Monday, September 19, 2011

Pre-FOMC: Trying to look decisive when there are no decisions to make

Some people have been weighing in with some interesting commentary prior to the FOMC meeting on Tuesday/Wednesday. John Taylor thinks that the Fed's dual mandate is a bad idea, and that the Fed should focus only on price stability. Of course, this is standard practice for the ECB, and for central banks with explicit inflation targets, e.g. the Bank of Canada, the Reserve Bank of Australia, the Bank of England, and the Reserve Bank of New Zealand. Taylor cites a paper by Dan Thornton (St. Louis Fed), who looks at the language in Fed policy statements, and finds a recent inclination of the Fed to get much more specific about the second part of the dual mandate. For example, the last FOMC statement says:
Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability.
This is much stronger than the vague pre-financial crisis language about "sustainable economic growth," for example.

This trend, and public statements like those of Charles Evans, Chicago Fed President, have Paul Volcker worried. The key part of Volcker's NYT op-ed is this:
My point is not that we are on the edge today of serious inflation, which is unlikely if the Fed remains vigilant. Rather, the danger is that if, in desperation, we turn to deliberately seeking inflation to solve real problems — our economic imbalances, sluggish productivity, and excessive leverage — we would soon find that a little inflation doesn’t work. Then the instinct will be to do a little more — a seemingly temporary and “reasonable” 4 percent becomes 5, and then 6 and so on.

What we know, or should know, from the past is that once inflation becomes anticipated and ingrained — as it eventually would — then the stimulating effects are lost. Once an independent central bank does not simply tolerate a low level of inflation as consistent with “stability,” but invokes inflation as a policy, it becomes very difficult to eliminate.
Volcker, as you remember, was the Fed Chairman faced with the task of eliminating the inflation caused by the well-intentioned Fed officials of the 1970s - the Ben Bernankes and Charles Evans of their day. Thus, his opinion should carry some weight, though Paul Krugman of course does not think so.

This WSJ piece suggests that Charles Evans and Charles Plosser (Philadelphia Fed President) have been assigned the task of working out explicit dual mandate targets, as envisioned by Evans. Of course, this leans further in a bad direction. It seems particularly dangerous to be making explicit statements about targets for the unemployment rate, for example. The Fed certainly has no control over the unemployment rate in the long run, and how much influence it can have even over short periods at the best of times is debatable. Further, right now there is absolutely nothing further the Fed can do to move the unemployment rate down.

Essentially, the Fed is faced with a non-decision. Under the current circumstances, with a large positive stock of reserves, all that can matter is a change in the interest rate on reserves (IROR). The Fed will not likely move the IROR to zero, for technical reasons that have to do with money market mutual funds. It will not move the IROR up, as it is not ready to tighten yet. It has already committed to to keeping the IROR at 0.25% for close to two years in the future, and extending that period is not only unlikely, but foolish (as indeed was the commitment made at the last meeting).

What will the Fed do? I think it unlikely that they will actually buy more long-maturity Treasury bonds - i.e. embark on QE3. The most likely outcome will be to lengthen the average maturity of assets in the Fed's portfolio through swaps of short-maturity for long-maturity Treasuries. This of course will accomplish absolutely nothing. However, I think most of the FOMC is convinced that it will.


  1. Steve: "It seems particularly dangerous to be making explicit statements about targets for the unemployment rate, for example. The Fed certainly has no control over the unemployment rate in the long run,...."

    Agreed. But a nominal GDP level-path target would be feasible, as an alternative to an inflation target or a price level-path target. And it would be one way to interpret the dual mandate. And if there were short-run non-neutralities of money, it would be different from a price-level path target, and might be superior in some circumstances.

  2. Nick,

    A nominal GDP target is essentially just a specific Taylor rule. There is no good theory that tells us why this rule would perform better than any other. Further, under current circumstances it may not even be feasible to attain a particular nominal GDP target, say 5% growth per year.

  3. Steve,

    What are your thoughts on price level targeting? As you probably know, the BoC has recently been revisiting this idea.

  4. If you think that an important cost of unanticipated inflation is associated with nominal debt contracts of differing maturities, then a price level target has good properties. With a price level target, you do not treat past errors as bygones, which is important.

