This paper investigates the relationship between insurance development and economic growth by employing GMM models on a dynamic panel data set of 77 economies for the period 1994–2005. Insurance density is used to measure the development of insurance. Controlled by a simple conditioning information set and a policy information set, we can draw a conclusion that insurance development is positively correlated with economic growth. The sample is then divided into developed and developing economies.

For the developing economies, the overall insurance development, life insurance and non-life insurance development play a much more important role than they do for the developed economies. The Geneva Papers (2010) 35, 183–199. doi:10. 1057/gpp. 2010. 4 Keywords: insurance development; economic growth; dynamic panel data Introduction There has been a great interest in the role of financial institutions in economic growth. Economists refer to some work of researchers in the late 19th and early 20th centuries who discussed the significance of finance for economic growth.

In recent times, a number of studies have analysed various issues with respect to the role of the banking sector in economic growth. The most prominent studies have been conducted by Levine and his colleagues. For instance, King and Levine1 demonstrated the connection between bank development and economic growth, which was confirmed by later studies such as Levine, Beck et al. , Levine et al. , Rousseau and Wachtel, and Beck and Levine. 2 The studies of the relationship between financial development and economic growth have been shown to be robust using different econometric methods.

For instance, Levine and Zervos3 used crosscountry regressions, whereas Levine4 used cross-country instrumental variables regressions. The recent studies by Beck et al. , Levine et al. , and Beck and Levine5 used dynamic panel GMM estimations, whereas Rousseau and Wachtel6 used panel Generalised Method of Moments (GMM) estimation for a Vector Autoregression (VAR) model. * Liyan Han and Yanhui Tian are especially grateful to National Natural Science Foundation of China (No. 70831001 and No. 70821061). 1 King and Levine (1993a, b).

2 Levine (1998, 1999), Beck et al. (2000), Levine et al. (2000), Rousseau and Wachtel (2000), and Beck and Levine (2004). 3 Levine and Zervos (1998). 4 Levine (1998, 1999, 2002). 5 Beck et al. (2000), Levine et al. (2000), and Beck and Levine (2004). 6 Rousseau and Wachtel (2000). The Geneva Papers on Risk and Insurance—Issues and Practice 184 Although comparing with studies on banks’ role in economic growth, the role of insurance is relatively less examined, there has been increasing literature on this issue recently.

Insurance is of great importance to a modern society by making many economic activities possible in addition to its contributions to the economies in terms of its size, employment, managed assets, and so on. 7 In fact, economic growth is characterised by the soundness of a national insurance market. 8,9,10 Outreville9 emphasised on the importance of property-liability insurance and life insurance, respectively, in developing economies and their growth. Skipper10 stated that insurance contributed to the economy from the following aspects: ‘‘ y (1). Promotes financial stability and reduces anxiety; (2).

Can substitute for government security programs; (3). Facilitates trade and commerce; (4). Mobilizes savings; (5). Enables risk to be managed more efficiently; (6). Encourages loss mitigation; (7). Fosters a more efficient capital allocation y ’’. 10 Sigma11, Enz12, and Ward and Zurbruegg13 described the relationship between insurance market development and economic development as an ‘‘S Curve’’, which stated the starting sharp and then smooth increase of insurance development corresponding to the lower and higher stages of economic development, respectively.

Ward and Zurbruegg14 argued the insurance contributions to economic growth from the following aspects: risk transfer and indemnification services and financial intermediary services. They further analysed the above two economic contributions in terms of the following factors: productivity improvement and innovation facilitation for the former services and production efficiency enhancement, investment opportunity increases, reduction in the waste of early monetary realisation, and insurance institutional monitoring benefits for the latter services.

Webb et al. 15 argued that life and property/liability insurers can contribute to economic growth from the following aspects: (1) Life insurance can increase productivity by reducing the demand for liquidity and by shifting from unproductive use to more productive use of resources. This is similar to the role of banks on investment quality documented by Pagano. 16 (2) Property/liability insurers provide an extra risk-financing choice, which potentially reduces the probability of firm financial distress and firm bankruptcy costs.

