Households' Vulnerability to Poverty in Ibadan

Journal of Rural Economics and Development vol. 20 No. 1

Rural Economics and Development

HOUSEHOLDS’ VULNERABILITY TO POVERTY IN IBADAN METROPOLIS, OYO STATE, NIGERIA

Adepoju, A. O. and F.Y. Okunmadewa Department of Agricultural Economics, University of Ibadan, Ibadan, Oyo State, Nigeria

Abstract This paper empirically assessed vulnerability to poverty at the household level using a two-period panel data set obtained from 150 households sampled from two local government areas within Ibadan Metropolis. Data were analysed using descriptive statistics, poverty indices and probit regression analysis. Analysis of the socio-economic characteristics and their relationship with vulnerability to poverty revealed that large-sized households headed by men who were old, widowed, self-employed, uneducated or who had only primary school education and no access to any form of credit, were more vulnerable than other households. The estimated probit regression model showed that marital status and tertiary education status of respondents reduced vulnerability to poverty while primary education status and household size enhanced households vulnerability to poverty. JEL-Classification: I31, I32, R20 Keywords: Vulnerability, Poverty, Oyo State, Nigeria

*Correspondence E-mail [email protected]; Telephone number: +2348033335958

Introduction projects has brought the issue of vulnerability The high incidence of poverty in to the attention of policy makers. Nigeria, despite myriads of interventions by Vulnerability, has been defined as the governments and NGOs to reduce it through likelihood that at a given time in the future, an poverty alleviation/reduction programmes and individual will have a level of welfare below

Journal of Rural Economics and Development vol. 20 No. 1

some norm or benchmark (Quisumbing, Poverty assessments draw on cross- 2002).One likely reason poverty has been on sectional household survey to provide a the increase may be that it has been seen by detailed profile of the poor, and to document several researchers as a static phenomenon the incidence of poverty in various segments rather than a dynamic one. Recent studies of the population. The incidence of poverty in have however observed movements in and out Nigeria however remains high. The World of poverty of households in developing Bank (1996) statistics on income and social countries (Adams and He,1995). This is an indicators show poverty in Nigeria to be indication that poverty is not a static widespread, severe and most certainly phenomenon as people can move out of and increasing. This brings to the fore, the issue fall into poverty. According to Baulch and of vulnerability which is defined as the McCulloch (1998) “a high percentage of probability that a household if currently poor, households move into poverty due to will remain in poverty or if currently non- temporary shocks (such as illness or loss of poor will fall below the poverty line in employment) that are reversed just one or two explaining the ever-increasing level of years later. Similarly, many of the people who poverty. escape poverty or who are not vulnerable now Vulnerability however, which suggests only succeed in doing or being so for one or exposure to the possibility of an adverse two years before a reverse in their outcome in the future, has not been widely circumstances forces them back below used alongside poverty in discussions of poverty line which makes them vulnerable”. poverty reduction strategies even though the Ligon and Schechter (2003) defined risks that households face are an important the essence of vulnerability as the uncertainty aspect of their wellbeing. This shows a of future income streams and associated loss limited understanding of a household of welfare caused by this uncertainty. They vulnerability to poverty. While it is noted that “a household with very low commonly asserted that the poor are among expected consumption expenditures but with the most vulnerable in any society (e.g. no chance of starving may well be poor but it World Bank, 2001), the overlap between still might not wish to trade places with a poverty and vulnerability is not perfect. There household having a higher expected seems to be general agreement that poverty is consumption but greater consumption risk”. a static concept, defined at a single point in However, it is not every time people are time, while the concept of vulnerability exposed to risk that they are vulnerable i.e. a situated in a dynamic context is less well shock might occur, but may not necessarily defined. Clarifying the distinction between lead to the households being vulnerable. The poverty and vulnerability is important concept of vulnerability therefore, is dynamic especially since social protection strategy is and is broadly an ex-ante or forward looking moving from ex-post poverty strategies to ex- measure of a household’s well being or (lack ante vulnerability considerations (Holzman, thereof). Hence, when thinking about 2001). forward-looking anti-poverty interventions In most developing economies, that aim to prevent rather than alleviate estimation of vulnerability has been mainly poverty, what really matters is the through the use of cross-sectional household vulnerability of households to poverty”. survey data but in principle the use of panel data permits the estimation of vulnerability

