Life Insurers in the Public and Private Security Markets

Abstract

An analysis of determinants of life insurer behavior in the market for publicly issued securities indicates that purchases of securities in this market are partly responsive to unexpected cash flow. Acquisitions in the private market (through the forward commitment process), however, are partly dependent on expected cash flow. It is suggested that the growth of life insurer participation in the market for publicly issued securities may be related to the rise in the volatility of financial markets which has increasingly engendered greater demand for liquidity. In tlurn, this rise curbed the willingness of life insurers to forward commit, evidenced not only by the decline in amounts committe(l, btit by the lengthening of thie lag between changes in expected cash flow and new commitments. Studies on life insurer investing have paid little attention to the market for publicly-issued securities. Such a neglect may stem from the fact that purchases of life insurers in this market have been, heretofore, small compared to their acquisitions in the private market {2, 3, 4, 5, 8, 9, 1 1}. In the last few years, however, life insurers have stepped tip their purchases of public issues. A sample of life insurers increased their purchases of bonds in tile public market from $3.5 billion during 1980 to $14.8 billion in 1982. This experience compares with average annual investments of $3.6 billion during 1977-1979. The rise in purchases of publicly issued bonds was accompanied by a decline in private placement acquisitions to $6.6 billion in 1982 from $8.5 billion in 1980, and from an annual average of $12 billion duiring 1977-1979. This note analyzes the (leterminants of the investment behavior of life insurers in the public market, extending thie analysis to the private market for purposes of comparison. Data and Research Methods The data used in this study are gross acquisitions taken from the new commitments survey conducted by the American Council of Life Insurance (ACLI). A partial measure of life insurer purchases of securities (bonds and common stocks) in the public market is provided by data on new commitments Senior Economist, American Council of Life Insurance. This content downloaded from 157.55.39.153 on Mon, 19 Sep 2016 04:59:12 UTC All use subject to http://about.jstor.org/terms Life lns.ureis in the Pildylic 349 that are made and taken down in the same month.' The methodology used is regression analysis. The model for purchases in the public market requires an estimate of unexpected cash flow. An expected cash flow series is first generated, and then the difference between actual and expected cash flow is used to represent cash flow surprises. In this study, the explanatory variables are four past values of cash flow itself and two interest rates, lagged once. Lagged values of the dependent variable for forecasting purposes can be rationalized by the fact that cash now is importantly determined by rather stable inlhfwk\s of premium payments under long-term contracts and by net investment flows. Cash flow, however, is also subject to prevailing conditions in capital markets. When interest rate levels are high, policy loans and other withdrawals tend to increase, reducing funds available for investment {71. To capture the influence of changing market conditions, the 91-day Treasury bill rate is included in the equation. Finally, the rate on seasoned BAA corporate bonds (Moody's) is included because it closely tracks the rate of return on the bond portfolios of life insurers. Information on these variables is readily available to life insurers, increasing the likelihood that these variables may be used to predict cash flow. A regression estimate of this model generated the expected cash flow variable, CFE. A series of tests indicated that the forecasts are rational in that they are unbiased and that the errors are orthogonal, i.e., current errors are not systematically related to previous errors {l(O, pp. 18-19).2 'An alternative source of data is the survey on gross yield on new investments which shows life insurer bond purchases in the public an(d private markets. However, the data are available only on an annual basis and the series, which began in 1977, is relatively short. Nevertheless, a comparison of the annual totals of the two series shows close correspondence, providing some basis for confidence that the data on public purchases froni the new commitment series may in fact capture the general trends. Moreover, the regression estimates using either series provide essentially similar results. 2To determine whether CFE was an unbiased predictor of actual cash flow, the following equation was estimated: