Effort: A Cross-Country Analysis of Armenia’s Stubbornly Low Tax Collection, IMF Working Paper wp/07/106.

De Boef, S. (2000). Modelling Equilibrium Relationships: Error Correction Models with Strongly Autoregressive Data, Political Analysis , 9:1

256

De Janvry, A and Sadoulet, E. (1993). Adjustment Policies, agriculture and rural development in Latin America. In A. Singh and H. Tabatabai (Ed) Economic crisis and third world agriculture: The Changing Role of Agriculture in Economic Development ; UK: Cambridge University Press

Department for International Development (DFID) (2008) “Elections in Sierra Leone in 2007 and 2008.” DFID.

Development Assistance Coordinating Office (DACO) (2008). Development Assistance to Sierra Leone 2006 . A report by DACO, Office of the Vice President, Republic of Sierra Leone

Devarajan, S., Dollar, D. and Holmgren, T. (2001). Aid and Reform in Africa: Lessons from Ten Case studies; Washington D.C.: The World Bank

DFID (2004). Poverty Reduction Budget Support, DFID Policy Paper , May.

DFID (Department for International Development) (2006). White Paper: Making Governance Work for Poor People, London: DFID. Available at: www.dfid.gov.uk/wp2006

DFID (2009). Country-Led Approaches to Poverty Reduction, Available at: http://www.dfid.gov.uk/About- DFID/Finance-and-performance/Making-DFIDs-Aid-more-effective/How-we-give-aid/Country-led-approaches- to-poverty-reduction

Dickey, D.A. and Fuller, W.A. (1979). Distribution of the Estimators for Autoregressive Time Series with a Unit Root, Journal of the American Statistical Association , 74: 427-431.

Dollar, D. and Easterly, W. (1999). The search for the key: aid, investment and policies in Africa, Journal of African Economies; 8(4): 546-577

Dollar, D. and Levin, V. (2005). Sowing and Reaping: Institutional Quality and Project Outcomes in Developing Countries, World Bank Policy Research Working Paper 3524 , The World Bank.

Dollar, D. and Kraay, A. (2000). Growth is Good for the Poor; Development Research Group, The World Bank

Domar, E. (1947). Expansion and Employment, American Economic Review , 37(1): 343-55.

Doucouliagos, H. and Paldam, M. (2008). Aid effectiveness on growth: A meta study, European Journal of Political Economy, 24: 1–24.

Dowling, J.M. and Hiemenz, U. (1985). Biases in the Allocation of Foreign Aid: Some New Evidence, World Development , 13(4): 535-S-11.

Downward, P. and Mearman, A. (2007). Retroduction as Mixed Methods Triangulation in Economic Research: Reorienting Economics into Social Science, Cambridge Journal of Economics, 31: 77-99

Dreher, A., Nunnenkamp, P., and Theile, R. (2008). Does aid for education educate children? Evidence from panel data, The World Bank Economic Review , 22:291-314.

Dutta, D. and Ahmed, N. (2004). Trade Liberalisation and Industrial Growth in Pakistan: A Cointegration Analysis, Applied Economics , 36: 1421-1429

Easterly, W. (2001). The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics ; London: The MIT Press.

Easterly, W. (2009). Can the West Save Africa?, Journal of Economic Literature, 47(2): 373–447

Easterly, W., Levine, R., Roodman, D. (2004). New data, new doubts: a comment on Burnside and Dollar’s ‘Aid, policies and growth (2000)’, American Economic Review 94(3): 781-4.

Easterly, W. R., and Rebelo, S.T. (1993). Fiscal Policy and Economic Growth: An Empirical Investigation, Journal of Monetary Economics, 32(3): 417-58.

257

Elbadawi, I.A and Soto, R. (1997). Real exchange rates and macroeconomic adjustment in sub- Saharan Africa and other developing countries, Journal of African Economies. Supplement to 6.3:74-120.

Engle, R. F., and Granger, C.W.J. (1987). Cointegration and Error Correction: Representation Estimation and Testing, Econometrica, 55:251–276.

Feeny, S. (2005). The impact of foreign aid on economic growth in Papua New Guinea, Journal of Development Studies , 14(6): 1092–1117

Feeny, S. (2007) Impacts of Foreign Aid to Melanesia, Journal of the Asia Pacific Economy , 12 (1): 34 — 60

Feeney, S., Ouattara, B., (2009). What type of economic growth does foreign aid support? Applied Economic Letters 16(7): 727 – 730

Feyzioglu, T., Swaroop, V., and Zhu, M. (1998). A Panel data analysis of the fungibility of foreign aid, World Bank Economic Review , 12(1), 29-58

Fielding, D. McGillivray, M. and Torres, S. (2006) A Wider Approach to Aid Effectiveness: Correlated Impacts on Health, Wealth, Fertility and Education, UNU-WIDER Research paper No. 2006/23

Fischer, S. (1993). The Role of Macroeconomic Factors in Growth, Journal of Monetary Economics, 32(3): 485-512

Franco-Rodriguez, S., McGillivray, M. and Morrissey, O. (1998). Aid and the Public Sector in Pakistan: Evidence with Endogenous Aid, World Development, 26:1241–50.

Garrett, G. and Mitchell, D. (2001). Globalization, Government Spending and Taxation in the OECD, European Journal of Political Research , 39: 145-177

Gillanders, R. (2010). The Effects of Foreign Aid in Sub-Saharan Africa, UCD School of Economics (August 4).

Gomanee, K. and Morrisey, D. (2002). Evaluating Aid Effectiveness against Poverty Reduction Criterion, Part of a PhD Thesis, Vienna University of Economics and Business Administration, Vienna, Austria. Available at: www.sti.ch/fileadmin/user_upload/pdfs/swap/swap153.pdf

Gomanee, K., Girma, S., Morrissey, O. (2005). Aid and growth in sub-Saharan Africa: accounting for transmission mechanisms , CREDIT Research Paper 02/05 , School of Economics, University of Nottingham.

Gomanee, K., Morrissey, O., Mosley, P. and Verschoor, A. (2005a). Aid, Government Expenditure, and Aggregate Welfare, World Development, 33(3): 355–370

Gomanee, K., Girma, S., and Morrissey, O. (2005b). Aid, public spending and human welfare: Evidence from quantile regressions, Journal of International Development , 17(3): 299-309.

Gounder R. (1994). Empirical results of aid motivations: Australia’s bilateral aid program, World Development 22(1): 99–113.

Gounder, G. (2001). Aid-Growth nexus: Empirical evidence from Fiji, Applied Economics, 33: 1009-1019.

Government of Sierra Leone (GoSL) (2005). Poverty Reduction Strategy Paper (SL-PRSP): A National Programme for Food Security, Job Creation and Good Governance (2005 – 2007). Available at: www.imf.org

Griffin, K. (1970). Foreign capital, domestic savings and economic development, Bulletin of the Oxford University Institute of Economics and Statistics 32(2): 99-112

Griffin, K., Enos, J (1970). Foreign assistance: objectives and consequences, Economic Development and Cultural Change 18: 313-27.

258

Guillaumont, P., Chauvet, L. (2001). Aid and performance: A reassessment, Journal of Development Studies 37(6):66-87.

Gupta, S., Verhoeven, M. and Tiongson, E. (2002). The effectiveness of government spending on education and health care in developing and transition economies, European Journal of Political Economy , 18(4): 717-738

Gupta, S., Powell, R. and Yang, Y. (2006). Macroeconomic Challenges of Scaling Up Aid to Africa: a Checklist for Practitioners, Washington D.C.: IMF. de Haan, A. and Everest-Philips, M. (2007). Can New Aid Modalities Handle Politics ?, UNU-WIDER Research Paper No. 2007/63

Hansen, H., Tarp, F. (2000a). Aid effectiveness disputed, Journal of International Development 12(3):375-98.

Hansen, H. and Tarp, F. (2001) .Aid and growth regressions, Journal of Development Economics, 64: 547-570.

Harris, R., and Sollis, R. (2003). Applied Time Series Modelling and Forecasting , West Sussex: Wiley

Harrod, R.F. (1939). An Essay in Dynamic Theory, The Economic Journal, 49(193): 14-33 .

Harrod, R. F. (1948). Towards a Dynamic Economics , London: MacMillan.

Harvey, D. (2003). The New Imperialism , Oxford: Oxford University Press.

Havrylyshyn, O. (1990). Trade policy and productivity gains in developing countries: A survey of the literature, World Bank Research Observer , 5:1-24.

Heady, D. (2007). Geopolitics and the effect of foreign aid on economic growth: 1970-2001. Journal of International Development , 20: 161–180.

Hefeker, C. (2006). Project Aid or Budget Support? The Interests of Governments and Financial Institutions. Review of Development Economics, 10(2): 241-252.

Hendry, D.F. and Doornik, J.A. (2001). Empirical Econometric Modelling using PCGIVE , Volume 1. London: Timberlake Consultanta Ltd.

Herzer, D. and Morrissey, O. (2011). Long-Run Aid Effectiveness, CSAE Conference 2011-Economic Development in Africa, Paper 144; Oxford.

Hickey, S. (2003). The Politics of Staying poor in Uganda, CPRC Working Paper 37 . At: www.chronicpoverty.org/pdfs/SH%20CPRC%20wp%2037.pdf

High Level Forum (2005). Paris Declaration on Aid Effectiveness; Paris, (February 28-March 2).