Journal of Rural Economics and Development vol. 20 No. 1

within a more general framework and allows occurring i.e. with less than certainty. for the inclusion of time-invariant household Individuals/households have some a priori effects and dynamic effect and in some cases sense of the likelihood of these events to get a sense of the magnitude of biases in occurring, without direct control over its estimates of vulnerability generated from likelihood. The lack of direct control over the cross-sectional data (Chaudhuri, 2000).This risk they face is crucial and distinguishes it study will therefore attempt to contribute to from the responses one can observe from an understanding of vulnerability of individuals, households and communities given households to poverty in Nigeria since a pre- the risk they face. While the concept of risk condition for successful anti-vulnerability refers to uncertain events that can damage the policies is the identification of the group of well being of people such as falling ill, vulnerable households, together with an (Christiaensen and Subbarao, 2001), understanding of the sources of vulnerability. vulnerability is a function of the risk Consequently, there is a need for government characterization of a person’s environment - to proactively take measures to protect the nature, frequency and severity of the shocks vulnerable households and in order to do so, he is exposed to, his exposure to these risks as vulnerable households have to be identified. well as his ability to cope with it when the The nature of their vulnerability also needs to shock materializes which is determined by his be examined. The main objective of the study asset endowments and his ability to insure is to assess the vulnerability of households to himself (formally or informally) (Alayande, poverty in Ibadan Metropolis. The rest of the 2002). Vulnerability is therefore the product of paper is in four sections. Section two presents risk, but also of household conditions and the literature review while section three actions (Dercon, 2001). describes the methodology of the study. A World Bank study on risk Section four presents the empirical findings management in South Asia also defines while section five concludes the paper. vulnerability as the likelihood of being adversely affected by a shock that usually 2. Literature review causes consumption levels, or other factors that affect well-being to drop (World Bank, 2.1 Risks, vulnerability and poverty 2001b). On the other hand, Chambers (1989) There are many definitions of opined that vulnerability is one among the vulnerability, and seemingly, no consensus on different dimensions of deprivation, which its definition and measurement. (Chaudhuri, include such other concepts as physical 2000; Christiaensen and Subbarao, 2001) weakness, isolation, poverty and define vulnerability as the ex-ante potential of a powerlessness. Therefore in addition to risk decline in future well being, or the ex-ante exposure, which signifies the probability that probability of falling below the poverty line at a person will be affected by uncertain events some future date. In support of this, McCulloch which may lead to welfare loss, vulnerability and Calandrino (2003) view vulnerability as reflects the lack of capacity to cope with a the probability of being below the poverty line shock ex-post. It is concerned with the ex- in any one year. Vulnerability is ante potential of a decline in well-being in the multidimensional, and households face a future. Thus, it is a dynamic concept that number of risk. The risk faced by an generally involves a sequence of events individual/household relates to events possibly following some shocks (Alayande, 2002).

