Monday, March 9, 2015

Fed Policy in 1995

I was interested in Paul Krugman's NYT column from this morning, mainly for his take on mid-1990s monetary policy in the United States. Krugman says:
Recent job gains have brought the Fed to a fork in the road very much like the situation it faced circa 1995. Now, as then, job growth has taken the official unemployment rate down to a level at which, according to conventional wisdom, the economy should be overheating and inflation should be rising. But now, as then, there is no sign of the predicted inflation in the actual data.
So, let's look at the data, restricting attention just to the 1990s:
You can see that, indeed, the unemployment rate in 1995 was about 5.6% vs. 5.5% in the last survey. But the pce inflation rate in 1995 was in the vicinity of 2%, versus 0.2% year-over-year in January. As Krugman notes, a Phillips curve view of the world would not help you much in making sense of the data, either in 1995, or now. For example, if we look at a scatter plot of the inflation and unemployment data for the 1990s, and link the observations in order, we get:
There are brief periods during the decade when inflation and unemployment actually move in opposite directions, but more often than not we get the reverse case. Indeed, from 1992 until 1999, the unemployment rate falls about 3 1/2 percentage points while the inflation rate falls about one percentage point. So much for the inflationary pressures of a tight labor market.

What was the Fed doing during the 1990s, and in 1995 in particular? Here's how Krugman describes it:
In the early-to-mid 1990s, the Fed generally estimated the Nairu as being between 5.5 percent and 6 percent, and by 1995, unemployment had already fallen to that level. But inflation wasn’t actually rising. So Fed officials made what turned out to be a very good choice: They held their fire, waiting for clear signs of inflationary pressure.
Here's what is in the data. We'll plot the fed funds rate and inflation, and the fed funds rate and the unemployment rate for the 1990s:
So, you can see that, by early 1995, the Fed had just finished a substantial tightening cycle - the fed funds rate had increased from about 3% at the beginning of 1994 to about 6% at the beginning of 1995. So, the Fed wasn't "holding its fire" in 1995, it had just launched a major artillery barrage, and had stopped shooting until the smoke cleared. And the fed funds rate in 1995 was at 6%, not at (effectively) zero, as is the case now.

Krugman's conclusion is:
What’s worrisome is that it’s not clear whether Fed officials see it that way [Krugman's way]. They need to heed the lessons of history — and the relevant history here is the 1990s, not the 1970s. Let’s party like it’s 1995; let the good, or at least better, times keep rolling, and hold off on those rate hikes.
Here's one way to see it as Krugman does. Apparently he thinks that 1995 Fed policy was appropriate. Suppose then that we fit a Taylor rule to 1990s data, of the form:

R = a + bi + cu,

where R is the fed funds rate, i is the pce inflation rate, u is the unemployment rate, and a, b, and c are coefficients we are going to estimate. OLS regression gives us

a = 9.2
b = 1.2
c = -1.2


And here are the raw data and the fitted values:
Note that policy is actually tighter in 1995 than average Fed behavior over the 1990s predicts. But what if the Fed followed Krugman's advice and behaved like it did in the 1990s? Well, the fitted Taylor rule, given an unemployment rate of 5.5% and a pce inflation rate of 0.2%, implies a fed funds rate of 2.8%. So if the Fed had been behaving like it did in the 1990s, it would have lifted off the zero lower bound long ago. Apparently Krugman is confused.

17 comments:

  1. Does it change much if you regress from 1995 onward?

    ReplyDelete
    Replies
    1. You'll get different Taylor rules depending on what data set you use, and what measures of inflation and output gap you use. A lot of them will predict a positive value for the fed funds rate given current inflation and output gap. But, it's certainly possible to find ones that will imply zero - a very aggressive response to inflation, for example (large b). But that creates problems. See this:

      http://newmonetarism.blogspot.com/2015/02/taylor-rules-zero-lower-bound-inflation.html

      Delete
  2. Steve, you and Krugman need to settle this on the track.
    http://krugman.blogs.nytimes.com/2015/03/09/wearables-and-self-awareness-personal/

    ReplyDelete
    Replies
    1. We're from totally different schools on that issue. I can't abide "fitness devices" and heart monitors. But, this might be a fair match. We're about the same age. Middle distance would be fine, if I get to have a choice.

