Sunday, August 2, 2009

More On the Importance of Nominal Spending Shocks

Scott Sumner has a new article that provides a nice follow up to my previous post where I argued that stabilizing nominal spending rather than inflation is the key to macroeconomic stability. Scott similarly argues that inflation targeting is a poor option for monetary policy compared to nominal income targeting. After reading his post, I was reminded of some quick and dirty empirical analysis I did on this blog that lends support to this view. Here is an excerpt:
The importance of stabilizing nominal spending can be seen in the graph below that plots the relationship between the output gap and nominal spending shocks. The output gap is calculated as the percent difference between actual real GDP and the U.S. Congressional Budget Office’s potential real GDP. The nominal spending shocks series is calculated as the deviation of the year-on-year growth rate of quarterly final sales to domestic purchasers from its preceding 10-year moving average. The data cover the period 1953:Q1 - 2008:Q1.

The scatterplot makes it clear there is a strong, positive relationship between nominal spending shocks and the output gap. As a comparison, I have constructed in the same way an inflation shock series from the PCE price index and plotted it below against the output gap.

These figures indicate nominal spending shocks are more closely related to the output gap than inflation shocks. Given these results, I went ahead and plugged the nominal spending shock and output gap series into a vector autoregression (VAR) to get a sense of their dynamic relationship. Five lags were used in the VAR, which is enough to remove serial correlation from the quarterly data (data already in growth rates so no unit roots). After estimating the model and imposing recursive ordering to identify the structural shocks, I got the following impulse response function (IRF) for the output gap given a 1 standard deviation shock to nominal spending:

In plain English, the above figure shows that the typical shock to nominal spending leads to about a 0.5 percentage point increase in the output gap--a positive output gap--that persist for about a year and then begins unwinding. Another interesting exercise is to look at the decomposition of the forecast error from the VAR. This exercise explains how much of the forecast error can be attributed to a certain shock. (It tells us whether the interesting results from the IRF really matter)

Here we see that nominal spending shocks account for about 50% of output gap forecast error, a significant amount. By comparison, if the VAR is reestimated with the above inflation shock series instead of the nominal spending shock series, only about 8% of the forecast error can be explained by the inflation shock. Nominal spending shocks matter greatly!

Now I do want to oversell the findings presented here since they are based on a two variable VAR, but they are highly suggestive that nominal spending shocks are more important to macroeconomic stability than inflation shocks. Hence, monetary authorities should pay more attention to nominal spending. Moreover, stabilizing nominal spending should do better than inflation targeting at preventing the buildup of financial imbalances and asset bubbles for reasons explained here. In short, I am big believer that there would be meaningful gains in macroeconomic stability should the Fed should adopt a nominal income targeting rule.
Here is the original post.


  1. You may have already answered this question in a post in the past, but do you have a target value for nominal GDP growth, and what instrument(s) do you see the Fed using to hit it ?
    Is there some kind of feedback rule we could use?

  2. ECB:

    I have alluded to George Selgin's Productivity Norm rule in the past. His rule would have nominal income grow at the expected growth rate of factor inputs. It would effectively stabilize input prices but give output price more flexibility. In other words, it would allow for productivity-driven deflation.

    I, however, have not seen it ever tested in dynamic general equilibrium model and can not be sure of all of its consequences.

  3. Ok right I forgot about the Selgin norm. I did find one paper on testing nominal GDP targeting in a dynamic optimization model, and that was McCallum & Nelson, NBER wp6675. Unfortunately its gated so I cannot read beyond the abstract.