Determinants of the Development

Document Sample
Determinants of the Development Powered By Docstoc
					Determinants of the Development of Insurance in China under the
Cuizhen Zhang (China Foreign Affairs University, China)*
Nong Zhu (INRS-UCS, University of Quebec, Canada)**

Abstract: China’s insurance industry has been undergoing extremely fast growth since the
enforcement of open-up policy. This article, using data for 225 cities, examines the determinants of
China’s insurance development, measured by premium volume, insurance density, and insurance
penetration. Our econometric results reveal that foreign direct investment is more significant for
property than for life insurance. Per capita GDP is the only variable significant for all measures of life
consumption, while the total population, savings deposit, education attainment, telephone ownership
per capita, social welfare expenditure, and young dependency are significant for life premiums.
Variables such as wage level, savings deposit, and investment in fixed assets, report their significant
effect on the demand for property insurance. In addition, the article finds that there exists obvious
regional difference in the development of China’s insurance sector, and the governing factors vary
from region to region.
JEL Classification: C21; D91; G22; O18
Key words: Life insurance; Property-liability insurance, China

   The insurance industry is, with no doubts, an increasingly important sector in China for
recent years. According to the statistics of Swiss Reinsurance Company (2005), China’s
premium income mounted to USD 52.17 billion, equaling to 3.26% of the gross domestic
products (GDP) in 2004, compared to that of USD 1.37 billion, 0.7% of GDP, in 1986 (Swiss
Reinsurance Company, 2005). Life insurance premiums, in particular, increased to USD 35.41
billion, contribution to GDP being 2.21% in the year of 2004, from premiums of USD 278
million, being 0.13% of GDP in 1986.
   Meanwhile, China’s insurance industry has been more and more significant in the world
market. In terms of total written premiums, the rank of China on the world list done by Swiss
Reinsurance Company rose from 29th in 1986 to 11th in the year of 2004, companied by the
share of the world market raising from 0.16 to 1.61% for the same period. 1.2% of non-life
premiums are contributed to China in 2004, while just 0.24% in 1986. For the line of life,
1.92% of written premiums in the whole world is owed to China, whereas only 0.07% in 1986
(Swiss Reinsurance Company, 2005).
   The insurance industry of China by 2008 is to reach a value of USD 119.58 billion in gross
premium income, equating to a compound annual growth rate (CAGR) of 20.4% in the 2003-
2008 period (Datamonitor, 2004). The already achieved considerable development of China’s
insurance industry and the forecasted great potential in the coming years have started to
capture the interests of the academic staff. What factors behind are supporting the substantial
growth? Whether the governing factors proved significant in the other countries also have
effect in case of China? Compared with the relatively extensive empirical literatures on other
countries, industrialized nations in particular, the literature on China is scarce. Zhuo (1999),

Hwang and Gao (2003), Hwang and Greenford (2005) examined the determinants of China’s
life insurance consumption. In contrast, to our knowledge, there is no literature on the
determinants of China’s aggregate property-liability insurance consumption, while Zou (2003)
and Zou et al. (2003) focus their interests on the determinants for corporate demand for
property insurance in China. They are definitely the advancers in the kind of research on
China and consequently make contributions to the literature.
   Nevertheless, we have to point out the shortage of the aforesaid studies on China’s life
consumption. Firstly, their results are not very reliable due to their limited observations. Some
studies build their econometric analysis either on time-series data over the period about ten
years, or on cross-sectional data collected for about twenty provinces. As a common sense, the
so short time series are not enough for econometric investigation, their results being unreliable
as a consequence. Secondly, the observed premiums in reality a joint result from both supply
(the insurer’s side) and demand (the consumer’s side), it is thereby essential that the
regression model have to reflect both of these considerations (Beenstock et al., 1988).
However, none of the variables reflecting the supply side is included in the published studies
on China’s life insurance consumption.
    Furthermore, any ignorance of the background such as the enforcement of open-up policy,
liberalization, and gradual participation in globalization, will necessarily lead to the failure in
telling the truth of China’s insurance development. One of the evidence is the increasing
openness degree, which is often measured by foreign direct investment (FDI). China, since
1995, has been the largest recipient of FDI in the developing countries. More than USD 501.4
billion amount of FDI, in total, flowed into China over the period between 1979 and 2003
(Department of Trade and External Economic Relations Statistics, National Bureau of
Statistics of China, 2005). Thomas (2002: 415) states that: “As foreign investment flowed into
China, the demand for insurance protection grew. The economic growth spurred by foreign
investment generated additional demand for insurance in the domestic economy.” As the
development of China’s insurance industry is motivated by China’s opening-up process (Wu
and Strange, 2000), openness degree is reasonably expected to have some kind of relationship
with the insurance consumption. However, any variables related to open economy are
excluded in the previous analysis on China.
   China, being a country with large dimension, differs in the level of economic development
from region to region, frequently measured by the division of the eastern, the central, and the
western area. As a service sector, regional insurance development shall mirror, to some
degree, the overall economic disparity. Regional difference along with its reasons concerning
insurance consumption within a nation has never been investigated, to the authors’
   This article, depending on the data collected for 225 cities, aims to analyze the prevailing
factors in the development of China’s insurance industry, including life and property-liability
line, under the macro-economic background, and to investigate the regional difference, too.
This article contributes literature in the main aspects as follows. First of all, some variables,
such as the openness degree and sales condition represented by telephone ownership, are for
the first time to be introduced in the econometric analysis on insurance consumption. Also,
market structure measured by Herfindahl index, life expectancy, private savings deposit, and
population count are firstly used to explain the development of China’s life insurance.
Additionally, the investment in fixed assets is never incorporated as an independent variable
in the previous research on property insurance expenditure. The second contribution of this
article goes to the regional difference in insurance consumption, and its causes. Sub-samples

for the eastern, the central, and the western cities in China make it possible to investigate the
governing determinants in different regions in case of China. Thirdly, three dependent
variables including total premiums, insurance expenditure per capita, and insurance
penetration are adopted in our analysis, of which the ratio of premiums to GDP is not used by
Hwang and Gao (2003), Hwang and Greenford (2005), and Zhuo (1999). Finally, the ever
largest and latest database, composed of 225 observations by city for the year of 2003,
guarantee the results more representative and reliable.
   In the following section, we outline the most essential characteristics of the development of
insurance sector in China. Section 3 provides the related literature, on both the theoretical and
the empirical perspectives, more interests will be paid to the prior studies concerning China.
Section 4 explains the data and the specification of our empirical model. Results are presented
in section 5. The conclusion reports in the last section.

   The establishment of the People’s Insurance Company of China (PICC) in October 1949
marks the very beginning of insurance industry in the People’s Republic of China. However,
the domestic business of PICC, the single insurance provider at that time, was suspended due
to the restrictions on private ownership of property and comprehensive social entitlement
programs for the 1959-1979 period (Lee, 2002; Thomas, 2002). It is the reform and open-up
policy initiated in late 1970s that resuscitated the insurance sector in China, with property
business being resumed in 1980 and life insurance in 1982.
   The monopoly of PICC remained till the foundation of the second insurer, Xinjiang Corps
Insurance Company, in 1986,1 followed by Ping An Insurance Company of China in 1988 and
China Pacific Insurance Company in 1991. In 1992, the American International Group (AIG)
was granted the license to write policy in Shanghai, signaling China insurance industry’ door
opening to the international insurers. Many more providers including local, wholly foreign-
owned, and joint ventures insurance companies have started operations in China, especially
since China’s accession to WTO in late 2001. In total, there are 61 insurance companies
providing loss protection for consumers in China by the end of 2003. Motivated by the
noticeable development of insurance sector, China Insurance Regulatory Commission (CIRC),
an exclusive watchdog, was then set up in November 1998.2 We now summarize the
development of China’s insurance sector from the following perspectives.
   Figures 1 and 2 tell the stunning development of the insurance sector in China over the
period of 1986-2004, in terms of total premiums, insurance density, and insurance penetration.
The per capita expenditure in property-liability and life insurance increased from USD 1.3 and
USD 0.3 in 1986, to USD 12.9 and USD 27.3 in 2004, respectively. Property-liability and life
insurance’s contribution to the gross domestic products grew to 1.05% and 2.21% in 2004,
compared with that of 0.52% and 0.13% in 1986.

                                   Figure 1 - Premium income growth
  Xinjiang Corps Insurance Company is in fact the second established insurer in China. But it is seldom regarded
as the mark for the end of the monopoly of PICC because of its business being limited in corps insurance in
Xinjiang Construction Force at the time of its establishment. Later its business region and products gradually
expand, and now it has already grown to be a national insurer named as China United Property Insurance
Company, covering all kinds of non-life risks across the country.
  The central bank, i.e., the People’s Bank of China, is the previous regulation body for insurance industry in


      Million US dollars





                                    1986 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

                                                                     Non-life business      Life business

Source: Swiss Reinsurance Company, Sigma.

                                           Figure 2 - Evolution of insurance density and penetration

                           45                                                                                               3.5

  Us dollars

                           25                                                                                               2.0

                           20                                                                                               1.5   %


                            0                                                                                               0.0
                                1986 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

                                                          Insurance density          Insurance penetration

Source: Swiss Reinsurance Company, Sigma.

   Whereas, in 2004, on the world list released by Swiss Reinsurance Company (2005),
depending on the per capita expenditure in insurance, China just holds the 72th position. The
big gap in insurance density and penetration whether compared to the world average or
industrialized countries, even the emerging markets, as Table 1 shows, on the other hand,
indicates the promising increase of China’s insurance industry. Premiums collected are
expected to exceed $100 billion in 2008, surpassing France and Germany (Binder et al.,


                                 Table 1- World and China Insurance in 2004
                                     Total Business              Life Business           Property-liability
                         Penetration Density               Penetration    Density     Penetration Density
                            (%)       (USD)                   (%)          (USD)         (%)         (USD)
World                       7.99      502.0                   4.55         288.7         3.44        213.3
Industrialized countries    9.02     2966.1                   5.14        1691.7         3.88       1275.0
Emerging markets            3.94       68.7                   2.41          42.1         1.52         26.6
China                       3.26       42.0                   2.21          27.3         1.05         12.9
Source: Swiss Reinsurance Company, 2005.

   Life insurance has been playing the leading role in the insurance sector since the year of
1997, when the life premiums for the first time surpassed property-liability. For the year of
2004, life insurance companies collected 77.59% of the total premiums in China’s market.3
Figure 3 describes the development of urban premium distribution between 1998 and 2003.
The peak value of the overall premium income is on the move towards the right side,
suggesting that its extremely fast growth for the studied period. Figure 3 also tells that the
increase in overall premium income is significantly owed to life business because there
obviously shows a rough superposition between the curve for life and that for the overall
premiums. Compared to that in 1998, the curve for property in 2003 slightly moves to the
right, expressing that the property premiums grew at a much lower speed than life insurance.

