UNIVERSITY OF NEW SOUTH WALES
SCHOOL OF BANKING AND FINANCE
INTERNATIONAL FINANCIAL SERVICES:
Determinants of Bancassurance Demand and Life
Insurance Consumption
CSABA SZABOLCS JOSA
This thesis is submitted in fulfillment of the requirements of the degree of Master of
Commerce (Honours) at the University of New South Wales.
2005 CERTIFICATION
I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.
Csaba Josa
2005
ii ABSTRACT
This thesis is a pioneer study that examines the growing importance of global insurance
markets and the factors that determine continued success and viability. A couple of issues
relating to risk and insurance that have not been examined to such an extent in previous
studies are represented through the examination of two of the fastest growing areas
within international insurance services, namely those of global bancassurance and life insurance markets. Firstly, this thesis establishes what determines the demand for bancassurance using a sample of 73 companies from 28 developed and developing countries from across the globe. Methodological improvements are made on previous studies through the use of the advanced estimation technique known as Generalized
Method of Moments (GMM) which helps account for such inconsistencies as measurement errors, heteroscedasticity, multicollinearity and endogeneity. The results obtained both extend and reconcile existing literature in the field of bancassurance.
Secondly, the determinants that influence the level of life insurance consumption
throughout the OECD are scrutinized with a particular focus on the influence of systems of law. Thus far no other study has attempted to discover the relationship between life insurance consumption patterns and the legal systems in place within a given nation. The key finding highlights the importance of systems of law on consumption patterns, and
specifies that there is a significant positive relationship between the French and German
civil-law systems and the level of life insurance consumption within the OECD. In
addition, the findings in regards to other demographic, macroeconomic and social
determinants extend as well as support the existing literature in the field of life insurance.
iii ACKNOWLEDGEMENTS
I would like to dedicate this thesis to my parents and my grandparents, who have never
wavered in their support of me throughout my university career. Their continued backing
and encouragement has made it possible for me to complete this monumental project.
Special thanks must also go to my two supervisors, Professor Fariborz Moshirian and Dr.
Donghui Li. I must thank Fari for his ongoing support and encouragement even during
times when I did not believe in myself to finish this project. His cheerfulness, motivation
and belief in me went a great deal towards helping me complete this thesis. To Donghui, I
would like to extend my thanks for providing the suggestions for the two chapters that are
covered within this thesis. I would also like to thank him for all his support and advice
without which I would most likely have strayed from my goals and failed to complete
this thesis.
I would also like to extend my thanks to all my fellow honours classmates at UNSW.
Your friendship and camaraderie have made these previous years some of the most
enjoyable of my life. Also, I would like to thank you for all the distractions that resulted
in this thesis taking a year longer to complete.
Most importantly, I would like to thank Mong Nhi for putting up with me during this
time. I am sure that my constant complaints about the progression of my thesis were not
easy to stand. Without your support, love and patience I truly believe that I would have given up on this thesis long ago.
iv TABLE OF CONTENTS
CERTIFICATION iiii
ABSTRACT iii iii
ACKNOWLEDGEMENTS iviv
TABLE OF CONTENTS vv
LIST OF TABLES & FIGURES xi
LIST OF ACRONYMS xii
CHAPTER 1
GENERAL INTRODUCTION
1.1. INTRODUCTION 1
1.2. ROLE OF INSURANCE AND A REVIEW OF GLOBAL TRENDS 3
1.3. PURPOSE AND SCOPE OF THIS STUDY 7
1.4. OBJECTIVES AND VALUE OF THESE CHAPTERS 8
1.4.1. The Determinants of Bancassurance 8
1.4.2. Law and the Determinants of Life Insurance in OECD Countries 11
1.5. THESIS STRUCTURE 15
v CHAPTER 2
THE DETERMINANTS OF BANCASSURANCE
2.1. INTRODUCTION 17
2.1.1. Chapter Outline 20
2.2. GENERAL DEFINITIONS 21
2.3. BANKS 22
2.3.1. Introduction 22
2.3.2. Current Vulnerabilities 23
2.3.3. Advantages of Banks over Insurance Providers 28
2.4. INSURERS 29
2.4.1. Introduction 29
2.4.2. Current Vulnerabilities 30
2.4.3. Advantages of Insurance Providers over Banks 32
2.5. BANCASSURANCE 33
2.5.1. Brief History 34
2.5.2. Why Does Bancassurance Occur? 36
2.5.2.1.Gains for Banks 43
2.5.2.2.Gains for Insurers 48
2.5.2.3.Gains for Consumers 51
2.5.3. Global Breakdown 52
2.5.4. Challenges Faced by Bancassurers 62
2.5.5. Quantitative Works of Major Researchers 66
vi 2.6. DATA AND METHODOLOGY 70
2.6.1. Data Sources 70
2.6.2. Determinants of Bancassurance 75
2.6.2.1.Explanatory Variables 77
2.6.2.1.1. Risk Proxy 77
2.6.2.1.2. Customer Base/Size Proxy 80
2.6.2.1.3. Cost Savings Proxy 82
2.6.2.1.4. Revenue Increase Proxy 85
2.6.2.1.5. National Banking Sector Proxy 86
2.6.2.1.6. Level of Deregulation (within a Country) 88
2.6.2.1.7. Demand for Insurance Proxy 90
2.6.2.1.8. Level of Inflation 92
2.6.3. Model for the Determinants of Bancassurance 95
2.6.4. Ordinary Least Squares (OLS) 99
2.6.5. Generalized Method of Moments (GMM) 99
2.7. EMPIRICAL RESULTS 103
2.7.1. Whole Sample 103
2.7.2. Whole Sample less the U.S. 106
2.7.3. European Sample 109
2.7.4. Legal Systems 112
2.7.5. Analysis and Implications of Results 116
2.8. CONCLUSION 120
2.8.1. General Conclusion 120
vii 2.8.2. Avenues for Further Study 124
CHAPTER 3
LAW AND THE DETERMINANTS OF LIFE INSURANCE IN OECD
COUNTRIES
3.1 INTRODUCTION 126
3.1.1 Chapter Outline 129
3.2 THE LIFE INSURANCE INDUSTRY AND SYSTEMS OF LAW 130
3.2.1 Life insurance in Current Times 130
3.2.2 Systems of Law (a Study of Legal Families) 135
3.2.3 Quantitative Works of Major Researchers 142
3.3 DATA AND METHODOLOGY 149
3.3.1 Data Sources 149
3.3.2 Law and the Determinants of Life Insurance 153
3.3.2.1 Explanatory Variables 154
3.3.2.1.1 Foreign Market Share 154
3.3.2.1.2 Dependency Ratio 156
3.3.2.1.3 Interest Rates 158
3.3.2.1.4 Inflation Levels 159
3.3.2.1.5 Life Expectancy 161
3.3.2.1.6 Foreign Direct Investment (FDI) 163
viii 3.3.2.1.7 Education 164
3.3.2.1.8 Uncertainty Avoidance Index (UAI) 166
3.3.2.1.9 Legal Rights 168
3.3.2.1.9.1 Minority Shareholder Protection 169
3.3.2.1.9.2 Investment Restriction 170
3.3.2.1.10 GDP per capita 172
3.3.2.1.11 Legal Systems 173
3.3.3 Model for the Determinants of Life Insurance 175
3.4 EMPIRICAL RESULTS AND THEIR IMPLICATIONS 179
3.4.1 Empirical Results 180
3.4.2 Analysis and Implications of Results 195
3.5 CONCLUSION 207
3.5.1 General Conclusion 207
3.5.2 Avenues for Further Study 211
CHAPTER 4
CONCLUSION
4.1 INTRODUCTION 213
4.2 THE DETERMINANTS OF BANCASSURANCE 214
ix 4.3 LAW AND THE DETERMINANTS OF LIFE INSURANCE IN OECD
COUNTRIES 217
4.4 SUGGESTIONS FOR FURTHER RESEARCH 220
APPENDIX 1 224
APPENDIX 2 227
REFERENCES 229
x LIST OF TABLES & FIGURES
Figure 2.1 Bancassurance Share of Life Insurance Business 35
Table 1.1 World Insurance Premiums (1995-2004) 6
Table 1.2 OECD Insurance Premiums (1995-2004) 6
Table 2.1 Sample Countries 70
Table 2.2 Descriptive Statistics 74
Table 2.3 Hypothesized Relationships 95
Table 2.4 Whole Sample Results 107
Table 2.5 Correlation Matrix for the Whole Sample 108
Table 2.6 Whole Sample (less the U.S.) Results 110
Table 2.7 Euro1 and Euro2 Results 113
Table 2.8 Civil-Law Nations Results 114
Table 3.1 Sample Countries 149
Table 3.2 Descriptive Statistics for the year 2003 152
Table 3.3 Hypothesized Relationships 176
Table 3.4 Pooled OLS and GMM Results 185
Table 3.5 Correlation Matrix 186
Table 3.6 Sub-Period Results 188
Table 3.7 Common-Law vs. Civil-Law Results 192
Table 3.8 Civil-Law Results 193
Table 3.9 Results Excluding the Systems of Law 194
Table 3.10 Premium Volumes 203
xi LIST OF ACRONYMS
APRA - Australian Prudential Regulation Authority
BHC - Bank Holding Company
BIC - Bayesian Information Criterion
CIA - Central Intelligence Agency
ECB - European Central Bank
GATS - General Agreement on Trades and Services
GDI - Gross Disposable Income
GDP - Gross Domestic Product
GLB - Gramm Leach Bliley act
GMM - Generalized Method of Moments
GNI - Gross National Income
IFS - International Financial Statistics (arm of IMF)
IMD - International Institute for Management Development
IMF - International Monetary Fund
OECD - Organization of Economic Co-operation and Development
OLS - Ordinary Least Squares
U.K. - United Kingdom
U.S. - United States (of America)
USD - United States Dollar
WTO - World Trade Organization
xii CHAPTER 1
GENERAL INTRODUCTION
"If I have the belief that I can do it, I shall surely acquire the capacity to do it even if I
may not have it at the beginning."
- Gandhi. Indian spiritual and political leader, called Mahatma "great soul".
1.1 INTRODUCTION
This thesis examines the growing importance of the current global insurance markets and
the factors that determine their overall success and viability. Meticulous attention is paid
to two of the fastest growing areas within international insurance services, namely those
of the global bancassurance and life insurance markets. Insurance acts as an important
contributor to the trade in financial services, providing benefits to the consumer, business owner and economy alike. In its simplest form, insurance involves the management and
pricing of risks that arise from the pooling of insurance contracts. As such, insurance
alleviates risks inherent in everyday activities for individuals and firms, allowing greater
risk-taking by these parties, which in turn promotes the demand for goods and services
1 and facilitates economic development. Insurance also acts as a vehicle for savings and investments thereby, channeling funds to their most productive economic uses.
With the continued increase in globalization and deregulations over the past few decades, the global economy has witnessed the proliferation of trade in international financial services. While many nations, especially those of developing countries, still employ restrictive regulations in regards to foreign participation in local insurance services, continued liberalization has seen the gradual opening of such markets to the forces of globalization. Insurance services comprise an ever increasing portion of the global financial services market, a trend which is set to continue with the sweeping deregulations that are taking place across the globe, the advancements made in technology that assist in improving distribution and product quality, and the obfuscation between the traditional roles of banks and insurers through trends such as bancassurance1.
As such, the importance of insurance services to the continued health of the global economy cannot be overstated. Studies such as those presented within this thesis aim to assist policymakers and regulators, as well as consumers and firms, in gaining a better understanding of what drives the success of trade in international financial services through analyzing the determinants of the global insurance industry. This allows for appropriate decisions to be implemented that promotes continued growth within
1 One of the major contributors to such trends is the furthering of the General Agreement on Trades and Service (GATS) as focused on during the latest Uruguay Round of Trade Negotiations. The furthering of GATS has paved the way for increased deregulations and the lowering of barriers to trade in the financial services sector, including the banking, insurance and securities sectors. Furthermore, initiatives such as GATS have assisted in the blurring of what comprises the traditional functions of banks and insurers, allowing for greater intermingling to occur. This then allows the differing sectors to benefit from the knowledge and expertise that is brought about through the union with members from other industries.
2 international insurance services, which will also ensure the lasting strong growth within the global economy.
1.2 ROLE OF INSURANCE AND REVIEW OF GLOBAL TRENDS
Since both of the chapters within this thesis focus on the determinants of international insurance services, we deem it useful for the reader to understand the definition and shape of the current global insurance industry. The role of insurance within the global marketplace is gaining in importance every year and currently forms a large part of the trade in international financial services. With the globalization, deregulations and industrializations of recent decades, this trend is set to continue in both developed and developing nations as the world moves towards an integrated marketplace that exists without barriers to trade.
Insurance provides many useful services to consumer, businesses and the economy alike.
On a broad level, insurance assists economic development via the central role it plays in the financial markets, allowing for the global spreading of risks, increasing the individual countries risk-taking capacity, and enabling businesses to undertake more innovative activities in relation to product development and making use of alternate manufacturing processes. However, the primary purpose is generally agreed to be the protection of individuals and firms against adverse events occurring. These events could encompass
3 death and disability, which would fall under the cover of life and health insurance
policies, and the destruction of property and workers-compensation, which would fall
under the banner of non-life insurance. Through taking on their roles in the risk transfer and indemnification process, insurance companies are compensated through the receipt of premiums from the insured policyholders, and as such act as a financial intermediary which in itself plays a significant role in promoting economic development.
The main difference of an insurer as a financial intermediary to that say of a commercial bank, lies in terminology, since bank customers are referred to as borrowers or savers,
whereas those of an insurer are known as the insured. Insurers thus sell contingent claim
contracts that pay-off the occurrence of a given event in some future period. Insurers also
assist in reducing the transaction search costs by matching sources and users of funds, in
turn improving the efficiency of the financial system. Other roles insurers are involved in
that assist economic development include the reduction of asymmetric information in the
financial marketplace, creation of liquidity, mobilizing savings and then channeling them
towards their most productive economic uses, and engaging in risk transformation.
Regardless of the possible benefits that insurance services may provide, it is argued by
Fukuyama (1995) that the characteristics of the individual country in terms of culture and
perceptions towards risks will considerably mitigate the full benefits inherent. This
argument is supported by Hofstede (1995) who argues that insurance levels will depend
on national culture and the willingness of individuals to use insurance as a means of
dealing with risks.
4 Below, we provide two tables that highlight the growth within the world and OECD
insurance industries respectively over the ten-year period of 1995-2004. Several
important trends can be derived from these tables including the fact that the growth in
both the world and OECD insurance industries has been increasing for the most part throughout this ten year period, with the growth rate averaging around 10 percent for both over the past three-year period. Life insurance premiums around the world (OECD)
have grown from $1239.6USD million ($1182.9USD million) in 1995 to $1848.7USD million ($1681.2USD million) in 2004, an increase of 49.14 percent (42.12 percent).
Non-life insurance premiums, while comprising a smaller amount of total insurance premiums have outgrown life insurance over the same period, with growth around the world (OECD) increasing from $908.2USD million ($845USD million) in 1995 to
$1395.2USD million (1286.4USD million) in 2004, which comprises a ten-year growth rate of 53.63 percent (52.23 percent). These figures clearly indicate the growing importance of insurance services throughout the world and OECD nations. In terms of distribution of premiums, industrialized countries still dominate, encompassing approximately 89 percent of total world insurance premiums in 2004. However, the share for emerging markets has been on the increase from 8.71 percent of total world premiums in 1995 to 11.47 percent in 2004, highlighting the fact that the importance of insurance services is on the increase in these areas as globalization and sweeping deregulations take hold2.
2 Source: Swiss Re.: Sigma World Insurance articles, various issues. Emerging markets comprise of Latin America, Central and Eastern Europe, South and East Asia, Middle East and Central Asia, and Africa.
5 Table 1.1: World Insurance Premiums (1995-2004) This table presents the gross insurance premiums written across the world as well as a separate breakdown for global life and non-life insurance premiums. Year Total Growth Life Growth Non-Life Growth 1995 2147770 -- 1239605 -- 908164 -- 1996 2105838 -1.95% 1196736 -3.46% 909100 0.10% 1997 2128671 1.08% 1231798 2.93% 896873 -1.34% 1998 2155269 1.25% 1264156 2.63% 891112 -0.64% 1999 2324025 7.83% 1412357 11.72% 911668 2.31% 2000 2443673 5.15% 1521253 7.71% 922420 1.18% 2001 2408252 -1.45% 1439177 -5.40% 969074 5.06% 2002 2626898 9.08% 1536122 6.74% 1090775 12.56% 2003 2940670 11.94% 1672514 8.88% 1268157 16.26% 2004 3243906 10.31% 1848688 10.53% 1395218 10.02% Note: Figures are in nominal USD Source: Swiss Reinsurance Company, Sigma World Insurance articles, various issues.
Table 1.2: OECD Insurance Premiums (1995-2004) This table presents the gross insurance premiums written across the group of OECD nation, as well as a separate breakdown for life and non-life insurance premiums written within the OECD. Year Total Growth Life Growth Non-Life Growth 1995 2028537 -- 1182936 -- 845001 -- 1996 1977816 -2.56% 1135562 -4.00% 842255 -0.40% 1997 1985761 0.40% 1160083 2.16% 825678 -1.97% 1998 2016084 1.53% 1190340 2.61% 825774 0.01% 1999 2183167 8.23% 1334444 11.21% 848723 2.78% 2000 2292755 5.02% 1436064 7.62% 856691 0.94% 2001 2240454 -2.28% 1342424 -6.52% 898030 4.83% 2002 2437555 8.80% 1423218 6.02% 1014337 12.95% 2003 2709757 11.17% 1533183 7.73% 1176574 15.99% 2004 2967552 9.51% 1681176 9.65% 1286376 9.33% Notes: a) Figures are in nominal USD. b) Figure for 1995-2001 consisted of 29 member countries, while for 2002-2004 they consisted of 30 member countries. Source: Swiss Reinsurance Company, Sigma World Insurance articles, various issues.
6 1.3 PURPOSE AND SCOPE OF THIS STUDY
This is a study that aims to forge the way in relation to how future studies are conducted in international financial services and can be argued to be unmatched in terms of both its depth and content. A couple of key issues in banking and risk and insurance are examined within the area of international financial services, where these topics are studied for the first time. Specifically, this thesis contributes to existing literature by examining issues relating to the determinants of bancassurance, and the determinants of life insurance with a particular focus on the impact of legal systems. Additionally, methodological improvements are also made over prior studies within these areas.
Chapter 2 provides an analysis of the determinants of bancassurance with the aim to provide a greater understanding to all concerned parties regarding the key demographic, social and macroeconomic determinants of demand to such an extent as has not been accomplished before. This study will be extremely useful to bankers, insurers, policymakers and regulators alike, since it will allow for a better understand of the driving forces behind the continued success of bancassurance markets, and will outline advantages inherent from the union of banking and non-banking operations.
Chapter 3 examines the determinants of life insurance consumption within OECD nations, paying particular heed to the impact that legal systems have on demand. Once again key demographic, social and macroeconomic determinants are examined in order to
7 assist policymakers in their decision-making process. In addition, systems of law are
examined in order to provide policymakers and regulators with a better understanding of
how their legislations and regulations impact on the health and continuing success of the
life insurance industry, and the economy on a whole. Also included are suggestions on how employing aspects from varying systems of law may improve the current economic climate within the nation.
1.4 OBJECTIVES AND VALUE OF THESE CHAPTERS
The two chapters within this thesis focus on issues relating to the demand of international banking and risk and insurance products and services that have hitherto not been given the appropriate level of attention in prior research works. Namely, we identify the key determinants of global bancassurance, and also provide insight into the determinants of life insurance consumption within OECD nations and the impact that systems of law have on demand.
1.4.1 The Determinants of Bancassurance
The bancassurance industry has been growing at an accelerated rate ever since its conception during the 1970s in Europe. The combination of inadequate national welfare initiatives, the changing demands of consumers for more variety in their insurance
8 products, and the benefits that participating companies receive has seen the continual
success of bancassurance markets. The concept has reached every corner of the globe and
in most places is being embraced with open arms, while in other areas a gradual reduction
in regulations is paving the way for a promising market for bancassurers, with the two
prime examples being the U.S. and Japan. With its increasing popularity, knowledge
regarding the determinants of bancassurance is increasingly important for banks, insurers,
policymakers and regulators alike.
This chapter addresses the gap in existing literature relating to the determinants of
demand of bancassurance products and services using current global data. The
importance of this study is based on the need to understand what drives bancassurance
demand within the current climate of intense global growth of bancassurance markets in
both developed and underdeveloped nations. The majority of past studies regarding
bancassurance have focused mainly on the risk and profitability effects resulting from the
union of a banking and non-banking firm, and hence have only examined a few key
variables while mainly overlooking the rest3. In turn, this chapter thus provides several key contributions, including:
The methodology used within this study improves on previous literature, since we
incorporate the use of the Generalized Method of Moments (GMM) estimation
technique in unison with the Ordinary Least Squares (OLS) regressions. GMM is
advantageous in the sense that it allows for consistent and efficient estimates to
3 Bancassurance studies tend to analyze variables that proxy levels of diversification, revenue generation and expense reduction.
9 still be derived even if some of the assumptions required for OLS do not hold.
Using GMM with the White variance-covariance matrix helps adjust for potential
heteroscedasticity while the use of instrumental variables mitigates the problems
posed by multicollinearity, measurement errors and endogeneity, which are
considerably difficult to diagnose and treat separately.
This study provides possibly one of the best reviews of prior descriptive and
quantitative literature in the field of bancassurance and an extensive review of
bancassurance operations in varied regions of the globe in conjunction with the
use of current data and industry trends.
The results of this chapter provide conclusive evidence in regards to the validity
of variables used in prior research on bancassurance4. We also utilize variables
that have proven to be significant in prior studies in the fields of banking and
insurance (proxies for company size, size of the national banking industry, level
of deregulation within the country, changes in national income levels, system of
law, and inflationary levels) as to the most influential determinants of demand in
order to provide a more complete picture regarding the factors that drive the
success of bancassurance operations. The sample incorporates data from both
developed and underdeveloped countries ensuring that a truly global subset of
countries is examined. All the hypothesized variables are statistically significant
4 Benefits in regards to risk diversification as found in Allen and Jagtiani (2000), Boyd and Graham (1986, 1988), Boyd, Graham and Hewitt (1993), Brewers (1989), Estrella (2001), Lown, Osler, Straham and Sufi (2000), Saunders and Walter (1994), and Wall (1987). Variables analyzing the wealth effects of
10 and meet with one of the expected hypotheses5. Specifically, it is found that
reductions in company risk resulting from diversification benefits, the decrease in
costs and increase in revenues following the implementations of bancassurance
operations, consumer demand as proxied by GNI per capita, company size that is
indicative of customer base, size of the national banking industry, the inflationary
environment, and the level of deregulation within the country are all significant
determinants of bancassurance demand. In addition, it is also discovered that the
system of law within a country will be influential on the overall success of the
bancassurance industry since it impacts directly on the determinants of
bancassurance.
1.4.2 Law and the Determinants of Life Insurance in OECD Countries
Life insurance is becoming increasingly important in the financial services industry across the globe with the current world life insurance premiums amounting to $1849USD billion in 2004 (57% of total world insurance premiums). Industrialized countries accounted for over 88% of global life insurance premiums in 2004 with the countries of the OECD amassing $1681USD billion in premiums (56.65% of total insurance premiums within the OECD). Life insurance is important to the community and economy alike, since it provides ways for the populace to alleviate uncertainty brought about by risks inherent in everyday activities and also provides a long-term savings mechanism
bancassurance as found in Carow (2001), Cown, Howell and Power (2002), Cybo-Ottone and Murgia (2000), Fields, Fraser and Kolari (2005), and Stiroh (2004).
11 that can be catered to the individuals’ particular needs, while also facilitating economic
growth by acting as financial intermediaries between investors and economic agents and
assuaging pressure on governments resulting from exhausted and ineffective pension
schemes.
In spite of the obvious importance life insurance offers in the way of risk management,
savings facilitation, and providing term finance, we do not have a clear understanding as
to the factors that drive the demand and supply of life insurance across countries and over
time. A number of previous researchers have proposed a variety of different
macroeconomic, demographic, social and psychographic factors as possible determinants
of life insurance consumption6. Sample sizes, availability of data, and the geographic
areas of study have constrained the testing of theoretical hypotheses and ensured that
results vary amongst the differing studies. In turn, this study provides evidence regarding the key determinants of life insurance consumption within the OECD paying particular heed to the importance of systems of law, in the aim of increasing the awareness of all concerned parties regarding the driving forces behind continuing success of the life insurance industry. More specifically, this study provides the following contributions:
This is the first study that examines the impact of the differing systems of law
within countries on life insurance consumption patterns. This adds to the only
5 For certain variables there is more than one hypotheses as to their influence as a bancassurance determinant based on differing situations. Please refer to Section 2.7.2.2.1 for an example relating to the Risk Proxy. 6 Research of note would include: Beck and Webb (2003), Browne and Kim (1993), Chen, Wong and Lee (2001), Hakansson (1969), Hammond, Houston and Melander (1967), Headen and Lee (1974), Hwang and Gao (2003), Lenten and Rulli (2005), Lewis (1989), Lim and Haberman (2004), Okura and Kasuga (2005), Outreville (1996), Truett and Truett (1990), and Yari (1965).
12 existing studies regarding insurance and legal systems, namely those of Esho,
Kirievsky, Ward and Zurbruegg (2004) and Browne, Chung and Frees (2000),
both of whom examined the relationship between non-life insurance and legal
systems. Here, it is found that only the French and German civil-law systems
express significance, in conjunction with the arguments that greater financial
intermediation promotes increased consumption through facilitating economic
growth; and that low trust societies place a greater emphasis on market-based
means, such as insurance, of dealing with risk and uncertainty.
This study also improves on previous methodology through the incorporation of
the more advanced GMM estimation technique in juxtaposition to the OLS
technique. This helps to account for possible inaccuracies that may arise as a
result of heteroscedasticity, multicollinearity, endogeneity and measurement
errors.
A number of tests are conducted to ensure that the highlighted relationships hold
in alternate cases and to provide any patterns over time7.
This study also offers support regarding the validity of numerous variables that
have been previously examined as possible determinants of life insurance
consumption while also providing insight into the importance of certain key
variables that have hitherto been overlooked, with a major focus on the affects of
7 One method of testing splits the whole sample into two periods of equal length in order to examine changes regarding the relationships between variables that occurred over time.
13 systems of law on the demand for life insurance. Evidence was found supporting the positive relationships of GDP per capita that reflects the levels of economic growth, economic development and disposable income, education which acts as a proxy for the level of risk aversion, the level of minority shareholder protection that assists in measuring the level of legal rights within a nation, and the number of dependents, all of which are in line with the findings of previous literature.
Both the level of foreign direct investment plus the level of government investment restriction proved to be significant, indicating that important benefits can be derived from the resources that flow in from abroad and any restriction of such inflows will be damaging to the industry on a whole. While the results regarding levels of uncertainty within the community, as measured by Hofstede’s
UAI, indicate that societies with considerably high uncertainty avoidance, past a given UAI level, experience a level of anxiety so high that insurance is seen as a passive way of coping with the additional uncertainty and a more aggressive approach of material wealth accumulation and personal savings is deemed more appropriate. Finally, the level of interest rates and the foreign market share of the life insurance industry are both deemed to be negative determinants of consumption.
14 1.5 THESIS STRUCTURE
This thesis explores issues related to the demand and overall success of international insurance operations and extends the existing literature accordingly. Moreover, this thesis provides methodological improvements over previous studies, adding to the validity of the findings herein. The remainder of this thesis is organized as follows:
Chapter 2 provides a global study examining the determinants of bancassurance using data from 1999-2003. A review of general definitions is followed by an in-depth analysis of the banking, insurance and bancassurance markets. Here we review the history of each, outline their respective strengths and weaknesses, and discuss the gains that banks, insurers and consumers obtain from the participation of bancassurers within the financial markets. A global breakdown is also provided through which we review the bancassurance markets within certain key areas around the world, followed by a detailed literature review of prior works in the field of bancassurance. An overview of the variables hypothesized as potential determinants is then provided in unison with their corresponding justifications based on existing literature. Next, we outline the methodology, explaining the model used as well as a comprehensive analysis of the two methods of testing utilized, namely, Ordinary Least Squares (OLS) and Generalized
Method of Moments (GMM). This is then followed by the chapter’s empirical results and a discussion of their implications; and finally we use the chapter conclusion to sum up our findings.
15
Chapter 3 looks at law and the determinants of life insurance demand/consumption within
OECD countries over the period of 1996-2003. Initially we provide a synopsis of the life insurance industry and the differing systems of law that exist. This is followed by a comprehensive literature review. Since the literature is so diverse, the literature review provides a summary of existing works on life insurance demand, systems of law and their effects on financial markets, and the limited studies performed analyzing the impact of legal systems on insurance markets. Next, the hypotheses and their theoretical justifications are presented in relation to the variables that are outlined as key determinants of life insurance consumption. This is followed by the methodological section through which we outline our model and provide a summary of the various methods of testing incorporated within the study. Finally, the empirical findings and their implications are presented before we sum up the chapter with a general conclusion and present areas for further study.
Chapter 4 provides the overall conclusion of the thesis based on the findings of the previous two chapters, in addition to supplying the implications and significance of these results to existing literature. Finally, we make recommendations in regards to possible avenues for further research that will help contribute to exiting literature.
16 CHAPTER 2
THE DETERMINANTS OF BANCASSURANCE
"There are risks and costs to a program of action. But they are far less than the long- range risks and costs of comfortable inaction."
- John F. Kennedy (1917-1963). American Statesman (35th US President)
2.1 INTRODUCTION
The bancassurance phenomenon has been growing at an accelerated rate ever since its conception during the late 1970s and early 1980s in Europe. The combination of inadequate national welfare initiatives (such as lackluster pension schemes), the changing demands of consumers for more variety in their insurance products, and the benefits that participating companies receive has seen the continual success of bancassurance markets.
The concept has rapidly reached most areas of the globe, sweeping through developed and developing countries alike, and in most places is being embraced with open arms, while in other areas a gradual reduction in regulations is paving the way for a promising market for bancassurers, with the two prime examples being the U.S. and Japan. With its increased popularity and the varying levels of success achieved by bancassurers in
17 differing countries, the possible factors influencing the demand and supply of
bancassurance have received a considerable amount of attention in recent decades from researchers and policymakers alike.
A large quantity of previous studies into the blossoming bancassurance markets have only been descriptive in nature, providing broad insight into believed reasons behind the success of this venture and potential benefits and disadvantages for all parties associated.
However only a few researchers such as Allen and Jagtiani (2000), Boyd and Graham
(1988), Boyd, Graham and Hewitt (1993), Carow (2001), Cowan, Howell and Power
(2002), and Estrella (2001) provided any quantitative findings, and most of these studies focused mainly on the potential risk diversification benefits associated with bank expansion into the insurance industry. The lack of quantitative studies can be argued to result from a lack of available data regarding bancassurance operations. Improved disclosure requirements on behalf of companies and improved collection of national statistics now provide the means for increased research into this rapidly growing area of the financial markets, knowledge of which is highly sought after by such parties as businesses, policymakers and regulators.
The determinants of the banking and insurance industries have been examined at length by researchers with recent focus being placed on the possible determinants of bancassurance. However, to date, no quantitative study has examined such a large number of economic, institutional and demographic determinants on the global bancassurance market as our study does.
18 This study provides several significant contributions to existing literature. This chapter
firstly reconciles existing literature on bancassurance with studies done in the area of life
insurance, non-life insurance and banking, and provides a more complete picture with
respect to the main factors that determine the overall success and popularity of
bancassurance operations around the world in both developed and developing countries.
This is important for policymakers, given the monumental growth in bancassurance markets in recent history as well as the continued positive forecast into developing regions such as mainland Asia as well as in countries such as the U.S. This study also makes methodological improvements on previous studies in the general field of insurance
(including bancassurance) by using the advanced estimation technique know as GMM8 that enables us to exploit the time series nature of the data and overcome potential data limitations such as measurement errors, heteroscedasticity, endogeneity and multicollinearity.
The results obtained from this study indicate that the reduction in company risk, size of the company, the reductions in company costs and increases in company revenues, size of the national banking industry, level of financial deregulation within a country, the level of national demand for insurance products, and the national inflation rate all play significant roles as determinants of bancassurance. These variables all display the correct hypothesized relationships in conjunction with existing literature9. We also find that the
system of law that exists within a country is also a contributing factor to the overall
success of bancassurance operations.
8 As proposed by Arellano and Bond (1991), Arellano and Bover (1995), Blundell and Bond (1998), Hansen (1982), and Kitazawa (2001).
19 The results of this study are important to policymakers and regulators in particular since it indicates that they should focus their attentions on further financial deregulation to encourage greater market accessibility since it is found that targeting this area has potential benefits to the bancassurance sector. For bank managers, the study indicates that they should strongly consider expansion into bancassurance operations due to the benefits provided through diversification of risk, reduction in company fixed costs, increase in fee-based revenues, and improvement in overall company competitiveness.
2.1.1 Chapter Outline
This chapter is structured as follows: Section 2.2 provides commonly accepted definitions of the main terms used within this chapter. Sections 2.3 and 2.4 provide an outline of the banking and insurance industries, noting their respective vulnerabilities and strengths.
This is important to provide a full understanding of why both parties would wish to partake in a bancassurance operation. Section 2.5 provides an outline of bancassurance to date. This section also reviews the past descriptive literature and quantitative research of researchers. Section 2.6 outlines the data and methodology used within the study. This includes looking at empirical hypotheses that are being tested, outlining model specification tests that are being used, reviewing the actual model employed in the study, and discussing the testing procedures utilized. Section 2.7 provides the empirical results for the differing test samples and also discusses the implications of said results. Finally,
Section 2.8 concludes with the main findings and resulting policy implications of this study, while also providing suggestions for further research into this field.
9 Results vary slightly depending on the samples being tested. This is further clarified in Section 2.7.
20 2.2 GENERAL DEFINITIONS10
Banks
“The essential function of a bank is to provide services related to the storing of
value and the extending of credit. The evolution of banking dates back to the earliest
writing, and continues in the present where a bank is a financial institution that provides
banking and other financial services. Currently, the term bank is generally understood to
be an institution that holds a banking license. Banking licenses are granted by financial
supervision authorities and provide rights to conduct the most fundamental banking
services such as accepting deposits and making loans. There are also financial institutions that provide certain banking services without meeting the legal definition of a bank, a so-
called non-bank. Banks are a subset of the financial services industry.”
Insurers
“Insurance is the business of providing protection against financial aspects of risk,
such as those to property, life, health and legal liability. It is one method of the overall
concept known as risk management.”
“Insurance refers to coverage by contract whereby one party undertakes to
indemnify or guarantee another against loss by a specified contingency or peril.”
10 Definitions for ‘Bank’, ‘Insurer’, ‘Bancassurance’ and ‘Assurbanking’ were obtained from Wikipedia- The Free Encyclopedia website.
21 Bancassurance
“Bancassurance is a banking and insurance structure in which insurance is sold
through the bank or the banks distribution channels.”
Universal Bank
“Universal Banks are financial institutions that may offer the entire range of
financial services. They may sell insurance, underwrite securities, and carry out securities
transactions on behalf of others.”11
Assurbanking
“Assurbanking is an insurance model whereby insurance companies sell banking products through their distribution channels.”
2.3 BANKS
2.3.1 Introduction
This section and the one that follows (Section 2.4) aims to provide a clearer
understanding about why both banks and insurers would be open to an idea of uniting
their strengths and working together under the banner of bancassurers. These two
sections will also give us a clearer insight into the determinants of bancassurance on
22 behalf of both banks and insurers by allowing us to understand both parties’ current
vulnerabilities and strengths. The subsequent advantages that banks and insurers receive
from participating in bancassurance operations are then highlighted in Section 2.5.
The current global climate of increased deregulation and globalization is ensuring the future of the banking industry. No longer are banks relegated to specified duties such as providing credit; now they are more involved with all facets of the market. In the latter
half of the 20th century, there has been a distinct move towards Universal Banks, and
even though not all of these are successful enterprises, it seems clear that this is where the
future for banks must lie if they are to overcome the many modern challenges that they
face.
2.3.2 Current Vulnerabilities12
A common question arises when we look at the banking industry today, which is: “Are
banks necessary for banking?” With the large amount of alternate suppliers of banking
products, such as insurance companies and even supermarkets, the answer would seem to
be NO. With this being said, the question must also be asked as to whether “banking is
necessary for banks?” Once again the answer would seem to be NO as a result of
decreased regulations allowing banks to expand into providing alternate services such as
securities and insurance products, in turn allowing banks to obtain an income source that is fundamentally different from that which they currently receive.
11 Benston George J. (1994) pg 121. 12 This section borrows heavily from Llewellyn (1999)
23 This section aims to show what the current vulnerabilities of banks are on a global scale.
We believe that it is important to understand what the vulnerabilities of banks are within
the global economy in order to understand why banks decide on starting bancassurance
operations. It should also be noted that each of the following vulnerabilities have a potentially positive impact as well.
Competition:
Banks have always faced competitive pressure, especially from amongst their own
industry participants. Recently however the pressure has also started coming from
competitors outside of the banking industry due to three main reasons. Firstly, entry
barriers into the banking industry have declined considerably allowing smaller firms to
gain access to markets they previously could not. Secondly, as a result of a wave of
deregulation that has swept the western world (and is following suit amongst less
developed nations), the regulatory protection that banks once were afforded is no longer
so strong. Finally, globalization means that banks are no longer simply competing against
national counterparts, but are also being forced to compete with international firms that
have alternate ways of achieving similar goals. These pressures have eroded the
monopoly power that banks have held for so long in the provision of banking services.
However, an advantage of increased competition is that it forces incumbent players out of
the market and may in turn leave well-established banks with a greater market share of
the industry. Furthermore, if the new ideas and products from new competitors can be
effectively harnessed and used, then existing banks can expand past their current size.
24 Contestability:
Resulting from the decreased entry barriers, the banking markets have become more
contestable. The Contestable Market Theory states that the removal of entry barriers can
reduce the long-run profits of existing firms in the industry (Carow (March 2001)).
Globalization:
In current times competition amongst banks has become truly global. The combination of
decreased entry barriers and regulations allows banks to expand offshore with limited restrictions being faced. Furthermore, the consumer in current times has a more global outlook. Banks must compete more aggressively to maintain their customer base since consumers have the option of gaining their banking needs from some offshore bank that offers better terms and conditions.
At the same time, this allows banks to easily expand overseas where their type of banking may be more favorable than the existing form. It also allows banks to promote their products to overseas consumers in order to expand their offshore customer base.
Entry and Exit Barriers:
Decreasing entry barriers allows smaller banks as well as nonbank companies to gain easier access to the banking market, something that did not exist in the past. Exit barriers are also at all-time lows, further facilitating competition within the industry.
25 Supply:
The way that competitors supply products means that banks have to face different
competitive pressures for each product they offer. In the current era, new entrants into the
banking market are no longer offering all the banking products that a bank generally
offers. Instead, new entrants may break up possible products and offer only those in
which they feel they will be competitive in. Hence, for each different type of product, a
bank will face different competitive pressures.
Regulations:
Regulations have been steadily declining over the past few decades, no longer affording
the same protection and economic rents that they once did for banks. The current
deregulation allows firms from different industries to gain access to the banking markets
and it also fosters globalization. Regulation, however, has its own benefits since it allows
banks to more easily move into other areas of business (namely bancassurance) and
expand abroad.
Technology:
This is turning into one of the most serious threats that banks currently face to their
‘traditional’ banking structure. Technology is changing the core business components of
banks by altering both the production and distribution side simultaneously. The main
problems caused are the decreased need for the traditional branch network, decrease in transaction costs that can be charged to the consumer, and the increased competition
26 provided by other firms that are making the best possible use of new technologies such as the Internet.
On the other hand, technology could prove to be extremely beneficial to banks should it be employed effectively. It could increase efficiency, decrease costs, and allow banks to reach a broader range of consumers both domestically and abroad.
Consumer Trends:
Consumer expectations and demand for products and services is changing continuously.
Consumers demand better quality products at lower costs as well as more efficient delivery services that minimize the time it takes to obtain a product or service. Greater choice in products and more information regarding the products are also being demanded by a more sophisticated consumer. Consumers are no longer afraid to go to alternate sources to fulfill their banking needs, whether it is firms from a different industry or offshore banks.
Costs:
Banks are facing increased pressure to decrease their costs due to a combination of decreased entry barriers, increased competition, and more cost-effective technologies existing in the current marketplace.
27 Capital Markets:
Capital markets are increasing pressure on banks to be more disciplined and efficient by demanding such things as increased disclosure in their financial statements.
