Finding the Equilibrium Real Interest Rate in a Fog of Policy Deviations

A large number of economic research papers have recently been written endeavoring to estimate the current level and trend in the equilibrium real interest rate. These studies are generally model-based: they either use semi-structural time-series and filtering methods or formal structural dynamic stochastic general equilibrium (DSGE) macroeconomic models to examine the relationship between the equilibrium real interest rate and its possible determinants. A common finding in these studies is that the equilibrium real interest rate has declined in recent years to a level not seen in decades.

In this paper we examine the underlying methodology used in these model-based estimates. First, we show that the estimates are subject to omitted variable or even omitted equation bias. What appear to be trends in the equilibrium interest rate may instead be trends in other policy variables that affect the economy. While such problems have always made it difficult to find and estimate concepts such as the equilibrium real interest rate, recent changes in policy variables have deepened the fog. Second, we show that methods used to adjust monetary policy rules to take account of shifts in the equilibrium interest rate alone are incomplete and misleading because they do not incorporate other shifts—such as changes in potential GDP—that are associated with the shifts. These shifts need to be accounted for when the results are applied to monetary policy decisions or rules. Finally, we show that alternative simulation techniques can radically alter the results. In sum we conclude that the estimates of time-varying real equilibrium interest rates that have emerged from recent research are not yet useful for application to current monetary policy.