The Geneva Papers on Risk and Insurance Vol. 27 No.

3 (July 2002) 395–412

Law, Politics and Life Insurance Consumption in Asia

by Damian Ward and Ralf Zurbruegg

This paper investigates the determinants of consumption for one of the fastest growing
financial products in Asia. We find evidence that increased provision of civil rights and
political stability leads to an increase in life insurance provision. However, by utilizing
various estimation procedures, a number of differences between the more developed
insurance markets and those in Asia are illustrated. In particular, the estimated income
effect is found to be far higher in Asia than in other countries. However, the size of this
difference is reduced once political and legal factors are controlled for, suggesting that
future insurance market growth in Asia may not exceed that in the rest of the world.

1. Introduction
Asia is fast becoming the world’s leading insurance sector. According to Sigma (1999),
the consumption of life insurance in Asia, on a per capita basis, is higher than the global
average, with Asians spending three times as much on life insurance as on non-life insurance.
As a consequence, during the last decade the consumption of life insurance within Asia has
grown at more than 10 per cent per annum and currently accounts for 40.1 per cent of total
global life insurance premiums. However, despite this growth in the Asian insurance market,
there has been little substantial research into the sources behind the purchase of life insurance
and, importantly, whether the motives for purchasing insurance differ from that of more
developed life insurance markets.
As a starting point, Sigma (1999) and Enz (2000) point to the ‘‘S curve’’ relationship
between economic development and insurance market development. Specifically, consump-
tion of life insurance is expected to accelerate as a developing economy grows, but then slows
as economic development becomes comparable to that in the developed world. As a
consequence, the income elasticity of demand for life insurance should be greater in the
emerging economies of Asia than in the economies of the developed world. This argument is
supported by Truett and Truett (1990), but only for the specific comparison between the U.S.
and Mexico. However, it is not only the income effect that is likely to differ. Asian growth,
according to Sigma (1999, 2000), stems from a set of regional specific characteristics. Savings
rates in Asia are higher than in the developed world. However, capital markets which provide
alternative savings vehicles, are less developed. In addition, the Asian population is extremely
large, but the provision of pensions and other social welfare provisions by the state are
relatively low. Taken together, all of these factors are likely to promote the consumption of life
insurance in Asia.
Unfortunately, existing research on aggregate level growth in the consumption of life

Damian Ward, Lecturer in Economics, Bradford University School of Management, Emm Lane, Bradford,
West Yorkshire, United Kingdom BD9 4JL. Ralf Zurbruegg, Associate Professor of Finance, School of Commerce,
University of Adelaide. The authors wish to thank Anatoly Kiriviesky and the anonymous referees for their helpful
comments and advice.

# 2002 The International Association for the Study of Insurance Economics.

Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK.
396 WARD AND ZURBRUEGG

insurance has primarily focused upon the economic and social factors driving global
insurance purchases, with little focus on Asia. Principal work includes that of Browne and
Kim (1993) and Outreville (1996), whose findings indicate that the level of life insurance
consumption within an economy can be explained by income levels, the price of insurance,
inflation, the dependency ratio, and financial development.
Also, a significant omission from previous studies and debates is the impact of legal and
political factors, particularly in the less developed markets, upon the consumption of life
insurance. This contrasts with growing evidence supporting the importance of legal and
political factors in economic and financial growth, where work by Laporta, Lopez-de-Silanes,
Shleifer and Vishny (1997, 1998, 2000) has highlighted the positive role national legal origin
can have on creditor and shareholder rights, and the provision of external finance. Moreover,
their research has been extended to show how a clearly defined and enforced legal system aids
economic growth by facilitating better financial intermediation (see Levine, 1998, 1999;
Levine, Loayza and Beck, 2000).
Unfortunately, the applicability of these arguments to the insurance sector, is as yet, not
well understood. However, Cummins and Danzon (1997) have shown that since insurance
companies can default on claims payable, then the offering of insurance is analogous to issuing
risky corporate debt. Therefore, given that insurance is conceptually a credit-based product,
there is ample reason to suspect that the legal environment will impact upon the growth of the
industry. However, it should also be acknowledged that the legal structure is also entwined with
the political environment. Knack and Keefer (2000) argue that economic polarization within an
economy leads to a greater probability of political intervention and redistribution of property
rights within an economy. Hence, the government’s willingness and ability to alter the legal
basis over time conditions the maintenance and quality of laws and contractual obligations.
This leads to the economic and legal stability required for the development of financial
intermediation.
To address the above issues, this paper analyses annual life insurance premium
consumption in a total of 37 countries for the period between 1987 to 1998. The full set of
countries are listed in the appendix. Premium data was kindly provided by Swiss Re. This
panel dataset will be the largest that has been currently examined. In addition to following the
previous methodologies, this paper also utilizes the General Method of Moments (GMM)
dynamic system estimator developed by Arellano and Bond (1991). The estimator is
considered preferable when persistence exists between sequential observations and when the
independent variables are potentially endogenous.
In summary, this paper as an aid to policymakers and the rapidly expanding international
insurance companies that are beginning to enter the region, provides an evaluation and
comparison of the determinants of life insurance consumption in Asia and the developed
world. In addition to the contemporary need for such a study, this paper will also incorporate
the now widely accepted role that legal and political variables have upon the life insurance
sector. To accomplish this, the next section provides a description of the determinants for life
insurance consumption alongside the conditioning set and additional legal and political
variables used for this study. Section 3 discusses the methodology used to obtain the results
tabulated in section 4. A conclusion is then presented in the final section.

2. The determinants of life insurance consumption

A large amount of modern theoretical work on the demand for life insurance can be
traced back to Yaari (1965) who introduced the concept of uncertainty of life into a framework

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 397

explaining an individual’s lifetime consumption allocation process. Other work has extended
Yarri’s theory by incorporating additional determinants. This includes Hakansson (1969) who
discussed the impact wealth, an income stream, interest rates and the price of insurance has on
the demand function. Karni and Zilcha (1986) also incorporated a measure of risk aversion
into the model. Alternative frameworks also exist such as Lewis’ (1989) model, which treated
the demand function slightly differently by defining the goal of a policyholder to maximize
the utility of any beneficiaries, rather than maximizing personal consumption utility.
From the above theoretical research, empirical tests for international life insurance
demand have attempted to measure the significance of these factors. Moreover, many have
also tried to account for cross-national differences in risk perception and consumer utility
maximization. Examples would include Ram and Schultz (1979), Szpiro and Outreville
(1988), Viscusi and Evans (1990), Browne and Kim (1993) and Outreville (1996) who
have examined how consumer utility functions can be dependent upon life expectancy,
human capital endowment, health, religion and differing lifecycle stages for cross-country
populations.
As it is the intention of this paper to examine the impact legal and political factors can
have on life insurance purchases, the basic set of economic and social variables used in these
previous studies are also incorporated into the possible explanatory factors that affect life
insurance density, LIFEDEN,1 in countries. Following Yuengert (1993), it is acknowledged
that total premiums are not a pure measure of output as they represent total sales which is
equal to price times output. Cummins, Tennyson and Weiss (1999) in a cost study preferred to
measure output by the value adding activities of risk transfer and financial intermediation.
Such measures would be useful here but are unfortunately unavailable in macro-level
datasets. However, the value added approach does not exclude the use of total premiums as the
amounts paid by policyholders should correlate with the value added by the insurance
company. For expository purposes, each of the factors contained within the final estimation
model are explained in depth below.

