Any risk measure which seeks to relate returns to risk should be able to do so in the securities marketplace. Therefore, we have statistically tested the relationship between risk and return on returns earned by investing in various industrial groups through the bond and equity markets. Figure 7 is a summary of these results in which we use "M" as the measure of return. The risk-return points, when plotted, form an upward sloping line; that is, those industries which have earned a high rate of return for their investors in the 15-year period have also experienced high values of risk. Statistically, the goodness of the fit of this line is highly significant, which means that this relationship is not a spurious one merely because of chance factors, but most likely is based on the economic realities of risk and return. The conclusion is strong that this is indeed a meaningful measure of risk. Before proceeding to analyze the results measuring return at book value, it is important to emphasize (1) what we have learned from testing the relationship at market value and (2) why we did not, even if we could, construct an industry point on the market value. As was indicated above, the market value test of the relationship allowed us to gain the necessary confidence in the general validity of the econometric measure of risk. However, the relationship found is of no use in deciding whether or not real assets are being efficiently employed in any industry. This relationship simply says that the securities of the various industries studied are being distributed so that the rewards and risks to security holders are as described by the relationship. This would not rule out the possibility that the assets which these securities represent in any industry are not earning monopolistic returns or, on the other hand, failing to earn minimal returns sufficient to protect the industry from capital outflow. For this reason, no attempt was made to construct a property and liability insurance industry point on this relationship. Even if it were desired to construct such a point, the task would be impossibly complicated by the fact that there are very few widely traded securities which represent solely assets devoted to the property and liability insurance business. Most of the widely known concerns have various proportions of life and other insurance businesses in their portfolios. Stock market investors evaluate each line differentially, and any industry composite of companies with different proportions of various lines would be indeed misleading and not at all representative of the stock market evaluation of the property and liability business. Figure 8 shows results for book rates of return. Although the book rates of return are lower (as shown by the lower intercept on the vertical axis) and the scale of risk is more 100 120 140 160 180 200 220 240 260 280 300 320 340 360 Risk (Percent2/10) Industry Return (Fixed Charges Dividends Change in Market Value/Market Value Source: Arthur D. Little, Inc., developed from Standard & Poor's annual industrial FIGURE 7 MODEL 1: INDUSTRY RISK/RETURN PATTERN Source: Arthur D. Little, Inc., developed from Standard & Poor's annual industrial COMPUSTAT tape. FIGURE 8 MODEL 1: INDUSTRY RISK/RETURN PATTERN restricted, the pattern is similar to that formed using the market value measure in Figure 7. There is a consistent upward sloping relationship between actually experienced industry rates of return and the measure of industry risk for the 1950-1965 period. Property and Liability Industry Risk and Return, 1955-1965 The average rate of return for the 43 property and liability stock companies studied for the 1955-1965 period was 4.4%; their risk, measured by the average dispersion of the individual companies about this mean, was 10.89 (percent squared).9(See Figure 8.) Therefore, the point for the property and liability insurance industry (marked I) falls at the intersection of the horizontal line 4.4 units above the origin and the vertical line 10.89 units to the right, considerably below the regression line. In fact, it can be statistically determined how far below the population comprising the regression line it does fall. In only one case in a hundred could an industry with a risk of 10.89 units have earned the 4.4% or less by chance factors alone. The more likely explanation is that this particular industry is "underearning." It is not earning a rate of return commensurate with those earned in other economic activities on assets placed in similar risks. It was explained earlier that the measure of risk is based on a cross-sectional concept. As such, it eliminates many of the statistical difficulties encountered by previous researchers who used measures of risk based on time-series concepts. We can draw our conclusions quite unequivocally because of the large number of degrees of freedom involved in this study coupled with the high degree of significance found for our statistical ratios. Comparisons with Other Reported Rates of Return It is appropriate to look at traditional comparisons of rates of returns in different industries to see how the property and liability industry fares. Throughout, however, it is necessary to be careful of definitions and to make sure that these comparisons are appropri ate. The Monthly Economic Letter, published by the