Johnson, D. and Farmer, T.A. (2007). Researching Educational Leadership: A Conceptual Framework for Doctoral Program Research, At: http://cnx.org/content/m14781/1.1

Hjelm, G. (2002). Effects of Fiscal Contractions: The Importance of Preceding Exchange Rate Movements, Scand. J. of Economics 104(3): 423-441

Hjertholm, P., and White, H. (2000). Survey of Foreign Aid: History, Trends and Allocation, University of Copenhagen Department of Economics Discussion Papers .

Hopkins, R. (2000). Political Economy of Foreign Aid. In Gilbert & Vines (eds), The World Bank: Structure and Policies , Cambridge: Cambridge University Press, P.19

Hulme, D. (2009. The Millennium Development Goals (MDGs): A Short History of the World’s Biggest Promise . BWPI Working Paper 100 , University of Manchester.

259

IDD and Associates (2006). Evaluation of General Budget Support: Synthesis Report, A Joint Evaluation of General Budget Support 1994-2004; School of Public Policy, University of Birmingham. At: www.idd.bham.ac.uk

Inder, B., (1993). Estimating Long-Run Relationships in Economics: A Comparison of Different Approaches, Journal of Econometrics , 57: 53–68.

International Monetary Fund (IMF) (2007). Sierra Leone: Poverty Reduction Strategy Paper - Annual Progress Report, IMF Country Report No. 07/17

Isenman, P. (1976). Biases in Aid Allocations Against Poorer and Larger Countries, World Development, 4(8): 631-641.

Isenman, P.J. and Singer, H.W. (1993). Food Aid: Disincentive Effects and their Policy Implications. In V.W. Ruttan (Ed) Why Food Aid? Baltimore and London: The John Hopkins University Press, Pp 99-128

Ishfaq, M. (2004). Aid effectiveness, debt capacity and debt management in the economy of Pakistan, PhD dissertation to Quad-e-Azam University, Islamabad.

Islam, A.N., (1992). Foreign Aid and Economic Growth: An Econometric Study of Bangladesh, Applied Economics , 24: 541–44.

Javid, M. and Qayyum, A. (2011). Foreign Aid-Growth Nexus in Pakistan: Role of Macroeconomic Policies, MPRA Paper No. 29498.

Jhingan, M.L. (2004). The Economics of Development and Planning; 37 th Edition . Delhi: Vrinda Publications Ltd.

Johansen, S. (1988), Statistical Analysis of Cointegrating Vectors, Journal of Economic Dynamics and Control , 12: 231-254.

Johansen, S. and Juselius, K. (1990). Maximum Likelihood Estimation and Inference on Cointegration –With Application to the Demand for Money, Oxford Bulletin of Economics and Statistics , 52(2): 169-211.

Johansen, S. (1991). Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models, Econometrica, 59:1551-1580 .

Johnson, D.G. (1980). Inflation, Agricultural Output, and Productivity. American Journal of Agricultural Economics , 62(5): 917-923.

Johnson, D. and T.A. Farmer (2007) ‘Researching Educational Leadership: A Conceptual Framework for Doctoral Program Research’; At: http://cnx.org/content/m14781/1.1

Jones, C.I. (1998). Introduction to Economic Growth , New York: W.W. Norton & Company.

Kamara, M.M. (2008). Libya Offers 30 tractors to Sierra Leone. Africa News , (February 20). Available at: www.africanews.com .

Kargbo, S.M. and Adamu, P.A. (2010). Financial Development and Economic Growth in Sierra Leone, Journal of Monetary and Economic Integration, 9(2): 30-61

Karras, G. (2006). Foreign Aid and Long-run Economic Growth: Empirical Evidence for A panel of Developing Countries, Journal of International Development , 18: 15–28.

Kenny, C. and Williams, D. (2001). What Do We Know About Economic Growth? Or, Why Don't We Know Very Much? World Development , 29(1): 1-22

Killick, T. (1998). Aid and the Economy of Policy Change . London: Routledge

260

Koeberle, S. G. (2003). Should Policy-Based Lending Still Involve Conditionality?, The World Bank Research Observer , 18(2): 259-273.

Koenker, R. and Basset, G (1978) Regression Quantiles. Econometrica , 46: 33-50

Koenker, R. and Basset, G. (1982). Robust Tests for Heteroscedasticity based on Regression Quintiles. Econometrica , 50: 43-61

Koenker, R. (2005). Quantile Regression , Cambridge University Press. ISBN 0-521-60827-9

Kosack, S. (2003). Effective Aid: How Democracy Allows Development Aid to Improve the Quality of Life, World Development, 31(1): 1–22

Knack, S. and Keefer, P. (1995). Institutions and Economic Performance: Cross-Country Tests Using Alternative Institutional Measures, Economics and Politics, 7: 207–227.

Knack, S. and Keefer, P. (1997). Does social capital have an economic impact? A cross-country investigation, Quarterly Journal of Economics , 112(4): 1252–88

Knack, S. (1999). Aid Dependence and the Quality of Governance, IRIS Center Working Paper , University of Maryland.

Knack, S. (1999). Social capital, growth and poverty: a survey of cross-country experience, World Bank Social Capital Initiative Working Paper 7 .

Krause, G. A. (1997). Voters, Information Heterogeneity, and the Dynamics of Aggregate Economic Expectations, American Journal of Political Science, 41:1170–1200.

Krueger. A.O. (1998). Why Trade Liberalisation is good for the Poor, The Economic Journal, 108:1513-1522.

Krugman, P. (1988). Financing vs. Forgiving a Debt Overhang. Journal of Development Economics , 29: 253- 268.

Kunz, D.B. (1997). Marshall Plan Reconsidered - A Complex of Motives, The [comments] Foreign Affairs , 76(3): 162-170

Lanjouw, P. and Ravallion, M. (1999). Benefit Incidence, Poverty Spending Reforms and the Timing of Program Capture, The World Bank Economic Review , 13(2): 257-274

Laurenceson, J. and Chai, J.C.H. (2003). Financial reform and economic development in China, Cheltenham: Edward Elgar .

Lawson, T. (1997 ). Economics and Reality , London: Routledge.

Levine, R., Renelt, D. (1992). A sensitivity analysis of cross-country growth regressions, American Economic Review, 82(4):942-63.

Levy, V. (1988). Aid and growth in sub-Saharan Africa: the recent experience, European Economic Review 32(9): 1777-95.

Lloyd, T., Morrissey, O. and Osei, R. (2001). Aid, Exports and Growth in Ghana, CREDIT Research Paper 01/01 , University of Nottingham

Loxley, J. and Sackey, H.A. (2008). Aid Effectiveness in Africa, African Development Review , 20(2): 163-199.

Lubker, M., Smith, G. and Weeks, J. (2002). Growth and the Poor: A Comment on Dollar and Kraay; Journal of International Development , 14: 555–571

Lucas, R.E. (1988) .On the Mechanics of Economic Development, Journal of Monetary economics , 22: 3-42.

261

Luke, D.F. (1985). Electoral politics in Sierra Leone: an appraisal of the 1982 elections. Journal of Commonwealth and Comparative Politics 23 (1): 30-42.

Luke, D.F. and Riley, S.P. (1989). The Politics of Economic Decline in Sierra Leone, The Journal of Modern African Studies , 27(1): 133-141

Macdonald, R. and Ricci, L.A. (2004). Estimation of the Equilibrium Real Exchange Rate for South Africa, The South African Journal of Economics , 20: 71-83

Magassa, H. and Meyer, S. (2008). The Impact of Aid Policies on Domestic Democratisation Processes: The Case of Mali: - Donor Harmonisation: Between Effectiveness and Democratisation: Case Study IV, FRIDE Working Paper 50

Maizels A, Nissanke M.K. (1984). Motivations for aid to developing countries, World Development , 12(9): 879–900

Mallaye, D. (2011). Foreign Aid and the Mobilization of the Factors of Growth in Sub-Saharan Africa, A conference paper presented at the AERC-UNU-WIDER conference on Macroeconomic Management of Foreign Aid, Nairobi 02-03 December 2011.

M’Amanja, D. and Morrissey, O. (2005). Foreign Aid, Investment and Economic Growth in Kenya: A Time Series Approach, CREDIT Research Paper No. 06/05 , University of Nottingham

Mann, F. (2007). Debt Burden Hindering Development, Global Envision Archives. Available at: www.globalenvision.org .

Marshall, M.G. and Jaggers, K. (2008). Polity IV Project: Political Regime Characteristics and Transitions, 1800-2008.

Masud, N. and Yontcheve, B. (2005). Does Foreign Aid reduce poverty? Empirical Evidence from Non- governmental and Bilateral Aid, IMF working Paper 05/100 .

Mauro, P. (1995). Corruption and growth, Quarterly Journal of Economics, 110 (3), 681–712.

Mavrotas, G. (2002). Aid and Growth in India: Some Evidence from Disaggregated Aid Data, South Asia Economic Journal , 3(19): 19-48

Mavrotas, G. (2005). Aid Heterogeneity: Looking at Aid Effectiveness from a Different Angle. Journal of International Development 17 (8): 1019–1036.