                                 Figure 3 – Kernel density of premium income
      Kernel density

                                                      Premium income

                       Total business in 1998         Non-life income in 1998       Life business in 1998
                       Total business in 2003         Non-life income in 2003       Life business in 2003

Source: China Insurance Regulatory Commission, Yearbook of China’s Insurance (various issues).

    The percentage is calculated according to the data released by CIRC,

   From the viewpoint of market structure, China’s insurance market still belongs to be
oligopolistic (Leung and Young, 2002) although the number of insurers, foreign ones in
particular, has been increasing ever since. In contrast to only one provider in 1980, 32 life and
24 non-life insurance companies write policy in China in 2003. The market concentration still
remains very high although has been gradually reducing (Leverty et al., 2004). In the year of
2004, the big-4 life and property-liability insurers collect nearly 90% and 85% of the gross
premiums in their individual line (see Table 2). What needs to notice is that the China Life
Insurance Company and the PICC still dominate more than half of the market , if measurd by
the premiums collected.

                    Table 2 - Market Concentration in China Insurance Market4
                           Life insurance                              Property-liability insurance
            Number         Share of premiums written            Number       Share of premiums written
            of firms                   (%)                      of firms                   (%)
                          1-Firm   2-Firm   3-Firm    4-Firm                 1-Firm    2-Firm   3-Firm   4-Firm
 1996         6a        84.9    92.3    97.9      99.8         8    77.34 89.59 96.91           -
 2001         21       57.08 85.23 95.31 96.93                19    74.11 86.55 95.93 96.86
 2002         23       56.59 80.13 91.08 94.59                23    70.88 84.06 94.79 95.77
 2003         31       53.98 73.86 86.56 92.35                25    69.43 82.14 92.20 94.71
 2004         33       55.17 72.35 83.15 89.02                26    58.09 70.41 79.87 85.69
Note a: The data concerning life insurance is for the year of 1995.
Source: The share of life for the year of 1995 see CIRC and Samsung Life (2003: 53); the data
concerning property insurance for the year of 1996 see Wu (2004: 105); calculation for life insurance
(2001-2003) depends on Chen (2004: 59); the others are computed on the data in Yearbook of China’s
Insurance (various issue).

   The involvement of foreign insurers is one of the evidence that China’s market is in the
midst of internalization. The international companies (37) including foreign-wholly owned
and foreign joint ventures outnumbered the domestic ones (24) in 2003, according to the
statistics from CIRC. However, foreign providers capture only a small part of the gross
premiums (Leung and Young, 2002; Leverty et al., 2004; Shen, 2000). For instance, in 2004,
just 2.27% of the total premiums are contributed to foreign insurance providers,5 in large part
due to the restrictions on them.6 But when we turn eyes to the earliest opened cities such as
Shanghai and Guangzhou, the case is greatly different. In Shanghai, the first opened city and
the financial center in China, the market share of foreign life and property insurers
respectively reached 8.87% and 14.08% in 2003, much higher than the national average. Also
in 2003, foreign life insurers in Guangzhou, the second opened city and also a foreign-
outnumbered city, captured 16.70% of the total life premiums (see Wu, 2004: 45).
   Another characteristic of China’s insurance market is related to the product composition.
Motor vehicle and third-party liability insurance has been the most important source for
property premium income for the latest years, as Table 3 states. In earlier period, the business
property and automobile insurance played nearly equal role in the premium income, whose
premium share was 42% and 48% in year 1985. Motor vehicle insurance replaced the business
  AIA Dongguan, Jiangmen, and Guangzhou Branch are counted as one company.
  The calculation is based on the data released by CIRC,
  According to China’s regulation, foreign insurers are only permitted to write policy in their company-based
geographic area. Non-life international insurers only can underwrite the risks of foreign-invested enterprises,
while they are banned from writing the compulsory insurance such as third-party liability for motor vehicle.
Foreign life providers are limited to sell individual life policies and prohibited from writing group life products
or any type of health, pension, and annuities products. Most of the restrictions are in the middle of being removed
owing to the fulfillment of the commitment for the accession to WTO.

property to be the dominating product for the first time in 1988, and its share increased to the
summit in 1997 (Wu, 2004: 107-108) and remains at 60% or so ever since then. The product
composition proves what Lee (2002:17) says: “To date, personal property and casualty
insurance - including individual automobile, dwelling, and third-party liability policies -
remains significantly underdeveloped.” For personal insurance, in 2003, life insurance,
pension policy, and healthcare insurance, separately accounts for 74.28%, 14.34% and 8.06%
of the total life premiums. And the policy covering medical expense has increased at a stead
pace for recent years. Lately, participating and unit-linked policy grab quite a large number of
Chinese consumers, which jointly contributes over 60% to the total premiums in 2003(see
Wu, 2004: 138-139). In reality, foreign providers’ failure in realizing that many Chinese
consumers’ interest in insurance or financial products is closely linked to invested and
financial management, partly lead to their low market share (see Ji and Thomas, 2001).

                  Table 3 - Product Composition of Property insurance in China
                                                   1985       2001        2002             2003
Automobile and third-party                         42.0        61.3       60.6             62.1
Business property                                  48.0        17.7       15.7             14.4
Cargo and transport                                 6.0         5.9        5.4              4.7
Private property                                    4.0         2.7        3.0              2.2
General Liability                                     -         4.0        4.7              4.0
Others                                                -         8.3       10.5             12.6

Total                                               100.0        100.0      100.0         100.0
Source: The composition for the year of 1985 and 2001 see Wu (2004: 108); the rest is calculated
from the data in Yearbook of China’s Insurance (various issues).

   Regional disparity in the insurance consumption obviously appears in China, with the
coastal area being the leader, followed by the central and the western region. Table 4
summaries the premium regional distribution in China. In 2003, more than 2/3 of gross
premiums were collected from the coastal region, and the central and western region just 1/3.
The similar situation shows even when we separate the overall business into life and property-
liability line. If analyzed the regional distribution from the perspective of annual growth rate,
the coastal region is also the leading one, especially in life business. It is obvious that the
coastal area was and will be the most important contributor for the development of insurance
industry in China.

                         Table 4 - Premium income distribution in 2003

                                                                       1998-2003 annual growth
                     Premium income                Percentage                    rate
                       (million yuan)                 (%)                        (%)
                  Total Property Life       Total Property Life         Total Property Life
                business business business business business business business business business
                   (%)       (%)      (%)    (%)      (%)      (%)       (%)     (%)      (%)
China           395472 90806 304650 100.0            100.0    100.0    25.3     11.1     32.7

Coastal region 252311 58101 194210         63.8     64.0     63.7      26.1      11.9      38.7
Central region 87107 16586 70506           22.0     18.3     23.1      25.8       8.0      34.6
Western region 56053 16119 39934           14.2     17.8     13.1      21.1      11.9      28.5
Source: China Insurance Regulatory Commission, Yearbook of China’s Insurance (various issues).

   We also can investigate the regional distribution, using provincial data. In 2003, overall
premiums collected in Jiangsu and Guangdong, located along the coastal line, accounted for
more than 9.5% of national premiums, respectively. Meanwhile, the market share of Tibet,
Qinghai, and Ningxia, belonging to the western region, is individually less than 0.5% for the
same year. Regarding life insurance, more than 10% of written premiums are contributed to
Jiangsu Province, whereas the joint share of the western region just 13.1%. For property
business, the market share of Guangdong Province reached 13.8% of the national premium
   Figure 4 presents the evolution and the decomposition between interior and coastal areas of
the Theil index, computed on the basis of the share of a given province’s premium income and
population, over 1997-2003 period. We can observe that, compared with the population’s
regional distribution, inter-provincial differences in terms of premium income decreases to
the year of 2000 and then goes up. The decomposition of the Theil index clearly indicates that
the decline in disparities that occurred over the period 1997-2000 resulted essentially from
the reduction of the gap within the region concerned. On the contrary, the surge in total
disparity in 2000 is rooted in the widening disparity between coastal and interior provinces.
In other words, premium income levels have diverged between coastal and interior provinces
of China, while premium income’s homogeneity has improved within each region.

                                       Figure 4 - Evolution of the Theil Index

     Theil index




                         1997   1998          1999            2000             2001               2002   2003
                                       Inter-regional Theil index    Intra-regional Theil index

   We firstly shed light on the determinants of life insurance demand. Nearly all the theories
on the demand for life insurance purchase have been developing on the basis of Yaari (1965).
Bequest motives and risk aversion are analyzed as the main factors to motivate individuals to
buy life insurance in the early theoretical literature.7 Among of them, Lewis (1989), from the

    See for instance Bernheim (1991), Campbell (1980), Fisher (1973), Fitzgerald (1987), Hakansson (1969), Karni

perspective of beneficiaries, presents a function of life insurance demand, which in reality
builds the theoretical starting point for many empirical works.
   Mantis and Farmer (1968) examine the determinants of America’s life insurance demand,
measured by total sales per year over the 1929-1964 period, which may be one of the earliest
empirical studies in this field. According to Zietz (2003), twenty-six academic empirical
studies examining the determining factors associated with a consumer’s life insurance demand
had been published. Five categories of parameters including demographic, macroeconomic,
socio-psychological, institutional, and insurer-side variables are investigated to be the
governing factors for the life insurance coverage. The literature specifies that disposable
income, the inflation rate, financial development, social security, and some population
variables such as young dependency, aged dependency, birth rate, educational level, life
expectancy, are verified as the robust explanatory variables for the life insurance
consumption.8 However, the determinants for life insurance demand vary from country to
country. In case of China, the determining factors are not conclusive in the published studies.
   Zhuo (1999) adopts three models to test the determinants of China’s life insurance
consumption. Using cross-sectional data for all provinces (29 observations) except Tibet for
the year of 1995, his study finds that per capita GDP, young dependency rate are the
significant determinants. He then employs the national time-series data for the period of 1986-
1995 (10 observations) and just per capita GDP proves significant, which also the only
significant determinants in his third model based on the cross-sectional data for 14 large cities
in China. Hwang and Gao (2003) introduce four variables of income, inflation, urbanization
and education to examine their influence on the life insurance consumption over the period of
1986-1996. GDP per capita, urbanization, and educational level are found significant and
positive, while the influence of inflation shows not significant. Lately in 2005, Hwang and
Greenford (2005) use variables including income, education, social security, social structure,
one-child policy, price of insurance, economic development to try to explain the development
of life insurance from the perspective of comparison among China mainland, Hong Kong and
Taiwan. Their study expresses that real GDP per capita, education level, social structure
measured by agricultural population ratio of total employed affect the insurance consumption
in a significant and positive way, whereas one-child policy is significantly negative.
   We now turn to the determinants for the property insurance consumption. Based on the
expected utility paradigm, the insurance demand theory suggest that such factors as
individual’s income and wealth, the price of insurance, the individual’s degree of risk
aversion, and the probability of loss decide the individual’s insurance purchase.9 Many
theories suggest that market imperfections motivate corporate to take out policy for loss
protection (e.g. see Smith, 1986; Grace and Rebello, 1993; MacMinn, 1987; Skogh, 1989). In
contrast to life insurance, the empirical studies on property-liability insurance consumption is
less extensive (see Beenstock et al., 1988; Zietz, 2003). Many prior researches focus their
interests on the corporate demand for insurance (Core, 1997; Davidson III et al., 1992; Mayers
and Smith, 1990; Yamori, 1999; Zou, 2003; Zou et al., 2003). Beenstock et al. (1988) proves
the influence of income on the property-liability expenditure. Browne et al. (2000), relying on
the data for OECD countries, verifies that income, wealth, the percent of a country’s insurance
market controlled by foreign firms, and the form of legal system in the country all have

