Saturday, September 11, 2010

Monetary Policy Issues

A key policy question facing us is: Should the Fed, based either on concerns about the inflation rate, the level of real aggregate economic activity, or both, engage in a more accommodating policy? If the answer to that question is yes, the next question is: Does such a policy exist and, if so, what is it?

Two days ago, the Wall Street Journal published a "symposium," titled "What Should the Federal Reserve Do Next," with short pieces by John Taylor, Richard Fisher (Dallas Fed President), Frederic Mishkin, Ronald McKinnon, Vincent Reinhart, and Allan Meltzer. The WSJ picked a group of conservative economists with a considerable amount of accumulated policy experience among them, and including one sitting Federal Reserve Bank President (Fisher). One would think we could get something useful out of these guys. Well, apparently not.

Let's start with the low point. Fisher should win the bad analogy contest with this:
One might assume that with more than $1 trillion in excess bank reserves and significant amounts of cash held by businesses, the gas tank of those who have the capacity to hire is reasonably full. One might also conclude that the Fed, having cut the cost of interbank overnight lending to near zero and used quantitative easing to coax the entire yield curve downward, has driven the cost of gas to virtually nil for businesses that are creditworthy. And yet businesses still aren't hiring.
So, the gas is in the tank, the Fed has done all it can by making the cost of gas zero. So why won't the car go? Fisher says:
If businesses are more certain about future policy, they'll release the liquidity they're now hoarding.
He's talking about fiscal policy:
Fiscal and regulatory authorities share significant responsibility for incentivizing economic behavior through taxes, spending and rule making.
So, apparently the person driving the car, which is full of cheap gas, is paralyzed with fear - he or she might get stopped at the toll both, have to obey speed limits, etc. If Fisher is worried about policy uncertainty, he should probably clean his own house first (to use another analogy). What does the Fed intend to do with the more than $1 trillion in mortgage-backed securities (MBS) on its balance sheet. Will it hold those forever? Will they be sold off slowly? If so, when, and at what rate? What's with that "extended period" language in the FOMC policy statement? How long is that period? How do we know when it is time for the Fed to tighten? When the time comes to tighten, how does the Fed intend to do it - raise the interest rate on reserves, sell Treasuries, sell MBS?

There is some talk among the WSJ Symposium guys about Taylor rules. The funny part of this is having Taylor discuss his rule, which of course he makes reference to as the "Taylor rule." Taylor warms up with this:
To establish Fed policy going forward, the best place to start is to consider what has worked in the past.
That sounds great. Here we are in an unprecedented situation. On the liabilites side, the Fed is holding an enormous quantity of excess reserves, which it did not have in the past, and those reserves are earning interest, like they never did in the past. The nominal interest rate has been close to zero since late 2008, as has never been the case in past. On the asset side of the Fed's balance sheet, the Fed is holding in excess of $1 trillion in essentially private assets (MBS - see above, again uprecedented), and its Treasury holdings are mainly long-maturity, which again is a novelty. Taylor has a lot of explaining to do to get me to understand how what appears to have worked in the past is going to help here.

Taylor's argument is summarized here:
Get back to the rule-based policy that was working before the crisis. To get there without causing more market disruption, announce and follow a clear exit rule, in which the Fed's bloated balance sheet is gradually pared back by predictable amounts as the economic recovery picks up.
First, it's not clear what "working" means. Taylor seems to want to tell us that periods where the Taylor rule fit actual Fed behavior in the past were good times, and periods where it did not fit were bad. We could certainly debate that. If you tell me the Taylor rule worked well, my question is "relative to what?" To give Taylor some credit, he's asking the Fed to map out an exit strategy, seemingly for reasons that I discussed above in connection with Fisher's comments. However, now he's discussing things in the context of the quantities (the "bloated balance sheet"), which is inconsistent with the Taylor rule concept (what "worked in the past") which ignores the quantities on the Fed balance sheet.