Mavrotas, G. (2007). Scaling Up of Foreign Aid and the Emerging New Agenda. Theory and Practice of Foreign Aid, 1 : 211-233.

Mavrotas, G., and P. Nunnenkamp (2007). Foreign Aid Heterogeneity: Issues and Agenda. Review of World Economics/Weltwirtschaftliches Archiv 143 (4): 585– 595.

Mavrotas, G. and Ouattara, B. (2003). The Composition of Aid and the Fiscal Sector in an Aid-Recipient Economy: a Model, Discussion Paper No. 03/07 , School of Economic Studies, University of Manchester.

Mavrotas, G., Ouattara, B. (2006a), Public fiscal behaviour and aid heterogeneity in aid-recipient economies. Journal of Developing Areas 39, 1–15.

Mavrotas, G. and Ouattara, B. (2006b). Aid Disaggregation and the Public Sector in Aid-Recipient Economies: Some Evidence from Cote D’Ivoire, Review of Development Economics, 10(3), 434–451

Mavrotas, G., and B. Ouattara (2007). Aid Modalities and Budgetary Response: Panel Data Evidence. Review of World Economics/Weltwirtschaftliches Archiv 143 (4): 720–741

Maxwell, J. A. (2005a). Qualitative research design: An interactive approach (2nd ed.), Thousand Oaks, CA: Sage Publications.

262

Maxwell, S. (1986a). Food Aid: Agricultural Disincentives and Commercial Market Displacement, IDS Discussion Paper 224

Maxwell, S. (1986b). Food Aid to Senegal: Disincentives Effects and Commercial Displacement, IDS Discussion Paper 225

Maxwell, S. (1986c). Food Aid to Ethiopia: Disincentive Effects and Commercial Displacement, IDS Discussion Paper 226

Maxwell, S. (1991). The Disincentive Effect of Food Aid: A Pragmatic Approach. In E. Clay and O. Stokke (ed.) Food Aid Reconsidered: Assessing the Impact on Third World Countries , London: Frank Cass, Pp. 66-90

Maxwell, S.J. and Singer, H.W. (1979). Food Aid to Developing Countries: A Survey, World Development, 7: 225-247

Maxwell, S. (2002), More aid? Yes – and use it to reshape aid architecture. In: ODI Opinions . ODI, London.

Maxwell, S. (2005b). The Washington Consensus is Dead! Long Live the Meta- narrative!, ODI Working paper 243

Mbaku, J.M. (1993). Foreign Aid and Economic Growth in Cameroon, Applied Economics , 25: 1309–14.

McGee, R. (2002). Assessing Participation in Poverty Reduction Strategy Papers: A Desk –based Synthesis of Experience in Sub- Sahara Africa, IDS Research Report 52 . Brighton: IDS.

McGillivray, M. (2004). Descriptive and prescriptive analyses of aid allocation: Approaches, issues, and consequences, International Review of Economics and Finance ; 13: 275–292

McGillivray, M., Feeny, S., Hermes, N., Lensink, R. (2006). Controversies over the impact of development aid: it works; it doesn't; it can, but that depends …., Journal of International Development, 18: 1031–1050.

McGillivray, M. and Noorbakhsh, F. (2007). Aid, Conflict and Human Development, Working Papers 2007-03 , Department of Economics, University of Glasgow

Mekasha, T.J. and Tarp, F. (2011). Aid and Growth: What Meta-Analysis Reveals, UNU-WIDER Working Paper No. 2011/22

Mellor, J.W. (1999). Impact Assessment of Agricultural Policy Reform on Employment and Productivity , Bethesda, MD: Abt Associates Inc.

Mellor, J.W. (2000). Faster, More Equitable Growth: The Relation between Growth in Agriculture and Poverty Reduction, CAER II Discussion Paper No.70 , Cambridge, MA: HIID.

Meyer, S and Schulz, N. (2007). Theoretical framework and methodology for country case studies; FRIDE working paper No. 41 .

Miles, M. B. and Huberman, A. M. (1994). Qualitative data analysis ( 2nd ed.). Thousand Oaks, CA: Sage.

Morrissey, O. (2007). Aid and Poverty Reduction: Whose Well-being and How to Increase It. Paper for the Well-being and Development Policy Sessions at the Well-being in International Development Conference , University of Bath, (28-30 June).

Morrisson, C. (ed.) (2002). Education and Health Expenditure and Poverty Reduction in East Africa: Madagascar and Tanzania , Paris: OECD Development Centre Studies.

Mosley, P. (1980). Aid, savings and growth revisited, Oxford Bulletin of Economics and Statistics, 42(2): 79-95.

Mosley, P., Hudson, J, Horrell, S. (1987). Aid, the public sector and the market in less developed countries, Economic Journal, 97(387): 616-41.

263

Mosley, P., Hudson, J., and Verschoor, A. (2004). Aid, poverty reduction and the new conditionality, The Economic Journal , 114: F217–F243.

Murthy, V.N.R., Ukpolo, V. and Mbaku, J.M. (1994). Foreign Aid and Economic Growth in Cameroon: Evidence from Cointegration Tests, Applied Economic Letters , 1: 161-163

Myrdal, G. (1968). Asian Drama : An Enquiry into the Poverty of Nations , Vol.1. New York: Penguin Press

Narayan, P. K. (2004). Reformulating critical values for the bounds F-statistics approach to cointegration: an application to the tourism demand model for Fiji, Department of Economics, Monash University, Australia

Ndulo, M. (2003). The Democratization Process and Structural Adjustment in Africa. Indiana Journal of Global Legal Studies , 10(1): 315-368.

Neville, P. (2002). Sierra Leone: The Libyan Connection. World Express Review , 49(6) (April 11).

North, D. C. (1990), Institutions, Institutional Change and Economic Performance , Cambridge, U.K.: Cambridge University Press.

North, D. C., and Thomas, R. (1973). The Rise of the Western World. A New Economic History , Cambridge, UK: Cambridge University Press.

Nurkse, R. (1953). Problems of Capital Formation in Underdeveloped Countries, Oxford: Blackwell.

Odhiambo, W., N’yangito, H. and Nzuma, J. (2004). Sources and Determinants of Agricultural Growth and Productivity in Kenya, KIPPRA Discussion Paper No. 34 , Kenya Institute for Public Policy Research and Analysis.

OECD (2006). The Development Effectiveness of Food Aid: Does Tying Matter? France: The development Dimension, OECD.

Olakojo, S.M. and Folawewo, A.O. (2011). Foreign Development Assistance and its Effect on Agricultural Growth in Nigeria, A conference paper presented at the AERC-UNU-WIDER conference on Macroeconomic Management of Foreign Aid, Nairobi 02-03 December 2011.

Oloka-Onyango, J. and Barya, J.J. (1997). Civil society and the political economy of foreign aid in Uganda, Democratization , 4(2), 113 — 138

Onwuegbuzie, A.J. and Leech, N.L. (2005). Taking the “Q” Out of Research: Teaching Research Methodology Courses Without the Divide Between Quantitative and Qualitative Paradigms, Quality and Quantity , 39: 267- 296.

Osei, R. and Quartey, P. (2001). The HIPC Initiative and Poverty Reduction in Ghana: An Assessment, UNU- WIDER Discussion Paper No. 2001/119.

Ottaway, M. and Chung, T. (1999). Debating Democracy Assistance: Toward a New Paradigm, Journal of Democracy , 10(4): 99-118.

Ouattara, B. (2004). Foreign Aid and Fiscal Policy in Senegal, Mimeo, University of Manchester

Ouattara, B. and Strobl, E. (2008). The Impact of Aid on Growth: An Aid Disaggregation Approach, Proceedings of the Development Economics Conference, Zurich.

Oyejide, T.A. (1986). The Effects of Trade and Exchange Rate Policies on Agriculture in Nigeria , IFPRI Research Report 55.

Pack, H. and Pack, J.R. (1993). Foreign aid and the question of fungibility, Review of Economics and Statistics , 75(2): 258-265.

264

Papanek, G.F. (1973). Aid, foreign private investment, savings and growth in less developed countries, Journal of Political Economy, 81(1):120-30.

The Patriotic Vanguard (2011). Maada Bio is SLPP Flagbearer. (August, 2). Available at: www.thepatrioticvanguard.com

Pavarala, V. (1993). Corruption as a Site for Contested Meaning: Elite Constructions in India, Qualitative Sociology, 16(4): 405-422.

Pesaran, M. H. and Pesaran, B. (1997). Working with Microfit 4.0: Interactive Econometric Analysis . Oxford, UK: Oxford University Press.

Pesaran, H.M. and Shin,Y. (1999). An Autoregressive Distributed-Lag Modelling Approach to Cointegration Analysis. In S. Strom (ed.) Econometrics and Economic Theory in the 20 th Century: The Ragna Frisch Centennial Symposium ; UK: Cambridge University Press; Pp. 371-413.

Pesaran, M. H., Shin, Y. and Smith, R.J. (2001). Bounds Testing Approaches to the Analysis of Level Relationships, Journal of Applied Econometrics, 16(3): 289-326.

Piron, L-H. with Evans, A.. (2004). Politics and the PRSP Approach: Synthesis Paper, ODI Working Paper 237 , London: Overseas Development Institute.