2.3.3 Advantages of Banks over Insurance Providers
Banks have a number of advantages over insurers that makes a possible union of the two so attractive, especially to the insurers.
Banks typically have a larger customer base than insurers since they offer more products and services. They provide a range of everyday services that the customer requires and looks favorably upon, unlike insurers whose products and services are based on adverse incidents occurring and hence are looked upon unfavorably by consumers. The potential market for banks is also far larger since certain banking services are seen by consumers as a necessity, more so than insurance products which are considered more of a luxury item.
The strong brand name of the banks is also an advantage that banks have (Holsboer
(1999)). Banks find it easier to promote new products to their customers by leveraging their already strong brand name. This is something that insurers lack, finding it much more difficult to bring new products or changes into effect.
28 Banks also prove to be more cost effective in the way that they provide products. This is
important to insurers since a substantial amount is spent on each policy in terms of
administration fees, mailing of policies, medical exams for potential clients, and so forth.
The distributional infrastructure of banks allows for products to be sold more cost-
effectively, regardless of what the product may be (Griffin (1996)).
The vast amount of information on their clients that banks are privileged to is advantageous in the sense that it allows for client segmentation. Banks can promote products and services that are intended for a particular segment of their customer base directly to that segment, avoiding costs involved with marketing to the entire customer base. This ensures that a substantial proportion of those consumers marketed will accept the offer.
2.4 INSURERS
2.4.1 Introduction
In modern times, insurance is much more prevalent in the economy as a result of changing regulations and an increased risk aversion of the general populace. Employers in many countries are now required to insure their employees against a variety of mishaps. In many cases, they offer group insurance policies that provide a number of
29 insurance products such as life and health. Currently, the growth in insurance far exceeds that of economic development globally.
2.4.2 Current Vulnerabilities
Similar to banks, the question has to be posed as to whether “insurers are necessary for insurance?” The answer to this is a bit more ambiguous than for banks. The answer is once again NO, but only in the long term. In the short term, other possible providers, such as banks, have to learn the basics of the provision of insurance products and services from insurers, including what constitutes a successful insurance product, how to calculate the risks, and so forth. The answer to the other question “is insurance necessary for insurers?” would as well have to be NO. In fact, if regulations did not come into play and both banks and insurers were on an equal footing, then it would undoubtedly be easier for insurers to offer banking products than for banks to offer insurance products. Regardless of this, the current trend in the financial services industry has resulted in a lot of problems for insurers. Below we discuss the main vulnerabilities that insurers currently face in the marketplace.
Balance Sheet Pressures:
As a result of numerous global financial crises, insurers have faced severe pressures on
their balance sheets and are in turn seeking ways to decrease their operating cost ratios.
Traditional agents are proving to be cost-inefficient in terms of the distribution of
products.
30 Globalization and Deregulation:
Globalization coupled with deregulation is causing insurers considerable concern,
prompting increased foreign competition and decreasing market share for existing
insurers. This is especially true for developing nations where the competitive advantage of incumbent players in the market is being nullified. In more developed countries such
as the U.S. and U.K., the deregulations are having similar results of increasing
competition by reducing the powers of the once ‘almighty’ insurance lobbies.
Deregulations and increased global competitions will also mean a squeeze on margins
due to premium erosion, and will make the correct pricing of products increasingly
important. The tax advantages that certain insurance products received are now
disappearing, reducing their attractiveness to the consumer. New types of products will also flood the market that better cater to the consumers needs, pushing those players not innovative enough to the sidelines.
This being said, the benefits of deregulations are also high, since it allows insurers to join with firms from other industries and offer products hitherto restricted. An opportunity cost thus exists between the loss in traditional insurance profits and the gain in profits from new products and other operations such as banking.
Consumer Sophistication:
Consumers in modern times are better educated about what they can ask for from their financial service provider. Growing individualization of the consumer is meaning that
31 products have to be developed in order to suit the specific needs of each individual consumer as opposed to traditional products that were constructed for the general populace.
Technology:
Just as with banks, new technologies are altering the production and distribution of insurance products. This means that certain players in the market can offer products at a faster rate and a lower cost than traditional insurance companies.
Capital Markets:
More stringent solvency standards are putting pressure on insurers to strengthen their financial statements. In the U.S. for example, this was brought about through the adoption of the risk-based capital system in 1993 (Cummins, Tennyson and Weiss
(1999)).
2.4.3 Advantages of Insurance Providers over Banks
There are a number of reasons why insurers seem attractive enough to banks to foster a union between the two. The first and most obvious factor is that insurers provide a fee- based income that banks have hitherto lacked. Furthermore, some of the products offered by insurers, namely life insurance products, are very closely related in structure to traditional banking products. This means that their integration into the existing banking system will be considerably easy. Other non-life products can also be easily offered in
32 unison with banking products since they complement one another. One example of such would be taking out a loan for a house or motor vehicle and at the same time taking out insurance on that investment. Another advantage relates to the actual insurance products and the fact that in certain countries these products may have a tax-advantage associated with them, which will foster increased sales to consumers. Finally, banks are also attracted by the fact that the insurance markets have been experiencing exceptional growth for some time and continue to offer attractive growth potential.
2.5 BANCASSURANCE
This section aims to provide a thorough analysis of the history of the bancassurance industry by detailing the reasons behind the swift shift towards bancassurance, encompassing the keys to success of any bancassurance operation, reviewing the gains to the differing parties involved in bancassurance, as well as the challenges that bancassurers face. Here we also provide a rundown of how differing areas in the world are responding to the growing success of bancassurance. This section is heavily influenced by the descriptive work of past bancassurance researchers and is in part a literature review of their work. Later in the section, we will also provide a brief literature review of quantitative works of researchers. This section aims at reviewing previous literature regarding bancassurance in order to give us a better understanding of what the determinants of bancassurance may be.
33 2.5.1 Brief History
Bancassurance is a relatively new concept on the global stage. Unlike banks and insurers,
which have been around in one form or another for centuries, bancassurance has only
been around for a few decades. Bancassurance has achieved its greatest success in the
European markets where it first originated, currently holding a 32% share of the
European life and pension market. In certain countries such as Portugal, Spain and
France, bancassurers possess the lion’s share of the life insurance business, as is shown in the table below, and have recently begun to make significant inroads into non-life industry as well. Its prominence in certain areas around the world, or lack thereof, is due largely to the challenges that still face bancassurers, which is something that is going to be discussed later in Section 2.5.4.
As with so many other concepts, it is argued over by many as to which country developed the idea of bancassurance. The general consensus amongst researchers is that bancassurance first reared its head in France in the late 1970’s, motivated by, amongst other things, changing consumer needs due to an inadequate pension scheme that existed at the time (Bonnet and Arnal (2000)).
A more in-depth definition of bancassurance than was given previously in Section 2.2 is that it involves the manufacture and/or distribution of insurance products by banks.
Distribution in this sense can mean either directly through the branches of the bank or
where the bank acts as an agent for the insurer and promotes the insurers products to its
34 clients even though the bank itself does not directly provide the product. There are a
number of ways that bancassurance can arise. The weakest form of union would involve
a distributional agreement through which the bank would distribute stand-alone insurance
products with little to no sharing of the customer base occurring13. A higher form of
integration would be a strategic alliance, which would see a greater sharing of the banks
customer base and also greater investments required on behalf of both parties. From here
Figure 2.1
Bancassurance Share of Life Insurance Business
100% 90% 81% 80% 77% 70% 63% 61% 60% 56% 50% 40%
30% 24% 20% 15% 10% 0% Portugal Spain Belgium France Italy Germany United Kingdom
Source: Swiss Re. various publications, and Datamonitor various publications.
we move to the parent bank establishing an insurance subsidiary as is common in the
U.K. according to Skipper (2000). This model is argued by Holsboer (1999) to be the most successful, especially if built from scratch since cultural barriers can be avoided. A
13 Swiss Re. (2002).
35 joint venture model on the other hand sees a long-term commitment from both the bank
and the insurer and is less prominent. Finally, full integration can occur where all
financial services are produced and distributed by a single corporation and all activities
supported by a single capital base. Skipper (2000) argues that this ‘one-stop-shop’ only exists in theory at present, since no financial institution is structured in this way legally.
2.5.2 Why Does Bancassurance Occur?
As has been mentioned previously, the lackluster pension scheme was a driving force
behind the growth of bancassurance. Not only in France, but also in the majority of
developed countries, the population growth rates have slowed significantly14, affecting
the age structure of the population (Klein (2001)). People have been found to be living
longer than in previous decades and this has become an increasing burden on the outdated
pension schemes within certain countries. Governments can no longer maintain the
funding that they were once able to as a result of added pressure from the workforce to
decrease taxes. As a result of this, people have begun to take a more active role in their
future entitlements by looking at alternatives to pensions. Bancassurance not only
provides an alternative to pensions but also caters to the current taste of consumers,
which is no longer satisfied by the traditional products offered by insurers. A study by the
Morgan Consulting Group in 2003 supports this argument by saying that ‘structural
changes and evolving customer attitudes and needs are causing traditional [insurance]
distribution channels to lose market share’. Though this was unquestionably a
predominant reason behind the increased popularity of bancassurance, it was not the only
36 one. Apart from pensions, governments have also been making severe cutbacks in
healthcare, disability and sickness benefits, paving the way for those companies that also
offer non-life insurance products, such as bancassurers.
From the banks perspective, the strong economic growth of recent years in an
environment of low interest rates, inflation rates, and consumer confidence meant that
banks were hit with substantial trading losses due to poor consumer savings and squeezed
margins (Holsboer (1999)). To alleviate this problem, one possible solution was
expansion into the insurance industry through bancassurance, which opened a new form
of fee-based income, and in turn allowed banks to move away from incomes generated by
interest spreads. From the insurers perspective, the extensive customer base and
distributional network that banks have are an attractive reason to partake in bancassurance operations. Improvements in technology that have forced increased competition through lowering of entry and exit barriers are also playing a fundamental role in the growth of bancassurance.
Another leading factor is the cost savings benefits that arise as a result of bancassurance operations. One particular study found that European bancassurers have eliminated 30%-
50% of the typical insurers distributional and administration costs (The Boston
Consulting Group (BCG) (1999)). Another study states that cost savings as a result of
bancassurance operations was 21.2% for the study group, while expected growth in
revenues was 4.4% (Swiss Re. (2002))15. Furthermore, bancassurers generally posses a
14 Another key example would be Australia which is currently faced with an ageing population. 15 Generally takes 3 years to realize these gains.
37 competitive advantage over their competition since they are able to carry out intra-group transactions (at times without correct risk evaluations), as well as to optimize requirements of solvency margin according to each type of financial operation within the group (Bonnet and Arnal (2000)), and are generally more efficient16.
The bancassurers competitive advantage within Europe became apparent during the transition to the common Euro currency. This period posed a major challenge to both bancassurers as well as insurers alike. But according to Tassin (2002) the bancassurers came out with a competitive advantage, thanks in part to their more modern and less costly software that is backed up by their banking systems. In areas where government regulations have thus far restricted the growth of bancassurance operations, such as mainland Asia, researchers argue that newcomers to the market are keen to start bancassurance operations in order to nullify the competitive advantage of the incumbent players that retain large agency networks (Swiss Re. (2002)).
Bancassurance operations can also provide tangible synergies and economies of scale and scope (Berberich (2000)) such as offering customers a one stop shopping experience, and the utilization of intangible assets that can be accessed simultaneously in more than one area without their value being diminished. Boot (2003) sees four possible sources of scale and scope economies that point to potential revenue (output) and cost (input) synergies.
These include:
16 Expense ratios (measuring efficiency) in bancassurance activities are very low (Sakr (2001)).
38 → Information technology related economies that result in revenue synergies
through increasing the cross-selling potential of banks. For example, life
insurance and home contents insurance products sold in unison with a mortgage.
A couple of cost synergies also result, which include a more efficient utilization
of databases; and improvements in the distributional network resulting in scale
economies (larger distributional networks can be managed that in turn facilitate
the management of a larger organization) and scope economies (greater array of
products can be offered through the same distributional network).
→ Reputation and marketing/brand name related benefits. By associating their name
and hence reputation with new products (such as insurance products) banks can
fix the costs of marketing and branding. At the same time, they can obtain
revenue synergies from the associated cross-selling potential and the acceptance
of new distributional channels such as the Internet.
→ Financial innovation related economies. Larger companies could make use of
their wider scope and larger scale in order to better recoup the fixed costs of
financial innovations while still maintaining a competitive advantage17. Revenue
synergies are brought about through superior innovations that can be offered as a
result of a broader information set, since banks and insurers will be able to
leverage each other’s product skills. For example, insurers may be able to benefit
17 Competitive advantages associated with financial innovations generally do not last past the short term. This is a result of competitors rapidly imitating those operations that prove successful in another institution.
39 from the derivative-innovations and securitization skills that come from the bank.
Finally, there will also be a better rent extraction due to a bigger network.
→ Benefits from diversification. This could facilitate an internal capital market
where cash flow generating businesses could help fund other activities. Benefits
regarding the proliferation of off-balance sheet banking can also be made since
these activities will impact on the bank’s credit rating. Diversification may lead
to an improved rating for the organization. Other benefits would include
improved asset management and more effective use of equity capital. The
mismatch between banks assets (long term) and liabilities (short term) might be
the reverse from that of an insurer (long term obligations).
Bancassurance is also an effective way of preventing job loss and in most cases actually creating employment opportunities (Benoist (2002)).
Let us now look at the reasons why banks and insurers would agree to unite in the first place before we examine the overall benefits of bancassurance.
All researchers believe that, as with any business, a successful bancassurance operation requires certain guidelines to be followed in order to ensure its success. From the start, a substantial up-front investment is generally required in terms of time and resources; subsequent investments will also be needed. Successful integration of the insurance operation with the existing bank operation is fundamental. The costs put into the
40 integration phase have to be able to be recouped. Current banking staff will need
additional training in how to sell insurance products, the new operation will need space in
the existing branch networks, and new products need to be manufactured and marketed.
These expenses will likely be extremely high at the outset due to the differing nature of
the insurance operation from those which the bank is used to dealing with. In order for
the integration to be successful, there should be a strong management structure employed
as well as a clear definition of company goals18. This should also assist in the resolution
of differences that may arise. Researchers such as Morgan (2003) argue that the more
heavily integrated the bancassurance operation is into the branch banking activity, the
lower the marginal costs of transacting bancassurance will be.
One of the major hurdles to the success of a bancassurance operation is the cultural
differences that exist between the insurance company and the bank. This specifically
relates to the employees that have been used to different working environments and
procedures and how they manage to work as a team within the new organization.
Differences that may cause conflict include insurance company employees receiving
commissions based on their sales, whereas banking staff do not, and how each type of
employee would go about their business. One way to overcome such differences is to
ensure that there is an effective remuneration package in place that rewards performance,
while another is to make use of external recruitments19 (Hubbard (Spring 2001)).
18 The relevance of this varies with the bancassurance structure employed. 19 External recruitment may cause more problems than it solves, since existing bank and insurance staff will not look favorably upon having their jobs filled by people from outside of the business, possibly at the cost of some of their colleagues.
41 Virtually all researchers state that a successful bancassurance operation requires a standardized and simple group of products to be offered to the consumer that possess low administration and acquisition costs (Bonnet and Arnal (2000), and BCG (1999)). Placing an insurance product as a rider to an already existing banking product will mean that the sale will be easier and faster (Heistermann (2000)). By making the products simple and standardized, the consumer can quickly decide on whether they want to purchase it and the salespersons can close each deal relatively quickly20. The pricing of the product is very important as well, since differences have to be allowed for in the calculation of premiums such as expense loading and more liberal underwriting. Wade (Fall 2000) advises that traditional products that can be offered by other insurers should be avoided by bancassurers, since they would be facing a competitive disadvantage due to their poor brand recognition in the insurance services industry. The most successful products are those that are sensitive to the differing characteristics between insurance and banking products and meet the majority of the consumers’ needs (Griffin (1996)). The products should be such that they are easy to understand from both the consumers and salespersons perspective, while at the same time meeting the basic needs of the consumer. It is important to keep things simple for the banking salesperson since the distribution of insurance products requires a more active sales approach than they would necessarily be used to with normal banking products. The sales staff should also be coached to understand that it is more important to focus on ‘value’ rather that ‘volume’ (Hubbard
(Spring 2001)). In addition to an effective sales support system, proper use of the banks varying distributional channels is critical (i.e. Internet) (Nigh (2000)).
20 This is important, since the commissions received from each transaction are relatively low (Hubbard (Spring 2001)).
42
In order for a bancassurer to make the most out of its advantages, it must segment the customers and market them accordingly. For example, banks can place more emphasis on targeting the young and uninsured consumers that currently have no brand loyalty to a particular insurer, as a middle aged person might have. Simple scripted questionnaires as well as the use of computers for needs analysis would also be beneficial in targeting the appropriate consumer (Nigh and Saunders (2003)). Administration costs and time could also be further reduced through the use of computers and technological advancements
(Bonnet and Arnal (2000)).
2.5.2.1 Gains for Banks
Previously, we have outlined the main vulnerabilities that banks currently face and how insurance companies may have an advantage over them. Here we will give a summarized version of how these factors influenced banks to form alliances with insurers, and then outline the main advantages offered to banks by utilizing the growing popularity of bancassurance.
Even though bancassurance was first founded in the 1970’s, it did not become mainstream in France until 1984 following a widespread deregulation of the financial markets within the country. For banks, the changing regulations meant more flexibility in relation to the types of products they could offer to the consumer (Bonnet and Arnal
(2000)). Banks could more freely expand their product lines into insurance, investments
43 and securities products, which they were hitherto limited in providing. This regulation change came at an opportune time, since banks were facing mounting pressure on their profit margins with each passing year. Profit margins came under attack as a result of the increased competition from both domestic and global financial service providers and also from the demand of consumers to cut costs and provide the same products and services at lower prices. Cost reductions were also stimulated by technological progress, which was accelerating at an exponential rate. This made the already more sophisticated consumer less complacent with traditional banking products, and encouraged them to search for cheaper and quicker ways to fulfill their banking requirements (i.e. making use of the
Internet). Another immediate result was a decrease in the interest rate margins that banks could impose on the consumer (Swiss Re. (2001b)). The market-wide deregulations further exacerbated the problem of shrinking interest rate margins by further decreasing costs (Swiss Re. (2002)).
In response to these pressures, banks could now de-emphasize the importance they placed on earning from interest spreads and rely more on fee-based income to supplement their core earnings, which was becoming available through sources such as bancassurance.
This type of income source was considerably more stable than the traditional income from interest spreads that was susceptible to dramatic fluctuations from one year to the next (Smith, Staikouras and Wood (2003)).
Another direct problem of the deregulations was the competition it promoted from smaller firms that were hitherto squeezed out of the market due to excessive entry and
44 exit fees (Swiss Re. (2001b)). Newer players could now enter the market and provide lower costs and more innovative products that were in line with the consumers changing tastes. For example, Assurebanking was created, which involved insurers making and/or distributing banking products without having to form any sort of relationship with a bank.
Even though this was not as successful (nor still is) in competing with banks, simply due to the small customer base and poor brand name recognition that insurers possessed, it still posed a serious threat to banks.
For all the above-mentioned reasons, and those of Sections 2.3 and 2.4, researchers argue that one of the main reasons (if not the main reason) that banks entered the bancassurance market was as a defensive strategy to avoid market share erosion by, amongst others, insurance companies (Klein (2001)).
Defensive strategy or not, bancassurance provides numerous benefits for the bank.
Below, we provide the main advantages that banks are offered from having bancassurance operation:
→ A partnership with an insurance company offers the bank attractive growth
opportunities both on a domestic and global front (BCG (1999)). If the bank
forms ties with foreign insurers, then it could reach customers that were difficult
if not impossible to reach. Unions with insurers allow banks to create economies
of scale, enabling them to gain a more dominant market share, not only of the
banking market, but also of the insurance market. Depending on the size of the
45 union, this could lead to the new operation being ‘Too Big To Fail’21. This also
aids in controlling competitive pressures from other market participants.
→ Economies of scale in technology are also a major benefit of bancassurance. New
technology is typically very expensive to purchase and install, and would only
prove feasible should the scale of operations be large enough. Insurers may be
able to provide banks with some of this new technology.
→ Banks earn another source of income, namely fee-based, which is more stable
than their current form of income (Smith, Staikouras and Wood (2003)).
→ Banks can leverage their strong brand name, positive image and strong presence
within the country to offer insurance products to consumers who would otherwise
not accept it from an insurance company (Tassin (2002)).
→ Banks gain an opportunity to retain their customers as well as attracting new
customers by offering an array of financial products (Berberich (2000)). No
longer are banks constrained to providing mainly banking products while only
being allowed to ‘dip-their-fingers’ into other areas. Now they can diversify their
product base and at the same time position themselves as a ‘one-stop’ purchasing
store where consumers can find a myriad of products to fulfill all their needs
(Nigh (2000)).
21 Term used when it is believed that a company has reached a given size large enough to ensure its continued survival.
46
→ Banks can gain possible economies of scope. They can transform insurance
products and make use of their well established distribution channels to provide
the insurance products at lower costs than insurers could. This will mean that
banks will not need to spend excessive amounts in order to achieve respectable
revenues. Furthermore, bancassurance reduces the effect of the bank’s fixed costs
(Klein (2001)) and makes the general branch network more cost effective and
efficient (Agrawal A. (Aug (2002))).
→ Banking and life insurance products are complementary in nature, both being
geared towards asset accumulation and the management of financial institutions.
Hence, banks do not have to alter their existing practices greatly in order to be
able to offer life insurance products22 (Swiss Re. (2002)).
→ Banks can now gain access to funds and other benefits that are otherwise kept
with the insurance industry (Swiss Re. (2002)). For example, the tax benefits that
certain insurance products receive may be beneficial to banks in allowing them to
more easily promote these products and in turn gain an easy source of revenue.
→ Since there is a lack of synchronization between banking and insurance company
profit cycles, banks can decrease their volatility of return on equity by forming a
22 General insurance products are another matter and require more effort in terms of producing and marketing.
47 bancassurance operation. Banks could also reduce their risk-based capital
requirement for the same level of revenue. (Swiss Re. (2002)).
→ Depending on how the bank begins its bancassurance operation, diversification
benefits may be close to immediate. For example, should a bank acquire an
insurer, then there would be fewer learning costs involved, the purchased entity is
already successful, and economies of scale are immediately attained.
→ By establishing a bancassurance operation, banks are taking the first step towards
the provision of fully integrated financial services and becoming Universal in
nature. They can also tailor products to the life cycle of the customer ensuring
that products reach only those segments of the consumers that will purchase them
and in turn reducing costs. (Swiss Re. (2002)).
2.5.2.2 Gains for Insurers
It seems curious that insurers would not only allow, but participate in bancassurance,
since the process apparently takes business away from the insurer and transfers it to the
bank. With the ever-changing regulations and advancement of globalizations, insurers have little choice but to go along for the ride. Warsow (1986) argues that regulations reduce the supply side of the insurance market, raising the price at which international insurance companies are willing and able to do business and also increasing the costs to the local insurance industry. As the supply and quality of services deteriorates, the
48 demand for insurance may also be detrimentally affected. Furthermore, due to outdated products, lack of trust and excessive costs, insurers have been facing market share erosion for many years now as a result of consumers seeking other, more modern, insurance suppliers. Bancassurers, for example, have become the leading suppliers of life insurers products in a number of European countries such as Spain and Portugal, with this trend set to continue as bancassurance becomes more prominent on a global scale.
Bancassurance can be viewed as a defensive strategy for insurers, just like banks.
However, by going along with the growing ‘craze’, insurers have numerous benefits to reap as well. Below, we provide the main benefits insurers obtain by participating in a bancassurance venture with a bank.
→ It is argued by some researchers (BCG (1999)) that insurers tend to focus the
majority of their efforts on one segment of the consumer (upper-class) instead of
diversifying everywhere23. This has resulted in insurers gaining poor brand
recognition amongst those segments they have hitherto ignored. It may also mean
that these segments of the consumer have developed a negative attitude towards
insurers. Through a joint bancassurance venture with a respectable bank, insurers
can reach clients that were previously unattainable through making use of the
banks strong brand name as well as extensive information on its clientele. Insurers
could align themselves with the strong public image of the bank, which tends to
be far more favorable than that of the insurers (Nigh (2000)). This results in
23 Meaning that this segment obtains preferential treatment in regards to insurance products whether it be through lower costs, better customer service, or a wider range of products.
49 insurers being able to drastically increase their prospects of future sales and
profits, since they now have access to a much larger consumer segment, namely
the middle-class consumer.
→ Bancassurance allows insurers to better segment clients and access high net-worth
sectors (Swiss Re. (2001)) in order to obtain optimal profits. Furthermore,
insurers would also be able to establish a relationship sooner in the consumers’
life than they previously could, allowing them to build trust (Nigh (2000)).
→ Bancassurance means that companies from two separate industries are being
unified, allowing for a sharing of ideas. Innovative products will thus be born,
which would not be available if the companies were operating separately. These
products could be specifically tailored to a particular consumer segment, allowing
for lower costs and greater profits (Klein (2001)).
→ The reliance on traditional agents could be decreased. These agents are generally
high in cost. Insurers can now make use of the banks extensive distributional
network to promote their products (Swiss Re. (2002)).
→ Through a union with a reputable bank, insurers would be able to establish a
market presence more rapidly without the need to build up a network of expensive
agents (Swiss Re. (2002)).
50 → Insurers can also make use of the banks capital in order to expand the business
(Swiss Re. (2002)).
→ Insurers can obtain business at a lower acquisition cost than they normally could.
A bancassurance venture will also help the insurer cope with competitive
pressures better since they have an established bank behind them24 (it also helps
the bank in a similar way).
2.5.2.3 Gains for Consumers
Consumers stand to gain a lot in the current marketplace. With the extensive globalization and deregulations fostering competition within the financial services industries, the days of one or two companies holding a stranglehold on the market are virtually over in most modernized countries. This is all to the benefit of the consumer.
Lower costs will be passed on in the form of reduced premiums on insurance products, especially if it is offered by a bancassurer. There will be a greater variety of products to choose from and these products will be better engineered to suit the needs of the consumer. Consumers will also be offered increased satisfaction resulting from the ability to do all their financial needs shopping in one area, namely a bancassurer.
24 The bank does not necessarily have to be established. Even smaller banks will help in coping with competitive pressures but not to the same extent as an established bank.
51 2.5.3 Global Breakdown
Considering how this is a global study on bancassurance and that not every nation is at
the same level of development in regards to this industry, we believed that it was
important to outline the impact that bancassurance has had on differing regions around
the world, as well as looking at the major regulations that impact the further growth of
bancassurance. If an in-depth and current analysis of bancassurance markets is required, then numerous companies such as Milliman Consultants and Actuaries and individuals such as Susan Drury have begun to publish comprehensive articles that examine bancassurance markets in different regions of the world25. Below, we provide you with a
brief synopsis of bancassurance markets in certain key areas.
EUROPE:
As has been mentioned earlier, bancassurance is a construct of Europe (France in
particular) and this perhaps helps explain why it is such a phenomenal success within
certain European markets. The large influx of banks into insurance within Europe in recent years was motivated largely by the 1989 Second Banking Coordination
Directive26. Currently, the penetration levels are fairly stable in Europe, since
bancassurance in the majority of Western European countries (France, Netherlands,
25 Morgan Consulting, an arm of Milliman Global, also provide such publications. Latest publications include European Bancassurance Review 2004 (Morgan Consulting), European Bancassurance Review 2005 (Milliman Consulting), and Bancassurance in the 21st Century (Susan Drury). 26 This came into effect in January 1993. The Directive allowed financial institutions in the European Union countries to operate in member countries without obtaining a license from the regulatory authorities within the guest country.
52 Portugal and Spain) has reached what studies such as Swiss Re. (2002) argue to be maturity. These penetration levels will only pick up once bancassurance manages to fully infiltrate Central and Eastern European countries such as Hungary and Poland, and the
Baltic nations. Currently, the final major hurdle for bancassurance in Western Europe seems to lie in the U.K. where a predominantly strong insurance board still attempts to resist the bancassurance trend even in the face of widespread deregulations.
One of the major differences between banks in Europe and those of the U.S. is that domestic banks in Europe are protected by a domestic flagship (Boot (2003)). This is not surprising, since control of the financial sector, central banks and government ministries, generally lies in the hands of a homogeneous group of executives. Thus in Europe, as is the case for other countries such as Japan, foreign ownership of domestic banks is very unlikely. However this still enables these banks to expand into other areas of the financial services industry. In contrast, the U.S. has lobbies that aim to preserve traditional demarcations between financial institutions.
France:
In France, the success of bancassurance is mitigated by a favorable tax treatment on life insurance products, lack of competition within the insurance industry, and an inadequate pension scheme (Bonnet and Arnal (2000)). The pioneer of bancassurance in France is argued to be Credit Mutual, which created its own life and non-life subsidiaries in the early 1970’s (Sakr (2001)). Bancassurance has seen the most success in the life insurance
53 market, something which is true for every nation, increasing from 52% in 1995 to account for 69% of life insurance business in 2000 (Durand (2003), and Turner (1998)).
However, as of late, the banking networks market share of the life insurance market has remained fairly stagnant, actually dropping over the years to 66% market share in 2001 and 61% in 2003 (Falautano and Marsiglia (2003), Datamonitor (2003)). This resulted from a combination of falling stock market prices and the banking network bearing the brunt of lower transfer prices according to Benoist (2002).
In contrast to the success bancassurers have seen in the life insurance market for so many years, the non-life market has proved to be a greater challenge. Banks only began offering non-life products in the early 1990’s and it represented only 1% of the total individual premiums in 1991 (Tassin (2002)). By 1997, this had grown to 6%, and in
1999 the banking networks were ranked 4th amongst the distributors of non-life insurance products with a market share of 8%. This trend is set to continue as bancassurers develop new more cost-effective ways of producing and distributing non-life insurance products.
Currently, the market share is around 20%. The slow growth in non-life products can be explained by the fact that they are unlike any traditional banking products and hence offer greater problems for those creating and distributing them (i.e. banking staff).
Regulations of financial institutions that offer a number of different products (banking, insurance, securities) are on functional basis. This means that banking and insurance companies are overseen separately within the country. For a conglomerate, the regulator will depend on who is the parent of the two. For example, if the bank is dominant, then it
54 is the job of the banking regulator to oversee the company. There are no separate
regulators for financial conglomerates, merely a strong cooperation between different
regulators.
United Kingdom:
Bancassurers have faced a tougher time in trying to penetrate the U.K. market, thanks in
large to a combination of restrictive regulations and a powerful insurance governing
body27. The first move for bancassurers came in 1985 when Standard Life purchased a stake in the Bank of Scotland. Changes in legislation soon followed in 1986 and 1988,
which made it legal for banks to market insurance products and set up their own
insurance subsidiaries (Sakr (2001)). Even then, the main type of union between the two
was a joint venture, since the banks placed an emphasis on maintaining the knowledge of
the insurer. Twenty years later, researchers argue that bancassurance is still in its infancy
within the U.K., currently accounting for 15% of new insurance premiums issued
(Benoist (2002)), but is nevertheless making inroads into certain areas such as motor
insurance (Datamonitor (2004)).
It is argued that restrictive regulations were detrimental to the growth of bancassurance
within the country and that due to the lack of experience the correct model for the U.K. is
still to be found (Hubbard (Spring 2001)). Two benefits of the regulatory system in the
U.K. are, firstly, that it is based on one almighty regulator that overseas the different
27 Restrictive regulations lasted longer than in other countries such as France, but even these are now being phased out of the system.
55 facets of the financial services industry (the Financial Services Authority). This leads to
more streamlined regulations than in other countries that employ functional form
regulatory systems. Secondly, prudential regulation means that there is little monitoring
of premium rates or coverage, allowing the market to determine for itself the correct price
of insurance28.
Spain:
Spain has one of the most developed markets in bancassurance (Datamonitor (2003)).
Current penetration of bancassurers is over 75% of life insurance business and an ever- increasing proportion of the non-life business. In Spain, the evolution of the bancassurance market is fostered by the phenomenal growth within the insurance services industry (life insurance alone has seen 30% growth per annum over the past 15 years
(Durand (2003)). The development of bancassurance in the Spanish market was facilitated by the well-established network of regional building societies, and also the cultural mentality that it is correct to take on risks (Goddard (1999)).
LATIN AMERICA:
Latin America in itself can be said to be an emerging region that is currently facing many changes, both financial and political. Throughout the region, economic stabilization has
decreased inflation rates and led to increased consumer savings. The regions population
28 Ward and Zurbruegg (2000) argue that this may actually be a potential problem. Even though it fosters a free market, Ward et al. argue that reliance on the market in such a way may introduce added volatility into
56 is growing at a rapid rate, and the average age is relatively young. Governments are
seeking to encourage personal savings by privatizing government pension schemes and
promoting financial deregulation in order to allow banks to underwrite insurance
products (Nigh (2000)). Furthermore, a wave of democratic reforms has been sweeping the region in the past few decades, meaning that governments have less of a say in the financial services industry and paving the way for increased competition. Certain researchers such as Nigh and Saunders (2003) argue that in response to the economic and political situation within the region, the main aim of bancassurance is to generate additional capital and create shareholder value29. Even though the maturity of
bancassurance in Latin America is greater than in a region such as Asia, to date the
bancassurance penetration as a percentage of combined GDP is still relatively low (Nigh
(2000)).
Each country within the Latin American region has differing regulations concerning
bancassurance operations. As can be seen from the three examples below, bancassurers
still face numerous obstacles if they wish to be a predominant force within the Latin
American insurance markets.
Brazil:
In Brazil the laws are in the bancassurers favor, and the banks within the country control
more than 65% of the insurance market (Nigh and Saunders (2003)), a size that rivals the
the system.
57 leading bancassurers in Europe. Furthermore, in Brazil, bancassurers are assisted by regulations that ban the development of agent networks (Benoist (2002)).
Chile:
In contrast to Brazil, Chile has more stringent regulations in place that prohibit the direct sale of insurance products to bank customers unless such sales are incidental to the banks sale of the banking product. In addition, a bank cannot directly own an insurance company unless they are a holding company.
Argentina:
In between the two extremes we have Argentina, which permits the sale of insurance products by banks to its customers only through an intermediary (Nigh (2000)).
NORTH AMERICA:
The North American financial services market is the largest in the world and bancassurance has developed in a differing manner in this region depending on the country in question. In Canada, there has been consolidated regulation for more than 15 years and banks are legally allowed to own insurance companies, but limitations are placed on the products that can be provided (Dorval (2002)). While in Mexico,
29 While generating additional capital is a driving force behind bancassurance in Europe, it is not one of the main ones.
58 bancassurance has been a flourishing industry due largely to the role played by banks in
the creation of pension funds since the 1997 pension reforms.
Bancassurance in the U.S. has, in contrast, faced a very tight regulatory and legislative
environment for many decades. The formation of financial conglomerates was greatly
hindered by the Banking Act of 1933 (Glass-Steagall Act)30 and the Bank Holding
Company Act of 195631. Only in 1999 did laws become more favorable to banks offering
insurance products, with the passing of the Gramm-Leach Bliley Act32. However, due to
the divergence between the state and federal laws regarding banks offering insurance
products, bancassurers still face a hard time ahead in relation to regulations (Scharfstein
(2000)) and attempting to overcome powerful lobbies that aim to maintain existing hierarchies (Boot (2003)). Currently, only around 7% of Americans purchase their
insurance products through bank branches (Thomson (Summer 2002b)). However, with the ever-continuing regulatory changes such as the demutualization of insurance companies coupled with an ageing population, it is widely believed that there will be
strong growth potentials for bancassurers in a mature market such as the U.S.
30 It was believed that the banks’ investment in stocks precipitated the Great Depression; hence a distinction was made between the functions of a commercial and an investment bank. 31 This act prohibited bank holding companies and their subsidiaries from marketing insurance. Other legislations still allowed this but only from a place of 5000 people or less or in relation to bank related activities. 32 Also known as the Financial Services Modernization Act (1999). It repeals the Glass-Steagall and Bank Holding Company restrictions on affiliations between banks, insurers and securities firms.
59 ASIA and the PACIFIC:
Bancassurance in the Asian region has been relatively slow to take off, with the exception
of countries such as Australia, Hong Kong and Singapore where regulations have been
considerably lenient (Swiss Re. (2002)). The trend in the majority of mainland Asian
countries has been for a bank to form ties with a foreign insurer in order to begin
bancassurance operations with around 80% of these being life insurers, and the financial
structure of the operation tends to be in the form of a distributional agreement. Since
bancassurance is still in its infancy in most Asian countries, it is very susceptible to
global changes. The Swiss Re. (2001) study argues that one of the major threats to the
growth of bancassurance in the region is a U.S. or EURO economic slowdown33.
Most countries within Asia have only recently begun allowing the formation of bancassurance operations with the main players listed below. Certain countries within the region are still holding out against the onslaught of the bancassurance trend. Vietnam still restricts banks from offering life insurance products, while South Korea has made certain rules that make it difficult to begin a bancassurance operation within the country34.
Nevertheless, bancassurers have made considerable advancements within the Asian region, having a positive outlook for future growth. The Swiss Re. (2002) study believes that by next year (2006) bancassurers could account for 13% of total premiums collected in Asia’s life insurance sector and 6% for the non-life sector, with around one-third of these being the result of new business.
33 These two are the major export regions of a number of Asian countries.
60 Australia:
Australia can be seen as a good benchmark for other Asian nations in terms of its mature
bancassurance industry. The country has never prohibited the creation of financial
conglomerates, with them constituting more than 80% of total financial system assets
today. The four major banks in the nation have operations in all areas such as banking,
insurance, investments and securities. The only major change in the country occurred in
1998 when the nation turned to a consolidated regulatory system under the government body referred to as the Australian Prudential Regulation Authority (APRA).
China:
China has begun allowing bancassurers greater access to its markets, but in a differing
manner to any country before it. China made several promises that upon its accession to
the World Trade Organization (WTO) it will begin to break down barriers and allow
foreign insurers greater reign within its borders. China recently became an official WTO
member and has done what it promised by allowing foreign insurers both into its coastal
and inland economic centers; however, the level of ownership allowed is still heavily
regulated35. Currently the results of the newly established bancassurance operations
within China have been positive. Sun (2003) finds that foreign insurers have brought with
them advanced technology and management expertise and have contributed to the
34 Each bank has to have three life and non-life partners and all those must receive less than 50% of new business generated in any quarter (Nigh and Saunders (2003)). 35 China will allow non-life bancassurance subsidiaries to be formed two years after WTO accession, while it will allow life bancassurance subsidiaries to be formed only five years after it accession.
61 creation of new jobs and greater levels of taxable income. It seems that in China, the last major hurdle bancassurers need to overcome is the superstitious nature of the populace, as is argued by Chu (2000)36.
Japan:
Until recently, Japan has had a very restrictive view in which it believed market access should be limited. It was not until 1998 that widespread reforms occurred within the country. A consolidated regulatory system was implemented (Financial Supervisory
Agency) and the ban on the establishment of holding companies was lifted, allowing affiliations amongst banking, insurance and securities businesses. Beforehand, Japan operated a ‘convoy system’ in which the regulator set rates at a level that enabled the weakest company to remain solvent and protected from competition (Ward and
Zurbruegg (2000)). Bancassurance in Japan has finally begun; however, it still has a long way to go before it can overcome all the restrictive regulations and change consumer attitudes.
2.5.4 Challenges Faced by Bancassurers
As with any relatively new business, bancassurers face their fair share of problems, relating to such things as the way the operation is financed to the restrictive regulations of certain countries and the ever-continuing competition from traditional insurers. The following section outlines what researchers agree to be the most problematic areas that
36 It is argued that purchasing insurance may be considered as cursing oneself and asking for misfortune.
62 bancassurers currently face. The importance of this section lies in the fact that it shows that differing countries pose differing challenges to bancassurers that in turn will affect the accuracy of our results when comparing bancassurers from various countries.
The authors of the 2002 Aviva Plc annual report argue that the first problem that the bancassurance operation faces is the way that it is actually financed. For example, Aviva
Plc.37 uses internal debt for financing purposes, which has a potential negative effect, since it leverages the group without being visible in the consolidated balance sheet. This is an important consideration, more so in the case of a banking business starting an insurance operation from scratch, since this will require heavy investments at the start.