2.1 Legal and political factors

Recent literature by Laporta et al. (1997, 1998, 2000) has highlighted the supportive
aspects of the legal environment for finance. A central recognition is that investors are at risk
of opportunistic behaviour unless the legal system offers protection. Not surprisingly, recent
empirical evidence (see Levine, 1998, 1999) shows that where legal environments provide
good investor protection then economic growth and financial intermediation will tend to be
higher.
Moreover, it can be further argued that there are even more reasons to believe the legal
environment may be more important for the life insurance market where a client’s financial
relationship with life companies tends to be long term. With life policies offering investment
opportunities, the variation in investment returns over prolonged time periods leads to an
element of risk being associated with the policies’ final value. Furthermore, when this risk is
coupled with the complexity of life insurance products, policyholders can suffer from
informational asymmetry. This combination of problems for policyholders can provide
managers with the ability to pursue opportunistic behaviour. This can invariably realize itself

1
Specifically, our dependent variable is the logarithm of the real value of life insurance premiums written in
any given year divided by the total population.

# 2002 The International Association for the Study of Insurance Economics.

398 WARD AND ZURBRUEGG

via increased expense loadings, reduced annuity rates or even higher dividend payments to
shareholders. All of which result in lower financial returns to policyholders.
Traditionally, the mutual mode of corporate governance, originally discussed by Mayers
and Smith (1981), has been seen as a way of controlling managerial opportunism and the
expropriation of policyholders’ wealth. However, more recently Baker and Thompson (2000)
have noted that the governance of managerial opportunism in life insurance companies
extends beyond modes of corporate ownership and includes the legal and regulatory
environment of the insurance industry.2 Specifically, life insurance is often subject to investor
protection regulation, where governments provide industry specific laws that seek to protect
investors from opportunistic behaviour by insurance companies. This provision of protection
should smooth the progress of efficient long-term contracting in the life insurance industry
and accordingly aid the development of the industry while encouraging the purchase of life
products. But importantly, given the existence of industry specific regulation, it is almost self-
evident that the legal environment as it relates to the consumption of life insurance is
tremendously important to policyholders.
Therefore, in order to examine the legal environment we use two variables to capture a
host of legal factors that may affect a consumer’s choice to purchase life insurance products.
The primary tool is CIVRIGHTS, with an alternative variable called ROL. CIVRIGHTS is an
index formed by the company Freedom House and incorporates human rights, the rule of law,
personal autonomy and economic rights. As a broad measure, it seeks to capture the
functioning of the legal system, the enforcement of property rights and the ability to contract
for commercial services, as they relate to individuals. The lower the score, the better the
provision of civil rights within the economy. We therefore expect that CIVRIGHTS will be
negatively related to the consumption of life insurance.
As it is arguable that the results obtained are a consequence of the legal variable chosen, a
second variable, ROL is also analysed. It is derived from the International Country Risk Guide
Dataset (ICRG) utilized by La Porta et al. (1998), among others, as a basic measure of the law
and order tradition (rule of law) within a country. As it is more specific, we use ROL as a
benchmark for the preferred measure of CIVRIGHTS which covers a broader range of civil
liberties. With improvements to the rule of law in a given country we would expect the
protection and enforcement of property rights to facilitate the transaction of insurance.
Therefore, it is hypothesized that there is a positive relationship between the consumption of
life insurance and ROL.
Unfortunately, even though the written law may be supportive of the insurance market,
the protective benefit of the law for policyholders needs to be recognized within the
constraints of the institutions and the political system that provides and enforces such rules.
For example, Pistor, Raiser and Gelfer (2000) show that in transition economies it is the
effectiveness of legal systems in enforcing the law, rather than the actual law on the books,

2
Indeed, as an example, two key features of the current U.K. life insurance market have been the use of the
courts in settling disputes between managers and policyholders at the Equitable and the AXA insurance companies. In
the high inflation, high interest rate environment of the 1970s, Equitable offered guaranteed annuity rates which it
never envisaged having to pay. In the recent low inflation, low interest environment, the company faced crippling
liabilities but was forced by the courts to honour its contracts. In the case of AXA, following unprecedented
investment returns in the 1950s and 1960s its life fund contained unallocated reserves, or so called orphan assets.
Again, the courts were used to allocate these between policyholders and shareholders. Such examples show clearly
how uncertainty in investment returns over long time periods can generate scope for opportunism and a reliance on the
courts is inevitable in many cases.

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 399

which has a greater impact on the level of financial intermediation. Furthermore, it is