Pottie, D. (2003). Party Finance and the Politics of Money in Southern Africa, Journal of Contemporary African Studies, 21 (1): 5-26

Przeworski, A. and Limongi, F. (1993). Political Regimes and Economic Growth, Journal of Economic Perspectives, 7: 51-69.

Radelet, S. (2006). A Primer on Foreign Aid, Working Paper 92, Washington, D.C.: Center for Global Development.

Rajan, R.G. and Subramanian, A. (2008). Aid and Growth: What Does the Cross-Country Evidence Really Show? The Review of Economics and Statistics, 90(4): 643-665.

Rajkumar, A. S., and Swaroop, V. (2002). Public Spending and Outcomes: Does Governance Matter? World Bank Policy Research Working Paper 2840. Washington, DC: World Bank.

Ram, R. (1985). Exports and economic growth: Some additional evidence, Economic Development and Cultural Change , 12: 59-74.

Ram, R. (2003). Roles of Bilateral and Multilateral Aid in Economic Growth of Developing Countries, Kyklos 56(1): 95–110

Randel, J., German, T., Ewing, D. eds. (2004). The Reality of Aid , London: Zed Books

Ravallion, M., and Datt. G. (1996). How Important to India’s Poor is the Sectoral Composition of Economic Growth, World Bank Economic Review, 10(1).

Ray, D. (1998). Development Economics , New Jersey: Princeton University Press

Reddy, S., and Pogge, T. (2002). How not to count the poor. Available from .

Reinhart, C.M. (1995). Devaluation, Relative Prices and International Trade: Evidence from Developing Countries, IMF Staff Papers , 42: 290-312

Riddell, R.C. (2007). Does Foreign Aid Really Work? Oxford: Oxford University Press

Robinson, J A. (2008) “Governance and Political Economy Constraints to World Bank CAS Priorities in Sierra Leone” Harvard University.

265

Rodney, W. (1972). How Europe Underdeveloped Africa, London: Bogle-L’Ouverture publications.

Rodrik, D. and Subramanian, A. (2004). From Hindu Growth to Productivity Surge: The Mystery of the Indian Growth Transition, IMF Working Paper, wp/04/77 ; Available at: http://www.imf.org/external/pubs/ft/wp/2004/wp0477.pdf

Rogerson, S. (2005). Aid Harmonisation and Alignment: Bridging the Gaps between Reality and the Paris Reform Agenda, Development Policy Review , 23 (5): 531-552.

Romer, P. (1986). Increasing Returns and Long-Run Growth, Journal of Political Economy , 92: 1002-1037.

Roodman, D. (2003). Letter to the Forum for applied and theoretical researchers in econometrics. Available at: http://www.stata.com/statalist/archive/2003-11/msg00756.html

Roodman, D. (2006). How to do xtabond2: an introduction to “Difference” and “System” GMM in Stata, Center for Global Development Working Paper Number 103 .

Rostow, W. W. (1959). The Stages of Economic Growth, Economic History Review, 13 (1): 450-8.

Ruttan, V.W. (1979). Inflation and Productivity. American Journal of Agricultural Economics , 61(5): 896-902.

Sachs, J., McArthur, J., Schmidt-Traub, G., Kruk, M., Bahadur, C., Faye, M. and McCord, G. (2004). Ending Africa’s Poverty Trap, Brookings Papers on Economic Activity, Volume 1.

Sachs, J. (2005). The End of Poverty: how we can make it happen in our lifetime. London.

Saunders, C. (2008). Investigate the Economic Effects of Foreign Aid on Economic Growth, Economics Project, Department of Economics, University of Essex

Schultz, T. W. (1960). Value of US farm surpluses to underdeveloped countries, Journal of Farm Economics, 42(5): 1019-1030

Sen, A. (1999) Development as Freedom . Oxford: Oxford University Press.

Shrestha, M.B. and Chowdhury, K. (2005). ARDL modelling approach to testing the financial liberalization hypothesis , Economics Working Paper Series 2005, University of Wollongong.

Sesay, M. (1995). State capacity and the politics of economic reform in Sierra Leone, Journal of Contemporary African Studies , 13(2): 165 — 192

Solow, R. (1956). A Contribution to the Theory of Economic Growth, Quarterly Journal of Economics , 70(1): 65-94 .

Solow, R. (2001). Applying Growth Theory across Countries, World Bank Economic Review, 15(2): 283-88.

Soto, M. (2010). System GMM Estimation with a Small Sample, Instituto de Análisis Económico, Barcelona

Sricharoena, T and Buchenriederb, G. (2005). Principal component analysis of poverty in northern Thailand, Conference on International Agricultural Research for Development .

Standard Times Press News Paper (2008). Micro Finance and Poverty Alleviation in Sierra Leone. (November 18). Available at: www.standardtimespress.net

Stewart, F. and Wang, M. (2003). Do PRSPs Empower Poor Countries and Disempower the World Bank, or is it the Other way Around? Queen Elizabeth House WP 108 . At: www2.qeh.ox.ac.uk/pdf/qehwp/qehwps108.pdf

Stiglitz, J.E. (2005). Post-Washington Consensus , Initiative for Policy Dialogue Working Series, New York: Columbia University.

Tanzi, V. and Chu, K. eds. (1998). Income Distribution and High-Quality Growth , Cambridge: MIT Press

266

Tarp, F. (2006). Aid on Development. Swedish Economic Policy Review , 13(2): 9-62

Temple, J.R.W. (2010). Aid and Conditionality. In Rodrik, D. and Rosenzweig, M.R. (ed.) Handbook of Development Economics Volume 5 . Pp. 4415-4523. Amsterdam: North-Holland.

Teorell, J., Charron, N., Samanni, M., Holmberg, S. and Rothstein, B. (2009). The Quality of Government Dataset, version 17June09. University of Gothenburg: The Quality of Government Institute, At: http://www.qog.pol.gu.se

Thirlwall, A.P. (2003). Growth and Development; Seventh Edition , New York: Palgrave MacMillan

Thompson, B. (2007). Sierra Leone: Reform or Relapse? Conflict and Governance Reform, A Chatham House Report. London: Royal Institute for International Affairs. At: www.chathamhouse.org.uk Thorbecke, E. (2000). “The Evolution of the Development Doctrine and the Role of Foreign Aid, 1950-2000”. In F. Tarp (ed.) Foreign Aid and Development: Lessons Learnt and Directions for the Future . pp 17-47

Timmer, C.P. (1997). How Well do the Poor Connect to the Growth Process? HIID Discussion Paper No.178, Cambridge, MA: HIID.

Tsikaka, T.M., (1998). Aid effectiveness: a survey of the recent empirical literature, IMF Paper on Policy Analysis and Assessment No. 98/1.

Tybout, J.R. (1992). Linking trade and productivity: New research directions, World Bank Economic Review , 11(2): 151-177.

UNCTAD (2000). Tax Incentives and Foreign Direct Investment A Global Survey, ASIT Advisory Studies No. 16.

UNDP (2002). Human Development report 2002 , New York: Oxford University Press.

UNDP (2007). Seminar Report Governance Assessments and the Paris Declaration: Opportunities for Inclusive Participation and National Ownership, Bergen Seminar, Norway. de Vaus, D. (2002). Surveys in Social Research , Fifth Edition, London: Routledge

Verschoor, A. (2002). Aid and the poverty-sensitivity of the public sector budget. Research Programme on Risk, Labour Markets and Pro-poor Growth: Occasional Paper 3 , Department of Economics, University of Sheffield.

Verschoor, and A. Kalwij, A. (2006). Aid, Social Policies and Growth; Journal of International Development , 18: 519-532

Villavicencio, L. and Bara, J.L.R. (2008). Short-run and Long-run Determinants of Real Exchange Rate in Mexico, The Developing Economies , XLVI: 52-74

Walton, O. (2010). Youth, armed violence and job creation programmes: A Rapid Mapping Study, GSDRC Research Papers , University of Birmingham and the Norwegian Peace building Centre (NOREF).

White, H. (1992). The macroeconomic impact of development aid: a critical survey, Journal of Development Studies, 28(2):163-240. White, H. (1998). Aid and Macroeconomic Performance: Theory, Empirical Evidence and Four Country Cases . London: Macmillan.

White, J. (1974). The Politics of Foreign Aid , London: The Bodley Head.

Williamson, C.R. (2008). Foreign Aid and Human Development: The Impact of Foreign Aid to the Health Sector, Southern Economic Journal, 75(1): 188-207

267

Windmeijer, F. (2006). GMM for Panel Count Data Models, Discussion Paper No. 06/591 , Department of Economics, University of Bristol.

Winters, L. A. (2002). Trade Liberalisation and Poverty: What are the Links? World Econ. 25(9): 1339–67.

Winters, L.A., McCulloch, N. and McKay, A. (2004). Trade Liberalization and Poverty: The Evidence So Far, Journal of Economic Literature, XLII : 72–115

Wolf, S. (2007). Does Aid Improve Public Service Delivery? UNU-WIDER Research paper No. 2007/71

World Bank (1989). Sub-Saharan Africa: From Crisis to Sustainable Growth: A Long-Term Perspective Study; Washington, D.C.: World Bank.

World Bank (1998). Assessing Aid: What Works, What Doesn’t, and Why . Washington DC: World Bank.

World Bank (2001). World Development report 2000/ 2001: Attacking poverty . New York: Oxford University Press.

