The masquerading problem
By Scott Sumner
For the past four decades, I’ve been complaining about the way the profession does empirical work on monetary policy. The studies often use “vector autoregressions” to estimate the impact of changes in the policy interest rate. Unfortunately, interest rates are not monetary policy. You can try to estimate the part of interest rate movements that are “exogenous” and hence reflective of monetary policy, but in practice this is almost impossible.
So after four decades of VAR studies by the best and the brightest in the economics profession, where are we? Here’s the abstract to a promising new paper by Christian Wolf of Princeton University:
I argue that the seemingly disparate findings of the recent empirical literature on monetary policy transmission are in fact all consistent with the same standard macro models. Weak sign restrictions, which suggest that contractionary monetary policy if anything boosts output, present as policy shocks what actually are expansionary demand and supply shocks. Classical zero restrictions are robust to such misidentification, but miss short-horizon effects. Two recent approaches – restrictions on Taylor rules and external instruments – instead work well. My findings suggest that empirical evidence is consistent with models in which the real effects of monetary policy are larger than commonly estimated.
Many previous VAR studies have found monetary shocks to have relatively weak effects on the economy. But these shocks tend to conflate reductions in interest rates with expansionary monetary policy. In fact, in the vast majority of cases a decline in interest rates reflects slower growth in aggregate demand, not easier money. So it’s no surprise that lower interest rates don’t seem to provide much of a boost to the economy. Lower interest rates are not easier money:
Sign restrictions, as in Uhlig (2005), are vulnerable to expansionary demand and supply shocks “masquerading” as contractionary monetary policy shocks, which then seemingly boost – rather than depress – output. . . . For monetary policy transmission, my results encouragingly suggest that, first, recent advances in identification effectively address the masquerading problem, and second, even small sets of macro observables may carry a lot of information about policy shocks. Viewed in this light, I conclude that existing empirical work quite consistently paints the picture of significant, medium-sized effects of monetary policy on the real economy.
This “masquerading problem” is sometimes called the identification problem; it’s what happens when people engage in reasoning from a price change.
After forty years, economists yet to develop a generally accepted VAR model of the monetary policy transmission mechanism. Like fusion power, there’s a small chance that it may happen some day. But there’s almost no chance I will live long enough to see this approach yield useful results. The profession would be much better off switching to an approach that used NGDP growth expectations as the primary indicator of monetary policy shocks, and then develop models to estimate those expectations using real time data from asset markets. Interest rates are not a useful variable when analyzing monetary policy.