The implementation and integration stage possess its own problems, such as the risk of substitution between similar banking and insurance products during which time certain products may inadvertently become lost (Turner (1998), and Swiss Re. (2002)). The consumer may not react favorably to such a substitution and hence the banks client base may be adversely affected. Neither is it uncommon during the implementation stage for the insurer and the bank to disagree on the correct business approach to use, which will inevitably create further communication problems within the bancassurance operation in the future (Klein (2001)). This conflict of interest results from the insurers long-term approach to premium revenue generation being at times incompatible with the banks need for instant fee-based income for services rendered (Scharfstein (2000)).
37 This company is more of an assurebanker, but remains valid for explanatory purposes.
63 Other conflicts of interest or agency problems will result when the incentives within the
bancassurance group do not align themselves with the consumer’s best wishes
(Staikouras (2005c)). Examples of such would include the salespersons not providing
objective advice on commission-based insurance products, and tie-in sales forcing
customers to purchase products they might not otherwise have chosen.
In relation to the business approach employed, the overall structure of the bancassurance operation also tends to provide bancassurers with numerous headaches. Some banks form a weak tie with the insurer and only rent out their customers in return for a small commission38 (BCG (1999)). The problem with this structure according to Berberich
(2000) is that it means that the banks client data cannot be used for cross-selling
purposes, and additionally the client may react unfavorably to a suggestion of releasing
their private information to others. On the other hand, if a full integration occurs, it may
be difficult to assess the real cost of the bancassurance operation (Morgan (2003)). The
Swiss Re. (2001b) study argues that one problem that arises in this instance is that the
operating risk of the company will be increased due to the difficulty of monitoring and
controlling the actions of staff. When full integration occurs, it generally means that bank
staff are handed the responsibility of distributing the various insurance products, even
though they may not be skilled enough for such a task, which will impact on the quality
of customer service provided (Klein (2001)).
Once the structure has been decided upon, Nigh (2000) argues that getting the
distributional approach right is the most important factor bancassurers have to consider.
64 Obviously, the structure and distributional approach will depend on what country the
bancassurers are operating from. McDermott and Saunders (2003) have found that the
reason behind the poor returns of some bancassurers is that they merely copy the
structure of other successful companies from different countries without taking into
account the specifics of the country that they are operating in.
The bank is also likely to face a major image risk problem from the outset that may be
potentially disastrous to its future prospects. This comes about from the bank aligning
itself with an insurer, and according to studies such as Benoist (2002) and Swiss Re.
(2002) many customers may not look upon this favorably, since insurers generally have a
‘bad image’ in the consumers’ mindset. In the end, not all customers can be satisfied, and
a successful bancassurer should focus on those consumers that are likely to participate in
the bancassurers operation in the future.
On a global scale, the greatest challenge bancassurers face is trying to get around the
countless restrictive legislations and regulations within different countries. Even though globalization is fostering an era of decreasing regulations around the world, only a few countries have fully embraced this trend39. Even in countries that are apparently
‘modernized’, bancassurers face a tough time due to the existence of powerful lobbies
and groups that attempt to hold on to earlier times with less integration between
industries. This being said, the regulations that exist are not totally unwarranted and are
bought about by a fear of the negative influence of bancassurance on the economy. These
38 The bank only refers the customers to the insurer. 39 Refer to section 2.6.4 Global Breakdown.
65 negativities include possible decrease in client confidentiality, polarization of markets,
restriction of entry to new market players, extrusion of existing smaller players, proper
disclosures not being adhered to, and regulatory arbitrage.
2.5.5 Quantitative Works of Major Researchers
Compared to the vast amount of descriptive work that has been published in the field of
bancassurance, there is only a limited amount of empirical studies conducted on the
effects that bancassurance actually has on the company once implemented40. As was
mentioned previously, this was largely due to the lack of information that resulted from
poor company disclosure statements and inadequate collections of national statistics. As
these problems are being rectified, more and more empirical research is being made by
researchers into the bancassurance practice; nevertheless, it is still in its early stages. The
following aims at highlighting the major quantitative findings of certain researchers that
have performed research into the union of banks and insurers41. We use the term
bancassurance sparingly here, since the majority of the works referred too do not look at merely bancassurance but a broader union between banking and non-banking firms
(insurance, securities and investments firms).
The majority of past studies have focused mainly on the risk and profitability effects
resulting from the union of a banking and non-banking firm. One of the earliest studies in
40 Most of the empirical studies are quite old and originate from before 2000. Current articles are much harder to come by. 41 Note that we include the majority of researchers here, while a few that are not mentioned here being referred to in section 2.7.2.1.
66 this area was performed by Boyd and Graham (1986). They conducted a risk-of-failure analysis42 and looked at two periods around a new Federal Reserve Policy (1974s go- slow policy). They found that bank holding companies (BHCs) involvement in non- banking activities is significantly positively correlated with the risk of failure over the period 1971-1977, while the period 1978-1983 showed no significance, thus indicating that the new policy had a considerable impact on bank holding company (BHC) expansion into non-banking activities. Wall (1987) conducted a similar risk-of-failure analysis and found that diversification into non-banking activities may slightly decrease
BHC risks, though the findings were not statistically significant. Boyd and Graham
(1988) followed their 1986 study with a paper that used a simulation approach, whereby they simulated possible mergers between banking and non-banking companies which were then compared to existing BHCs in order to determine whether the risk of bankruptcy will increase or decrease should expansion be allowed into the non-banking industry, and also to determine the concurrent effect on company profitability. Their main finding was that the risk of bankruptcy only declined should the BHC expand into the life insurance practice. Brewers (1989) study finds similar risk reduction benefits existing, however cannot specify whether they originate as a result of diversification, regulation or efficiency gains. Boyd, Graham and Hewitt (1993) build on Boyd et al. (1988) by conducting another simulation study. They once again conclude that mergers of BHCs with insurance companies may reduce risk, whereas those with securities or real-estate firms will not. Saunders and Walter (1994) and Lown, Osler, Strahan and Sufi (2000) use a similar method to Boyd and Graham (1988) and obtain similar results with more current
42 Advantage of this over variance-covariance analysis is that the risk of failure reflects both the equity position and the expected earnings of the company.
67 data. Allen and Jagtiani (2000) do likewise, but do not find any benefit from bank expansion into the insurance field. Estrella (2001) examines diversification benefits for banks by using proforma mergers. In contrast to previous studies that incorporate accounting data, Estrella uses market data and a measure of the likelihood of failure that is derived through the application of option pricing theory to the valuation of the firm.
The findings indicate that banking and insurance companies are likely to experience gains on both sides in the majority of cases. However, unlike previous studies, Estrella finds that property and casualty (P&C) insurance companies lead to larger diversification gains than life insurance companies.
The other major series of studies on banks expansion into non-banking activities focus on the wealth effects of such a move. Cybo-Ottone and Murgia (2000) analyzed the stock market valuations of mergers and acquisitions in the European banking industry over the period 1988-1997, and found the existence of significant positive abnormal returns associated with the announcement of product diversification of banks into insurance.
Furthermore, they found that country effects do not significantly affect their overall results, suggesting a homogeneous stock market valuation and institutional framework across Europe. Carow (2001) looked at the abnormal returns of bank and insurance companies following the changing legislation brought about as a result of the Citicorp-
Travelers Group merger, and discovered that investors expect large banks and insurance companies to gain significantly from the legislation removing barriers to bancassurance.
In an event study released later in the same year, Carow (Mar 2001) found in support of
68 the contestable market theory43 that insurance companies became worse off and banks
had no long-term gains following legislations further supporting bancassurance within the
U.S. Cowan, Howell and Power (2002) conducted a similar event study surrounding four separate court rulings and discovered that on average only larger, riskier BHCs with fee- based income gain the most, while smaller, riskier insurers sustain the highest wealth losses. Fields, Fraser and Kolari (2005) find that bancassurance mergers are positive wealth creating events by examining abnormal return data. They further deduced that scale and scope economies were a contributing factor in these results.
As is clear, the majority of past studies have found risk reduction and wealth creation
benefits associated with the expansion of banks into the insurance industry. However,
there are always opponents to any given theory. In this case, two such opponents would
have to be Amel, Barnes, Panetta and Salleo (2004) and Stiroh (2004). Amel et al. (2004)
examined the benefits associated with consolidation within the financial sector. They
found that consolidation in the financial sector (results hold for commercial banks and
insurance companies) is beneficial up to a relatively small size in order to reap economies
of scale, and that there is no clear evidence supporting cost reductions stemming from
improvements in managerial efficiencies. Stiroh (2004) finds non-banking income more
volatile and that there is little evidence of diversification benefits existing.
43 The contestable market theory shows that the removal of entry barriers can reduce the long run profits of existing companies within the industry.
69 2.6 DATA AND METHODOLOGY
2.6.1 Data Sources44
This study employs cross-sectional data for the five-year period of 1999-2003
encompassing 28 countries. The countries in our sample include:
Table 2.1: Sample Countries45
Argentina Estonia Italy Peru Taiwan Australia France Japan Philippines Turkey Belgium Germany Luxembourg Portugal U.K. Canada Greece Malaysia South Africa U.S.A. Czech Republic Hong Kong Netherlands Sweden Denmark Ireland Norway Switzerland
A large number of countries (and banks) were excluded due to lack of available data over the period of study, creating potential selection bias. We found that some country national bank websites and a considerable amount of bank (offering bancassurance) websites did not have data that went more than 3 years back in time. The most notable exclusions are Spain and Brazil, which are bancassurance superpowers. A potential survivorship bias may also exist, since we only include firms with data for the required
44 For a summarized description of the sources for all the variables being used as well as a description of each variable please refer to Appendix 1. 45 An extended form of this table is presented in Appendix 1 depicting the differing samples each country belongs to within our study.
70 five-year period of our study (plus two lag years) and excluded those that did not have bancassurance operations over this period due to: the firm starting the bancassurance operations during the study period, firms ending their bancassurance operations during the study period, and firms going insolvent during the study period. We also found it very difficult to obtain the required data directly from the bank.
Data for premium income generated from bank insurance operations was obtained from a combination of Compustat and bank annual reports46. This includes both domestic and international bancassurance operations of the bank (should any exist). It must be noted straight from the outset that we do not include commissions generated from insurance operations that suggest a weak bancassurance model where the bank is merely an agent of an insurance company47. We do this because of two reasons. Firstly, it is hard to distinguish the different sources of commission income for a bank, making it difficult to pinpoint the exact amount originating from insurance ties. Secondly, we believe that an agency relationship is too weak of a bancassurance model to be used in conducting tests on the determinants of bancassurance. Most banks, in fact, have this sort of relationship with an insurer whereby they receive insurance commissions by promoting the insurer to their customers. We argue that these banks would not be adversely affected should that relationship be severed. Thus, we believe that including banks with weak ties to insurers will only distort our results.
46 This measure is symbolic of the size of the bancassurance operation within the bank while also being symbolic of the demand for insurance within a country. 47 We compare Compustat figures for premium income with the figures from the respective banks annual reports to ensure that the stated amount is indeed premium income and not commission income.
71 The total premium income variable is measured as the ratio of total premium income generated from the bancassurance operations of a bank, over the total revenue the bank earns. This gives a clear indication of how much bancassurance contributes to the overall company. However, using premium income on a global study such as this is problematic for a number of reasons, which we feel is important to make note of here. Firstly,
Outreville (1996) argues that since contracts are of differing nature (i.e. life and non-life) and duration, it may not be meaningful to group them all into one (for a company) by adding the respective premium income they generate, given how the overall value of each contract will be different, regardless of whether they are of the same duration or not.
Secondly, Outreville (1990) argues that the fact that considerable variation in premium rates exists between different countries, it may not be valid to compare premium income generated by exactly the same insurance product from two different countries. Browne and Kim (1993) agree with this, stating that government regulations and the competitiveness of the insurance market will further exacerbate this problem. While another problem lies in the fact that premiums can be seen to measure revenue and not strictly insurance output (Esho et al. (2004), Yuengert (1993)). An alternative, value- added, approach is suggested by Cummins et al. (1999) in which you sum the face value of insurance policies in effect. Here, outputs are measured in terms of services provided and include risk transfer and financial intermediation. However, due to the difficult procurement of this type of data we remain with our initial premium income measure regardless of its shortcomings.
72 Bank specific expenses, revenues and total assets are also obtained from a combination of
Compustat and bank annual reports. Total asset figures are then converted to USD to
allow for comparisons of banks in differing countries. Since total assets accumulate over
a given year, we used the average exchange rates for conversion purposes. This, however,
introduces a slight measurement error, as argued by Park, Borde and Choi (2002), since
exchange rates can have a confounding effect and obscure the true picture regarding
bancassurance operations given the current volatile currency exchange systems
throughout the world. The exchange rate figures were obtained from the International
Monetary Funds (IMF) International Financial Statistics (IFS) website.
The competitiveness figures for each country were obtained from the International
Institute for Management Developments (IMD) World Competitiveness Yearbook. Both the hardcover version and online site were utilized to obtain this information.
The daily stock price data for each company which is used in the calculation of the risk proxy was obtained from DataStream. Total banking assets of a country were obtained from each country’s respective national bank website and then converted to USD based on the same IFS exchange rates previously used. For inflationary figures, we used the
IMFs World Economic Outlook. While for GNI and GDP figures, we used data from the
World Bank.
73 The following table presets the summary statistics for the regression variables used.
Table 2.2: Descriptive Statistics48
Stand. Minimu Maximu Variable Mean Dev. m m Total premium income (BA) 0.1354 0.1510 0.0002 0.5855 Risk proxy (RISK) 28.2698 146.7488 0.0000 1240.119 Customer base/Size proxy (SIZE) 10.9818 2.1139 5.8375 14.0498 Cost savings proxy (EXP) -0.0294 0.1284 -0.8891 0.2158 Revenue increase proxy (REV) -0.0129 0.2632 -0.3443 1.9051 Nations banking sector proxy (SIZE(N)) 2.4798 3.5402 0.3059 28.8457 Level of deregulation (REGL) 19.5753 17.4864 1.0000 60.0000 Demand for insurance proxy (GNI) 0.0971 0.0428 -0.0972 0.2160 Level of inflation (INFL) 2.7644 4.3110 -2.6000 25.3000
Total Premium Income represents the ratio between total premium income for the bank generated by insurance operations and the total revenue for the bank generated from all operations. All figures are for the year 2003
This study incorporates the OLS and GMM estimation techniques49 and makes use of both STATA and TSP statistical packages. We employ pooled cross-sectional analysis over all the years of study. Further explanations of these estimation techniques and the reasons for their use are outlined later in this section.
48 Appendix 1 has an expanded Descriptive Statistics table in which we provide the statistics based on the differing samples we use in our tests.
74 2.6.2 Determinants of Bancassurance
The prior works of researchers such as Outreville (1996) and Browne and Kim (1993) have looked at a number of variables when examining demand for both life and non-life insurance domestically and abroad. Our aim here is to employ a similar technique when looking at bancassurance by using a number of company and national variables that we consider to be most relevant in helping us gain a better understanding of what factors impact bancassurance operations. Theoretically, there is a multitude of factors that could have been tested for, and we believe that the variables that we incorporate in this study represent the key determinants of bancassurance.
One of the major problems that we came across when trying to include different variables was a lack of data for all the countries in our sample50. For this reason, we had to leave
out certain variables that we believe may also influence the success of bancassurance
operations. Some of the most notable exceptions would include:
• Social Security Expenditure: The social security expenditure is argued by
researchers such as Browne et al. (1993) to be influential in the demand for life
insurance. Furthermore, one of the most noted reasons for the birth of
bancassurance, as discussed earlier, was the inadequate pension schemes of
countries such as France that forced consumers to look for alternative savings
49 Since lag instruments are used when applying GMM estimation we include another two years to our sample period making our period of study 1997-2003 (five years of study plus two lag years). 50 Or even for a large portion of the countries which would have allowed us to run significant tests.
75 methods to support their latter years. Browne et al. (1993) uses the government’s
aggregate public social expenditure to proxy this.
• Education: Browne et al. (1993) argues that higher levels of education would
cause an increased level of risk aversion through a better understanding of the
uncertainties in life and the necessity for protecting oneself through means such as
life insurance. This will be influential in bancassurance operations, since the
majority of products sold are life related, with non-life products only recently
gaining in popularity. At the same time, a confounding viewpoint can also be
posed in which it can be argued that through education, the individual will be
better able to determine risks inherent with everyday activities and hence avoid
them, resulting in a negative impact on the success of bancassurance operations.
• Life Expectancy: The effect of this on bancassurers is uncertain. Outreville (1996)
concludes that a longer life expectancy is likely to have a positive impact on the
demand for life insurance if a longer life span results in a lower price of
insurance, leading to greater incentives for human capital accumulation. However,
if lower levels of life expectancy coincide with a greater chance of death, people
would have an incentive to protect against this, and there would be a negative
relationship between average life expectancy and life insurance demand.
In this study, the model that is used incorporates variables from the literature on bancassurance and life and non-life insurance. While it is unfortunate that we had to
76 exclude certain explanatory variables, we believe that the eight variables included in our study form a key part of explaining the success of bancassurance operations, and hence can be classified as key determinants of bancassurance. The variables comprise:
• Company Risk
• Company Size
• Reduction in Costs
• Increase in Revenues
• Size of National Banking Industry
• Level of Deregulation within a Country
• Changes in National Income
• Level of Inflation
2.6.2.1 Explanatory Variables.
2.6.2.2.1 Risk Proxy
Measure: Standard deviation of daily share price over a given year, for each year.
Because the risks faced by banks are different to that faced by insurers, one common belief is that a risk diversification benefit exists when a bank takes on an insurance operation. One of the major problems of loan-based income from banks is that is it very sensitive to economic downturns, whereas income from another source such as insurance
77 may not be affected to such a great extent. Portfolio theory advocates that the expansion
of activities will reduce the risk of bank failure as a result of the diversification benefits
that arise. On the other hand, nontraditional banking activities may be considerably
riskier and less regulated than traditional activities, causing the risks faced by the bank to
increase regardless of any diversification benefits that may exist. From this is it clear to
surmise that nontraditional activities have an affect on the riskiness of banks. However, it
is hard to say whether the affect is a positive or a negative one.
Past researchers findings in relation to the effect of nontraditional banking activities on
the overall riskiness of the bank are mixed. Boyd, Graham and Hewitt (1993), Boyd and
Graham (1988), Brewer (1989), Estrella (2001), and Smith, Staikouras and Wood (2003) find risk reduction benefits associated with nontraditional banking activities. Boyd et al.
(1988) looks at what happens when banks expand into the insurance, real estate and securities industry. They simulate mergers between BHCs and firms in these industries and then compare the risk of bankruptcy for the simulated firms with those of unmerged
BHCs. They find mergers with life insurance companies reduce overall risk, while those mergers with securities or real estate firms increase risk. Boyd et al. (1993) finds similar results to Boyd et al. (1988) in that mergers of BHCs with life or P&C insurance companies may reduce risks, while those with securities or real estate firms would likely increase risk. Estella (2001) finds diversification gains from all combinations of banks with insurers. Smith et al. (2003) find negative correlations between the interest and non- interest income streams, which in turn leads them to conclude that non-interest income stabilizes bank earnings.
78 On the other hand, Allen and Jagtiani (2000), Boyd and Graham (1986), Demsetz and
Strahan (1997), DeYoung and Roland (2001), DeYoung and Rice (2004), Nurullah and
Staikouras (2004) and Stiroh (2004) find that diversification into nonbanking activities
increases the riskiness of banks. Boyd and Graham (1986) find that when BHCs are more
stringently regulated, the positive association between risk and nontraditional banking
activities disappears. Demsetz and Strahan (1997) find that while large BHCs are better
diversified than small BHCs, but this diversification does not translate into reductions in
risk. DeYoung and Roland (2001) examine non-interest income using a U.S. sample
during the 1990’s and find that income from traditional activities was less volatile than
income from fee-based activities. Stiroh (2004) looks at the impact of non-interest
income on bank revenue volatility and finds non-interest income is more volatile. He also
finds the growth rates of interest and non-interest income have become more correlated in
recent years and in turn finds little evidence of diversification effects of non-interest
income. Allen and Jagtiani (2000) find that while banks overall risk is reduced with new
banking powers, insurance activities only help in reducing interest rate risk and not
general market risk.
As a result of these past studies, we feel it appropriate to leave the possible relationship
between our risk proxy and the demand for bancassurance in question until we have
conducted our own set of tests. We agree that diversification benefits exist as a result of
the differing risk structures that banks and insurance companies have. However, we
cannot accurately determine whether or not this diversification benefit outweighs all the
other inherent risks associated with a bank having an insurance operation. Risks such as
79 the integration of the new operation, successfully training staff, reaction of the customer
base, success of the products offered, and so forth.
Hypothesis: Company risk may be either a positive or negative determinant of
bancassurance.
→ If a negative relationship exists, then it means diversification benefits are
predominant, i.e. risk reductions are synonymous with growth of bancassurance
operations.
→ If a positive relationship exists, then it means diversification is outweighed by
other risks brought about by the implementation of a bancassurance operation.
2.6.2.1.2 Customer Base/Size Proxy
Measure: Total banking assets (company figure) as a gross amount.
Through this proxy we aim to determine if the size of a company is influential on the
success of a bancassurance operation. The size of the bank is also indicative of the
customer base that it has. Theoretically, a larger bank can take advantage of a greater customer base in which it can promote new products. Larger banks would then have a greater chance of selling more insurance products, while spending proportionately less amount on promotions and advertisements51. Canals (1998) argues that the size of the
51 Assuming an advertising campaign costs a similar amount to both a small and large bank, say a T.V. advertisement, the profit that the larger bank receives would be considerably more than the smaller bank
80 bank is also important when gaining new customers since it indicates maturity and
solvency. Chen and Wong (2004) agree with this, arguing that the size of total assets are
indicative of the health of an organization, since regulators are less likely to liquidate
larger companies. However, at the same time, it may mean that managing changes within
the organization is more difficult. Amel et al. (2004) is one opponent who argues that
consolidation in the financial sector, especially amongst commercial banks and insurance
companies, is only beneficial up to a certain (relatively small) size in order to reap
economies of scale.
We believe that larger banks would be able to operate a more successful bancassurance
operation due largely to the greater experience they would have within the market and at
managing different operations, and at being able to effectively handle competition that will be offered by other insurers, bancassurers and boutique companies.
Hypothesis: Bank size, which is indicative of customer base, is a positive determinant
of bancassurance.
The best measure of customer base or company size is the total assets of a company. We initially use a gross figure here, but as will later be shown, we use the log of this figure for calculation purposes.
since it is reaching a larger customer base. So if a similar ratio of the existing customer base takes up the offer in both the small and large bank then the large bank would have profited more from the
81 2.6.2.1.3 Costs Savings Proxy
Measure: Annual percentage change in (total bank expenses) / (total bank revenue)52.
As was mentioned earlier in Section 2.5.2.1, one of the major benefits of starting a bancassurance operation is the resulting reduction in company expenses. The multitude of researchers such as Cowan et al. (2002) and Berberich (2000) agree that there would be a positive relationship between this proxy and our dependant, meaning that as the bancassurance operation grows the expenses generated by each new output53 should be
proportionately less. Canals (1998) argues that one of the advantages of universal banks
is the economies of scope they create by allowing costs to be shared amongst different
business units. Lown et al. (2000) find that banks have a cost advantage over life insurers
through their sales staff who work without commissions, and the economies of scope that
exist resulting from the branch systems, customer information and trust, and
administration within a banking system. Webersinke (2000) says that due to the high cost
ratios faced by banks, the attractiveness of bancassurance in South Africa is increasing.
Klein (2001) finds that banks enter the insurance market in an aim to reduce the effect of
the banks fixed costs.
As with any theory, opponents exist to the common view. An example in this case would
be Rime and Stiroh (2002) who find weak evidence of economies of scope. However,
bancassurance operation. This assumes customers behave similarly in all banks. 52 Total bank revenue includes banking revenue from all sources including bancassurance. 53 Output=Insurance Product. Here we refer to all costs associated with the product from conception to final sale to the consumer.
82 since their sample was based on banks from a small country, Switzerland, it is not
indicative of other countries with larger banking systems. Morgan (2003) mentions
another problem, which highlights the difficulty in assessing the real costs of
bancassurance, especially where the bancassurers is wholly owned by the bank.
This paper agrees with the consensus amongst the majority of researchers, in that
bancassurance results in cost benefits, and hence our hypothesis becomes:
Hypothesis: Cost Savings Proxy is positively related to the demand for bancassurance.
However, it must also be noted that as far as we are aware (based on all the papers we have researched), no researcher factors into their calculations the actual length of time that the bancassurance operation has been running. This means that no clear distinction was made for the period of time that it takes for companies to integrate new operations.
The current body of work merely assumes that cost savings benefits will be evident straight away. It is true that over the long-term, integration costs would have been overcome by the cost savings from the actual bancassurance operations. However, it is less clear as to how such operations affect costs over the immediate short-term54. We
propose that the integration period will be influential, in that there are significant start-up
costs involved whenever a bank takes on a new operation, especially one such as
54 Generally argued to take around two to three years for the effects of integration to dissipate. However, we believe that it could take even longer, up to five years, due to the fact that the operations of banks and insurers are so fundamentally different (Swiss Re. (2002)).
83 insurance that is inherently different from the current bank operations55. Start-up costs refers to such things as resource allocation (advertising, computing equipment, time spent re-training current staff, tailoring products56, etc.), re-distribution of finances to support the new operation, overcoming the differences between the bank and the insurer (Wade
(2000)), and the impact on company reputation (current client base as well as the markets in general may not react favorably to the expansion). Canals (1998) agrees with this theory by arguing that economies of scope may be difficult to achieve since, the company has to amalgamate, manage and coordinate different businesses under a single umbrella.
Morgan (2003) argues that the marginal cost of transacting bancassurance decreases with time as the operation becomes more heavily integrated into the existing banking activities.
Since our sample contains a number of companies that have recently begun their bancassurance operations, namely companies from the U.S. and Japan, we believe that this may have some influence on our results57. To obtain clear insight into the impact on start-up costs, tests would have to be run on the period immediately following the expansion and compared to the longer-term results. Since the number of new companies in our sample is too small to provide any significant results, we leave this up to other researchers to look at.
55 The cost of integration will depend on the bancassurance model employed. Morgan (2003) argues that certain models such as the Northern European Model have lower integration costs than others. 56 Thomson (2001) argues this is especially important in the U.S. where it is difficult to sell insurance products in the mass market, through any sales channel, because of the cumbersome product delivery system that insurers use. To overcome this, banks have to tailor insurance products and support services for the bank-customer mass market.
84 As a measure we use the ratio of the total bank expenses to its total revenues, since this will clearly show whether expenses have risen or fallen in regards to new operations.
2.6.2.1.4 Revenue Increase Proxy
Measure: Annual percentage change in (total bank revenues) / (total bank assets).
As with the Cost Savings Proxy, the general consensus amongst researchers is that one of the major benefits of bancassurance is the increase in revenues that it will bring to the bank. Wepler, Linn and Linnert (2004) state that one of the major reasons banks in the
U.S. entered the insurance industry so aggressively following the GLB act was to expand their non-interest income. Heymowski (2000) agrees with this and argues that this was due to shrinking markets, decreasing net interest margins and the ability of insurance products to have low renewal margins while still providing a strong income source.
Agrawal M. (2002) and Klein (2001) both argue that one of the major advantages of bancassurance is the potential for a substantial increase in bank revenue. Once again opponents exist, such as Berger, Humphrey and Pulley (1996) who find no revenue economies of scope in their study.
This paper agrees with the consensus and argues that there is a clear relationship between increases in revenue and the level of bancassurance operations a bank has.
57 We expect a positive relationship between this proxy and the dependant, but the significance of the relationship may be dampened by the influence of the initial start-up costs.
85 Hypothesis: There is a positive relationship between increases in bank revenue and the
commencement of bancassurance operations.
As with the Cost Savings Proxy, we believe that time plays an important part in the trend
of revenue increases facilitated by bancassurance operations. While it is not our purpose
to test the influence of time, we believe it is important to mention it so that all the angles
are covered. Our belief is that revenue increases may be minimal at start-up and become
progressively more important as bancassurance becomes an integral part of the banks operations. At the start, factors such as reputation and customer sentiment towards the new, ‘foreign’ operation will impact on revenues generated. Only with aggressive advertising and promotions will positive impacts on revenues be seen, which in turn links this proxy to the Cost Savings Proxy. Only through an initial outlay will revenues from bancassurance become significant in relation to the overall revenues of the bank.
We measure this variable by observing the difference over years in the ratio of a banks total revenue (from all operations) to its size.
2.6.2.1.5 Nations Banking Sector Proxy
Measure: Annual percentage change in the (total banking assets of a country) / (GDP).
The size of the banking sector within a country is symbolic of the ability of the member
banks to run efficiently and productively in all the business operations they partake in.
86 Banks within a country with a larger, and in turn more established, banking sector would undoubtedly have more opportunities to begin bancassurance operations, finding it easier to compete against any negative sentiment towards such a move that may come from bodies of power such as the insurance lobby58. The size of the banking sector is also indicative of the faith and loyalty that customers place in banks within that country, something that will prove to be extremely important for the commencement of a successful bancassurance operation. On the other hand, having a large banking sector means that there is little room for error. Banks will only begin bancassurance operations if they are absolutely sure of its success, since the consequences of failure for that one particular operation could prove disastrous for the rest of the bank.
We hypothesize that due to the benefits that banks gain from operating in a country with a well established banking system, the likelihood of bancassurance operations beginning are going to be higher, and these operations will also be more successful since banks will only begin such a course of action if they are certain of success.
Hypothesis: There is a positive relationship between the size of a country’s banking
sector and the success of bancassurance operations.
For this measure we use the total banking assets of a country as indicated on its respective national bank website divided by GDP. We use this ratio since we believe it
58 Banks in the U.S. are finally overcoming restrictive pressures applied by the insurance lobby when it comes to expansion into bancassurance.
87 serves as a better comparative measure between countries than simple gross figures
would.
2.6.2.1.6 Level of Deregulation (within a Country).
Measure: Dummy variable symbolizing the overall competitiveness of a country.
Regulation has been one of the major impediments to the growth of bancassurance across
the globe. As was previously discussed in Section 2.5.3, countries such as the U.S. and
Japan, while being global banking superpowers, have only recently begun allowing banks
to expand into the fields of bancassurance following regulatory changes that removed constraints to possible expansion across industries59. This by no means is limited to these
two countries, but is a common occurrence in other areas of the world60. Whenever
deregulation is discussed by researchers, it is always found that benefits exist for the
domestic bancassurance markets, whether it is from increased competition that forces
incumbent players out of the market or through added product innovations that are
brought to the marketplace. Carow (2001), Lown et al. (2000), and Morgan (2003) find
benefits from the removal of regulatory barriers to all parties involved (banks, insurers and consumers). Wepler et al. (2004) finds that seven of the top thirty insurance brokers were bank owned in 2003, whereas this is only true for seven out of the top one hundred brokers in 1998, clearly showing the benefit that deregulation had in the U.S. market.
59 In the case of the U.S. (as well as the U.K.) even though it is considered a deregulated economy, the presence of powerful unions and insurance lobbies have caused significant hurdles towards the successful implementation of bancassurance (White (1990)).
88 Another benefit of deregulation is the ability of established foreign firms to gain access to
local markets where they can employ their expertise in the field of bancassurance to
facilitate growth within the industry by enabling local firms to apply tried and tested
methods from abroad. In this context, deregulation is not limited to the financial services
industry, but relates to the whole economy. Bos and Kolari (2003) argue that profit
efficiencies exist with cross-Atlantic expansion. Holland, Lockett and Blackman (1998)
agree that deregulation and the concurrent globalization facilitate the spread of new innovative ideas.
It is hard to argue against the benefits that exist for the bancassurance industry resulting from the deregulation of the financial services market, and hence our hypothesis is as
follows:
Hypothesis: The greater the level of deregulation within a country, the more successful
the bancassurance industry should be within that country. Hence, the
success and size of individual banks insurance operations should, on
average, also be more pronounced and profitable.
Deregulation is an extremely hard concept to quantify. Since we believe that deregulation in not only the financial markets but the economy on a whole is important to the growth of bancassurance markets we have used the IMDs competitiveness indicators as our proxy for deregulation. Deregulation allows for increased competition within an industry
60 China, India and South Korea are three other nations that have begun the slow progression towards deregulation of the financial services industry.
89 or country, hence the level of competitiveness should also be indicative of the deregulation within the country in question (this will of course not be an exact measure, merely an adequate proxy). The competitiveness indicator reveals how competitive a country is based on a series of factors both economical and cultural, for example the state of the financial services industry, lifestyle of the populace, and education, to name a few61.
2.6.2.1.7 Demand for Insurance Proxy.
Measure: Annual percentage change in GNI per capita62.
The overall demand for insurance products is obviously going to affect how successful the bancassurance industry is within a country and hence how profitable each bancassurer is as well. Browne and Kim (1993) and Outreville (1996) have both found a positive correlation between national income and the overall demand for life insurance within a country. Outreville (1990) also found a positive relationship between property-liability insurance market development and income.
Both life and general insurance can be characterized as a luxury item. It is not an item that the consumer places at the top of their list, instead insurance products are only
61 For more specific information, the IMD website should be consulted. 62 Gross disposable income (GDI), which equals national income minus corporate profits, social security contributions, transfer payments, and personal income taxes, as argued by Brown and Kim (1993), would undoubtedly be a more suitable measure. However, due to lack of data, we chose to use GNI figures instead. Since we are using annual percentage changes, this measure should still be suitable, since percentage-wise the two should follow a similar pattern from year to year, i.e. if GNI per capita increases
90 purchased if the consumer has enough disposable income63. GNI per capita represents how much income each individual within a country has. As GNI per capita increases, the consumers are more likely to spend their disposable income on such luxury items as insurance products. For example, life insurance becomes more affordable as income rises and there is also a greater need to protect the dependant from unexpected future income losses due to the wage earners’ premature death.
Hypothesis: As GNI per capita increases premium income levels obtained by banks
should also increase as more consumers purchase insurance products.
In some countries, insurance products have to be purchased in relation to owning certain items as a result of the legislation. In Australia, for example, upon purchasing a car, the consumer has to obtain ‘compulsory third party insurance’ and renew it on a yearly basis.
Our belief is that products required by legislation such as this would not significantly influence the relationship outlined in the above hypothesis.
Outreville (1996) uses nominal GDP per capita as a proxy for disposable income.
However, we feel that a measure of national income would provide a better proxy since it measures income earned by the factors of production. Hence, we use GNI per capita64
from one year to the next, then GDI per capita would follow suit and the percentage change would be fairly similar, since cost of necessities does not drastically change from one year to the next. 63 Disposable income is the remaining income after all ‘necessities’ have been purchased. It is what the consumer can spend on products that are not essential to sustain their life, rather improve their overall lifestyle. 64 Using the Atlas Method to calculate GNI.
91 figures for this variable. Furthermore, we take the annual percentage change to ensure the
best possible proxy for increases in disposable income.
2.6.2.1.8 Level of Inflation.
Measure: Countries level of inflation averaged over each year.
Inflation is a difficult aspect to test for, since it affects every part of the economy and not
always in the same predictable way. Two things must be looked at when analyzing the
impact of inflation of the bancassurance industry. Firstly, we must analyze the impact it
has on the company that offers the bancassurance products. Cummins (1991) found that
inflation increases an insurance company’s premium income as well as reducing the real
value of insurance liabilities. This influence would similarly be found in companies that
offer bancassurance. However, the degree to which this would be true would depend on
how established the operation is within the company65.
Secondly, we must focus on how inflation impacts the consumer of the bancassurance
product and in turn how it affects demand. A similar argument can be used as in the
previous section, where we studied GNI per capita. As inflation increases, the overall cost
of a basket of ‘necessities’ increases, however a rise in inflation does not necessarily
mean that the income of the consumer will also increase. This, in turn, would mean that
consumers will have less disposable income, and hence, their demand for bancassurance
92 products would undoubtedly diminish. Browne and Kim (1993) and Outreville (1996) both found that inflation would be detrimental on savings through insurance products as it erodes the value of the product (any final claim payment that one receives in a year of high inflation would be drastically decreased).
As a result of these arguments, it is not clear to say how inflation affects the bancassurance industry. Furthermore, one cannot specify which impact (positive impact on the company or negative impact on the consumer) would take precedence over the other. In our belief, the overall impact on the consumer would inevitably be more influential, since they are the ones that in the end determine how well any product in the marketplace performs. It should finally be noted that during the period of study, the levels of inflationary rates within our sample countries can be said to be at manageable levels, thus an insignificant inflation variable (whether positive or negative) would not be of a great surprise either66.
Hypothesis: The impact of inflation on the demand for bancassurance is uncertain.
→ If positive, it means that the positive impact it has on the company’s premium
income and liabilities more than makes up for the negative impact it has on
consumer demand.
65 Those banks that have only recently started up a bancassurance operation could not possibly gain as much benefit from inflations positive effect on premium income and liabilities unlike those banks where insurance now comprises a key component of everyday operations. 66 While the variable may prove insignificant, the argued positive or negative relationship will still shed some insight into the overall effect that inflation has on the demand for insurance products.
93 → If negative, it means that consumer demand decreases to such an extent that the
company is adversely influenced regardless of any positive affects it may achieve
for an increase in inflation.
We use the actual average inflation rate per annum for each country akin to Chen and
Wong (2004). Researchers, such as Browne and Kim (1993), tend to use an averaged figure over a number of years. However, we propose that the consumer in particular does not take into consideration what the inflation rate will be in later years and does not plan ahead for anticipated inflationary changes. Some would argue that a change in inflation would not be influential for some time and the consumer would be able to continue their lives without change. We, on the other hand, believe that any change would have an immediately impact on how consumers perceive insurance products, mainly due to the fact that they are a luxury item that people tend to neglect should they be faced with high prices in other areas of their lives which they deem more important. Any rise in inflation will have an immediate impact and hence, we aim to determine how premium incomes behaved in immediate response to inflationary changes.
94 2.6.3 Model for the Determinants of Bancassurance
Based on the hypotheses presented in the previous section, the following model for the determinants of bancassurance is proposed:
BA = f [RISK, SIZE, EXP, REV , SIZE(N), REGL,GNI, INFL]
where,
Table 2.3: Hypothesized Relationships
Expected Hypothesis Variable Name relationship with dependant Total Premium Income BA N/A
Company Risk proxy RISK ?
Company Size proxy SIZE +VE
Cost decrease proxy EXP +VE
Revenue increase proxy REV +VE
Size of the National Banking Industry SIZE(N) +VE
Level of National Deregulation Dummy REGL +VE
Changes in Gross National Income GNI +VE
Changes in the National Inflation Rate INFL ?
Previous literature in this field uses either a linear or log-linear model when running tests.
A linear model provides a relationship between variables in absolute terms whereas a log- linear model does it in terms of elasticity. Outreville (1996) uses both forms in his testing
95 procedures. In addition, Browne and Kim (1993) use a log-linear model in their studies of life insurance demand. (Add a couple of more researches you have from current times)
We did not think it appropriate to immediately specify the model we will use without running further tests, since an incorrect functional form will result in biased and inconsistent estimators that will lead to the wrong conclusions being drawn. Hence, we employ model specification tests for the functional form through the use of the Box-Cox transformation of the linear form of our model.
Linear form of our model:
BA = β 0 + β1 (RISK) + β 2 (SIZE) + β 3 (EXP) + β 4 (REV ) + β 5 (SIZE(N))
+ β 6 (REGL) + β 7 (GNI) + β8 (INFL) + ε
The Box-Cox or power transformation is a flexible method for transforming a variable, whether it is the dependant or the independents. The Box-Cox considers a parametric
family of transformations of the dependant variable BA of the form BAθ , where θ can be
positive or negative. Similarly, for each independent variable ( X i ) Box-Cox considers
transformations of the form X i,λ , where λ can be positive or negative and i represents the i th independent variable. The values of θ and λ are then estimated using maximum likelihood estimation procedures, which will in turn give a clearer insight into what the
96 correct functional form of the model should be67. A value of θ (λ ) closer to 0 indicates that the log form would be more appropriate for the dependant (each independent) whereas a value closer to 1 indicates a linear form would be more appropriate.