important to recognize the dynamic variability that can exist in the enforcement and provision
of legal rights. Knack and Keefer (2000) argue that increasing polarization or inequality in a
society leads to a higher probability that the government will act to redistribute wealth. In
doing so they may increase taxation, withdraw property rights, or pursue changes to the
regulatory environment for business. In all cases such policies have the ability to reduce asset
values. Moreover, in the specific case of the East Asian financial crisis, Bongini, Claessens
and Ferri (2000) argue that in providing support to distressed companies on political rather
than economic grounds, governments extended the resolution process and hence the financial
crisis. However, even on a general basis, when faced with the potential for legal changes, both
firms and individuals will be less willing to develop business relationships. This will be
especially true for long-term relationships.
Therefore, in order to understand the role of the legal system in promoting life insurance
consumption, it is important to recognize the stability of such systems over time. In the
developed world where the degree of polarization is less and the political environment is more
stable, we can also expect the legal system to be stable. In contrast, within the emerging
economies of Asia it is more likely that polarization is greater, political stability is reduced
and changes in the legal environment more likely. In order to capture the political
environment and potential for legal instability we consider the measure CHECKS, and an
alternative CO, both of which are described in Beck, Clarke, Groff, Keefer and Walsh (2000).
CHECKS3 is a measure of checks and balances within a country’s political system that
broadly measures the number of individuals with a veto within the political apparatus of a
country. The CHECKS index increases with the number of checks in the political system.
Such checks should promote the effective running of the political system and lead to
improvements in political actions and enforcement. For this reason we also expect CHECKS
to be positively related to the consumption of life insurance.
As an alternative index we also examine CO, a measure of political cohesion originally
devised by Roubini and Sachs (1989). This variable captures the extent to which government
control is vested in one or more political parties. Control vested in one party is associated with
greater political cohesion and is expected to be positively related to greater stability and
enforcement across political and legal institutions. As such, CO is expected to be positively
related to the consumption of life insurance.
Finally, as an alternative to utilizing political checks and balances a religious dummy,
RELIGION, is included as an indicator of whether the population is predominately Muslim.
In some Muslim-dominated societies there are possible legal or political conditions that may
inhibit growth of the insurance sector. In fact, some Islamic laws do not encourage certain
forms of insurance, which facilitate speculation of future events. Takaful, mutual insurance, is
an exception but not widely common.

2.2 Economic factors

Probably the most influential determinant for purchasing life insurance is income. Work
dating back to Fortune (1973) has shown a direct positive link between income levels and the
purchase of life insurance. In line with previous studies, income is measured as real GDP per

3
The measure applied in this paper also takes into account the policy orientation of the parties.

# 2002 The International Association for the Study of Insurance Economics.

400 WARD AND ZURBRUEGG

capita, GDPCAP. However, the significant issue for this study is the extent to which the
income effect is homogeneous across the emerging and developed economies, particularly
those in Asia. The ‘‘S curve’’ relationship highlighted by Sigma (1999) and Enz (2000)
illustrates how the consumption of life insurance rapidly accelerates as an economy begins to
develop. Their evidence seems to suggest this is particularly true for populations that have a
GDP per capita of between U.S.$1,000 and U.S.$10,000. After this point the insurance
industry undergoes diminishing returns to economic growth. This is not surprising, for as
income levels grow individuals will begin to save. Truett and Truett (1990) provide support
for these arguments by showing that the income elasticity for insurance in Mexico is around
three times higher than in the U.S.
A further explanation can be found in the bequeath motive for life insurance. Hau (2000),
working only with U.S. data, shows that when an individual’s holdings of net liquid assets are
low and government-backed annuity wealth is high, there is little scope to bequeath financial
assets to dependents. Hence, consumption of life insurance increases. Therefore, when
focusing on emerging economies, individuals will ordinarily have accumulated few assets and
will be drawn to life insurance for its bequest capabilities. Accordingly, when we take all these
ideas together we expect income elasticity to be higher in the rapidly growing economies of
Asia than in the developed economies of the world.
Second, the real inflation rate, INFLATION, is also included as there is substantial
evidence to suggest a very strong, negative relationship exists. As inflation erodes the cash
value of any sums received in the future, the benefits from purchasing life insurance
diminish. Further, higher levels of inflation are associated with macro-economic
uncertainty and as a result the discounted value of financial assets, including life
insurance, will be less. Indeed the impact of inflation is so strong that a study by Babbel
(1981) noted that demand was less in periods of high inflation for indexed cash value life
insurance products.
Finally, Outreville (1996) has shown that financial development, DEPTH, as an aid to
financial intermediation can be an important source of growth in the insurance industry. A
common measure of financial development is the ratio M2/GDP. However, the more specific
measure of Private Credit from Banks and other Financial Institutions over GDP, as deployed
by Levine et al. (2000), is used in this study as it focuses on financial institutions involved in
financial development.

2.3 Social factors

As with the other empirical studies, life expectancy, LIFE, is included. It is expected to
hold a positive relationship with life insurance consumption. Outreville (1996) argues that it
reflects the actuarially fair price of life insurance, as the longer one is expected to live, the
more premium payments will be made. However, as a proxy for the price of insurance life
expectancy has a number of problems associated with it. First, it refers to the general
population, not the pool of risks insured by the average life insurance company. Second, life
expectancy is notoriously inaccurate (see McDonald, 1996; Blake and Burrows, 2001), with a
need to price into life insurance the additional risk of life expectancy error. Third, long-
running insurance will have been underwritten and priced with very different assumptions of
mortality risk from those currently used. Fourth, life expectancy does not account for the
ability to discount life insurance liabilities by investment returns. Therefore, an improved
proxy for price would take account of the statistical distribution associated with mortality risk
over the population and the expected distribution of investment returns. However, such

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 401

measures do not currently exist.4 Therefore, following the previous literature we will continue
using life expectancy while suggesting the development of a better pricing proxy would be an
excellent topic for further research.
WELFARE measures average social welfare expenditure as a percentage of GDP. It is
expected that as a government increases its social provision, the need for individuals to make
private provision against longevity or early death, via life insurance, is reduced.
The final two variables are annual education and dependency ratios for each country. The
ratio, EDUCATION, measures the proportion of the population completing secondary
schooling. Following Browne and Kim (1993), it serves a dual purpose of not only measuring
risk aversion (on the assumption that an individual’s education level is positively related to
greater risk aversion) but signifies that with more people completing secondary education, the
more individuals are dependent on family income earners. The dependency ratio, YOUNG,
measures this last point more succinctly. The chosen measure is the number of young
dependents over the total working population. As this ratio grows, the more likely it is that
income-earners will purchase life insurance in the event of a premature death.5
The full list of variables and their hypothesized signs are tabulated in Table 1. Also, with

Table 1:
Summary of Hypotheses
Expected relationship with
Hypotheses Proxies demand for life insurance

Economic and social factors

• Real GDP per capita GDPCAP +ve
• Inflation rate INFLATION ve
• Financial development DEPTH +ve
• Social welfare expenditure WELFARE ve
• Young dependency ratio YOUNG +ve
• Risk aversion EDUCATION +ve
• Price of insurance LIFE +ve
Political rights
• Checks and balances CHECKS +ve
• Political cohesion CO +ve
Legal rights
• Civil rights CIVRIGHTS ve
• Rule of law ROL +ve
• Muslim country RELIGION ve

4
For an indication of the problems associated with discounting a stochastic variable with a stochastic discount
factor see Klumpes (2001).
5
Recent empirical studies that have employed a similar dependency ratio include, among others, Truett and
Truett (1990), Browne and Kim (1993) and Outreville (1996).