This influences investment decisions in a particular economy. (3) Insurers may potentially increase expected investment returns by reducing the costs of risk financing, because insurers can: ‘‘(a) Excel in offering risk-pooling services through the identification of standardised risks and simplification of contracts, (b) Provide optimal investments and asset-liability matching, (c) Provide valuable and cost-effective administrative services related to 7 8 9 10 11 12 13 14 15 16 Liedtke (2007).

UNCTAD (1964). Outreville (1990, 1996). Skipper (1997). Sigma (1999). Enz (2000). Ward and Zurbruegg (2002). Ward and Zurbruegg (2000). Webb et al. (2002). Pagano (1993). Liyan Han et al. Insurance Development and Economic Growth 185 risk management and claims payments, and (d) Offer products that are tax-deductible business expenses in many markets’’ (p. 6).

15 Regarding the theoretical relationship between insurance and economic growth, Webb et al. 16 has a detailed argument. According to Webb et al.,15 based on a Solow-Swan neoclassical growth model, assuming a Cobb-Douglas type of production model, which states that production growth is due to labour, capital, and technology, the following factors should be added in the augmented growth model: financial activities of property/liability insurers and life insurers, which with banks may measure the differences in productivity and investments based on institutional factors and savings rate.

As can be seen from the above analysis, it is expected in this paper that insurance activities should have a positive impact on economic growth.

However, this impact may vary across different countries and across different lines of insurance business. The empirical results of this paper, by employing GMM models on a dynamic panel data set of 77 economies for the period 1994–2005 and controlled by a simple conditioning information set and a policy information set, have shown that insurance development is positively correlated with economic growth. This paper reports the analysis of insurance development and economic growth by breaking them into life and non-life insurance as well as developed and developing economies.

It has been shown that for the developing economies, the overall insurance development, life insurance and non-life insurance development play a much more important role than they do for the developed economies. This finding has significant policy implications in that ‘‘it could give empirical ground to the micro-insurance policy strategy of the World Bank and the UN-ISDR and nicely complement the theory of the wealth effects of insurance’’. 17 The remaining parts of this paper are structured as follows: the following section discusses the data used in the empirical analysis and the econometric methodology.

The penultimate section discusses the empirical results. The final section concludes. Data in the insurance industry and empirical methodology This paper evaluates the long-run relationship between insurance development and economic growth. In doing so, it will differentiate between developed and developing economies and the role that life and non-life insurance development could play for economic growth. This paper uses a panel data set of 77 economies over the period 1994–2005. Table A1 in the Appendix lists the names of the economies used in this study.

Table 1 reports summary statistics for 77 economies used in this study by referring to the information on economic growth, insurance density, life and non-life insurance density. As can be seen from Table 1, the development of insurance is largely different in life and non-life insurance lines and across different economies. Table 2 reports global insurance income from 1998 to 2005. As can be seen from Table 2, in 2005, there were a total amount of US$3426 billion in worldwide insurance premiums, with life insurance US$1974 billion and non-life insurance US$1452 billion.

17 Thanks to the anonymous referee to point this out. The Geneva Papers on Risk and Insurance—Issues and Practice 186 Table 1 Summary statistics: 1994–2005 Descriptive statistics Economic growth Insurance density Total business Life business Non-life business Mean Median Maximum Minimum Std. dev. Skewness Kurtosis 3. 480 3. 700 31. 100 A22. 900 3. 459 A0. 825 13. 545 4. 982 4. 808 8. 534 0. 182 2. 001 A0. 096 2. 005 3. 747 3. 507 8. 313 A2. 302 2. 504 0. 001 1. 920 4. 428 4. 433 7. 660 A0. 511 1. 823 A0. 338 2. 340.

