In defined benefit pension fund investments, the funding ratio is affected by bond yields through both the asset and liability sides. This paper evaluates the impact of misspecification uncertainty on the funding ratio forecasted 10 years ahead, using a prediction interval incorporating misspecification uncertainty to account for using a wrong model. We focus on some commonly-used affine term structure models to investigate the impact of misspecification uncertainty. These models include those taking into account macroeconomic factors, and those not explicitly excluding arbitrage opportunities but performing well from an empirical point of view. As an application, we consider a stylized defined benefit pension. The empirical analysis shows that the impact of misspecification uncertainty on the forecasted funding ratio is non-negligible, especially when a simple-structured nominal model is preferred in practice.

JEL Classification: C52; C58; C68; G12
Keywords: prediction interval, misspecification uncertainty, misspecification interval, pension fund investment, funding ratio