and Zilcha (1985; 1986), Lewis (1989), Moffet (1979), Pissariades (1980), Yaari (1965).
  See for instance, Beck and Webb (2003), Beenstock et al. (1986), Browne and Kim (1993), DePamphillis
(1977), Diacon (1980), Hwang and Gao (2003), Hwang and Greenford (2005), Lewis (1989), Lim and Haberman
(2004), Outreville (1996), Schwebler (1984), Truett and Truett (1990), Zhuo (1999).
  See for instance, Cleenton and Zellner (1993), Mossin (1968), Schlesinger and Graf (1987), Smith and Buser

governing effect on the property-liability insurance consumption. As our interest lies in the
determinants for China’s non-life insurance consumption as a whole, we follow Browne et al.
(2000) that the governing factors for corporate demand for insurance are the same as the

   In this study, we use the Yearbook of China’s Insurance (various issues), China Urban
Statistical Yearbook (various issues), China Labor and Social Security Yearbook (various
issues) and China Statistical Yearbook for Regional Economy (various issues) as the sources
of our data. The data cover 225 cities in total, either at the provincial level or at the prefecture
level. Our objective is to study the determinants of the development of insurance industry in
China, as well as their regional difference.
   The choice of the dependant variable requires some indicators to measure the development
of insurance industry. Although the number of insurance policies, or the premium volume, or
the sums insured is the simplest approach to measure insurance demand, in the literature, three
widely adopted indicators related to macroeconomic or socio-economic variables are (i)
premium volume, (ii) insurance expenditure per capita (insurance density) and (iii) insurance
penetration (insurance premiums divided by gross domestic product).
   Premium volume represents total insurance premiums written in a given region/country and
is a major indicator of the importance of the insurance industry in the economy of that
region/country, which is popularly adopted as the dependent variable in prior empirical
studies.10 Insurance density is calculated by dividing direct gross premiums by the population
and represents average insurance spending per capita in a given region/country, which is often
used as the dependent variable in the previous studies.11 Insurance penetration is the ratio of
direct gross premiums to GDP, indicating the relative importance of the insurance business in
the given economy, which is less frequently used in the prior research.12 This article at the
same time employs the penetration together with aggregate premiums and per capita
expenditure in insurance as the dependent variable.
   For the independent variables, we introduce the following indicators, which are supposed
to determine the development of insurance industry. The independent variables common for
both life and property are composed of total population, per capita GDP, wage level, private
savings deposit, FDI, education attainment, telephone ownership, Herfindahl Index. In
addition, such variables as investment in fixed assets and dwelling space per capita are special
for non-life insurance, while social security expenditure, life expectancy, young and old
dependency ratio special for life line.
    (i) Total population. Mantis and Farmer (1968) use population as an independent variable
to explain the life insurance demand in U.S for the period of 1929-1964. Schlag (2003) also
tells that the size of population determines the operating background, that is to say, the size of
market, for the life insurance industry in the long term. Besides, size of population is often an
element for the potential of China’s insurance market, life insurance in particular (e.g., see
Lai, 2002; D’arcy and Xia, 2003). We, therefore, include the total population for each city into

   For instance, Babbel (1985), Beenstock et al. (1986), Browne and Kim (1993), DePamphillis (1977), Diacon
(1980), Lim and Haberman (2004), Mantis and Farmer (1968), Schwebler (1984), Ward and Zurbruegg (2002),
Zhuo (1999).
   For example, Beck and Webb (2003), Browne and Kim (1993), Hwang and Gao (2003), Hwang and Greenford
(2005), Outreville (1996), Truett and Truett (1990), Zhuo (1999).
   See for instance Beck and Webb (2003).

our regressions and assume that its effect on the life and property insurance demand is
   (ii) Per capita GDP. We adopt the per capita GDP to represent the level of regional
economic development and expect it has positive effect on life insurance consumption, as
quite a number of researches proved. From the perspective of property insurance, previous
studies consistently verify GDP among the most important determining factors for the
insurance consumption (Beenstock et al., 1988; Browne et al., 2000; Outreville, 1990a;
1990b). This article also hypothesizes it positive to China’s property insurance consumption.
   (iii) Average wage of staff and workers. The income variable is previously represented by
various proxy such as normalized disposable personal income - disposable personal income
divided by total household net worth (Cargil and Troxel, 1979), the real disposable income
(Babbel, 1985), national income - GNP minus depreciation and indirect business tax (Browne
and Kim, 1993), GDP (Outveville, 1996), per capita GDP (Hwang and Gao, 2003; Lim and
Haberman, 2004; Hwang and Greenford, 2005). As for property-liability insurance
consumption, Beenstock et al. (1988) and Browne et al. (2000) proves the significant and
positive effect of income, while the latter of which takes GNP per capita as the proxy for
income. In our study, the wage level is used to measure the level of income, and per capita
GDP work as an indicator for macroeconomic development for a given city. And we suppose
that the wage level is positively related to both the life and the property insurance demand.
   (iv) Savings. Researchers have no common opinion about the relationship between savings
and demand for life insurance. Rose and Mehr (1980) believe that savings will reduce the life
insurance coverage because of it being a protection deductible for surviving dependents.
However, Headen and Finley (1974) argues that, on the perspective of increasing household
assets resulting from savings, savings may play a pushing role in life insurance demand. Some
other scholars also suggest that the level of savings tells the private household’s propensity to
save, and therefore also the background condition for the life insurance industry (Schlag,
2003). Schwebler (1984) and Beck and Webb (2003) prove the significant and positive
influence of savings on life insurance demand. We introduce both total savings deposit and
per capita savings deposit, as measures of wealth, into our analysis. Considering China being
a nation with high savings propensity, the positive effect on insurance consumption is
therefore expected.
   (v) Foreign direct investment (FDI). A number of studies have investigated issues related to
the determinants of FDI in various industries, but only a few have included insurance-specific
content. Ma and Pope (2003) use per capita FDI to examine the premium written per capita by
international insurer in host markets by supposing that the inflow of FDI favors the level of
international insurer participation in a given host market, depending on the client-following
suggestions by Price Waterhouse Coopers (1991) and Skipper (1998). FDI can also be used to
measure the openness degree of a city. The open-door policy and the export-oriented sectors
greatly contribute to the modernization and marketization of local economy, and hence would
have effect on the demand for insurance, as stated in Section 1. In particular, D’Arcy and Xia
(2003) partly contribute the continuous growth of property-liability insurance to more foreign
investment in the eastern coastal area. We, in our regressions, introduce respectively FDI
amount and ratio of FDI to GDP to examine whether the foreign investment has significant
influence on the demand for insurance. Taking the unequal regional distribution of FDI in
China into consideration, FDI is hypothesized positive to both non-life and life consumption,
especially in the coastal region.
   (vi) Fixed assets investment to GDP ratio. The variable has a direct impact on production

power, urban infrastructure and, in particular, it has an impact on property. We introduce this
variable to measure the potential of the local economy and examine whether it influences the
demand for non-life insurance. Positive result is expected in this article.
   (vii) Education. The individual’s high education attainment determines his or her level of
risk aversion (Schlag, 2003), leading to higher probability to buy life insurance as Karni and
Zilcha (1985; 1986) suggests. Pratt (1964), Arrow (1971), and Szpiro (1985) show that in
theory the more risk averse an individual is, the higher the amount insured. Some studies also
find the positive effect of education on China’s life insurance consumption in China (Hwang
and Gao, 2003; Hwang and Greenford, 2005). We adopt the variable, measured by university
enrollment per million population, to investigate its influence in our sample. The positive
result is assumed.
   (viii) Per capita number of telephones. This variable can serve as a proxy of the living
standard for a given region; it can also reveals the conditions of the local communication
channels as well as the power of information transmission in the city in question.
Communication convenience has so far never been included in the analysis of the demand for
insurance. We, here, incorporate it into our regressions to represent the sales condition in the
current China. The saying that insurance is sold not to buy is very popular in insurance
industry. It is the very case in China because of the low awareness of insurance even among
well-educated population. To contact the potential clients easily with low cost is crucial for
salespersons to sell policy. Number of telephones per capita therefore can indicate the level of
sales condition, to a great extent. The positive effect is therefore expected.
   (ix) Per capita dwelling space. Arrow (1965) suggests that the insurance demand increases
with wealth when individuals are characterized by increasing relative risk aversion. In
contrast, Mossin (1968) postulates conditions under which the optimal level of insurance
coverage decreases with increases in wealth. The conflicting theoretical suggestions of Arrow
(1971) and Mossin (1968) on the influence of wealth on insurance demand make it difficult to
judge the variable’s empirical results. This study employ per capita dwelling space as a
measure of private wealth and suppose its effect is positive, although Browne et al. (2000)
finds the negative influence of wealth on the consumption of property-liability insurance.
    (x) Herfindahl Index (HI).13 The relationship between competitive situation and insurance
ownership is inconclusive and has not been widely proved in empirical research. Following
the National Association of Insurance Commissioners’ (NAIC) criteria for defining a
competitive marketplace, Ma and Pope (2003) consider countries with HI measurements
greater than 1 800 to be noncompetitive and assigned a value of one for the dummy variable.
They conclude that comparably competitive markets are more likely to include the
participation of international insurers. Outreville (1996) takes a dummy variable indicating
either whether the market is a monopolistic one or whether foreign companies are writing
business in the market into his regressions, with data collected from 48 developing countries,
concluding that “monopolistic markets are significantly less developed than competitive
markets”. Simultaneously, he suggests that the appropriate variable to test the influence of
foreign insurers’ involvement would be a concentration ratio instead of the dummy variable.
Kwon (2002) use Herfindal Index as one of the means to measure market structure. In case of
China, measured in written premiums’ share, the insurance market is typical oligopoly in most
cities, along with the very low foreign share. The dummy variable is thereby of no use in

