SWOV homepage
Nederlands  

The effect of a no-claim bonus system on reported claims

This report (R-98-47) attempts to provide an answer to the question of whether negative or positive financial consequences of insurance claims really do lead to fewer accidents. The research question was phrased as follows: what are the effects of financial stimuli on reported damages?

To answer this research question, two groups of insured motorists were compared: one group insured according to a no-claim bonus system, and one group who was not insured according to a no-claim bonus system.

From the claims data made available by the Association of Dutch Insurers, a selection was made in order to create the two groups. The selection was made in such a way that a certain association of claims could be made with 'accidents'. Non-accident related claims were removed form the analysis, for example claims with regard to stolen property. Injury accidents were also excluded from the analysis because they make up only 5% of all claims, but respresent extensive loses. If not excluded they would have dominated the analysis due to third-party liability claims.

The analysis was conducted in three steps:

  1. An analysis of the claim amounts made to investigate the similarity of the groups showed no marked differences between the two.
  2. A paired comparison was based on groups differing only in insurance policy. Claims data associated with groups having only different insurance policies are assumed to be similar were compared. In these analyses, no systematic differences were found which could indicate that one of the two kinds of policies would score significantly better than the other.
  3. A comparison was made based on individual policies by applying a generalised linear model.

The most conspicuous conclusion from this analysis was that it is not true that a no-claim bonus arrangement generally leads to fewer reported claims and/or less third-party loss. This finding was based both on the paired comparison analysis and the analysis using the generalised linear model.

SWOV Research Activities 11 - March 1999

Knowledge base