World Bank (2002). Global Economic Prospects and the Developing Countries . Washington D.C.: The World Bank. Available at www.worldbank.org

World Bank (2009). African Development Indicators (Edition: January 2009) . ESDS International, University of Manchester.

World Bank (2011). World Development Indicators (Edition: April 2011) . ESDS International, University of Manchester. DOI: http://dx.doi.org/10.5257/wb/wdi/2011-04

Younas, J. (2008). Motivation for bilateral aid allocation: Altruism or trade benefits, European Journal of Political Economy, 24: 661–674

Zack-Williams, A.B. (1999). Sierra Leone: The Political Economy of Civil War, 1991-98. Third World Quarterly , 20(1): 143-162

268

Appendix

Chapter 1: Introduction and Research Background

Appendix 1.1: History of Foreign Aid Context Development Donor Focus New forms/rising Orthodoxy forms of aid

1960s Major bilateral aid Role for state in Productivity, Technical assistance, programs established planning and infrastructure project aid investment, perhaps also in production

1970s Expansion of As in 1960s Agriculture, basic Greater focus on low multilaterals (IMF, needs income countries World Bank)

1980s Debt crisis, fading of Washington Macroeconomic Program aid, central planning, rise consensus reform, growth structural adjustment, of NGOs lending, debt relief

1990s HIPC debt problems, Market friendly Poverty alleviation, Sector-wide support, aid fatigue, post-Cold institutions, export governance, HIPC initiative War transition and FDI promotions investment climate

2000s Anti-globalization Uncertainty Millennium Poverty reduction movements, high Development Goals, strategy papers, media profile of world Governance, Selective aid poverty, “war on Health/HIV-AIDS allocation, Budget terror” support, post-conflict aid

Source: Temple (2010: 4424)

Appendix 1.2: Historical Human Development Ranking for Sierra Leone (1990-2007)

Year Total No. of Rank for Countries Sierra Leone 1990 160 159 1991 173 172 1992 174 173 1993 174 173 1994 175 175 1995 174 174 1996 - - 1997 174 174 1998 174 174 1999 162 162 2000 173 173 2001 175 175 2002 177 177 2003 - - 2004 176 177 2005 177 177 2006 - - 2007 182 180 Source: Annual UNDP Human Development Reports (Various)

269

Chapter 2: Foreign Aid and Economic Development in Sierra Leone: An Overview

Appendix 2.1: Post-Colonial Regimes in Sierra Leone Head of State Date Type of Regime Comments Sir Milton Margai 1961-1964 Conservative and Established the SLPP. Relied on chiefs' pragmatist; multi- patronage for support. Died of natural causes party system in 1964 while in office. Sir Albert Margai 1964-1967 Reformist; non- Preferred a popular appeal for SLPP. Accused of corruption, tribalism and ideologue authoritarianism in his last days. Andrew Juxon-Smith 1967-1968 Military junta Implemented an austere IMF reform package; set up commissions of inquiry; removed from power by a counter-coup. Siaka P. Stevens 1968-1985 Personalist; Founded the APC party; introduced republic authoritarian; and one party state; corrupt; enlarged the accommodationist bureaucracy; failed to introduce economic reforms; managed the succession. Joseph S. Momoh 1985-1992 Personalist; benign Abolished one party in 1991 and introduced authoritarianism; some liberal reforms; intervened in Liberia nationalist in 1990; rebel activity started in 1991; in rhetoric overthrown by a military coup d’état. Valentine Strasser/Maada 1992-1996 Military junta Initially populist; carried out the largest series of political executions in December Bio 1992; set up commissions of inquiry on ex- ministers; rebel activity intensifies. Ahmed Tejan Kabbah 1996-1997 Multi-party First political regime with some features of modern democracy. Was however toppled system/democratic by a Military junta Headed by Major Johnny governance Paul Koroma on the 25 th May 1997. Johnny Paul Koroma 1997-1998 Military junta Ruled for barely a year as the toppled democratic presidency of Tejan Kabbah fought back with assistance from ECOMOG and British Forces. Ahmed Tejan Kabbah 1998-2007 Multi-party The democratic regime resumed power before mid 1998 and ruled unto 2007 system/democratic winning another election in 2002 following governance. Notable the declaration of the end of the civil war few months before the elections. However, political and in January 1999 the government had been governance reforms temporary disjointed as the rebel fought into the capital city but could not establish full control of governance and was again driven out with assistance of ECOMOG forces Ernest Bai Koroma 2007-Date Multi-party In 2007, the first peaceful transition of governance took place between two system/democratic democratic civil regimes. The opposition governance. More of APC party that had earlier ruled the country for 23 years won the elections under the a populist leader with leadership of Ernest Bai Koroma. To date, much political will in he remains the president of the country controlling corruption Source: Updated from Sesay (1995:168)

270

Appendix 2.2: Per Capita GDP Growth: Sierra Leone versus Underdeveloped Countries Overall 1970- 1975 1980- 1985- 1990- 1995- 2000- 2005- Period 74 - 79 84 89 94 99 04 07 Average HIPC 1.10 0.25 -1.52 -0.72 -2.74 1.63 1.44 3.18 0.33 Low income -0.10 -0.61 -0.69 0.58 -1.23 1.86 2.07 3.99 0.73 Developing 3.70 2.68 1.06 1.71 1.15 2.43 3.65 6.62 2.87 Sub-Saharan Africa 2.87 -0.34 -1.22 -0.25 -2.05 0.72 1.52 3.53 0.60 Sierra Leone 1.85 -0.48 0.76 -3.17 -2.67 -6.28 9.28 3.61 0.36 Source: Computed from World Bank ‘s (2011) World Development Indicators

Appendix 2.3: Human Development Index: Sierra Leone versus Africa Overall 1980-84 1985-89 1990-94 1995-99 2000-04 2005-07 Average Sierra Leone 0.23 0.24 0.23 0.23 0.27 0.30 0.25 SSA 0.37 0.32 0.41 0.44 0.46 0.49 0.42 Africa 0.39 0.35 0.43 0.46 0.48 0.51 0.44 Source: Compiled from UNDP’s Human development reports (various). Coverage for Africa and SSA is exclusive to the countries that constitute the sample for the econometric analysis in Chapter 7

Appendix 2.4: Infants Mortality Rate: Sierra Leone versus Africa Overall Period 1980-84 1985-89 1990-94 1995-99 2000-04 2005-07 Average S/L 188.80 184.00 192.60 171.00 166.32 159.45 177.0 SSA 115.25 104.69 98.07 91.19 88.37 86.83 97.4 Africa 112.34 99.91 92.12 84.89 81.72 79.09 91.7 Source: Compiled from UNDP’s Human development reports (various) and the World Bank’s World Development Indicators. Coverage for Africa and SSA is exclusive to the countries that constitute the sample for the econometric analysis in Chapter 7

Appendix 2.5: Sierra Leone: Sectoral Contribution to GDP (1970-2007) (in %) Period Agriculture Industry Services Pre-war 35.1 18.5 40.0 war 46.5 30.9 17.5 Post-war 43.9 23.5 27.4 Total (1970-2007) 39.8 22.9 31.5 Source: Computed from World Bank (2011)

271

Chapter 4: Conceptual Framework and Overall Research Strategy

Appendix 4.1: List of Fieldwork Interviews

No. Designation Office/Agency Public Audit/Monitoring Institutions 1 Director of Public Education Anti-Corruption Commission 2 Deputy Auditor General Auditor General’s Office 3 Director Internal Audit Department 4 Ex-Chairman Parliamentary Oversight Committee on Finance 5 Deputy Chairman Parliamentary Oversight Committee for National Commission for Social Action 6 Deputy Chairman Parliamentary Oversight Committee, Public Accounts Committee 7 Director Central Planning, Monitoring and Evaluation Department, MO Donor Agencies 8 Economic Adviser European Union 9 Country Representative International Monetary Fund (IMF) 10 United Nations Development Programme (UNDP) 11 Country Economist African Development Bank (AfDB) 12 Economic Adviser UK department for International Development (DFID) 13 Country Economist The World Bank Politicians 14 Clerk of Parliament and Chairman, Parliamentary Oversight Committee for Presidential Affairs

Civil societies 15 Chairman Coalition of Civil society groups 16 Head Network Movement for Justice and Development (NMJD) – Mines Division 17 Head Network Movement for Justice and Development (NMJD) – Budget Oversight Division 18 ENCISS and Westminster Foundation 19 Programme Manager Campaign for Good Governance 20 Manager, Budgets Analysis National Accountability Group 21 Coordinator Budget Advocacy Network 22 Action Aid Media Elites 23 Acting Editor Awareness Times Newspaper 24 Managing Editor Eagle Radio, Citizen Radio 25 Secretary General Sierra Leone Association of Journalists (SLAJ) 26 Managing Editor For the People newspaper 27 Managing Editor Peep Magazine 28 Managing Editor The Independent Observer Newspaper Legal Elites 29 Researcher Law Reforms commission 30 Director of Research Law Reforms Commission 31 Deputy Master and Registrar Sierra Leone Law Courts Policy Experts/Elites 32 Director Economic Policy and Research Unit

