A Survey of the Effects of Liberalization of Iran Non-Life Insurance Market by Using the Experiences of WTO Member Countries Marufi Aghdam Jalal 1, Eshgarf Reza 2 Abstract Today, globalization is prevalent and joining World Trade Organization (WTO) is one of the symbols of globalization and development and this organization has 185 permanent members and supervisors. Joining WTO had various effects on different sectors of economy of Iran including non-life insurance industry. The present study by panel data econometric method investigated the performance of non-life insurance indices of 10 countries that were member of WTO from the beginning. The countries were selected based on similar human development index in Iran at the establishment year of WTO (1995). The results showed that joining WTO had negative effect on non-life insurance indices of the selected countries. It is expected that by joining Iran to WTO, due to the inconformity of the insurance rules with WTO, we observe negative effects of membership in this organization on non-life insurance indices in Iran at least in short term. Also, inconformity with WTO rules is possible. Keywords: World Trade Organization. Insurance penetration, Insurance density (premium per capita), Degree of open economy 1. Introduction Today, globalization is prevalent phenomenon. It is a process of change underestimating the economic and political borders and develops the communication and culture interaction and it has considerable effects on social, economic, political, legal, and military and technology activities. 1 MA, Industrial engineering school, Iran science and Industry University; E-mail: Jalal.mahabad@gmail.com 2 MA, Industrial engineering school, Iran science and Industry University 33
One of symbols of globalization now is joining WTO, with 185 permanent members. WTO purpose is liberalization and competitiveness of goods and services among the countries as the economic enterprises and companies of member countries can compete in a fair and equal condition. Iran due to the necessity of development of non-oil export is obliged to join this organization as oil revenue providing major financial resources of the country, that is now (per capita and fixed price) is reduced rapidly (Behkish, 2001). Insurance role is important as some of the economic theorists including Ghulam Rasool, Pakistani author and won gold medal award in writing of articles of insurance journal in Pakistan in 1990 and he considered insurance as the legal child of any country (Sehat, 2005). Financial services in which there are insurance services are manifested in General Agreement on Tariff and Trade. This agreement obliged member states to eliminate the existing barriers to foreign collaboration and liberalization of insurance market. According to the definition of central insurance, all the insurance fields except life insurance are recognized as non-life insurance. This is including the following insurances: accident, health, fire, cargo, car (third party and extra, driver accident insurance and body insurance), ship, airplane, liability, engineering, oil and energy, money, credit and other types of insurances (Iran central insurance statistical report, 2010). As the rules are inclined to gradual liberalization of insurance industry, the probable outcomes of liberalization of non-life insurance in Iran as the major part of insurance industry market are investigated. 2. Study Purposes And Hypotheses The present study aimed to investigate the effect of membership of some of the developing countries in WTO as one of the indices of liberalization of insurance services trade on non-life insurance of these countries. Also, the probable outcomes of future membership of Iran in this organization are investigated from the view of international standard indicators of evaluation of insurance industry performance including insurance penetration index (receive premiums ratio to GDP) and premium per capita of non-life insurance. The present study hypothesis is that the membership in WTO had positive effect on performance of non-life insurances of selected countries from the view of global indices. 34