Allan Meltzer discusses the Taylor rule as well:
Adopting and following a rule, like the Taylor rule, is an effective way to regain independence [for the Fed].
It's not clear exactly what he has in mind here. "Adopting" could mean that you write the Taylor rule into the Federal Reserve Act, or I suppose it could be an amendment to the Constitution (I know little about the law, maybe someone can help me out here). In any case, he seems to have in mind that the Fed be explicit about the rule, agree with the branches of government on what it is, and follow it. Then, for example, if there is a public outcry for the Fed to buy more MBS to stimulate the housing market, the Fed can say: "sorry, can't do that, my rule says no."

The Taylor rule specifies how the target fed funds rate should be determined in response to the deviation of the current inflation rate from its target, and in response to the "output gap," which is the difference between some measure of aggregate real economic activity, and what that measure "should" be. What is the argument against the Taylor rule as a formal rule for the Fed to follow?

1. The role of the output gap in the Taylor rule is problematic. First, we could never get agreement among economists on how we should measure this object. At one extreme is the Old Keynesian view that essentially all of the deviation of aggregate economic activity from trend is inefficient, and so that is the output gap. At the other extreme is a Prescott view of the world, under which monetary policy is close to irrelevant for the output gap, and reducing the gap is a matter of getting rid of all of the inefficient fiscal interventions. Some people measure output gaps in terms of real GDP, others in terms of the unemployment rate. Second, even if we could agree on what the output gap is, we would not be able to measure it well in real time, or communicate that information well to the public.

2. There are several things that are treated as invariant in the Taylor rule, which may not be in practice. First, the rule is typically written with the nominal fed funds rate as the relevant policy target. However, the currently relevant policy rate is the interest rate on reserves which, with a positive quantity of excess reserves in the system, determines short-term interest rates. Currently, the fed funds target is irrelevant. There is an important governance issue as well. The fiction is that the FOMC determines monetary policy, but it is the Board of Governors that determines the interest rate on reserves. Second, there is no good reason to think that the long-run real interest rate, or the optimal long-run inflation rate, should be constants. The savings behavior of Chinese households may matter for the world real interest rate, for example, and there are good reasons to think that the optimal inflation rate can fluctuate over time.

However, there is a special case of the Taylor rule that is in fact in use by other central banks - the Bank of Canada, for example - and that is an explicit inflation target. This ignores the output gap as a criterion for setting monetary policy, and makes no mention of how the instruments available to the central bank are to be used in achieving the goal. Bernanke is on record as supporting inflation targeting in the past, and the question is why this does not appear to be part of the current agenda. Is the Fed too distracted by unconventional monetary policy, is it too difficult to negotiate this with Congress, or what?

Now, suppose that we accept that the optimal inflation rate is a constant, 2%, as the Fed seems to believe. For me, it's hard to make the case that 2% is better than -1%, 0%, or 4%, but let's go with 2% anyway. Now, what should be the Fed's measure of the price level? Many central banks focus on "core" measures of inflation. I think that's nonsense. The idea is that we should ignore volatile prices when we think about inflation targeting, which seems akin to ignoring investment and consumer durables expenditures during recessions. Some people draw distinctions between prices that are "sticky" and those that are not, which seems like a related, and equally bad, idea. Since the costs of inflation are related to the fact that we write contracts in nominal terms, which makes inflation uncertainty bad, it seems we should aim for predictability in the rate of change in the broadest possible measure of the price level, which for me is the implicit GDP price deflator. Now, by that measure (see the chart), we are well below the target. Year-over-year, the inflation rate is below 1% (though increasing).

What should the Fed do about an inflation rate that appears to be too low? Possibly it has already done the trick. The last FOMC statement specified that the Fed would essentially hold the size of its balance sheet constant as MBS run off through prepayments on the underlying mortgages. This is perhaps a bigger policy move than it seems. MBS are private sector assets backing a portion of the stock of outside money. Tom Sargent might think of these as "real bills" which under some conditions we could think of as having no inflationary consequences. For example, if the Fed bought $100 billion in MBS today, with the promise that they would sell them one year from now, thus retiring the outside money that was issued to finance the original purchases, that could have no consequences for prices. Interventions like this were studied extensively, both in theory and empirically, in the 1980s. You see some of this in Sargent and Wallace's "Unpleasant Monetarist Arithmetic" paper, in Sargent's "Four Big Inflations" paper, and in Bruce Smith's work on colonial money. Now, once the Fed makes a promise to replace MBS as they run off with Treasury securities, they are promising that the outside money issued to finance the MBS will be held by the private sector indefinitely, and that then has consequences for inflation. Possibly it increases the current inflation rate enough to get us up to 2%.

