Tuesday, May 26, 2020

Extensions to the NGDP Gap

The monetary policy program at the Mercatus Center recently released a new measure called the NGDP gap. We created it as an alternative way to gauge the stance of monetary policy and have provided a website that will update the measure as new data become available. In this post, I will briefly summarize the NGDP Gap and then highlight a few extensions that some readers may find useful. 

Summary of the NGDP Gap 
As mentioned above, the NGDP Gap provides a cross check on the stance of monetary policy. Its use does not require the Fed to adopt a NGDP target, but it does draw upon the fact that NGDP is comprised of both real GDP and the price level and therefore captures both elements of the Fed’s dual mandate. Moreover, since NGDP is a nominal variable it can be shaped by the Fed over the medium to long run. 

The basic idea behind this measure is to construct a benchmark growth path for nominal GDP (NGDP) where monetary policy is neither expansionary nor contractionary. Deviations of actual NGDP from this neutral level of NGDP provide a way to assess the stance of monetary policy. These deviations, in percent form, are called the NGDP gap. 

The NGDP gap can also be called the nominal income gap since NGDP equals NGDI. In fact, the construction of the neutral level of NGDP can be most easily understood from a nominal income perspective. To see this, consider that people make many economic decisions based on forecasts of their nominal incomes. Examples include households’ decisions to take out mortgages and car loans or firms’ decisions to finance with debt and commit to multiyear contracts on plants, raw materials, and labor. Sometimes, however, actual nominal incomes may turn out very different from what people expected and, as a result, may be disruptive for households and firms that are not able to quickly adjust their economic plans. These disruptions can be minimized by maintaining nominal income on the growth path expected by the public.

The neutral level of NGDP, then, is the public’s expected growth path of nominal income. Both this measure and the NGP Gap are shown below up through 2020:Q1 and come from a NGDP Fact Sheet we will be publishing each quarter. 

To be clear, non-zero NGDP gap outcomes need not be the result of Fed policy but of monetary conditions more generally. For example, the current NGDP gap exists because of the severe nominal income shortfall that has emerged from the COVID-19 shock. Consequently, the job of the Fed and U.S. Treasury during this crisis is to close this gap and avoid the secondary spillover effects (e.g. mass insolvency) this shortfall could create. Failure to close it would indicate a failure of countercyclical policy. This measure, then, provides a useful guide for the economic relief efforts during the pandemic. 

Extension I: Blue Chip Forecast Version
A key goal of this project was to provide a measure that is relatively simple to calculate and uses publicly available data. To that end, the neutral level of NGDP is based off of forecasts from the Philadelphia Fed's Survey of Professional Forecasters (SPF) and BEA data on NGDP. There is no use of r-star or u-star and therefore no "navigating by the stars" in this measure. The neutral level of NGDP is just an averaging of NGDP level forecasts from accessible data sources. Below is the formula for the neutral level of NGDP:

where NGDPt* is the neutral level and NGDPt-iSPF forecast(t) are NGDP level forecasts for period t coming from the past 20 quarters. NGDPt*, in short, is just a rolling average of NGDP level forecasts for a particular period. The difference between it and actual NGDP is the NGDP gap. 

Given the five-year (20 quarter) window in creating NGDPt*, there is a need for long-term NGDP forecasts. They are available in the SPF, but begin only in 1992 and therefore limit our series to a start date of 1997. 

The Blue Chip forecast database provides a long-term NGDP forecast that goes back further than the SPF. Alexander Schibuola and Andrew Martinez (2020) use it to construct an even longer time series of the NGDP gap. It is shown in the figure below along with the SPF version we use at Mercatus. The two NGDP gaps are very similar. 

Interestingly, Schibuola and Martinez use the data to construct a forecasted NGDP gap and it is disturbingly large. Even the recovery looks nasty. 

The use of Blue Chip data is a nice extension of the NGDP gap. However, we still plan to use the SPF version as our baseline version since the data is free and we can show the underlying calculations to the public. Eventually, we plan to provide the Blue Chip version as a complement to our baseline SPF version, but since it uses proprietary data only the final measure will be available. 

Extension II: Precision Version
Schibuola and Martinez also provide another useful extension of the NGDP gap that looks at its precision. They motivate this by noting two potential issues: (1) the forecasters in the SPF sample change over time and (2) individual forecasts in the SPF may be very different. Accounting for these two issues they produce the following chart that shows the range of individual forecasts for a semi-fixed sample of forecasters in the SPF. 

The median of the semi-fixed sample provides a very similar result to the overall median of all the forecasters. Also, the range of forecasts provides a way to better think about the stance of monetary policy. For example, one could make the case that monetary policy was neutral in 2019 since the range of estimates span both sides of 0 percent. 

Extension III: NGDP Targeting Application
As noted above, the use of the NGDP gap does not require the adoption of a NGDP target by the Fed. Nonetheless, a closer look at the forecasts used in constructing the neutral level of NGDP reveal that it could be used by the Fed as the target growth path for a NGDP target. For it would amount to a NGDP level target that slowly changes the target NGDP growth path based on changes to forecasts of potential real GDP.

To see why this is the case, note that we use a combination of short-run and long-run forecasts of NGDP to construct the neutral level estimate of NGDP. The SPF provides distinct quarterly NGDP forecasts for five quarters out: t+1 to t+5. After that, we use the average annual NGDP forecast over the next 10 years adjusted to a quarterly basis for quarters t+6 to t+20. This is seen in the table below. 

What this means is that three-fourths of each NGDP neutral level estimate is being shaped by a long-term forecast of NGDP. This long-term forecast, in turn, is the sum of a 10-year average GDP deflator inflation forecast and a 10-year average real GDP growth rate forecast. The long-term inflation forecast is determined by the Fed's inflation target while the long-term real GDP growth rate forecast is shaped by expected changes in the potential real GDP growth rate. 

Consequently, as the neutral level of NGDP series moves through time, it can be seen as a rolling average of expected changes to potential real GDP growth plus the Fed's inflation target. This is the kind of NGDP level target some advocates, like Jeff Frankel, would like to see implemented.  

The figure below shows the neutral level of NGDP constructed with the Blue Chip data, complements of Schibuola and Martinez. This version allows us to see a hypothetical NGDP level target from late 1987 to present based on the neutral level measure of NGDP. 

Again, the original intent of the neutral level of NGDP and the NGDP gap is simply to provide a crosscheck against other measures of the stance of monetary policy. The discussion of a NGDP level target is simply an extension of this work. 

Here's hoping, though, that the Fed and Treasury keep this measure front and central in their efforts to provide economic relief during the COVID-19 crisis.