Box-Cox considers the transformation for each variable in the following form:
BAθ −1 BA = θ θ
This allows for the log transformation since:
BAθ −1 log(BA) = lim θ →0 θ
The Box-Cox transformation of our model will thus be:
BAθ −1 ⎛ RISK λ −1⎞ ⎛ SIZE λ −1⎞ ⎛ EXP λ −1⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ = β 0 + β1 ⎜ ⎟ + β 2 ⎜ ⎟ + β 3 ⎜ ⎟ θ ⎝ λ ⎠ ⎝ λ ⎠ ⎝ λ ⎠
⎛ REV λ −1⎞ ⎛ (SIZE(N))λ −1⎞ ⎛ REGLλ −1⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ + β 4 ⎜ ⎟ + β 5 ⎜ ⎟ + β 6 ⎜ ⎟ ⎝ λ ⎠ ⎝ λ ⎠ ⎝ λ ⎠
⎛ GNI λ −1⎞ ⎛ INFLλ −1⎞ ⎜ ⎟ ⎜ ⎟ + β 7 ⎜ ⎟ + β8 ⎜ ⎟ + ε ⎝ λ ⎠ ⎝ λ ⎠
67 It must be noted that method is not conclusive, however it provides a much better approach to understanding what the signs of the variables in an equation should be, rather than simply guessing and
97 When using STATA to apply the transformation to both the left and right hand sides of the equation at the same time we obtained a value for θ that was not significantly different from 1, hence we continue to use a linear transformation on the dependant variable. We obtained similar results for λ on the right hand side. What we did next was to check each of the independent variables on the right hand side of the equation separately. So we left all the variables in their linear form except for one that we transformed in the above manner. We repeated this process for each independent variable
(that did not contain negative values) and found that only one of the variables had a λ that was not significantly different from 0, while all the rest had λ values close to 1, suggesting linear transformations for these variables would be appropriate. When transforming the SIZE variable we obtained a λ of 0.14, which is shown to be significant at 0, hence indicating that the log-linear transformation is valid for this variable68.
Given the above specification tests, for estimation purposes, the model that we shall use for testing purposes becomes:
BA = β 0 + β1 (RISK) + β 2 LOG(SIZE) + β 3 (EXP) + β 4 (REV )
+ β 5 (SIZE(N)) + β 6 (REGL) + β 7 (GNI) + β8 (INFL) + ε
taking the best result.
98 2.6.4 Ordinary Least Squares (OLS)
The OLS estimation technique has numerous assumptions that have to be adhered to in order for efficient estimates to result. This can also be argued to be one of the greatest weaknesses with this type of estimation technique. The assumptions in relation to the model being tested include linearity, normal distribution, homoscedasticity, no serial correlation, no multicolinearity, no simultaneity, no unit roots and the expected value of residuals always being equal to zero. Given that the data used is finite and these assumptions are adhered to, then it can be said that the OLS estimator is unbiased, consistent and BLUE (Best Linear Unbiased Estimate). In order to test that the main assumptions required for our method of testing are adhered to, we incorporate diagnostic tests to check for heteroscedasticity, normality of residuals and multicollinearity. These are briefly explained at the start of Section 2.7.
2.6.5 Generalized Method of Moments (GMM) 69
This type of estimation procedure is used in conjunction with OLS, in view of the fact that it improves on simple OLS calculations and corrects some of its inconsistencies. It allows for consistent and efficient estimates to still be derived, even if some of the
68 During our testing phase we still employed models with a log transformed dependant, as well as one with log transformation for both dependant and relevant independents. We found that both these models provided inefficient results when compared to The Model, having much higher BIC values and p-values. 69 An in-depth analysis and description of the GMM model goes beyond the scope of this study. We merely employ the process and provide a summarized incite into its functionality. For full working out of GMM efficient and consistent estimates, please refer to Davidson R. and MacKinnon J.G. Chapter 9 “The Generalized Method of Moments,” at the following website: http://russell.vcharite.univ- mrs.fr/Doctoral/chp09big.pdf
99 assumptions required for OLS do not hold. With a study such as this, numerous problems arise in terms of correct calculations. Outreville (1996) suggests that with models such as the ones that we employ, endogenous variables may be present, whereby some of the explanatory variables are a function of others or of past values of themselves. GMM assumes a weak exogenity, which means it considers how explanatory variables (as well as the dependant) may be affected by past and current values, but not affected by future values. With a cross-sectional study of countries such as ours where we incorporate a large amount of country figures for the explanatory variables, we face the problem of not using the correct proxy for the variable. An obvious example of this is our proxy for insurance demand in a country (GNI). As we stated, disposable income is a better measure than national income. However, due to the lack of available data, we cannot use the preferred proxy for the demand for insurance. Heteroscedasticity70 may also be present in our calculations and this is something that GMM corrects for.
Hansen (1982) shows that GMM instrumental variable estimation allows more robust calculations by helping avoid data problems such as errors in the variables, endogeneity and multicollinearity. GMM is also superior to OLS, since it does not require any pre- specified knowledge of the distribution of the error terms; in turn we can remove the normality requirement. Instead of this, all the GMM requires is that orthogonality conditions be met. If these conditions are met, then the estimator is consistent, and if variance of the moment conditions is consistent, then the estimator is also efficient.
70 The White variance-covariance matrix is used to ensure efficiency in regards to heteroscedasticity.
100 Valid instruments should comprise two key attributes. They should have little to no correlation with the error term, and be highly correlated with the explanatory variables.
Most researchers use lagged values of the explanatory variables, since they satisfy these two conditions, and in turn that is what this study uses. In order to determine whether the instruments we use are valid, we incorporate the Hansen J-test for overidentifying instruments. This test is Chi-Square distributed with j-k degrees of freedom, where ‘j’ is the number of instruments and ‘k’ is the number of regressors. The estimates are robust to heteroscedasticity (Beck, Levine and Loayza (2000)).
Suppose we have a linear model, as below, but we are concerned about the correlation
between the residuals ε i and the explanatory variables X i . In this case we cannot estimate β by OLS methods.
Yi = X i′β + ε i
If we have a set of instruments Z i that are uncorrelated with the error termsε i , and
correlated with the regressors X i , we can exploit the moment conditions:
Ε[Z(Y − X ′β )] = 0
The dimension of Z i in this case is not necessarily the same as the dimension of X i .
101 GMM exploits moment conditions in order to obtain consistent estimates and allows for more moment conditions to be used than actual parameters to be estimated. This is advantageous, since we can overidentify parameters in order to improve estimation efficiency and the overall consistency of the model by mitigating the effects of multicollinearity on the GMM estimates71. However, it also presents the problem of data mining, since different moment conditions can be used to improve the accuracy of results.
The GMM estimate is obtained by minimizing Q(⋅) with respect to the parameter estimates:
Q(⋅) = ε ′ZW −1Z′ε
where,
Z = Matrix of Instruments
W = Consistent variance-covariance matrix.
You then solve for the coefficients and obtain the following efficient estimates
−1 β (GMM ) = [X ′Z()Z ′ΩZ −1 Z ′X ] X ′Z()Z ′ΩZ −1 Z ′Y
and
102
Var[]β (GMM ) = (Z′X )−1 Z ′σ 2ΩZ(X ′Z )−1
where Ω = Ε(εε ′)
For estimation purposes in this chapter, we estimate the Q(⋅) using the White variance- covariance matrix in order to adjust for any possible heteroscedasticity. Then to correct for endogenous variables and other measurement errors, an overidentified estimation technique is used, reflected in equations β (GMM ) and Var[β (GMM )].
2.7 EMPIRICAL RESULTS72
Below we provide a brief overview of our empirical results, providing only the main attributes of each test conducted. We will then compare and sum up the implications and significance of these results in Section 2.7.5.
2.7.1 Whole Sample
The results of testing our whole sample of 73 companies from 28 countries using both regression analysis and GMM tests are presented in Table 2.4. The first set of tests that
71 Results from the smaller condition index in the Z’X matrix as compared to the OLS X’X matrix.
103 were run made use of a pooled OLS estimation procedure, whereby you pool all the years that you are observing and test them all at once. The benefits of this type of testing is that our sample size is larger than if we perform year by year testing, allowing us to obtain more accurate estimates, and allowing us to observe relationships over a given number of years. For diagnostic purposes, we use the LM heteroscedasticity test of the residual variances to check if heteroscedasticity is evident. This is done by regressing the squared residuals on the squared fitted values of the regression. The resulting test statistic has a
Chi-squared distribution with one degree of freedom. Another diagnostic that is performed tests the residuals for normality, since this indicates whether or not the use of the t-statistic is suitable. The Jarque-Bera test is used for this purpose, which compares the skewness and kurtosis of the residuals from the model to that of a normal distribution73. The null for this test is that the residuals are normally distributed with the test statistic being Chi-squared distributed with two degrees of freedom. Finally, we also look at whether a multicollinearity problem exists by observing the correlations amongst the explanatory variables (Table 2.5). We can see that there is apparently no great problem with multicollinearity with only the correlation between the cost savings and revenue increase proxies seeming excessive (-0.6074)74. This can be explained by the fact that they are interlinked in the way we calculate the cost savings proxy (EXP)75. These high correlation coefficients indicate the possibility of a multicollinearity problem in the
72 For testing purposes we used a combination of the STATA and TSP statistical packages 73 Skewness of a normal distribution is 0 and its kurtosis is 3. 74 This is true for all the different samples that we test. The highest this correlation becomes is -0.8449 when testing the Civil-Law sample. 75 Refer to section 2.7.2.3.
104 OLS tests. To reduce this problem, instrumental variable estimation is used through the application of GMM tests76.
For the whole sample, the LM heteroscedasticity test shows the existence of non- homogeneity in the residual variances, while the Jarque-Bera test for normality is insignificant at the 5 percent level indicating the suitability of the t-statistic. The results all have the expected signs and the adjusted R 2 for the test is 0.1543. At the 1 percent level of significance only the risk proxy (RISK), nations banking sector proxy (SIZE(N)), level of deregulation (REGL), and the level of inflation (INFL) are shown to be significant. While at the 5 percent level of significance the customer base/size proxy
(SIZE), the cost savings proxy (EXP), and revenue increase proxy (REV) are significant.
The only variable that is insignificant is the demand for insurance proxy (GNI), which is a result in contrast to those obtained by most other researchers. The constant term is also insignificant, indicating that our model does well in explaining the determinants of bancassurance.
Following the regression analysis we run GMM tests due to the advantages they possess over the more simplistic OLS type of testing, as was explained previously. The Hansen J tests for overidentifying instruments are insignificant, indicating that the instruments that are being used are appropriate for our tests. The constant is also insignificant, indicating the suitability of our model. The GMM results are fairly consistent with the above OLS
76 Instrumental variable estimation reduces the impact of multicollinearity problems, since the large condition index of the OLS X’X matrix is reduced, since we use the instrumental variable matrix Z’X which has a smaller condition index. The combination of the proper instruments (checked through the
105 analysis results with the adjusted R 2 for the test being 0.1556. The only difference being that the revenue increase proxy (REV) is significant at the 1 percent level, and the company size proxy (SIZE) is significant at the 10% level. Using the GMM approach, we also find that the demand for insurance proxy (GNI) becomes significant close to the 5 percent level (p-value = 0.052), a result that is more consistent with previous researchers.
2.7.2 Whole Sample less the U.S.
We have a large number of bancassurers that originate from the U.S. and we believed that this may distort our overall results, since this is aimed at being a global study. We ran similar tests to those above after removing the U.S. sample of bancassurers in order to see whether our previous results were actually global results, or if they were significantly influenced by the U.S. bancassurers in our sample. The sample size was reduced to 57 companies after removing the 16 U.S. organizations. The results for both the OLS analysis and GMM tests using this smaller sample are presented in Table 2.6.
Running OLS tests first, we find that the LM heteroscedasticity test is insignificant, indicating the absence of heteroscedasticity, while the Jarque-Bera test for normality is also insignificant at the 5 percent level. The results have signage identical to those that were obtained when testing the whole sample and the adjusted R 2 for this test is 0.0864.
This suggests that the results are not as strong as when we tested the whole sample and that the U.S. sample of bancassurers is influential on the results of our tests.
Hansen J-test) with the smaller condition index will result in a mitigating impact of the multicollinearity on the GMM estimates.
106 Table 2.4: Whole Sample Results
This table presents the pooled OLS and GMM results for the whole sample of 73 companies from 28 countries over the five-year period of 1999-2003. For the GMM tests an additional two years of lags we also incorporated for each variable (i.e. years 1997 and 1998 act as lags).
Ordinary Least Squares Generalized Method of Moments
Variables Estimate t-stat p-value Estimate t-stat p-value
Intercept -0.02335 -0.55814 0.577 -0.03850 -0.67787 0.498 RISK -0.00010 -3.49092 0.001*** -0.00011 -4.06849 0.000*** SIZE 0.01946 2.28497 0.023** 0.02155 1.77196 0.076*
EXP 0.06581 2.01365 0.045** 0.06298 2.00039 0.045** REV 0.00022 2.49510 0.013** 0.00021 3.39018 0.001*** SIZE(N) 0.01231 5.40629 0.000*** 0.01041 5.91786 0.000*** REGL 0.00248 5.54565 0.000*** 0.00224 3.70387 0.000*** GNI 0.01037 0.07268 0.942 0.06211 1.94500 0.052* INFL -0.00335 -6.15509 0.000*** -0.00298 -6.35780 0.000***
Adjusted 0.15427 0.15559 R 2 Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.011 0.446 Test Test JB Test 0.365
* Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. RISK=Company Risk Proxy, SIZE=Company Size/Customer Base Proxy, EXP=Cost Decrease Proxy, REV=Revenue Increase Proxy, SIZE(N)=Size of National Banking Sector, REGL=Level of National Deregulation, GNI=Proxy for the Level of Insurance Demand Within a Country, INFL=Proxy for Inflationary Changes Within a Country.
107 Table 2.5: Correlation Matrix for the Whole Sample
This table presents the correlation matrix amongst variables for the whole sample of 73 companies from 28 countries in order to allow for the detection of severe multicollinearity in the data. (Similar results are obtained for the other samples and since they do not add any further insight into multicollinearity problems they are not reported). BA RISK SIZE EXP REV SIZE(N) REGL GNI INFL
BA 1.0000 RISK -0.1330 1.0000 SIZE 0.1300 0.0164 1.0000 EXP 0.0999 0.0496 0.1624 1.0000 REV -0.0653 0.0281 -0.3504 -0.6074 1.0000 SIZE(N) 0.3093 -0.0247 0.1161 0.0521 -0.0580 1.0000 REGL 0.1288 0.0510 -0.1262 -0.2954 0.2490 -0.0908 1.0000 GNI 0.2471 -0.1622 0.2878 0.3149 -0.5548 0.1560 0.1561 1.0000 INFL 0.1362 -0.1083 -0.1241 -0.2078 0.2354 -0.1696 0.4798 -0.0847 1.0000 Note: BA = (Total Premium Income) / (Total Banking Revenues); RISK = Standard deviation of daily share price over a given year, for each year; SIZE = Log (Total Banking Assets); EXP = Annual percentage change in (Total Banking Expenses) / (Total Banking Revenues); REV = Annual percentage change in (Total Banking Revenues) / (Total Banking Assets); SIZE(N) = Annual percentage change in the (Total Banking Assets of a Country) / (GDP); REGL = Dummy variable symbolising the overall competitiveness of a country; GNI = Annual percentage change in GNI per capita; INFL = Countries level of inflation averaged over each year.
108 We find all variables significant except for the demand for insurance proxy (GNI), which is similar to our previous OLS results. However, the significance of the variables has changed, with the company size proxy (SIZE) and the level of deregulation (REGL) significant at the 10 percent level; cost savings proxy (EXP) significant at 5 percent, and the remaining variables showing significance at the 1 percent level.
Hansen’s J test for overidentifying instruments is once again insignificant while the adjusted R 2 for the GMM test is 0.0891. The results once again have the correct signage and are more conclusive than before with all variables except the company size proxy
(SIZE), the level of deregulation (REGL), and the cost savings proxy (EXP) being significant at the 1 percent level. Here, the cost savings proxy (EXP) is significant at the
5 percent level and the company size proxy (SIZE) is significant at the 20 percent level.
2.7.3 European Sample
Since bancassurance is a European concept, we felt it appropriate to test a sample of
European countries as well. From the outset, we believe that our results will not be that conclusive, since our sample of European bancassurers is not large enough and also does not accurately represent the major bancassurance nations in the region77. We ran tests on two samples of European nations. The first sample (Euro1) included all mainland
European nations in addition to the U.K., Ireland and Turkey. The second sample (Euro2) included only mainland European nations (above sample ex the U.K. and Ireland).
77 As was previously mentioned, this was largely due to lack of readily available data that existed for countries such as Portugal and Spain.
109 Table 2.6: Whole Sample (less the U.S.) Results
This table presents the pooled OLS and GMM results for the whole sample less those observations originating from the U.S. (57 companies in total from 27 countries) over the five-year period of 1999-2003. For the GMM tests an additional two years of lags we also incorporated for each variable (i.e. years 1997 and 1998 act as lags). Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Intercept 0.02212 0.36731 0.714 0.01052 0.13143 0.895 RISK -0.00010 -3.56397 0.000*** -0.00012 -4.06434 0.000*** SIZE 0.01797 1.68541 0.093* 0.01947 1.32388 0.186 EXP 0.06795 2.10367 0.036** 0.06350 1.98277 0.047** REV 0.00017 1.92031 0.056* 0.00019 2.92105 0.003*** SIZE(N) 0.00998 4.59712 0.000*** 0.00820 4.68341 0.000*** REGL 0.00125 1.77093 0.078* 0.00105 1.13106 0.258 GNI 0.04912 0.33549 0.738 0.06898 2.95633 0.003*** INFL -0.00268 -4.90850 0.000*** -0.00262 -5.25789 0.000***
Adjusted 0.08636 0.08691 R 2
Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.168 0.535 Test Test JB Test 0.278
* Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. RISK=Company Risk Proxy, SIZE=Company Size/Customer Base Proxy, EXP=Cost Decrease Proxy, REV=Revenue Increase Proxy, SIZE(N)=Size of National Banking Sector, REGL=Level of National Deregulation, GNI=Proxy for the Level of Insurance Demand Within a Country, INFL=Proxy for Inflationary Changes Within a Country.
110 We decided it may be interesting to view how the U.K. and Ireland influence the
European results, since they have a different legal system in force than the remaining
European countries that tend to be based on the Roman system of law. The results for these samples are presented in Table 2.7.
For both samples, the LM Heteroscedasticity and Jarque-Bera tests are insignificant. The average adjusted R 2 is 0.1617. For the Euro1 sample, the risk proxy (RISK), company size proxy (SIZE) and nations banking sector proxy (SIZE(N)) are significant at the 1 percent level, with the level of inflation (INFL) significant at 5 percent. In the Euro2 sample, the company size proxy (SIZE) is no longer significant.
When running GMM tests, we find that Hansen’s J test is insignificant for both Euro1 and Euro2, and the adjusted R 2 is 0.1598 and 0.1483 respectively. The results of the
Euro1 sample confirm those of the OLS tests in addition to having the level of deregulation (REGL) significant at the 10 percent level, while the significance of the inflation variable (INFL) moves to 10 percent as well. The Euro2 results are more encouraging with only the cost savings (EXP) and revenue increase (REV) proxies not being significant at the 10 percent level. This indicates that the U.K. and Ireland, with their differing systems of law, exerted influence on the European results. It also establishes the idea that the U.K. and Ireland are considered to be outside of the European family of countries.
111 The poor results in respect to these tests can be attributed to the small sample size that existed, which in turn meant that the degrees of freedom are lower than we would ideally like. However, we felt it important to represent the European family of nations in separate tests, since the bancassurance concept was first founded and has seen some of its greatest success within this region.
2.7.4 Legal Systems
As was previously mentioned, bancassurance is a concept that first came to life in France.
While many countries have had success with bancassurance, thus far others have found it hard to maintain a flourishing bancassurance industry. Numerous factors undoubtedly contribute to this problem, and here we seek to determine whether the legal system within the country has any influence on the success of bancassurance. Generally, bancassurance markets have found it easier to expand within civil-law nations as opposed to common- law nations. In turn, we divided our sample into civil-law nations78 and common-law nations based on the La Porta, Lopez-de-Silanes, Shleifer and Vishy (1998) paper79. The results for the civil-law nations sample are presented in Table 2.8.
78 We do not distinguish between different types of civil-law nations; instead we combine nations from the French, German and Scandinavian civil-law systems. We believe this is appropriate since they all originate from the same Roman system of law. This was mitigated by the fact that our sample sizes would become too small should we incorporate distinctions between the civil-law families. 79 Appendix 1 shows the distribution of countries based on legal system.
112 Table 2.7: Euro1 and Euro2 Results
This table presents the pooled OLS and GMM results for the first and second European samples (Euro1 = mainland European nations plus the U.K., Ireland, and Turkey, amounting to 37 companies from 17 countries. Euro2 =Euro1 ex the U.K. and Ireland, amounting to 27 companies from 15 countries) over the five- year period of 1999-2003. For the GMM tests an additional two years of lags we also incorporated for each variable (i.e. years 1997 and 1998 act as lags). Euro1 Sample Euro2 Sample Ordinary Least Squares Generalized Method of Moments Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Estimate t-stat p-value Estimate t-stat p-value
Intercept 0.01304 0.25287 0.801 0.02823 0.58541 0.558 0.055105 0.44927 0.654 0.03672 0.28491 0.776 RISK -0.00105 -4.50457 0.000*** -0.00110 -3.98262 0.000*** -0.00054 -2.85506 0.005*** -0.00043 -2.95276 0.003*** SIZE 0.03013 3.35866 0.001*** 0.02997 3.24189 0.001*** 0.03094 1.45732 0.148 0.03745 1.68104 0.093* EXP -0.02145 -0.31276 0.755 -0.01722 -0.56081 0.575 0.04429 0.54880 0.584 0.01386 0.41847 0.676 REV 0.00002 0.19115 0.849 -0.00001 -0.25652 0.798 0.00015 1.06754 0.288 0.00005 0.89859 0.369 SIZE(N) 0.00512 4.11840 0.000*** 0.00464 4.03899 0.000*** 0.00460 3.41947 0.001*** 0.00437 3.17344 0.002*** REGL -0.00033 -0.25654 0.798 -0.00170 -1.88234 0.060* -0.00142 -0.99695 0.321 -0.00242 -2.04563 0.041** GNI 0.10293 0.67549 0.500 0.06209 1.18768 0.235 0.24377 1.35331 0.178 0.15214 1.68639 0.092*
INFL -0.00188 -2.11362 0.036** -0.00099 -1.82972 0.067* -0.00195 -1.92062 0.057* -0.00117 -1.90583 0.057*
Adjusted 0.16755 0.15985 0.15588 0.14826 R 2
Diagnostic P-Value Diagnostic P-Value Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J LM Het. Hansen 0.164 0.862 0.455 0.995 Test Test Test J Test JB Test 0.278 JB Test 0.568
* Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. RISK=Company Risk Proxy, SIZE=Company Size/Customer Base Proxy, EXP=Cost Decrease Proxy, REV=Revenue Increase Proxy, SZIE(N)=Size of National Banking Sector, REGL=Level of National Deregulation, GNI=Proxy for the Level of Insurance Demand Within a Country, INFL=Proxy for Inflationary Changes Within a Country.
113 Table 2.8: Civil-Law Nations Results
This table presents the pooled OLS and GMM results for the Civil-Law Nations sample of 29 companies from 18 countries over the five-year period of 1999-2003. For the GMM tests an additional two years of lags we also incorporated for each variable (i.e. years 1997 and 1998 act as lags).
Ordinary Least Squares Generalized Method of Moments
Variables Estimate t-stat p-value Estimate t-stat p-value
Intercept 0.00510 0.04293 0.966 -0.02064 -0.13635 0.892 RISK -0.00010 -4.59547 0.000*** -0.00010 -5.74971 0.000*** SIZE 0.04036 2.04741 0.043** 0.04260 1.62895 0.103*
EXP 0.05678 1.95872 0.052* 0.057582 2.57342 0.010*** REV 0.00022 1.53554 0.127 0.00020 2.57833 0.010*** SIZE(N) 0.00765 3.33890 0.001*** 0.00536 2.82456 0.005*** REGL -0.00141 -1.11657 0.266 -0.00110 -0.77255 0.440 GNI 0.45289 2.58509 0.011*** 0.28766 1.98727 0.047** INFL -0.00191 -1.88736 0.061* -0.00197 -2.62322 0.009***
Adjusted 0.18753 0.17914 R 2
Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.021** 0.834 Test Test JB Test 0.075*
* Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. RISK=Company Risk Proxy, SIZE=Company Size/Customer Base Proxy, EXP=Cost Decrease Proxy, REV=Revenue Increase Proxy, SIZE(N)=Size of National Banking Sector, REGL=Level of National Deregulation, GNI=Proxy for the Level of Insurance Demand Within a Country, INFL=Proxy for Inflationary Changes Within a Country.
114 The LM heteroscedasticity tests for the civil-law sample are significant at the 5 percent level, indicating that the problem of heteroscedasticity exists. The Jarque-Bera test, on the other hand, is insignificant at the 5 percent level. The signs of the significant variables are appropriate and the adjusted R 2 of the OLS test is 0.1875. All the variables except the revenue increase proxy (REV) and level of deregulation (REGL) are significant at the
10 percent level with company size proxy (SIZE) and revenue increase proxy (REV) significant at 5 percent; and the risk proxy (RISK), nations banking sector proxy
(SIZE(N)), and the demand for insurance (GNI) significant at the 1 percent level.
The GMM results have an insignificant Hansen’s J test value and an adjusted R 2 of
0.1791. The results concur with the OLS results just mentioned with only the level of deregulation (REGL) being insignificant. The company size proxy (SIZE) is significant at the 10 percent level, while demand for insurance (GNI) is significant at the 5 percent level and the remaining variables are all significant at the 1 percent level.
We have not reported the results for the set of common-law nations due to the fact that our sample size was too small to provide meaningful results. From the results that were obtained, it was revealed that common-law nations do not facilitate bancassurance demand. Ideally, further tests should be conducted in this area to provide conclusive results in regards to the relationship between differing systems of law and the demand for bancassurance.
115 2.7.5 Analysis and Implications of Results
The results obtained support the findings of previous researchers in regards to the determinants of bancassurance while also providing some new insight on a couple of key determinants.
As expected, the size of a nations banking sector is consistently significant and positive80.
This is consistent with the belief that banks find it easier to compete with negative sentiment towards bancassurance, and are able to make use of the strong customer loyalties when beginning a bancassurance operation in countries with established banking sectors.
The risk proxy is a consistently significant and negative determinant of bancassurance.
This means that as the level of bancassurance operations increase within the company, the company risk decreases. This supports the findings of researchers such as Boyd,
Graham and Hewitt (1993), Boyd and Graham (1988), Brewer (1989), and Smith,
Staikouras and Wood (2003) in regards to insurance operations acting as a stabilizer for bank earnings by providing diversification benefits.
The negative significance of the inflation proxy supports the idea that the negative impact an increase in inflation has on the consumer outweighs the positive impact it has on the company in terms of increasing premium income values and further discounting the
116 insurance liabilities. This supports researchers such as Browne and Kim (1993) and
Outreville (1996) who both found that inflation would be detrimental on savings through insurance products as it erodes the value of the product. Furthermore, it validates the notion that the subsequent reduction in disposable income will reduce the demand for insurance products, since they are not necessity items.
The results indicate that the income of the populace is significant and positively related to the size of and demand for bancassurance operations. As disposable income increases, luxury items such as insurance products become more affordable which leads to greater consumption. A larger income also increases the need for protection, since the costs regarding lost income in the case of an untimely death are now greater. This finding supports the results of Browne and Kim (1993), Campbell (1980), Lewis (1989) and
Outreville (1990 and 1996).
The size of the company and its customer base, as proxied by its total assets, proves significant and positively related to level of bancassurance operations within the company, as was expected. This is indicative of larger banks having a larger customer base to which they can market new products to, as is supported by Canals (1998). At the same time, we can argue that Amel et al. (2004) was correct in the belief that consolidation between commercial banks and insurance companies is only beneficial up to a certain size in order to reap economies of scale. We argue for this since the significance of our results is not that strong in certain cases, in particular when we
80 In the analysis of our findings we exclude the results of the Common-Law tests since they distort the overall results. We believe this is the result of a less than perfect dataset, biased by excessive representation
117 remove the U.S. sample that contains a large number of small sized firms. This suggests that the remaining larger firms do not reap as much benefits from the start of bancassurance operations as smaller firms do.
The level of deregulation has mixed results. The results of the whole sample indicate that deregulation is beneficial in terms of the amount of bancassurance within a company which supports the findings of the majority of researchers such as Bos and Kolari (2003),
Carow (2001), Holland et al. (1998), Lown et al. (2000), Morgan (2003), and Wepler et al. (2004). However, once the U.S. sample is removed, the significance disappears. This can be explained by pointing out that the large sample from the U.S. provides probably the best representation of the benefits for reductions in regulations, since the U.S. has only recently removed restrictions to the formation of large bancassurance operations.
The small sample size from other nations means that it is inherently more difficult to conclude whether benefits exist from deregulation, since the sample of companies we have may not be representative of those firms that have prospered from deregulation. The results for Europe can be explained by the fact that bancassurance operations have reached maturity in a number of key countries of the sample such as France, Belgium and the Netherlands. Hence, the subsequent impact of further reductions in regulations is moot. The negative significance indicates that the growth prospects for bancassurers are greater for those countries that have not reached maturity yet and can still gain substantial benefits from deregulations (such as Estonia, Turkey and the Czech Republic). It also
from one particular country.
118 symbolizes the failure of bancassurance in a number of countries that can be considered deregulated in relation to their peers, such as Germany and the U.K.81
The consensus amongst the two larger samples that were tested is that the revenue increase proxy is positively significant. This is to be expected and is in conjunction with the findings of researchers such as Agrawal M. (2002), Heymowski (2000), Klein (2001), and Wepler et al. (2004). This shows that the subsequent increases in fee-based revenues are a key determinant of the growing popularity of bancassurance operations. It also symbolizes the fact that significant revenue gains exist regardless of time.
The cost savings proxy is significant and positively related to the size of the bancassurance operation. This indicates that cost benefits are a key determinant of bancassurance, in line with the results of Cowan et al. (2002), Berberich (2000), Canals
(1998), and Webersinke (2000). By utilizing bancassurance operations, banks can share existing costs with the new operation while also being able to produce and sell cost- effective products.
Finally, we can also provide weak evidence that the legal system of a country significantly impacts the determinants of bancassurance. We say weakly because we feel that the tests of common-law countries are not conclusive enough due to a poor sample
81 Even with considerable deregulations the U.K. has not shown too much success for bancassurers with their market share falling in recent years even with the levels of deregulations increasing. A simple case exists in Germany. In countries such as Estonia, Turkey and the Czech Republic, bancassurance is flourishing even though they are still considerably regulated and further deregulations are possible.
119 size82. However, we feel confident in the results of civil-law countries, since they complement the results from the whole sample. From our results we can say that civil- law countries have more successful bancassurance operations, since they exhibit significance for all the determinants of bancassurance that were tested for in the expected way (except for the level of deregulation which is not significant). Hence, we argue that banks in these countries gain more from starting bancassurance operations than those in common-law countries. Of course, further research into this area should be conducted to obtain more concrete results.
2.8 CONCLUSION
2.8.1 General Conclusion
Ever since the founding of the bancassurance concept, it has received much attention amongst both researchers and policymakers, since it is a major step towards the creation of universal financial markets that are no longer segregated based on industry operations.
This attention has been intensified over the past decade with the emergence of new players within the industry as a result of changing legislations and regulations, increased competition from both home and abroad, and the ever-intensifying globalization of industry. Bancassurance serves as another medium for trade in insurance services and has helped to transfer expertise and knowledge between industries, increase product variety
82 This being the reason we neglected to include the results of the common-law sample.
120 and product innovation, and improve the efficiency of insurance markets. While previous studies have looked at the benefits and impacts that bancassurance has on the company and in turn focused on a few key determinants such as diversification of risk and wealth effects, none have paid attention to such a wide range of important bancassurance determinants as we do in this study.
This study aims to fill a gap in existing literature by investigating a number of key determinants of bancassurance using a sample of both developed and underdeveloped countries from around the globe. This study also aims to provide an in-depth analysis of previous literature, both quantitative and descriptive, in order to give a thorough review of the bancassurance market to date.
This study examines the determinants of bancassurance across 28 countries. Our results complement the existing literature on bancassurance demand, insurance demand, and international trade in insurance services, while also providing additional insight in key areas. We use a cross-sectional time-series approach by making use of panel analysis. In particular we incorporate the use of the GMM estimation technique which overcomes the inherent weaknesses found in the simple OLS approach. That is to say, the GMM estimator exploits the time-series variation in data, allowing for the inclusion of lagged variables as regressors, and controls for endogeneity of all explanatory variables. For
GMM purposes, we made use of internal instrumental variables only.
121 The results obtained in this study provide several contributions to our understanding of the determinants of bancassurance and confirm the findings of some of the previous literature in this area over which debate has since ensued. In this study, eight hypotheses have been established. It is found that the union of banking and insurance operations serves as a way of diversifying risk for the parent bank, since the two operations have considerably different risk structures. Thus, a negative relationship exists between our dependant (premium income) and the risk proxy. This also highlights the fact that diversification benefits outweigh additional risks bought about through the integration of insurance operations (such as loss of customers and inefficient allocation of resources).
Insurance demand as proxied by GNI per capita is found to be a positive determinant, since luxury items such as insurance products become more affordable, and the need for protection against premature death of the wage earner becomes higher with added income. The inflation rate is shown to be a negative determinant, supporting the argument that the subsequent reduction in disposable income outweighs company gains made through increased insurance premiums and discounts of insurance liabilities. The level of financial deregulation, on average, is found to be a positive determinant of bancassurance, highlighting the fact that more regulated nations around the world have weaker bancassurance industries. It was also found that in Europe, unlike the global results that we obtained, more regulated countries such as Estonia and Turkey offer lucrative prospects to bancassurers as opposed to the mature, deregulated markets of countries such as France. The size of the company, which indicates the breadth of the company’s customer base, proves to be a positive determinant, demonstrating that firms make use of their size and that it is an important consideration when first beginning
122 bancassurance operations. In a similar manner, the size of the national banking industry is also a positive determinant, since it allows banks that want to expand into different industries, such as insurance, to overcome any negative sentiments that may exist towards such moves. Both the decrease in costs, through spreading of costs amongst different operations, and increase in revenues, from the new fee-based income streams, resulting from the implementation of insurance operations within the bank, are found to have positive implications in regards to banking organizations beginning bancassurance operations. Finally, it is also found that the legal system existing within the host country will be influential on the overall success of the bancassurance industry, since it impacts directly on the determinants of bancassurance.
The results from this study lead to several policy implications for regulators and policymakers alike. Promoting the further deregulation of the financial services industries will enhance the development and growth of both the banking and insurance industries through allowing the augmentation of distributional channels such as bancassurance that incorporate the knowledge and expertise of both industries. Through the greater efficiency of product distribution and the increased innovation utilized in product development, consumer welfare will be improved by satisfying the consumers’ need for greater variety in products offered, which come about as a result of differentiated tastes.
Companies that fully utilize the changing atmosphere towards the provision of differing financial products and services are set to reap the benefits in terms of diversified risk, reduced costs, and increased revenues, enabling them to be more competitive on both a domestic and global front than their counterparts that fail to establish themselves in this
123 new style of industry (namely the provision of multiple financial services from across industries).
2.8.2 Avenues for Further Study
While this study does verify previous topics of discussion and provides further insight into bancassurance determinants and the overall bancassurance markets than previous studies have, it also opens further avenues of research. Throughout the study, we have mentioned where additional research could be focused. Here, we will revise the main areas in which future research can be coordinated.
While we have looked at a number of key determinants of bancassurance we have also been unable to examine a few due to data restrictions based on our sample. In order to expand this study, possible determinants of bancassurance such as social security expenditure of governments, level of education, taxation, and the life expectancy of the populace should be examined. Exchange rates could also be incorporated as an explanatory or even instrumental variable in order to account for the slight measurement error they pose to such a study as ours (outlined in Section 2.6.1). Similarly, an expansion of the dataset to incorporate key countries which this study has omitted for reasons previously specified, would provide a clearer insight into the global bancassurance industry. Further research should also be done into the determinants of bancassurance based on legal systems, and developing bancassurance economies versus those that have already established maturity. This study focuses mainly on what determines the demand
124 for bancassurance around the world, studies focusing on the supply side, which this study does not encompass, would also be able to provide useful research avenues. Another logical step would be to segment the bancassurance industry and conduct separate analysis on life and non-life bancassurance products, in contrast to this study that makes no distinction between the two. Finally, a natural extension of this study would be to examine the determinants of assurebanking and determine which is a more viable union of the banking and insurance industries.
125 CHAPTER 3
LAW AND THE DETERMINANTS OF LIFE INSURANCE IN OECD
COUNTRIES
"Human life is proverbially uncertain; few things are more certain than the solvency of a life-insurance company."
- Sir Arthur Stanley Eddington (1882-1944). British astronomer & physicist,
director of Cambridge observatory. From J. R. Newman (ed.) The World of
Mathematics, New York: Simon and Schuster, 1956.
3.1 INTRODUCTION
Insurance, with life insurance in particular, is an increasingly important factor within the financial sector and community on a whole, providing ways for the populace to alleviate uncertainty brought about by risks inherent in everyday activities and also providing a long-term savings mechanism that can be catered to the individuals’ particular needs. It also facilitates economic growth by acting as financial intermediaries between investors and economic agents and assuaging pressure on governments resulting from exhausted
126 and ineffective pension schemes. Over the ten years from 1993-2003, the life insurance industry has maintained a 4.3% growth rate per annum within the OECD with life insurance premiums amounting to 57% of total insurance premiums in 200383. Life insurance penetration rates (ratio of premium volume to GDP) within the OECD have averaged 4.1% over the five-year period of 1999-200384. The increased importance of life insurance as a provider of financial services and investment funds on capital markets is evident throughout the OECD, and differing nations within the OECD experience distinct levels of life insurance consumption. Given this variation within consumption levels, the question of the causes of this diversity and therefore the determinants of life insurance consumption arises.
In spite of the obvious importance that life insurance offers in the way of risk management, savings facilitation, and providing term finance, we do not have a clear understanding as to the factors that drive the demand and supply of life insurance across countries and over time. A number of previous researchers have proposed a variety of different macroeconomic, demographic, social and psychographic factors as possible determinants of life insurance consumption. Sample sizes, availability of data, and the geographic areas of study have constrained the testing of theoretical hypotheses and ensured that results vary amongst the differing studies.
This study improves on existing literature regarding the determinants of life insurance consumption in several ways. Foremost, this study reconciles and adds to the existing
83 Source: Swiss Re. SIGMA World Insurance publications, various issues.
127 literature in the field of life insurance demand with studies conducted in the field of legal systems and corporate governance, and in turn provides a more complete picture of the determinants of life insurance demand than previous studies accomplish. Hitherto, the impact of legal systems, corporate governance and the level of investor protection on the demand for life insurance products have been completely overlooked, and this study aims to assert the importance of these factors on consumer consumption levels. This study also makes methodological improvements on numerous previous studies in the field of life insurance by using the advanced estimation technique known as GMM, which enables us to overcome potential data limitations such as measurement errors, heteroscedasticity, endogeneity and multicollinearity.
The results obtained within this study provide support to a number of existing theories while at the same time also affording new insight into certain key areas. The most notable addition this study makes is to prove that civil-law nations, in particular French and
German civil-law, exert a positive influence on the level of life insurance consumption, whereas English common-law nations are found to be insignificant when it comes to promoting life insurance demand. Additionally, it is discovered that the level of investment restrictions imposed by regulators and the level of investor protection afforded are significant determinants of consumption.
The results presented within this study are expected to assist policymakers and regulators in their decision-making processes if they endeavor to ensure a viable and progressive
84 4.48% in 2003. Source: Swiss Re. SIGMA World Insurance publications, various issues; and the OECD Factbook 2005.
128 life insurance industry and economy on a whole. The results should shed additional light on economic, institutional and demographic determinants that drive the levels of life insurance consumption, allowing for effective policy decisions to be promoted that develop the industry and benefit the whole population.
3.1.1 Chapter Outline
This chapter is structured as follows: Section 3.2 provides an outline of the life insurance industry, paying attention to advantages it offers to the community and economy as well as focusing on the challenges the industry faces in current times. An analysis of the major legal systems is also performed, encompassing their history and major differences.
Finally, this section also provides an in-depth literature review in regards to works conducted in systems of law, life insurance demand and consumption, and legal systems and the determinants of insurance demand. Section 3.3 outlines the data and methodology used within the study. This includes looking at empirical hypothesis that are being tested, outlining model specification tests that are used, reviewing the actual model employed in the study, and discussing the testing procedures utilized. Section 3.4 provides the empirical results for all the conducted tests as well as an analysis of said results and their corresponding implications. Finally, Section 3.5 concludes with the main findings and resulting policy implications of this study, while also providing suggestions for further research into this area of study.