# 2002 The International Association for the Study of Insurance Economics.

402 WARD AND ZURBRUEGG

the exception of WELFARE and figures relating to Taiwan, all the data for the social and
economic variables were collected from the World Bank. For WELFARE, some of the World
Bank statistics were supplemented with welfare expenditure data obtained from national
statistical reports. Taiwanese government statistical yearbooks were also used for the
collection of data not available from the World Bank.

3. Descriptive statistics and methodology

Table 2 provides descriptive statistics for the full unbalanced6 panel dataset used. The
choice of countries included within the sample was based on the availability of life insurance
premium data for the countries between 1987 and 1998. As a comparison, the sample is also
split into two: one for Asian countries and one for developed/OECD countries. There are five
exceptions to this rule. Due to the highly developed nature and size of the insurance market in
Japan, it is excluded from the Asian sample. Also, the samples are not mutually exclusive,
with Singapore,7 South Korea, Israel and Turkey being represented in both sub-samples,
primarily due to the fact that they also have developed economies. Despite this, it is noticeable
that the level of life insurance consumption is higher in the developed economies. This is not
surprising given the higher level of GDP per capita and the lower rate of inflation. In addition,
financial development, civil rights, political checks and balances, rule of law and political
cohesion, all thought to promote insurance consumption, are more favourable in the OECD
sub-sample.
The table also contains the correlation scores for the entire sample. With the exception of
the young dependency ratio, all the figures support our hypothesized relationships between
the independent variables and life insurance density. However, it is also notable that the
correlation between GDPCAP and LIFE is 0.91. This figure is not surprising when one
considers that income and general living standards will inevitably lead to a longer life
expectancy. Including both of these measures as independent regressors will increase the risk
of generating spurious results due to the high level of multicollinearity. Although previous
studies have not dealt with this potential problem, we will exclude LIFE in order to alleviate
the problems of collinearity. However, we also conduct ancillary regressions that include
LIFE, which show its significance as an explanatory variable. The reason for the negative
relationship between YOUNG and life insurance density will be explored in the empirical
section, as part of the explanation for this lies in differences in population growth between the
developed and developing countries.
To provide a more comprehensive examination of the relationships between these
explanatory factors and life insurance density, two specific methodologies are applied. In both
cases logarithmic values are taken. This creates linearity in the data and also provides for the
estimation of elasticities. First, pooled cross-section OLS regressions are performed for the
two primary sub-samples with only the basic economic and social set of variables included.
These regressions include time-specific dummies to account for time variation, but no country
specific effects in order to maximize cross-country variation. The results are presented to
make a direct comparison with the existing literature on life insurance which has primarily
used cross-sectional datasets. Secondly, to determine whether there are any transitory or

6
Where possible, data was interpolated to maximize the number of observations.
7
Singapore is not a member of the OECD, although for the purposes of this study we also place it in the OECD
sub-sample.

# 2002 The International Association for the Study of Insurance Economics.

 LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA
Table 2:
Descriptive statistics
Asia

LIFEDEN INFLATION GDPCAP EDUCATION DEPTH WELFARE YOUNG CIVRIGHT CHECKS ROL LIFE CO
Mean 219.58 12.78 5263.29 63.79 0.67 3.02 61.87 4.11 3.65 3.64 68.78 0.94
Median 34.61 07.19 3916.50 60.00 0.67 1.46 60.32 4.00 3.00 4.00 69.02 1.00
Std. Dev. 302.19 18.36 3846.52 22.49 0.40 3.61 18.04 1.35 2.44 1.40 4.88 1.00
OECD
LIFEDEN INFLATION GDPCAP EDUCATION DEPTH WELFARE YOUNG CIVRIGHT CHECKS ROL LIFE CO
Mean 551.04 07.49 12060.79 99.95 0.89 11.41 51.28 1.76 4.03 5.27 75.41 1.18
Median 505.20 3.41 12969.00 100.00 0.81 11.53 49.76 1.00 3.00 6.00 76.40 1.00
Std. Dev. 382.69 14.34 3927.72 20.61 0.47 6.48 7.00 1.19 2.12 1.18 3.57 1.11
# 2002 The International Association for the Study of Insurance Economics.

Full sample
LIFEDEN INFLATION GDPCAP EDUCATION DEPTH WELFARE YOUNG CIVRIGHT CHECKS ROL RELIGION LIFE CO
LIFEDEN 0.6109 0.8372 0.7271 0.7126 0.0183 0.5450 0.6850 0.1916 0.6203 0.6467 0.7236 0.1554
INFLATION 0.5427 0.4317 0.5671 0.0449 0.3941 0.4238 0.1640 0.5531 0.3984 0.5222 0.1812
GDPCAP 0.7534 0.5824 0.1652 0.6145 0.7402 0.2276 0.7888 0.4844 0.9144 0.2168
EDUCATION 0.4300 0.2397 0.5002 0.6597 0.0419 0.6434 0.5820 0.7548 0.1615
DEPTH 0.0430 0.4787 0.3147 0.0723 0.5258 0.4657 0.5507 0.0170
WELFARE 0.0112 0.0024 0.1546 0.1526 0.1175 0.2047 0.2059
YOUNG 0.4263 0.1457 0.6196 0.6686 0.6461 0.0960
CIVRIGHTS 0.2039 0.5643 0.5585 0.6905 0.2558
CHECKS 0.0907 0.0024 0.1981 0.6054
ROL 0.4330 0.7892 0.1836
RELIGION 0.5494 0.0467
LIFE 0.2053
CO