272

No. Designation Office/Agency (EPRU) – Ministry of Finance 33 Deputy Commissioner National Commission for Social Action (NaCSA) 34 Ex-Development Secretary Ministry of Economic Planning and Development 35 Senior Budget Analyst Budget Bureau 36 Strategic Policy Unit – Office of the President 37 Director Multilateral Aid Department, Ministry of Finance and Economic Development 38 Monitoring and Evaluation Specialist Development Assistance Coordinating Office (DACO) 39 Acting Director Public Debt Unit, Ministry of Finance and Economic Development National Researchers/Consultants 40 Director Centre for Economic and Social Policy Analysis (CESPA)/BEST

273

274

Appendix 4.2: Summary of the Overall Research Design No. Research Specific Research Proposition/ The Type of Data to Data Collection Method Method of Data Analysis Objective Research Hypothesis be collected and Sources Question Has foreign aid Foreign aid to Sierra To examine had any Leone has had a Annual official statistics Desk research of official Quantitative analysis involving within- 1 the impact of significant significant impact on the consisting of foreign aid statistics obtained from World method triangulation of time series aid on contribution on country’s economic inflows, investment, Bank’s WDI databases, IMF’s econometric analysis with ARDL bounds economic economic growth growth over the years macroeconomic variables, IFS and GFS databases, test approach to cointegration and growth in Sierra Leone? and institutional quality OECD databases, ICRG Johansen Maximum Likelihood data databases, and POLITY IV Cointegration approach databases

Does foreign aid Foreign aid significantly Annual official statistics Desk research of official Quantitative analysis involving time- improves pro-poor growth on agricultural GDP statistics obtained the OECD, series econometric analysis with ARDL significantly in Sierra Leone growth, macroeconomic World Bank’s WDI, IMF’s approach to cointegration and the contribute to pro- variables, and political IFS, POLITY IV database. Johansen Maximum Likelihood To determine instability variables Cointegration approach. poor growth in 2 the impact of aid on Sierra Leone? poverty

reduction

275

Appendix 4.2: Summary of the Overall Research Design No. Research Specific Research Proposition/ The Type of Data to Data Collection Method Method of Data Analysis Objective Research Hypothesis be collected and Sources Question Annual official statistics Desk research of official Quantitative analysis involving panel on human welfare statistics obtained from data regression analysis with interactive indicators, macroeconomic UNDP’s HDR reports, OECD, Sierra Leone country dummy. Is foreign aid a variables, pro-poor social World Bank’s WDI, POLITY Some qualitative analysis involving significant expenditures, military IV database, and ICRG analysis of interviews with donors, civil determinant of Foreign aid is effective in expenditure, and database society, public audit institutions and aggregate human improving human institutional quality policy officials; and content analysis of welfare in Sierra development in Sierra variables. relevant reports to provide explanations Leone Leone to the findings from the quantitative Qualitative data include assessment. views and perceptions of public oversight and audit institutions, donor agencies, government officials, aid managers, the national elite groups and civil societies.

3 Annual official statistics Desk research of official Largely quantitative analysis involving on human development statistics obtained from the inclusion of an aid*political Is the system and The System and Quality of indicators, macroeconomic UNDP’s HDR reports, OECD, rule*country dummy interaction term in quality of politics Domestic politics matter in variables, pro-poor social World Bank’s WDI, POLITY the Welfare model. detrimental to the making aid effective in expenditures, military IV database, and ICRG Some qualitative analysis involving To assess the effectiveness of Sierra Leone expenditure, democracy, database analysis of interviews with donors, civil effect of foreign aid in autocracy and polity society, public audit institutions and domestic promoting human quality variables Qualitative method comprises policy officials; and content analysis of politics on development in the expert interviews with public relevant reports to complement the aid’s case of Sierra Qualitative data include officials managing and findings from the quantitative analysis. effectiveness Leone? views and perceptions of monitoring aid, political in improving public oversight and audit leaders, representatives of human institutions, donor government audit and

welfare agencies, government oversight institutions, civil officials, aid managers, the society representatives, media

276

4 • In a low-income Annual official statistics Desk survey of official Largely quantitative analysis Country like Sierra consisting of disaggregated statistics obtained from the complemented with qualitative analysis. Leone, grants rather foreign aid inflows, sources indicated in objectives The quantitative analysis involves time than loans are more investment, and 1 and 2 above; and series econometric analysis, and panel important for macroeconomic variables, documentary resources and cross country analysis for the impact of improving economic trade statistics, reports from government aid disaggregated aid types on human growth and reducing institutional quality, public managers, auditors, monitors development. The qualitative analysis poverty. expenditure, military and oversight institutions as involves discourse analysis of in-depth • Technical assistance expenditure, politics, and well as donor agencies; expert key informant interviews and content Does aid has not had a aggregated human welfare. interviews from these same analysis of documentary evidence to disaggregation significant impact in sources ; and researcher provide explanations to the findings from matter in the promoting economic Qualitative data include observation. the quantitative analysis. examination of the growth and reducing views and perceptions of impact of aid in poverty in Sierra public oversight and audit Sierra Leone? Leone. institutions, donor • In a low-income agencies, government country with less officials, aid managers, the significant political national elite groups and importance (as is the civil societies. case with Sierra Leone), bilateral assistance can have a favourable development impact. • In cognisance of its role and form, food aid does have a positive and moderately significant impact in fostering pro- poor growth.

277

Chapter 5: Foreign Aid and Economic Growth Appendix 5.1: Principal Component Analysis for the generation of the Policy Index

Principal Component Eigen Values % of Variance Cumulative %

1 1.736 57.873 57.873

2 0.638 21.263 79.136 3 0.626 20.864 100.000 Variable Component Loadings Component Score

INF -0.763 -0.440

TOPEN 0.763 0.439 GEX 0.757 0.436 Where INF = inflation rate; TOPEN = trade openness; and GEX = government expenditure share of GDP

Appendix 5.2A: The Impact of Aid on Economic Growth: Summary statistics Note :- Variable definition and sources are as defined in text.

Variable Obs. Mean Std. Dev. Min Max LRGDP 38 20.61 0.18 20.23 21.04 LAID 38 2.27 0.99 0.43 3.75 LPI 38 1.65 0.74 -1.31 2.46 LIQ1 38 1.53 0.21 1.08 1.79 POLICY 38 -0.0008 1.00 -2.03 1.77

Appendix 5.2B : Plots of the log level variables as used in the model

21.5 21.0 LRGDP 20.5 20.0 1970 1975 1980 1985 1990 1995 2000 2005 2007 Years

4 3 2 LAID 1 0 1970 1975 1980 1985 1990 1995 2000 2005 2007 Years

278

2.5 2.0 1.5 1.0 0.5 LPI 0.0 -0.5 -1.0 -1.5 1970 1975 1980 1985 1990 1995 2000 2005 2007 Years

1.8 1.6 1.4 LIQ1 1.2 1.0 1970 1975 1980 1985 1990 1995 2000 2005 2007 Years

2.0 1.5 1.0 0.5 0.0 -0.5 POLICY -1.0 -1.5 -2.0 -2.5 1970 1975 1980 1985 1990 1995 2000 2005 2007 Years

279

Appendix 5.3: Long-run Impact of aid on economic growth – Robustness Specifications using ARDL approach DEPENDENT VARIABLE IS LOG REAL GDP (LRGDP) Model B Model C MODEL D: Foreign Aid 0.542** 0.290** 0.159*** (2.070) (2.222) (3.355) Private Investment 0.374*** 0.200* 0.171*** (2.959) (1.928) (2.853) Polity 0.012 (1.131) Governance 1.727* (1.690) Property Rights 0.886 (0.951) Macro Policy 0.238 0.079 -0.020 (1.318) (0.840) (0.435) CONSTANT 17.416*** 18.424*** 20.141*** (11.237) (10.316) (147.671) Crisis -0.270 -0.292* (1.136) (2.033)

F-TEST FOR COINTEGRATION 4.500** 6.202*** 3.862* SERIAL CORRELATION 2.73 4.24** 1.26 FUNCTIONAL FORM 0.001 0.004 0.52 NORMALITY 3.41 0.90 2.45 HETEROSCEDASTICITY 0.66 1.82 0.27 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

Appendix 5.4: Short-run Impact of Aid on Economic Growth using the ARDL Approach - Robustness Check Specifications

Dependent variable is LOG REAL GDP (dLRGDP) Regressors MODEL B: MODEL C: MODEL D: First lag of Real GDP -0.473** -0.379* (2.720) (1.889) Foreign Aid 0.008 0.068*** 0.056*** (0.315) (4.083) (3.355) Private Investment 0.073*** 0.047** 0.060* (3.484) (2.061) (2.084) Polity 0.002 (0.392) First lag of Polity -0.014*** (2.785) Governance 0.047(0.536) Property Rights 0.206 (1.441) Macro Policy 0.047*** 0.018 -0.007 (2.918) (1.229) (0.404) Constant 3.414* 4.291* 7.079*** (1.756) (1.767) (3.386) Crisis -0.063 -0.103** (-0.937) (2.690) Ecm(-1) -0.196** -0.233** -0.351*** (2.065) (2.076) (3.386) R2 0.71 0.67 0.71 Adjusted R 2 0.62 0.58 0.62 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

280

Appendix 5.5: Long-run Impact of Foreign Aid on Economic Growth using Johansen Approach-– Robustness Specifications Johansen Long Run Estimates and Diagnostic Tests DEPENDENT VARIABLE IS GROWTH IN REAL GDP (LRGDP) Model B Model C Model D Foreign Aid 0.298* 0.112** 0. 276*** (0.1510) (0.050) (0.068) Private Investment 0.279*** 1.000 1.000 (0.0570) (none) (none) Polity -0.025 (0.0155) Governance 0.787 (0.6363) Property Rights -1.181** (0.571) Macro Policy 0.090 0.010 0.125* (0.1083) (0.037) (0.074) Constant 18.846*** 22.419*** 20.144*** (0.8666) (0.990) (0.172) Crisis -0.137*** -0.142*** (0.042) (0.0262) Short-Run Result ECM (-1) -0.222*** -0.403*** -0.244*** (0.0492) (0.084) (0.082)

Var Order (SBC) 1 1 1 Cointegration Test R (Maximal Eigen) 1** 2** 2* R (Trace ) 1** 2** 2** R (used in Regression) 1 2 2 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance – inclusive of F-Test for cointegration. Standard errors in Parenthesis. Note that the signs of the long-run estimates have been appropriately adjusted in the table presentation to allow easy comparison with the ARDL estimates.