Journal of Rural Economics and Development vol. 20 No. 1

Concepts of vulnerability and poverty elements. First is one due to a low level of (which is also multidimensional) are linked and limited variance in consumption and a but not identical. For example, Chaudhuri, et second due to high level of and much al. (2002), submit that vulnerability is an ex- variance of consumption. However, ante (forward-looking) rather than an ex-post measuring income and consumption concept. Whereas poverty status can be dynamics and variability requires specific observed at a specific time period, given the types of data: These include cross sectional welfare measure and the poverty threshold, data and longitudinal data. Relying on single household vulnerability is not directly cross-sectional data requires making stringent observed, rather it can only be predicted. The assumptions regarding the stochastic process observed poverty status of a household generating consumption e.g. that cross (defined simply by whether or not the sectional variability proxies interpersonal household’s observed level of consumption variation. These sets of data are always expenditure is above or below a pre selected available because they are relatively cheaper poverty line) is the ex-post realization of a to obtain especially for developing countries. state, the ex-ante probability of which can be According to Hoddinot and Quisumbing taken to be the household’s level of (2003a), carefully collected cross-sectional vulnerability. Therefore, while it is possible data reveal much about risk and vulnerability, to make statements about whether or not a particularly if they are augmented by use of household is currently poor, it is not possible secondary sources, community and make statements about household’s level of qualitative field work. On the other hand, the vulnerability. Also, while we can estimate or scope of risk and vulnerability assessment is make inferences about whether a household is greatly enhanced if longitudinal household currently vulnerable to future poverty, we data are available because longitudinal data can not directly observe a household’s current allow the same household to be tracked over vulnerability level (Chaudhuri, et al. 2002). a sufficient length of time. These permit the According to Ligon and Schechter (2003), direct estimation of the inter-temporal traditional poverty measures neglect several variance of consumption at the household important dimensions of household welfare level without the need for strong assumptions. while vulnerability measures allow the However, this should not be taken to imply quantification of welfare loss associated with that longitudinal data are both necessary and poverty as well as the loss associated with sufficient for vulnerability assessments any of a variety of different sources of because their dearth and limited cross- uncertainty. Again, while poverty is sectional coverage render them not quite concerned with not having enough now, useful for policy analysis that requires vulnerability is about a high probability now nationally representative samples (Chaudhuri, of suffering a future shortfall (Christiaensen 2000). They are also time-consuming to and Boisvert, 2000). However, it is pertinent collect and their collection requires strong to say that though in practice, the poor are data documentation skills so that interviewers often also vulnerable, both groups (poor and can find individuals and households in order vulnerable) are not typically identical (Sen to re-interview them. However, the consensus 1998, Baulch and Hoddinot 2000). in literature (e.g. Glewwe and Hall, 1998; According to Alayande (2002), the Chaudhuri 2000) is that longitudinal data are measurement of vulnerability has two most appropriate for the study of

Journal of Rural Economics and Development vol. 20 No. 1

vulnerability. It is in this context that this The study used a multistage sampling study utilized longitudinal data in the technique in selecting the representative examination of vulnerability of households in households. The first stage was the selection Ibadan Metropolis. of two local government areas in Ibadan city, namely, Ibadan North and Ibadan South 3. Methodology West. The second stage involved random sampling of areas within these local 3.1 Data government areas. These areas include: The study was conducted in Ibadan University of Ibadan, Agbowo, Bodija, UCH, metropolis, the capital of Oyo State. The Orogun for Ibadan North, and Odo Ona, metropolis is composed of 11 Local Oluyole, Oluyole Extension, Iyaganku for government areas, 6 at the outskirts and 5 at Ibadan South West. In the third stage, the the centre. The latter are: Ibadan South East, households surveyed were randomly selected Ibadan North East, Ibadan North West, to make them representative of the two Ibadan South West and Ibadan North Local Local Governments. In the second survey Government Areas. Ibadan is located between round, the same households were selected in longitude 70 20” and 70 40” East of the order to track the characteristics of the Greenwich meridian and between latitude 30 households at the two different periods. A 55” and 4010” North of the equator. The city is hundred and fifty (150) households were in the equatorial rain forest belt and has a interviewed in the first survey exercise but land area of between 445 and 455km2 with an only 133 households could be re-interviewed estimated population of 1,991,367 persons in the second round. Hence, only the data according to the 1991 population census. from these 133 households were used for Ibadan metropolis is an important commercial analysis in this study. centre and it comprises of people of different cultural and socio-economic backgrounds. 3.2 Analytical tools Predominantly, food crops such as yam, 3.2.1 Poverty measures maize, cowpea, okro, melon which reflect the dietary habits of the inhabitants are grown as The Foster, Greer and Thorbecke (1984) is clearly seen in the type of meals taken by weighted poverty index was used for the the people. Data used in this survey were quantitative poverty assessment, the FGT collected from a two-round panel survey measure for the ith sub group (Pai) is given as: undertaken at 3-month interval to allow q  measurement of seasonal variation in i 1 P  z y  i  Z behaviour and outcome and to balance both n i1   the cross-sectional and time series 1 q 0 q when  0, P  z y   Povertyincidenceorheadcount 0  Z requirements of panel data. The first round n i1   n was in May 2005, while the second survey 1 q 1  1, P  z y  Povertygapordepth 1  Z was in August 2005.The primary source of n i1   data were collected with the use of structured 1 q 2  2, P  zy  Povertyseverity 2  Z questionnaire containing both open- and n i1   close-ended items. The questionnaire administration was cross-sectional in nature. Where,

