Total defined-benefit pension assets of U.S. companies today are substantial and run into several trillion dollars. For a typical firm, these pension assets are over 15.0 percent of the firm's total assets, over 35.0 percent of its market value, and over 55.0 percent of its book value. (1) The pension asset management policy significantly affects the future returns on these assets and, therefore, the contributions to be made by the firm. Understanding the incentives of pension fund managers helps us to predict asset allocation decisions and fund performances better. Knowing how plans adjust to past performance also helps in the use accounting numbers to forecast future earnings.
Performance measurements are indispensable to hiring fund managers, checking their performance, and firing them if necessary (Arnott, 1997). Plan sponsors are increasingly interested in establishing performance-related termination thresholds for investment managers. Managers are terminated when they underperform the applicable benchmark by either an absolute amount or by a certain percentage of the benchmark return. A major criterion to judge the pension fund manager's performance is the return earned by his or her pension asset portfolio. Higher returns imply lower employer contributions to the defined-benefit pension fund in the future. For higher expected returns, fund managers have to invest in riskier portfolios and on the efficient frontier. Most sponsoring firms establish fund management policies that state the desirable risk-return profiles for their pension funds. These policies are typically guidelines and fund managers have adequate freedom to determine the riskreturn profile of their portfolios.
We find that dismal performance (relative to funds with similar risk profiles) prompts managers to reallocate their assets in the subsequent periods. This result persists even after controlling for riskiness of the asset portfolio, firm and plan sizes, funded status, profitability, leverage, and age distribution of participants. We also observe that these actions of the managers result in improved portfolio performances even after controlling for investment risk and the mean-reversion phenomenon of asset returns. Further, we observe that the market responds negatively to pension asset reallocations to riskier portfolios. Finally, we observe that shareholder value is positively affected by good performance. These results persist even after controlling for a firm's risk, profitability, leverage, and funded status. Thus, we find that the pension gains that ensue from improvements in plan performance and that are included in accounting income are a fundamental and predictable part of doing business. As such, these gains impact future pension costs and indirectly impact what has become known as "core earnings" (Standard & Poor, 2002). Hence, we question Standard & Poor's treatment that excludes pension gains from core earnings.
Theory and Development of Hypotheses
According to McGill and Grubbs (1984), the total accumulation of assets in a pension plan comes from contributions and returns on investments of pension assets, after adjustments for benefits paid and administrative expenses. The return on investments reduces the future costs of a defined benefit pension plan and increases the benefits under a defined contribution plan. (2) Thus, productive deployment of pension assets is a major concern of both the sponsoring firm and regulatory authorities. For the sponsoring firm a sound investment policy reduces the future cash outflows from the firm to the defined-benefit plan. For the regulatory authorities it reduces the chances of default on the part of defined benefits plan and subsequent payments of benefits by the Pension Benefit Guaranty Corporation (PBGC). Management of pension plan assets is regulated at both federal and state levels. The federal controls arise from the statutes of the Employee Retirement Income Security Act of 1974 (ER1SA) and the Internal Revenue Code (IRC). The full responsibility for the management of assets rests upon the sponsoring firm. (3) Other restrictions include segregation of plan assets, (4) identification of plan assets, diversification of investments, (5) and prudence standard. (6) Investment management is also governed by the Securities Act of 1933, the Securities Exchange Act of 1934, the Investment Company Act of 1940, and the banking laws (in cases of investments in banks and trusts). In addition to the federal restrictions, several states also have promulgated laws to regulate pension investments in trusts and insurance companies.
Unlike defined-contribution plans, where the investment risks are borne by the beneficiaries, in defined-benefit plans the plan sponsor bears most of the investment risk. For defined-contribution plans, a poor return on assets lowers the future benefits to participants but future costs to the sponsoring firm are unaffected. On the other hand, for defined-benefit plans poor returns on assets do not affect future benefits to participants (unless plans terminate prematurely and entire benefits are not insured by the PBGC) but future costs to the plan sponsor are increased.
Thus, sponsors of defined-benefit plans can invest in riskier assets (follow a more aggressive investment policy) without violating their fiduciary obligations under ERISA, unlike sponsors of defined-contribution plans who have to be more cautious and conservative in their investment policy. Good and Love (1990) also report that defined-benefit plans typically are more actively managed and have riskier portfolios with higher long-term returns. This gives fund managers of defined-benefit plans more investment freedom but added responsibility for results, in comparison to their counterparts in defined-contribution plans. (7) In the current paper, we examine the investment policies of defined-benefit plan managers. (8)
Given the emphasis on pension portfolio performance, we hypothesize that managers of defined-benefit pension funds with poor performance (relative to benchmarks) will attempt to improve their performances. On the other hand, managers with good performances are unlikely to make extra efforts to improve more, because this might raise the expectations about their future performances. One reason for performance below the benchmark could be that the portfolio is not on the efficient investment frontier, i.e., portfolio holds an interior, less efficient position. Managers of pension portfolios can improve their performance by reshuffling their asset-mix to get closer to the efficient frontier. On the other hand, pension fund managers with good performances have no such incentives to reallocate. Thus, the first pair of alternative hypotheses is:
Hypothesis 1: Defined-benefit pension funds with poor performances are more likely to reallocate their assets in comparison to those with good performances. Hypothesis 2: Defined-benefit pension funds with poor performances are more likely to improve their performances in comparison to those with good performances.
Shareholders are assumed to be risk averse. News that the sponsored pension plan holdings are riskier will be priced negatively. On the other hand, shareholders will react positively to improvement in performance, driving up share values, as higher returns on pension assets imply reduced cash outflow from the firm in the future as employer's contribution. Thus, the last pair of alternative hypothesis can be written as:
Hypothesis 3: Share prices are negatively affected by asset reallocations to riskier investment categories by defined-benefit pension funds. Hypothesis 4: Share prices are positively affected by improvements in performances of defined-benefit pension funds.
Research Design …