3. Methodology The method of the study was analytical-descriptive and it was conducted based on the models of panel data with Eviews6 software. The selection criterion of the countries was Human Development Index (HDI) that is calculated for people of each country by UN. It is obvious that the countries are selected that have similar HDI or close index with Iran. The period was 1995-1999 with 10 countries. Country Index Turkey 0.588057 Tunisia 0.585043 Philippine 0.586053 Indonesia 0.527244 Panama 0.579736 Egypt 0.539227 Iran 0.59575 Brazil 0.634297 Colombia 0.627274 Dominican Republic 0.60808 Malaysia 0.674227 Table 1: HDI in 1995 Based on time series data as a part of panel data, the present study at first tested the probability of the long-term relationship among the existing variables in the model. To do this, stationary and cointegration tests were applied. Based on the various methods of estimation of panel data, to obtain the best method, Haussmann test was used. Generally, the following model indicates a model with pooled data: 1 Where i=1, 2, n denote cross section groups (countries) and refers to time. Denotes dependent variable for its cross section unit in year t and is kith non-random independent variable for its cross section unit in year t. 35
Unit root test: stationary tests are the most important tests to estimate a regression with reliable coefficients. To present spurious regression, stationary tests are applied. To determine stationary of panel data, various tests are used. IPS test is formed based on the mean of Augmented Dickey- Fuller test among cross section samples. The present study applied Im Pesaran Shin tests to determine stationary of the variables (Baltagi, 2005). Co-integration test: If based on unit root test, it was found that the variables are non- stationary, co-integration test is performed. Only the results are reliable when the variables are co-integrated (Granger, 1986). To determine the co-integration, there are various tests as Kao, Pedroni's and Fisher tests. The preset study applied Pedroni's test (Eviews 7 User s Guide II). Error Correction Model (ECM): ECM model applies the combination of the first rank difference and lag value for co-integrated variables. 2 The above model is error correction model or equilibrium correction model. The term is called error correction term on condition that and are co-integrated with y coefficient. Thus, has I (0), and using OLS method is good and statistical inferences are not biased. The above model showed the short-term relation between Y, X changes. Thus, is the coefficient relating the Y changes during t period with X changes during period t. Some of the Y changes are due to the nonequilibrium correction that were in the previous period. Is adjustment speed in return to equilibrium and it shows which percent of equilibrium error of the previous period is corrected in current period (Ashrafzade and Mehregan, 2010). Haussmann test: The statistic of this test is calculated to determine the fixed or random differences of cross section groups as following that is having chi-square distribution with degree of freedom equal to the number of independent variables (K). 3 4 Null hypothesis of Haussmann test is the equality of estimator of both methods of generalized least squares and dummy variable. Thus: 36
If the calculated test statistics is greater than of table, H0 is rejected. Thus, equality of estimations of this method is rejected and it is recommended that random method is used to receive in cross section groups [8]. Fixed effects: In fixed effects, intercept of groups is varied. The following regression with assuming the independence of all from is considered. 5 By entering dummy variables for each group in the above equation, the following equation is obtained: 6 In the above equation, if, we have and in other cases. This equation is estimated by OLS and it is called Least Square Dummy Variable (LSDV). In this method, intercept is only different across the sections and time has no effect on them. Fixed effects are investigated in the conditions of intercept change with time or the change of intercept with the changes of groups and time changes (Shujie, 2007). Random effects: The fixed effects model is logical when we are sure that the difference across the sections is shown as regression transition function. We are not sure about this issue. Thus, other methods are used. Another method of estimation is random effects method assuming that fixed component determining the various sections as randomly across the groups and distributed regions. 7 Where it has K regress or plus an intercept. denotes the random part of its group and it is fixed over time. It should be say that in this state the variances of various sections are not equal and there is variance heteroscedasticity and GLS method is applied. By introducing these two methods, the question is which methods are used? For decision making, Haussmann test is used (Zaranejad and Anvari, 2005). - Explication of empirical model to investigate the effects of joining WTO on non-life insurance penetration index of the selected countries (Model 1) 37