Journal of Rural Economics and Development vol. 20 No. 1

Pai = weighted poverty index for the ith sub poor and always poor i.e. vulnerable = group (becoming poor + always poor). Table 1 ni = total numbers of the ith subgroup in show the implicate transitional matrix. poverty Where, = yji Per capita expenditure of households in n1 = numbers of households that were subgroups vulnerable in the two survey rounds

Zi = Poverty line for the subgroup n2 = numbers of households that were vulnerable in the first survey round but αi = degree of concern for the depth of non- vulnerable in the second survey poverty round.

α = 0 gives incidence of poverty (Head count n3 = numbers of households that were non- index) and is used to determine the vulnerable in the first survey round but percentage of the poor. vulnerable in the second survey.

α = 1 gives depth of poverty which is defined n4 = numbers of households that were non as the difference between poverty line vulnerable in the two survey rounds. and mean expenditure of the poor as a Y = Total numbers of respondents i.e. ( n1 + ratio of the poverty line. n2 + n3 + n4). The households were subdivided into two based on the measures of poverty as follows:- Vulnerability index - The probability of being always poor Vulnerability index for each subgroup was defined as being poor in the two survey calculated as: rounds. Number of vulnerable households in the - The probability of becoming poor defined subgroup as being non-poor in the first round but poor in the second survey. Total numbers of households in the subgroup Vulnerable households were then defined as a combination of those becoming

Table 1: Transitional matrix box Vulnerable Non-vulnerable Total

Vulnerable n1 n2 n1 + n2

Non-vulnerable n3 n4 n3 + n4

Total n1 + n3 n2 + n4 Y

Journal of Rural Economics and Development vol. 20 No. 1

3.2.1 Model specification for vulnerability Rearranging terms measurement Pr (Yi=1) = Pr [Ui>-(Bo + B1 X1i + B2 X2i + In order to ascertain the effect of certain …. Bk + Ki)] fctors on the vulnerability of households to = 1 – Pr [Ui< - (Bo + B1 X1i + B2 X2i poverty, a probit model was estimated using + ….. Bk + Ki]………..(3) data from the panel. The probit regression analysis was used since the OLS estimation If we make the usual assumption that U is procedure will not be appropriate, especially normally distributed, we have when most of the independent variables are Pr (Y=1) = 1 -  [-(Bo + B1 X1i + B2 X21 + ….. dichotomous. This arises due to the following Bk Xki)] reasons:- Non normality of the disturbances = 1 -  (-X B ) ui; Heterescedasticity of the disturbance-term; 1 3 The predictions of the logit – probit model =  (X1B)……..(4) offered by OLS lack boundedness since nothing constrains it from being either less where than 0 or greater than 1. Backward regression  = standard cumulative normal distribution was carried out in which the most using data from panel insignificant variable was excluded from the model and the regression run again at each X1 = vector of independent variables time until the model consists only of B’s = estimates of coefficients which give the significant variables. The probit model impact of the independent variables on assumes that while we observe the values of 0 the latent variable Y*. and 1 for the variable Y1 there is a latent, unobserved continuous variable Y* that The model is stated explicitly as:- determines the value of Y, we assume that Y* Y = f (X1, X2, X3, X4, X5, X6, X7, X8, X9, X10, can be specified as follows: X11, X12, X13) ……… (5)