   Summing the squared market shares of each insurer in the market and multiplying the result by 10 000
determine this index. In a monopolistic market, the HI would be 10 000, and the index would become smaller as
the market becomes more competitive - suggesting either more companies competing in the market and/or market
share becoming more evenly distributed among the competitors (Ma and Pope, 2003).

analyzing the effect of market structure within China. For these reasons, following Bajtelsmit
and Bouzoutia (1998) and Kwon (2002), we calculate for each city the HI of non-life
insurance and that of life insurance, and, use directly the value of HI to measure the degree of
market concentration, expecting a negative result. Moreover, as the observation in our study is
“city” but not “country”, the value of HI is much higher than other studies because of lower
number of insurers in most of cities in China. .
    (xi) Social security expenditure. Life insurance and government-provided social security
systems interact each other. Some early theoretical research suggests that the increase of social
security will result in the decrease of private life insurance (see Yaari, 1965; Lewis, 1989;
Berheim, 1991). However, Fitzgerald (1987) says that social security has little net effect on
life insurance demand empirically.14 The contrasting suggestions encourage us to incorporate
the variable, indicated by social security expenditure ratio to GDP, into our regressions to
examine its effect in the case of China. We still expect it is negatively associated with private
life insurance ownership, even though the previous study on China does not prove it
significant (Zhuo, 1999; Hwang and Greenford, 2005).
    (xii) Life expectancy. Societies with high expectancy tend to have lower demand for pure
life insurance policies on the one hand; on the other hand, life expectancy may have positive
relationship with products bearing high savings element. Beenstock et al. (1986) and
Outreville (1996) in their empirical analysis corroborate the positive influence of life
expectancy. However, this variable is never examined in case of China. We expect the positive
effect also exists in our regressions.
   (xiii) Dependency ratio of children. The influence of young dependency on life insurance
consumption is inconclusive. Some research Schlag (2003) states that lower proportion of
young people in the working population tends to curtail demand for pure life insurance
policies via a reduction in the present value of consumption by the beneficiaries. On the other
hand, the high ratio of the young to the working population makes it reasonable a negative
effect on the demand for life policy with high savings element (see for instance, Beck and
Webb, 2003; Schlag, 2003). We, following Zhuo (1999), Beenstock et al. (1986) and Browne
and Kim (1993), expect the variable positive in our model.
   (xiv) Dependency ratio of aged. The higher ratio of aged to working population can
increase the demand for insurance with savings components and annuities, while can decrease
the demand for mortality risk coverage. Beck and Webb (2003) proves this variable
significantly related to the life insurance demand. Taking the reduced pension benefits from
the Social Security into consideration, we still expect the aged dependency ratio has positive
influence on China’s life insurance consumption.
   It’s obvious that the “effect of feedback” exists between the growth of premium income
and its explanatory variables. For example, the increase of income level would lead to the rise
of insurance consumption; but on the other hand, the development of insurance industry may
exert effect on the development of other sectors. It’s difficult to identify this causality in the
regressions. In order to avoid the problem of endogenity, we follow the method of lagged
variables. This method supposes that premium income responds to regional development with
a time difference. In other words, any improvement in socio-economic conditions at the
present time will encourage the growth of the premium income in the future. For the above

   Fitzgerald (1987) states that social security survivor benefits are found to decrease the demand for life
insurance on an earner, while social security benefits conditional on the earner’s survival increase the demand.
And empirically the two effects largely offset one another and so that social security has little net effect on life
insurance demand.

reasons, in our estimation, the dependant variables take their values at the year of 2003, and
all the explanatory variables take their values at 1998.15 More precisely, we assume that,
during a period of five years, the growth of premium income is determined by certain socio-
economic characteristics at the beginning of that period.
    All premium data including total written premiums, per capita life insurance expenditure,
life premium ratio to GDP, are drawn from Yearbook of China’s Insurance 2004, compiled by
CIRC. The independent variables such as population, GDP, wage, savings, FDI, fixed assets
investment, university student enrollment, number of telephones, per capita dwelling space,
come from the China Urban Statistical Yearbook. The above variables are based on city data.
The four other variables are based on provincial data, including social insurance expenditure
from China Labor and Social Security Yearbook, and life expectancy, dependency ratio of
children and dependency ratio of aged from China Statistical Yearbook for Regional
   Table 5 reports the summary statistics for the main variables concerned. On average,
Chinese consumers expend significantly more money in life than property insurance. The life
sector, bearing more providers, is comparatively more competitive, although the market as a
whole belongs to oligopoly with Herfindahl Index for both lines being more than 7000.

                                          Table 5 - Descriptive statistics
                                                                                   City size
                                                                        2.0                     Below
                                 All the Coastal Central Western millions 1.0 – 2.0 0.5 – 1.0     0.5
                                 cities provinces provinces province or above millions millions millions
Premium income
(million yuan)
  Total                          1604       2502        861      1064       7152      1,752       905         620
  Property insurance              464        753        161       401       2703        434       181         149
  Life insurance                 1140       1750        700       663       4450      1,317       723         471
Insurance density (yuan)
  Total                           354        488        245       271        767        401       277         279
  Property insurance               85        121         46        81        174        106        58          73
  Life insurance                  269        368        199       189        593        296       219         206
Insurance penetration (%)
  Total                          2.53       2.48       2.56       2.61      3.21       2.45       2.63        2.29
  Property insurance             0.57       0.58       0.48       0.73      0.77       0.58       0.54        0.55
  Life insurance                 1.97       1.89       2.11       1.86      2.44       1.91       2.08        1.73
Herfindahl index
  Property insurance             7903       7396      8269       8298       6020      7,708     7,991        8,553
  Life insurance                 7582       6836      8165       8085       5462      7,115     7,662        8,565
Number of insurance
  Property insurance              3.70       4.51       2.91      3.39       6.57       4.05      3.28        2.91
  Life insurance                  4.03       4.82       3.50      3.30       6.38       4.39      3.88        3.08

Number of cities               225          96         82       47        21         61       78          65
Note: The data for premium income, insurance density, insurance penetration, and the number of insurers, are for
the year of 2003. The Herfindahl index is for the year of 2000.

   The regional difference also describes in Table 5. The eastern coastal cities are absolutely
the leader in the China’s insurance development whether measured by premium income, per

     Due to the data limitation, the HI is calculated from the data of Yearbook of China’s Insurance 2001.

capita expenditure, or insurance penetration. The western and the central cities rank the
second in property and life insurance respectively. In addition, there are much more providers
writing policies in the coastal line than the rest of the country, indicating that competition is
more severe in coastal areas than in inland areas (see Sun, 2003). If we divide the cities
according to the size of population, we will find that the insurance consumption gets the
highest in cities where there is population of 2.0 million or above, and then decreases
gradually with the reducing size of city. For instance, citizens living in cites of more than 2
million population spend 593 yuans in life insurance policy, over 2 times of that for cities with
population of 1.0-2.0 million. Meanwhile, insurance companies prefer larger cities to sell
policy. For example, there are, on average, more than six insurers writing non-life policy in
the largest cities, compared with less than three providers in the smallest cities, as Table 5

   Insurance is, both in theory and in practice, generally classified into life and property-
liability. The estimations are correspondingly run for life and for property respectively. Table 6
and Table 7 present the results of the regressions. The dependant variables are premium
income, insurance density and insurance penetration. For each measure, the regressions are
run separately for the whole sample and for the three sub-samples of the coastal provinces, the
central provinces and the western provinces, which help us examine the effect of the
explanatory variables on the insurance industry in different areas.
   For life insurance, we find that per capita GDP is the only variable which has significant
influence on all three dependent variables. Meanwhile, the governing factors for different
region also prove different. The results are discussed in details hereinafter.
    As reported in Table 6, the effect of population size on written premiums is positive and
statistically significant, in particular for central and western provinces. This result consists
with the expectations that the population count determines the operating background for the
life insurance industry in the long term (Schlag, 2003) and the empirical results by Mantis and
Farmer (1968). That indicates, with the expanding size of population, the capacity of life
insurance market in China is likely to increase in the future. The effect of total population on
per capita life insurance expenditure proves positive and significant for the whole sample and
the sub-sample for coastal cites. This finding supports insurer’s preference for populous
metropolitan centers with established service networks (see Wu and Strange, 2000). Actually
large cities in China have more insurers than small ones. However, the population size doesn’t
exert significant influence on insurance penetration.
   The coefficient of per capita GDP is by and large positive and significant in the regressions
with written premium and insurance density as dependant variable. This is also in harmony
with the results from previous studies on life insurance consumption in China (Hwang and
Gao, 2003; Hwang and Greenford, 2005; Zhuo, 1999). It proves the conclusion that life
insurance sector will significantly benefit from the dramatic increase of the national economy.
In particular, the effect of per capita GDP is significant for western province, suggesting that
the rise of per capita GDP is actually a fundamental factor for the development of life
insurance in less developed regions. In contrast, the effect of per capita GDP is negative and
statistically significant for insurance penetration in the whole sample and the sub-sample for
central provinces. A possible reason may lie in that insurance penetration is an indicator
influenced by both the insurance sector and the gross domestic products for a given region.

