Distributional properties of some financial ratios in insurance

This paper presents a distributional analysis of some financial ratios in the insurance industry. Although previous studies have established a reasonably good understanding of the frequency distributions of financial ratios, to date most of our knowledge has been derived from cross-sectoral studies, or from the manufacturing sector in particular. This paper aims to contribute to the literature by further applying and extending this knowledge into the financial services sector. The focus is on four key profitability ratios: the combined ratio, the expense ratio, the loss ratio and return on assets. A sample of 115 global insurance conglomerates, 64 from the United States and 55 from the EU, is used to test the effects of an outlier-removal procedure and data transformation on the normality of the distributions. The study successfully applies a modest normalisation procedure for all four ratios, highlights the dangers of conducting parametric tests on non-normal data series, and finds that lognormal transformation is beneficial just for the Return on Assets ratio.