Jaque-Bera Probability 4348. 392 0.000 37. 427 0. 000 42. 103 0. 000 32. 468 0. 000 916 77 875 77 866 77 874 77 Observations Cross-sections Table 2 Global insurance income, 1998–2005 (billion) 2005 Life insurance Non-life insurance Total insurance 2004 2003 2002 2001 2000 1999 1998 1,974 1,452 3,426 1,849 1,395 3,244 1,673 1,268 2,941 1,536 1,091 2,627 1,439 969 2,408 1,521. 3 922. 4 2443. 7 1,412 912 2,324 1,264 891 2,155 Source: Sigma (various issues). From Figure 1, it can be seen that the growth patterns between 1994 and 2005 for life and non-life insurance differ from each other.

For instance, there was a large fluctuation for life insurance, while the fluctuation in non-life insurance is relatively small. As can also be seen from Figure 1, there have been apparently unbalanced growth patterns between life and non-life insurance over this period. For instance, in 2005, the growth rates for total, life and non-life premiums were 2. 5 per cent, 3. 9 per cent, 0. 6 per cent, respectively, after deducting inflation. 18 Indicators To measure insurance development, we use the insurance density, measured by annual premium payments divided by population and converted into U.S. currency.

As a way of assessing the independent connection between insurance development and economic growth, we control for other potential factors influencing economic growth in this paper. In order to control for convergence, in the simple conditioning information set, we include the initial real gross domestic product (GDP) per capita. As a way of 18 Sigma (2006). Liyan Han et al. Insurance Development and Economic Growth 187 10 8 6 4 2 0 -2 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 -4 Life Non-life Total Figure 1.

Global growth in premium income from 1994 to 2005. Source: Sigma (various issues). controlling for human capital accumulation, we use gross enrolment ratio of tertiary students. In the other model that deals with the policy conditioning information set, we use one of the following factors: inflation rate, trade balance, and gross fixed assets investment. All of the above factors are supposed to have an impact on the conventional growth model by influencing one or more of the growth factors, such as labour, capital, and technology.

Empirical methodology In order to empirically test the relationship between insurance development and economic growth and also to avoid the statistical problems associated with the use of ordinary least squares, the panel data technique is used in this paper. We use the same methodology used in Beck and Levine19 for banks, stock markets, and economic growth for the purpose of testing the role of insurance development in economic growth. In this paper, we follow Beck and Levine19 to employ the following regression equation, Eq.

(1): yi;t A yi;tA1 ? ?a A 1? yi;tA1 ? b0 Xi;t ? Zi ? ei;t ? 1? where y is GDP after taking into account inflation and by the logarithm transformation, X is the independent variables other than lagged y, Z, and e is unobservable country-specific effect and error term, respectively, i means country, and t means time period. We use two-step GMM estimators of dynamic panel data with fixed effects to estimate the model, which is promoted by Holtz-Eakin et al. ,20 and Arellano and Bond.

21 In the model, we also include period dummy variables to control for time-specific effects. 19 20 21 Beck and Levine (2004). Holtz-Eakin et al. (1988). Arellano and Bond (1991). The Geneva Papers on Risk and Insurance—Issues and Practice 188 Empirical results Regression without insurance density and endogeneity test of insurance density For testing the impact of insurance development on the growth of an economy, we conduct a series of regressions excluding the insurance density from the explanatory variables.

The statistics of those regressions are given in Table 3. If the regressors include all economic factors, which are gross enrolment ratio of tertiary students lagged 2, inflation rate, trade balance, and gross fixed assets investment, the sign of the coefficient of gross enrollment ratio of tertiary students lagged 2 is positive, which is reasonable. If regressing inflation rate and gross fixed assets investment separately with gross enrollment ratio of tertiary students lagged 2 one by one into the regressors, we can have reasonable regressions.

If we put trade balance and gross enrollment ratio of tertiary students lagged 2 together into regressors, the coefficient of trade balance is negative, which is not reasonable. However, the R2 of all those regressions are lower than the corresponding regressions including insurance density. It has demonstrated that insurance development really improves the economic growth. In fact, in the history of human economic society, investment and demand take the first position to promote the economic growth, and then for the long run come the education and technique innovation.