   Table 6 – Determinants of the
   development of life insurance
                                                                                                                          Dependant variable
                                                         Logarithm of total premium income                           Logarithm of insurance density                                Insurance penetration
                                                                Coastal       Central     Western                        Coastal       Central        Western                      Coastal      Central    Western
                                               All the cities provinces      provinces   province       All the cities provinces      provinces       province   All the cities   provinces    provinces   province
Logarithm of total population                   0.342***        -0.007       0.350***    0.566***        0.160***        0.182**        0.033          -0.168        0.049          -0.021       -0.169     -0.261
                                                  (3.53)        (-0.03)        (2.63)      (2.83)          (3.16)         (2.54)        (0.45)         (-1.21)      (0.62)         (-0.18)       (-1.32)    (-1.18)
Logarithm of per capita GDP                      0.257**         0.119         0.097      0.449*         0.301***         0.175         0.166          0.367*     -0.368**          -0.399     -0.761***     0.485
                                                  (2.46)         (0.46)        (0.78)      (1.90)          (3.05)         (1.03)        (1.46)          (1.76)      (-2.42)        (-1.45)       (-3.94)     (1.49)
Logarithm of average wage                          0.194         0.412         0.128      -0.058         0.558***       1.184***        0.130           0.620        0.142         0.943*        -0.005     -0.725
                                                  (1.13)         (0.90)        (0.71)     (-0.11)          (3.61)         (3.78)        (0.82)          (1.53)      (0.58)          (1.90)       (-0.02)    (-1.08)
Logarithm of total savings deposit              0.301***       0.604**        0.217**      0.181
                                                  (3.55)         (2.62)        (2.18)      (0.88)
Logarithm of per capita savings deposit                                                                   0.234**      0.425***        -0.018          -0.355       0.212          0.215        0.114      -0.696*
                                                                                                           (2.47)        (2.64)        (-0.15)         (-1.44)      (1.44)         (0.87)       (0.55)     (-1.70)
Logarithm of FDI                                  -0.009         0.033         0.056        -0.028
                                                  (-0.36)        (0.53)        (1.41)       (-0.72)
FDI to GDP ratio                                                                                          -0.139        0.944**        -1.118          -0.893     -1.843***        -0.362       -4.555     -3.886
                                                                                                          (-0.32)        (2.17)        (-0.61)         (-0.39)      (-2.72)        (-0.52)      (-1.46)    (-1.03)
University student enrollment per million
population                                       0.188***      0.224***         0.103        0.203         0.001         0.023          0.064           0.039      0.132**        0.234***       0.150       0.111
                                                   (4.58)        (3.45)        (1.63)        (1.66)        (0.03)        (0.53)        (1.13)           (0.40)       (2.28)         (3.30)      (1.55)       (0.69)
Per capita number of telephones                  0.910***        0.175        1.762**        0.392       0.890***       -0.593*       1.948**         4.817***       0.503          -0.727       0.947       3.376
                                                   (2.71)        (0.34)        (2.10)        (0.27)        (2.86)       (-1.69)        (2.56)           (3.84)       (1.03)        (-1.31)      (0.73)       (1.62)
Herfindahl index of life insurance (/10000)       -0.109         0.385         -0.084       -0.668        -0.354        -0.656*      -1.077***        -1.116**     -0.904**        -1.004*    -2.200***    -2.060**
                                                  (-0.42)        (0.75)        (-0.23)      (-1.05)       (-1.53)       (-1.85)        (-3.33)         (-2.17)      (-2.49)        (-1.76)      (-4.00)     (-2.43)
Expenditure for social insurance and welfare
to GDP ratio                                     -0.083**      -0.149*       -0.154**        0.122          0.029       -0.117**       -0.051         0.333***    0.157***        0.200**       -0.212*     0.342*
                                                  (-2.39)       (-1.86)        (-2.04)       (0.92)         (0.95)       (-2.19)       (-0.76)          (3.22)      (3.29)         (2.40)        (-1.89)     (2.00)
Life expectancy                                   -0.001        -0.148       0.166***        0.002          0.028        -0.031        0.090*         0.100**      0.077*          -0.031      0.344***    0.146**
                                                  (-0.02)       (-0.82)        (3.04)        (0.03)         (1.05)       (-0.26)        (1.87)          (2.12)      (1.94)        (-0.16)        (4.19)      (2.04)
Dependency ratio of children;                   -0.024***     -0.053***         0.014       -0.019        -0.014**     -0.049***      0.055***         -0.024      -0.002         -0.031*      0.111***     -0.025
                                                  (-3.40)       (-3.45)        (0.96)       (-0.77)        (-2.09)       (-4.55)        (4.31)         (-1.16)     (-0.16)        (-1.91)        (5.15)     (-0.73)
Dependency ratio of aged                           0.036         0.135         -0.038        0.115        -0.063**        0.051      -0.288***          0.090      -0.052           0.111     -0.426***      0.013
                                                  (1.34)         (1.44)        (-0.71)       (0.92)        (-2.46)        (0.80)       (-5.93)          (0.91)     (-1.29)         (1.08)        (-5.14)     (0.08)
Constant                                           2.057         9.090        -7.618*        1.471       -6.113***       -7.256        -2.278          -8.509      -2.963          -2.644     -12.987**      1.247
                                                  (0.83)         (0.76)        (-1.83)       (0.21)        (-2.72)       (-0.90)       (-0.61)         (-1.50)     (-0.87)        (-0.20)        (-2.05)     (0.14)

R                                                   0.790         0.765         0.788         0.845        0.694         0.804         0.676           0.855        0.302          0.570        0.514       0.528
Number of observations                               217            90            82            45          217            90           82               45          222             94           82          46
The t-students are in brackets. *** significance at 1%; ** significance at 5%; * significance at 10%.

                                                          Table 7 – Determinants of the development of property-liability insurance

                                                                                                                   Dependant variable
                                                      Logarithm of total premium income                       Logarithm of insurance density                                Insurance penetration
                                                             Coastal       Central     Western                    Coastal       Central        Western                      Coastal      Central    Western
                                            All the cities provinces      provinces   province   All the cities provinces      provinces       province   All the cities   provinces    provinces   province
Logarithm of total population                0.564***      0.566***       0.315***    0.574***    0.231***       0.323***        0.089          -0.020        0.017           0.040       -0.029     0.001
                                               (7.43)        (4.24)         (2.69)      (3.35)      (4.40)         (4.49)        (0.95)         (-0.16)      (0.73)           (1.35)     (-0.84)     (0.02)
Logarithm of per capita GDP                  0.400***       0.416**        0.287**     0.430**    0.392***        0.429**      0.454***          0.364       -0.069        -0.210***      -0.017     0.088
                                               (4.71)        (2.58)         (2.61)      (2.09)      (3.69)         (2.31)        (2.93)          (1.68)      (-1.45)         (-2.71)     (-0.29)     (0.68)
Logarithm of per capita wage                 0.499***      0.890***         0.138       0.537     0.734***       1.082***        0.105         0.976**     0.277***        0.512***       0.039      0.066
                                               (3.49)        (2.84)         (0.87)      (1.19)      (4.33)         (3.24)        (0.48)          (2.36)      (3.61)           (3.67)      (0.48)     (0.26)
Logarithm of total savings deposit            0.147**         0.198        0.200**      0.272
                                               (2.20)        (1.51)         (2.31)      (1.66)
Logarithm of per capita savings deposit                                                           0.267***        0.231         0.097           -0.102      0.088**        0.176***      0.057      -0.043
                                                                                                    (2.86)        (1.58)        (0.64)          (-0.40)      (2.06)          (2.88)      (1.00)     (-0.27)
Logarithm of FDI                              0.037**      -0.015        0.065*       0.001
                                               (2.06)      (-0.36)       (1.86)       (0.03)
FDI to GDP ratio                                                                                    0.039         0.504         1.451           -0.380       -0.247         -0.013        1.431     -1.358
                                                                                                    (0.09)        (1.09)        (0.54)          (-0.16)      (-1.23)        (-0.07)      (1.42)     (-0.88)
Fixed investment to GDP ratio                0.478***      0.638**      0.469**       0.061        0.420**        0.049         0.156          2.145***      0.174*          0.061       -0.006     0.634*
                                               (2.82)       (2.30)       (2.19)       (0.11)        (2.08)        (0.16)        (0.49)           (4.07)      (1.89)          (0.48)      (-0.05)     (1.92)
University student enrollment per million
population                                   0.084**      0.102**       0.149***       0.104       -0.063         -0.046        0.033           -0.008       -0.010         -0.013        0.017     -0.020
                                               (2.45)       (2.05)        (2.67)       (0.96)      (-1.57)        (-0.87)      (0.43)           (-0.08)      (-0.55)        (-0.61)      (0.61)     (-0.32)
Per capita number of telephones              1.142***     1.211***        0.554       -0.005       0.571*          0.164        1.275          2.855**       -0.039          0.026       -0.098      1.087
                                               (4.00)       (3.37)        (0.77)       (0.00)      (1.69)          (0.43)      (1.25)            (2.29)      (-0.25)         (0.16)      (-0.26)     (1.39)
Per capita dwelling space                      0.010        0.013         0.012       -0.023        0.012        0.024**      -0.057**           0.003       -0.001          0.003       -0.016     -0.002
                                               (1.27)       (1.19)        (0.60)      (-1.27)      (1.21)          (2.09)      (-2.19)           (0.19)      (-0.23)         (0.55)      (-1.63)    (-0.16)
Herfindahl index of non-life insurance
(/10000)                                       -0.100       0.660*       -0.216       -0.434        -0.232        0.038         -0.343          -0.566      -0.294**         -0.085      -0.257*    -0.371
                                               (-0.47)       (1.70)      (-0.81)      (-0.88)       (-0.91)       (0.09)        (-0.94)         (-1.17)      (-2.53)         (-0.47)      (-1.89)   (-1.22)
Constant                                     -3.582***    -8.124***       0.754       -4.616      -9.736***    -13.531***       -1.983         -7.668**    -1.900***       -3.821***       0.289    -0.274
                                               (-3.04)      (-3.22)      (0.46)       (-1.45)       (-6.96)      (-5.13)        (-0.86)         (-2.37)      (-2.98)         (-3.47)      (0.33)    (-0.13)

R2                                             0.853        0.869        0.828        0.863         0.682         0.771         0.480           0.858        0.238           0.450        0.190      0.519
Number of observations                          221           94          82            45           221            94           82               45          222              94           82         46
The t-students are in brackets. *** significance at 1%; ** significance at5%; * significance at10%.

