Williamson on the Fed, by Tim Duy: This weekend on Mark’s blog I stumbled across the link to Stephen Williamson’s piece The Fed and Inflation. Rarely does one encounter a piece with so many inconsistencies and misrepresentations that it requires an almost line by line retort. The core of Williamson’s piece begins:
While it is certainly true that there is nothing alarming on the inflation front if we look at what forecasters are predicting and the yields on TIPS, maybe we should examine some other evidence.
This sentence alone should raise red flags, as he might as well say: “The overwhelming evidence is that inflation is not a problem, but the evidence is obviously lying because it is inconsistent with my opinion.” He continues:
What's the goal here? Fed people, including Bernanke, make vague statements about a 2% inflation rate being what the Fed is shooting for. But what does that mean? Do we set the target at 2% in January and then say that we achieved our goal if the year-over-year inflation rate as of the next January falls between 1% and 3%?
I don’t find the Fed vague at all. Indeed, Federal Reserve Chairman Ben Bernanke seems quite clear. From October:
The longer-run inflation projections in the SEP indicate that FOMC participants generally judge the mandate-consistent inflation rate to be about 2 percent or a bit below. In contrast, as I noted earlier, recent readings on underlying inflation have been approximately 1 percent. Thus, in effect, inflation is running at rates that are too low relative to the levels that the Committee judges to be most consistent with the Federal Reserve's dual mandate in the longer run.
If the current inflation rate and its forecast are below the target, consider additional easing. If above, tighten. How complicated is this? The point that is arguably vague is the definition of the “longer run,” but I think it is safe to assume this is the period beyond the three year forecast horizon. More from Williamson:
If we exceed the target are we going to have a lower inflation target for a while?
Note the bias that ignores the recent period of low inflation. The question can be reversed. Since we have been under target recently, shouldn’t we be willing to accept a period of higher than target inflation?
I thought about the costs of inflation earlier, in this post, and would make the case that price level targeting might work well.
Same problem. If Williamson believes that price level targeting might work, then shouldn’t he want a period of above target inflation to return the price level to trend?
Suppose, for example, that our target price level path is 2% inflation forever. Then, intervention by the Fed which always aims to hit the target path within a relatively short period of time (say a quarter or two) should minimize the uncertainty in real interest rates over any horizon. I think real interest rate uncertainty is a key cost of variable inflation, if not the primary one. Most debt is denominated in nominal terms, default is costly, and uncertainty is costly. Just ask Paul Krugman.
Is there anyone that believes the lags in monetary policy are sufficiently short to allow the Fed to always hit an inflation target within two quarters? Anyone? We usually think of long and variable lags to monetary policy. Further on:
In the chart, you can see that headline CPI inflation is about 1% above target, core CPI inflation is about 1.5% below target, and pce deflator inflation is about on target.…Of course, what the Fed should do in this context depends on what measure of inflation it wants to focus on. The Fed, and many economists, make the case that we should focus on some core inflation measure, either core cpi, or the pce deflator with food and energy prices stripped out. The argument is either that movements in volatile prices tend to be temporary, so we should ignore them, or an appeal to New Keynesian ideas, whereby we only care about the sticky prices, which are the non-volatile ones.
No, not quite, the story is more nuanced. Policymakers focus on core inflation because it measures what Paul Krugman describes as "inflation inertia":
Now imagine an economy in which everyone is doing this [resetting prices on the basis of inflation expectations]. What this tells us is that inflation tends to be self-perpetuating, unless there’s a big excess of either supply or demand. In particular, once expectations of, say, persistent 10 percent inflation have become “embedded” in the economy, it will take a major period of slack — years of high unemployment — to get that rate down. Case in point: the extremely expensive disinflation of the early 1980s.
Now, the measurement issue: we’d like to keep track of this sort of inflation inertia, both on the upside and on the downside — because just as embedded inflation is hard to get rid of, so is embedded deflation (ask the Japanese). But in the real world, while some (many) goods behave like this, some don’t: their prices rise quickly with supply and demand changes, and don’t display inertia. So we need a measure that extracts the signal from the noise, getting at the inertial part of the story.
Consider also the actual behavior or core vs. headline inflation:
Notice in particular the relative stability of core in recent years vs. the instability of headline. Attempting to manage policy via headline inflation would have made policy even more volatile. Evidently, policymakers are clearly doing a better job managing "inflation inertia" than in the 1970s despite a decade of steadily rising commodity prices. More on that later. Williamson continues:
Bernanke, in his Congressional testimony, takes the first route:
...the most likely outcome is that the recent rise in commodity prices will lead to, at most, a temporary and relatively modest increase in U.S. consumer price inflation...
My concerns here are the following: (i) Maybe these price increases are not temporary; with the world economy coming back (problems in Japan aside) and very strong growth in some places, resources in the world are becoming increasingly scarce.