8 IPI is non-life insurance penetration index, perm is insurance density or non-life premium per capita, APT is the ratio of foreign trade volume to GDP (open economy), V is model disturbance, and Ln is natural logarithm and Dum is dummy variable that is zero for the years before membership and it is one for the years after membership. - Explication of empirical model to investigate the effects of joining WTO on premium per capita of non-life insurance of the selected countries (Model 2) 9 is premium per capita of non-life insurance, GDP is Gross Domestic Product per capita, V is model disturbance and Dum is dummy variable that is zero for the years before membership and it is one for the years after membership. 4. Results Level 1st difference Statistic Prob Statistic Prob L IPI -0.72332 0.2347-2.20066 0.0139 L PREMP -1.99756 0.0229 L APT 0.23062 0.5912-4.25133 0.0000 Table 2: The Summary Of The Results Of IPS Test For Selected Countries Based on the results in the above table, premp variable that its Prob is smaller than 5%. Null hypothesis of non-stationary is rejected but other two variables are not stationary but when when the first rank differentiation is used, null hypothesis is rejected and IPI and Apt stationary of first rank I(1). The optimum lags are determined by Schwartz criterion. Statistic Prob Panel adf-stat -4.464388 0.000 38
Group adf-stat -6.27472 0.0000 Table 3:The Results Of Co-Integration Test For Insurance Penetration Index Based on the results of two statistics and their Prob is smaller than 5% and null hypothesis of the lack of co-integration relation between the variables is rejected. It can be said that there is a longterm relation between the variables. Chi-sq.statistic PROB Chi-square statistics 79,747966 0.000 Table 4: The Results Of Haussmann Test For Insurance Penetration Index Based on the results in the previous section, a long-term relation between the variables is supported. It is required that a suitable method is used to determine the method of this relation and its value. To do this, Haussmann test was used to determine a good method in estimation of model parameters. By this test, a good estimation method is used among fixed effects and random effects methods. As in this test, H0 shows that estimation of weights of fixed effects and random effects have no difference. The results of this test and the probability of supporting null hypothesis are shown in the above Table. The obtained λ2 for selected countries was 79.747966. By comparing this statistics with statistics value of critical value Table and statistics probability that is smaller than 0.1, it can be said that the best method for estimation of model is fixed effects method, as null hypothesis is rejected. Based on the results for model estimation, fixed effects method is used. β1 β2 β3 β4 DW R2 0.7909-0.38062-0.3611 0.4948 2.24 0.616 Table 5:The Results Of Model Estimation By Fixed Effects Method For Insurance Penetration Index 39
As the number of coefficients except intercept is equal to 4 and the number of observations is 90, based on Durbin-Watson Table, we have: dl =1.57 4-du=2.25 du=1.75 4-dl=2.43 As DW (2.24) is greater than dl=1.57, non-autocorrelation is supported. - Estimation of coefficients by fixed effects method M_IPI = -2.18266405583 + 0.790926942858*PREMP - 0.380628654644*M_APT - 0.361139465412*DUM + 0.494849911558*DUM*M_APT + [CX=F] Coefficient is income elasticity of demand for non-life insurance and it shows the percent of changes of premium received to the percent of changes in premium per capita. Based on the positive relation between demand for insurance services and insurance penetration index, coefficient is positive. Coefficient denotes elasticity of open economy to GDP before membership in WTO. Here, is equal to -0.38 and it shows that one percent increase of open economy, insurance penetration index of the selected countries is reduced as -0.38. Coefficient is dummy variable and it shows the effect of membership in WTO. In this model, coefficient is -0.36 and it shows that membership of selected countries in WTO, reduced insurance penetration index of these countries. In other words, membership in WTO had negative effect on insurance penetration index. Coefficient showed the changes in open economy before and after joining WTO and its coefficient value is 0.49 and it shows that the increase of open economy increased the insurance penetration index of the selected countries. - The investigation of the effects of joining WTO on insurance density index of the selected countries Variables L PREMP L GDPP Level Statistic prob -1.99756 0.0229 0.48589 0.3135 Table 6: The Results Of Unit Root Test 1 st difference Statistic prob -4.88401 0.0000 40