    As our results show, the relationship between insurance premium and income level,
measured by average wage per employee, is very interesting. The coefficient is significantly
positive in the regression in which the insurance density acts as the dependent variable, while
not significant for total premiums. In addition, the independent variable shows its significant
and positive effect on the insurance density and penetration in the sub-sample for coastal
cities. This result proves the matter of fact that the increase of income will not serve as a
pushing factor for insurance purchase unless the income reaches a certain level, which is also
verified by Beenstock et al. (1988) in property-liability. That is to say, the income is not high
enough in the central and western cities that it cannot encourage consumers to buy life
coverage. On the other hand, the disposable income instead of wage level shall be a better
indicator to measure income.
   Our results show that total savings deposit and per capita savings are significantly and
positively associated with total premium and per capita consumption in life insurance,
respectively. Its effect is particularly significant for the eastern coastal region that achieves the
highest level of savings rate.16 This gives support to Headen and Finley (1974) and is also
consistent with the conclusions from prior studies (Beck and Webb, 2003; Eisenharuer and
Halek, 1999; Schwebler, 1984). Compared to other countries, the savings rate in China is
extremely high. By the end of 2004, the total savings account amounted to 10361.8 billon
yuan, equaling to 88.9% of GDP (National Bureau of Statistics of China, 2005). We therefore
can forecast that the big volume of savings will provide great potential for the growth of life
   We use two indicators to represent the level of foreign direct investment in our regressions,
one is logarithm of FDI, and the other is FDI to GDP ratio. FDI has, in statistics, not
significant influence on life insurance development, which is in contrast with our expectation.
A possible explanation is that the sector distribution of FDI in China. As statistics reports,
most of FDI has been invested in manufacturing industries in China. In 2003, for instance,
69% of FDI, USD 36 936 million, flowed into the manufacturing industry (Department of
Trade and External Economic Relations Statistics National Bureau of Statistics, 2005), whose
major risk lies in property and liability but not life. Also in the year of 2003, just 0.43% of
FDI went to finance and insurance, compared to 22.6% for the Association of South-East
Asian Nations (ASEAN) in 2001 (ASEAN, 2002). We can come to a conclusion that, on the
supply side, FDI has no direct effect on the development of life insurance.
   However, we have to notice that FDI ratio to GDP shows positive and significant effect on
per capita life insurance consumption in the east of China. This can be explained by FDI’s
regional distribution in China. According to the figure released by China National Statistics
Bureau, in 2003, the eastern regions along the coastal line in China captured the FDI USD 45
386.22 million, equaling more than 85.7% of the gross amount (National Bureau of Statistics
of China, 2005). Ten per cent of China’s labor force are employed by foreign-invested
enterprises (see Ji and Thomas, 2001), and their wage is usually much higher than their
domestic counterparts. On the other hand, the replacement ratio of their pension benefits from
the social insurance system remains the lowest among enterprises, being 55.2% in 2003
(National Bureau of Statistics of China, 2003). The contrast between far higher income and
the lower social security benefit encourage them to buy more life insurance to increase their
expected utility after retirement.
   University student enrollment per million population expresses to be positive and
significant in the two first regressions (the whole sample and the sub-samples of the coastal
  In 1998, per capita savings deposit reached 13396 yuans for the coastal provinces, compared to 8007 yuans for
the central provinces and 7095 yuans for the western provinces.

provinces), in which the total written premiums and insurance penetration are dependent
variables respectively. This not only proves the theory (see for instance, Karni and Zilcha,
1985; 1986) and our hypothesis that education attainment is positively related to life insurance
development, but also supports previous empirical findings (Beck and Webb, 2003; Browne
and Kim, 1993; Hwang and Gao, 2003; Hwang and Greenford, 2005; Truett and Truett, 1990).
The result indicates that, in China, individuals with higher educational attainment are more
likely to buy life insurance.
   Number of telephones owned by per capita affects the premiums in a significant and
positive way for the whole sample and the sub-sample for the central provinces. The
coefficient is positive and statistically significant for the expenditure per capita in the whole
sample and all three sub-samples. The result verifies our expectation that the improvement in
communication will encourage the sales of insurance policy in China. That is to say, insurance
is easy to sell in the regions with more telephones the sales condition is also a robust
explanatory variable for the development of insurance.
   The result of the effect of Herfindahl index is very interesting. HI has no significant
relationship with the premiums volume, even in the coastal provinces, where most of the
insurers writing policies in China have established entities. As Section 2 states that most of
premiums are concentrated on the top 3 firms, the market influence of new entrants is
comparatively low, owing to their low market share. Therefore, we believe that the change in
market structure resulting from the increasing number of entrants is not enough to push the
growth of the total amount of written premiums significantly.
   Contrary to the above results, the association of HI with per capita consumption and ratio
of premiums to GDP is significant and negative especially in the central and western part of
China, being consistent with our hypothesis that competitive market is a pushing factor for the
insurance ownership.
   The earlier theory suggests that the relationship between social security benefits and private
insurance demand is substitutive (see Berheim, 1991; Lewis, 1989; Yaari, 1965). Our results
verify that social security expenditure has significant and negative effect on private life
insurance premium income. This is in harmony with the previous studies (Beck and Webb,
2003; Beenstock et al., 1986; Browne and Kim, 1993; DePamphilis, 1977). The reform of the
social insurance system over recent years gives supports to our results. We here take pension
or old age insurance as an example. The replacement ratio of pension benefits provided by the
currently implemented system is about 60% of their working wages,17 compared to the
previous percentage of 80% or so (Leung, 2003; Li, 2000; Saunders and Shang, 2001; Song,
2001).18 The reduced replacement ratio requires employees to seek financial support from
other insurance schemes, which is actually regarded as one pillar to supplement to the social
security system. The other type of social insurance system such as health care is nearly the
   The old age insurance system has been implemented since the year of 1997. The new system replaced the
previous pay-as-you-go system with three-pillar system. Pillar 1: a mandatory, defined-benefit social pension
from social account funded by payroll tax. Pillar 2: a mandatory, defined-contribution, individual pension with
individual accounts funded by contributions from employee and employer. Pillar 3: a voluntary supplementary
pension (see State Council, 1997).
   The old social insurance system of China, only covering the employees in state-owned enterprises and
government employees in urban area, is a defined-benefit, pay-as-you-go type. The system provides benefits for
retirement, disability, health care, unemployment, housing, and all other risks employees may face (see Leung,
2003; Li, 2000).
   A new health social insurance system called three-tier system has been in force since 1999, which combines
social pooling with individual accounts. The system covers all sectors and is pooled above the county level. Tier

   Social security expenditure ratio to GDP does not have significant influence on the per
capita consumption in the whole sample, consistent with the suggestion of Fitzgerald (1987).
However, the independent variable is significantly and positively related to the insurance
density in the west of China. The local economy in the western part of China is still dominated
by state-owned enterprises, whose employees are well covered by the social insurance system.
What’s more, the employees in such enterprises as oil, gas, or steel companies often are paid
much higher than the other sectors. That is to say, individuals with higher benefits are also
ones with higher income, who are the main purchasers of insurance policy in the west of
   The influence of life expectancy appears different for different dependent variables. As
Table 4 shows, life expectancy only affect significantly and positively the premium volume in
the central cities in China. For cities in both the central and the western part of China, life
expectancy’s influence on per capita life insurance expenditure is positive and significant. In
the regressions with premiums ratio to GDP as the dependent variable, life expectancy appears
positive and significant in all samples. Our results are in consistent with the theoretical
suggestions and the previous empirical findings (Beck and Webb, 2003; Beenstock et al.,
1986; Outreville, 1996), indicating that life expectancy is actually a determining factor for life
insurance demand in China.
   Consistent with the positive effect of ratio of young dependents in previous studies,20
young dependency ratio is also related to life insurance expenditure per capita and premiums
ratio to GDP in the central area of China, proving the bequest motives in theory. In contrast,
the variable appears a significantly negative determinant to the life premium volume for the
whole sample and eastern sub-sample, which supports the Schlag’s expectation that a high
proportion of young people possibly have a negative effect on the demand for insurance
policies with a high savings element (Schlag, 2003). The results may come from the two
possible reasons. First, a large number of families, especially those in the coastal area, spend a
large part of household income into children’s education and daily consumption, reducing the
purchase power for insurance, to some degree. Second, as reported in Section 3, due to the
scarcity of data by city, we use the dependency ratio by province, which may lead to some
   The aged dependency ratio to working population affects per capita consumption in a
negative and significant way while its effect in east of China is positive but not significant.
The significant and negative effect is different from the positive and significant result of Beck
and Webb (2003). We owe the contradicting results to the following reasons. Normally with
the increase of aged population, more will buy insurance to protect their economic
requirement for retirement. However, the currently aged (who are retirees at present) often can
be given relatively higher pension from the social security system, reducing the necessity to
purchase private insurance. From the perspective of private insurers, the aged, the high aged
in particular, generally do not meet the underwriting requirements such as age and health, and
therefore cannot get coverage even if they would like to. Even so, we still believe that, with
the rising portion of aged population,21 the reducing family size, and the furthering reform of

1: individual health account funded by employee’ whole contribution and a small portion of employer’s
contribution. Tier 2: social health account funded by the rest of employer’s contribution. Medical expenses up to
4 times the local average wage will be covered by the social account. Expenses above the ceiling will have to be
financed by commercial insurance, which is the third tier (see State Council, 1998).
   See Beenstock et al. (1986), Beck and Webb (2003), Browne and Kim (1993); Diacon (1980), Truett and
Truett (1986), Zhuo (1999).
   The proportion of the aged population, 65 years old or over, will reach 9% by 2010 and further to 21% by 2050
(Zhang, 2000).

social insurance system, the premiums collected from products for the aged, particularly from
insurance bearing savings components and annuity, will greatly increase in the coming days.
   Now we shed lights on the property-liability line. Table 7 presents the results of the
regressions. Variables such as wage level, savings deposit, and fixed assets investment, are
found significantly related to the property insurance consumption. Concerning the regional
difference, the determining elements vary from region to region.
   The coefficient of the population volume is positive and significant for all the regressions
with the total premiums as dependant variable. The result verifies our expectations and
indicates that population size is a robust explanatory factor not only for life insurance but also
for property insurance in China. However, if we replace the premiums with insurance density,
the variable of total population shows positive and significant only for the whole sample and
the sub-sample for eastern cities. Furthermore, the variable is not statistically significant when
used to explain insurance penetration as Table 7 reports.
   Per capita GDP is, as expected, significantly and positively related to both premiums and
insurance density, supporting the results from previous studies (Beenstock et al., 1988;
Browne et al., 2000; Outreville, 1990a; 1990b). From the regional perspective, the variable is
positive and significant statistically for cities in the coastal and central provinces, whatever
the insurance consumption is measured by. However, the effect of per capita GDP on property
insurance density is not significant for the western provinces.
   In the regressions for the whole sample, average wage per employee appears is significant
and positive. The results not only corroborate the already published studies but also prove it a
pronounced element to explain the achieved development of insurance. Obviously, our results
verify that its effect on non-life more robust than that on life insurance. Meanwhile, we have
to give caution to the results in different regions. The coefficient is positive and significant in
eastern cities for all three indicators to measure the insurance development, whereas only
significant and positive for the insurance density in the west. The result is consistent with the
matter of fact we touched earlier that income is able to push the insurance consumption only
when it reaches to a certain level (see Beenstock et al., 1988).
   As to the savings deposit, if we incorporate total savings deposit into the regressions with
total premiums as dependant variable, the results show that the coefficient is significantly
positive for all-city sample and the central sample. The results appear different when we use
the savings deposit per capita as a substitute for the volume. For the all-city sample, insurance
density and insurance penetration are both significantly and positively influenced by the
average level of savings deposit. What has to be pointed out is that this variable is positive
and significant only for the measure of insurance penetration in eastern cities in China, while
insignificant in all other sub-samples. The results are consistent with Arrow’s theory that the
demand for insurance increases with wealth when individuals are characterized by increasing
relative risk aversion, while different from the previous negative result from Browne et al.
   The effect of the logarithm of FDI is positive and significant for the total premiums
collected from property insurance in the whole-sample regression, also slightly positive in the
central area. No significant results are found among the regressions with insurance density as
dependant variable in which we adopt the variable of FDI ratio to GDP. The result basically
supports the study by Ma and Pope (2003), in which the FDI is not statistically significant,
while contrasting with our expectations. Nevertheless, the result is explainable. According to
the product composition stated earlier, the business property and liability protection, which

