Defined benefit pension plans play a significant role in financial markets and have a significant impact on the financial decisions taken by firms that provide them. The authors identify some key factors that determine the level of investment risk taken by corporate sponsors of defined benefit pension plans. They show that these factors can change over time, particularly because of changes in the level of plan funding and in the fund sponsor’s probability of default.

This research is particularly useful to individuals involved in risk-based regulatory supervision because the authors identify a set of risk factors that can be monitored over time as pension investment risk changes. The findings, in conjunction with plan funding levels, could be used as a basis for Pension Benefit Guaranty Corporation (PBGC) premium rates: Corporate sponsors operating at elevated risk levels could be charged higher risk premiums than sponsors operating at more conservative levels. Other practitioners can use these findings as a potential gauge of future credit deterioration.

The authors use pension beta as a proxy for pension investment risk in a series of univariate regressions to identify the key factors that drive investment risk. Pension beta is defined as the difference between pension asset beta and pension liability beta, weighted by plan assets and pension liabilities, respectively. Pension beta has the advantage of taking asset/liability mismatching into account.

The first set of factors tests the authors’ risk shifting and risk management hypotheses. Risk shifting (or moral hazard) can occur as a result of the safety net provided by the PBGC. Firms with defined benefit pension plans pay a premium to the PBGC, which then pays out a minimum level of pension benefit in the event of the failure of a corporate sponsor. The firm can treat this insurance as a put option and maximize firm value by increasing the level of investment risk. The authors investigate whether low funding levels and high risk of credit default are consistent with high investment risk.

The risk management hypothesis considers an alternate view. Because restrictions are placed on firms with low funding levels (such as capital expenditure and dividend payments), the risk management hypothesis postulates that firms have an incentive to adopt low-risk investment strategies to improve the funding position of their plans.

Other factors the authors test include taxation (firms with higher tax rates have an incentive to invest in such lower-risk assets as bonds, which are taxed at lower rates within pension funds), financial buffers (firms that are trying to build a buffer in the pension plan during favorable economic conditions tend to invest in liquid, low-risk assets that can be used in a downturn), an accounting effect hypothesis (to justify the use of a high expected rate of return on pension plan assets, an aggressive investment policy must be in place), and a labor unionization factor (firms with high levels of labor unionization tend to offer higher benefits, which require higher investment returns). Finally, a risk synchronicity hypothesis is tested, which takes into account the correlation between firm performance and pension plan asset performance.

The regressions cover the period of 1990–2007. Unique sources of data include Form 5500 filings with the IRS and union membership data from the Union Membership and Coverage Database.

The univariate analysis shows that the risk management hypothesis dominates the risk shifting hypothesis. On average, firms with lower plan funding levels and higher credit default risk tend to adopt a more conservative strategy with lower investment risk (as proxied by pension beta) rather than the strategy suggested by the risk shifting hypothesis. The single-variable regressions support the risk management hypothesis; the evidence does not support the risk synchronicity or risk shifting hypotheses. The authors do find evidence to support the tax benefit, accounting effect, financial slack, and labor union hypotheses.

The authors’ findings from multivariate regressions show that investment strategy decisions can change as a firm’s circumstances change. For example, although firms with low funding levels and high default risk usually assume a low level of risk, firms on the brink of bankruptcy or those about to convert to a defined contribution plan, in which members’ contributions rather than benefits are pre-defined, often adopt a very aggressive strategy (taking advantage of the PBGC safety net). Other examples of dynamic strategy implementation include funds recovering from a position of underfunding.

Unless senior management has a stake in the defined benefit pension plan, its financial incentives may be aligned with those of shareholders rather than pension plan members. Management is obligated to maximize firm value, which implies a high level of pension investment risk. Consequently, the stakeholders most affected by the authors’ findings are likely to be the plan participants and their representatives, who may be relatively powerless to act because of the nature of defined benefit pension plans. The findings are also relevant to regulators using a risk-based supervision approach, but this method of supervision tends to be more resource intensive compared with the assessment of such quantitative-based measures as funding levels and proportion of equities.

Future research may include an exploration of whether there is a relationship between pension investment risk and subsequent movements in credit ratings, as well as an investigation of the time period after the financial crisis, non-U.S. funds, and differences between and within public and private sector funds.