Accurate estimation and reporting of loss reserves is essential for property-casualty insurance companies. The loss reserve is the principal liability and its measurement affects both income and owners’ equity. The topic has received considerable regulatory attention recently because inadequate loss reserves may result in insurance company failures. These failures damage both policyholders’ and shareholders’ interests and may destabilise financial markets. The principal focus in this thesis is the error in estimating loss reserves, i.e. the difference between the initial estimate of total losses incurred and the total actual claim payments. On the one hand, it is an inevitable consequence of the uncertainties sustained within an insurance business. On the other hand, this uncertainty may create opportunities for management manipulation. These may be respectively described as nondiscretionary and discretionary loss reserve errors.
This thesis is divided into three individual studies. The first study develops a model to separately estimate the nondiscretionary and discretionary components of total loss reserve errors. This helps to overcome the problem of indirect measurement of the dependent variable in regression estimates of the relationship between management discretion and incentives. The second study examines these relationships in detail, using three levels of financial information, reported, adjusted, and unbiased, to compute incentive variables and uses three different estimation techniques. It explains discretionary loss reserve errors in terms of income smoothing, tax planning, capital adequacy, investment income, reinsurance purchases, and auditor type. The last study examines the effects of institutional factors on discretionary loss reserve errors. Both the direct effects and the indirect moderating effects are examined. The institutional factors considered include legal origins, shareholder protection, general enforcement environment, public and private enforcement, policyholder protection, the integration level in insurance regulators, and the twin-peak model of prudential supervision.
This study furthers our understanding of the discretionary behaviour in loss reserve reporting. The sample includes insurers from 40 countries, thus extending an existing literature that has focused on the US and, to a lesser extent, the UK market. I find that there is significant international diversity in reserving behaviour and that it reflects complex institutional factors that vary across countries