The market for catastrophe risk: a clinical examination

Abstract

This paper examines the market for catastrophe event risk – i.e., financial claims that are linked to losses associated with natural hazards, such as hurricanes and earthquakes. Risk management theory suggests protection by insurers and other corporations against the largest cat events is most valuable. However, most insurers purchase relatively little cat reinsurance against large events, and premiums are high relative to expected losses. To understand why the theory fails, we examine transactions that look to capital markets, rather than traditional reinsurance markets, for risk-bearing capacity. We develop eight theoretical explanations and find the most compelling to be supply restrictions associated with capital market imperfections and market power exerted by traditional reinsurers.

Introduction

Hurricanes, earthquakes, wind and ice storms, floods, etc. have long been known to cause large and unexpected losses to owners of physical capital. Recently, it has become widely appreciated that a single hurricane or earthquake could result in damages of $50–$100 billion. Given the growth rates in physical asset values and in population in high-risk zones, distribution of catastrophe event losses continues to grow¹.

Because households are risk averse, they have a strong incentive to share their risks with others through the purchase of insurance. Corporations, depending on their level of concern with risk management, also have an incentive to purchase insurance. Insurers are particularly motivated purchasers of reinsurance since they would otherwise become concentrated warehouses of catastrophic exposures. If corporations in general and insurers in particular behave in a risk averse manner with respect to these exposures, then they treat severe losses as expectationally more costly than moderate losses. One would therefore expect that reinsurance is focused on catastrophic outcomes. Moreover, since cat event losses are uncorrelated with (and perhaps even independent of) financial wealth, the premiums for such catastrophic protection should, if markets are perfect, approximate expected losses.

This paper explores these propositions, provides both large-sample and clinical reinsurance data on their veracity, and then attempts to understand why they fail. We first use evidence from a large and unique dataset of reinsurance transactions to show that protection in general is far more limited and prices are far higher than can be readily explained by the theory. We then try to understand these deviations better using clinical information. We look at several recent, widely-discussed transactions by USAA (one of the largest insurance companies in the US) and the California Earthquake Authority (a state insurance pool set up to help fund earthquake losses). They are among the first to back reinsurance with dedicated collateral supplied by bondholders. Traditionally, reinsurance contracts have been backed by the general credit of reinsurers, who use equity in ongoing reinsurance businesses to fund themselves. These newer bond transactions display the features mentioned above, i.e., high prices and limited quantities. But the process of innovation that they represent is in itself interesting, reveling clues about the imperfections and inefficiencies of the reinsurance marketplace that economists’ models tend to miss.

The paper then turns to develop alternative hypotheses, eight in total. The majority of these focus on distortions on the supply side, but several suggest problems with the demand side as well. The most important explanations concern supply-side phenomena of capital market imperfections facing reinsurers and the exercise of market power by reinsurers.

In many respects, the evidence is most interesting for its implications beyond that of the cat risk market itself. In the conclusions, we discuss several lessons drawn from this evidence for the broader behavior of capital markets and corporate risk management.