Using near-VARs to examine phase-dependent monetary and fiscal policy

Economic policies are known to have different effects on the economy depending on the size of policy changes and on business cycle conditions. For example, monetary policy might be more stimulative in recessions and around turning points in the business cycle than during expansions. Such dependencies, however, are usually ignored in most empirical research or are placed under the rubric of "long and variable lags." Accounting for phase-dependent policies holds out the possibility not only of better forecasting performance with our macroeconomic models and of more accurate methods to identify policy effects, but also of broadening our view of how to understand business cycles. The phases are estimated using threshold autoregressive (TAR) methods suggested by Hansen (1996). ; This paper extends the policy framework of Christiano, Eichenbaum, and Evans (1994), and empirically estimates phase-dependent behavior using a variant of a popular empirical business cycle model. Commonly referred to as the near-VAR (NVAR), this model helps to estimate dynamic interrelationships and to conserve degrees of freedom when dealing with short data samples. Despite the rather simple empirical measures of the business cycle phases, these near-VARs provide a better understanding of the relationship between cyclical phases and monetary and fiscal policies.