Bernanke did not say the commodity price increases were temporary, he said the impact on inflation would be temporary. See my post here. Moreover, Williamson claims that higher commodity prices are simply an artifact of an improving economy. Does this mean he was happier when the global economy was in recession and commodity prices depressed? Shouldn’t the rebound of commodity prices be seen a vindication of monetary policy? Why be so worried about current inflation yet have no concern about past deflation? On top of that, Williamson even adds that an additional problem is a supply constraint, which is something beyond the purview of the Fed and thus one of the reasons we expect the Fed to keep an eye not on headline inflation but the pass-through from higher commodity prices to core-inflation. More:
(ii) Inflation is inflation. I make the case here for looking at broader measures than core price indices.
All inflation is not the same. Spencer at Angry Bear:It is starting to look like the cyclical bottom in inflation may be at hand. But there is much confusion over current price changes. There are two types of price changes to consider. One is the change in relative prices. The other is the change in the overall price level, or inflation. Currently we are seeing large change in relative prices, but little change in the overall price level, or inflation.
Spencer’s position is not controversial. More Williamson:
(iii) Public relations: What people see is headline inflation; indeed the prices they observe most frequently are food and energy prices. If the Fed speaks to core inflation, it can end up looking devious.
All right, I am tempted to throw Williamson a bone on this one, as the recent experience of New York Federal Reserve President William Dudley speaks to the challenge of differentiating between headline and core inflation. Still it appears that Williamson’s solution is that we should just give up and pursue the wrong policy because the right policy is hard to do. This is clearly irrational, and a rational response would be to construct an alternative measure of inflation that more theoretically identifies the volatile and temporary components of inflation.
Of course, such measures exists, see Calculated Risk for the latest update on the Cleveland Fed trimmed mean CPI, which also shows no evidence that inflation is currently a problem. Williamson is ignorant of such measures. And that is not just my opinion. His responses in the comments:
I've heard about "trimmed mean PCE," but don't know what it is. How do you modify the raw PCE to get that?
Maybe I'm good at faking that I know something. The Atlanta Fed sticky price index uses some kind of trimming procedure, but it's not period-by-period, but based on the Bils-Klenow data, I think. I think you drop specific prices from the index, much like the CPI ex food and energy. Do you have a theory for the PCE trimming procedure?
What about the future? What causes inflation? Some people, including Christina Romer, think that inflation is all about Phillips curves.
Why a derisive attack on Romer? Albeit I don't know her personally, just via her research and press interviews, I just can’t see why she should be singled out for attack.
According to these people, if we get more inflation than we bargained for, that would be no problem. In their view, we have a long way to climb up the Phillips curve, a huge output gap, and plenty of spare capacity.
No, “some people” do not believe that missing inflation targets is without consequence (on either side of the target, I might add, as Williamson appears to fear only one-sided errors). “Some people” do believe, however that because of the spare capacity inflation is not currently a threat. That is not the same thing as saying higher than anticipated inflation in "no problem." Williamson then brings out the 1970s to dismiss “some people”:
Of course Christina and her fellow Phillips-curve types seem to want to ignore the 1970s.
When did it become an insult to be compared to or grouped with Christina Romer? What exactly did I miss? Slogging forward:
The mid-1970s US economy was of course not identical to what we see today, but some of the same factors were at play. Recently there has been a runup in commodity prices, real GDP and employment have taken a hit, and there is an accommodative Fed working in the background, all of which were also true of the mid-70s…
It is not the similarities that are important, but the differences. And the key difference is the inability of the labor force to translate higher prices into higher wages. A wage-price spiral cannot exist on increasing prices alone. A look at the record:
Apparently, we don’t have much to worry about until wage gains put more substantial upward pressure on unit labor costs. This is key element of propagating inflation expectations. The Fed very much needs to keep an eye on actual wage pressures, not hypothetical inflation pressures.
Then we get to the money issue. You knew it was coming.
What about the stock of currency? That has been growing substantially as well, though what we are seeing may or may not alarm you…
…you can see that the stock of currency has been growing at an increasing rate, indeed at an annual rate of 10.6% since the inception of the QE2 program. If sustained, and assuming constant velocity and 4% GDP growth, we would get something approaching 7% inflation, which I think would start to make people unhappy.
Notice that Williamson expresses concern about the growth of the stock of currency, but presents the chart in levels. Let’s look at the historical record on currency growth:
Exactly where is the reason for concern? Move along folks, nothing to see here. Williamson feels a need to create a problem where none exists. Of course, he ignores data that reasonably should raise eyebrows. A money multiplier less than one, for example:
Or declining velocity in the face of massive monetary easing, a combination consistent with a liquidity trap scenario:
I will leave the criticisms of using simple versions of the quantity theory of demand to Brad Delong and Jim Hamilton.