129 3.2 THE LIFE INSURANCE INDUSTRY AND SYSTEMS OF LAW
3.2.1 Life Insurance in Current Times85
Overview:
The role of life insurers is becoming increasingly important in the financial services industry around the world. Current world life insurance premiums amounted to
$1849USD billion in 2004 (57% of total world insurance premiums) up from $1412USD billion in 1999 (60.8% of total world insurance premiums), with the industrialized countries accounting for over 88% of global life insurance premiums in 200486. In
Europe, positive growth in the life insurance industry, fostered in numerous countries by tax and pension reforms, has continued with the U.K. recording a 3.6% growth over the
2003-2004 period and France recording growth of 10.6% over the same period. Oceania also recorded a strong 6.3% growth (2003-2004 period) following a double-digit plunge in the Australian markets in the previous reporting period. In contrast, the world’s two largest markets in terms of life insurance consumption, namely the U.S. and Japan, recorded dismal growth over the same period, with the U.S. posting a 0.1% growth while
Japan recorded a loss of 1%.
85 This information in this section is based on figures provided by the SIGMA World Insurance in 2004 and World Insurance in 1999 publications, and the information within the OECD Secretariats (2000) article on Developing Life Insurance in Economies in Transition. . 86 Countries of North America, Western Europe, Japan and Oceania.
130 With significant foreign direct investment inflows, low labor costs and attractive GDP forecasts coupled with changes in taxation and pension systems, emerging markets are becoming increasingly important players in the global life insurance industry, with the growth in many of these areas far outstripping that of their industrialized counterparts. In
South and East Asia, growth amounted to 9.8% over the 2003-2004 period, fueled by areas such as India (10.5%), Taiwan (17.6%) and Hong Kong (30.8%). Latin America also recorded strong growth of 17.1% in light of the growth in areas such as Chile
(9.9%), Brazil (16%), Mexico (21.7%) and Argentina (40.9%). In contrast, life insurance premiums in Central and Eastern Europe fell by 11.2% in 2004, following a rise of 17.7% in 2003, as a result of 38.4% decrease in real terms within Russia due to the drop in short- term policies aimed at tax optimization. Other nations within the region posted a 10% growth, lead by Baltic and South-Eastern European countries where life insurance premiums increased by about 21%.
Advantages Offered:
The growing popularity of life insurance stems in large part from the numerous advantages it offers to the community and economy on a whole. One of the most noted uses of life insurance is as a form of protection against adverse financial consequences associated with an untimely death, in turn alleviating worry and distress and increasing initiative. This is something that no other financial instrument can perform (OECD
Secretariat (2000)). Another use of life insurance can be as a vehicle for saving purposes, especially over the long-term, even for those people who do not save on a regular basis,
131 and in a similar manner they provide a means by which individuals can make financial provisions for retirement. In certain countries (majority of the OECD), life insurance also offers tax concessions for the purchase, ownership or execution of policies (Swiss Re.
(1998a)).
According to the OECD Secretariat (2000), life insurance is beneficial to business in the sense that it can minimize the disruption brought about by the death of key employees, and it can also permit more favorable credit terms to borrowers and can decrease the risk of default. To the industry, foreign life insurers are advantageous since they bring with them their extensive experience from abroad that include such factors as managerial experience, differing products and more efficient production and distribution methods, as well as encouraging competition within the local industry.
Economy-wise private life insurance can supplement, if not substitute for, benefits provided by governments. The trend of rising life premiums in OECD countries can, in large part, be attributed to the mounting pressure that existing government pension schemes are under and the alternate investment strategy that life insurance products offer
(Dickinson (2000), OECD Secretariat (2000) and Swiss Re. (1998a)). The existence of life insurers also enables government to concentrate on other key policy areas such as core social protection benefits, while individuals can decide for themselves the degree of additional protection they require for later on in life. Another key benefit afforded is that life insurance can promote economic development in general, and the development of financial markets in particular. Insurance companies can amass funds, which are long-
132 term in nature that become important in supporting investment and the national economy, thereby serving as financial intermediaries between investors and economic agents87.
Challenges:
As with any other financial service provider, life insurers face numerous difficulties that have to be surmounted in order to ensure the success of their business. High inflation rates are of great concern especially in emerging economies for two major reasons (Swiss
Re. (1998a)). Firstly, the level of consumer disposable income would be diminished, and since insurance products are not seen as a necessity, their consumption will also be adversely affected. Secondly, policyholders run the risk of not receiving what they expect since the expression of guarantees is generally outlined in nominal amounts, which would give rise to mistrust within the community in regards to insurers.
Consumers lack of familiarity with insurance products is also a problem that life insurers need to overcome. Individuals unfamiliar with the benefits of possessing life insurance may seek other methods of protecting themselves or saving for the future. In response to this, considerable outlays are required by life insurers at the outset in the form of advertisements and community education programs (Dickinson (2000)). In the end, the products, services and distribution channels need to be aligned with the local markets needs. The costs involved, however, may not warrant such action (Swiss Re. (1999)).
87 In the OECD nations, insurance companies are the largest institutional investors (OECD Secretariat (2000)).
133 The domestic markets may also pose numerous setbacks to life insurers, especially in emerging economies. One such problem relates to the lack of reliable actuarial data in emerging economies on which life insurers will be able to base their calculations on.
Insufficiencies in this regard will cause life insurers to substantially increase their margins of error to ensure that pricing does not cause a deficit. The subsequent increase in premiums, however, will be viewed negatively by the consumer (OECD Secretariat
(2000)). A second setback relates to the lack of experience that domestic life insurers in emerging economies have in regards to estimating risk, providing a diverse set of products and correctly pricing policies. This is something that can easily be overcome by allowing foreign participants into the industry. Finally, should domestic financial markets prove to be underdeveloped, then insurers may find it difficult to conduct business confidently and effectively in regards to investing their assets, which can hinder the development of extensive insurance operations and adversely affect the domestic economy.
Required Regulatory Environment:
The regulatory framework with a country is absolutely crucial to the development and ongoing success of the life insurance industry. Governments must ensure that the rules and regulations are available to all parties concerned, i.e. the consumer and insurer, regardless of whether they are domestic or foreign. Consumers must be protected by supervising the solvency of the life insurance companies, since an insurance contract is only an obligation to pay a certain amount in the future. With such transactions, a
134 considerable amount of trust is involved on behalf of the consumer that the contract will in fact be honored. This is particularly true for life insurance where the amounts of money involved are large and the commitment spans a long period of time. Should an insurance company become insolvent and the contracts not honored, detrimental consequences can result for the whole insurance industry due to a plunge in consumer confidence levels (Dickinson (2000)). Regulators must also ensure that the industry remains competitive enough to ensure fair prices prevail, that will in turn further facilitate consumer demand. Similarly, regulators must ensure the transparency of the industry in regards to reporting and contract terms. Should policy features be too complicated for the consumer to understand, they will not conduct business with the insurer. In this regard, regulations must also be stable to guarantee that both the insurer and consumer may enter into contract knowledgeably and with confidence that the contract terms will not fluctuate.
3.2.3 Systems of Law (a Study of Legal Families)
It is commonly agreed by legal scholars that while no two nations systems of law are exactly alike, there are sufficient similarities in certain respects to allow for classifications of the legal systems into a number of major families of law.
La Porta et al. (1998) came out with probably the most well-known article regarding legal systems. They collected and summarized information on the legal systems of 49 nations around the world, and based on the works of legal scholars, La Porta et al. (1998) placed
135 these 49 nations into four major legal families, these being English, French, German, or
Scandinavian. Other classifications could have been made that would include law based on religious traditions, but these were argued by La Porta et al. (1998) to be less relevant in the matter of investor protection. In turn, religious traditions such as Jewish Law
(Halakha), Canon Law, Hindu Law, and Muslim Law (Sharia) were grouped into the four major families of law (as previously listed) based on which colonial power subjugated the nation in the past88. For example, even though India belongs to Hindu Law, it is grouped with the English family of law, since it is argued that the economic law of this country is heavily impressed by the legal thinking of the colonial power, which was Great Britain in this case. Arab nations are similarly grouped with the French system of law, since the
French were the major colonial power89.
The system of law within a country and how it came about is a complex concept to comprehend, and we feel it is important to understand the origins of each legal system before we conduct any study focusing on the impact of said systems of law on financial markets90.
88 Jewish Law is followed by Conservative and Orthodox Jews; Canon Law is followed by Catholics and Anglicans with a similar system used in the Eastern Orthodox Church; Hindu Law is followed largely by Indians; and Muslim Law which is derived from the Qur’an is predominantly used in Middle Eastern Nations. 89 Questions arise as to the validity of grouping Muslim nations with those nations that follow the French system of law. This is especially true when conducting a study on insurance markets since the Muslim faith does not view insurance products favorably, arguing that it is immoral since it indicates a lack of faith in the protective care of God. 90 The following is based on information collated from a combination of sources which include Microsoft Encarta Online Encyclopedia 2005, Wikipedia Free Online Encyclopedia, Collier’s Encyclopedia, and AbsoluteAstronomy.com Encyclopedia.
136 Common-Law
Common-law systems of law constitute the majority of the Commonwealth91 and other nations where the British Empire once held sway such as the U.S. and Hong Kong. In many of these countries, the common-law system has been intermixed with other legal and religious systems, with the two clear examples being Scotland, which mixes outdated versions of Roman civil-law and common-law, and India, which unites common-law with
Hindu law.
The origins of the common-law system can be traced back to Henry II in 1154 who was responsible for institutionalizing common-law by creating a unified system of law
‘common’ to the country that incorporated local customs, ending peculiarities and arbitrary remedies92, and reinstating a jury system sworn to investigate criminal accusations and civil claims. By the 15th century a system of equity had been developed, which often came in conflict with the system of law, through which claimants would demand financial retribution. It was not until the Judicature act of 1873 that the courts of equity and law were combined.
91 Exceptions include Louisiana and Quebec which follow French civil-law, and South Africa which follows Roman Dutch law.
137 Civil-Law
Civil-law systems are far more widespread than the common-law systems and form the basis of law in continental Europe as well as Japan, South Korea, Latin America, and other former colonies of continental European countries.
Civil-law stems from the laws of ancient Rome. Emperor Justinian I who managed to temporarily regain control of parts of the Roman Empire that had been lost to Germanic tribes set out to restore Roman law through the combination of legislation and the views of prominent legal scholars by commissioning the production of three great law codes.
→ Codex Lustinianus (the Code): Published in 529AD and was a collection of
imperial laws from the time of Hadrian (117AD) to Justinian.
→ Digesta (the Digest): Published in 533AD, it was a collection of excerpts from
classical jurists and was given the force of law.
→ Intitutiones (the Institutes): Published in 533AD and was basically a textbook for
beginners, which was also given the force of law.
This comprehensive code embodying the accumulated wisdom and experience of many generations of Roman jurists was called the Corpus Juris Civilis (Body of Civil-Law, or, the Justinian Code).
92 Previously, if consensus could not be reached, then one form of test to prove innocence that could be administered was to have the defendant carry a red-hot iron; should the wounds then heal in a pre-specified
138 Justinian’s code had virtually no impact in Western Europe from its conception to the
10th century, which is characterized by a steep cultural decline in the area. It was not until the intellectual reawakening in the latter 11th century that the Corpus Juris Civilis was rediscovered in Western Europe. The law professors at the newly founded university of
Bologna based their teachings on the Corpus Juris, with other European universities following suit. Only England did not partake in this favorable reception for the newly rediscovered Roman law, predominantly due to the fact that the English legal system was more developed than their continental counterparts by the time the Roman system of law was revived. By the 16th century a mixture of the old Roman law combined with elements of cannon law and feudal law dominated continental Europe. This was known as Jus
Commune; and when its popularity began to wane in the 17th and 18th centuries, in the light of reason, modern forms of civil-law systems were constructed through a systematic and comprehensive codification process.
German Civil-Law
The first attempts at codification were made in the latter 18th century when the German states of Prussia, Bavaria and Saxony began to codify their laws. Meanwhile in Austria, codification was begun in 1753 and the Austrian Civil Code was completed by 1811.
Unlike Austria, France, and the Scandinavian countries, who had discovered their systems of law long before, it was not until the unification of Germany under Bismarck in
1871 that the German system as we know it today came into realization. It took a further twenty years to complete the code, which was eventually called the Burgerliches
period of time, innocence was proven, and if not then execution would usually follow.
139 Gesetzbuch. According to Levine (1998), the German code was unprecedented in terms of consistency, technical precision, and detail. As a result, this type of law exerted influence over the systems of law in place within its neighbors of Austria and
Switzerland, as well as countries such as Hungary, Czechoslovakia, Greece, Italy,
Yugoslavia, Japan and Korea.
French Civil-Law
The most influential codification effort was the enactment of the five basic codes of
France during the Napoleonic period93. The French Civil Code was enacted in 1804 after only a few years of preparation and was truly a child of the French Revolution. It is concise, accessible, and relies on the Jus Commune. Napoleon ensured that this system of law was grounded in every territory that was conquered, which included parts of Europe, the Near East, Northern Africa, Indochina, Oceania, French Guinea, and the French
Caribbean Islands94. The French system was also spread to Central and Southern America thanks to the large influence it had on the Portuguese and Spanish legal systems.
Scandinavian Civil-Law
From the name, it is obvious that this system of law is only prominent in Scandinavian countries. This system was established at the start of the codification era in the late 17th
93 The five codes included: The civil code (Code Napoleon of 1804), the code of civil procedure (1807), the commercial code (1808), the penal code (1811), and the code of criminal procedure (1811). 94 La Porta et al. (1998).
140 and early 18th centuries, though large amounts of the Scandinavian systems were left uncodified.
In current days, Roman law is no longer applied in legal practice, yet it still exerts considerable influence, since no legal code completely broke with the Roman tradition.
Instead, Roman law is fitted into more coherent systems.
Comparisons
One of the major differences between common-law and civil-law lies in their origins.
Civil-law developed out of old Roman law proceeding from broad legal principles and the interpretation of doctrinal writings, whereas common-law was developed by custom, beginning before there were any written laws in the area, and involves the application of facts to legal fictions. In functionality, the difference is that in civil-law countries, legislation is seen as the primary source of law, and courts base their judgments on the provision of codes and statues, whereas in common-law countries, cases are the primary source of law where statutes are only interpreted narrowly. This basically means that common-law societies are ruled by court decision while civil-law societies are ruled by written code. Recent articles into the impact of legal systems on financial markets have yielded mixed results regarding which system of law is most beneficial to financial markets. Some, such as La Porta et al (1998), argue that common-law countries tend to afford a higher level of protection of creditors rights than civil-law countries do, with the
French civil-law system offering the lowest level of protection. Others such as Koch
141 (2003) and Levine (1997), however, argue the opposite to this. The following section highlights the major work by researchers, not only in the field of life insurance, but also in relation to the effects that legal systems have on financial markets.
3.2.4 Quantitative Works of Major Researchers
The aim of this chapter is to establish the determinants of life insurance consumption using a sample of OECD countries, paying particular attention to the way that the differing legal systems affect consumption. In order to do this, a review of existing quantitative literature in the fields of life insurance and legal systems will help us better understand the work already done, and, more specifically, the determinants that are relevant to this study.
Research examining the impact of legal systems on the functionality of the insurance industries has been very limited to date. To our knowledge, the papers by Browne et al.
(2000) and Esho et al. (2004) are the only ones with a specific focus on how differing systems of law affect the demand/consumption for insurance products. Both these papers focus their attention on the non-life insurance industry and their results diverge considerably. Browne et al. (2000) analyze the determinants of property-liability insurance consumption in OECD nations for the period of 1987-1993 by conducting least squares regression tests on both the motor vehicle and general liability industries. They find a significant positive relationship existing between the common-law dummy variable and the level of consumption in both tests. Other key findings include income being
142 positively correlated with the purchase of both lines of insurance and the market share for foreign insurers being statistically negatively related to motor vehicle insurance consumption. Meanwhile, Esho et al. (2004) examine the importance of legal rights and enforcement in influencing property-casualty insurance consumption through a series of
GMM tests conducted on a set of unbalanced panel data from the years of 1984-1998 covering 44 developed and developing countries. The major determinants that are examined include legal origin and property rights, economic development, risk aversion determined by the level of education and the level of uncertainty within the society, the price of insurance, and the probability of loss based on the level of theft and urbanization within the community. Esho et al. (2004) separate the systems of law into three categories which include English common-law, French civil-law and German civil-law, and unlike
Browne et al. (2000) no significance is found in relation to the systems of law and the level of consumption.
In contrast to the limited research performed on how legal systems influence the insurance industries, there has been extensive research conducted into legal systems and their influence on the overall financial services industry, with a particular focus on the banking industry in certain cases. The most notable work in this area is provided by La
Porta et al. (1998) entitled Law and Finance in which the effects of legal systems are examined in relation to how well they afford protection to creditors and shareholders. Of the 49 countries examined, it was found that those nations with common-law systems generally had the strongest legal protections in force, with the French civil-law nations having the weakest. The other major finding of this paper was that ownership of shares is
143 concentrated in those nations that afford low levels of investor protection. La Porta,
Lopez-De-Silanes, Shleifer and Vishny (2000) support these findings and go further by arguing that strong investor protection is associated with broadening financial markets and dispersed ownership of shares. More importantly, they argue that financial markets require a certain degree of intervention in the form of protection of outside investors in order to encourage the markets growth. In an earlier study, La Porta Lopez-De-Silanes,
Shleifer and Vishny (1997) similarly found that French civil-law nations offer the weakest form of investor protection, but more importantly they concluded that countries with weaker investor protection tend to have smaller capital markets (both debt and equity markets).
Levine (1997) looked at the impact legal systems have on the overall economy and found that the system of law is positively associated with economic growth. The main finding of the paper asserted that financial intermediaries are better developed in countries with legal systems that ensure the protection of shareholder and creditor rights through requiring accurate and comprehensive financial reporting and enforcement of contracts, with this being true for German civil-law nations in particular. This is supported by
Levine (1998) who finds identical results when conducting GMM tests on the banking industry over the period of 1976-1993, and Beck et al. (2000) who determine that GDP growth is helped by the positive impact that financial intermediaries exert in enhancing total factor production growth through better resource allocation. Levine, Loayza and
Beck (2000) conduct a series of tests over the period 1961-1995 utilizing 74 countries and in addition to finding that developed financial intermediaries facilitate economic
144 growth, they also conclude that cross-country differences in legal and accounting systems help account for differences in financial development.
According to Outreville (1996), nearly all theoretical work on the demand for life insurance products identify Yaari (1965) as the genesis. Lewis (1989) extends the seminal paper of Yaari (1965) that first theorized the uncertain lifetime life cycle model so that the level of life insurance consumption is based on the beneficiaries expected lifetime utility as opposed to the actual consumer (as proposed by Yaari (1965)). Yaari
(1965) found that wealth levels, income streams, interest rates and prices all influence the demand function derived from the maximization of the utility function of the consumer.
Lewis (1989), though, determines that the demand for life insurance will be a function of several factors that include the wage earner’s probability of death, level of risk aversion, policy loading factor, size of the bequest, and the present value of the survivors’ consumption stream. Hakansson (1969) also extends Yaari’s (1965) work, suggesting that the level of demand is also a function of wealth, income, interest rates, premium rates, and the discount function for current consumption. Zietz (2003) provides a very in-depth summary of existing literature in the field of life insurance demand and identifies 26 related academic empirical studies.
Headen and Lee (1974) propose that the demand determinants of life insurance can be separated into three categories. The first being variables that stimulate demand as a result of the marketing effort of the insurer in terms of advertising, size of the sales force, and the introduction of new products and policies that cater to the needs of specific consumer
145 segments. The second set of determinants involve variables that influence the size of the potential market and the ability to buy, which includes the level of disposable income and population-related factors such as birth rates, death rates, dependency rates, marriages and life expectancy. The final set of demand determinants involve variables that affect household decisions to save and accumulate financial assets along with variables that determine the composition of those assets, such as education and risk aversion, net household savings rates, short and long term yield rates, consumer expectations concerning future economic conditions, and the flow of funds into alternative financial assets.
Empirical research in the area of life insurance demand can be classified into cross- country and individual country studies, with the cross-country studies being more relevant to this study. Browne and Kim (1993) investigate the determinants of life insurance demand over a range of 45 countries for the years of 1980 and 1987. They find that the level of national income, inflation, dependency ratio, education, religion, price of insurance, and the level of government social security expenditure have significant impact on the level of life insurance consumption, with the average life expectancy proving to be insignificant. In a later study by Outreville (1996), a set of 48 developing countries from 1986 were examined via OLS regressions in order to investigate the relationship between life insurance premiums and financial development and other socio- economic variables. The paper provides empirical evidence of the negative effect of a monopolistic market on life insurance growth, and concludes that for developing countries, life insurance markets are significantly related to the market structure of the
146 country, disposable income, anticipated inflation rates and the level of financial development. In a more recent article by Beck and Webb (2003), demand factors are examined using both a cross-sectional dataset spanning 63 countries and a panel dataset over 23 countries, where it is highlighted that life insurance demand is low in developing nations and variations exist even amongst developed nations.
In addition to cross-country studies, there are numerous researchers that examine the determinants of life insurance demand within one or two particular countries. Hammond,
Houston and Melander (1967) examine the relationship between life premiums and income, net worth holdings, life cycle, and the occupation of the wage earner across the
U.S. Headen and Lee (1974) examine the short-term behavior of life insurance demand in response to changes in financial market conditions as well as the demand for alternative financial assets. Meanwhile, Truett and Truett (1990) examine the factors affecting life insurance demand in the U.S. and Mexico, while Chen, Wong and Lee (2001) examine the U.S. and discover that baby boomers tend to purchase less life insurance than earlier counterparts which led to the decline of recent life insurance consumption within the U.S.
More recent studies by Hwang and Gao (2003), Lim and Haberman (2004, Oct 2004),
Lenten and Rulli (2005) and Okura and Kasuga (2005), have examined the influence of economic variables on life insurance consumption within China, Malaysia, Australia and
Japan respectively. Hwang and Gao (2003) determine that the factors contributing to the growth of the life insurance industry in China can be associated with the 1978 reforms95
95 Factors that include higher incomes, improved education levels and better economic security.
147 96. Lim and Haberman (2004) unsuccessfully attempt to link the 1998 economic downturn to a decline in the performance of the Malaysian life insurance sector, while
Lim and Haberman (Oct 2004) find that savings deposit rates and price change in insurance are two important macroeconomic variables in relation to the demand for life insurance within Malaysia97. Lenten and Rulli (2005) utilize a structural time series model in order to analyze the life insurance demand in Australia over the period of 1981-
2003. They find evidence to suggest that demand is influenced by such things as deregulations and reforms, price levels, income, unemployment, and population variables, and that life insurance demand has a deterministic seasonal component98.
Okura and Kasuga (2005) find that in Japan, income, amount of financial assets already held, children, pensions and existing knowledge all influence the decision as to whether to purchase life insurance products99.
96 Jappelli and Pistaferri (2001) found that similar reforms in the Italian taxation system did not alter the consumers decision to invest in life insurance or how much to invest. 97 However, the savings deposits rate does not have the hypothesized negative sign. 98 There appears to be a surge in the first quarter of each year and a subsequent downturn in the fourth quarter. 99 Tachibanaki and Shimono (1994) and Urata et al. (1999) have also found similar results in relation to the factors that influence the demand for life insurance products within Japan.
148 3.3 DATA AND METHODOLOGY
3.3.1 Data Sources100
This study employs cross-sectional data for the eight-year period of 1996-2003 encompassing the 30 countries that constitute the OECD. The countries in our sample include:
Table 3.1: Sample Countries101
Australia Finland Ireland Netherlands Spain Austria France Italy New Zealand Sweden Belgium Germany Japan Norway Switzerland Canada Greece Korea (South) Poland Turkey Czech Republic Hungary Luxembourg Portugal U.K. Denmark Iceland Mexico Slovak Rep. U.S.A.
Data for per capita life insurance consumption was obtained from various issues of the
SIGMA World Insurance publication102. This measure is symbolic of the demand for life insurance within a country. However, it must be noted that premiums may vary significantly from one country to another due to differences in the types of policies, costs of writing life insurance, regulations, and the overall competitiveness of the insurance market. For these reasons, life insurance premiums are not an accurate measure of
100 For a summarized description of the sources for all the variables being used as well as a description of each variable please refer to Appendix 2. 101 An extended form of this table is presented in Appendix 2 depicting the differing legal systems of each country. 102 The actual description according to SIGMA is ‘Insurance Density – Premiums per capita (life)’.
149 insurance consumption and coverage103. An alternative measure, as used by Esho et al.
(2004) is life insurance in force, which is the face value of life insurance policies in effect in each country. However, this measure is difficult to obtain, and current databases are far from complete104. In order to have valid comparisons to previous literature and due to the fact that premium data is so readily available for most countries, we will use life insurance premiums in order to proxy life insurance consumption.
Foreign market share and foreign direct investment figures were both obtained from various issues of the OECDs Insurance Statistics Yearbook. Foreign direct investment figures are then converted into USD to allow for meaningful comparisons to be made between countries. The exchange rate figures used for conversion purposes were obtained from the IMFs International Financial Statistics website.
Interest rate, inflation rate, per capita gross domestic product, and life expectancy figures were all obtained from the OECD Factbook. For the purposes of interest rates we also made use of data contained in the ECB Convergence Reports in regards to Hungary and
Poland.
103 Quantity wise, the same amount of insurance sold in one country may not have the same value as that of another country. Hence, even though the amount consumed is identical, we would not see it that way since the monetary value between the two countries is different for the same amount of insurance consumed. 104 The American Council of Life Insurance (ACLI) has such data, however there are numerous errors that exist within the figures, which may include: annuities being included, insurance issued by foreign insurers not being included and insurance issued by government agencies being included.
150 Dependency ratio figures were found in the OECDs Labor Force publications.
Enrollment in secondary education was attained from the World Banks website. And
Hofstede’s uncertainty avoidance index was found on the ITIM International website.
We made use of La Porta, Lopez-de-Silanes, Shleifer and Vishny’s 1998 article in the
Journal of Political Economy entitled Law and Finance for our proxy on the level of minority shareholder protection, as well as to identify the legal system that each of the countries within our sample belong to. In regards to legal systems, we also utilized information contained within the CIAs The World Factbook.
Golub’s 2003 article published in the OECDs Economic Studies series entitled Measures of Restrictions on Inward Foreign Direct Investment for OECD Countries provided us with information regarding the level of investment restrictions that exists within member countries of the OECD.
The following table presets the summary statistics for the regression variables used. Here, only data for the year 2003 are presented.
151 Table 3.2: Descriptive Statistics for the year 2003
Variable Mean Stand. Minimu Maximu Dev. m m Life insurance consumption per capita 1145.933 958.5892 8.4000 3431.800 (LFI) Foreign market share (FMS) 23.0910 23.7821 0.0000 84.5589 Dependency ratio (DR) 0.4948 0.0482 0.4006 0.6155 Interest rate (IR) 4.6031 1.3761 1.0030 8.1000 Inflation rate (INF) 3.1934 4.4348 -0.3049 25.0000 Life expectancy (LE) 77.9571 2.7031 68.7667 82.0067 Foreign direct investment (FDI) 41218.38 102032.8 0.0000 458568 Education (EDU) 88.9333 10.1300 51.0000 100.000 Uncertainty avoidance index (UAI) 68.0333 23.3983 23.0000 112.000 Minority shareholder protection (MSP) 0.3000 0.4661 0.0000 1.0000 Gross domestic product per capita (GDP) 25544.40 9321.302 6457.857 49663.33 Investment restriction (INR) 0.1788 0.0867 0.0640 0.3900 Common-Law (Common) 0.2000 0.4068 0.0000 1.0000 French Civil-Law (Civil(F)) 0.3667 0.4901 0.0000 1.0000 German Civil-Law (Civil(G)) 0.2667 0.4498 0.0000 1.0000 Scandinavian Civil-Law (Civil(S)) 0.1667 0.3790 0.0000 1.0000
All figures are for the year 2003
This study incorporates the OLS and GMM estimation techniques105 and makes use of both STATA and TSP statistical packages. We employ pooled cross sectional analysis over all the years of study. A brief review of these estimation techniques is provided later in this section. However, for a more in-depth rundown on their validity and usefulness, please refer to Chapter 2 on the Determinants of Bancassurance.
105 Since lag instruments are used when applying GMM estimation, we include another two years to our sample period, making our period of study 1994-2003.
152 3.3.2 Law and the Determinants of Life Insurance.
The prior studies of researchers such as Browne, Chung and Frees (2000), Browne and
Kim (1993), Esho et al. (2004), Outreville (1996) and Park, Borde and Choi (2002) have looked at a number of variables when examining demand for both life and non-life insurance. Our aim here is to use a number of national variables that we consider to be most relevant in helping us gain a better understanding of what determines the level of life insurance consumption and the way that legal systems influence overall demand.
In this study, the model that is used incorporates variables from the literature on life and non-life insurance. We believe that the eleven variables included in our study form a key part of explaining the success of life insurance operations and the overall impact of the system of law within the country. The variables comprise:
• Foreign market share
• Dependency ratio
• Interest rate
• Inflation rate
• Life expectancy
• Foreign direct investment
• Education
• Uncertainty avoidance index
• Legal rights
153 Minority shareholder protection
Investment restriction
• Gross domestic product per capita
• Legal system
Common-Law
French Civil-Law
German Civil-Law
Scandinavian Civil-Law
3.3.2.1 Explanatory Variables
3.3.2.1.1 Foreign Market Share
Measure: Market share of (foreign controlled undertakings) and (branches/agencies of foreign controlled undertakings) in total domestic business expressed as a percentage.
The level of foreign participation within the domestic market is an important determinant of insurance and is used to proxy the price of insurance. The extent of foreign presence in the domestic market is a clear indicator of the openness of a country’s insurance industry.
However, the relationship that the level of foreign market participation has with the demand for insurance can be either positive or negative (Browne, Chung and Frees
(2000)), as we shall now explain.
154 Foreign market penetration could be limited as a result of governments employing restrictive regulations that limit the amount of foreign ownership within the country.
Skipper (1987) agrees with this and argues that the main method governments utilize is the implementation of trade barriers. While governments aim to act in the best interests of domestic suppliers, they are in fact causing a detrimental effect on the insurance industry due to the resulting loss of competitiveness. This will directly impact on the insurance prices within the country and in turn the overall demand for insurance products. If this were to be the case, then the level of foreign market share will have a positive impact on life insurance consumption according to economic theory, which states that more protective trade barriers lead to higher prices through a less competitive market
(Outreville (1996)).
The alternative way to look at a lack of foreign market share would be to argue that it comes as a result of the domestic market already being highly competitive and not as a result of protectionism. The idea is that the domestic market is so competitive that foreign insurers will not be able to find profitable prospects, and hence, not enter the market. The relationship between foreign market share and insurance consumption within the market will thus be negative, since a more competitive market leads to higher levels of demand but lower levels of foreign participation. Browne et al. (2000) find a negative relationship exists between foreign market share and the motor vehicle line of insurance, while they find a positive relationship existing with the general liability line. Another argument would be that domestic customers prefer products from domestic distributors as a result
155 of a sense of nationalism or due to the fact that the products offered by foreign companies are inferior to domestically produced products.
Hypothesis: The relationship between foreign market share and the level of life
insurance consumption may be either positive or negative.
→ If positive, it means that low levels of foreign market participation stem from
protectionism and barriers to entry, directly resulting in loss of market
competitiveness.
→ If negative, it means the lack of foreign market participation stem from a more
competitive domestic market that reduces the incentives of foreign entry.
For this variable we incorporate actual foreign market share figures akin to Browne et al.
(2000) and unlike Outreville (1996) who uses a dummy variable. We use the combined foreign undertakings, which are inclusive of foreign controlled branches and agencies, within the domestic life insurance market.
3.3.2.1.2 Dependency Ratio
Measure: Sum of the population aged under 15 and over 64 divided by the population between the ages of 15 and 64.
156 One of the core reasons that drive the consumption of life insurance products is the need to protect dependents against financial hardships in the case of the wage earners premature death. This is becoming increasingly important in many nations that are faced with an aging population and a burdened pension system. Researchers such as Lewis
(1989),Truett and Truett (1990), Browne and Kim (1993) and Showers and Shotick
(1994) find a significant positive relationship between the country’s dependency ratio and the subsequent demand for insurance products. Furthermore, it is argued that this increase in demand resulting from the increase in dependents occurs regardless of whether the insurance coverage is supplied by domestic or foreign providers. Other researchers, on the other hand, such as Outreville (1996) and Goldsmith (1983) fail to find such significance in their studies, while Auerbach and Kotlikoff (1989) actually find a negative relationship exists.
In this paper, we hypothesize that there is indeed a positive relationship between the number of dependants within a country and the demand for life insurance products. The greater the number of dependants, the greater the adverse affects would be from the primary wage earners premature death, since the present value of the family’s consumption is greater. In turn, there will be an increased need to guard against the death of the wage earner.
Hypothesis: There is a positive relationship between the number of dependents and the
demand for life insurance.
157 We base the measure of this variable on the generally accepted definition of the term
‘dependents’ as those who are too young or too old to work. The dependency ratio is expressed as the percentage of dependents to the working aged population.
3.3.2.1.3 Interest Rates
Measure: Long term interest rates as provided by the OECD Factbook106.
The level of interest rates will cause a direct impact on not only the spending patterns of the populace but also on the products that are offered by insurance companies. Since both new borrowings and existing repayments become more expensive should interest rates increase, people will avoid spending on products that are not truly necessary, such as insurance products. At the same time, increases in interest rates will dissuade businesses from investing in such things as new equipment (OECD Factbook 2005 online version).
This will mean that there will not be many new products brought to the market, which, in these times of the sophisticated consumer107, will mean that the consumption of life insurance products will decrease. Thus, in line with the hypothesis of Outreville (1996)108 and the findings of Cargill and Troxel (1979) and Rubayah and Zaidi (2000), we hypothesize there to be negative relationship existing between the level of interest rates and the consumption of life insurance products.
106 Government bonds with a maturity of about 10 years. 107 Current consumers are more sophisticated in modern times and demand greater variety in the products that are offered to them. Should there be a lack of variety then it is reasonable to argue that overall consumption will be adversely affected.
158 Hypothesis: Interest rates will be negatively related to the level of life insurance
consumption within a country.
We use the long term interest rate as provided by the OECD Factbook as opposed to the short term cash rate used by such as Rubayah and Zaidi (2000), since it better represents the relationship between the just-mentioned level of business investment and its impact on life insurance consumption, while at the same time still being able to represent changing consumer spending patterns109.
3.3.2.1.4 Inflation Levels
Measure: Annual average inflation rate for a country.
For inflation the same argument is presented here as in Chapter 2 on The Determinants of
Bancassurance, and to avoid excessive repetition we provide a summary of the argument presented therein. Cummins (1991) contends that inflation can increase an insurance company’s premium income as well as reducing the real value of insurance liabilities.
This may result in subsequent cost reductions or new products being brought to the market that will, in turn, mean that a positive relationship could exist between the level of inflation and the demand for insurance.
108 The hypothesized relationship proved to be insignificant and not strictly negative. 109 We believe that a ten year period is not a considerable amount of time and any changes in these ‘long term’ interest rates will still cause consumers to change their spending patterns.
159 However, when we focus on how inflation affects the consumer, we find that a negative relationship will exist. This is a direct result of inflation increasing the overall cost of a basket of ‘necessities’ while not necessarily contributing to an increase in the overall income of the consumer. This in turn would mean that consumers will have less disposable income, and hence their demand for insurance products would undoubtedly diminish. Additionally, the real value of any insurance product will be severely diminished should inflation rates rise, which would adversely impact their attractiveness to the consumer. Browne and Kim (1993) and Outreville (1996) both found that inflation would be detrimental on savings through insurance products as it erodes the value of the product.
As a result of these arguments, it is not clear to say how inflation affects the life insurance industry. However, based on our results from Chapter 2, we believe that the overall impact on the consumer would inevitably be more influential.
Hypothesis: Inflation is negatively related to the demand for life insurance.
We use the actual average inflation rate per annum for each country as provided by the
OECD Factbook. Academics such as Browne and Kim (1993) tend to use an averaged figure over a number of years. However, we propose that the consumer in particular does not really take into consideration what the inflation rate will be in later years. Some would argue that a change in inflation would not be influential for a while and the consumer would be able to live their lives without change. We, on the other hand, believe
160 that any change would have an immediate impact on how consumers perceive insurance products, mainly due to the fact that they are a luxury item that people tend to neglect should they be faced with high prices in other areas of their lives which they deem more important. Any rise in inflation will have an immediate impact, and hence, we aim to determine how premiums per capita behave in immediate response to inflationary changes.
3.3.2.1.5 Life Expectancy
Measure: Average number of years of life remaining to a person at a particular age, based on a given set of age-specific mortality rates.
Life expectancy which expresses the average lifespan of the individuals within a country has been found by many researchers to be an important determinant of the demand for life insurance products. However, the relationship that exists between it and the actual demand for life insurance has been found to be somewhat ambiguous, in that it can be either positive or negative.
A negative relationship will exist between life expectancy and the demand for life insurance if life expectancy is seen as an adequate proxy for the probability of death. If life expectancy increases, it means that people will be living for longer periods of time, on average, and consequently the probability of death will decrease. All this will mean that the need to insure ones life will not become paramount until later years. Both
161 Browne and Kim (1993) and Lewis (1989) find life insurance consumption increases as the probability of the wage earner’s death increases110.
Alternatively, life expectancy may be positively related to the level of insurance consumption as highlighted by Outreville (1996), where life expectancy is used as a proxy for the actuarial fair price of insurance in a developing country. The positive relationship stems from the rationale that a longer life span reflects a lower actuarial price for life insurance as well as a greater incentive for human capital accumulation, all of which will bolster demand. Beenstock, Dickinson and Khajuria (1986) have also found that increases in average life expectancy raises demand and decreases the supply of life insurance products.
Given existing research, we believe that changes in average life expectancy will have an impact on the overall consumption of life insurance products, and this impact can be either positive or negative. A positive impact will occur if higher life expectancy results in lower life insurance prices, leading to greater incentives for human capital accumulation. On the other hand, a negative relationship will exist if higher levels of life expectancy indicate a reduction in the probability of death, resulting in wage earner’s having less incentive to protect their dependents against the outcome of the wage earner’s untimely death.
Hypothesis: The relationship between average life expectancy and the level of life
insurance consumption may either be positive or negative.
110 The life expectancy variable lacked significance in some of Browne and Kim’s (1993) models.
162 → If positive, it means that increases in average life expectancy are coupled with
decreases in the price of insurance.
→ If negative, it means average life expectancy is viewed as an adequate proxy for
the probability of death.
3.3.2.1.6 Foreign Direct Investment (FDI)
Measure: Amount of foreign investments within the life insurance industry as provided by the OECD Insurance Statistics Yearbook (various issues).
This measure is similar to our proxy on Price that viewed the share of the market held by foreign participants. This measure provides a glimpse as to the growth within the life insurance industry. Greater amounts of FDI will mean more capital for the domestic industry as well as technological transfers and improved products, all of which will go toward boosting insurance consumption. At the same time, an increase in FDI will mean that the domestic arena is competitive and productive enough to justify such investments, which will once again mean that the level of insurance demand will increase. This shows that a positive relationship shall exist between life insurance consumption and the level of
FDI.
Hypothesis: The relationship between the level of foreign direct investment and life
insurance consumption will be positive.
163 3.3.2.1.7 Education
Measure: Net secondary enrollment rate expressed as a percentage.
The level of education is an important factor in helping people better understand the risks involved with everyday life and the overall benefits of insurance coverage. As a result, the levels of education becomes an important determinant of the demand for both life and non-life insurance, none more so than in the past three decades111. Greater levels of education will mean that the populace can better determine the risks involved with any given action and have a clearer understanding of the uncertainties that exist within life, which will cause an increase in their overall risk aversion and the requirement for better protection via insurance products112. Browne and Kim (1993) agree with this view of education as a measure of the overall risk aversion of the populace as does Browne,
Chung and Frees (2000), Beck and Webb (2003) and Esho et al. (2004). Truett and Truett
(1990) add to this common argument by suggesting that this will, in turn, increase the recognition of various types of life insurance products offered, leading to high levels of demand.
In addition, the greater the length of education, the greater the length and cost of dependency (Browne and Kim (1993)), consequently increasing the need of protection
111 Miyamoto (2003) argues that educational attainment has steadily increased over the past three decades. 112 The argument can also be made that increases in education will enable the individual to assess the level of risk associated with a course of action and hence avoid the risk. This will, in turn, result in a negative relationship existing between the level of education and the level of life insurance consumption. Szipro and Outreville (1988) have argued in favor of this, while Outreville (1990) has found an insignificant negative relationship between education and life insurance demand.