403
404 WARD AND ZURBRUEGG

persistence effects within the life market from previous life insurance consumption, panel
regressions are performed that examine within-country differences. Specifically, rather than
conduct a standard fixed effects model, a General Method of Moments (GMM) system
estimation is performed that allows for the inclusion of a lagged dependent variable in order to
distinguish between transitory and permanent influences upon life insurance consumption.
Specifically, this model can be notionally written as:
yi,t ¼ Æ yi, t1 þ  xi, t þ  i þ  t þ  i, t (1)
where y is the dependent variable, yi, t1 is the lagged dependent variable, x is the set of
explanatory variables,  is a country-specific effect,  is a time-specific effect,  is a time-
varying error term, and i and t denote country and time periods, respectively.
While the fixed effects model is able to account for both time and country-specific
effects, estimates will be biased and inconsistent when a lagged dependent variable is
included.8 For this reason this procedure is not appropriate and the preferred estimation
technique is the GMM dynamic system estimator developed by Arellano and Bond (1991)
and Arellano and Bover (1995), and also employed by Levine et al. (2000), among others.
The GMM estimator provides the ability to model lagged dependent variables as
regressors within a dynamic framework, while assuming weak exogeneity.9 In brief, the
GMM system estimator involves the simultaneous estimation of the model in differences and
levels, with lagged levels of the independent variables used as instruments in the differences
equation and lagged differences used as instruments in the levels equation. Blundell and Bond
(1998) provide appropriate moment conditions for the estimation of this system and go onto
show the GMM estimator offers significant improvements in terms of bias and precision over
non-system estimators.
Importantly, the consistency of the GMM system estimator depends on a number of
assumptions. Specifically, the assumption of no serial correlation in the residuals,  i, t , and on
the validity of the instruments. Under the null hypothesis of no serial correlation, the test of
first-order serial correlation of the differenced residuals should be significantly negative, and
the second-order test should be insignificant. The Sargan test of over-identifying restrictions
can also be conducted, and is based on the null hypothesis that the instruments are valid
(Arellano and Bond, 1991). Failure to reject the null hypothesis supports the model
specification.10 For this reason, both the Sargan and serial correlation test statistics are
presented alongside the GMM system results to test the validity of the estimation procedure.
Although the purpose of including the GMM estimator results are also to partially
provide a comparison with the cross-sectional regressions, slightly different datasets are used.
Despite annual data being collected, many of the variables do not experience much time
variation on an annual basis. This is particularly true of the legal and political factors for the
more developed countries. In order to deal with this problem, one technique is to segment and
average the data. In this paper non-overlapping two-year averages are calculated and used for

8
See Verbeek (2000) for details on the limitation of fixed effects models with lagged dependent explanatory
variables.
9
Strict exogeneity implies that the explanatory variables are uncorrelated with current, future and past values
of the error term. This is in contrast to weak exogeneity that allows for current explanatory variables to be affected by
past and current values of the dependent variable, but are not affected by future changes in the dependent variable.
10
We use the Dynamic Panel Data (DPD) for Ox program written by Doornik, Arellano and Bond (1999) to
implement the GMM system estimator.

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 405

the GMM estimation. Although this halves the number of observations, it also increases the
time variation within the data.

4. Empirical results
Table 3 presents the pooled cross-section OLS regression results.11 The first set of
regressions only include a subset of social and economic variables which can lead to a
comparison of the results from previous research by, for example, Browne and Kim (1993)
and Outreville (1996). With the exception of the young dependency ratio in the ASIA sample,
the estimated co-efficients have the expected sign and are statistically significant to at least the
10 per cent level. The negative sign for the YOUNG dependency ratio may in our dataset
proxy for the rise in the young adult population over the period, rather than proxy for the
number of dependents. Unlike the more developed countries, population growth is still
evident in most parts of Asia. Therefore, a growing young adult population, with few family
commitments, is unlikely to seek life insurance. However, aside from this, these initial results
are similar to previous research on life insurance markets and support the premise that our
underlying data is comparable.
Since all of the variables have undergone a logarithmic transformation, the estimated co-
efficients in Table 3 are measures of elasticity. Any value greater than 1 indicates that a change
in that variable will drive an even bigger change in the consumption of life insurance. So for
the first set of results the Asian sub-sample has an estimated co-efficient of 1.313 for
GDPCAP, signifying that consumption is income elastic. More specifically, an increase in
income of 10 per cent will increase the consumption of life insurance by 13.13 per cent. At this
point it is therefore worth noting the differences in the magnitude of the estimated co-
efficients across the two samples. In contrast to Asia, the consumption of life insurance in the
OECD sample is around three times less sensitive to changes in income with an estimated co-
efficient of 0.4485. This result is not surprising given the higher average income level in the
OECD sample and is supportive of the ‘‘S curve’’ hypothesis where, at higher levels of
income, insurance consumption becomes less sensitive to income growth.
Following the above reasoning, inflation and, we would hazard, economic uncertainty, is
around two and a half times more important in the ASIA region as opposed to the OECD
sample. Conversely, in the developed economies a 10 per cent improvement in financial
development, DEPTH, will lead to approximately a 12 per cent increase in life insurance
consumption. In the ASIA sample the improvement in consumption would only be around 2
per cent. This is an interesting result as it would imply that the importance of the financial
intermediary sector is to complement insurance consumption in developed economies. In the
Asian economies where financial intermediation is not as broad the impact is not as great.
The second group of results augments the earlier specification with legal and political
variables. In the first instance we use CIVRIGHTS and CHECKS as our measures of the legal
and political environment, while in the later specification we opt for our alternative measures
of ROL, CO and RELIGION. These alternative regressions using differing legal and political
determinants will help to provide an indication of the robustness of the results, independent of
the variables used.
Taking ASIA first, improvements in civil rights have a significant and positive impact on

11
All regressions are de-trended for time by the inclusion of time dummies and co-efficient tests are corrected
for heteroskedasticity using White’s standard errors.

# 2002 The International Association for the Study of Insurance Economics.

# 2002 The International Association for the Study of Insurance Economics.

406
Table 3:
Determinants of life insurance consumption in ASIA and OECD OLS pooled cross-section estimatesa
Basic conditioning variables Legal and political (1) Legal and political (2) Exclusion of GDPCAP Asia sub-samples

ASIA
ASIA OECD ASIA OECD ASIA OECD ASIA OECD (non-Tiger) Asia Tigers

Variable
Constant 7.7810 6.2220 0.2716 13.3770 7.8081 8.8901 17.7297 16.9031 0.7921 15.3662
(2.6242)


(2.0290)


(2.8888) (2.6379)


(3.0640)


(1.9591)


(6.0718)


(8.6980)

(2.3522) (2.1775)



INFLATION 0.6354 0.2450 0.5841 0.2113 0.6172 0.2354 0.6164 0.2642 0.8134 0.2008
(0.1302)


(0.0589)


(0.0961)


(0.0606)


(0.1062)


(0.0606)


(0.1033)


(0.0665)


(0.1542)


(0.1702)
DEPTH 0.1850 1.1974 0.4614 1.0677 0.2824 1.0715 0.5037 1.1676 0.3363 0.0567
(0.1119)
(0.0933)


(0.1464)


(0.1007)


(0.1674)
(0.1038)


(0.1897)


(0.0970)


(0.2074)
(0.5016)
EDUCATION 1.1055 1.2188 0.8609 2.0602 0.8010 1.4778 0.9931 1.9872 0.5810 0.3166
(0.2633)


(0.2561)


(0.2123)