Just hold tight, it's almost over:
Ben Bernanke says not to worry, though…So, what could go wrong? Plenty...
All of the necessary conditions are there for a substantial inflation. I'm thinking of something on the order of 5% to 10%, and once it gets going it will be costly to stop it. The total stock of reserves will rise to about $1.6 trillion by the end of June, and that represents an accident waiting to happen.
All of the necessary conditions are not met – note lack of wage pressures above.
Right now, as Bernanke points out, financial markets are taking the Fed at its word. The margin between TIPS yields and nominal Treasury yields is not very large, reflecting modest anticipated inflation. But if people start to anticipate that the Fed will not be reducing the size of its balance sheet any time soon, and start to anticipate higher inflation, then nominal rates of return on assets will rise, making reserves undesirable to hold.
People already anticipate the Fed is not poised to reduce the balance sheet – the general consensus is that the Fed will not consider a rate hike until early next year, and the rate hike seems likely to come before an effort to whittle away at the balance sheet – yet 10 year Treasury rates remain well below 4%.
The Fed can make reserves more desirable to hold by increasing the interest rate on reserves, thus cutting off the incipient inflation, but it will have a hard time doing that. Unemployment may still be high and employment growth slow, and the Fed may not want to be blamed for slowing the recovery.
Here Williamson is making an argument that much of the increase in unemployment is structural. Despite another study suggesting the opposite, I concede there is a reasonable chance that some of the unemployment increase is structural (see David Altig here and here). What I don't agree with is the assumption the Fed will unwittingly miss the shift. Of course, they won't if they are watching wages, which should start to rise earlier than anticipated in the event of a significantly higher structural rate of unemployment.
Further, the Fed has locked itself into a portfolio of long-maturity assets which will drop in value if short-term interest rates go up substantially. Ultimately, the Fed may not have the stomach to fight the inflation, in which case the higher anticipated inflation becomes self-fulfilling.
The Fed has not locked itself into a portfolio; the Fed has carefully considered how to reduce that portfolio when necessary. And it has always been the case that if monetary authorities do not act in a manner consistent with price stability, inflationary expectations and actual inflation will move higher. in another time, this could have occurred in the absence of massive balance sheet expansion. Also note the argument works in reverse - if the Fed did not act to fight deflation, it would become self-fulfilling. Basically, Williamson argues that we should deliberately pursue the wrong policy today because we might pursue the wrong policy tomorrow.
Bernanke makes it sound like the Fed has a lot of tools. Surely with so many tools, things have to work, right? However, there are really only two instruments that matter in these circumstances: the interest rate on reserves, and the quantity of assets in the Fed's portfolio.
The tools are intended to ensure an orderly withdrawal of the monetary expansion should it become necessary, and those tools specifically relate to managing the size of the portfolio should interest rates paid on reserve alone prove insufficient. Fed staff have clearly spent substantial time and effort understanding how to manage the balance sheet (including its duration), yet Williamson simply dismisses that effort without explanation, without any reason to believe it won’t work.
If all hell breaks loose, there could be conflict on the FOMC. With a positive stock of excess reserves in the system, it is the interest rate on reserves that matters. The fed funds rate target is irrelevant. But the Board of Governors sets the interest rate on reserves. What if the Board and the regional Fed Presidents who vote on the FOMC disagree? What happens then?
A nonsensical hypothetical war within the Fed. The FOMC is comprised of seven board members plus 5 regional presidents, one of which is always the New York Fed president. The latter can be assumed to vote with the Board, in which case it is virtually impossible for the will of the Board of Governors to be different from that of the FOMC. Simply put, the interest rate on reserves will be managed in such a way as to be consistent with the policy stance of the FOMC – Williamson has nothing here but an irrelevant straw man. Moreover, a few years ago it sure looked like all hell was breaking loose, and the Fed managed to minimize conflict. Why so little faith that same will not be true in the future?
Ben Bernanke wants you to think that everything is OK. Policy is proceeding in a normal fashion. We understand what is going on. I think the truth is that he is making it up as he goes along. The Fed has undertaken a risky experiment. I can see ways in which this will not turn out OK. Maybe I'm wrong, and that would be good.
I am not sure that Bernanke believes in any way, shape, or form that everything is OK. I am pretty sure the opposite is true.
Bottom Line: This is uncharted territory for the central bank. Could mistakes be made? Yes – but I would prefer that possibility above of the certainty of ongoing recession in the absence of correct policy today. And don't fall into the trap that some hyperinflation scenario will suddenly sneak up on us. The signs will be evident long in advance, notably sharply rising unit labor costs and nominal interest rates. Those conditions simply have not been yet, leaving no reason to believe we need to rush into policy reversal, either monetary or fiscal.