Y* = Bo + B1Xii + B2 X 2i + … Bk Xki + Where U …………………(1) i Y = 1 if vulnerable (becoming poor + always and that: Yi = 1 if Y* > 0 poor)

Yi = 0 otherwise = 0 if otherwise

Where X1 = Sex of household head (D =1 If male, 0 if otherwise) Yi = poverty level (poor = 1, 0 = non poor)

X2 = Age of household head (years) X1i …. Xki = Vector of Independent variables

X3 = Marital status of household head (D = 1 Bo = constant if married, 0 if otherwise) B1 = coefficient estimates X4 = Marital status of house hold head (D = 1 Ui = random disturbance term if widowed 0 if otherwise)

Pr (Yi = 1) = (Bo + B1 X1i + B2 X2i ….. Bk Xki X5 = Household size (number) + Ui > 0…….. (2)

Journal of Rural Economics and Development vol. 20 No. 1

X6 = Primary Education of household head household that had MPCHHE below or equal (D=1 if primary education and 0 if to N5,528.14 or N6,612 was considered to be otherwise) poor for first and second survey rounds

X7 = Secondary Education of household head respectively, while households with per (D=1 if secondary education and 0 if capita expenditure above the amounts were otherwise) considered to be non-poor. Table 2 presents

X8 = Tertiary Education of household head the transitional matrix of households in the (D=1 if tertiary education and 0 if study area. The table reveals that 63 otherwise) households were vulnerable in both surveys

X9 = Occupation of household head (D = 1 if while 7 households were vulnerable in the wage earning and 0 if otherwise) first survey but non-vulnerable in the second survey round. Further, 7 households which X10 = Exposure to Covariate shocks dummy of household head were non-vulnerable in the first survey round had become vulnerable in the second survey X11= Exposure to Idiosyncratic shocks dummy of household head round while fifty six households were non- vulnerable in the 2 survey rounds. In all, the X12= Number of risks exposed to by the household head total number of vulnerable households in the study area stood at 70 and that of non- X13 = Access to formal credit dummy of the vulnerable households at 63. household head (D= if yes and 0 if The poverty and vulnerability profile otherwise) of households as presented in Table 3 4. Results and discussion revealed that poverty and vulnerability incidence were higher in the second survey 4.1 Poverty status of households period when compared with the first survey Poverty line was computed differently period. Specifically, in the first and second for the two survey rounds. On the basis of survey rounds, households with heads older than 65 years were found to be the poorest. relative poverty, the mean per capita Also, while male- headed households were household expenditure (MPCHHE) for the found to be poorer compared with their respondents stood at N8, 292.21 while the female counterparts in the first round, the two-thirds MPCHEE amounted to N5, 528.14 reverse was the case in the second round. for the first survey round. Likewise, in the This implies that both male and second survey, the MPCHHE stood at N9, female-headed households can indeed be poor 917.95 while the two-thirds MPCHHE depending on their level of exposure to risks. amounted to N6, 612. This means that any

Journal of Rural Economics and Development vol. 20 No. 1

Table 2: Transitional matrix of households in study area Vulnerable Non-vulnerable Total