Appendix 5.6: ADF Unit Roots Test for Aid Disaggregates with Intercept but without trend: 1970-2007 Levels 1st Difference Variable SIC ADF t-stat. SIC Lag ADF t-stat. I(d) Lag LGRANT 0 -1.236322 0 -6.612102 I(1) LOAN 0 -3.731575 I(0) LLOAN 0 -3.717416 I(0) LTCA 0 -1.874788 0 -5.926002 I(1) LNTCA 0 -2.383794 0 -8.395034 I(1) LBA 0 -1.789344 0 -7.841462 I(1) LMA 0 -2.217629 0 -6.799115 I(1) LFA 0 -3.660756 I(0) LNFA 0 -2.069593 0 -7.428356 I(1) Critical Value 1% 2 -3.632900 0 -3.626784 5% -2.948404 -2.945842 Critical Value 1% 0 -3.621023 5% -2.943427

281

Appendix 5.7: Short-run Impact of aid Structure on economic growth –Grant versus Loan using ARDL Approach Dependent variable is growth in LOG real GDP (dLRGDP) Regressors MODEL A: First Lag of Real GDP -0.686*** (4.097) Grants -0.009 (0.334) Loans 0.007** (2.337) Private Investment 0.046** (2.253) Governance Property Rights 0.292*** (3.061) Macro Policy 0.023* (1.943) First Lag of Macro Policy CONSTANT 3.270* (1.753) Ecm(-1) -0.207 (2.276)** R2 0.78 Adjusted R 2 0.69 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

Appendix 5.8: Short-run Impact of Technical Assistance on economic growth using ARDL Approach - Robustness Check Dependent variable is growth in Log Real GDP (dLRGDP) Regressors MODEL A: First Lag of Real GDP -0.592*** (3.278) Tech Coop Assist -0.005 (0.137) Non-Tech Coop Assist 0.028 (1.292) Private Investment 0.053** (2.451) Governance Property Rights 0.352*** (4.889) Macro Policy 0.033** (2.164) CONSTANT 3.605* (1.978) Ecm(-1) -0.210** (2.339) R2 0.70 Adjusted R 2 0.61 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

282

Appendix 5.9: Short-run Impact of Foreign Aid Structure on Economic Growth - Bilateral aid versus Multilateral Aid using ARDL Approach Dependent variable is growth in Log Real GDP (dLRGDP) Regressors MODEL A MODEL B First Lag of Real GDP -0.565*** -0.474** (3.006) (2.655) Multilateral Aid 0.013 -0.003 (0.691) (0.157) Bilateral Aid 0.017 0.009 (0.718) (0.393) Private Investment 0.055** 0.072*** (2.586) (3.372) Governance 0.041 (0.471) Property Rights 0.323*** (4.152) Macro Policy 0.024* 0.048*** (1.810) (2.938) CONSTANT 3.187* 3.571* (1.785) (1.802) Ecm(-1) -0.187** -0.199** (2.149) (2.059) R2 0.69 0.72 Adjusted R 2 0.60 0.61 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

283

Appendix 5.7.1: CUSUM and CUSUMSQ Plots for Model on Impact of Grants and Loans on Economic Growth

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

284

Appendix 5.8.1: CUSUM and CUSUMSQ Plots for Model on Impact of Technical Assistance on Economic Growth

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

285

Appendix 5.9.1: CUSUM and CUSUMSQ Plots for Model on Impact of Bilateral and Multilateral Assistance on Economic Growth

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Chapter 6: Foreign Aid and Pro-poor Growth Appendix 6.1: Principal Component Analysis for the generation of the Trade Policy Index (TPOLI)

Principal Component Eigen Values % of Variance Cumulative % 1 1.463 73.163 73.163 2 0.537 26.837 100.000 Variable Component Loadings Component Score TOPEN 0.855 0.585 RER 0.855 0.585

Where TOPEN = trade openness; and RER = Real Exchange Rate. Both variables are standardized.

286

Appendix 6.2: Error Correction representation (Short-run Estimates) of the impact of Food aid on Pro- poor Growth Using ARDL Approach Dependent Variable is Log of Real Agricultural GDP (dLRAGDP) Regressor Model 1:– Model 2: First Lag of Real Agric GDP - - Food Aid 0.006(0.382) 0.004(0.238) Non-Food Aid -0.051(1.132) -0.048(1.071) First Lag of Non-Food Aid -0.095(2.477)** -0.098(2.452)** Private Investment -0.013(0.467) -0.015(0.493) Trade Policy 0.007(0.418) Agric. Trade -0.014(0.225) Constant 8.242(4.356)*** 8.402(4.432)*** Political Crisis - - 0.162(3.082)*** 0.165(3.167)*** Ecm(-1) - - 0.421(4.312)*** 0.429(4.416)*** R2 0.62 0.62 Adjusted R 2 0.49 0.49 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

Appendix 6.3: Error Correction representation (Short-run Estimates) of the impact Technical Cooperation Assistance on Pro-poor Growth Using ARDL Approach Dependent Variable is Log of Real Agricultural GDP (dLRAGDP) Regressor Model 1 Model 2 Model 3:

First Lag of Real GDP -0.254(1.855)* - -0.287(2.124)** Tech. Assistance 0.006(0.130) 0.043(0.847) -0.004(0.088) Non-Tech. Assistance -0.100(2.668)** -0.092(2.553)** -0.090(2.463)** First lag of Non-Tech. Assistance -0.112(3.528)*** -0.095(3.278)*** -0.123(3.855)*** Private Investment -0.023(0.897) -0.021(0.773) -0.028(1.139) Trade Policy 0.001(0.073) Trade Openness 0.043(0.678) First Lag of Trade Openness -0.139(2.048)** Agric. Trade -0.072(1.303) Constant 6.487(3.600)*** 7.00(3.879)*** 6.874(3.938)*** Political Crisis -0.165(3.396)*** -0.139(2.954)*** -0.172(3.647)*** Ecm(-1) -.329(3.544)*** -0.365(4.131)*** -0.342(3.841)*** R2 0.67 0.70 0.69 Adjusted R 2 0.55 0.58 0.58 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

Appendix 6.4: Error Correction representation (Short-run Estimates) of the impact of Grants on Pro- poor Growth Using ARDL Approach Dependent Variable is Log of Real Agricultural GDP (dLRAGDP) Regressor Model 1: Model 2 First Lag of Real Agric GDP - - Grant -0.022(0.441) 0.058(2.245)** Loan -0.011(1.996)* Log (Loan) -0.002(0.098) Private Investment -0.015(-0.425) -0.026(0.840) Trade Openness 0.012(0.166) -0.029(0.409) Constant 6.061(2.273)** 5.423(2.584)** Political Crisis -0.117(1.802)* -0.189(3.441)*** Ecm(-1) -0.313(2.362)** -0.276(2.644)** R2 0.53 0.54 Adjusted R 2 0.39 0.43 Note :- * significant at 10%, ** significant at 5%, and *** significant at 1% level of significance; T-statistic in parenthesis

287

Appendix 6.2.1: CUSUM and CUSUMSQ Plots for Model on Food Aid and Pro-poor Growth

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0

-5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

288

Appendix 6.3.1: CUSUM and CUSUMSQ Plots for Model on Technical Assistance and Pro-Poor Growth

Plot of Cumulative Sum of Recursive Residuals 15

10 5 0

-5

-10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

289

Appendix 6.4.1: CUSUM and CUSUMSQ Plots for Model on Grants versus Loans on pro-poor growth

Plot of Cumulative Sum of Recursive Residuals 15 10 5 0 -5 -10 -15 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Plot of Cumulative Sum of Squares of Recursive Residuals 1.5

1.0

0.5

0.0

-0.5 1972 1977 1982 1987 1992 1997 2002 20072007 The straight lines represent critical bounds at 5% significance level

Chapter 7: Foreign Aid and Welfare

Appendix 7.1: Construction of the pro-poor expenditure (PPE) index Following the methodology of the literature (Gomanee and Morrissey, 2002, Mosley et al., 2004; Gomanee et al., 2005a; 2005b; and Verschoor and Kalwij, 2006), the index of pro-poor government expenditure has mostly been derived by combining the expenditures on education, health, and sanitation. However, for the majority of the countries in the panel including Sierra Leone, the country of interest, data on public expenditures on sanitation for the period of study is insufficiently available. Thus, in this study, we construct the pro-poor

290 public expenditure index using the two crucial welfare expenditures of health and education related in the following form:

PPE = P e + P h

Where P e is government expenditure on education; and P h is government expenditure on health. The choice of these public expenditure categories as those found to be pro-poor, is in tune with most of the literature (e.g. Barro, 1991; Chu and others, 1995; Tanzi and Chu, 1998; Gupta et al., 1999; Rajkumar and Swaroop, 2002; Verschoor, 2002; Dabla-Norris et al., 2004). We use the unweighted PPE index. Though in the studies by Gomanee and Morrissey (2002) and Gomanee et al (2005a), they used both weighted and unweighted indices of PPE, yet it did not significantly affect outcome of the results. Moreover, the evidence of the impact of these individual social expenditures on welfare is mixed and therefore establishing which one maybe more important than the other is not a priority for this study; and we hence use the unweighted PPE index in our empirical model. This implies the index is constructed by simply adding the health and education expenditures for each corresponding period for which observations are available.

