هنگامی که سیاست پولی غیر سیستماتیک بر تورم ماثر است؟

An important stylized fact to emerge from VAR estimates is that exogenous monetary policy shocks (also labelled unsystematic monetary policy) have a delayed, persistent, hump-shaped effect on inflation. I argue that this empirical pattern is fragile. In particular, it disappears when one examines periods without large shifts in the level of inflation (such as 1984–2005). An important consequence is that the hump-shaped VAR estimated response of inflation is not appropriate to fit stylized models of the response of inflation around a stable steady state inflation level.

One of the most widely accepted stylized facts of monetary economics is that US inflation has a hump-shaped response to exogenous monetary policy shocks. For instance, Christiano et al. (2005, pp. 5–8) state, “after an expansionary monetary policy shock […][…] inflation responds in a hump-shaped fashion peaking after about two years.” Likewise Mankiw (2001) writes, “According to the consensus view among central bankers and monetary economists, a contractionary monetary shock raises unemployment, at least temporarily, and leads to a delayed and gradual fall in inflation.” This stylized fact is supported by a series of robustness checks along several dimensions, though all within the VAR framework, reported in the second chapter of the 1999 Handbook of Macroeconomics: “Monetary Policy Shocks, What have we learned and to what end?” (also by Christiano et al., 1999a; thereafter CEE-99a) and in many contributions reviewed therein. However, this VAR based characterization of the effects of unsystematic monetary policy1 on inflation is very sensitive to the choice of the sample period. In particular, if one considers the last twenty years, VAR estimated monetary policy shocks have no effect on inflation nor the price level. The hump-shaped response of inflation is obtained only if either the building up or the collapse of the 1970 Great Inflation is included in the samples over which the VAR is estimated. One important implication is that models that are consistent with the evidence estimated over long sample periods may be mixing up the response of inflation to monetary policy shocks in periods of large adjustments of inflation, such as the so-called Volker disinflation, and periods when the mean of inflation is stable, e.g. from 1984 to 2005. There is therefore a risk that these models provide a poor approximation of inflation dynamics for both periods of large adjustments and periods when the mean of inflation is stable. The paper proceeds as follows. Section 2 reviews the literature on VAR identifications of US monetary policy shocks. Section 3 focuses on the changes in the inflation impulse responses estimated on the 1984–2005 sample period relative to the ones obtained for the 1960–2005 period. Section 4 concludes.

This paper shows that the hump-shaped response of US inflation, which is obtained with VAR based identification and widely considered a stepping stone for the development of structural models, is not robust. In particular, the post 1984 experience is one where VAR estimated money supply shocks have no significant effect on inflation and the response of the price level is flat. This change in the response of inflation points to time variation in the effects of monetary policy. Outside periods of large and persistent adjustments, the VAR based measures of non-systematic monetary policy do not affect inflation. While this result may simply reflect some limitations of the VAR methodology, it nevertheless establishes a new benchmark stylized fact for the calibration of macroeconometric models of the US business cycle.