(0.2885)


(0.1732)


(0.2866)


(0.2617)


(0.2805)


(0.1668)


(0.4274)
YOUNG 0.6738 1.2267 0.8720 1.2062 0.6925 0.8268 1.2226 0.3130 0.6360 1.3889
(0.3809)
(0.3875)


(0.4282)

(0.3864)


(0.5180) (0.4004)

(0.3771)


(0.4201) (0.4033) (0.4069)



GDPCAP 1.3130 0.4485 0.8984 0.5728 1.6989 0.7098 0.7328 1.7990
(0.1575)


(0.1405)


(0.1364)


(0.1619)


(0.1235)


(0.1552)


(0.2088)


(0.2461)



LIFE 6.1675 3.1300
(1.4007)


(1.9459)
WELFARE 0.0024 0.0227 0.0259 0.0162 0.0826 0.0613 1.6379 0.3930
(0.0131) (0.0061)


(0.0130)

(0.0065)


(0.0397)


(0.0085) (0.3891)


(0.1610)



CIVRIGHTS 0.5492 0.2324 0.6210 0.0628 0.5069 0.2873
(0.0532)


(0.1476) (0.0492)
(0.0584)


(0.0630)


(0.0866)


CHECKS 0.0345 0.0357 0.0610 0.0960 0.0517 0.2246
(0.0313) (0.0146)


(0.0355)

(0.0175)


(0.0398) (0.0810)



ROL 0.4958 0.2176
(0.0714)


(0.0521)



CO 0.2573 0.0123
(0.0667)


(0.0343)

WARD AND ZURBRUEGG

RELIGION 0.7291
(0.1428)



Observations 156 273 150 273 150 273 151 273 95 55
F test /Chi Square 39.2241


53.8027


70.4938


52.6007


56.2903


49.5857


57.3720


52.2066


32.7183


19.0047



2
Adjusted R 0.7872 0.7444 0.8936 0.7735 0.8758 0.7628 0.8647 0.8291 0.8516 0.8334

a


,

and
indicate significance at the 1%, 5% and 10% levels respectively.
LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 407

life insurance consumption. With a 10 per cent improvement generating a 5.5 per cent
increase in consumption of life insurance. However, the make-up of the political system is
found to be statistically insignificant. It should also be noted that the inclusion of legal and
political variables markedly reduces the income elasticity measure in ASIA, indicating that
after controlling for legal and political effects income elasticity for life insurance in Asia is
inelastic. Finally, for the Asia region WELFARE is insignificant which provides little support
for the argument that consumption of life insurance in the Asia region is driven in part by low
public sector provision of social insurance.
By contrast, in the OECD sample, welfare provision has a significant and detrimental
effect on the consumption of life insurance, with a 10 per cent increase in welfare provision
leading to a 0.2 per cent reduction in life insurance consumption. However, for the OECD
sample, civil rights are found to be statistically insignificant, perhaps reflecting that the level
of civil rights do not vary across these economies very much and have reached an acceptably
high standard. This is despite the fact that the political environment is found to be a positive
and significant determinant of life insurance consumption, with a 10 per cent improvement in
the political process generating a 0.4 per cent improvement in consumption.
In order to check that these results are not spurious to our choice of variables we also
present results for our alternative measures of the legal and political environment. For the rule
of law (ROL), it turns out to be a significant determinant of life insurance consumption in both
the OECD and ASIA samples. However, with an estimated co-efficient of 0.4958, the impact
of the law is two and a half times more important in ASIA than in the OECD sample.
Moreover, with higher estimated co-efficients for the ASIA sample, the impact of the political
system (CO and RELIGION) and the welfare state are also of greater importance for
insurance consumption in ASIA, as opposed to the OECD sample.12
In conjunction, the results presented do sometimes contrast with each other. In
particular, welfare provision seems important in one regression for ASIA, but not the other,
while evidence of legal determinants, as measured by CIVRIGHTS, is weak for the developed
countries. However, these results may be more reflective of the differences between what the
various political and legal variables are measuring. Despite this, the results do show broad
evidence to support the hypothesis that the legal and political environment are key
determinants of life insurance consumption in both developed and emerging economies.
This is clearly important to policymakers and new entrants to the Asian insurance markets, as
it highlights the legal and political environments as key structural determinants of future
expansion in these markets. Policymakers can use this knowledge to aid the development of
these markets and entrants can use this knowledge to select markets that they should and
should not enter.
Significantly, the results so far are restricted through the omission of LIFE, which was
seen in Table 2 to be highly correlated with GDPCAP. To recognize this problem the model is
re-estimated with LIFE taking the place of GDPCAP. The results in Table 3 show that an
improvement in LIFE expectancy is important in ASIA, but not in the OECD sample.
Whether it is possible to consider LIFE as being a measure of the price of insurance is
questionable, given its close relationship with income and the wealth of a country. Indeed, the
results suggest that if life expectancy does increase by 10 per cent there would be a
corresponding rise of over 60 per cent in life insurance consumption. In the OECD economies

12
RELIGION was excluded from the OECD sample due to the dummy variable only being equal to 1 for
Turkey.

# 2002 The International Association for the Study of Insurance Economics.

408 WARD AND ZURBRUEGG

this relationship is probably not displayed because of the fact that life expectancy has
stabilized and thus there is little variation in the data. However, this alternative specification
does also confirm the importance of legal and political factors in the ASIA and OECD
samples which continue to be significant determinants of life insurance consumption,
regardless of whether LIFE or GDPCAP is used.
As a recognition of the fact that economic development varies across Asia and because
insurance has developed most markedly within the so-called ‘‘Tiger Economies’’ of South-
East Asia, we also split our Asian sample into Tiger and Non-Tiger economies. Interestingly,
from the results in Table 3 it is clearly evident that even after controlling for legal and political
factors, income elasticity is much higher in the Tiger countries than elsewhere. In fact, out of
all our specifications the Tiger economies gain the highest and therefore most elastic measure
for income. The inelasticity of the non-Tiger economies is probably due to the low purchases
of insurance products in these countries to start with, as life density is an average of only
$45.86 per capita. Furthermore, public sector provision of welfare systems is seen to have a
larger and negative impact in the Tiger economies. This possibly suggests that these
governments need to recognize the negative impact that welfare provision can have on the
development of life insurance, and financial intermediation more generally. This is
particularly true where welfare provision is a viable alternative to private insurance. Finally,
as with the previous results, even within these further sub-samples there remains strong
evidence for the role of legal and political factors, measured by CIVRIGHTS and CHECKS,
in the consumption of life insurance.
Finally, as a means of examining the time dimension of the data and analysing any long-
run persistence effects within the consumption of life insurance, Table 4 presents the GMM
system estimator results for the panel data. It is important to realize that with the inclusion of
country-specific effects, our political and legal measures are not in themselves proxies for
other country-specific characteristics.
The results show the beneficial role of the legal environment in determining life
insurance density. However, no evidence is shown to support the impact of the political
environment. For the elasticity of income, there is no significant difference, at the 1 per cent
level, between the sub-samples. Also, not all of the remaining social and economic variables
are significant, although where they are, they hold the right hypothesized sign.
It is difficult to make any direct comparisons between these results and the previous
cross-sectional regressions due to the inclusion of country-specific dummies in the equations
and, most importantly, the dynamic specification using the lagged dependent variable.
However, the results for both sub-samples do indicate that current life insurance is dependent
upon approximately 40 per cent of its previous level, over a two-year period. This may be
diluting the impact obtained from the other socio-economic variables within the regressions.
This is particularly true of the Asian countries.

