2007, Vol.18, No.2

Journal of Insurance and Finance 2023 KCI Impact Factor : 0.59

pISSN : 2384-3209
https://journal.kci.go.kr/kiri

This study estimates and analyzes the technical efficiency of Korean general insurance firms for the period 1988-2005 by utilizing a recently developed stochastic frontier model. This model differs from traditional stochastic frontier models in that factors affecting efficiency are included in the model in efficiency estimation rather than treated as independent variables. Results show that most general insurers need significant efficiency improvement for future growth and survival in the industry. Most large insurers exhibit higher level of technical efficiency than midium or small insurers. The result implicates the size of insurers play an important role in insurance operation. Marketing channels, asset portfolios, and the diversification of insurance lines are found to be crucial factors for the technical efficiency of general insurers.

Unlike previous studies which estimated an underwriting cycle using AR(2) time-series models with aggregated loss ratio data, this study measures the underwriting cycle using a dynamic panel model with the yearly loss ratio data of property-liability insurance companies. We also used the one-step and two-step GMM, developed by Arellano and Bond (1991), to estimate our model parameter values. We analyzed models using the loss ratio data of entire property-liability insurance companies and the each line of property-liability insurance, respectively. The results of our paper are as follow. In case of entire property-liability insurance, the coefficients of loss ratio to each of the time lag and are significant at 1% level. In addition, there exists an underwriting cycle and its length is 5.4~6.9 years. Also, economic growth rates and growth rate of interest rate have a significant negative effect on loss ratio while growth rate of CPI have a significant positive influence. Next, by line of property-liability insurance, the underwriting cycle length of long-term insurance is estimated to be 7.2 years. Moreover, macroeconomic factors have only a statistically significant effect on the loss ratio of auto insurance.

This paper examines the Bernoulli Principle when an individual faces both a background risk and an interdependent insurable risk. The background risk is modelled as an initial random wealth. Within this model, using "expectation dependence" developed by Wright(1987), we provide the necessary and sufficient conditions for the Bernoulli Principle to hold or violate. That is, it is shown that the rational individual would buy less than, full or more than full insurance if and only if the expectation dependence is positive, zero or negative, respectively. These results extend the recent conclusion of, for example, Aboudi.Thon(1995) or Hong(2001, 2004) in two ways: 1.The notions of expectation dependence are less restrictive than those of regression dependence in Aboudi.Thon(1995) or those of Brumelle(1974)'s interdependence in Hong(2001, 2004); 2.We give conditions that are both necessary and sufficient, while Aboudi.Thon(1995) or Hong(2001, 2004) suggested only the sufficient, but not necessary, conditions for the Bernoulli Principle.