These factors do not exclude the positive impact of the development of insurance on economic growth, at least it improves the economic growth, in term of promoting the society stability and security. This paper argues that insurance density with other important variables really improve the economic growth, in another words, it has a positive impact on economic growth. For testing the endogeneity of insurance density, we implement the Hausman test. The result shows the insurance density is very significantly endogenous. Therefore in the next regression we use GMM method in the dynamic panel modeling. Insurance density and economic growth.

Using the econometric methods outlined above, this section presents regression results concerning the relationship between economic growth and insurance density. The regressions simply include the logarithm of initial real per capita GDP, the logarithm of gross enrolment ratio of tertiary students, the logarithm of gross fixed assets investment, the average inflation rate over the period, and trade balance of the economies. We present the panel estimator regressions in Table 4. The results in Table 4 show a statistically and economically significant relationship between the insurance development and economic growth.

The first column reports the result of the pure regression without the use of the variables forming the policy conditioning information set (trade balance, inflation rate, gross fixed assets investment). Insurance density is positively correlated with economic growth at the 5 per cent significance level in the columns 1–5, where the following potential econometric problems are absent: simultaneity bias, omitted variables, serial correlations, and over-fitting problems. Inflation rate and trade balance have negative signs and enter the regression significantly. Gross fixed assets investment is positively correlated with the economic growth.

In the last column, we include all the policy Liyan Han et al. Insurance Development and Economic Growth 189 Table 3 Economic growth excluding insurance density, two-step GMM estimator Regressors 1 Dependent variable: Real per capita GDP growth GDP (A1) A0. 011 (0. 750) Logarithm of initial income per capita A33. 912*** (0. 000) 2. 050*** Gross enrolment ratio of tertiary students lagged 2a (0. 000) Inflation rateb A2. 089** (0. 031) Trade balanceb 0. 046 (0. 195) Gross fixed assets investmentb 0. 303*** (0. 000) R2 Serial correlation test (P-value)c J-statisticd Countries Observations

0. 248 0. 000 25. 713 73 417 2 3 4 0. 019*** (0. 000) A36. 039*** (0. 000) 1. 034*** 0. 006 (0. 328) A35. 387*** (0. 000) 1. 121*** A0. 014 (0. 694) A33. 952*** (0. 000) *** 1. 916*** (0. 000) 0. 149* (0. 096) (0. 000) (0. 000) A0. 073*** (0. 000) 0. 271*** (0. 002) 0. 233 0. 000 50. 187 73 535 0. 245 0. 000 42. 186 73 535 0. 246 0. 000 26. 486 73 417 a In the regression, this variable is included as log (1+variable). In the regression, this variable is included as log (variable). The null hypothesis is that the errors in the regression do not have second-order serial correlation. d.

The null hypothesis is that the instruments do not have over-fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b c conditioning information set and conclude that economic growth is also positively influenced by insurance density. Regarding the magnitude of the measured effects, it has been shown in the last column that there is about 4. 781 per cent increase in economic growth given 1 per cent increase in total insurance density. For the purpose of comparison, the enhancement impact of banking activities on economic growth is no more than 1.

8 per cent given 1 per cent increase in bank credit (Table 4, Beck and Levine19). 22 Life insurance density and economic growth The results for the panel regressions in Table 5 show that life insurance density has a significantly positive impact on economic growth. The results do not reject the close connection between economic growth and life insurance density with the coefficient ranging from 1. 657 to 2. 640. Serial correlations in the error term 22 We use logarithm of insurance density and Beck and Levine (2004) use logarithm of banking credit. So it should be interpreted with precautions when comparing two different measures.