164 through life insurance, as was expressed in Section 3.3.2.1.5. Yet another argument is that the increased education causes people to be less risk averse, and at the same time take greater risks that will, in turn, need to be protected against113 (Esho et al. (2004)).
Inconsistencies exist in regards to the results concerning educations impact on demand.
Two proponents who have found a negative relationship include Anderson and Nevin
(1975) and Auerbach and Kotlikoff (1989). This is not surprising when you consider education as helping individuals to better determine the level of risks involved with an activity, and thus avoid said activity if the level of risk proves too high, which will then cause a decrease in the demand for insurance products. However, we agree with the majority of previous researchers that argue for a positive relationship existing between the level of education and the demand for life insurance products.
Hypothesis: There is a positive relationship between the level of education within a
country and the level of life insurance consumption.
Akin to Esho et al. (2004) we utilize World Bank figures that express the secondary enrollment rate within a country to proxy the level of education. Esho et al. (2004),
Outreville (1990) and Browne, Chung and Frees (2000) all use the proportion of the population that has completed secondary education. We believe that the enrollment rate is just as good a measure, if not better, since those that have enrolled but have not
113 This means that they need a different sort of coverage to that of the average individual, since they are facing risks most people would seek to avoid.
165 completed secondary education are still in an advanced position to understand risks and the reasons behind insurance coverage114.
3.3.2.1.8 Uncertainty Avoidance Index (UAI)
Measure: Measure of 8 to 112 based on Hofstede (1995) survey data as highlighted on the ITIM website115.
It is argued by Fukuyama (1995) that insurance is related to the cultural context of a given economy. Fukuyama further states that insurance consumption will depend on the level of trust within an economy where high-trust economies such as the U.S., U.K.,
Japan and Germany are indifferent to risk and are open to alternative risk transfer mechanisms (alternate to insurance), while low-trust economies like that of France and
Italy would prefer insurance coverage, since people in these nations find it harder to transact with unknown counterparties. Hofstede (1995) constructs a similar taxonomy by characterizing societies as either low-group or high-group societies with low-group societies preferring to support market based means of dealing with uncertainty such as insurance coverage. A set of data that Hofstede constructed to characterize a society’s dependence on market based means of dealing with risk is known as the Uncertainty
Avoidance Index (UAI)116.
114 We have also tested tertiary enrollment rates obtained from the OECD Factbook as a proxy for education, which is symbolic of the proportion of the population completing secondary education, and have obtained similar results to those obtained when secondary enrollment rates were used. 115 www.geert-hofstede.com
166 The UAI is another measure of risk aversion, as is education, and is based on survey data that is constructed through the collection of employee attitudes towards workplace stress, enforcement of company rules, and longevity of employment. The UAI ranges between
8, for the lowest uncertainty avoidance country, to 112, for the highest. The higher scores are generally associated with those cultures that are still in the process of modernization.
The general definition for UAI as found on the ITIM website is as follows:
Uncertainty Avoidance refers to the extent to which a culture feels
threatened by ambiguous, uncertain situations and tries to avoid them by
establishing more structure. The high positive scores on the uncertainty
avoidance index (UAI) indicate low tolerance for ambiguity. These
cultures prefer to avoid uncertainty and dissent as a cultural value and
desire consensus. As a result, HIGH uncertainty avoidance cultures prefer
formal rules and any uncertainty can express itself in higher anxiety than
those from low uncertainty avoidance cultures. Cultures with low UAI
scores have a high tolerance for uncertainty and ambiguity, believe in
accepting and encouraging dissenting views among cultural members and
in taking risks and trying new things. Thus, cultures which ranked low
(compared to other cultures), feel much more comfortable with the
unknown.
The relationship between UAI and the level of life insurance consumption is assumed to be positive. Park, Borde and Choi (2002) and Esho et al. (2004) both hypothesize a
116 This is an idea promoted by Hofstede (1995) who saw insurance as a product of national values.
167 positive relationship existing, but in tests conducted, both sets of researchers discovered that the evident relationship proved to be insignificant. While Esho et al. (2004) find the hypothesized positive relationship, Park et al. (2002) find a negative relationship existing between the UAI variable and the level of insurance pervasiveness. Since not much research has been conducted in relation to the effect of UAI on the demand of insurance, we cannot be sure which of these researchers’ results are accurate. We will use the same argument presented in both the Park et al. (2002) and Esho et al. (2004) studies in that the relationship between UAI and the level of insurance consumption should be positive.
Hypothesis: The relationship between the Uncertainty Avoidance Index (UAI) and the
level of life insurance consumption is positive.
3.3.2.1.9 Legal Rights
Legal rights are an important determinant when viewing the openness of an industry, and will have a direct impact on the overall demand for the products on offer through affecting competitiveness and prices. We use two separate measures when analyzing the openness of the life insurance industry, namely, the level of minority shareholder protection and the level of investment restriction. Both can be argued to influence life insurance consumption by inevitably altering the prices at which insurance products are supplied.
168 3.3.2.1.9.1 Minority Shareholder Protection
Measure: Dummy variable of 0 or 1 indicating whether laws are in place which provides a certain degree of protection to minority shareholders with 1 indicating that some level of protection exists.
The protection afforded to the minority shareholder is indicative of the competitiveness within an industry. Should there be no protection, then minority shareholders will have limited to no say on the functioning of the company and can easily be muscled out of their ownership. This will mean a concentration of ownership in the hands of a few individuals and may also result in management making decisions which may be uncompetitive and not in the interests of all the stakeholders, merely in the interests of the major shareholders.
For this variable we make use of the dummy variable used in La Porta et al. (1998). The variable is defined as follows:
Equals one if the company law or commercial code grants minority
shareholders either a judicial venue to challenge the decisions of
management or of the assembly or the right to step out of the company by
requiring the company to purchase their shares when they object to
certain fundamental changes, such as mergers, asset dispositions, and
changes in the articles of incorporation. The variable equals zero
169 otherwise. Minority shareholders are defined as those shareholders who
own 10 percent of share capital or less.
Based on the arguments above, we believe that a positive relationship should exist between the protection afforded to the minority shareholder and the overall level of life insurance consumption.
Hypothesis: A positive relationship exists between the level of minority shareholder
protection and the demand for life insurance.
3.3.2.1.9.2 Investment Restriction
Measure: Indices of FDI restriction over time for the whole economy as presented by
Golub (2003).
Restrictions on FDI are identified by Markusen and Maskus (2001) as a key variable in theoretical knowledge-capital models of multinational corporations, i.e. multinational insurance organizations. This is directly related to the Foreign Direct Investment variable and provides a different way of analyzing the overall openness of, and competition within, local markets. Data on the levels of investment restriction is notoriously difficult to obtain and we believe that the level of FDI restriction is an adequate proxy for the total investment restriction within a nation. It must be noted that we do not merely use the level of FDI restriction within the insurance industry; instead we make use of the overall
170 restriction of FDI within the economy117 118. We believe this is important since the level of restrictions, and hence competitiveness, of other sectors may either directly or indirectly impact on the life insurance industry as well. For example, excessive restrictions in the banking sector will mean less competitiveness and higher prices; this will cause individuals who purchase insurance products from banks (bancassurers) to look elsewhere for cheaper options, which will cause an increase in the demand of insurance products offered by traditional insurers (in our case life insurers). Based on these arguments, we hypothesize there to be a negative relationship between the level of investment restriction and the amount of life insurance consumption.
Hypothesis: A negative relationship exists between the amount of investment
restriction and the demand for life insurance products.
As was mentioned, it is hard to quantify and calculate the level of investment restrictions, and as a result, a portion of Golub’s (2003) data is extrapolated based on methods described within his paper. This may cause some bias in our results since the data is not complete119.
117 The FDI restriction within the insurance industry is closely in line with that of the total economy. We tested the FDI restriction in the insurance industry separately and found that our results did not differ much. 118 The total FDI investment restriction within the economy is calculated through separately viewing the restrictions evident in: Business Services (Legal, Accounting, Architecture, Engineering), Telecommunications (Fixed, Mobile), Construction, Distribution, Finance (Insurance, Banking), Hotels and Restaurants, Transportation (Air, Maritime, Road), Electricity, and Manufacturing; and then averaging them out based on how much each sector contributes to the whole economy. 119 For a full explanation on how FDI restriction was calculated, please refer to Golub (2003): Measures of Restrictions on Inward Foreign Direct Investment for OECD Countries, OECD Economic Studies No. 36.
171 3.3.2.1.10 GDP per capita
Measure: Amount of GDP divided by the number of individuals in the population.
The life insurance market within any country will depend heavily on the economic development that exists. As economic growth increases, the consumption of all types of insurance products should also increase as a result of increased income and greater opportunity costs in terms of income forgone to dependents in the event of an untimely death (Esho et al (2004)). Beck and Webb (2003), Beenstock, Dickenson and Khajuria
(1988), Babbel (1981), Browne, Chung and Frees (2000), Browne and Kim (1993),
Campbell (1980), Cargill and Troxel (1979), Gandolfi and Miners (1996), Hwang and
Gao (2003), Lewis (1989), Lim and Haberman (2004), Outreville (1990, 1996) and Truett and Truett (1990) have all documented a positive association between economic development (and the amount of income) and the level of insurance consumption. Ward and Zurbruegg (2000) have shown that a similar relationship exists amongst OECD countries. As a result of the past research, we hypothesize that there will indeed be a positive relationship between the level of insurance consumption and the amount of economic development and domestic income.
Hypothesis: The relationship between the level of economic development and life
insurance consumption will be positive.
172 Akin to Outreville (1996) and Esho et al. (2004) we use real GDP per capita to proxy economic development. This will also be an adequate proxy for the level of disposable income within the country120.
3.3.2.1.11 Legal Systems
Legal systems and their influence on the overall level of life insurance consumption are considerably difficult to determine. As yet, no study has been done in this field, and only a handful have been conducted analyzing legal systems and their impact on non-life insurance demand. Esho et al. (2004) and Browne, Chung and Frees (2000) provide the two most noted works in this regard. However, their results are far from conclusive, as we will focus on when we investigate our results in Section 3.4.2. Overall Esho et al.
(2004) finds that legal systems do not unduly influence the demand for property and casualty insurance, while, in contrast, Browne et al. (2000) conclude that the common- law legal system that is common amongst English colonial nations is significantly positively related to the consumption of property-liability insurance in OECD countries.
As has been discussed earlier in Section 3.2.4, some of the most notable works regarding legal systems has been conducted by researchers such as La Porta et al. (1998, 2000),
Levine (1998, 1999) and Levine, Loayza and Beck (2000). La Porta et al. (1998) showed that common-law nations provide a higher level of legal protection of shareholder and creditors’ rights. Since insurance contracts can be viewed as analogous to risky corporate
120 If looking primarily at income, a more suitable measure would be GNI per capita as was used in Chapter 2. However, since we aim to look at economic development and income in unison GDP per capita is a
173 debt (Cummins and Danzon (1997)), those nations that afford a higher level of protection should in all rights facilitate greater insurance demand. From the results of La Porta et al.
(1998) we would conclude that common-law countries promote insurance consumption more so than civil-law nations. Levine (1998, 1999) and Levine et al. (2000) find that a well-defined and enforced legal system promotes greater financial intermediation which in turn promotes economic growth. However, they find this only to be true for countries that incorporate the German civil-law legal system. In contrast to what was previously inferred from La Porta et al. (1998), this would suggest that only the German civil-law system helps in promoting the consumption of insurance products.
From the works just mentioned, it is very hard to make any clear distinction as to how the legal system within a country will affect the overall level of life insurance consumption.
Inferences can easily be made based on the results of La Porta et al. (1998, 2000) and
Levine (1998, 1999), but these can just as easily prove to be false as they do true. In addition, these works do not include factors regarding the economy that would also be influential in regards to levels of insurance consumption. For example, increased competition within the industry should improve product efficiency and in turn insurance consumption However, Cummins et al. (1999) and Hardwick (1997) both state in the
U.S. and U.K (the two paramount common-law nations) product inefficiencies are high for life insurers, which will adversely influence consumption patterns. The works of those such as Esho et al. (2004) and Browne et al. (2000) do not provide us with conclusive results either, and are based on a different industry to that which we analyze in this chapter. As a result of all this, we believe it justified to leave the relationship between
suitable measure.
174 each legal variable tested as questionable121 in regards to how they influence the level of life insurance consumption.
Hypothesis: The relationship between the system of law within a nation and the level of
life insurance consumption may either be positive or negative, with this
hypothesis holding true for all the systems of law tested.
3.3.3 Model for the Determinants of Life Insurance
Based on the hypotheses presented in the previous section, the following model for the determinants of life insurance consumption is proposed:
LFI = f [FMS, DR, IR, INF, LE, FDI, EDU,
UAI, MSP,GDP, INR,Common,Civil(F),Civil(G),Civil(S)]
where,
121 Questionable in the sense that the relationship may be either positive or negative.
175 Table 3.3: Hypothesized Relationships
Expected Hypothesis Variable Name relationship with dependant Life insurance consumption per capita LFI N/A Foreign market share FMS ? Dependency ratio DR +VE Interest rate IR -VE Inflation rate INF -VE Life expectancy LE ? Foreign direct investment FDI +VE Education EDU +VE Uncertainty avoidance index UAI +VE Minority shareholder protection MSP +VE Gross domestic product per capita GDP +VE Investment restriction INR -VE Common-Law Common ? French Civil-Law Civil(F) ? German Civil-Law Civil(G) ? Scandinavian Civil-Law Civil(S) ?
Since a large amount of literature in the field of insurance demand/consumption uses either a linear or log-linear model when running tests, we again run a series of Box-Cox transformations through STATA to see if the linear form of our model is suitable or not.
Please note that for brevity we do not provide a full description of the Box-Cox testing procedure, OLS method of testing or the GMM method of testing - instead we provide a brief rundown. Should you require more information, please refer back to Chapter 2 on the Determinants of Bancassurance.
176 Linear form of our model:
LFI = β 0 + β1 (FMS) + β 2 (DR) + β 3 (IR) + β 4 (INF) + β 5 (LE) + β 6 (FDI) + β 7 (EDU)
+ β8 (UAI) + β 9 (MSP) + β10 (GDP) + β11 (INR)
+ β12 (Common) + β13 (Civil(F)) + β14 (Civil(G) + β15 (Civil(S)) + ε
When transforming both the left and right hand sides of the equation at the same time we obtained a value for θ that was not significantly different from 1, hence we use a linear transformation on the dependant variable. We obtained similar results for λ on the right hand side. What we did next was to check each of the independent variables on the right hand side of the equation separately. In order to do this, we left all the variables in their linear form except for one that we transformed in the above manner. We repeated this process for each independent variable and found that only one of the variables had a λ that was not significantly different from 0, while all the rest had λ values close to 1, suggesting linear transformations for these variables would be appropriate. When transforming the UAI variable we obtained a λ of 0.252, which is shown to be significant at 0, hence indicating that the log-linear transformation is valid for this variable122 123.
122 The tests also revealed that the GDP variable showed significance at either 0 or 1 (with it favoring the value of 1 in terms of significance); as a result we ran our tests using both the linear form of the variable as well as the log-transformed version. Tests run with the log-transformed version of the GDP variable proved to yield poorer results with lower BIC values and p-values.
177 Given the above specification tests, for estimation purposes, the model that we shall use for testing purposes becomes:
LFI = β 0 + β1 (FMS) + β 2 (DR) + β 3 (IR) + β 4 (INF) + β 5 (LE) + β 6 (FDI) + β 7 (EDU)
+ β8 LOG(UAI) + β 9 (MSP) + β10 (GDP) + β11 (INR)
+ β12 (Common) + β13 (Civil(F)) + β14 (Civil(G) + β15 (Civil(S)) + ε
Once we have determined the functional form of our model, we firstly employ the OLS method of testing. This testing procedure is quick and easy and can provide estimators that are unbiased, consistent and BLUE124. However, this will only prove true should a number of assumptions be adhered to125. In order to test that the main assumptions required for this method of testing are adhered to, we incorporate diagnostic tests to check for heteroscedasticity, normality of residuals and multicollinearity.
The second method of testing we use is GMM incorporating instrumental variables. This method of testing improves on the simple OLS, since it allows for consistent and efficient estimates to still be derived, even if some of the assumptions required for OLS do not hold. Hansen (1982) shows that GMM instrumental variable estimation allows more robust calculations by helping avoid data problems such as errors in the variables, endogeneity and multicollinearity. GMM also is superior to OLS, since it does not
123 Similar to Chapter 2, we also ran our tests with a log-transformed dependent in line with tests performed by other researchers. However, we once again found that the results yielded lower BIC values and p-values, indicating that such a transformation adversely affects our tests. 124 Best Linear Unbiased Estimate. 125 Please refer to Section 2.6.4.
178 require any pre-specified knowledge of the distribution of the error terms, allowing us to remove the normality requirement. Instead of this, all the GMM requires is that orthogonality conditions be met. If these conditions are met, then the estimator is consistent, and if variance of the moment conditions is consistent, then the estimator is also efficient126.
Valid instruments should have two key attributes. They should have little to no correlation with the error term, and be highly correlated with the explanatory variables. In line with previous research, we use lagged values of our independent variables as instruments.
3.4 EMPIRICAL RESULTS AND THEIR IMPLICATIONS127
Below, we provide a brief overview of our empirical results, providing only the main attributes of each test conducted. We will then compare and sum up the implications and significance of these results later in this section.
126 For a more complete explanation on GMM please refer to Section 2.6.5. 127 For testing purposes we used a combination of the STATA and TSP statistical packages
179 3.4.1 Empirical Results
The result for our test of all 30 OECD countries using both regression analysis and GMM tests are presented in the following tables. The first set of tests makes use of a pooled
OLS estimation procedure which allows for a larger sample size and observation of the relationships between variables over a number of years. For diagnostic purposes, we make use of the LM heteroscedasticity test of the residual variances in order to check for the presence of heteroscedasticity. In order to accomplish this, squared residuals are regressed on the squared fitted values and the subsequent test statistic is Chi-square distributed with one degree of freedom. We also incorporate the Jarque-Bera test to test for the normality of the residuals, and hence, determine whether the t-statistic is suitable or not, with the null being that the residuals are in fact normally distributed128. Lastly, we also examine whether multicollinearity problems exist by observing the correlations amongst the explanatory variables, which is highlighted in Table 3.5. Numerous groups of variables show significant correlations, and we feel that these must be explained before we move any further.
The level of life insurance consumption (LFI) shows significant correlations with both interest rates (IR) and gross domestic product per capita (GDP). This does not seem all that surprising considering how the amount of life insurance consumed by an individual is going to be considerably affected by the amount of disposable income that they
128 The corresponding test statistic is Chi-square distributed with two degrees of freedom.
180 poses129. Is it not surprising either, that there appears to be significant negative correlation between inflation rates (INF) and both the level of education (EDU) and life expectancy (LE) as rises in inflation rates would mean necessities become more expensive, making it harder to sustain one’s lifestyle, thus decreasing average life expectancy within the population130. Education will similarly be effected due to the combined effects of parents not being able to afford tuition, and students seeking to leave school early in order to obtain employment. The relationship between the common system of law (Common) and the level of minority shareholder protection (MSP) is indicative of the relationship found in La Porta et al. (1998), whereby common-law nations tend to offer greater levels of protection to investors and creditors. The corresponding relationship between minority shareholder protection (MSP) and the two civil systems of law (Civil(F)) and Civil(G)) is also in line with the arguments of La Porta et al. (1998) in the sense that French civil-law offers the weakest form of protection, while German civil-law is somewhere in between. These results are not surprising considering how we utilize the La Porta et al. (1998) paper in constructing our minority shareholder protection and legal systems variables. This difference in the level of protection offered is also represented in the correlations between the three systems of law and the level of foreign direct investment (FDI) with only common-law having a positive relationship, indicating that foreign investors are attracted by the protection offered in common-law states and not so much by those offered in civil-law states. Finally, it is also
129 Disposable income is directly affected by both GDP and interest rates. GDP is a rough measure of national income, while the level of interest rates will impact on the individuals borrowing and savings pattern, and hence their spending pattern as well. 130 A similar argument can be used to explain the correlation between life expectancy (LE) and gross domestic product per capita (GDP), since rises in GDP will generally result in an increase in disposable income which will better allow individuals to maintain their lifestyles.
181 not surprising that we obtain such a strong positive correlation between the French civil- law system (Civil(F)) and the uncertainty avoidance index (UAI) considering how, on average, French civil-law nations tend to experience the most uncertainty, with the average UAI being 87.64, while those of German civil-law and English common-law nations are far lower, 75.25UAI and 44UAI respectively.
These highly correlated variables indicate the possibility of multicollinearity errors surfacing in our OLS results, and consequently, to reduce this problem, instrumental variable estimation is used through the application of GMM tests131.
In our initial tests, we aim to capture the effect of legal systems on the consumption of life insurance by following Esho et al. (2004), Levine et al. (2000), and La Porta et al.
(1997) and including the three legal system dummies for English common-law, French civil-law and German civil-law, and excluding Scandinavian civil-law countries from our tests. The results for this are presented in Table 3.4. In later tests, we make use of the
Scandinavian sample when analyzing the effects of common-law versus civil-law132 and tests using only the three civil-law systems.
In the initial tests that incorporate the three legal systems of English common-law, French civil-law and German civil-law, the LM heteroscedasticity test shows the existence of
131 Instrumental variable estimation reduces the impact of multicollinearity problems, since the large condition index of the OLS X’X matrix is reduced, since we use the instrumental variable matrix Z’X which has a smaller condition index. The combination of the proper instruments (checked through the Hansen J-test) with the smaller condition index will result in a mitigating impact of the multicollinearity on the GMM estimates. 132 Civil-law in this instance is a combination of the French, German, and Scandinavian civil-law countries.
182 non-homogeneity in the residual variances, while the Jarque-Bera tests for normality is insignificant at the 5 percent level, indicating the suitability of the t-statistic. The adjusted
R 2 for the test is 0.6034 and the constant is insignificant, indicating that our set of variables captures the major affects that determine the level of life insurance consumption within OECD countries. At the 1 percent level of significance, the foreign market share
(FMS), foreign direct investment (FDI), uncertainty avoidance index (UAI), minority shareholder protection (MSP), gross domestic product per capita (GDP), and German civil-law legal system (Civil(G)) prove to be significant. Meanwhile, at the 5 percent level of significance, the dependency ratio (DR), level of interest rates (IR), level of education (EDU), and the French civil-law legal system (Civil(F)) show significance. The level of investment restriction (INR) shows significance at the 10 percent level, with the level of inflation (INF), life expectancy (LE), and common-law legal system (Common) showing no significance at all. The signage of the variables is surprising in certain cases; the level of inflation (INF) and the uncertainty avoidance index (UAI) both have signs contrary to those anticipated, while the signs of those variables which we were uncertain about during our initial predictions reveal certain insight into how they affect the level of life insurance consumption in OECD countries. We fully analyze and discuss the implications of these results, as well as all others presented within this section, in the following section.
GMM tests utilizing instrumental variable estimation were run since they have certain advantages over the more simplistic OLS method, as was discussed in the previous section. The Hansen J test for overidentifying instruments is insignificant, indicating that
183 the instruments that are being used are appropriate for our tests. The constant is also insignificant indicating the suitability of our model. The adjusted R 2 for the test is
0.5862 and the results support those of the OLS tests fairly closely. The only differences being that the level of education (EDU) becomes significant at the 1 percent level, and the German civil-law legal system is now significant at the 5 percent level.
Following our initial tests, we also performed tests on two sub-periods within our sample in order to determine whether the results vary with time. For this, we divided our eight- year sample period into two four-year groups, 1996-1999 (1st period) and 2000-2003 (2nd period), and ran the same tests as before. The results for both periods and both methods of testing are presented in Table 3.6. The adjusted R 2 for each OLS test is 0.5447 and
0.6514 respectively, and the constant in both cases is insignificant. During the 1st period, only the foreign market share (FMS), gross domestic product per capita (GDP), and
German civil-law legal system (Civil(G)) exhibit significance at the 1 percent level.
Minority shareholder protection (MSP) shows significance at the 5 percent level, the dependency ratio (DR), level of foreign direct investment (FDI) and the uncertainty avoidance index (UAI) show significance at the 10 percent level, while the rest of the variables are insignificant. The 2nd period results are somewhat different, with the dependency ratio (DR), interest rates (IR), level of foreign direct investment (FDI), uncertainty avoidance index (UAI) and gross domestic product per capita (GDP) showing significance at the 1 percent level; the German civil-law legal system (Civil(G)) becomes significant at the 5 percent level, while the remaining variables show no significance.
184 Table 3.4: Pooled OLS and GMM Results This table presents the pooled OLS and GMM results for the initial sample of 25 OECD countries (excluding Scandinavian civil-law) over the eight-year period of 1996-2003 (excludes the 5 Scandinavian countries). For the GMM tests an additional two years of lags we also incorporated for each variable. Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Intercept 399.198 0.2132 0.831 1580.15 0.8108 0.417 FMS -11.0678 -5.7977 0.000*** -9.0920 -3.9992 0.000*** DR 2659.92 2.5061 0.013** 2933.33 2.2828 0.022** IR -38.9278 -2.4045 0.017** -35.6795 -2.1265 0.033** INF 4.1583 1.0813 0.281 3.2321 1.1704 0.242 LE 5.8055 0.2282 0.820 -14.8287 -0.5239 0.600 FDI 0.0028 3.8454 0.000*** 0.0030 2.8083 0.005*** EDU 12.0088 2.3092 0.022** 17.7825 3.1794 0.001*** UAI -1711.23 -3.6728 0.000*** -1841.73 -2.8193 0.005*** MSP 468.210 2.8841 0.004*** 551.464 2.5928 0.010*** GDP 0.0321 6.0615 0.000*** 0.0282 4.5719 0.000*** INR -826.416 -1.6373 0.103* -955.332 -1.7673 0.077* Common -250.268 -1.2017 0.231 -251.237 -0.8216 0.411 Civil(F) 617.048 2.3950 0.017** 702.012 1.9926 0.046** Civil(G) 1057.44 3.810 0.000*** 907.566 2.2770 0.023**
Adjusted 0.6034 0.5862 R 2 Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.000 0.568 Test Test JB Test 0.710 * Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction, Common=Dummy for common-law societies, Civil(F)=Dummy for French civil-law societies, Civil(G)=Dummy for German civil-law societies.
185 Table 3.5: Correlation Matrix This table presents the correlation matrix amongst variables for the whole sample of 30 OECD member nations in order to allow for the detection of severe multicollinearity in the data.
LFI FMS DR IR INF LE FDI EDU UAI LFI 1.0000 FMS -0.4384 1.0000 DR 0.0421 -0.4551 1.0000 IR -0.6195 0.3041 0.1207 1.0000 INF -0.3541 0.0434 0.1532 0.1481 1.0000 LE 0.5250 -0.5166 0.0461 -0.4399 -0.7446 1.0000 FDI 0.3873 0.0072 0.0319 -0.2179 -0.1086 0.1239 1.0000 EDU 0.3845 -0.0422 -0.2823 -0.2022 -0.7360 0.6477 0.1240 1.0000 UAI -0.3763 0.2272 -0.0081 -0.0711 0.1547 -0.2686 -0.2757 -0.2539 1.0000 MSP 0.2133 -0.1743 -0.2282 -0.1541 -0.1097 0.2784 0.5629 0.2016 -0.2697 GDP 0.5937 -0.3133 0.0360 -0.4308 -0.4800 0.6811 0.2945 0.3801 -0.4254 INR -0.4312 -0.0213 -0.0358 0.3099 0.3244 -0.1399 -0.0628 -0.473 0.0622 Common 0.1623 -0.1346 -0.0308 0.0919 -0.0619 0.1686 0.5573 0.0953 -0.5224 Civil(F) -0.2576 -0.0166 0.2949 0.0080 0.2447 -0.2211 -0.2940 -0.3491 0.6484 Civil(G) 0.0165 0.3901 -0.5008 -0.1907 -0.1007 -0.0852 -0.0543 0.0873 0.1892
186 MSP GDP INR Common Civil(F) Civil(G) MSP 1.0000 GDP 0.1853 1.0000 INR 0.1400 -0.3105 1.0000 Common 0.7638 0.2875 0.0444 1.0000 Civil(F) -0.3472 -0.2112 -0.1764 -0.3804 1.0000 Civil(G) -0.0658 -0.2444 0.0794 -0.3015 -0.4588 1.0000
FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction, Common=Dummy for common-law societies, Civil(F)=Dummy for French civil-law societies, Civil(G)=Dummy for German civil-law societies.
187 Table 3.6: Sub-Period Results This table presents the pooled OLS and GMM results for the sample of 25 OECD countries (excluding Scandinavian civil-law) for the 1st and 2nd periods in our sample (1st Period = 1996-1999. 2nd Period=2000-2003). For the GMM tests an additional two years of lags we also incorporated for each variable. 1st Period – 1996-1999 2nd Period – 2000-2003 Ordinary Least Squares Generalized Method of Moments Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Estimate t-stat p-value Estimate t-stat p-value
Intercept -3716.20 -1.1906 0.236 -1100.04 -0.4259 0.670 4085.58 1.3345 0.185 4422.59 1.6919 0.091*
FMS -8.8857 -3.1166 0.002*** -6.3027 -3.5175 0.000*** -13.1845 -4.6727 0.000*** -10.1554 -4.2301 0.000*** DR 3147.70 1.8689 0.064* 4026.78 3.5731 0.000*** 2115.31 1.3797 0.171 3594.15 2.1332 0.033** IR -24.5139 -1.0585 0.292 -23.5844 -2.8281 0.005*** -91.4029 -2.6594 0.009*** -96.1562 -3.3450 0.001*** INF 9.0553 1.2838 0.202 4.9614 1.2939 0.196 -1.0020 -0.1232 0.902 -1.3499 -0.2612 0.794 LE 39.8279 1.0064 0.317 11.5718 0.3410 0.733 -25.8080 -0.5949 0.553 -40.4680 -1.1074 0.268 FDI 0.0026 1.7799 0.078* 0.0019 1.7914 0.073* 0.0030 3.1181 0.002*** 0.0025 2.2972 0.022** EDU 12.1098 1.3071 0.194 16.4740 3.5326 0.000*** 13.0676 1.5625 0.121 18.5858 2.1114 0.035** UAI -1199.49 -1.7145 0.089* -1961.23 -5.6199 0.000*** -1909.91 -2.9155 0.004*** -2371.98 2.7301 0.006*** MSP 577.233 2.3405 0.021** 764.093 3.7667 0.000*** 318.059 1.5034 0.136 478.982 2.0257 0.043** GDP 0.0415 3.2422 0.002*** 0.0352 4.3647 0.000*** 0.0246 3.2910 0.001*** 0.0259 3.1073 0.002*** INR -810.814 -1.0963 0.275 -899.889 -2.7995 0.005*** -1294.61 -1.4591 0.148 -798.970 -0.7820 0.434 Common -277.326 -0.8212 0.413 -347.189 -1.3318 0.183 -225.117 -0.8264 0.410 -141.139 -0.4145 0.679 Civil(F) 557.477 1.4432 0.152 934.345 4.4501 0.000*** 514.039 1.3870 0.168 803.341 1.5678 0.117 Civil(G) 1095.42 2.6333 0.010*** 1142.18 4.0290 0.000*** 892.854 2.2805 0.025** 989.131 1.8714 0.061*
Adjusted 0.5447 0.5168 0.6514 0.6377 R 2
188
Diagnostic P-Value Diagnostic P-Value Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J LM Het. Hansen 0.000 0.997 0.000 0.538 Test Test Test J Test JB Test 0.453 JB Test 0.682
* Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments.
FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction, Common=Dummy for common-law societies, Civil(F)=Dummy for French civil-law societies, Civil(G)=Dummy for German civil-law societies.
189 The GMM tests seem to improve on the initial OLS results considerably. In the 1st period all variables except inflation rates (INF), life expectancy (LE), foreign direct investment
(FDI) and the common-law legal system (Common) show significance at the 1 percent level, with foreign direct investment (FDI) being significant at the 10 percent level. For the 2nd period the GMM results add to the OLS results by also finding significance in the dependency ratio (DR), level of education (EDU) and the level of minority shareholder protection (MSP)133. The adjusted R 2 for each test is 0.5168 and 0.6377 respectively, and no significance was found in regards to Hansen’s J test for overidentifying instruments.
We also conducted further tests analyzing the impact of differing mixes of legal systems within our results. In the first two tests we deviate from the method of testing conducted by Esho et al. (2004), Levine et al. (2000), and La Porta et al. (1997), since we no longer use the three dummy legal system variables of Common, Civil(F), and Civil(G). In the first test we group all the 3 civil-law nations into one variable entitled Civil; in the second test we exclude the common-law variable all together and run tests on the three civil-law nations separately, with Scandinavian civil-law represented by Civil(S); in the third test we exclude the legal dummies all together in order to obtain a better understanding of their impact on the remaining determinants for life insurance consumption. The results for these tests are presented in Tables 3.7, 3.8 and 3.9 respectively.
In the first set of tests, viewing the common-law family versus the combined civil-law family, we find results akin to those of our initial test with all the same variables
133 With each being significant at the 5 percent level.
190 exhibiting significance including the combined civil-law variable (Civil), while the common-law variable (Common) once again remains insignificant. This test was conducted to allow for comparisons to be made to Browne et al. (2000). However, we feel that this form of test, which does not separate civil-law systems, results in a degree of bias, since it means that we are testing a sample of six common-law countries versus a sample of twenty-four civil-law countries.
The second set of tests also reveal the same variables exhibiting significance as in our initial test, including the French civil-law legal system (Civil(F)) and German civil-law legal system (Civil(G)), while it appears that the Scandinavian civil-law legal system
(Civil(S)) does not influence our results in any great way.
The final set of tests that were conducted without the legal dummies exhibit the same signage for the remaining variables as in our initial tests. However, the significance of a number of variables is no longer that explicit. Here, only the foreign market share (FMS), foreign direct investment (FDI), level of education (EDU), GDP per capita (GDP), and level of investment restriction (INR) exhibit significance.
191 Table 3.7: Common-Law vs. Civil-Law Results This table presents the pooled OLS and GMM results for the common-law vs. civil-law test encompassing all 30 OECD nations, whereby the civil-law nations are grouped into one (i.e. French, German and Scandinavian civil-laws are united), over the eight-year period of 1996-2003. This basically compares the common-law nations to those that are founded on old Roman law. For the GMM tests an additional two years of lags we also incorporated for each variable. Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Intercept 2446.30 1.2878 0.199 2397.72 1.2952 0.195 FMS -10.3404 -5.4106 0.000*** -8.5608 -4.2953 0.000*** DR 1705.12 1.5475 0.123 2573.68 2.0493 0.040** IR -43.0273 -2.4016 0.017** -39.2130 -2.1093 0.035** INF 1.6089 0.4010 0.689 1.8100 0.6979 0.485 LE -6.8229 -0.2663 0.790 -19.6085 -0.6544 0.513 FDI 0.0031 4.0142 0.000*** 0.0030 2.7476 0.006*** EDU 15.5876 3.1890 0.002*** 18.2911 3.2252 0.001*** UAI -2323.88 -4.5275 0.000*** -2034.73 -2.8966 0.004*** MSP 511.030 2.7014 0.007*** 547.034 2.4872 0.013** GDP 0.0312 5.8618 0.000*** 0.0281 4.3767 0.000*** INR -233.060 -0.4422 0.659 -823.986 -1.4586 0.145 Common -299.470 -1.2977 0.196 -239.062 -0.7791 0.436 Civil 967.287 3.4030 0.001*** 813.897 2.1107 0.035**
Adjusted 0.5775 0.5642 R 2 Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.000 0.664 Test Test JB Test 0.564 * Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction, Common=Dummy for common-law societies, Civil=Dummy for civil- law societies.
192 Table 3.8: Civil-Law Results This table presents the pooled OLS and GMM results for only the civil-law sample of OECD countries (24 countries) for the eight-year period of 1996-2003. For the GMM tests an additional two years of lags we also incorporated for each variable. Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Intercept 148.931 0.0781 0.938 1727.95 0.8762 0.381 FMS -11.0678 -5.7977 0.000*** -9.3714 -4.3666 0.000*** DR 2659.92 2.5061 0.013** 3140.46 2.6926 0.007*** IR -38.9278 -2.4045 0.017** -34.0072 -2.2289 0.026** INF 4.1583 1.0813 0.281 3.9270 1.2093 0.227 LE 5.8055 0.2282 0.820 -27.0091 -0.9739 0.330 FDI 0.0028 3.8454 0.000*** 0.0029 2.8745 0.004*** EDU 12.0088 2.3092 0.022** 17.9793 3.2877 0.001*** UAI -1711.23 -3.6728 0.000*** -1632.56 -2.7530 0.006*** MSP 468.210 2.8841 0.004*** 559.349 2.8594 0.004*** GDP 0.0321 6.0615 0.000*** 0.0316 4.9715 0.000*** INR -826.416 -1.6373 0.103* -1015.26 -2.0716 0.038** Civil(F) 867.316 3.7231 0.000*** 872.537 2.9479 0.003*** Civil(G) 1307.70 5.8511 0.000*** 1085.59 3.8638 0.000*** Civil(S) 250.268 1.2017 0.231 306.948 1.0734 0.283
Adjusted 0.6034 0.5835 R 2 Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.000 0.839 Test Test JB Test 0.753 * Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction, Common=Dummy for common-law societies, Civil(F), Civil(G), Civil(S)=Dummy for French, German and Scandinavian civil-law societies.
193 Table 3.9: Results Excluding the Systems of Law This table presents the pooled OLS and GMM results for our whole sample of 30 OECD countries for the eight-year period of 1996-2003 with the exclusion of the dummy variables representing the various systems of law. For the GMM tests an additional two years of lags we also incorporated for each variable. Ordinary Least Squares Generalized Method of Moments Variables Estimate t-stat p-value Estimate t-stat p-value Intercept -1281.72 -0.5257 0.600 2498.14 1.2180 0.223 FMS -10.8079 -6.3381 0.000*** -9.7118 -6.4194 0.000*** DR -2109.53 -1.8081 0.072* 450.560 0.3793 0.704 IR -30.1883 -1.4216 0.156 -25.0561 -1.4786 0.139 INF 6.1498 1.3026 0.194 1.3966 0.5240 0.600 LE 48.9142 1.4160 0.158 -33.3853 -1.0901 0.276 FDI 0.0033 3.6336 0.000*** 0.0032 2.7497 0.006*** EDU 4.4081 0.6791 0.498 14.5536 2.2663 0.023** UAI -384.652 -1.5559 0.121 -312.359 -1.2087 0.227 MSP 41.4138 0.3380 0.736 112.217 0.8494 0.396 GDP 0.0168 2.8056 0.005*** 0.0238 3.7390 0.000*** INR -1494.25 -3.4513 0.001*** -1634.41 -3.5095 0.000***
Adjusted 0.5067 0.5066 R 2 Diagnostic P-Value Diagnostic P-Value LM Het. Hansen J 0.000 0.988 Test Test JB Test 0.251 * Indicates 10 percent level of significance, ** Indicates 5 percent level of significance, *** Indicates 1 percent level of significance. JB = Jarque-Bera test statistic for normality, LM = Lagrange Multiplier test for heteroscedasticity, Hansen J = Hansen J test for overidentifying instruments. FMS=Foreign Market Share, DR=Dependency Ratio, IR=Interest Rates, INF=Inflation Rates, LE=Life Expectancy, FDI=Foreign Direct Investment, EDU=Education (secondary enrollment rate), UAI=Uncertainty Avoidance Index, MSP=Minority Shareholder Protection, GDP=per capita Gross Domestic Product, INR=Level of investment restriction.
194 3.4.2 Analysis and Implications of Results
The results obtained support the findings of previous researchers in regards to the major determinants of life insurance demand, while also adding to the limited work done analyzing the impact that legal systems have on the level of life insurance consumption.
Within this section, we aim to analyze, compare and contrast the results presented in the previous section and discuss the major implications of these results.
The level of foreign market share was found to be significantly negatively related to the level of life insurance consumption in all the tests that were conducted. This is consistent with the argument that the domestic market is competitive to such an extent that it is able to promote demand while dissuading foreign participants from entering the market due to unattractive profit margins. This result is consistent with Browne et al. (2000) who found a negative relationship existing between foreign market share and motor vehicle insurance. Other arguments that would support such a finding would be that domestic consumers prefer domestic products as opposed to those provided by international companies due to a sense of nationalism or due to foreign products being inferior to domestic ones in terms of price and quality.
As expected, the dependency ratio is found to exert a significant positive influence over the amount of life insurance consumption within OECD countries. This supports the argument that as the number of dependents increases, the level of insurance coverage will also rise in the aim to protect the dependents from the adverse affect of the primary wage
195 earners premature death. This is consistent with the arguments of Lewis (1989) and
Browne and Kim (1993).