Appendix 7.2: AID and Pro-poor Expenditure Sample Africa Sub-Saharan Africa Dependent Variable/Model PPE (Model 1) PPE (Model 2) Regressors PPE(-1) 0.738*** 0.713*** (3.80) (3.54) GDPPC(-1) -0.096 -0.088 (0.79) (0.75) AID -0.104 -0.113 (1.40) (1.39) TGR 0.042 0.046 (0.65) (0.60) Constant 1.279 1.274 (1.51) (1.53) No. of Observations 150 130 No. of Instruments 13 13 F-stat. 14.38 10.19 AR(2) 0.565 0.584 Hansen (OIR) Test 0.060 0.092 Note :- All variables are in logs. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%. OIR denotes Overidentifying Restrictions.

291

Appendix 7.3: Foreign Aid and Human Development Sample Africa Sub-Saharan Africa Dep. Variable/Model HDI HDI HDI HDI (Model 1) (Model2) (Model 3) (Model4) Regressors HDI(-1) 0.190 0.200 0.171 0.181 (1.45) (1.64) (1.22) (1.37) AID -0.039 -0.064 -0.091 -0.115 (0.63) (1.21) (1.26) (1.50) AID*DSL - 0.264*** - 0.298*** (2.90) (2.95) PPE 0.048 0.049 0.072* 0.075* (1.06) (1.15) (1.88) (1.94) ME -0.057 -0.069 -0.086 -0.101* (1.21) (1.39) (1.57) (1.78) GDPPC(-1) 0.137 0.106 0.069 0.045 (1.63) (1.31) (0.76) (0.47) IQU 0.200* 0.216* 0.247* 0.261* (1.83) (1.90) (1.95) (1.99) DSL -0.362*** -1.114*** -0.270** -1.132*** (3.32) (4.34) (2.27) (4.46) Constant -1.747** -1.479** -1.230 -1.008 (2.42) (2.14) (1.64) (1.28) No. Of Obs. 135 135 116 116 No. of Instruments. 20 24 20 24 F-Stat 228.74 402.47 99.77 230.08 AR(2) 0.532 0.703 0.648 0.839 Hansen Test for (OIR) 0.314 0.727 0.635 0.982 Note :- All variables are in logs except dummy. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%. Regression with time dummies added but not reported

292

Appendix 7.4: Foreign Aid and Infant Mortality Rate Sample Africa Sub-Saharan Africa Dep. Variable/Model IMR IMR IMR IMR (Model1) (Model 2) (Model3) (Model4) Regressors IMR(-1) 1.203*** 1.203*** 0.853*** 0.851*** (7.67) (7.69) (3.36) (3.37) AID 0.016 0.017 -0.050 -0.046 (0.25) (0.26) (0.82) (0.77) AID*DSL - -0.072 - -0.022 (1.03) (0.35) PPE -0.005 -0.005 -0.009 -0.011 (0.11) (0.11) (0.28) (0.32) ME -0.006 -0.005 -0.005 -0.005 (0.18) (0.14) (0.18) (0.20) GDPPC(-1) 0.067 0.067 -0.069 -0.066 (0.84) (0.86) (0.90) (0.90) IQU -0.011 -0.012 -0.009 -0.0.010 (0.15) (0.16) (0.12) (0.14) DSL -0.130 -0.085 0.187 0.250 (1.12) (0.39) (1.06) (1.15) Constant -1.440 -1.441 1.259 1.246 (1.40) (1.40) (0.80) (0.81) No. Of Obs. 153 153 133 133 No. of Instruments 20 24 20 24 F-Stat 453.92 748.81 448.55 775.99 AR(2) 0.646 0.682 0.470 0.487 Hansen Test for (OIR) 0.060 0.181 0.060 0.173

Note :- All variables are in logs except dummy. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%. Regression with time dummies added but not reported

293

Appendix 7.5: Structure of Aid and Human Development Sample Africa Sub-Saharan Africa Dep. Variable/Model HDI(M1) HDI(M2) HDI(M3) HDI(M4) Regressor HDI(-1) 0.216* 0.209* 0.213 0.186 (1.75) (1.67) (1.64) (1.37)

GRANT -0.038 - -0.081 - (0.89) (1.53) LOAN -0.003 - -0.004 - (0.77) (0.88) TCA - -0.018 - -0.025 (0.24) (0.25) NTCA - -0.039 - -0.058 (0.99) (1.41) GRANT*DSL 0.285*** - 0.288*** - (4.75) (4.12) LOAN*DSL -0.008 - -0.001 - (0.43) (0.06) TCA*DSL - 0.013 - -0.007 (0.20) (0.09) NTCA*DSL - 0.208*** - 0.236*** (3.24) (3.16) ME -0.068 -0.072 -0.098 -0.108** (1.33) (1.62) (1.65) (2.08) PPE 0.048 0.054 0.071* 0.077* (1.04) (1.24) (1.82) (1.91) GDPPC(-1) 0.139* 0.108* 0.086 0.075 (1.69) (1.66) (1.14) (1.00) IQU 0.182* 0.211 0.219* 0.245* (1.66) (1.63) (1.86) (1.67) DSL -0.079*** -0.926*** -1.047*** -0.931*** (4.64) (6.12) (4.66) (5.57) Constant -1.733** -1.546*** -1.317* -1.342** (2.41) (2.82) (1.99) (2.42) No. of Obs. 134 133 115 114 No. of Instruments 29 29 29 29 F-Stat 492.59 139.44 345.56 118.72 AR(2) 0.764 0.733 0.943 0.695 Hansen Test for OIR 0.997 0.775 0.993 0.646 Note :- All variables are in logs except dummy and loan variable. T statistic in parenthesis. *** significant at 1% ** significant at 5% and * significant at the 10%. Regression with time dummies added but not reported

294

Appendix 7.6: Politics and aid effectiveness (SSA Sample) Dependent Variable is HDI Model Model 1 Model2 Model3 Model4 Model5 Model6 Regressors HDI(-1) 0.142 0.146 0.144 0.145 0.131 0.135*** (1.05) (1.08) (1.06) (1.06) (0.92) (0.94) AID -0.115 -0.117 -0.060 -0.111 -0.115 -0.117 (1.50) (1.57) (1.21) (1.49) (1.48) (1.53) POLIT 0.004 0.004 - - - - (1.21) (1.21) POLIT*DSL 0.023*** - - - - - (4.61) AID*POLIT*DSL - 0.009*** - - - (7.05) DEM - - 0.011 0.011 - - (1.13) (1.13) DEM*DSL - - 0.062*** - - - (6.99)

0.020*** AID*DEM*DSL - - - (7.38) - -

- AUT - - - -0.015 -0.015 (1.43) (1.41)

AUT*DSL - - - - -0.017* - (1.68)

AID*AUT*DSL ------0.009*** (2.94)

PPE 0.108*** 0.106*** 0.101*** 0.098*** 0.108*** 0.105*** (3.16) (3.08) (2.81) (2.77) (3.17) (3.06) ME -0.105** -0.107** -0.105** -0.106** -0.101** -0.104** (2.07) (2.07) (2.12) (2.15) (2.08) (2.10) GDPPC(-1) 0.076 0.076 0.071 0.078* 0.071 0.071 (0.87) (0.89) (0.79) (0.88) (0.80) (0.82) DSL -0.267*** -0.267*** -0.420*** -0.434*** -0.268*** -0.243*** (3.12) (3.27) (4.71) (5.16) (3.12) (2.89) Constant -1.191 -1.174 -1.171 -1.223** -1.125 -1.110 (1.62) (1.63) (1.56) (1.66) (1.50) (1.51) No. Of Obs. 116 116 116 116 116 116 No. of Instruments 17 20 17 18 17 20 F-Stat 84.65 72.96 638.89 642.38 90.92 62.27 AR(2) 0.637 0.572 0.745 0.734 0.630 0.623 Hansen Test for 0.339 0.649 0.361 0.473 0.425 0.723 (OIR) Note:- all variables are in logs except dummy and ‘Polity regime’, democracy and autocracy variables. T statistic in parenthesis. *** Significant at 1% ** significant at 5% and * significant at the 10%.

295