actually are the main risks for enterprises to transfer by means of insurance, constitutes a very
small part of the aggregate premiums. In addition, Zou et al. (2003) finds that, in China, there
is no significant relationship between company demand for property insurance and ownership
of company.
   The ratio between investment in fixed assets and GDP is, in statistics, significant and
positive for all three measures of insurance development in the whole sample. This is in
harmony with our expectations that more fixed investment in a previous period forecasts the
increase of property premiums later. Concerning the eastern and the central cities, the increase
of fixed investment will lead to the increase of total premiums only. For cities located in the
west of China, the increased fixed investment will, as statistical results show, remarkably push
the rise of per capita expenditure in property coverage, while a slight rise of insurance
penetration. The fact that the importance of fixed investment in the overall economy varies
from region to region in China provides a possible explanation for the results. The pushing
effect of fixed investment on the growth of the macro-economy in the west is much larger
than in the coastal and central area, leading to the different effect in insurance industry. The
finding is also important to forecast the growth of insurance sector for a given city.
   The education level, measured by university student enrollment in this article, as a proxy
for risk aversion, is significant and positive only for total collected premiums in the whole
sample and sub-samples for eastern- and central-city separately. This result proves the
theoretical suggestions by Pratt (1964), Arrow (1971) and Szpiro (1985). At the same time, it
is different from the results in the study by Browne et al. (2000), where significant and
negative for general liability insurance while not significant for motor vehicle. Based on the
present literature, the risk aversion, represented by education level, is not a consistent factor to
be used for the explanation of the demand for property coverage.
   Sales condition is found to be significant and positive for the premium volume for all-city
sample together with the eastern sample. The effect of the variable is positive but not very
significant (significant at the level of 10%) for per capita consumption for all cites concerned
in our analysis, proving our hypotheses that convenient communication condition is good for
the development of insurance in a given region. The variable of number of telephones per
capita does show no significance for the property insurance development measured by
insurance penetration.
   Per capita dwelling space, used to proxy the wealth level, never show significant for all-
city samples. For regional samples, the exception appears in the eastern and central cities,
where the per capita expenditure in property insurance protection proves increase and
decrease significantly with the expansion of dwelling space per capita, respectively. The
contradicting results may result from that the per capita dwelling space is not an appropriate
indicator to proxy the wealth level in China, owing to the welfare housing allocation system.
However, we cannot easily say that this variable has significant effect on insurance demand.
One reason for this is that, the housing market in China is not mature enough to make the
exact measurement of real prices of houses and rent possible.
   The market structure, which is measured by Herfindahl Index in this article, just shows
negative significantly for the insurance penetration for all-city sample. The result is totally in
contrast with our expectations that the increasing competition deriving from the changing
market structure will enhance the development of insurance industry. However, on the other
hand, it is consistent with the results from the study by Browne et al. (2000). That indicates
that the changing market structure led by the increase of insurance providers does not
necessarily lead to the lowering of price of insurance in case of China for our studied period at

least. It also implies that the insurance market of China is generally noncompetitive as we
stated in Section II. As Ma and Pope (2003) verifies in their study that noncompetitive market
affect the development of insurance sector negatively and significantly, China’s insurance
consumption is, to some extent, depressed by the oligopolistic market. However, we have to
notice that, as Table 3 tells, the market structure has a positive and significant, though not very
robust, effect on the total collected premiums in the coastal provinces in China.

   China is in the midst of the transition from planned to market economy, from closed
economy to open economy. On the other hand, China is also an economy with big regional
difference. Determinants of China’s extremely fast development of insurance will definitely
mirror the characteristics of the national economy as a whole. This article, relying on data for
225 cities, examines the determinants of China’s insurance consumption and the regional
disparity in insurance sector.
   The current article finds some significant results. Foreign direct investment, as a measure
of openness degree for a given region, proves more significant to the development of
property-liability insurance than to that of life insurance. Besides, the variables of fixed assets
investment and number of telephones per capita, which are for the first time incorporated as
explanatory variables in empirical investigation, verify their significant influence on China’s
insurance consumption.
   Regarding life insurance, such variables as population count, per capita GDP, total savings
deposit, education attainment, telephone ownership per capita, social welfare expenditure and
young dependency ratio are found significantly related to the life premium volume written in a
given city. The insurance density is verified to be significantly affected by the following
factors as population size, per capita GDP, wage level per capita, private savings deposit per
capita, number of telephones per capita, young dependency ratio and aged dependency ratio.
Among the variables employed in our models, only per capita GDP, FDI, education level,
market structure, social welfare expenditure, and life expectancy express significant to the
insurance penetration. When the regional difference is concerned, per capita number of
telephones and Herfindahl Index are both significant to the development of insurance
measured by insurance density for all three regions. For the coastal, the central and the
western areas in China, the insurance penetration is significantly influenced by market
structure and social welfare expenditure. The effect of other explanatory variables varies from
region to region.
   Regarding property-liability insurance, premium income is significantly affected by the
variables including population count, per capita GDP, per capita wage, total private savings
deposit, FDI, fixed assets investment, education attainment, and per capita number of
telephones. The following variables composed of population size, per capita GDP, wage level,
per capita private savings deposit, fixed investment, and telephone number per capita, are
proved significant statistically to the per capita expenditure in property insurance. As for the
insurance penetration, it is just affected by wage level, FDI, fixed investment, Herfindahl
Index. In addition, total population and per capita GDP are found significant for the
explanation of premium income for all three regions.

   China’s insurance sector will keep growing for the near future mainly owing to her
economic development. The influence of foreign direct investment and the market structure is
likely to increase with the fulfillment of commitments for accession to WTO and the
furthering of market internationalization.

Arrow, K. J. (1971), Essays in the Theory of Risk-Bearing, Amsterdam, North Holland.
ASEAN (2002), Statistics of Foreign Direct Investment in ASEAN (2002) : Comprehensive Data Set,
   2002 Edition, http://www. aseansec. org/13100. htm.
Babell, D. F. (1985), “The Price Elasticity of Demand for Whole Life Insurance”, The Journal of
   Finance, Vol. 40, No. 1, 225-239.
Bajtelsmit, V. L. and R. Bouzoutia (1998), “Market Structure and Performance in Private Passenger
   Automobile Insurance”, The Journal of Risk and Insurance, Vol. 65, No. 3, 503-514.
Beck, T. and I. Webb (2003), “Economic, Demographic, and Institutional Determinants of Life
   Insurance Consumption across Countries”, World Bank Economic Review, Vol. 17, No. 1, 51-88
Beenstock, M., G. Dickinson and S. Khajuria (1986), “The Determinants of Life Premiums: An
   International Cross-Section Analysis 1970-1981”, Insurance Mathematics and Economics, Vol. 5,
   No. 4, 261-270.
Beenstock, M., Dickinson, G., S. Khajuria (1988), “The Relationship Between Property-Liability
   Insurance Premiums and Income: An International Analysis”, The Journal of Risk and Insurance,
   Vol. 55, No. 2, 259-272.
Bernheim, B. D. (1991), “How Strong are bequest motives? Evidence based on estimates of the
   demand for life insurance and annuities”, Journal of Political Economy, Vol. 99, No. 5, 899-927.
Binder, S., B. Tab and Y. Winston (2004), “Selling Life Insurance to China”, McKinney Quarterly,
   00475394, 2004 Special Edition, 83-87.
Browne, M. J. and K. Kim (1993), “An International Analysis of Life Insurance Demand”, Journal of
   Risk and Insurance, Vol. 60, 616-634.
Browne, M. J., J. Chung and E. W. Frees (2000), “International Property-Liability Insurance
   Consumption”, The Journal of Risk and Insurance, Vol. 67, No. 1, 73-90.
Campbell, R. A. (1980), “The Demand of Life Insurance: an Application of the Economics of
   Uncertainty”, Journal of Finance, Vol. 35, No. 5, 1155-1172.
Cargil, T. F. and T. E. Troxet (1979), “Modeling Life Insurance Savings: Some Methodological Issues,
   The Journal of Risk and Insurance, Vol. 46, No. 2, 391-410.
Chen, W. (2004), Report on China’s Life Insurance Development (in Chinese), China Financial and
   Economic Press.
Cleenton, D. L. and B. B. Zellner (1993), “Income, Risk Aversion, and The Demand for Insurance”,
   Southern Economic Journal, Vol. 60, No. 1, 146-156.
CIRC and Samsung Life Co. (2003), China-Korea’s Insurance Market and Regulation, Beijing, China
   Financial Publishing House.
Core, J. E. (1997), “On the Corporate Demand for Directors’ and Officers’ Insurance”, The Journal of
   Risk and Insurance, Vol. 64, No. 1, 63-87.
D’Arcy, S. P. and H. Xia (2003), “Insurance and China’s Entry into the WTO”, Risk Management and
   Insurance Review, Vol. 6, No. 1, 7-25.
Datamonitor (2004), Insurance in China, Insurance Industry Profile: China, November 2004,
   Reference Code 0099-2087.
Davidson III, W. N., M. L. Cross and J. H. Thornton (1992), “Corporate Demand for Insurance: Some
   Theoretical and Empirical Results”, Journal of Financial Services Research, Vol. 6, No. 1, 61-72.
DePamphlis, D. M. (1977), “Variation in Individual Life Insurance Premium Revenues: An
   Econometric Approach”, Journal of Risk and Insurance, Vol. 44, 67-76.
Department of Trade and External Economic Relations Statistics, National Bureau of Statistics
   (2005), China External Economic Statistical Yearbook (2004), Beijing, China Statistics Press.
Diacon, S. R. (1980), “The Demand for U. K. Ordinary Life Insurance: 1946-1968”, Geneva Papers