Interest rates are consistently found to be significant and negatively related to the level of life insurance consumption. These results are in line with both our predictions and, to an extent, the results of Outreville (1996)134. Thus, as interest rates increase, the populace on average will spend less on products such as life insurance. This relationship also supports the argument that higher interest rates deter business investment in such things as new equipment, which will result in poorer products being offered to the populace, hence decreasing consumption.
Both the level of inflation within a country and the life expectancy of a populace are shown to have no significance in relation to the consumption of life insurance products within the OECD. This is surprising in the sense that researchers such as Brown and Kim
(1993) and Outreville (1996) have discovered significance in relation to these variables and the demand for insurance products. However, a difference between their studies and ours is that our period of analysis spans close to a decade, whereas they analyzed individual years135, meaning that the influence they observed regarding inflation and life expectancy does not take time into consideration. We conclude that averaged over time, both the level of inflation and the average life expectancy of the populace exert no
134 Outreville (1996) also found a negative relationship between life insurance demand and interest rates, however he found no significance. 135 Browne and Kim (1993) ran tests on the years of 1980 (census date) and 1987, while Outreville (1996) ran tests based on information from the year 1986.
196 discernable influence over the populations’ consumption patterns of life insurance products.
With that being said, it is also important to discuss the signage of both these variables, since they do not seem to conform to the findings of others. The level of inflation is found to be positively related to life insurance consumption, which is a finding that seems to contradict those of researchers such as Browne and Kim (1993) and Outreville (1996).
During our period of study, inflation rates have remained fairly low and stable in comparison to historical rates. It can be argued that consumers continue purchasing insurance products even if the inflation rates increase, because the small increase in rates does not greatly impact the opportunity cost of being left without insurance136; nor does it greatly impact the consumers’ ability to purchase their required basket of necessities each year. It is interesting to note that when the two sub-periods were analyzed, the 1st period revealed a positive inflationary variable, while the 2nd period results revealed a negative signage for the variable. According to our arguments, this means that inflation would have caused a greater impact on an individual’s ability to purchase their basket of necessities in the 2nd period rather than in the 1st period, since the relationship between inflation and the level of insurance consumption became negative137.
The average life expectancy is found in most cases to have a negative relationship to the level of life insurance consumption, which supports the argument of life expectancy
136 Opportunity cost being that the small amount of money you save from not purchasing life insurance should inflationary rates increase slightly does not match up to the loss you may incur in the event of the untimely death of the primary wage earner.
197 being a proxy for the probability of death. However, the sign of this variable once again changes, just like with inflation, from positive to negative when we look at the two sub- periods. The 1st period results support the arguments of Outreville (1996) and Beenstock et al. (1986) in that a longer life span reflects a lower actuarial price for life insurance as well as a greater incentive for human capital accumulation, all of which will bolster demand. The change in the sign of this variable can be attributed to a change in the consumers mind in how they view their life and what they hold important.
Both the level of foreign direct investment and education have the expected significant positive relationship with the amount of life insurance consumption in all the tests that were conducted. The positive sign for education supports the arguments of researchers such as Browne et al. (1993), Browne et al. (2000), and Esho et al. (2004), in that education raises the level of risk awareness of individuals, as well as their overall level of risk aversion. While the positive sign for the foreign direct investment variable supports the notion that there is more capital for the domestic industry coming from offshore, as well as technological transfers and improved products, all of which go about to boosting insurance consumption.
When initially making our predictions, we hypothesized a positive relationship existing between the level of life insurance consumption and the UAI of Hofstede (1995). In our results, the UAI proves to be consistently significant; however the resulting sign is negative instead of the predicted positive. This result is in line with that of Park et al.
137 However, the impact was still not great enough to warrant inflation becoming a significant determinant of life insurance consumption.
198 (2002) who predicted a positive relationship between the UAI and the level of insurance demand but found the results to be negative. This can be explained by noting that
Hofstede (1983) pointed out that societies with high uncertainty avoidance, such as
Greece, may have a high level of anxiety. Park et al. (2002) then go about arguing that this may impose pressure to seek a more aggressive way to relieve the anxiety, and to people within these societies purchasing insurance may be seen to be a very passive way of dealing with risks. Working hard in order to accumulate material wealth and personal savings may be seen as a more desirable way to cope with the uncertainty rather than buying insurance. In this case, self-insurance through asset accumulation can be considered as a mutually exclusive alternative to buying life insurance. This, in turn, helps to explain the negative relationship found regarding the UAI and the level of life insurance consumption.
The level of minority shareholder protection is shown to be significantly positive in all the tests that were run, which is in line with our predictions and supports the idea of greater protection for minority shareholders encourages management to act in the best interests of all stakeholders. The benefits of this will be passed down to the consumer in terms of better products, policies and prices. The other measure of legal rights that we employ, namely, the level of investment restrictions, also confirms our predictions, in that it is negatively related to the level of life insurance consumption. This is shown to be significant in all cases except when we test common-law nations versus all three civil-law nations (14.5% significance). This supports the argument that restricting investment reduces competitiveness, increasing prices and thus reducing demand. It is also
199 interesting to note how the level of restrictions is significant in the first of our sub-period tests but not in the second. This could highlight the fact that either there were less restrictions in the period of 2000-2003, or that there were less participants trying to enter the market, which would mean that any restrictions would have less of an impact on the overall functionality of the market. The second argument supports the results that were obtained for foreign market share, in the sense that less foreign companies seek to enter the market, since the market is already competitive and the profit margins are diminished.
In line with researchers such as Esho et al (2004), Beenstock, et al. (1988), Browne et al.
(2000), Browne and Kim (1993), Campbell (1980), Lewis (1989), Outreville (1990,
1996), and Ward and Zurbruegg (2000), we find a significant positive relationship between the level of life insurance consumption and gross domestic product per capita.
This indicates that life insurance consumption is affected by the amount of economic development, economic growth and disposable income within a country.
Finally we must analyze how the level of life insurance consumption is affected by the law system in place within a nation, and whether this relationship is positive or negative.
From the set of tests conducted that exclude the system of law dummy variables (Table
3.9), it is clear to surmise that legal systems do have a significant impact on the determinants for life insurance consumption, since the significance of the remaining variables is severally dampened by the exclusion of the legal system dummies. Also, from the relatively low adjusted R 2 values, we can see that model for the determinants of life insurance consumption is no longer that strong once the legal system dummy
200 variables are removed. It can be said that without taking into account the various systems of law, the significance of the remaining determinants can not accurately be calculated and, therefore, a clear understanding of what forces drive the consumption of life insurance will not be able to be determined.
Compared to the two other major studies that focus on legal systems and their impact on the demand for insurance that were conducted by Esho et al. (2004) and Browne et al.
(2000), our results can be said to be somewhat different. In contrast to Esho et al. (2004) who discovered no significance in relation to legal systems, and Browne et al. (2000) who discovered a significant positive relationship between the common-law legal system and the demand for insurance while finding all other legal systems insignificant, we find significance only in the French and German civil-law legal systems. In nearly all our tests, we found that these two systems of law exert a positive influence on the overall consumption of life insurance products138. In addition, we find that the Scandinavian civil-law system is an insignificant determinant of the demand for life insurance products within a country. This is a new finding, since Esho et al. (2004) neglected to incorporate a sample of Scandinavian civil-law nations, while Browne et al. (2000) grouped all civil- law nations into one that was in turn entitled statutory law. Finally, we found that while common-law was an insignificant determinant of life insurance consumption, it was also a negative determinant. On the surface, this result would seem to contradict the findings of Esho et al. (2004) and Browne et al. (2000), since they both found common-law to be a
138 The exception being when the period of 2000-2003 was tested, here we found the French civil-law variable to have a significance of 11.7 percent.
201 positive determinant of insurance. However, a comparison between their results and ours should not be made lightly, as we shall now explain.
Legal systems and their impact on demand in insurance markets can be argued to be somewhat of a controversial topic, in the sense that there has been an extremely limited amount of papers published in this regard, each of which offers differing results. Esho et al. (2004) and Browne et al. (2000) to our knowledge are the only two papers that have done any work examining legal systems and their effect on insurance demand. However, both of these papers focus their attention on a different industry to that which is examined here; they analyze determinants of property and casualty insurance while we analyze life insurance. The differences in the nature of the industry examined may well be the cause for some of our differing results, but since no prior research has been conducted into how legal systems affect the demand of life insurance, we can only speculate in this regard.
One possible indicator that individuals view life and non-life insurance products differently is to look at the consumption patterns. Life insurance premiums in OECD countries have consistently been higher than that of non-life insurance, as is illustrated in the following table, with a growth rate of around 52% (49% for non-life), indicating that individuals consistently place greater emphasis on the protection of life139. In addition to the dissimilarities between the industries studied, the results of both Esho et al. (2004) and Browne et al. (2000) seem quite inconclusive and questionable. Browne et al. (2000) conducts tests in which a dummy variable is used to proxy the legal system with a value of 1 should it be common-law and 0 otherwise. As we have previously mentioned, we
202 believe that this incorporates a degree of bias since the sample of common-law nations within the OECD is too small to provide for meaningful tests when compared to civil-law nations within the OECD140. The results of Esho et al. (2004) can be said to be inconclusive in the sense that no significance is found when testing legal systems, which, unlike Browne et al. (2000), separate the civil-law system into French civil and German civil, against the demand for property-casualty insurance. The results suggest that common-law, French civil-law and German civil-law each influence the demand for property and casualty insurance in virtually the same way in terms of significance and the corresponding coefficient. This would suggest that the legal systems are virtually the same, which we know to be false141.
Table 3.10: Premium Volumes
OECD Premium Volume (USD million) in 1993, 1998, and 2003 Year Life (%Share) Non-Life (%Share) Total (100%) 1993 1010490 (56%) 792087 (44%) 1802731 1998 1204089 (59%) 824750 (41%) 2028839 2003 1533183 (57%) 1176574 (43%) 2709757 10 yr growth rate 52% 49% 50%
Source: Swiss Reinsurance Company, Sigma World Insurance articles, various issues.
139 This could in turn cause the consumer to behave differently when viewing life insurance products as opposed to non-life insurance products within a given legal system based on the differing rules and regulations the legal system has towards alternate types of insurance. 140 This bias could be rectified if the civil-law nations were further separated into the three systems of French, German, and Scandinavian, allowing for a better spread of the sample over the differing legal systems. 141 While there is considerable integration happening amongst legal systems within the world at present, a clear distinction can still be made between common-law and civil-law systems (even between the differing civil-law systems) and how they influence the financial services industries (Koch (2003)).
203 Our results regarding the significant positive influence that both French and German civil-law has on the consumption of life insurance is in line with the findings of Levine
(1998, 1999) and Levine et al. (2000), in the sense that the increased life insurance consumption can be argued to be brought about by greater financial intermediation that facilitates growth in the economy, which has been proven to be a positive determinant of insurance consumption142. It is also in line with Fukuyama’s (1995) argument that low- trust societies tend to place a greater emphasis on market based means of dealing with risk such as insurance143; as well as Smith’s (2003) arguments that civil-law nations, especially French civil-law, tend to face higher fatality rates from motor vehicle and other types of accidents144.
La Porta et al. (1998) argued that common-law nations do a better job of protecting the rights of policyholders and shareholders. This can be extended to the insurance industry only weakly, since policyholders cannot be deemed to be creditors in the same sense that
La Porta et al. (1998) meant. Creditors usually extend credit after thoroughly checking the credibility of a debtor and are responsible for their decisions – however, it is difficult to expect non-professional policyholders, such as individuals, to assume this responsibility in full (Yasui (2001)). Common-law nations have been argued by others to provide poorer protection for policyholders than civil-law nations. The Wikipedia encyclopedia states the following:
142 Researchers supporting this argument include: Beenstock, Dickenson and Khajuria (1988), Campbell (1980), Lewis (1989), Outreville (1990, 1996), and Ward and Zurbruegg (2000). 143 A large portion of French and German civil-law states fall into the classification of ‘low-trust’ societies, while those of English common-law fall into ‘high-trust’ societies. 144 This would increase people’s willingness to purchase life insurance products due to the realization that they face greater mortality rates resulting from accidents.
204 One of the differences between civil- and common-law is that the civi- law
systems go further in implementing the principle of freedom of contract, by
specifically upholding almost all contractual promises and by enforcing
penalty clauses.
Should an insurance company go into bankruptcy, the whole insurance industry would be affected by how the legal system handles compensation to existing policyholders. If an ineffectual job is done, then the public may be discouraged from seeking insurance in the future and look towards other forms of protection (Yasui (2001)). The common-law lawyer Koch (2003) also supports the seeming superiority of the civil-law system in the way it brings about an independent judiciary145. As a result of all this, the negative relationship between the common-law variable and the level of life insurance consumption can be explained by the fact that common-law nations do not offer adequate protection to the policyholder in cases where the insurance company becomes insolvent, which will then facilitate lower life insurance consumption in the future due to lack of trust in the judicial system.
Now that we have explained how it is possible to obtain a negative sign for the common- law variable, we must look at why it is insignificant. Apart from the arguments just presented which go about explaining the variables insignificance, an alternate argument would be that the level of uncertainty within OECD common-law nations is not high enough to warrant excessive spending on life insurance products as compared to the
205 French and German civil-law nations where uncertainty is much greater. An indicator of this would be to view the corresponding Hofstede UAI figures146. The average for common-law nations is around 44UAI, while for German and French civil-law nations it is 75UAI and 88UAI respectively. This clearly indicates that the level of uncertainty within common-law nations is fairly low and, hence, it can be argued that the populations’ requirement for protection in the form of life insurance would also be much lower147. It must be noted that this is only a partial contributor in explaining the insignificance of the common-law variable and is not necessarily a valid explanation for the significance found in the French and German civil-law variables148. Another explanation for the insignificance evident in the common-law variable can be based on
Smith’s (2003) argument that common-law nations face much lower fatality rates as a result of accidents than civil-law nations do, thus meaning there is less of a need to protect ones life against such an occurrence.
145 Koch (2003) states that in addition to this, other major advantages are the way judges mediate cases and the fact that civil-law nations such as Germany and France (unlike the U.S.) have detached certain tasks and created specialized court systems to deal with them. 146 While arguing similar points, Hofstede’s UAI figures are not a clear reflection (in quantitative terms) of the arguments presented by Fukuyama (1995). 147 A similar argument can be used to explain the insignificance of the Scandinavian civil-law nations, since their average UAI is approximately 42. 148 Previously we argued that the negative relationship between UAI and the level of life insurance consumption may be the cause of high anxiety nations looking for more aggressive means than insurance coverage to decrease their level of anxiety (Park et al. (2002)). For this reason we cannot use a similar argument to that just presented for the insignificance of the common-law variable for French civil-law nations, since on average they have the highest UAI. Based on the previous research of Hofstede (1995) and Park et al. (2002) we conclude that there is a certain UAI range in which the level of insurance consumption will increase while outside of this range the level of consumption will begin to decrease for two reasons. Firstly, if the country’s UAI is low the level of consumption decreases because the need for protection through insurance products is minimal, and secondly, if the country’s UAI is excessive, then the high anxiety argument comes into play that in turn causes a decrease in the consumption of insurance products.
206 3.5 CONCLUSION
3.5.1 General Conclusion
In recent times, the life insurance industry has received a great deal of attention from both researchers and policymakers alike, given the staggering growth the industry has achieved in many countries. The life insurance industry is an extremely important part of the economy, since it provides a means by which the populace can alleviate uncertainties and risks and at the same time provide a means of long-term savings that can assist local governments in relieving pressures from outdated and exhausted pension schemes. With the globalization of industry, the life insurance industry has received added attention in the previous decades, especially in emerging markets where growth prospects are highest.
While previous studies have focused attention on a few key determinants of life insurance consumption, none have scrutinized such a wide range of determinants as we accomplish within this study. Furthermore, no previous study has analyzed the influence of national legal systems on the level of life insurance consumption as we do here, which can be argued to be of paramount importance to regulators and policymakers.
This study aims to fill a gap in the existing literature by investigating a number of key determinants of life insurance consumption within OECD nations with a primary focus on the influence of corporate governance and shareholder protection. To date, existing literature has been lacking in the attempt to explain how consumption levels are affected
207 by the legal systems in place within a given country. Only Browne et al. (2000) and Esho et al. (2004) have made attempts at shedding insight into this relationship, however their primary focus was on the non-life insurance industry, and both sets of results were far from conclusive. In this study, we provide a clearer insight into the effects of legal systems on the levels of life insurance consumption, supported by sound reasoning.
This study examines the determinants of life insurance consumption over the 30 OECD countries, which include both developed and emerging markets. Our results complement certain areas of existing literature on life insurance demand and international trade in insurance services, while also providing additional insight in key areas that have hitherto been neglected by other researchers. We utilize a cross-sectional time-series approach by making use of panel analysis. In particular, we incorporate the use of the GMM estimation technique which overcomes inherent weaknesses existing within the simple
OLS approach.
The results of this study provide several contributions to our understanding of the determinants of life insurance consumption and confirm the findings of some of the previous literature in this area over which debate has since ensued. In this study, eleven hypotheses have been established and tested. It was found that the foreign market share of the life insurance industry is a negative determinant of consumption, promoting the notion that domestic markets within the OECD are competitive enough to facilitate consumer demand, while at the same time dissuading foreign participants through unattractive profit margins. In line with previous literature, it is found that the number of
208 dependents is a positive determinant, while the level of interest rates is a negative determinant of life insurance consumption. In contrast, the level of inflation rates and the average life expectancy of the populace are found to exert no significant influence on life insurance demand within the OECD. The level of foreign direct investment and education both yield significant results in line with our predictions and the consensus amongst prior researchers. This indicates that life insurance consumption is bolstered as a result of the technological transfers, additional capital and improved products that are made available through foreign direct investments, and the additional risk awareness and changes in the levels of risk aversion that education brings about. The results regarding the level of uncertainty within a community and the corresponding influence it has on the level of consumption do not meet with the hypothesis presented. Instead, the results indicate that societies with considerably high uncertainty avoidance, past a given UAI level, experience a level of anxiety so high that insurance is seen as a passive way of coping with the additional uncertainty and a more aggressive approach of material wealth accumulation, and personal savings is deemed more appropriate. The two measures of legal rights, namely, the level of minority shareholder protection and the level of investment restriction, meet with expectations in that they are positive and negative determinants of life insurance demand respectively. These results highlight the significant influence that legal rights have on the level of prices and in turn the consumption patterns of the populace. In a similar manner, we find the expected positive relationship between the level of consumption and GDP per capita, indicating that economic growth, economic development and the level of disposable income within a nation are all valuable determinants. Finally, it is also found that the civil-law legal systems are positive
209 determinants of the level of life insurance consumption, aligned with the arguments that greater financial intermediation promotes increased consumption through facilitating economic growth, and that low trust societies place a greater emphasis on market based means, such as insurance, of dealing with risk and uncertainty. These results proved valid for only the French and German civil-law systems, while the Scandinavian civil-law system demonstrated no significant influence on consumption levels. Surprisingly, the common-law legal system was found to be an insignificant negative determinant, in contrast to both Browne et al. (2000) and Esho et al. (2004). One possible supporting argument is that common-law nations do not offer adequate protection to life insurance policyholders when an insurance company becomes insolvent, which in turn leads to mistrust of the whole insurance industry within the community when policyholders do not recoup their investments. Another argument is that the level of uncertainty within common-law nations is so low that the requirement for protection in the form of life insurance would be minimal. However, we cannot attest to which of these arguments holds sway, since it was not within the constraints of this study.
The results from this study lead to several policy implications. The most important of which is the need for regulators and governing bodies to ensure the protection of shareholders and policyholders in terms of enforcing contracts, allowing foreign access to domestic markets and ensuring the honesty of existing insurers regarding product compositions, and company investment and management strategies in order to promote a positive atmosphere with regards to insurance. If need be, aspects of differing systems of law should be implemented within the current system if it can assist in enhancing
210 effective protection and bolstering public confidence. Economic stability must also be maintained in terms of continuing growth and preserving manageable levels of interest and inflationary rates if the augmentation of a healthy life insurance industry is sought.
All in all, effective corporate governance is vital for the continued success and growth of the life insurance industry, and in turn the entire economy, within OECD nations149.
3.5.2 Avenues for Further Study
While this study covers the key determinants of life insurance consumption and at the same time verifies previous topics of discussion and provides further insight into the life insurance industry and its key determinants than previous studies have, it also opens further avenues of research. Most importantly of all, further research is warranted into the effects that legal systems have on the life insurance industry, not just in OECD nations but also outside of them. Specifics of legal systems that were not encompassed in this study and their impact on life insurance consumption should be further analyzed in order to provide clear insight into what parts of corporate governance within each legal system are most influential in promoting the growth of the industry. This would, in turn, assist policymakers in passing regulations that will ensure the successful growth of the life insurance industry and the economy on a whole. This study focuses mainly on the determinants of demand. Studies that cover the supply side of the equation, which this study does not cover, would also be able to provide useful research avenues. A natural extension of this study would also be to examine the determinants of the entire insurance
149 The functioning of an efficient and effective life insurance industry promotes economic growth. Should the industry crumble, dire consequences can be forecast for the economy on a whole.
211 industry within the OECD, with a particular focus on corporate governance, and see how the results compare to those obtained when life and non-life insurance industries are analyzed separately. Finally, minimization of trade barriers and promotion of foreign direct investments will bring benefits to the life insurance industry in the form of technological transfers, additional capital and improved products, which assist in raising consumption levels and promoting industry-wide growth.
212 CHAPTER 4
CONCLUSION
"All scholarship, like all science, is an ongoing, open-ended discussion in which all conclusions are tentative forever, the principal value and charm of the game being the discovery of the totally unexpected."
- Hugh W. Nibley. Professor emeritus of ancient history & classical languages at
BYU, acclaimed author, 14 languages from "Common Carrier".
4.1 INTRODUCTION
International trade in financial services has gained much momentum in recent decades with the deregulation and liberalization of many national markets, decreased distribution costs resultant from technological advances, changing consumer attitudes towards foreign products, and the support of multinational agencies such as the WTO. With such growth, there is an ever-increasing need for added research, based on current data, into issues relating to the factors that influence the demand for international financial services. This thesis identifies a couple of important topics in relation to international banking and risk and insurance within the area of international financial services. To that extent,
213 significant contributions are made to the existing literature on global bancassurance markets, and the determinants of life insurance consumption. Extending on prior research in the areas of banking, and insurance, Chapter 2 examines demographic, social and macroeconomic determinants of the demand for bancassurance across the globe. The remainder of this thesis, encompassed in Chapter 3, scrutinizes the key determinants plus impact that systems of law have on the level of life insurance consumption within OECD nations.
Within this chapter, the following two sections summarize the main findings and contributions of the two topics examined within this thesis, while Section 4.4 concludes this thesis by making suggestions in relation to possible avenues for further research.
4.2 THE DETERMINANTS OF BANCASSURANCE
The results of this chapter provide several significant contributions to existing literature on the determinants of bancassurance. The study investigates a number of key determinants of bancassurance demand using a sample of both developed and underdeveloped countries from across the globe. A number of different tests are conducted to establish whether the relationships found are consistent over sundry samples. We utilize a cross-sectional time-series approach by making use of panel analysis. We make methodological improvements on a number of previous studies
214 through incorporating a pooled-GMM estimation technique in unison with simple OLS regressions. The GMM estimator exploits the time-series variation in data, allowing for the inclusion of lagged variables as regressors, and controls for endogeneity of all explanatory variables. For GMM purposes we made use of internal instrumental variables. The GMM results highlight the significance of all the eight variables examined with their expected signs.
In turn, it is established through this chapter that company risk, company size, reductions in expenses, increases in revenues, the size of the national banking industry, the level of deregulation existent within the country, changes in gross national income, and changes in inflationary rates are all significant determinants of bancassurance demand. In conjunction with the findings of researchers such as Boyd et al. (1993) and Estrella
(2001), it is found that nontraditional banking activities help to reduce the average risks faced by banks, signifying that the diversification benefits that exist from the unions of banks and insurers, with their different risk structures, outweigh the risks brought about in regards to integrating the two business areas. Both the decrease in costs, through spreading of costs amongst different operations, and increase in revenues, from the new fee-based income streams, resulting from the implementation of insurance operations within the bank are found to have positive implications in regards to banking organizations beginning bancassurance operations. Evidence supporting the positive relationships of consumer demand, as proxied by GNI per capita, the size of the company, which is indicative of the breath of the company’s customer base, and the size of the national banking industry are also provided and consistent with existing literature.
215 The inflation rate is shown to be a negative determinant, supporting the argument that the subsequent reduction in disposable income resulting from an inflationary increase outweighs company gains made through increased insurance premiums and discounts of insurance liabilities. The level of financial deregulation, on average, is found to be a positive determinant of bancassurance, highlighting the fact that more regulated nations around the world have weaker bancassurance industries. However, evidence also exists that more regulated nations such as Estonia afford greater potential to bancassurers than mature bancassurance nations such as France. Finally, it is also found that the legal system existing within the host country will be influential on the overall success of the bancassurance industry, since it impacts directly on the determinants of bancassurance.
The results of this study lead to several important policy implications. The significance of the deregulation proxy indicate that promoting the further deregulation of financial services industries will help enhance the development and growth of both the banking and insurance industries through allowing the expansion of distributional channels such as bancassurance which combine the knowledge, skills and experiences of both industries. Through improved distributional methods and innovative products that cater to changing consumer tastes, consumption levels can be increased benefiting both bank and insurer alike. Finally, companies that fully utilize the changing atmosphere towards the provision of differing financial products and services are set to reap the benefits in terms of diversified risk, reduced costs, and increased revenues, enabling them to be more competitive on both a domestic and global front than their counterparts that fail to
216 establish themselves in this new style of industry that promotes the provision of multiple financial services from across industries through a common organizational body.
4.3 LAW AND THE DETERMINANTS OF LIFE INSURANCE IN OECD
COUNTRIES
This chapter investigates the determinants that influence the level of life insurance consumption within OECD nations and has several important extensions to existing literature on both the demand for life insurance and the impact of legal systems on the financial markets. We make use of a number of demographic, social and macroeconomic variables that we believe form the foundation for explaining the movements in consumer demand for life insurance products as well as establishing the link between consumption levels and the system of law in place within a country. Yet again, we make use of the
GMM instrumental variable estimation technique in conjunction with the OLS method of testing when analyzing our specified model. The evidence provided by this chapter documents the relevance and validity of the variables used within past research articles while also providing the first link between the system of law within a country and the demand for life insurance products and services.
The findings of this study suggest that systems of law do indeed influence life insurance consumption patterns with the French and German civil-law legal systems proving to be
217 positive determinants aligned with the arguments that greater financial intermediation promotes increased consumption through facilitating economic growth, and that low trust societies place a greater emphasis on market based means, such as insurance, of dealing with risk and uncertainty. In contrast to two previous studies analyzing the impact of legal systems on non-life insurance products, it was found within this study that the common-law legal system is an insignificant negative determinant of life insurance demand. Numerous explanations can be provided for this relationship including difference of industries examined, lack of adequate protection in common-law nations for policyholders in cases of company insolvency leading to mistrust, and low uncertainty levels within common-law nations (as measured by the Hofstede UAI) reducing the need for protection through insurance products. However, we cannot attest to which of these arguments holds sway, since it was not within the constraints of this study.
The results of this study provide contributions to our understanding of the determinants of life insurance consumption and support the findings of a number of previous research papers. It was found that all variables examined proved to be significant except for the inflation rate and life expectancy variables150. Both the level of foreign direct investment plus the level of government investment restriction proved to be significant, indicating that important benefits can be derived from the resources that flow in from abroad and any restriction of such inflows will be damaging to the industry on a whole. Evidence was also found supporting the positive relationships of GDP per capita that reflects the levels of economic growth, economic development and disposable income, education which acts as a proxy for the level of risk aversion, the level of minority shareholder
218 protection that assists in measuring the level of legal rights within a nation, and the number of dependents, all of which are in line with the findings of previous literature.
The results regarding the level of uncertainty within a community and the corresponding influence it has on the level of consumption do not meet with the hypothesis presented.
Instead the results indicate that societies with considerably high uncertainty avoidance, past a given UAI level, experience a level of anxiety so high that insurance is seen as a passive way of coping with the additional uncertainty and a more aggressive approach of material wealth accumulation and personal savings is deemed more appropriate. Finally, the level of interest rates and the foreign market share of the life insurance industry are both negative determinants of consumption. The relationship that foreign market share has with the level of consumption supports the notion that domestic markets within the
OECD are competitive enough to facilitate consumer demand, while at the same time dissuading foreign participants through unattractive profit margins.
The results of this study highlight the need for regulators and governing bodies to ensure the protection of shareholders and policyholders in terms of enforcing contracts, allowing foreign access to domestic markets and ensuring the honesty of existing insurers regarding product compositions, and company investment and management strategies in order to promote a positive atmosphere with regards to insurance. Failure to do such will have dire consequences resulting from a plunge in consumer confidence, even if one life insurance company becomes insolvent and existing stakeholders fail to recoup their investments. If need be, aspects of differing systems of law should be implemented within the current system if there is a chance that it can assist in enhancing effective
150 The level of significance varies amongst the differing tests conducted. 219 protection and bolstering public confidence151. Economic stability must also be maintained in terms of continuing growth and manageable levels of interest and inflationary rates if the augmentation of a healthy life insurance industry is sought. All in all, effective corporate governance is vital for the continued success and growth of the life insurance industry, and in turn the entire economy, within OECD nations. Finally, trade barriers should be minimized and foreign direct investment promoted in order to allow for technological transfers, additional capital and improved products to enter the domestic market, thereby promoting growth within the industry.
4.4 SUGGESTIONS FOR FURTHER RESEARCH
While this thesis goes a long way in filling gaps within existing literature on the determinants of bancassurance and law and the determinants of life insurance consumption, in other ways it has opened up numerous avenues for further research which we now aim to highlight.
Chapter 2 – Possible avenues for further research:
Chapter 2 analyzes the determinants of bancassurance across a sample of 27 developed and developing countries using current data. While this paper greatly ads to existing
151 This has been gradually happening in recent times, and there is no longer any country that purely has one system of law totally untouched by the influences of others.
220 literature, it also opens a multitude of avenues for further research into global bancassurance, due in large part to the increased popularity of bancassurance and minimal amount of research conducted thus far in explaining the determinants of bancassurance152. The most immediate extension possible would be to expand the sample size and ensure that the relationships found within our study still holds true. Another extension would be to include certain variables that our study omitted due to the unavailability of data, which may be influential in explaining the demand for bancassurance. Possible influential variables may include government social security expenditure, the level of education within the country, taxation rates and benefits, and the life expectancy of the populace. Exchange rates could also be incorporated as an explanatory or even instrumental variable in order to account for the slight measurement error they pose to such a study as ours (outlined in Section 2.7.1). Further research should also be made into the impact that systems of law have on the determinants of bancassurance. Our results highlight the fact that the legal system within a nation does influence the demand for bancassurance – however, we did not test the significance of said relationship.
This study focuses on the determinants of demand for bancassurance. However, the supply-side of the equation should not be neglected and should in future studies also be analyzed. A segmentation of the bancassurance industry through which the determinants of life insurance and non-life insurance demand and supply are scrutinized separately is also warranted. For a long period of time, the major revenue source for banks operating
152 The majority of past research papers have focused on the risk and wealth effects resulting from the merger of banks and insurance companies while not focusing on the bigger picture.
221 insurance arms has been life insurance, resulting from the similarity of existing banking products to life insurance products. However, in recent times, banks have begun expanding into the non-life insurance area more aggressively and are set to continue this trend. Hence, it would prove beneficial to examine what determined the success of each type of insurance operation provided by banks. Finally, a logical extension would involve applying similar methodology and experimentation techniques as used within this study in order to examine the determinants of assurebanking. Results from such further studies may then lead in concluding which is the most viable and effective union of banks and insurance companies.
Chapter 3 – Possible avenues for further research:
Chapter 3 succeeds in identifying the determinants of life insurance consumption within
OECD nations using current data, and in particular expresses the relationship inherent between consumption patterns and the systems of law in place within a given country.
This paper is extremely beneficial in the sense that it is the first study examining the impact that systems of law have on life insurance consumption, as well as further scrutinizing a number of other key demographic, social and macroeconomic variables using current data. This said, it also opens many research avenues, since it is the first paper to examine a number of key relationships. Numerous extensions can be made in regards to the effects that legal systems have on the level of life insurance consumption.
Data should be analyzed encompassing non-OECD nations, developed and developing nations, as well as an overall global study. Specifics of legal systems that were not
222 encompassed in this study and their impact on life insurance consumption should also be further analyzed in order to provide clear insight into what parts of corporate governance within each legal system are most influential in promoting the growth of the industry.
This would, in turn, assist policymakers in passing regulations that will ensure the successful growth of the life insurance industry and the economy on a whole.
Additionally, studies should be conducted detailing the impact that systems of law have on the overall insurance industry, both within and outside of the OECD, as well as further studies into the non-life insurance industry153. Finally, future papers that place greater emphasis on the supply-side determinants of life insurance consumption within OECD nations should also be prepared in order to provide researchers with the complete picture in regards to what determines the demand for life insurance products and services.
153 Currently, a limited amount of studies exist examining the relationship between legal systems and non- life insurance consumption, and the results of these studies can be said to be far from conclusive.
223 APPENDIX 1
Data Sources and Descriptions
Variable Source Description BA Compustat + Bank operated insurance premium income as a Company Annual percentage of total bank income for the period 1 . Reports RISK DataStream Standard deviation of the banks daily share price measured on an annual basis. SIZE Compustat + The log of the banks total assets. Company Annual Reports EXP Compustat + Measured as the annual percentage change of the ratio Company Annual of the banks total expenses over the banks total Reports revenues. REV Compustat + Measured as the annual percentage change of the ratio Company Annual of the banks revenues over the banks total assets. Reports SIZE(N) Each countries Annual percentage change in the total banking assets respective National of a country. Bank website REGL World Ranking from 1 to 60 symbolizing the overall Competitiveness competitiveness of a country based on a number of Yearbook (IMD) pre-specified factors 2 . GNI World Bank Annual percentage change in a country’s GNI per capita based on the Atlas Method 3 . INFL World Economic Annual average inflation rate for a country. Outlook (IMF)
1Premium Income does not include any form of commissions received from insurance operations. 2For more information on the calculation of rankings please visit the IMD website. 3In calculating GNP, GNI and respective per capita figures in U.S. dollars for certain operational purposes, the World Bank uses a synthetic exchange rate commonly called the Atlas conversion factor. The purpose of the Atlas conversion factor is to reduce the impact of exchange rate fluctuations in the cross-country comparison of national incomes. The Atlas conversion factor for any year is the average of a country’s exchange rate (or alternative conversion factor) for that year and its exchange rates for the two preceding years, adjusted for the difference between the rate of inflation in the country and that in the Group of Five (G-5) countries (France, Germany, Japan, the United Kingdom, and the United States). A country’s inflation rate is measured by the change in its GNP deflator.
224 Country Breakdown based on Test Samples
Argentina b Estonia a b Italy a b Peru c Taiwan b Australia c France a b Japan b Philippines c Turkey a b Belgium a b Germany a b Luxembourg a b Portugal a b U.K. a c Canada c Greece a b Malaysia c South Africa c U.S.A. c Czech Rep. a b Hong Kong c Netherlands a b Sweden a b Denmark a b Ireland a c Norway a b Switzerland a b a Indicates countries that comprise the Euro1 sample. The Euro 2 sample is these same countries less the U.K. and Ireland. b Indicates countries that have a civil-law legal system. This unifies the three systems of French civil-law, German civil-law, and Scandinavian civil-law. c Indicates countries that have a common-law legal system.
Note: Some nations such as the Philippines, Turkey and South Africa employ a combination of common- law and civil-law within their legal systems. Here we only allocate one system of law to each of these countries based on which legal system is argued to have more influence within the nation during the time of our study.
225 Summary Statistics for 2003 based on Test Samples
PREM RISK SIZE EXP REV SIZE(N) REGL GNI INFL Whole Sample Mean 0.1354 28.2698 10.9818 -0.0294 -0.0129 2.4798 19.5753 0.0971 2.7644 Median 0.0659 2.4406 11.1766 -0.0166 -0.0554 1.1996 16.0000 0.1054 2.3000 St. Dev. 0.1510 146.7488 2.1139 0.1284 0.2632 3.5402 17.4864 0.0428 4.3111 Correlation 1 1.0000 -0.1330 0.1300 0.0999 -0.0653 0.3093 0.1288 0.2471 -0.1362 Observations 73 73 73 73 73 73 73 73 73 Whole Sample 2 2 Mean 0.1600 35.3856 11.1320 -0.0358 0.0043 2.9876 24.7895 0.1051 2.8947 Median 0.1037 2.4300 11.5231 -0.0195 -0.0482 2.7968 22.0000 0.1080 2.2000 St. Dev. 0.1615 165.6874 2.1788 0.1428 0.2945 3.8622 16.3366 0.0455 4.8802 Correlation 1.0000 -0.1719 0.1196 0.1388 -0.1096 0.2481 -0.0595 0.1567 -0.1635 Observations 57 57 57 57 57 57 57 57 57 European Sample Mean 0.1647 19.9690 11.8992 -0.0253 -0.0673 3.4433 25.0811 0.1236 3.0865 Median 0.1415 3.4899 12.5737 -0.0166 -0.0579 3.0252 22.0000 0.1150 2.1000 St. Dev. 0.1520 39.4037 1.8976 0.0823 0.1003 4.4479 12.5839 0.0297 5.4465 Correlation 1.0000 -0.2922 0.1946 0.0626 -0.0623 0.1641 -0.1128 -0.0379 -0.1855 Observations 37 37 37 37 37 37 37 37 37 Civil-Law Sample Mean 0.2126 52.3965 5.1885 -0.0496 -0.0067 3.4374 26.1724 0.1163 2.9379 Median 0.1883 2.9253 5.4428 -0.0164 -0.0634 2.7968 23.0000 0.1152 2.2000 St. Dev. 0.1713 231.2464 0.6917 0.1837 0.3836 5.1830 14.3728 0.0554 4.9137 Correlation 1.0000 -0.2562 0.3073 0.1731 -0.2309 0.2928 -0.2710 0.1737 -0.2659 Observations 29 29 29 29 29 29 29 29 29
Above statistics are presented in raw figures. 1 Correlation with respect to PREM. 2 Refers to the whole sample less those observations from the USA.
226 APPENDIX 2
Data Sources and Descriptions
Variable Source Description LFI World Insurance Real per capita life insurance consumption. (SIGMA) FMS Insurance Statistics Market share of (foreign controlled undertakings) and Yearbook (OECD) (branches/agencies of foreign controlled undertakings) in total domestic business expressed as a percentage. DR Labor Force Sum of the population aged under 15 and over 64 (OECD) (dependents) divided by those aged between 15 and 64 (working aged population). IR OECD Factbook + Average of daily or monthly long term interest rates ECB Convergence expressed as a percentage2. Reports1 INF OECD Factbook Annual average inflation rate for a country. LE OECD Factbook Average number of years of life remaining to a person at a particular age, based on a given set of age-specific mortality rates. FDI Insurance Statistics Foreign investment + Foreign controlled companies + Yearbook (OECD) Branches & agencies of foreign companies. EDU World Bank Net secondary enrollment rate expressed as a percentage3. UAI ITIM International Uncertainty Avoidance Index as measured by Geert Website4 Hofstede. MSP La Porta et al. Dummy variable with values of 0 or 15. (1998) GDP OECD Factbook Per capita GDP. INR Golub (2003) Foreign direct investment restriction6. Common La Porta et al. Dummy. 1=Common-law, 0=other. (1998) + The World Factbook (CIA) Civil(F) La Porta et al. Dummy. 1=French Civil-law, 0=other. (1998) + The World Factbook (CIA)
227 Civil(G) La Porta et al. Dummy. 1=German Civil-law, 0=other. (1998) + The World Factbook (CIA)
1Used in order to obtain the figures for Hungary and Poland. 2Interest rates on bonds (at least 10 years) whose capital repayment is guaranteed by government. Averages of daily rates for all countries except Japan, Australia, France, Iceland, Ireland, Switzerland and the United States. 3Net enrollment rate, secondary is the number of pupils in the theoretical age group for secondary education enrolled in secondary education expressed as a percentage of the total population in that age group. (Data Source: UNESCO Institute for Statistics). 4www.geert-hofstede.com 5“Equals one if the company law or commercial code grants minority shareholders either a judicial venue to challenge the decisions of management or of the assembly or the right to step out of the company by requiring the company to purchase their shares when they object to certain fundamental changes, such as mergers, asset dispositions, and changes in the articles of incorporation. The variable equals zero otherwise. Minority shareholders are defined as those shareholders who own 10 percent of share capital or less.” La Porta et al. (1998). 6The total FDI investment restriction within the economy is calculated through separately viewing the restrictions evident in: Business Services (Legal, Accounting, Architecture, Engineering), Telecommunications (Fixed, Mobile), Construction, Distribution, Finance (Insurance, Banking), Hotels and Restaurants, Transportation (Air, Maritime, Road), Electricity, and Manufacturing; and then averaging them out based on how much each sector contributes to the whole economy.
