The Return of the String

It’s been a while since we’ve read about the FOMC pushing on a string, but that string is getting renewed attention. Although there is broad consensus that monetary expansions lead to increased output and contractions lead to decreased output, in their new paper, “Gaussian Mixture Approximations of Impulse Responses and the Nonlinear Effects of Monetary Shocks,” Christian Matthes of the Richmond Fed and Regis Barnichon of Universitat Pompeu note there is little agreement on asymmetric and non-linear effects of monetary policy. (link below) That leaves two key questions unanswered. For the first, they employ the string metaphor—is a contractionary shock, pulling the string, more effective than an expansionary one, pushing the string? Second, how much does the state of the business cycle change the effectiveness of monetary policy?

The authors note that the standard approach to answering such questions, the use of Vector Autoregessions (VARS), is inherently flawed; such models are linear and cannot answer questions about asymmetrical shocks. Replacing VARS with Gaussian approximations (GMAs) the authors are able to estimate a moving average for the economy directly from the data, and then apply Gaussian basis functions in order to capture non-linear responses. Testing the GMA model against VAR results shows the former detects nonlinearities and estimates their magnitudes well.

And the string has it. On average, no matter how they identify monetary shocks, the contractionary ones have a “strong adverse” effect on unemployment, while expansionary ones have “little” effect. And the state of the business cycle does indeed play a role: when there is slack in the labor market, an expansionary shock has an expansionary effect, but in a tight labor market, an expansionary shock has no significant effect on unemployment, causing instead a “burst” on inflation. If the unemployment rate is 4%, an expansionary shock increases inflation by twice that suggested by VAR estimates. If the unemployment rate is 8%, there is no effect on inflation, probably because of downward wage rigidities. They note this is in line with the Keynesian narrative where a monetary authority working to expand an economy already operating above potential would achieve only higher inflation. (We discussed problems with how we measure potential, and its weak trajectory, in our last issue: we’re moving toward potential because the bar keeps falling.)

The authors suggest we employ their models to ascertain nonlinear effects of fiscal policy shocks. Indeed. A coherent fiscal policy would be such a shock in itself it might blow out the Gaussian models. But, joking aside, monetary policy has been pushing that lonesome string for a long time now, and if we want a more vibrant economy and the higher rates that come with that territory, we’d better try something else, something with some real pull.

https://www.richmondfed.org/publications/research/working_papers/2016/wp_16-08