Vulnerable 63 7 70

Non-vulnerable 7 56 63

Total 70 63 133

Table 3: Poverty incidence and vulnerability by socio-economic characteristics Age Poverty Incidence Vulnerability Index May August 25 – 45 0.29 0.48 0.48 46 – 65 0.44 0.61 0.62 > 65 0.90 0.75 0.75 Sex: Male 0.38 0.45 0.56 Female 0.20 0.52 0.44

Household size 1 – 4 0.25 0.35 0.42 5 – 9 0.32 0.50 0.65 7 – 9 0.39 0.53 0.75 Education Level No education 0.45 0.64 0.54 Primary 0.47 0.47 0.87 Secondary 0.30 0.44 0.46 Tertiary 0.28 0.35 0.44 Marital status Single 0.58 0.60 0.58 Married 0.34 0.43 0.51 Widowed 0.80 0.80 0.60 Occupational Status Wage earners 0.26 0.36 0.51 Non-wage earners 0.32 0.44 0.54 Access to credit None 0.39 0.50 0.63 Formal 0.33 0.34 0.56 Informal 0.31 0.47 0.39 All 0.36 0.44 0.47(May), 0.53(Aug)

Journal of Rural Economics and Development vol. 20 No. 1

Poverty was also found to increase with were found to be more vulnerable than those increase in household size. This may be with access to informal credit. This may be owing to the fact that a large household size due to the timely access to informal credit as tends to reduce per-capita expenditure against the lengthy appraisal of applications although it could enhance it depending on the for formal credit and requests for collateral distribution of household members between made by financial institutions which is adult and children and whether such adults practically non-existent for the poor. are working. This means that having a family, which includes more income earning 4.2 Household Vulnerability to poverty members thus a lower dependency ratio, reduces poverty. In support of this, Following the outlined analytical vulnerability of a household to poverty was procedure, the probit model was used. This found to increase with age of the household model has been used in many vulnerability head with vulnerability index highest for both studies e.g. Skoufias and Quisumbing (2002) household head aged above 65 years and also in their work on consumption insurance and large sized households. vulnerability and Byett (2002) in his work on The educational status of the measures of household vulnerability in which respondents showed that poverty and the probit regression was used to model the vulnerability decreased with increase in probability of a bank crisis. The results of the educational attainment, although vulnerability Probit Analysis (backward regression) are index was found to be highest for those with presented in tables 4 and 5. Table 4 presents primary education. This may be because their the initial analysis while table 5 presents the level of education may tempt them to seek final result of analysis after all the paid employment. Consequently, they end up insignificant variables had been excluded. in low cadre positions with a low level of Large household size invariably reduces income. Households where the household welfare of household members and therefore heads were married were found to be less can be said to enhance vulnerability to poor and less vulnerable than those with poverty. Expectedly, the result of the probit either single or widowed heads. The analysis showed that married household occupational status of the respondents heads were less vulnerable than single or revealed that wage earners were less poor and widowed household heads. This is so, less vulnerable than non-wage earners in the considering the negative sign of the two survey rounds. This may be connected to coefficient representing marital status. The the fact that being employed with a stable possible reason for this could be the ease of income reduces the likelihood of being poor risk sharing and pooling of resources together and of severe welfare loss whenever to jointly cater for household needs better. confronted with a risk. Also, the sign of the coefficient of tertiary Also, household heads without access education status was negative. This is an to any form of credit in the two survey rounds indication that the higher the years of formal were found to be the poorest and the most education obtained by household heads, the vulnerable. This implies that access to credit lower the odds of the household heads being reduces the likelihood of being poor. vulnerable. However, those with access to formal credit

Journal of Rural Economics and Development vol. 20 No. 1

Table 4: Determinants of Vulnerability to poverty of households (Initial) Variable Coefficient Z

Constant 0.282 -0.215

Sex 0.526 1.447

Age -0.032 -1.819*

MS1 -0.607 -1.147

MS2 0.743 0.631

HHS 0.436 3.870***

EDUST1 1.623 2.343**

EDUST2 -0.596 -0.879

EDUST3 -1.228 1.299

HHOCC -0.228 -0.754

EXPTOCO 1.166 1.299

EXPTOIDI 0.045 0.113

NOSOFRIS- -0.203 -1.272

ACCESSTO CR 0.093 0.321

Log likelihood function -63.946 Restricted log likelihood -92.004 Chi-squared 56.115 Degrees of freedom 13 significance level 0.000 ***, ** and * denote significance of coefficient at 1, 5 and 10 percent respectively.