Country Breakdown based on Systems of Law
Australia a Finland d Ireland a Netherlands b Spain b Austria c France b Italy b New Zealand a Sweden d Belgium b Germany c Japan c Norway d Switzerland c Canada a Greece b Korea (South) c Poland b Turkey b Czech Rep. c Hungary c Luxembourg b Portugal b U.K. a Denmark d Iceland d Mexico b Slovak Rep. c U.S.A. a a Indicates countries that utilize the English common-law legal system. b Indicates countries that utilize the French civil-law legal system. c Indicates countries that utilize the German civil-law legal system. d Indicates countries that utilize the Scandinavian civil-law legal system.
Note: Many countries have begun implementing aspects of differing systems of law into their current legal system. As such, the distinction as to which system of law governs a nation may not be clear-cut. For example, the Greek legal system can be argued to be heavily influenced by both the German and French civil-law systems. The above classifications are thus based on which system of law exerts the greatest influence within the country during the period of this study.
228 REFERENCES
• (Apr 1997): “The New American Universal Bank”, Harvard Law Review,
Volume 110, No. 6: 1310-1327.
• (Apr 2004): “European Bancassurance Review”, Morgan Consulting.
• (May 2005): “European Bancassurance Review 2005”, Milliman Consultants and
Actuaries.
• 1995: “The Supervision of Financial Conglomerates”, A report by the Tripartite
Group of Bank, Securities and Insurance Regulators.
• 2001: “Creating Breakthrough Value in Life Insurance. Moving From Vertical to
Virtual Integration”, A.T. Kearney.
• 2003: “Interview with the Regulator – Bancassurance in Korea: Future
Supervisory Directions”, Asia Insurance Review.
• Agrawal Abhishek (Aug 2002): “Distribution of Life Insurance Products in
India”, Insurance Chronicle.
• Agrawal Mohan (2002): “Distributing Insurance in India: The Tata AIG
Experience in India”, Tata AIG Insurance.
• Allen Linda & Jagtiani Julapa (2000): “The Risk Effects of Combining Banking,
Securities, and Insurance Activities”, Journal of Economics and Business,
Volume 52: 485-497.
• Allen Linda & Rai Anoop (1996): “Operational efficiency in banking: An
international comparison”, Journal of Banking & Finance, Volume 20: 655-672.
229 • Amel Dean, Barnes Colleen, Panetta Fabio & Salleo Carmelo (Jan 2004):
“Consolidation and efficiency in the financial sector: A review of the
international evidence”, Journal of Banking & Finance, Volume 28: 2493-2519.
• Anderson D.R. and Nevin D. (1975): “Determinants of Young Marrieds Life
Insurance Purchasing Behaviour: An Empirical Investigation”, Journal of Risk
and Insurance, Volume 42: 375-387.
• Arellano M. and Bond S. (1991): “Some Tests of Specification for Panel Data:
Monte Carlo Evidence and an Application to Employment Equations”, The
Review of Economic Studies, Volume 58, No. 2: 277-297.
• Arellano M. and Bover O. (1995): “Another Look at the Instrumental Variable
Estimation of Error-Components Models”, Journal of Econometrics, Volume 68:
29-51.
• Auerbach A.T. and Kotlikoff L.J. (1989): “How Rational is the Purchase of Life
Insurance?” National Bureau of Economic Research, Working Paper No. 3063.
• Aviva Plc. (2002): Annual Report.
• Babbel D.F. (1981): “Inflation, Indexation, and Life Insurance Sales in Brazil”,
Journal of Risk and Insurance, Volume 48: 111-135.
• Battista A. and Guzzo V. (2005): “Life Insurance Companies and the Economics
of Demand for Long-Term Bonds: the Italian Case”, Morgan Stanley Research,
April 2005.
• Beck T. and Webb I. (2002): “Determinants of Life Insurance Consumption
across Countries”, The World Bank, Development Research Group, Working
Paper No. 2792.
230 • Beck T. and Webb I. (2003): “Economic, Demographic, and Institutional
Determinants of Life Insurance Companies Across Countries”, The World Bank
Economic Review, Volume 17: 51-88.
• Beck T., Levine R. and Loayza N. (2000): “Finance and the Sources of Growth”,
Journal of Financial Economics, Volume 58: 261-300.
• Beenstock M., Dickinson G. and Khajuria S. (1986): “The Determination of Life
Premiums: An International Cross-Section Analysis 1970-1981”, Insurance
Mathematics and Economics, Volume 5: 261-270.
• Beenstock M., Dickinson G. and Khajuria S. (1988): “The Relationship between
Property-Liability Insurance Premiums and Income: An International Analysis”,
The Journal of Risk and Insurance, Volume 55, No. 2: 259-272.
• Benoist Gilles (2002): “Bancassurance: The New Challenges”, The Geneva
Papers on Risk and Insurance, Volume 27, No. 3: 295-303.
• Benston George J. (Summer 1994): “Universal Banking”, The Journal of
economic Perspectives, Volume 8, No. 3: 121-143.
• Berberich Kerstin (May 2000): “Bancassurance in Europe: Concept and Market
Overview”, General & Cologne RE.: Risk Insights For Life & Health Insurance
Executives, 12-16.
• Berger Allen N. & Humphrey David B. (1997): “Efficiency of financial
institutions: International survey and directions for future research”, European
Journal of Operational Research, Volume 98: 175-212.
231 • Berger Allen N., Humphrey David B. & Pulley Lawrence B. (1996): “Do
consumers pay for one-stop banking? Evidence from an alternative revenue
function”, Journal of Banking & Finance, Volume 20: 1601-1621.
• Berger Philip G. & Ofek Eli (1995): “Diversification’s effect on firm value”,
Journal of Financial Economics, Volume 37: 39-65.
• BernHeim B.D., Forni L., Gokhale J. and Kotlikoff L.J. (2001): “The Mismatch
Between Life Insurance Holdings and Financial Vulnerabilities: Evidence for the
Health and Retirement Survey”, National Bureau of Economic Research,
Working Paper No. 8544.
• Blom F., Nicholson J., Kuenen J.W. and Hekster J. (2004): “Opportunities for
Action in Financial Services: Grow with the Flow in Insurance”, The Boston
Consulting Group, BCG
• Blundell R. and Bond S. (1998): “Initial Conditions and Moment Restrictions in
Dynamic Panel Data Models”, Journal of Econometrics, Volume 87: 115-143.
• Blundell R. and Bond S. (1999): “GMM Estimation with Persistent Panel Data:
An Application to Production Functions”, The Institute for Fiscal Studies,
Working Paper Series No. W99/4.
• Bokans J. (2000): “Life Insurance Market Development in Baltic Countries:
Pension Reforms – Potential Market for Life Insurers”, OECD Publication.
• Bokans J. (2000): “Life Insurance Market Development in Baltic Countries:
Regulatory and Supervisory Framework”, OECD Publication.
• Bonnet Yanick, Arnal Pierre (2000): “Analysis and Prospects of the French
Bancassurance Market”, Working Paper.
232 • Boot Arnoud W. A. (Jan 2003): “Restructuring in the Banking Industry with
Implications for Europe”, Presented at 2003 EIB Conference on Economics and
Banking.
• Boot Arnould W.A. & Thakor Anjan V. (Winter 1997): “Banking Scope and
Financial Innovation”, Volume 10, No.4: 1099-1131.
• Bos J.W.B. & Kolari J.W. (July 2003): “Large Bank Efficiency in Europe and the
United States: Are There Economic Motivations for Geographical Expansion in
Financial Services?” De Nederlandsche Bank, Banking and Supervisory
Strategies, Research Series Supervision, No. 61.
• Boyd John H. & Graham Stanley L. (1986): “Risk, Regulation, and Bank Holding
Company Expansion into Nonbanking”, Quarterly Review, Federal Reserve Bank
of Minneapolis, No. 10: 2-17.
• Boyd John H. & Graham Stanley L. (1988): “The Profitability and Risk Effects of
Allowing Bank Holding Companies to Merge With Other Financial Firms: A
Simulation Study”, Quarterly Review, Federal Reserve Bank of Minneapolis, No.
12: 3-20.
• Boyd John H., Chang Chun & Smith Bruce D. (Aug 1998): “Moral Hazard under
Commercial and Universal Banking”, Journal of Money, Credit and Banking,
Volume 30, No. 3: Part 2: Comparative Financial Systems: 426-468.
• Boyd John H., Graham Stanley L. & Hewitt R. Shawn (1993): “Bank holding
company mergers with nonbank financial firms: Effects on the risk of failure”,
Journal of Banking & Finance, Volume 17: 43-63.
233 • Boyer Martin M. & Nyce Charles M. (Sept 2002): “Banks as Insurance Referral
Agents? The Convergence of Financial Services: Evidence from the Title
Insurance Industry”, CIRANO Centre interuniversitaire de recherché en analyse
des organisations.
• Brewers Elijah (III) (1989): “Relationship Between Bank Holding Company Risk
and Nonbank Activity”, Journal of Economics and Business, Volume 41: 337-
353.
• Browne M.J. & Kim K. (1993): “An International Analysis of Life Insurance
Demand”, Journal of Risk and Insurance, Volume 60: 616-634.
• Browne M.J., Chung J. and Frees E. (2000): “International Property-Liability
Insurance Consumption”, Journal of Risk and Insurance, Volume 67, No. 1: 73-
90.
• Campbell R.A. (1980): “The Demand for Life Insurance: An Application of the
Economics of Uncertainty”, Journal of Finance, Volume 35, No. 5: 1155-1172.
• Canals Jordi (1998): “Universal Banks: The Need for Corporate Renewal”,
European Management Journal, Volume 16, No. 5: 623-634.
• Cargill T.F. and Troxel T.E. (1979): “Modelling Life Insurance Savings: Some
Methodological Issues”, Journal of Risk and Insurance, Volume 46: 391-410.
• Carow Kenneth A. (2001): “Citicorp-Travelers Group merger: Challenging
barriers between banking and insurance”, Journal of Banking & Finance, Volume
25: 1553-1571.
234 • Carow Kenneth A. (March 2001): “The Wealth Effects of Allowing Bank Entry
into the Insurance Industry”, Journal of Risk & Insurance, Volume 68, No. 1:
129-150
• Catalan M., Impavido G. and Musalem A.R. (2000): “Contractual Savings or
Stock Markets Development: Which Leads?” The World Bank Financial Sector
Development Department, Working Paper.
• Chamberlain G. (1984): “Panel Data”, Handbooks in Economics, Chapter 22:
1247-1318.
• Chen R. and Wong K.A. (2004): “The Determinants of Financial Health of Asian
Insurance Companies”, The Journal of Risk and Insurance, Volume 71, No. 3:
469-499.
• Chen R., Wong K.A. and Lee H.C. (2001): “Age, Period and Cohort effects on
life insurance purchases in the U.S.”, Journal of Risk and Insurance, Volume 68,
No. 2: 303-327.
• Chu J. (2000): “All Eyes on China”, Best Review, Volume 101, No. 4: pg 67.
• Chun & Stitt (Feb 2001): “Allfinanz or Bancassurance – Which Route for Asia?”,
Asia Insurance Review.
• Colenutt Dennis (June 1979): “The Regulation of Insurance Intermediaries in the
United Kingdom”, The Journal of Risk and Insurance, Volume 46, No. 2: 77-86.
• Cowan Arnold R., Howell Jann C. & Power Mark L. (Autumn 2002): “Wealth
Effects of Banks’ Rights to Market and Originate Annuities”, The Quarterly
Review of Economics and Finance, Volume 42, No. 3: 487-503.
235 • Cummins J.D. (1991): “Static and Financial Models of Insurance Pricing and the
Insurance Firm”, Journal of Risk and Insurance, Volume 58: 261-302.
• Cummins J.D. and Danzon P.M. (1997): “Price, Financial Quality, and Capital
Flows in Insurance Markets”, Journal of Financial Intermediation, Volume 6: 3-
38.
• Cummins J.D., Tennyson S. and Weiss M.A. (1999): “Consolidation and
Efficiency in the US Life Insurance Industry”, Journal of Banking and Finance,
Volume 23: 325-357.
• Cybo-Otone Alberto & Murgia Maurizio (2000): “Mergers and shareholder
wealth in European banking”, Journal of Banking & Finance, Volume 24: 831-
859.
• Datamonitor (2003): “Opportunities in European Life Bancassurance”, A
Datamonitor Report.
• Datamonitor (2004): “The Role of Banks and Building Societies in UK General
Insurance: Current and Future Trends”, A Datamonitor Report.
• Demsetz R.S. and Strahan P.E (1997): “Diversification, Size, and Risk at Bank
Holding Companies”, Journal of Money, Credit and Banking, Volume 29, No. 3:
300-313.
• DeYoung R. and Rice T. (2004): “Noninterest Income and Financial Performance
at U.S. Commercial Banks”, The Financial Review, Volume 39, No. 1: 101-127.
• DeYoung R. and Roland K.P. (2001): “Product Mix and Earnings Volatility at
Commercial Banks: Evidence from a Degree of Total Leverage Model”, Journal
of Financial Intermediation, Volume 10, No.1: 54-84.
236 Diamond Douglas W. (Aug 1998): “Comment on Moral Hazard under
Commercial and Universal Banking”, Journal of Money, Credit and Banking,
Volume 30, No. 3: Part 2: Comparative Financial Systems: 469-471.
• Dickinson G. (2000): “Encouraging a Dynamic Life Insurance Industry:
Economic Benefits and Policy Issues”, OECD Publication.
• Dodd-Walker E., Shook C.L. and McGee J.E. (2000): “Do Minority and
Nonminority Business Owners Evaluate Firm Performance Differently?”
Working Paper.
• Dorval Bernard (July 2002): “Development of Bancassurance in Canada”, The
Geneva Papers on Risk and Insurance, Volume 27, No. 3: 304-306.
• Durand Romain (Feb 2003): “Bancassurance across the globe meets with very
mixed response”, SCOR Technical Newsletter.
• Ernst and Young Study (2001): “The European Bancassurance Experience”.
• Esho N., Kirievsky A., Ward D. and Zurbruegg R. (2004): “Law and the
Determinants of Property-Casualty Insurance”, The Journal of Risk and
Insurance, Volume 71, No. 2: 265-283.
• Estrella Arturo (Mar 2001): “Mixing and matching: Prospective financial sector
mergers and market valuation”, Journal of Banking & Finance, Volume 25:
2367-2392.
• Falautano I. & Marsiglia E. (July 2003): “Integrated Distribution of Insurance and
Financial Services and Value Creation: Challenges Ahead”, The Geneva Papers
on Risk and Insurance, Volume 28, No. 3: 481-494.
237 • Fields L. Paige, Fraser Donald R. & Kolari James W. (Feb 2005): “What’s
Different About Bancassurance? Evidence of Wealth Gains to Banks and
Insurance Companies”, Texas A&M University, May Business School, Working
Paper.
• Financial Services Factbook 2005: “Convergence”, Chapter 4.
• Fukuyama F. (1995): “Trust: The Social Virtues and the Creation of Prosperity”,
London: Hamish Hamilton.
• Gandolfi A.S. and Miners L. (1996): “Gender Based Differences in Life
Insurance Ownership”, Journal of Risk and Insurance, Volume 63: 683-694.
• Goddard S. (1999): “Globals Face Cultural Challenges: Differences in Attitude
can Complicate Risk Management Efforts”, Business Insurance, March 29,
G4(1).
• Goldsmith A. (1983): “Household Life Cycle Protection: Humna Capital Versus
Life Insurance”, Journal of Risk and Insurance, Volume 50, No. 1: 33-44.
• Golub S.S. (2003): “Measures of Restrictions on Inward Foreign Direct
Investment for OECD Countries”, OECD Economic Studies, Paper No. 36
2003/1.
• Greene M.R. (1972): “The Spanish Insurance Industry. An Analysis”, The
Journal of Risk and Insurance, Volume 39, No. 2: 221-243.
• Griffin K. (1996): “The Retail Banking Industry in Australia”, The Institute of
Actuaries of Australia.
238 • Hakansson N.H. (1969): “Optimal Investment and Consumption Strategies Under
Risk, and Under Uncertain Lifetime and Insurance”, International Economic
Review, Volume 10: 443-466.
• Hammond J.D., Houston D.B. and Melander E.R. (1967): “Household Life
Insurance Premium Expenditure: an empirical approach”, Journal of Risk and
Insurance, Volume 34, No. 3: 397-408.
• Hansen L.P. (1982): “Large Sample Properties of Generalized Method of
Moment Estimators”, Econometrica, Volume 50, No. 4: 1029-1054.
• Hardwick P. (1987): “Measuring Cost Inefficiency in the U.K. Life Insurance
Industry”, Applied Financial Economics, Volume 7, No. 1: 37-44.
• Hardwick P. and Adams M. (2002): “Firm Size and Growth in the United
Kingdom Life Insurance Industry”, The Journal of Risk and Insurance, Volume
69, No. 4: 577-593.
• Headen R.S. and Lee F.L. (1974): “Life Insurance Demand and Household
Portfolio Behaviour”, Journal of Risk and Insurance, Volume 41, No. 4: 685-698.
• Heistermann Bernd (May 2000): “Development of Bancassurance Products”,
General & Cologne RE.: Risk Insights For Life & Health Insurance Executives,
9-11.
• Heymowski M. (May 2000): “Selling Over the Counter: How Bancassurance
Works”, General & Cologne RE.: Risk Insights For Life & Health Insurance
Executives, 1-5.
239 • Heymowski M. (May 2000a): “Comprehensive Bancassurance Services Provided
by General & Cologne RE.”, General & Cologne RE.: Risk Insights For Life &
Health Insurance Executives, 5-7.
• Hishikawa M. (2002): “Saefty Net for Life Insurance Policyholders Should Be
Drastically Reformed”, PRANJ Newsletter, December 12th 2002.
• Hislop Angus, Peterson Ole & Ziegler Ralf (2002): “Making Bancassurance
Really Work: From Product-Oriented Cross-Selling to Customer Focused Cross-
Buying”, IBM Business Consulting Services.
• Hofstede G. (1983): “National Cultures in Four Dimensions”, International
Studies of Management and Organization, Volume 13, No. 1-2: 46-74.
• Hofstede G. (1995): “Insurance as a Product of National Values”, The Geneva
Papers on Risk and Insurance, Volume 77, No. 20: 423-429.
• Holland C.P., Lockett A.G. & Blackman I.D. (1998): “Global Strategies to
Overcome the Spiral of Decline in Universal Bank Markets”, Journal of Strategic
Information Systems, Volume 7: 217-232.
• Holsboer Jan H. (July 1999): “Repositioning of the Insurance Industry in the
Financial Sector and its Economic Role”, The Geneva Papers on Risk and
Insurance, Volume 24, No. 3: 243-290.
• Hong J. and Rios-Rull J.V. (2004): “Life Insurance and Household
Consumption”, Centro de Altisimos Estudios Rios Perez (CAERP), Working
Paper No. 23.
240 • Hubbard Chris (Spring 2001): “Bancassurance Operations in the UK: Insights
Gained on the Road to Success”, Thomson Management Solutions Inc.: Financial
Services Executive Letter, pg 1-5.
• Hunter William C. & Timme Stephan G. (Feb 1995): “Core Deposits and
Physical Capital: A Reexamination of Bank Scale Economies and Efficiency with
Quasi-Fixed Inputs”, Journal of Money, Credit and Banking, Volume 27, No.1:
165-185.
• Hussels S., Ward D. and Zurbruegg R. (2003): “How Do You Stimulate Demand
for Insurance?” Bradford University School of Management, Working Paper
Series No. 03/32.
• Hwang T. and Gao S. (2003): “The Determinants of the Demand for Life
Insurance in an Emerging Economy: the Case of China”, Managerial Finance,
Volume 29, No. 5: 82-97.
• Jagannathan R., Skoulakis G. and Wang Z. (2002): “Generalized Method of
Moments: Applications in Finance”, The Journal of Business and Economic
Statistics, Volume 20: 470-481.
• Jappelli T. and Pistaferri L. (2001): “Tax Incentives and the Demand for Life
Insurance: Evidence from Italy”, Centre for Studies in Economics and Finance,
Working Paper No. 52.
• Kitazawa Y. (2001): “Recent Development in Panel Data Econometrics”,
EKONOMIKUSU, the bulletin of Economics Association in Kyushu Sangyo
University, Volume 6, No. 1.
• Klein Roger A. (2001): “Bancassurance in Practice”, Muich Re. Group.
241 • Koch Jr. C.H. (2003): “The Advantages of the Civil-Law Judicial Design as the
Model for Emerging Legal Systems”, Working Paper.
• Koutsomanoli-Fillipaki N. and Staikouras C. (2004): “Competition and
Concentration in the New European Banking Landscape”, Athens University of
Economics and Business, Research Paper.
• La Porta R., Lopez-De-Silanes F., Shleifer A. and Vishny R. (2000): “Investor
Protection and Corporate Governance”, Journal of Financial Economics, Volume
58: 3-27.
• La Porta R., Lopez-De-Silanes F., Shleifer A. and Vishny R.W. (1997): “Legal
Determinants of External Finance”, The Journal of Finance, Volume 52, No. 3:
1131-1150.
• La Porta R., Lopez-De-Silanes F., Shleifer A. and Vishny R.W. (1998): “Law and
Finance”, The Journal of Political Economy, Volume 106, No. 6: 1113-1155.
• Lafferty (Dec 2001): “Farewell, annus horribilis”, News Article.
• Leflaive V. (2001): “The Supervision of Insurance Solvency: Comparative
Analysis in OECD Countries”, OECD Insurance and Private Pensions
Compendium for Emerging Economies, Book 1, Part 1: 2)a.
• Legrand Corinne (2004): “New Trends in World Bancassurance”, Milliman
Consultants and Actuaries.
• Lenten L.J.A. and Rulli D.N. (2005): “A Time-Series Analysis of the Demand for
Life Insurance in Australia: An Unobserved Components Approach”, Research
Paper: The University of New South Wales, School of Banking and Finance
Seminar Program, March 2005.
242 • Levine R. (1997): “Financial Development and Economic Growth: Views and
Agenda”, Journal of Economic Literature, Volume 35: 688-726.
• Levine R. (1998): “The Legal Environment, Banks, and Long-Run Economic
Growth”, Journal of Money, Credit and Banking, Volume 30, No. 3, Part 2:
Comparative Financial Systems, 596-613.
• Levine R. (1999): “Law, Finance, and Economic Growth”, Journal of Financial
Intermediation, Volume 8: 8-35.
• Levine R. and Zervos S. (1998): “Stock Markets, Banks, and Economic Growth”,
The American Economic Review, Volume 88, No. 3: 537-558.
• Levine R., Loayza N. and Beck T. (2000): “Financial Intermediation and Growth:
Causality and Causes”, Journal of Monetary Economics, Volume 46: 31-77.
• Lewis F.D. (1989): “Dependents and the Demand for Life Insurance”, American
Economic Review, Volume 79, No. 3: 452-467.
• Lim C.C. and Haberman S. (2004): “Modelling Life Insurance Demand from a
Macroeconomic Perspective: The Malaysian Case”, Research Paper: The 8th
International Congress on Insurance, Mathematics and Economics, Rome.
• Lim C.C. and Haberman S. (Oct 2004): “Macroeconomic Variables and the
Demand for Life Insurance in Malaysia”, Research Paper: The 2nd Malaysian
Research Group (MRG) Annual Conference, Manchester England.
• Llewellyn D.T. (1999): “The new Economics of Banking”, SUERF Studies, No.
5.
243 • Lown Cara S., Osler Carol L., Strahan Philip E. & Sufi Amir (Oct 2000): “The
Changing Landscape of the Financial Services Industry: What Lies Ahead?”
FRBNY Economic Policy Review, 39-55.
• Mamun A., Hassan M.K. and Maroney N. (2005): “The Wealth and Risk Effects
of the Gramm-Leach-Bliley Act (GLBA) on the US Banking Industry”, Journal
of Business Finance & Accounting, Volume 32(1) & (2).
• Markusen J. and Maskus K. (2001): “General-Equilibrium Approaches to the
Multinational Firm: A Review of Theory and Evidence”, National Bureau of
Economic Research Department, NBER Working Paper: No. 8334.
• Marois Bernard (1997): “French Banks and European Strategy”, European
Management Journal, Volume 15, No. 2: 183-189.
• McDaniel Dave (June 1996): “Bancassurance Lessons From Abroad”, Best’s
Review / Property-Casualty Insurance Edition, Volume 97, No. 2: 23-31.
• McFadden D.L. (1984): “Econometric Analysis of Qualitative Response Models”,
Handbooks in Economics, Chapter 24: 1395-1457.
• Miyamoto K. (2003): “Human Capital Formation and Foreign Direct Investment
in Developing Countries”, OECD Development Centre, Technical Paper No. 211.
• Morgan Consulting Group (2003): “Accessing New Distribution Opportunities in
the French Savings Market”.
• Morgan Ed (2003): “European Bancassurance Review”, Milliman Global
Insurance Article, 1-4.
• Nigh J.O. & Saunders M.V.T. (2003): “Bancassurance Around the World”,
Emphasis.
244 • Nigh J.O. (2000): “Bancassurance in Latin America”, Emphasis.
• Nurullah M. and Staikouras S.K. (2004): “The Separation of Banking from
Insurance: Evidence from Europe”, Case Business School, Working Paper.
• OECD Secretariat (2000): “Developing Life Insurance in the Economies in
Transition”, OECD Publication.
• Okura M. and Kasuga N. (2005): “Financial Instability and Life Insurance
Demand”, Working Paper, Nagasaki University.
• Outreville J.F. (1990): “The Economic Significance of Insurance Markets in
Developing Countries”, Journal of Risk and Insurance, Volume 57, No. 3: 487-
498.
• Outreville J.F. (1996): “Life Insurance Markets in Developing Countries”,
Journal of Risk and Insurance, Volume 63, No. 2: 263-278.
• Ovrut Barnett D. Esq (Summer 2000): “Gramm-Leach-Bliley Act: Opportunities
and Challenges”, Thomson Management Solutions Inc.: Financial Services
Executive Letter, 6-7.
• Park H., Borde S.F. and Choi Y. (2002): “Determinants of Insurance
Pervasiveness: A Cross-National Analysis”, International Business Review,
Volume 11: 79-96.
• Rankin J.C. (2004): “A World of Opportunity”, Resource, LOMA.
• Regional Symposium (Nov 1999): “Enhancing Life Insurance Regulatory
Regimes in ASIA”, Australian APEC Study Centre.
245 • Rime Bertrand and Stiroh Kevin J. (Mar 2002): “The Performance of Universal
Banks: Evidence from Switzerland”, Journal of Banking & Finance, Volume 27:
2121-2150.
• Rubayah Y. and Zaidi I. (2000): “Prospective Insurance”, Utara Management
Review, No. 1: 69-79.
• Sakr Gamal (2001): “Special Report About Bancassurance”, InsureEgypt.
• Saunders A. and Walter I. (1994): “Universal Banking in the United States: What
We Could Gain? What We Could Lose? Oxford University Press.
• Scharfstein Eva (May 2000): “The Childhood Years:1999 Bancassurance in the
United States”, General & Cologne RE.: Risk Insights For Life & Health
Insurance Executives, 19-23.
• Showers E.V. and Shotick A.J. (1994): “The Effects of Household Characteristics
on Demand for Insurance: A Tobit Analysis”, Journal of Risk and Insurance,
Volume 61: 492-502.
• Shulansky Ralph M. (Dec 1967): “The Case against Increasing the Savings Bank
Life Insurance Policy Limit in Connecticut”, The Journal of Risk and Insurance,
Volume 34, No. 4: 628-633.
• Sijbrands Simon & Eppink D. Jan (1994): “The Internationalization of Dutch
Banks: A New Beginning?” Long Range Planning, Volume 27, No. 4: 35-47.
• Skipper H.D. (1987): “Protectionism in the Provision of International Insurance
Services”, Journal of Risk and Insurance, Volume 54, No. 1: 55-85.
• Skipper Jr. H.D. (2000): “Financial Services Integration Worldwide: Promises
and Pitfalls”, North American Actuarial Journal, Volume 4, No. 3: 71-107.
246 • Skipper Jr. H.D. (2001): “The Taxation of Life Insurance Policies in OECD
Countries: Implications for Tax Policy and Planning”, OECD Insurance and
Private Pensions Compendium for Emerging Economies, Book 1, Part1: 7)b.
• Smith Bruce D. & Stutzer Michael (Summer 1995): “A Theory of Mutual
Formation and Moral Hazard with Evidence from the History of the Insurance
Industry”, The Review of Financial Studies, Volume 8, No. 2: 545-577.
• Smith R., Staikouras C. and Wood G. (2003): “Non-Interest Income and Total
Income Stability”, Bank of England, Working Paper No. 198.
• Soifer Ray (2001): “Bancassurance: The Revolution that Hasn’t Come”, Working
Paper.
• Soifer Ray (2001): “Bancassurance: The Revolution That Hasn’t Come”, Soifer
Consulting.
• Staikouras C. and Wood G. (2003a): “The Determinants of Bank Profitability in
Europe: New Trends”, The European Applied Business Research Conference,
Venice, Italy.
• Staikouras C. and Wood G. (2004): “The Determinants of European Bank
Profitability”, International Business and Economics Research Journal,
Forthcoming.
• Staikouras C. and Wood G. (2004a): “Sources of Income and Stability in the
European Banking Industry”, Ekonomika, Malaysian Economic Association:
16(1).
• Staikouras S.K. (2005a): “De facto versus de jure bank-insurance ventures in the
Greek market”, Case Business School, Working Paper.
247 • Staikouras S.K. (2005b): “The Greek Bank-Insurance Corporate Model: A Look
at a not-so-new Structure”, Case Business School, Working Paper.
• Staikouras S.K. (2005c): “Qualitative Aspects of European Financial
Conglomerates”, Case Business School, Working Paper.
• Staikouras S.K. and Dickinson G.M (2005): “An Examination of the Bank
Incursion into Insurance Business: The case of Greece”, Multinational Finance
Society Conference, Athens, Greece.
• Staikouras Sotiris K. (2005): “Business opportunities and market risks in financial
conglomerates”, Working Paper.
• Stiroh K.J. (2004): “Diversification in Banking: Is Non-Interest Income the
Answer?” Journal of Money, Credit and Banking, Volume 36, No. 5: 853-882.
• Sun Qixiang (2003): “The Impact of WTO Accession on China’s Insurance
Industry”, Risk Management and Insurance Review, Volume 6, No. 1: 27-35.
• Susan Drury (2005): “Bancassurance in the 21st Century”, Lafferty.
• Swiss Re. (1998): “Upheaval in Insurance Markets – Results Still Good Despite
Increased Competition”, Sigma-Prospect Article, No. 5.
• Swiss Re. (1998a): “Life and Health Insurance in the Emerging Markets:
Assessment, Reforms and Perspectives”, Sigma Research Article, No. 1.
• Swiss Re. (1998b): “World Insurance in 1996: Modest Growth in the Insurance
Industry”, Sigma Research Article, No. 4.
• Swiss Re. (1999): “Life Insurance: Will the Urge to Merge Continue?” Sigma
Research Article, No. 6.
248 • Swiss Re. (1999a): “World Insurance in 1997: Booming Life Business, but
Stagnating Non-Life Business”, Sigma Research Article, No. 3.
• Swiss Re. (1999b): “World Insurance in 1998: Deregulation, Overcapacity and
Financial Crises Curb Premium Growth”, Sigma Research Article, No. 7.
• Swiss Re. (2000): “Emerging Markets: The Insurance Industry in the Face of
Globalisation”, Sigma Research Article, No. 4.
• Swiss Re. (2000a): “World Insurance in 1999: Soaring Life Insurance Business”,
Sigma Research Article, No. 9.
• Swiss Re. (2001): “Insurance Markets in Asia: Sanguine Outlook Despite Short-
Term Uncertainties”, Sigma Research Article, No. 4.
• Swiss Re. (2001a): “World Insurance in 2000: Another Boom Year for Life
Insurance; Return to Normal Growth for Non-Life Insurance”, Sigma Research
Article, No. 6.
• Swiss Re. (2001b): “World Financial Centres: New Horizons in Insurance and
Banking”, Sigma Research Article, No. 7.
• Swiss Re. (2002): “Bancassurance Developments in Asia – Shifting into a Higher
Gear”, Sigma Research Article, No.7.
• Swiss Re. (2002a): “World Insurance in 2001: Turbulent Financial Markets and
High Claims Burden Impact Premium Growth”, Sigma Research Article, No. 6.
• Swiss Re. (2003): “World Insurance in 2002: High Premium Growth in Non-Life
Insurance”, Sigma Research Article, No. 8.
• Swiss Re. (2004): “World Insurance in 2003: Insurance Industry on the Road to
Recovery”, Sima Research Article, No. 3.
249 • Swiss Re. (2005): “World Insurance in 2004: Growing Premiums and Stronger
Balance Sheets”, Sigma Research Article, No. 2.
• Szipro G.G. and Outreville J.F. (1988): “Relative Risk Aversion Around the
World”, Studies in Banking and Finance, Volume 6: 123-129.
• Tachibanaki T. and Shimono K. (1994): “Demand Analysis of Life Insurance –
Asset Selection for Safe Asset, Risk Asset and Life Insurance”, Individual Saving
and Life Cycle, Nikkei Shimbun Press, Chapter 9: 220-243.
• Tassin Emmanuel (Aug 2002): “Success Factors for Bancassurance in France”,
Milliman Global Insurance Article, 9-12.
• The Boston Consulting Group, BCG (1999): “Putting it Together: Convergence
Strategies for Banking, Insurance, and Investments”, Research Highlights.
• Thom Michael (1999): “The Prudential Supervision of Financial Conglomerates
in the European Union”, North American Actuarial Journal, Volume 4, No. 3.
• Thomson Maria (Spring 2001): “Simpler Products, Streamlined Delivery”,
Thomson Management Solutions Inc.: Financial Services Executive Letter, 6-8.
• Thomson Maria (Summer 2002): “Interesting Developments in Bank Insurance”,
Thomson Management Solutions Inc.: Financial Services Executive Letter, pg 3.
• Thomson Maria (Summer 2002b): “The Survivial of Most Insurers Hinges on
Underwriting Faster, Cheaper, Better”, Thomson Management Solutions Inc.:
Financial Services Executive Letter, 8-11.
• Thomson Maria (Winter 2000): “Proven Approaches to Bancassurance Success”,
National Insurance Executive Letter, 1-3.
250 • Thomson Maria, Wade Alan (Summer 2001): “What’s Needed to Create the Ideal
Bancassurance Model in the U.S.?”, Thomson Management Solutions Inc.:
Financial Services Executive Letter, 1-5.
• Truett D.B. and Truett L.J. (1990): “The Demand for Life Insurance in Mexico
and the United States: A Comparative Study”, Journal of Risk and Insurance,
Volume 57, No. 2: 321-328.
• Turner M. (Mar 1998): “Bancassurance No More: Australian Banks’ Mindset
Offers Clues for Years Ahead”, The Society Of Actuaries Newsletter, Volume 32,
No. 3: 3-13.
• Urata F., Komamura K. and Shibuya T. (1999): “Purchasing Behaviour for Life
Insurance Product”, Management of Life Insurance Company, Volume 67, No. 1:
3-16.
• Verweire K. (1999): “Performance Consequences of Financial Conglomeration
with an Empirical Analysis in Belgium and the Netherlands”, Dissertation.
• Viswanathan K.S. and Cummins J.D. (2003): “Ownership Structure Changes in
the Insurance Industry: An Analysis of Demutualization”, The Journal of Risk
and Insurance, Volume 70, No. 3: 401-437.
• Wade Alan R. (2000): “Bancassurance in Europe: Role Model or Irrelevant?”,
Thomson Management Solutions Inc.: Financial Services Executive Letter, 1-3.
• Wade Alan R. (Fall 2000): “Lessons From Europe: Bancassurance Thriving in
Many Markets Overseas”, Thomson Management Solutions Inc.: Financial
Services Executive Letter, 1-3.
251 • Wall L.D. (1987): “Has Bank Holding Companies’ Diversification Affected Their
Risk of Failure?” Journal of Economics and Business, Volume 39: 313-326.
• Walter Ingo (1997): “Universal Banking: A Shareholder Value Perspective”,
European Management Journal, Volume 15, No. 4: 344-360.
• Ward D. and Zurbruegg R. (2000): “Does Insurance Promote Economic Growth?
Evidence from OECD Countries”, Journal of Risk and Insurance, Volume 67,
No. 4: 489-506.
• Webersinke Andres (May 2000): “Bancassurance in South Africa”, General &
Cologne RE.: Risk Insights For Life & Health Insurance Executives, 16-18.
• Wepler John M., Linn Thomas R. & Linnert Patrick T. (3rd Quarter 2004): “Banks
in Insurance: A Five-Year Retrospective Since the Passage of GLB”, Mash Berry.
• White M.D. (Mar 1990): “U.S. Bancassurance in the U.S. and Abroad”,
Proceedings of the 2nd International Life Insurance Conference: “Marketing
Peace of Mind”.
• Wooldridge J.M. (2001): “Applications of Generalized Method of Moments
Estimations”, The Journal of Economic Perspectives, Volume 51, No. 4: 87-100.
• Yaari M. (1965): “Uncertain Lifetime, Life Insurance and the Theory of the
Consumer”, Review of Economic Studies, Volume 32: 137-150.
• Yasui T. (2001): “Policyholder Protection Funds: Rationale and Structure”,
OECD Insurance and Private Pensions Compendium for Emerging Economies,
Book 1, Part 1: 2)b.
252 • Yuengert A.M. (1993): “The Measurement of Efficiency in Life Insurance:
Estimates of a Mixed Normal-Gamma Error Model”, Journal of Banking and
Finance, Volume 17: 483-496.
• Zietz E.N. (2003): “An Examination of the Demand for Life Insurance”, Risk
Management and Insurance Review, Volume 6: 159-191.
Internet Sites:
• Absolute Astronomy Encyclopaedia: http://www.absoluteastronomy.com
• Bank for International Settlements – Central Bank Websites:
http://www.bis.org/cbanks.htm
• Blackwell Publishing: http://www.blackwellpublishing.com
• CIA – The World Factbook: http://www.cia.gov/cia/publications/factbook
• Davidson R. and MacKinnon J.G.: “The Generalized Method of Moments”,
http://russell.vcharite.univ-mrs.fr/Doctoral/chp09big.pdf.
• ECB – Convergence Report:
http://www.ecb.int/pub/convergence/html/index.en.html
• Financial Shopper Network:
http://www.financial-shopper-network.com/life_insurance_history.htm
• IMD – World Competitiveness Center: http://www02.imd.ch/wcc
• IMF – International Financial Statistics: http://ifs.apdi.net/imf/logon.aspx
• IMF – The International Monetary Fund: http://www.imf.org
253 • IMF – World Economic Outlook:
http://www.imf.org/external/pubs/ft/weo/2005/02/data/index.htm
• ITIM International: http://www.geert-hofstede.com
• JSTOR: http://www.jstor.org
• Marples Gareth: “History of Insurance”,
http://www.thehistoryof.net/the-history-of-insurance.html
• McDermott M. Ben and Saunders Mark (2003): “Bancassurance: Doing It Right”,
www.asiainsurancereview.com.
• Microsoft Encarta Online Encyclopaedia: http://encarta.msn.com
• OECD – Organization for Economic Co-Operation and Development:
http://www.oecd.org
• OECD – Source:
http://puck.sourceoecd.org/vl=2416267/cl=15/nw=1/rpsv/home.htm
• Science Direct: http://www.sciencedirect.com
• Smith Kathy (2004): “The History of Life Insurance”, Articles-For-Free,
http://www.financial-shopper-network.com/life_insurance_history.htm
• Smith M.L. (2003): “Deterrence and Origin of Legal System: Evidence From
1950 – 2000”, http://ssrn.com/abstract=414365, Dice Centre Working Paper No.
2003-9.
• The World Trade Organization: http://www.wto.org
• Wikipedia-The Free Encyclopaedia: http://en.wikipedia.org/wiki/Main_Page
• World Bank: http://www.worldbank.org
254