Working Paper: NBER ID: w24643
Authors: Narayana R. Kocherlakota
Abstract: In the wake of the Lucas Critique, the study of appropriate macroeconomic policy has largely focused on the comparison of different regimes/rules. In practice, few policymakers are faced with making those kinds of choices. In this paper, I examine the problem of a policymaker making but one in a long sequence of similar decisions (like to raise or cut interest rates by a quarter percentage point). I model the policymaker as playing a dynamic game against a forward-looking private sector. My main result is that, under relatively weak conditions, the policymaker's optimal within-equilibrium response to the current state can be found by applying statistical regression methods to past macroeconomic data. Theory is only useful as a source of information about credible functional form restrictions on these regressions. Based on this result, I argue that macroeconomic policy evaluation intended to be of practical value should rely considerably less on putatively structural macroeconomic models and considerably more on regression-based approaches.
Keywords: macroeconomic policy; dynamic game; statistical regression; policy evaluation
JEL Codes: E58; E60; E61
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| use of regression analysis (C29) | improved policymaking outcomes (D78) |
| variation in policymakers' private information (D82) | all possible policy choices (D72) |
| past data (Y10) | estimated fixed exogenous payoff function (C51) |
| past outcomes (G41) | past choices (D01) |