The demand for life insurance in OECD countries

Journal of Risk and Insurance, Sept, 2007 by Donghui Li, Fariborz Moshirian, Pascal Nguyen, Timothy Wee

Contrary to expectations, life expectancy exhibits a positive relationship with life insurance demand, whereas the dependency ratio appears to influence life insurance demand negatively. This pattern is likely to reflect a wealth (income) effect. As income increases, so does life expectancy while the number of children tends to decrease. Controlling for wealth (income) effect appears necessary in order to reestablish the right direction of influence produced by these two variables. Increases in life insurance demand do not seem to be related to differences in social security expenditure and real interest rates. Education level and financial development present the postulated positive influence on insurance sales. In the opposite direction, high inflation rates appear to restrain the demand for life insurance. Finally, foreign participation displays a nonlinear relationship with life insurance that seems to warrant the inclusion of the squared form of foreign participants' market share in the regression models.

Multivariate Analysis

Table 3 presents the regression results using pooled data. We use both OLS and GMM estimation methods. GMM corrects estimates for possible heteroscedasticity of residuals and cross-correlation among regressors (see Greene, 2000). The difference between the two methods is more likely to be seen in Models 1 and 2, given that all independent variables can be used as instruments in the GMM procedure. All regressions appear to produce a reasonably good fit, with adjusted [R.sup.2] between 66 and 85 percent. In addition, the coefficients usually have the predicted signs. Both the White and Breusch-Pagan tests indicate significant heteroscedasticity problems that justify the use of the GMM estimators.

As suggested in the univariate analysis, the income variable has a positive and significant influence on life insurance demand. However, the income elasticity of life insurance demand displays significant variations across models, depending on whether product market conditions are considered or not. Adopting a conservative interpretation, the results suggest that a 1% increase in aggregate income is associated with an increase of about 0.6 percent in life insurance sales. The results are consistent with the models of Campbell (1980) and Lewis (1989). Truett and Truett (1990), Browne and Kim (1993), Outreville (1096), and Beck and Webb (2003) obtain similar income elasticity coefficients.

Socioeconomic variables appear to play a less significant role in total insurance sales in comparison with product market characteristics. Consistent with Browne and Kim (1003) and Outreville (1996), we find that life expectancy is not always significant although it displays the postulated negative influence in Model 3. One reason may be the difficulty to separate the positive wealth effect associated with a higher life expectancy from the negative effect associated with the lower probability of the income earner's premature death. The results do not necessarily contradict the theoretical premise that protecting the dependents' expected future consumption motivates life insurance purchase. In fact, demand for life insurance displays a significant correlation with the number of dependents. The results are robust across estimation methods for Model 3. The negative influence suggested by the first OLS regression is considerably reduced when the more consistent GMM method is used. The result supports the insight of Campbell (1980) and Lewis (1089) that the number of dependents increases the value of the dependents' future consumption, which in turn stimulates the demand for life insurance. The results are also consistent with the findings of Browne and Kim (1993) and Outreville (1996), who find coefficients only slightly higher. Similarly, the education level is a positive and significant contributor to the demand for life insurance. The higher demand appears to reflect the increased level of risk aversion suggested by Browne and Kim (19c)3), which may reveal a greater awareness of life's uncertainties as well as the higher disutility associated with the dependents' loss of future consumption.