UNDERWRITING CYCLES IN EMERGING AND NEWLY INDUSTRIALIZED ASIA AND THE CAUSES

Abstract

This study aims to provide empirical answers to two main sets of questions. First, do underwriting cycles exist in the emerging and newly industrialized Asian general insurance markets of interest? If they do, what are the approximate cycle lengths? Second, what are the factors influencing the underwriting cycles on an aggregate sample countries-basis as well as on individual sample country- and line- bases? For the countries that were studied in previous studies, we seek to provide more robust results by using longer sample periods and more dependent variables as well as by conducting robustness checks. We use a second-order autoregressive model to detect cycle existence and to measure their periods. Further, we conduct robustness tests to verify the stability of the estimates obtained. We also employ models which regress annual premium changes on independent variables such as (annual changes in) lagged losses, real gross domestic product, interest rates, stock market performances and catastrophic losses to examine the relative importance of these factors in influencing underwriting cycles in the various markets. Out of five insurance lines for each country, an average of four was found to exhibit cyclical behaviours. Underwriting cycles were especially prevalent in the Fire and Motor lines. Cycles were found for these two lines in all countries examined. For the UAE, cycles were detected in all the emirates examined. With regards to factors that could explain the underwriting cycles, we found several significant relationships between hypothesized factors and annual premium changes. By aggregating data, we found changes in real gross domestic product (among other factors) to be highly significant, both statistically and economically, in its relationship with annual premium changes. By focusing on the individual countries and their respective lines, we found annual changes in lagged losses to be significantly related to the premium changes of at least one line for all the countries examined.