First National City Bank of New York, lists for each quarter and by year the net income and net worth of leading industrial groups in the United States. Its tables are widely distributed and these data are cited as the standard for rates of return in the industries it covers. Included in this series are a number of financial industries, such as commercial banks, fire and casualty insurance (to use their term), investment trusts, and sales finance. For example, Table 2 shows an average rate of return on net worth of 11.1% for all industries in 1965. Of all the industries reported in that year, the lowest rate of return was 2.8% by the fire and casualty industry. The only 8 Lack of necessary data prevented our using the 1950-65 period for the insurance companies. However, analysis of available industry data indicates that the results are insensitive to this difference in time periods. 9Unaudited data for 1966, which became available later in the course of this study, show a return of -2.49% and a risk of 9.93 units. This changes the overall results to a rate of return of 3.84% and a risk of 10.81 units for the period 1955-1966. TABLE 2 RETURN ON NET WORTH OF LEADING CORPORATIONS FOR THE YEARS 1964 AND 1965 a Net worth (listed previously as "Book net assets") is based upon the excess of total balance sheet assets over liabilities; the amounts at which assets are carried on the books may not represent present-day values. Due to the large proportion of capital investment in the form of funded debt, rate of return on total property investment would be lower than that shown on net worth only. d Federal Reserve Board tabulation of all member banks; number of banks (6,221) is not included in total number of companies, earnings include profits and losses on sale of securities; net worth figures are annual averages. proximate rate is the 2.9% reported for investment trusts. It is immediately clear that, by the standards of the First National City Bank, the fire and casualty industry is at the bottom in 1965. A closer analysis shows that the situation is even worse. Table 3 concentrates on the four financial industries reported by the First National City Bank and gives the statistics for 1955 through 1965 for net income over net worth (or rate of return on net worth). For the period as a whole, the fire and casualty industry averages 4.5%, well below banks at 9.2%, and sales finance companies at 13.3%, but above the investment trusts at 3.4%. However, it is necessary to look carefully at the definitions of net worth and net income for each industry. Fire and casualty insurance companies have their net worth defined as policyholders' surplus, which is equivalent to the net equity for industrial corporations. Net income for these companies consists of underwriting profits, interest and dividends, and realized capital gains, all after tax. The situation in commercial banks is much the same. Investment trust companies, however, present an altogether different picture. Net worth for investment trust companies is not some fraction of total assets, but is defined by the First National City Bank as net assets at market value. In addition, net income is defined as operating profits and interest and dividends; it does not include even realized capital gains. Therefore, it is seen that by definition the asset base is larger for investment companies and that the income stream is smaller. Even if the two industries were operating under exactly the same profitability conditions, the fire and casualty insurance industry would report a higher rate of return by these definitions. Looking further into sales finance companies which reported the highest rate of return, it is seen that, although the net worth figure is correctly defined, net income does not include capital gains. This may be entirely appropriate in that sales finance companies do not experience a great amount of capital gain, but it is not at all appropriate for investment companies in the sense that such gain is one of their principal goals (as it is with fire and casualty companies). Thus, the rate of return for the fire and casualty companies is overstated in these comparisons. We have, therefore, used Best's Aggregates and Averages to construct Table 4, which shows a number of rates of return for each of the years 1955-1965 and the definitions of these rates of return. We can note immediately that the definition closest to the one used for investment companies by the First National City Bank is N1/D2, i.e., operating income including interest and dividends over total investable funds. Here the income item is about the same, but the total investable funds asset base is somewhat less than net assets would be for the investment company. Conclusion on Risk/Return and Adjusted Return Comparisons We have looked at risk and return in the property and liability insurance industry in two ways. First, by principles we believe to have general economic validity and to be in line with modern regulatory, judicial, and legislative practice, we have related rates of return earned on assets to the risk environment in which the assets are placed. We have looked at the standard measures of return as reported by numerous sources to see where the particular industry falls in a ranking of these rates of return. With respect to this latter, we can |