Journal of Rural Economics and Development vol. 20 No. 1

Table 5: Determinants of Vulnerability to poverty of households (Final) Variable Name Coefficient Z

Constant -0.520 -1.115

MS1 -0.823* -1.890

HHS 0.329*** 3.740

EDUST1 1.362** 2.150

EDUST 3 -0.789*** -3.068

Log likelihood function -67.667 Restricted log likelihood -92.004 Chi-squared 48.672 Degrees of freedom 4 Significance level 0.000

***, ** and * denote significance of coefficient at 1, 5 and 10 percent respectively.

This suggests that improving the educational The study concludes that most of the attainment of household heads assists in households in the study area were vulnerable. getting good jobs and taking opportunities However, household heads were found to be which otherwise would not have been more vulnerable in the second survey round possible. The overall effect of this is with 53% vulnerability compared with the increased income, which translates to first survey round with 47% vulnerability. increased per capita expenditure and Also, vulnerable household heads were found consequently improved welfare and standard to be mostly uneducated or have at least of living of household members. On the other primary education. While primary education hand, the coefficients of household size and status and household size enhanced primary education were positive implying vulnerability to poverty, marital status and that large household size and low level of tertiary educational status reduced it. Thus, education attainment enhanced vulnerability considering the level of vulnerability in the in the study area. In sum, it can be inferred two periods, having in mind the various risks from the result obtained that low level of exposed to by the respondents, (which may educational attainment, large household size easily reverse their situations – especially and widowed or single – all increase or macroeconomic risks) a lot needs to be done enhance vulnerability to poverty in the study to improve the factors that reduce area. vulnerability to poverty. If the currently poor are targeted, a large proportion of the

households will move out of poverty between 5. Conclusion and Recommendations one period and the other and policy

Journal of Rural Economics and Development vol. 20 No. 1

interventions that help the currently poor interest rates to reduce the impact of income cannot be assumed to lead to a reduced risks and Government could assist through incidence of poverty in the next period ahead relaxation of any stringent guidelines in (i.e. vulnerability to poverty). This suggests securing such assistance (especially in the that different policies may be needed for case of formal credit). poverty reduction because focusing anti- poverty efforts on the correlates of current References poverty status (which could be as a result of exposure to a shock at that time) may not Adams, R.H. and J. J. He (1995). “Sources of have any significant impact on the probability Income Inequality and Poverty in Rural of being poor in the future but forward Pakistan, Research Report, No. 102, looking anti-poverty interventions that aim to International Food Policy Research prevent rather than alleviate poverty could be Institute (IFPRI): Washington, D.C. embarked upon. Alayande, B.A. (2002). “Determinants of The implication of the above findings Vulnerability to Poverty”, a World is that large sized households with old, Bank Commissioned Study. widowed heads who have no access to credit, earn low income and have no or low Byett, A. (2002). “Measures of Household educational qualification are the most Vulnerability Project on Indicators and vulnerable to poverty in the study area. Thus, Analysis of Vulnerabilities to following from the above, improvement in Economic Crises”. Final Report EADN the level of educational attainment is a major Regional Conference, September 2002. policy prescription emanating from this Baulch, B. and J. Hoddinott (eds) (2000) study. This is pertinent since having primary Economic Mobility and Poverty education predisposed people to vulnerability Dynamics in Developing Countries. in urban areas according to this study. It is Frank Cass, London. envisaged that the government and all concerned will follow through the Universal Baulch, B. and N. McCulloch (1998). “Being Basic Education Programme (UBE) which Poor and Becoming Poor: Poverty prescribed a nine-year mandatory education Status and Poverty Transitions in Rural for all citizens. This will enable people to Pakistan,” IDS Working Paper No. 79. acquire better education which can lead to Chambers, R. (1989). “Vulnerability, Coping improved income and by extension reduced and Policy” Institute of Development vulnerability to poverty. This is amply Studies Bulletin, 20(2): 1-7. demonstrated by the fact that those that have tertiary education are less vulnerable to Chaudhuri, S. (2000). “Empirical Methods poverty. In addition, there is need for for Assessing Household Vulnerability increased awareness on benefit of small to Poverty”, Mimeo Department of family size which can be incorporated into Economics and School of International family planning activities, while safety net and Public Affairs. Columbia programmes should be specifically targeted at University. aged and widowed heads. Lastly, credit/loan Chaudhuri, S.; Jalan, J. and A Suryahadi facilities should be made available and (2002). “Assessing Household accessible to target households at moderate

