Federal Reserve Policy

Warning: Very, very wonky post on interest rate policy at the Federal Reserve.

There has been a lot of discussions and academic research on optimal interest rate policies. How do and how should central banks determine interest rates? The most common explanation in recent decades is to use a Taylor Rule approach to setting rates. The rule is named after Stanford professor John Taylor who first published his paper on the topic in 1993. Basically what the Taylor Rule does is look at economic growth relative to potential GDP and actual inflation relative to targeted/expected inflation. The further the economy is above potential or if inflation is running higher, the higher the interest rate should be – to cool the economy and head off inflation. If the economy is below potential and/or inflation is below target, then lower interest rates should be used to boost economic growth.

If you examine the data, it does appear that the Federal Reserve set its Fed Funds rate according to such a policy in the late 1980s and the 1990s. The dotcom bubble and recession caused the Fed to lower the Fed Funds rate from over 6 percent in 2000 all the way to 1 percent in 2003/4. A common criticism of Fed policy in the past decade was that it left interest rates too low for too long during the middle of the 2000s, helping to fuel the housing bubble. The most common way to back up such claims is to show how the Fed Funds rate was much lower than a Taylor Rule approach says it should have been during that time (notice how the blue line in the graph is significantly above the black line from 2000 through 2008). So far, so good in running through a brief overview of conventional thought on the subject.

Where this gets interesting, in a wonky way, is new research that shows the actual Fed Funds rate can be closely tracked if you were to use the Federal Reserve’s own economic forecasts. The Taylor Rule is an outcome based calculation (how far is actual growth from potential growth, e.g.) while the new research by Volker Wieland (former senior economist at the Washington DC Fed, and first published in 2008) calculates the interest rate conditional on the Fed forecasts. Now that the Federal Reserve has not only committed to keeping interest rates low until 2014, it has also begun publishing forecasts of the Fed Funds rate itself, this research and its findings by Mr. Wieland (and co-author Athanasios Orphanides) takes on more importance.

The graph below shows the actual Fed Funds rate (black), the Mankiw Rule (a simpler Taylor Rule approach) and then Mr. Wieland’s research findings (with and without interest rate smoothing). I have recreated Mr. Wieland’s findings and associated forecasts using the publicly released Federal Reserve forecasts and any differences between his work and the graph below are solely attributable to my errors. The actual Fed Funds rate forecast is from IHS Global Insight and the Mankiw Rule forecast is calculated based on Global Insight’s unemployment rate and core inflation forecasts.

While the new research is interesting and provides further insight into the interest rate decisions at the Federal Reserve, the biggest takeaway right now is that it does not matter which forecast one uses or which equation specifications one chooses, the end result is the same. Given that all three specifications say the Federal Reserve will not tighten policy until at least early 2014 and maybe not until late 2014, it is no wonder that the Federal Reserve has come out publicly stating that it will not raise interest rates until 2014.

Finally, even though this research shows that the Federal Reserve was setting interest rates appropriately in the 2000s given its own forecasts, that does not mean that they actually set interest rates appropriately. Whether they did or not remains heavily debated in academic circles and is certainly not answered here. What is shown in this new research is that one can reasonably explain the Fed’s policy choices by looking at the Fed’s own forecasts. Whether those forecasts were reasonable or not is another question entirely.