A Quick Look at Fed Forecasting
During the Chairman Bernanke’s recent press conferences, the first topic that he addressed was the Federal Open Market Committee’s (FOMC) set of economic projections. He outlined the Committee’s expectations for economic growth, inflation, and the unemployment rate for the next few years and the longer run. The numbers that he presented, however, offer only a snapshot of the Committee’s views.
A more complete picture is painted in the Summary of Economic Projections (SEP), which is released with the FOMC meeting minutes. The SEP reports the range of FOMC participants’ projections, along with a central tendency, which excludes the top three and bottom three estimates, kind of like a trimmed range. The SEP also contains detailed information on the uncertainty associated with the projections and the perceived risks to them. A close look at that information shows that the uncertainty in forecasts has increased over the past few years, but the dispersion of forecasts across FOMC participants has narrowed and returned to historical norms.
One way to measure the dispersion in the forecasts is to look at the difference between the top and bottom of the ranges. For example, charting the differences in ranges for the real GDP forecasts in the year the projections were made shows that the dispersion in the projections widened during the years of the financial crisis. There were varying views on the Committee about how severely the crisis would affect real economic growth. However, after the brief spike in the width of ranges, the dispersion of the forecasts returned to historical averages. Note that we split the projections by month since FOMC participants clearly will have more information about that year’s economic growth in November than in January. For example, the projections for 2009 economic growth that were made in January 2009 will be much more dispersed than those made in November 2009. It is expected that the ranges later in the year will be narrower, which is what we see.
To gauge uncertainty, we make use of the average historical projection error ranges that are reported for each variable in the SEP. These are simple averages of the root mean squared error for private and government forecasts over the previous 20 years. As noted in the SEP, “under certain assumptions, there is about a 70 percent probability that actual outcomes will be in the ranges implied by the average size of projection errors made in the past.”
We’ve pulled out the average projection error ranges for real GDP forecasts over the past four years. Looking specifically at the error ranges reported for the GDP forecasts in the June SEP in each of these years, one can see that the error ranges in the current-year GDP forecasts have stayed roughly the same size. However, as the projections shift further out into the future, the recent recession and financial market strains have significantly increased the error ranges.
|Projection year||One year forward||Two years forward|
Note: The central tendency excludes the three highest and three lowest projections for each variable in each year. The range includes all participants’ projections, from lowest to highest, in that year.
Sources: Federal Reserve Bank of Philadelphia; Summary of Economic Projections, January 2012 and April 2012, Federal Reserve Board; Bureau of Economic Analysis; Bureau of Labor Statistics; authors’ calculations.
Just in the last year, more information has been provided about Committee participants’ uncertainty in their forecasts. In addition to submitting their projections, participants have been asked to give an opinion on whether they thought the historical error ranges were an accurate portrayal of the amount of uncertainty in their projections. In the past four meetings where projections were submitted, most Committee participants viewed the uncertainty around their forecasts as higher than normal. At most, only four participants believed that the amount of uncertainty in today’s economic projections is similar to the past. No Committee participants believed there was less uncertainty about economic conditions than had prevailed over the past 20 years.
The SEP also includes information on whether FOMC participants believe outside risks to the economy are more likely to cause their projections to miss above or below the actual outcome. Currently, no participants see the balance of economic risks creating conditions in which economic performance would exceed their projections. Until the April meeting, most participants felt that their projections would turn out to be too rosy, if they missed at all. In April, those risks shifted such that the perception among the majority of participants was that it is equally likely that positive or negative shocks would affect economic growth.
What does this information on uncertainty say about the current outlook for economic growth? If we use the midpoint of the central tendency for real GDP growth projections in the June SEP to represent the Committee’s average projection, we would say that the Committee generally projects a 2.2 percent increase in 2012 and a 2.5 percent increase in 2013. When we factor in historical forecast accuracy, we would say that there’s roughly a 70 percent probability that real GDP will grow between 1.3 percent and 3.1 percent in 2012 and 0.9 percent and 4.1 percent in 2013. But since most FOMC participants believe that the historical forecast accuracy overstates how certain they are about their projections, a slightly larger range is implied. Given the risks to the economy, which participants reported to be either weighted to the downside or roughly balanced over the past year, we would put a little more probability on growth falling short of the midpoint forecast than on growth exceeding the midpoint forecast.