Claim patterns of private health insurance for individual and group contracts and the risk selection mechanisms

• Main Author: Mokienko, Anastasia
• Other Authors: Sverre Kittelsen
• Format: Master Thesis
• Language: English
• Published: 2011
• Subjects: VDP::210; Norway; North Norway
• Online Access: http://hdl.handle.net/10852/17055; http://urn.nb.no/URN:NBN:no-30303

This study is based on the analyses of the claim patterns in the light of the theory of risk-selection in the private health insurance market. Our aim is to study whether insures with group contracts have higher claims than the individual insures. However the limits in our data make it difficult to give a clear answer. The data is received from one of the Norwegian insurance companies and contains approximately 6300 processed claims from their customer portfolio in the period 2007-2010. We have analyzed both separate claims and aggregate claims per person for individual and group policyholders. We have controlled for type of the contract (group/individual), gender, age, geographical area, Oslo/other big city, industry sector (for group contracts) and reservations. Surprisingly, the type of the contract does not have any significant effect on the claim, neither considered separately nor aggregate per person. The analysis of separate claims has shown that the age of the policyholder has a significant positive impact on the size of the claims independently of the contract type. Reservations have an effect as well. 1% of group and 12% of individual policyholders have received reservations not to cover expenses connected to some particular illness they have. The presence of reservations has contrary effect on the size of the separate claims: negative for groups and positive for individuals. Further, we have analyzed what affects the likelihood of claiming for greater than average sums. We have found that age and gender (male) affect the likelihood positively for group contracts. Individual policyholders are more likely to claim for higher sums the older they are, if they have reservation, for children (i.e. being under 18 years old), while the likelihood is reduced for those who are registered in East Norway. The second part of the analysis is dedicated to the effects on the aggregate claims in the period 2007-2010 per person. It has shown that only the age of the policyholder has positive effect on the aggregate claims for group contracts. However, age does not influence size of aggregate claims of individual policyholders. They claim for bigger aggregate sums if they have reservations, come from North Norway or a big city other than Oslo. The analysis of the likelihood of higher aggregate claims has shown quite different results for individuals and groups. Group policyholders are more likely to ask for higher sums in the long run the older they are, if they are males and if they come from West Norway or a big city. Individual policyholders are more likely to claim higher sums in long run the older they are and in a presence of reservations. We have calculated relative numbers for the average claim ratio among the insured in 2010 and have found that it is smaller for the group contracts than for the individual. The ratio is equal to the number of persons who claimed in 2010 divided by the number of persons who were registered in 2010 as customers. The ratios contradict with the theory of risk selection. The theory predicts that “selected” customers are more profitable, and in our case the “selected” ones are individual policyholders because they are checked and they have to deliver health statement that can be used for giving them a reservation or even being rejected. The size of the aggregate claims that we have got from descriptive statistics for years 2007-2010 differs from the results of 2010. We assume that it is caused by tendencies in the development of PHI market that we did not take into account. The 2007-2010 data shows more or less the same numbers for group and individual policyholders, while the 2010 data shows much smaller number for group contracts than for individual. Aggregate data in year 2010 contradicts with the theory of risk selection and indicates that group policyholders (which are not selected) are more profitable customers. This shows that either risk-selection is not as profitable as expected or that group policyholders are a special type of people with special socio-economic characteristics (like that they have full-time job, and they are in the working age 18-68 years old and more). In addition group policyholders are usually subscribed automatically and therefore not all of them are clear that they have a PHI. Another factor is that the turnover of the customers with group contracts is smaller because of the automaticity of the group insurance registration process. Individual policyholders do have bigger turnover and they are more likely to drop insurance if they do not use it, thus, older and sicker are in the pool and, thus, the ratio of people with claims is higher for individuals than for groups. Even though the individual policyholders are selected on the basis of health status statements they might be an adversely selected group because of unobservable health differences. The aggregate data for 2007-2010 shows that there is not that much difference between aggregate claims of persons with group contracts and persons with individual contracts. In addition, the coefficient of the contract type in the regression of the aggregate data set shows that group policyholders spend less than individuals (although not significantly less). We conclude that we did not get a clear indication for or against profitability of risk-selection in our study, and we need more research in this field.