As is shown in the above table, we have GDPPI(1) variable. It is stationary in first rank difference. Statistic Prob Panel adf-stat 0.040921 0.5163 Group adf-stat -1.516005 0.0648 Table 7: The Results Of Co-Integration Test For Insurance Density Index Based on the results of co-integration test, long-term relationship between the variables is not supported and ECM method or error correction method is used. Chi-sq. Statistic PROB Chi-sq. Statistic 9.843285 0.0432 Table 8: The Results Of Hausman Test For Insurance Density Index Based on ه statistics that is greater than table statistics and Prob smaller than 0.1, fixed effects method is used for model estimation. β1 β2 β3 β4 DW R2 0.121748 0.2367732-0.00109 0.335688 0.801 1.5 Table 9: The Results Of Estimation Of Model By Fixed Effects Method For Insurance Density Index The calculation of Durbin-Watson: As the number of coefficients except intercept is equal to 4 and the number of observations are 80, Thus: dl =1.53 4-du=2.26 du=1.74 4-dl=2.47 As obtained DW (1.5) is equal to d1 of table, non-autocorrelation is supported. Estimation of the coefficients by fixed effects method 41
PREMP = 2.90633927433+ 0.121748330589*ECM + 0.236773166446*M_GDPP - 0.001093086444*DUM + 0.335688202554*DUM*M_GDPP + [CX=F] Coefficient is estimated as 0.12 and it shows that in each period, 0.12 of non-equilibrium in premium per capita is adjusted and it is approaching the long-term trend. Coefficient is demand income elasticity for non-life insurance for the years before membership that is 0.23. In other words, during the years before membership, one percent increases in GDP per capita, the demand for non-life insurance of the selected countries is increased as 0.23. Coefficient is dummy variable and it shows the effect of membership in WTO. In this model, coefficient is -0.001 and it shows that membership of the selected countries in WTO reduced the premium per capita of these countries. In other words, membership in WTO had negative effect on premium per capita. Coefficient showed the changes in GDP per capita before and after joining WTO and its coefficient is 0.33 and it shows that the increase in GDP per capita increased premium per capita index of the selected countries. 5. Discussion and Conclusion As it was said, the selected countries are all developing countries and they were encountered with negative effect of membership in the organization (negative value of Dummy variable (DUM)) in the first five years of joining WTO, thus the main hypothesis regarding the positive effect of membership in confidence interval 95% was rejected. The reason is that at the beginning of joining WTO, due to the lack of preparation of economic infrastructures as the lack of competition freedom, it caused that the non-life insurance companies had state supports and lost their efficiency at the beginning before being adapted to WTO rules. Another issue is that based on the study results, membership in WTO increased open economy of the selected countries. Based on the positive correlation between human development index of each country and insurance industry, the findings of the study are generalized to Iran. Based on the above items, it can be said that Iran membership in WTO had effect on performance and indices of Iran non-life insurance. As the non-life insurance in Iran is encountered with special problems, fulfilling the positive effect of membership in WTO on Iran non-life insurance indices in short-term is impossible. 42
References Behkish, M. (2001). Iran economy in globalization process. Tehran. First edition. Nashr Ney. 24-26. Sehat, S. (2005). The study of the effective factors on insurance demand by the companies. Insurance industry journal. 20, 78-45. Iran central insurance statistical report. (2010). First chapter. The global position of insurance industry in 2010. Tehran. Central insurance in Iran. 23. www.un.org Baltagi, B. H. (2005). Econometric Analysis of Panel Data", 3rd ed, England, John Wiley & Sons Ltd. Eviews 7 User s Guide II, chapter 20, panel estimation, pp. 372-373. Suri, A. (2011). Econometric with application of Eviews 7. Culture studies edition. Tehran. Second editon. 221-241. Ashrafzade, H. R. Mehregan. N. (2010). Econometric of panel data. Tehran. Second edition. Cooperation researches of Tehran University. 135-138. Shujie Y., Zhongwei H., Genfu F. (2007). on technical efficiency of China's insurance industry after WTO accession", China Economic Review, 18:66 86. Zaranejad, M., Anvari, A. (2005). Application of pooled data in regression analysis method in various sciences. First international conference of research methods in sciences, techniques and engineering. Tehran. 43