The Geneva Papers on Risk and Insurance—Issues and Practice 190 Table 4 Economic growth and insurance density, two-step GMM estimator Regressors 1 2 3 4 5 Dependent variable: Real per capita GDP growth GDP (A1) 0. 038 0. 059*** 0. 0265*** A0. 001 0. 023206 (0. 000) (0. 000) (0. 004) (0. 971) (0. 583) Logarithm of initial income per capita A42. 100*** A42. 93909*** A42. 635*** A42. 889*** A42. 794*** (0. 000) (0. 000) (0. 000) (0. 000) (0. 000) 1. 252* 0. 975*** 1. 340*** 0. 990*** 1. 023* Gross enrolment ratio of tertiary students lagged 2a (0. 000) (0. 000) (0. 000) (0. 0599) (0. 079)

Inflation rateb A0. 878** A1. 572 (0. 0140) (0. 135) Trade balanceb A0. 051*** 0. 092 (0. 000) (0. 261) Gross fixed assets investmentb 0. 317*** 0. 402*** (0. 000) (0. 000) 5. 414*** 5. 455*** 5. 329*** 4. 872*** 4. 781*** Premium densityb (Insurance density) (0. 000) (0. 000) (0. 000) (0. 000) (0. 000) R2 Serial correlation test (P-value)c J-statisticd Economies Observations 0. 260 0. 000 52. 982 73 517 0. 257 0. 000 52. 869 73 517 0. 267 0. 000 52. 125 73 517 0. 276 0. 000 32. 022 73 405 0. 300 0. 000 35. 560 73 405 a In the regression, this variable is included as log (1+variable).

In the regression, this variable is included as log (variable). c The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b and over-fitting problems are absent. Total fixed assets investment plays a significant role in economic growth, and inflation rate and trade balance do not pass the test in the last column.

Regarding the magnitude of the measured effects, it has been shown in the last column that there is about 1. 728 per cent increase in economic growth given 1 per cent increase in life insurance density, which is very close to the enhancement impact of banking activities on economic growth (no more than 1. 8 per cent given 1 per cent increase in bank credit documented in Table 4 by Beck and Levine19). Non-life insurance density and economic growth In Table 6, the panel results are more robust than life insurance as described in Table 5. The coefficient value ranges from 4. 180 to 4. 962.

Non-life insurance density is positively and significantly correlated with economic growth when using the same conditioning information sets and policy information sets. Inflation rate and trade Liyan Han et al. Insurance Development and Economic Growth 191 Table 5 Economic growth and life insurance density, two-step GMM estimator Regressors 1 2 3 4 5 Dependent variable: Real per capita GDP growth GDP (A1) 0. 004 0. 017*** A0. 002 A0. 100*** A0. 078** (0. 591) (0. 001) (0. 739) (0. 005) (0. 016) Logarithm of initial income per capita A40. 941*** A40. 435*** A41. 570*** A41. 144*** A42. 249*** (0. 000).

(0. 000) (0. 000) (0. 000) (0. 000) 0. 697*** 0. 460** 0. 901*** 0. 647 0. 657* Gross enrolment ratio of tertiary students lagged 2a (0. 004) (0. 041) (0. 000) (0. 186) (0. 080) Inflation rateb 0. 179 A1. 414 (0. 598) (0. 118) Trade balanceb A0. 070*** 0. 0496 (0. 000) (0. 340) Gross fixed assets investmentb 0. 439*** 0. 454*** (0. 000) (0. 000) 2. 640*** 2. 502*** 2. 478*** 1. 657*** 1. 728*** Premium densityb (Insurance density) (0. 000) (0. 000) (0. 000) (0. 000) (0. 000) R2 Serial correlation test (P-value)c J-statisticd Economies Observations 0. 255 0. 000 55. 620 73 511 0. 251 0. 000 58. 782.

73 511 0. 264 0. 000 54. 043 73 511 0. 315 0. 000 27. 194 73 405 0. 316 0. 000 26. 516 73 405 a In the regression, this variable is included as log (1+variable). In the regression, this variable is included as log (variable). c The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b balance have negative signs and enter the regression significantly.

The gross fixed assets investment is positively correlated with economic growth. Once again, serial correlations in the error term and over-fitting problems are absent. Regarding the magnitude of the measured effects, it has been shown in the last column that there is about 4. 180 per cent increase in economic growth given 1 per cent increase in non-life insurance density. For the purpose of comparison, the enhancement impact of banking activities on economic growth is no more than 1. 8 per cent given 1 per cent increase in bank credit (Table 4, Beck and Levine19).

