Investors favor business-line concentration over broad diversification. But high levels of geographic concentration are viewed negatively, as is undue concentration in areas prone to natural disasters.

Companies that keep expenses in check and GAAP asset returns high are rewarded with higher-than-average market-to-book ratios.

Our analysis, based on 37 publicly- traded insurance groups mainly engaged in p-c business, explored the relationships between companies stock market values and various financial and non-financial variables. By applying a regression analysis approach to year-end GAAP and statutory data for the years 1997 through 2000, we were able to explain much of the variation in market valuation over the four years studied.

Market-value assessments were based on two different measures: total dollar market capitalization and market-to-book (M/B) ratios.

The potential value drivers we examined included business decision variables (size, business mix, diversification and leverage), and performance variables (earnings, returns, efficiency measures).

Statutory data (data filed with insurance regulators) on reserve levels and premiums written were grouped into four major business linespersonal lines, commercial lines, workers compensation and reinsuranceand a number of subcategories. From this data we constructed variables that allowed us to describe a companys attainment of size thresholds (overall and by type of business), its concentration levels in specific businesses, and its diversification across businesses.

Premiums-written data, available by state, were segmented into eight regions to measure geographic diversification. The same data were then regrouped to construct another set of variables reflecting the degree of catastrophe exposure in those regions.

To gauge exposure to asbestos and environmental risk, we used the loss estimates found in the notes to the financial statements.

Our analysis confirms the view of industry analysts that balance-sheet measuresespecially book valueare more consistent drivers of market value than earnings or cash flow. For example, GAAP book value alone explains 93 percent of the variation in market capa somewhat better result than for GAAP assets (80 percent).

GAAP variables have much greater explanatory power than statutory balance-sheet measures like policyholders surplus (65 percent) and net admitted assets (62 percent).

Net income explained only 36 percent of the variation in total dollar market capitalization, we found. Cash-flow measures did little better, with EBIT (earnings before interest and taxes) and EBITDA (earnings before interest, taxes, depreciation and amortization) explaining only 39 percent and 44 percent of the differences in market values, respectively.

These results do not imply that profit and cash do not matter for p-c companies. Rather, they indicate that the capital markets do not rely on a single years results to project permanent differences in future earnings, and hence, in relative market values.

We also explored how the capital markets weight companies business-line and geographical choices and recent performance. We found that certain performance variables (returns-on-assets, expense ratios) and business decision variables (size, leverage, business-line concentration) appear to be statistically significant in explaining differences in market/book (M/B) ratios.

In particular, nine such variables showed a confidence level greater than 80 percent and with fairly stable magnitudes over both the entire four years studied and for the two shorter subperiods (2000 to 2001 and 1998 to 1999).

In other words, for each of these nine variables, theres an 80 percent or better chance that the variable does indeed have an impact on the M/B ratio.

The nine variables were ROA, expense ratio, business line concentration index, revenues, high leverage, percentage of personal auto reserves to total reserves, percentage of commercial auto reserves, percentage of reinsurance reserves, and percentage of asbestos reserves.

The accompanying chart shows whether the impact on the M/B of each variable was found to be positive or negative. For example, the “negative” impact of expense ratios means that higher expense ratios correlated with lower M/B ratios, all other things being equal. The reverse is true of business line concentration index, where a higher index (in other words, a more concentrated underwriting portfolio) implies a higher M/B ratio.

Two other variables that had a negative impact on the M/B ratiohomeowners reserve percentage and medical malpractice reserve percentagewere also above the 80 percent confidence level over the four-year period and for the years 2000 to 2001, but were less significant from 1998 to 1999. This suggests that business-line preferences may change over time.

These results send some strong messages from the market regarding how they can position their companies to increase and maintain shareholder value.

On the strategy side, the markets are telling p-c insurers to diversify across regions in order to spread risks efficiently, but to “stick to your knitting” when it comes to business lines. This is not surprising, given the unique and rapidly changing knowledge and skill sets required in underwriting different p-c lines.

Larger size also seems to produce an advantage, and leveraging debt does not appear to be disfavored.

The market appears to consistently reward companies whose business lines are more weighted to reinsurance and personal auto (the latter finding may surprise some), while penalizing companies with commercial auto and asbestos exposures.

On the performance side, companies are rewarded for keeping expenses under control. Maintaining a strong ROA is also a plus, but our regressions imply that realistic variations in ROA do not lead to economically significant differences in market/book premiums.

Finally, a number of variables often used to evaluate performance and assess relative value actually showed no statistically significant relationship to M/B ratios. This does not imply that these measures should be ignored, but they appear to be less important in explaining market-value differences than had been assumed. The list of such variables includes:

Loss ratio

Gross/net premiums to surplus

Change in writings or surplus

Investment yield

Liabilities to liquid assets

Agent balances to surplus

Many might wonder how the loss ratio, regarded by many as the best single measure of a companys ability to select, underwrite and price risks, could not significantly influence the markets assessment of a companys value.

Loss ratios are, of course, notoriously volatile, and one years historical result tells an investor little about the companys ability to achieve a sustained level of performance.

This interpretation is consistent with the relatively higher importance of book value compared with earnings or income measures in explaining differences in total market caps, and with the statistically significant, but economically miniscule, impact of ROA on M/B ratios.

In the volatile world of the p-c insurer, this points to the importance of the expense ratio as a measure of the efficiency with which the business can be conducted. While investors may forgive volatile loss ratios, which so often seem to be beyond managements control, they appear to place much weight on the companys ability to run an efficient business.

Perhaps the market is saying that an efficient company will do well when industry loss trends are favorable, while an inefficient player will underperform even when market conditions are right.

Perry Quick, Ph.D. (top), [email protected], is a partner in Ernst & Young, LLPs Economics and Business Analytics Group in Washington, D.C. Daniel Kahn, Ph.D., [email protected], is a senior manager in Ernst & Youngs Corporate Finance practice in Washington, D.C.

Reproduced from National Underwriter Property & Casualty/Risk & Benefits Management Edition, July 22, 2002. Copyright 2002 by The National Underwriter Company in the serial publication. All rights reserved.Copyright in this article as an independent work may be held by the author.