Did Pension Plan Accounting Contribute to a Stock Market Bubble?

During the 1990s the assets of corporate defined-benefit (DB) pension plans ballooned as a result of the booming stock market. Because of accounting rules for DB plans put in place in 1986, this robust growth provided a substantial, although stealthy, boost to the profits reported by sponsoring corporations. In particular, the extraordinary returns earned on pension assets flowed to the bottom line on corporate income statements through lower net pension cost accruals included in general corporate expenses.

These developments may have misled many investors about the value of corporate equities, because pension cost accruals provide a fairly convoluted signal of the underlying value of net pension assets, in two ways. First, the accounting rules allow firms to smooth the effect of volatility in asset returns in calculating net pension expense; this smoothing both hides the variation inherent in the realizations of risky returns and tends to make current accruals of pension cost a stale measure of a pension plan's net asset value. Second, the net costs of financing outstanding pension liabilities are effectively understated when pension sponsors assume a future rate of return on plan assets that far exceeds the discount rate they use to calculate the present value of plan obligations. In effect, the costs associated with providing a pension plan to employees are offset on the sponsoring firm's financial statements by a smoothed and inflated stream of income flowing from the pension portfolio.

In this paper we assess the degree to which investors may have been fooled by current pension accounting rules and practices. We do so using two alternative models of pension valuation. The first, known as the standard transparent model, holds that investors gauge the contribution of a pension plan to the sponsoring firm's value by looking at the plan's marked-to-market net asset value, which is reported in footnotes to the firm's financial statements. The second, the "opaque" model proposed by Jeremy Gold, presumes that the market's assessment of net pension value is driven instead by the pension cost accruals reported in the body of the firm's income statement.¹

Our analysis applies a standard framework for equity valuation based on abnormal earnings and shows that, under the current accounting regime, the market appears to pay more attention to the flow of pension-induced accruals reported in the body of the income statement than to the marked-to-market value of pension assets and liabilities reported in the footnotes. These findings strongly support the predictions of the opaque model of pension valuation. We then perform a second battery of tests to determine whether the market prices a firm's pension accruals any differently than it prices the firm's core business earnings. The results suggest that investors do not distinguish between these two sources of earnings, at least not in the way that one would expect in an efficient market. If anything, the earnings associated with pension accruals appear to receive a higher valuation multiple than do core earnings.

Finally, we bring this evidence to bear on the question of whether there was a substantial pension-induced bubble in equity prices by simulating firm-level valuation errors using one of the empirical models employed in our hypothesis tests. The simulations suggest that, for the average firm in the Standard & Poor's (S&P) 500 with a DB plan, pension-induced valuation errors added 2 to 3 percent to the stock price during the late 1990s. Although significant in terms of absolute dollars of wealth, this source of error thus explains little of the runup in stock prices over this period. However, the estimated pension-induced distortions rise considerably in 2001, when the plunge in pension net asset values had not yet shown through to pension cost accruals. In particular, we estimate that, as of early 2002, one-tenth of the firms in our sample that sponsored a DB pension plan were at least 20 percent overvalued, relative to otherwise similar firms, and that the unweighted average level of overvaluation of the stocks of...