In terms of the magnitude of the impact, it is apparent from the above analysis that non-life insurance has a much more significant impact on economic growth than life insurance. In addition, in Tables 4–6, initial income per capita has a negative effect on economic growth, which indicates that the higher the historical economic growth is, the slower the economy boosts. This supports the ‘‘S Curve’’ argument by Sigma11 and Enz. 12 Gross enrolment ratio of tertiary students lagged 2 periods has positively influenced economic growth in most equations. So, the education factor is quite important for a country’s long-term development.

The Geneva Papers on Risk and Insurance—Issues and Practice 192 Table 6 Economic growth and non-life insurance density, two-step GMM estimator Regressors 1 2 3 4 5 Dependent variable: Real per capita GDP growth GDP (A1) 0. 043*** 0. 065*** 0. 034*** 0. 032 0. 0548 (0. 000) (0. 000) (0. 000) (0. 340) (0. 105) Logarithm of initial income per capita A42. 491*** A42. 667*** A42. 690*** A41. 221*** A40. 028*** (0. 000) (0. 000) (0. 000) (0. 000) (0. 000) 1. 129* 0. 926*** 1. 238*** 1. 336*** 1. 083* Gross enrolment ratio of tertiary students lagged 2a (0. 000) (0. 001) (0. 000) (0. 009) 0. 061.

Inflation rateb A0. 933** A0. 381 (0. 016) (0. 768) Trade balanceb A0. 049*** 0. 0851** (0. 000) (0. 042) Gross fixed assets investmentb 0. 305*** 0. 366*** (0. 000) (0. 000) 4. 962*** 4. 938*** 4. 909*** 4. 197*** 4. 180*** Premium densityb (Insurance density) (0. 000) (0. 000) (0. 000) (0. 000) (0. 000) R2 Serial correlation test (P-value)c J-statisticd Economies Observations 0. 250 0. 000 55. 619 73 516 0. 246 0. 000 55. 628 73 516 0. 257 0. 000 55. 986 73 511 0. 263 0. 000 31. 505 73 405 0. 259 0. 000 28. 153 73 404 a In the regression, this variable is included as log (1+variable).

In the regression, this variable is included as log (variable). c The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b The comparison between developed economies and developing economies This section of the paper attempts to separately investigate the relationship between insurance development and economic growth for industrial and developing economies over the period 1994–2003.

Classification of economies is based on 2002 Gross National Income (GNI) per capita calculated using the World Bank Atlas method. The income groups are: low income, $735 or less; lower middle income, $736–$2,935; upper middle income, $2,936–$9,075; and high income, $9,076 or more. According to our empirical analysis, we divide 77 economies discussed in the paper into two kinds: developed economies and developing economies. The developed economies are selected from high-income economies publicised by the World Bank.

Low-income and middle-income economies are sometimes referred to as developing economies. We also put the upper middle-income economies into the developing economies group. Finally, we get 32 industrialised economies and 45 undeveloped economies. The means of these economies’ indicators can be seen in Tables A2 and A3 in the Appendix. According to Tables 7–9, the results in developing markets provide strong support for the insurance services: overall insurance density is strongly associated with Liyan Han et al. Insurance Development and Economic Growth 193.

Table 7 A comparison between developed economies and developing markets: Economic growth and insurance density Regressors Developing economies Developed economies 0. 109*** (0. 001) A49. 900*** (0. 000) 2. 435** A0. 157*** (0. 000) A19. 311*** (0. 000) 1. 656* Premium densityb (Insurance density) (0. 040) 9. 172*** (0. 000 ) (0. 065) 1. 873*** (0. 001) R2 Serial correlation test (P-value)c J-statisticd Economies Observations 0. 299 0. 000 26. 949 41 291 0. 231 0. 000 24. 307 32 216 Dependent variable: Real per capita GDP growth GDP (A1) Logarithm of initial income per capita.

