|
|
|
Brainstorms
Stock Companies, Mutuals, and Risk Measures
by Stephen W. PhilbrickI'm happy to see that all of the Proceedings of the CAS (since 1914) are now available online. I was fortunate to start my career with The Hartford, which had a complete set of Proceedings in its library. In one of my many breaks from studying the Syllabus, I would often peruse an old volume to see how our predecessors developed the profession. I noticed that we used to make more of an issue about the distinction between stock companies and mutuals. Expense loadings were different for the two classes of companies. But I recall few other references to stocks versus mutuals in the quarter century I've been in the business. That changed last week, when two people mentioned in it two different contexts.
At the DFA meeting in Boston, Glenn Meyers made the tongue-in-cheek suggestion that of two competing algorithms for capital allocation, perhaps one would appeal to stock companies and the other to mutual companies.
Shortly thereafter, I was startled to hear another stock versus mutual comparison. Several of us in my office were discussing measures of risk, such as variance, semi-variance, expected policyholder deficit, and the newest measure, tail value at risk (TVaR). We were trying to decide which measure was best. John Burkett suggested that perhaps measures such as variance are best suited for stock companies, and measures such as TVaR are best suited for mutual companies. He elaborated by saying that stock companies tend to be more interested in the variability of their entire results because they are concerned with quarterly earnings reports to stockholders. Mutual companies are free from the pressure of quarterly results, but might be more concerned about threats to capital since they are less able to raise capital in the marketplace.
I'm still not convinced that the demarcation between the business planning of stocks and mutuals is that dramatic, but I wonder if the distinction could be viewed another way. Companies have to plan over different time horizons. For example, short-term planning must be done to decide whether to write a particular piece of business at a particular price. Long-term planning is needed to decide what lines of business the company will write. Perhaps the underwriting decision on a specific risk is based more on the potential impact on short-term earnings, so variance matters more. Perhaps decisions about the capitalization level of the company are inherently more long term, and a risk measure emphasizing the tail is more relevant.
The theoretical underpinnings of the tail measures of risk seem persuasive from a mathematical point of view. But it is difficult to explain to company management that the most extreme 1 percent of the distribution matters more than the typically experienced results. The extreme tail events may be most important to regulators and a few policyholders concerned about the company's ability to pay claims, but the variability of the rest of the distribution often matters more to the owners of the firm.