Suppose that the announced change in policy does not work. The inflation rate stays below 2%. Maybe it even falls - maybe into the negative range that Paul Krugman seems to be terrified of. What can the Fed do then? Some people, including Krugman, argue that we're trapped and we turn into Japan. I don't think so. If the Fed buys Treasury securities at a rate sufficient to induce growth in the size of its balance sheet, I think this has to ultimately induce more inflation in the present. Why? Given opportunities for lending, and positive yields on long-term Treasuries, the willingness of banks to hold reserves is not unlimited. But outside money can also be held as currency. However, in turn, the willingness of the public to hold currency is not unlimited. For reasons of safekeeping, people are not going to put it in the mattress. It seems that, if the Fed floods the system with enough outside money, they can create all the inflation they might want. I don't think it matters whether they Fed buys T-bills or T-bonds in order to do this. Indeed, buying T-bills may reduce the yields on long-maturity bonds in the circumstances we now find ourselves in, simply by altering the liquidity premia in Treasury yields.


  1. I thought this post was a good one, and basically right. Yet unlike all the other posts it gets no comments (perhaps because of that?). So I'm just going to leave a comment just to say that.

  2. Thanks Nick. I was wondering about that too. Maybe this is entirely uncontroversial, or something everyone knows?

  3. because his fans have no interest in Economics at all unless you find somethings wrong in PK again

  4. Stephen: it was less controversial, but I wouldn't say it's something everybody knows. And it was well and clearly argued. (I had some minor disagreements, but basically agreed with the general thrust.)

    If we write a good post, that is clear, with a well-reasoned argument, there is often little that a commenter can add.

    A controversial post, with unclear arguments, can sometimes get way more comments.

    Number of comments doesn't always correlate well with the perceived (by the commenters) quality of the post. The sign can go either way. Or even how much it changes people's minds (whether they learn anything from it). That's true for all of us.

  5. great post. title could be more sexy. Steve might have added a bit more about the uncertainty of fed motives. sexy title: Why should we trust a captured regulator?

  6. There was a lot to chew on in this piece, and it was one of the better ones I have come across.

    The part on the "limited" bank appetite for reserves sparked some questions. Perhaps you could clarify them.

    It seems there are "opportunity cost" and "signaling" explanations for why QE should boost credit growth.

    The first holds that banks will lend because the opportunity cost of holding Excess Reserves is positive. The explanation is plausible but debatable (is bank lending reserve or capital constrained?). More important, it does not describe recent system behavior: it has accumulated QE as ER's despite the prevalence of historically high lending spreads.

    The recent behavior of banks in reaction to QE leads to the second explanation--signaling. The Fed effectively demonstrated that the response of credit growth to QE is unknown: it was zero for the first trillion, and you argue at some incremental trillion it will be positive. Therefore, we could conclude that this function is both unknown and non-linear. Therefore an open-ended QE promise on the part of the Fed would have two effects: 1) it would increase the range of estimates for future inflation; and 2) it would render the probability distribution asymmetric to higher inflation. In short, it would signal that the Fed is willing willing to risk higher and more volatile inflation in the future to fight deflation today, and this signaling would help the Fed achieve its goal of sparking velocity. The problem with this view of "signaling" is that it does not fit well with the argument that the Fed can maintain and meet a stable inflation target. If the Fed knew exactly how much QE is necessary to get to 2%, then it wouldn't be signaling that it accepts the risk of higher, more volatile inflation. Absent that signaling, how would QE have any effect?

  7. I don't see why the signal is creditable? it goes back to the literature of cheap-talk in last decade

  8. 2nd to last anonymous:

    I think it matters what the "QE" is. Is it purchases of MBS or Treasuries? If the first, under some circumstances it could be irrelevant for prices. With the second, maybe part of your point is that we have no idea what the quantitative effect is, which is correct. It's hard to make credible and well-understood policy commitments if the policymaker does not have a clue.