Gross enrolment ratio of tertiary students lagged 2a a In the regression, this variable is included as log (1+variable). In the regression, this variable is included as log (variable). c The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over-fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b Table 8 A comparison between developed economies and developing markets: Economic growth and life insurance density.

Regressors Dependent variable: Real per capita GDP growth GDP (A1) Logarithm of initial income per capita Gross enrolment ratio of tertiary students lagged 2a Premium densityb (Insurance density) R2 Serial correlation test (P-value)c J-statisticd Economies Observations a Developing economies Developed economies 0. 150*** (0. 000) A36. 563*** (0. 000) 0. 023 (0. 980) 2. 495*** (0. 000 ) A0. 178*** (0. 000) A44. 525*** (0. 000) 0. 627 (0. 681) 0. 812 (0. 328) 0. 206 0. 000 28. 793 41 297 0. 420 0. 000 22. 175 32 214 In the regression, this variable is included as log (1+variable).

In the regression, this variable is included as log (variable). c The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over-fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b The Geneva Papers on Risk and Insurance—Issues and Practice 194 Table 9 A comparison between developed economies and developing markets: Economic growth and non-life insurance density Regressors Dependent variable: Real per capita GDP growth.

GDP (A1) Logarithm of initial income per capita Gross enrolment ratio of tertiary students lagged 2a Premium densityb (Insurance density) R2 Serial correlation test (P-value)c J-statisticd Economies Observations Developing economies Developed economies 0. 121*** (0. 000) A48. 039*** (0. 000) 2. 323* (0. 059) 8. 760*** (0. 000) A0. 162*** (0. 000) A19. 220*** (0. 000) 1. 653* (0. 072) 1. 309** (0. 028) 0. 247 0. 000 25. 030 41 289 0. 230 0. 000 23. 537 32 216 a In the regression, this variable is included as log (1+variable). In the regression, this variable is included as log (variable).

The null hypothesis is that the errors in the regression do not have second-order serial correlation. d The null hypothesis is that the instruments do not have over fitting problem. P-value in parentheses. ***,**,* indicate significance at 1%, 5%, and 10% level in the two-step GMM regression, respectively. b c economic growth. This close relationship holds after controlling for potential simultaneity bias and omitted variable bias. More specifically, as shown in Table 7, overall insurance density is closely associated with economic growth, with a coefficient of 9.

172 significant at 1 per cent level for developing economies, much larger than that for developed economies (a coefficient of 1. 873 at 5 per cent significance level). This means there is about 9. 172 per cent increase in economic growth given 1 per cent increase in overall insurance density for the developing economies, compared to 1. 873 per cent increase in economic growth given 1 per cent increase in non-life insurance density for the developed economies. It has been shown in Table 8 that life insurance only has a significant impact on economic growth for the developing economies, not for the developed economies.

There is about 2. 495 per cent increase in economic growth given 1 per cent increase in life insurance density for the developing economies. As shown in Table 9, non-life insurance density is closely associated with economic growth, with a coefficient of 8. 76 significant at 1 per cent level for the developing economies, much larger and significant than that for the developed economies (a coefficient of 1. 309 at 5 per cent significance level). This means that there is about 8. 76 per cent increase in economic growth given 1 per cent increase in non-life insurance density for the developing economies, compared to 1.

309 per cent increase in economic growth given 1 per cent increase in non-life insurance density for the developed economies. As can be seen from the above analysis, for developing economies, life insurance, non-life insurance, and total insurance play a much more import role than they do for the developed economies. Liyan Han et al. Insurance Development and Economic Growth 195 Conclusion In this paper, we have combined cross-sectional and time series data to examine the relationship between insurance development and economic growth in 27 economies over the period of 1994–2005.

We used GMM models on dynamic panel data to conclude that there is fairly strong evidence in favour of the hypothesis that insurance development contributes to economic growth. This relationship is more significant for non-life insurance than for life insurance. We then divide the economies into two groups and compare the different roles of insurance in the developed and developing economies.