Evan Soltas
Mar 16, 2012

On Target, Redux

Using the monetary policy framework for historical analysis

Earlier today I introduced a method of describing monetary policy by trying to fit an equation in the form of aΔY + bπ = k, such that central banks used the rate of inflation to stabilize a quantity, which could describe an NGDP target or flexible inflation target readily.

As I promised, today I'm applying that framework more thoroughly to the central banks I listed on Friday's post -- the Bank of Canada, the U.S. Federal Reserve, the Reserve Bank of Australia, the Bank of England, and the Bank of Japan -- all of whom have publicly committed to flexible inflation targeting.

Before I go ahead, I'll note that I see two outstanding problems with flexible inflation targeting. First, since there is an explicit commitment only to an inflation target, such central banks tend to be insufficiently flexible in correcting deviations in real output. This is perhaps because any deviation in inflation jeopardizes the credibility of the target. Second, such central banks have not managed to make explicit how they will be flexible -- i.e. what deviation in inflation they are willing to tolerate in order to stabilize real output. This means that during recessions, a flexible inflation target often introduces considerable uncertainty in monetary policy at the worst possible moment. In fact, I would encourage central banks which do flexible inflation targeting to announce H-values.

Now to the analysis. The first thing I did was a backward looking check to confirm that my analysis comes close to describing real-world flexible inflation targets. Since this could get repetitive, I'll show you the results from Australia, and then I'll move on to the next topic.In the graph above, the blue points are actual monetary policy outcomes, and the red is the forecast as fit from historical data. Another way to look at the strength of the representation of monetary policy is by direct comparison of the actual and forecasted inflation:In both cases, we see that the model does a qualitatively good job describing the monetary policy of the RBA from 1993 to 2011.

The model, however, struggles to fit the historical data when there are major divergences in the pattern of monetary policy during the time span analyzed. It's actually amazingly consistent for most of this range for any single bank, although some banks will tend to be more or less inflexible when it comes to deviations in inflation. This all changes in 2008, when the recession forced real output growth sharply negative. Central banks did not raise inflation enough to match the monetary policy rules implied by their histories, which caused their target rules to "bend" towards more inflexible inflation targets. Consider the Fed, or the Bank of Canada, both below. The first fit line comes from their data pre-2008, the second includes all data.We can expand this historical analysis further than a pre- and post-2008 look, though. What I've done next is try to fit the policy rule quarter-by-quarter over the broader time spans, such that the ε term was minimized, and we can watch the policy evolve by tracking the H-value, H = log(b/a), over time.This graph tells us a lot. First, you can see that monetary policy was relatively consistent before 2008. Second, you can see the wildly different postures of the central banks during this time span, with the Fed leading the way on inflexible inflation targeting and the Bank of England and the RBA maintaining an almost perfect NGDP targeting regime. Then comes 2008, at which point we can see all heck break loose in the data. In particular, what you should notice is that both the Fed and the Bank of England responded by adopting in effect more inflexible inflation targets by ignoring the massive deviation in real growth. Only the RBA responded in such a way that it became more sensitive to real output growth deviations during the recession.

Out of personal interest, I extended this method of analysis even further back for the Fed, where I have FRED data running back to the late 40s.I find this absolutely amazing -- we can look at the sweep of American monetary policy since the beginning of independence for our central bank. And the quantitative analysis confirms what we know qualitatively. In the 1950s, monetary policy was very inflexible about inflation, as the Romers showed in a study. Then we saw a major shift in the mid-60s through the 70s towards a policy that tolerated higher and more volatile inflation in order to stabilize real growth. Just a note here: my problem with the policy during this time is that their NGDP rate target was too high -- 15 percent annually, as opposed to 5 percent. Then we see a well-known and major shift in monetary policy: the appointment of Paul Volcker and the return of the inflation hawks. As we might expect, the H-value turns sharply positive, as it has remained for the past three decades.

I'll keep working on applying my framework in future posts.