      Delete
  3. I think this is an interesting contrast between the casual interpretation of economic history (often based on a hunch) that too many economic analysts pursue and the scientific interpretation which is based on a model and hard data. The first is much easier and more appealing to the masses, especially if you are an eloquent writer like Krugman. Unfortunately, it also tends to be misleading way more often than the latter.

    ReplyDelete
    Replies
    1. Most of what I've written here is well-known, I think. If you were trying to argue in favor of sooner, rather than later, liftoff, you might argue that this is consistent with pre-financial crisis Fed behavior. So, it's rather strange that Krugman would argue that postponement of liftoff is consistent with past behavior, as it's not. Typically, if you want to argue in favor of postponement of liftoff, you might argue that the long-term interest rate is low (a is low), that there should be a more aggressive response to low inflation (b is high and a is low), or that the unemployment rate is the wrong measure of the output gap, for example. I'm not saying any of those arguments have any merit, but that's how you could frame it, with the Taylor rule as a backdrop.

      Delete
  4. I think Krugman was just off by a year. In 1996 there was intense discussion of further increases in the Fed Funds target. See Laurence Meyer's book "A Term at the Fed" chapter 3,4,6. Essentially they hold back from the beginning of 1996 through mid 1997 and were very tentative with regards to further tightening thereafter. The entire debate about raising rates during that period revolved around NAIRU, very similar to today.

    ReplyDelete
    Replies
    1. "The entire debate about raising rates during that period revolved around NAIRU, very similar to today."

      Are you saying that debate made sense?

      Delete
    2. I'm saying the debate is similar in the sense that both revolved around the usefulness of the concept of NAIRU which included discussion of the degree to which low unemployment would contribute to inflation.

      Delete
    3. I can't remember all my Fed history, but this was also the age of "irrational exuberance." Greenspan said that in 1996. Also, I think this may have been the period when the Fed was trying to sort out productivity growth. Note also that the real fed funds rate is quite high by current standards - about 3% for the latter part of the 1990s.

      Delete
    4. Interesting to hear you describe this era as "irrational exuberance". Quite a few people describe the Great Moderation as a bit of a drunken party. I wonder whether you think that monetary policy was targetting the right things? Goods price inflation was low (I think this had a lot to do with exceptional events in the world economy - the entry of the Soviet Union and Eastern Bloc finally breaking the OPEC cartel, and the entry of China (a quarter of the world's population) overnight becoming the manufacturing workhouse of the world with both inflationary and deflationary impacts here as well).

      But what about assets prices? I was reading Yutaka Kosai about the origins of the Japanese bubble (Kosai is a fantastic economist but unfortunately only writes in Japanese) and he describes how financial deregulation and the formation of a secondary bond market and pressures for loose monetary policy from the US during the Plaza Accord to close the bilateral imbalance led to a huge amount of liquidity chasing few good investment opportunities, resulting in overblown real estate prices. Is monetary policy that revolves around an inflation and an operating target of overnight rates really the appropriate weapon to deal with this? And of course did representative agent models give the insights to see what was fundamentally going on?

      Delete
    5. "Interesting to hear you describe this era as "irrational exuberance"."

      No no. I'm just quoting Greenspan. Just trying to get into the thought processes of the Fed officials at the time. Some of those ideas live on. There are people who think that central banks should pay attention to "bubbles" though I'm not sure they necessarily have a clear idea what a bubble is.

      Delete
  5. But, to add to my previous point, it is impossible to define what "tight" or "lax" policy meeans without using some policy rule as a reference point. Steve shows that according to the Taylor rule the Fed's policy was in fact tight, contrary to PKs claim. So PK might be having some other rule in mind, but then what is it?

    ReplyDelete
    Replies
    1. Now yes, but what about 1995? When he writes, "They held their fire, waiting for clear signs of inflationary pressure" what does he have in mind? As you show, not the Taylor rule and certainly not R = 0.

      Delete
    2. In 1995, he's happy with what they were doing. So, my point is that he must have been happy with the Taylor rule I estimated, which is inconsistent with other things he said. Conclusion: He doesn't have his act together.

      Delete
  6. Taking these data into account, If you don't have a stable/reliable Phillops curve, and you are under inflation targeting, it does not make any sense to talk about NAIRU... I'm no economist, just a practitioner interested in economics, so, what am I missing here? Should the focus of the discussion be something else ?

    ReplyDelete