Journal of Rural Economics and Development vol. 20 No. 1

Vulnerability to Poverty from Cross- Holzmann, R., (2001). Risk and Sectional Data: A Methodology and Vulnerability: The Forward Looking Estimates from Indonesia”, Discussion Role of Social Protection in a Paper No. 0102-52, Department of Globalizing World. Paper Prepared for Economics, Columbia University. “The Asia and Pacific Forum on Poverty – Policy and Institutional Christiaensen, Lue and K. Subbarao (2001). Reforms for Poverty Reduction,” Asian “Towards an Understanding of Development Bank, Manila, February Vulnerability in Rural Kenya”. Mimeo. 5-9, 2001. The World Bank. Washington, D.C. Ligon, E. and L. Schechter (2003). Measuring Christiaensen, L.J. and R.N. Boisvert (2000). Vulnerability. Economic Journal 113: “On Measuring Household Food 95-102. Vulnerability: Case Evidence from Northern Mali.” Working Paper 2000- McCulloch, J. and C. Calandrino (2003). 05, Department of Applied Economics Poverty and Vulnerability, American and Management, Cornell University Economic Review Papers and Proceedings, 84 (2): 221-25 Dercon, S. (2001). “Assessing Vulnerability to Poverty”. Mimeo, Centre for the Quisumbing, A. (2002). “ Consumption Study of African Economies, Smoothing, Vulnerability, and Poverty University of Oxford. in Rural Bangladesh”, Draft. International Food Policy Research Ellis, F. (1998). “Household Strategies and Institute (IFPRI): Washington, D.C. Rural Livelihood Diversification.” Journal of Development Studies, 35 Sen, A., (1998). “The Concept of (1):1-38. Development” Handbook of Development Economics. Vol. 1, Foster, J.J., Greer, and E. Thorbecke (1984). Amsterdam: North Holland. “A Class of Decomposable Poverty Measures.” Econometrica 52 (3):761- Skoufias, E. and A.R. Quisumbing (2002). 766. “Consumption Insurance and Vulnerability to Poverty: A Synthesis Glewwe, P. and G. Hall (1998). ‘Are Some of the Evidence from Bangladesh. Groups More Vulnerable to Ethiopia, Mali, Mexico and Russia”. Macroeconomic Shocks than Others”? International Food Policy Research Hypothesis Tests Based on Panel Data Institute (IFPRI): Washington D.C. From Peru’. Journal of Development Economics. 56 (1) :181-206. World Bank (1996). “Poverty Reduction Handbook” The World Bank, Hoddinott, J. and A. Quisumbing (2003a) Washington, D.C. Data sources for Microeconometric Risk and Vulnerability Assessments, World Bank (2001). The World Development Report, Washington, D.C. World Bank Social Protection Discussion Paper No.0323, Washington, D.C.