    on Risk and Insurance, June.
Eisenhauer, J. G. and M. Halek (1999), “Prudence, Risk Aversion, and The Demand for Life
    Insurance”, Applied Economics Letters, Vol. 6, No. 4, 239-242.
Fisher, S. (1973), “A Life Cycle Model of Life Insurance Purchases”, International Economic Review,
    Vol. 14, No. 1, 132-152.
Fitzgerald, J. (1987), “The Effects of Social Security on Life Insurance Demand by Married Couples”,
    Journal of Risk and Insurance, Vol. 54, No. 1, 86-99.
Grace, M. F. and M. J. Rebello (1993), “Financing the Demand for Corporate Insurance”, The Geneva
    Papers on Risk and Insurance (Theory), Vol. 18, No. 2, 147-172.
Hakansson, N. H. (1969), “Optimal Investment and Consumption Strategies Under Risk, and Under
    Uncertain Lifetime and Insurance”, International Economic Review, Vol. 10, No. 3, 443-466.
Headen, R. S. and L. J. Finley (1974), “Life Insurance Demand and Household Portfolio Behaviour”,
    Journal of Risk and Insurance, Vol. 41, No. 4, 685-698.
Hwang, T. and B. Greenford (2005), “A Cross-Section Analysis of the Determinants of Life Insurance
    Consumption in Mainland China, Hong Kong, and Taiwan”, Risk Management and Insurance
    Review, Vol. 8, No. 1, 103-125.
Hwang, T. and S. Gao (2003), “The Determinants of Demand for Life Insurance in an Emerging
    Economy- the Case of China”, Managerial Finance, Vol. 29, No. 5/6, 82-96.
Ji, C. and S. Thomas (2001), “The Role of Foreign Insurance Companies in China’s Emerging
    Insurance Industry: An FDI Case Study”, Paper presented at the Conference on Financial Sector
    Reform in China, September, China.
Karni, E. and I. Zilcha (1985), “Uncertain Lifetime, Risk Aversion and Life Insurance”, Scandinavian
    Actuarial Journal, 1985(1), 109-123.
Karni, E. and I. Zilcha (1986), “Risk Aversion in the Theory of Life Insurance: the Fisherian Model,
    Journal of Risk and Insurance”, Vol. 53, No. 4, 606-620.
Kwon, W. J. (2002), “Toward Free Trade in Services: Emerging Insurance Markets in Asia”, The
    Geneva Papers on Risk and Insurance, Vol. 27, No. 4, 638-668.
Lai, G. C. (2002), “The Future Direction of the Chinese Insurance Industry after Enters the WTO”,
    China & World Economy, No. 4, 50-58.
Lee, Y-K. (2002), “Insurance In China: Pre-and Post-WTO”, The China Business Review, Vol. 29, No.
    3, 16-21.
Leung J. C. B. (2003), “Social security reforms in China: issues and prospects”, International Journal
    of Social Welfare, Vol. 12, No. 2, 73–85.
Leung, M-K. and T. Young (2002), “China’s Entry to the WTO: Prospects and Managerial
    Implications for Foreign Life Insurance Companies”, Asia Pacific Business Review, Vol. 9, No. 2,
Leverty, T., Y. Lin and H. Zhou (2004), “Firm Performance in the Chinese Insurance Industry”,
    http://rmictr. gsu. edu/papers/wp04_10. pdf.
Lewis, F. D. (1989), “Dependents and the Demand for Life Insurance”, American Economic Review,
    Vol. 79, No. 3, 452-466.
Li, H. (2000), “Economic Efficiency and Social Insurance Reforms in China”, Contemporary
    Economic Policy, Vol. 18, No. 2, 194-204.
Lim, C. C. and S. Haberman (2004), “Modeling Life Insurance Demand from a Macroeconomic
    Perspective: The Malaysian Case”, Paper presented at The 8th International Congress on Insurance:
    Mathematics & Economics, Rome, 2004,
Ma, Y-L. and N. Pope (2003), “Determinants of International Insurers’ Participation in Foreign Non-
    Life Markets”, The Journal of Risk and Insurance, Vol. 70, No. 2, 235-248.
MacMinn, R. D. (1987), “Insurance and Corporate Risk Management”, Journal of Risk and
    Insurance, Vol. 54, No. 4, 658-677.
Mantis, G. and R. N. Farmer (1968), “Demand for Life Insurance”, Journal of Risk and Insurance,
    Vol. 35, No. 2, 247-256.
Mayer, D. and, C. W. Jr. Smith (1990), On the Corporate Demand for Insurance: Evidence from the
    Reinsurance Market, Journal of Business, Vol. 63, No. 1, 19-40.

Moffet, D. (1979), “An Analysis of the Demand for Life Insurance: The Consumer’s Problem”, The
   Journal of Risk and Insurance, Vol. 46, No. 2, 99-112.
Mossin, J. (1968), “Aspects of Rational Insurance Purchasing”, Journal of Political Economy, Vol. 76,
   No. 4, 553-568.
National Bureau of Statistics of China (2003), China Labour Statistical Yearbook (2003), Beijing,
   China Statistics Press.
National Bureau of Statistics of China (2005), China Statistical yearbook (2004), Beijing, China
   Statistics Press.
Outreville, F. J., (1990a), “The Economic Significance of Insurance Markets in Developing
   Countries”, The Journal of Risk and Insurance, Vol. 57, No. 3, 487-498.
Outreville, F. J., (1990b), “The Relationship between Insurance, Financial Development and Market
   Structure in Developing Countries: An International Cross-Section Study”, UNCTAD Review, 53-
Outreville, F. J., (1996), “Life Insurance Markets in Developing Countries”, Journal of Risk and
   Insurance, Vol. 63, No. 2, 263-278.
Pissariades, C. A. (1980), “The Wealth-Age Relation with Life Insurance”, Economica, Vol. 47, No.
   188, 451-457.
Pratt, J. W. (1964), “Risk Aversion in the Small and in the Large”, Econometrica, Vol. 32, 122-136.
Price Waterhouse Coopers (1991), “Trends in the Internationalization of the Insurance Industry”,
   August, 3-4.
Rose, T. and R. I. Mehr (1980), “Flexible Income Programming”, Journal of Risk and Insurance, Vol.
   47, No. 1, 44-60.
Saunders, P. and X. Shang (2001), “Social Security Reform in China's Transition to a Market
   Economy”, Social Policy and Administration, Vol. 35, No. 3, 274-289.
Schlag, C-H. (2003), “Determinants of demand for life insurance products - Theoretical concepts and
   empirical evidence”, Working Paper, http://www. sgvs. ch/pdf_basel_04/B2-SchlagC. pdf.
Schlesinger, H. and J. M. Graf (1987), “Risk Aversion of The Purchase of Risky Insurance”, Journal
   of Economics, Vol. 47, No. 3, 309-314.
Schwebler, R. (1984), “Identification and Quantification of the Overall Impact of the National
   Economy on Life Assurance in the Federal Republic of Germany”, The Geneva Papers on Risk
   and Insurance, No. 32, 280-298.
Shen, Y. (2000), “China’s Insurance market: Opportunity, Competition and Market Trends”, The
   Geneva Papers on Risk and Insurance, Vol. 25, No. 3, 335-355.
Skipper, H. D., Jr. (1998), “Insurance Operations and Management”, International Insurance
   Management, Chapter 24, Boston, Massachusetts, Irwin MacGraw-Hill.
Skogh, G. (1989), “The Transaction Cost Theory of Insurance: Contracting Impediments and Costs,
   The Journal of Risk and Insurance, Vol. 56, No . 4, 726-732.
Smith, C. W. Jr. (1986), “On the Convergence of Insurance and Finance Research”, The Journal of
   Risk and Insurance, Vol. 53, No. 4, 693-717.
Smith, M. L. and S. A. Buser (1987), “Risk Aversion, Insurance Costs and Optimal Property-Liability
   Coverage”, The Journal of Risk and Insurance, Vol. 54, No. 2, 225-245.
Song, X. (2001), Report of the Reform and Development of China’s Social Security System, Beijing,
   People’s University Press.
State Council (1997), The Act on Establishing a Unified Basic Pension System for Workers, Document
   No. 26, P. R. China.
State Council (1998), The Act on Establishing a Basic Health Insurance System for Urban Workers,
   Document No. 44, P. R. China.
Sun, Q. (2003), “The Impact of WTO Accession on China’s Insurance Industry”, Risk Management
   and Insurance Review, Vol. 6, No. 1, 27-35.
Swiss Reinsurance Company (2005), World insurance in 2004: growing premiums and stronger
   balance sheets, Sigma No. 2/2005.
Szpiro, G. G. (1985), “Optimal Insurance Coverage”, The Journal of Risk and Insurance, Vol. 52, No.
   4, 705-710.
Thomas, J. E. (2002), “The Roles and Powers of the Chinese Insurance Regulatory Commission in the

   Administration of Insurance Law In China”, The Geneva Papers on Risk and Insurance, Vol. 27,
   No. 3, 413-434.
Truett, D. B. and L. J. Truett (1990), “The Demand for Life Insurance in Mexico and the United
   States: A Comparative Study”, The Journal of Risk and Insurance, Vol. 57, No. 2, 321-328.
Ward, D. and R. Zurbruegg (2002), “Law, Politics and Life Insurance Consumption in Asia”, The
   Geneva Papers on Risk and Insurance, Vol. 27, No. 3, 395-412.
Wu, D. (2004), Report for the Development and Reform of China’s Insurance Industry: 1979-2003 (in
   Chinese), Beijing, China Economic Publishing House.
Wu, X. and R. Strange (2000), “The Location of Foreign Insurance Companies in China”,
   International Business Review, Vol. 9, No. 3, 383-398.
Yaari, M. E. (1965), “Uncertain Lifetime, Life Insurance, and the Theory of the Consumer”, Review
   of Economics Studies, Vol. 32, No. 2, 137-150.
Yamori, N. (1999), “An Empirical Investigation of the Japanese Corporate Demand for Insurance”,
   The Journal of Risk and Insurance, Vol. 66, No. 2, 239-252.
Zhang, W (2000), The Problems and Responses of Population Aging in China in the First Half of the
   21st Century, Beijing, Hauling Publishers.
Zhuo, Z. (1999), “Demand for Life Insurance: An Empirical Analysis on China”, Zeitschrift fur
   Versicherungwesen, No. 6-7.
Zietz, E. M. (2003), “An Examination of the Demand for Life Insurance, Risk Management and
   Insurance Review”, Vol. 6, No. 2, 159-191.
Zou, H. (2003), “Corporate Insurance Market in the People’s Republic of China: Some Macro and
   Micro Perspectives”, Managerial Finance, Vol. 29, No. 5-6, 65-81.
Zou, H., M. B. Adams and M. J. Buckle (2003), “Corporate Risks and Property Insurance: Evidence
   From the People’s Republic of China”, The Journal of Risk and Insurance, Vol. 70, No. 2, 289-