Risk Taking and Incentives of Institutional Investors: Evidence from US Defined Benefit Plans

This paper investigates the volatility of defined benefit of pension plans over the period 1999-2006. We document economically significant relationships between the asset-liability structure and the risk-taking of pension plans. We find that increasing the underfunding of a bottom size-decile plan by one-standard deviation on average implies a 34 % increase in the volatility of the plan. In contrast, the relationship between funding status and risk is reversed for top size-decile plans. The higher risk-taking of underfunded plans is confined to higher upside-risk exposure, consistent with plans hedging pension liabilities via structured finance instruments with non-linear payoffs. Defined benefit (DB)1 pension plans today seem less secure than a decade ago. The total amount of US retirement assets in private DB plans was $2.328 trillion as of March 2007 (According to the Investment Company Institute). There is a growing concern about the funding status of DB pension plans as many of them are experiencing declining solvency. While regulators and policy makers are calling for reform of the institutional framework of DB pension plans, understanding the determinants of DB risk-taking is of critical importance. This paper investigates the relationship between the risk-taking of US DB pension plans and their asset-liability structure and characteristics. We report evidence that the total risk exposure of privately-held pension plans is explained by the interaction of the funding status of the pension plan and the plan size, while the risk-taking of publicly-held corporate plans is driven by their commitment to accrue future obligations. Moreover, we find evidence of asymmetric risk exposure of the average pension plan that is consistent with hedging of pension liabilities. These relationships between funding and risk can be interpreted as evidence of different managerial incentive structures across different plans. Theoretical arguments for optimal pension-plan risk-taking have produced conflicting predictions about both high and low risk (for example, Sharpe (1976), Treynor (1977), Blac