5. Conclusion
Recent work by LaPorta et al. (1997, 1998, 2000) and Levine (1998, 1999) has
highlighted how the financial sector is influenced by the legal system within an economy.
Unfortunately, there has been little empirical attention given to such factors influencing the
insurance industry. This study begins to bridge this gap in the literature while at the same time
emphasizing the influence of such factors within the rapidly growing insurance markets of
Asia.
Using a larger dataset than has been previously examined, we find evidence that the

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 409

Table 4:
Determinants of consumption for life insurance in ASIA and OECD
Panel data resultsa
GMM system estimator

ASIA OECD

Variable
CONSTANT 3.0616 16.5848
(5.173) (7.1580)


LIFEDEN 1 0.3924 0.3844
(0.0640)


(0.0872)



INFLATION 0.0201 0.0546
(0.1888) (0.1174)
DEPTH 0.9635 1.0675
(0.2398)


(0.1568)



YOUNG 0.1671 1.8838
(0.7090) (0.9397)


EDUCATION 0.5088 1.7415
(0.3313) (0.6027)



GDPCAP 0.5255 0.4991
(0.2570)

(0.2539)


WELFARE 0.0088 0.0083
(0.0204)
(0.0083)
CIVRIGHTS 0.2289 0.2595
(0.0821)


(0.1582)

CHECKS 0.0318 0.0047
(0.0593) (0.0278)

Observations 66 124
F test/Chi Square
Adjusted R2
Wald joint test [pvalue] 0.0000


0.0000



Sargan test 12.5221 13.2700
AR (1) test [N(0,1)] 1.6653 1.7361
AR (2) test [N(0,1)] 0.5492 0.7549
a


,

and
indicate significance at the 1%, 5% and 10% levels respectively.

improved provision of civil rights and political stability leads to an increase in the
consumption of life insurance in both OECD countries and the ASIA region. More telling
is the difference in the income elasticity between the developed economies and the emerging
markets of Asia. This is consistent with Sigma’s (1999) ‘‘S curve’’ for insurance growth and
the findings of Truett and Truett (1990).
Beyond the academic merit of this comparative study between Asian and the more
developed markets of the OECD is the real significance these results have for policymakers

# 2002 The International Association for the Study of Insurance Economics.

410 WARD AND ZURBRUEGG

and foreign entrants within some of the world’s most rapidly expanding insurance markets.
For instance, the attraction of the Tiger economies to many insurance companies from the
developed world has been the anticipation of high future growth rates linked to greater
economic development. The estimated income elasticities from this study after controlling
for legal, political and dynamic effects would tend to caution against such outright optimism.
For while economic development may drive faster rates of insurance market development in
Asia, the link may not be as strong as first thought. Indeed, economic stability conditioned on
political and legal stability, as in the developed economies of the world, appears just as
important for long run success.

Appendix

Countries included in the study are:

OECD Asia Asian Tigers

Australia China Hong Kong1

Austria Hong Kong1 Korea, South
Belgium India Malaysia
Canada Indonesia Singapore
Denmark Iran Taiwan
Finland Israel Thailand
France Jordan
Greece Korea, South
Ireland Kuwait
Israel Malaysia
Italy Pakistan
Japan Philippines
Korea, South Singapore
Netherlands Taiwan
New Zealand Thailand
Norway Turkey
Portugal
Singapore2
South Africa
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
1
Some of the regressions exclude Hong Kong due to missing political and legal variables.
2
Singapore is not a member of the OECD, although for the purposes of this study it is also
placed in the OECD sub-sample.

# 2002 The International Association for the Study of Insurance Economics.

LAW, POLITICS AND LIFE INSURANCE CONSUMPTION IN ASIA 411

REFERENCES
ARELLANO, M. and BOND, S., 1991, ‘‘Some Tests of Specification for Panel Data: Monte Carlo Evidence and an
Application to Employment Equations’’, Review of Economic Studies, 58, pp. 277–297.
ARELLANO, M. and BOVER, O., 1995, ‘‘Another Look at the Instrumental Variable Estimation of Error-
Components Models’’, Journal of Econometrics, 68, pp. 29–51.
BABBEL, D., 1981, ‘‘Inflation, Indexation, and Life Insurance Sales in Brazil’’, The Journal of Risk and Insurance,
48, pp. 111–135.
BAKER, A. and THOMPSON, S., 2000, ‘‘Governance in Financial Mutuals’’, Managerial Finance, 26,9, pp. 36–48.
BECK, T., CLARKE, G., GROFF, A., KEEFER, P. and WALSH, P., 2000, ‘‘New Tools in Comparative Political
Economy: The Database of Political Institutions’’, Policy Research Working Paper #2283, World Bank.
BEENSTOCK, M., DICKINSON, G. and KHAJURIA, S., 1986, ‘‘The Determination of Life Premiums:
An International Cross-section Analysis 1970–1981’’, Insurance, Mathematics and Economics, 5, pp.
261–270.
BLAKE, D. and BURROWS, W., 2001, ‘‘Survivor Bonds: Helping to Hedge Mortality Risk’’, The Journal of Risk and
Insurance, 68,2, pp. 339–348.
BLUNDELL, R. and BOND, S., 1998, ‘‘Initial Conditions and Moment Restrictions in Dynamic Panel Data Models’’,
Journal of Econometrics, 87, pp. 115–143.
BONGINI, P., CLAESSENS, S. and FERRI, G., 2000, ‘‘The Political Economy of Distress in East Asian Financial
Institutions’’, Policy Research Working Paper #2265, World Bank.
BROWNE, M.J. and KIM, K., 1993, ‘‘An International Analysis of Life Insurance Demand’’, Journal of Risk and
Insurance, 60:4, pp. 616–634.
CUMMINS, D.J., TENNYSON, S. and WEISS, M., 1999, ‘‘Consolidation and Efficiency in the US Life Insurance
Industry’’, Journal of Banking and Finance, 23, pp. 325–357.
CUMMINS, D.J. and DANZON, P.M., 1997, ‘‘Price, Financial Quality, and Capital Flows in Insurance Markets’’,
Journal of Financial Intermediation, 6, pp. 3–38.
DOORNIK, J.A., ARELLANO, M. and BOND, S., 1999, Panel data estimation using DPD for Ox. http://
www.nuff.ac.uk/users/Doornik/papers/dpd.pdf.
ENZ, R., 2000, ‘‘The S-Curve Relation between Per-capita Income and Insurance Penetration’’, Geneva Papers on
Risk and Insurance: Issues and Practice, 25:3, pp. 396–406.
FORTUNE, P., 1973, ‘‘A Theory of Optimal Life Insurance: Development and Tests’’, Journal of Finance, 28:3,
pp. 587–600.
HAKANSSON, N., 1969, ‘‘Optimal Investment and Consumption Strategies under Risk, an Uncertain Lifetime, and
Insurance’’, International Economic Review, 1:3, pp. 443–466.
HAU, A., 2000, ‘‘Liquidity, Estate Liquidation, Charitable Motives, and Life Insurance Demand by Retired Singles’’,
Journal of Risk and Insurance, 67:1, pp. 123–142.
KARNI, E. and ZILCHA, I., 1986, ‘‘Risk Aversion in the Theory of Life Insurance: The Fisherian Model’’, Journal of
Risk and Insurance, 53:4, pp. 606–620.
KLUMPES, P., 2001, ‘‘Alternative Valuation Models and their Association with Insurance Share Prices’’, 23rd U.K.
Insurance Economists’ Conference, 2001, University of Nottingham.
KNACK, S. and KEEFER, P., 2000, ‘‘Polarization, Politics and Property Rights, Links between Inequality and
Growth’’, Policy Research Working Paper #2418, World Bank.
LAPORTA, R., LOPEZ-DE-SILANES, F., SHLEIFER, A. and VISHNY, R.W., 1997, ‘‘Legal Determinants of
External Finance’’, Journal of Finance, 54:6, pp. 1131–1150.
LAPORTA, R., LOPEZ-DE-SILANES, F., SHLEIFER, A. and VISHNY, R.W., 1998, ‘‘Law and Finance’’, Journal of
Political Economy, 106:6, pp. 1113–1155.
LAPORTA, R., LOPEZ-DE-SILANES, F., SHLEIFER, A. and VISHNY, R.W., 2000, ‘‘Investor Protection and
Corporate Governance’’, Journal of Financial Economics, 58, pp. 3–27.
LEVINE, R., 1998, ‘‘The Legal Environment, Banks, and Long-Run Economic Growth’’, Journal of Money, Credit
and Banking, 30:3, pp. 596–620.
LEVINE, R., 1999, ‘‘Law, Finance, and Economic Growth’’, Journal of Financial Intermediation, 8, pp. 8–35.
LEVINE, R., LOAYZA, N. and BECK, T., 2000, ‘‘Financial Intermediation and Growth: Causality and Causes’’,
Journal of Monetary Economics, 46:1, pp. 31–77.
LEWIS, F., 1989, ‘‘Dependents and the Demand for Life Insurance’’, American Economic Review, 79, pp. 452–466.
MACDONALD, A., 1996, ‘‘United Kingdom’’, in A. Macdonald (ed.), The Second Actuarial Study of Mortality in
Europe. Groupe Consultatif des Associations d’Actuares des Pays des Communautés Européennes.
MAYERS, D. and SMITH, C., 1981, ‘‘Contractual Provisions, Organizational Structure, and Conflict Control in
Insurance Markets’’, Journal of Business, 54, pp. 407–434.

# 2002 The International Association for the Study of Insurance Economics.

412 WARD AND ZURBRUEGG

OUTREVILLE, J.F., 1996, ‘‘Life Insurance Markets in Developing Countries’’, Journal of Risk and Insurance, 63:2,
pp. 263–278.
PISTOR, K., RAISER, M. and GELFER, S., 2000, ‘‘Law and Finance in Transition Economies’’, Economics of
Transition, 8:2, pp. 325–368.
RAM, R. and SCHULTZ, T., 1979, ‘‘Life Span, Health, Savings, and Productivity’’, Economic Development and
Cultural Change, 27:3, pp. 399–421.
ROUBINI, N. and SACHS, J., 1989, ‘‘Political and Economic Determinants of Budget Deficits in the Industrial
Democracies’’, European Economic Review, 33, pp. 903–938.
SIGMA, 1999, ‘‘Asia’s Insurance Markets after the Storm’’, No.5, Swiss Re.
SIGMA, 2000, ‘‘Emerging Markets: The Insurance Industry in the Face of Globalisation’’, No. 4, Swiss Re.
SZPIRO, G. and OUTREVILLE, J.F., 1988, ‘‘Relative Risk Aversion Around the World’’, Studies in Banking and
Finance, 6, pp.123–129.
TRUETT, D. and TRUETT, L., 1990, ‘‘The Demand for Life Insurance in Mexico and the United States: A
Comparative Study’’, Journal of Risk and Insurance, 57:2, pp. 321–328.
VERBEEK, M., 2000, A Guide to Modern Econometrics. Chichester: John Wiley.
VISCUSI, W. and EVANS, W., 1990, ‘‘Utility Functions that Depend on Health Status: Estimates and Economic
Implications’’, American Economic Review, 80:3, pp. 353–374.
WARD, D. and ZURBRUEGG, R., 2000, ‘‘Does Insurance Promote Economic Growth? Evidence from OECD
Countries’’, Journal of Risk and Insurance, 67:4, pp. 489–506.
YAARI, M., 1965, ‘‘Uncertain Lifetime, Life Insurance, and the Theory of the Consumer’’, Review of Economic
Studies, 32, pp. 137–150.
YUENGERT, A.M., 1993, ‘‘The Measurement of Efficiency in Life Insurance: Estimates of a Mixed Normal-Gamma
Error Model’’, Journal of Banking and Finance, 17, pp. 483–496.

# 2002 The International Association for the Study of Insurance Economics.