  5. Steve,

    Suppose that you were given absolute control of the FOMC. Steve Williamson runs the Fed. Period. What would monetary policy look like under normal circumstances? Would you have an inflation target? A price level target? Would you adopt the Friedman rule?

  6. I'm going to suppose you are going to take Congress off my back too.

    Step 1: Hire a staff that actually knows money and banking. You would think that this would be obvious for a team of economists advising central bank management, but apparently some people don't get it.

    Step 2: It might be useful to have some people working on sticky wages and prices, but it seems to me that the quantity of person hours devoted to refining NK models is too great. People should be working on the details of how credit markets work, how assets are used in transactions, alternative payments technologies, intraday and overnight central bank intervention and its effects, etc. We don't want to be too heavy-handed, as good research is produced by individuals following their instincts, but incentives need to be changed.

    3. Pay interest on all reserve accounts, including those held by the GSEs. Permit negative interest on reserve accounts. Establish a channel system, with a standard margin between the target overnight rate and the discount rate, and between the overnight rate and the interest rate on reserves.

    4. Establish a price level target. There are three key questions here: (i) What is the appropriate price level measure? (ii) What should the trend growth rate in the price level be? (iii) How quickly should the Fed correct deviations from the target price level path? On (i), I think that inflation is inflation, and this should be some broad measure, such as the raw pce deflator. On (ii), I think this should be an inflation rate above the Friedman rule rate, determined essentially by how you should tax currency transactions (and there are plenty of reasons why we should do that). On (iii), I have no idea. In general, though, (i), (ii), and (iii) are key questions that the research staff should be working on.

  7. "If you think that an important cost of unanticipated inflation is associated with nominal debt contracts of differing maturities..."

    surely this can be addressed more efficiently with inflation-linked contracts? i'd guess there is a precedent for those, most likely brazil.

  8. You've been favorably cited by a quasi^H^H^H^H^H "market monetarist" for explicitly giving money a place as medium of exchange in your work.

    I remember years ago when I was younger nodding along as Austrians waved away money for barter-like stories about the "real economy" (like Bob Murphy's sushi story). Now it seems rather silly to me.

  9. Well, that's the big question. Why is it that we tend to see widespread indexation only when there is high inflation? In the United States there is not much that is indexed, and what is indexed is typically indexed by the government. So, we could say that, if inflation uncertainty were a big deal, we would have a lot of private indexation, but we don't see it, so inflation uncertainty is not a big deal. Is that because the Fed is doing a good job, or would there be no indexation even if the Fed were doing a really bad job?

  10. Stephen,

    I love this casual recommendation of yours: "Permit negative interest on reserve accounts."

    Would you mind tracing out some of the consequences of (say) charging 1% negative interest on excess reserves?

  11. See this:

  12. Stephen,

    Glad to see you endorse level targeting. Price level targeting would certainly be an improvement over what we have now, but it is not without problems. The key one, I see, is that would do a poor job dealing with supply shocks. A NGDP level target does not have that problem.

  13. What do you mean by "dealing with supply shocks?" I have probably said this before, but NGDP targeting is operationally just a particular Taylor rule. I'm not sure what model delivers that as my optimal policy rule.

  14. Stephen,

    How sure are you that the law doesn't allow the Fed to charge negative IROR? (remember, the actual judges' interpretation of a law can be very different from how it looks literally).

    Because Alan Binder thinks it's legal:

    "How about minus 25 basis points? That may sound crazy, but central bank balances can pay negative rates of interest. It’s happened."


  15. Richard,

    No, it wouldn't fly. It would either be interpreted as a tax, which the Fed cannot levy, or a fee, and my understanding is that any fees the Fed collects are carefully defined in the law. I don't think they can do it without a change in the law.

  16. Brazil: Indexation in contracts introduced inertia, because by its very nature, it was backward looking (prices and wages today get readjusted by yesterday's inflation). This made the cost of deflation through monetary contraction very large. More than that, inflation would tend to feed on itself and accelerate.

    Essentially disinflation involved changing the indexation system to a daily basis and then adopting the new index as a new currency and unit of account. ``disindexing'' all contracts was an important part of the stabilization plan and to some extent residual inflation in the economy can still be traced to some indexed contracts (utility prices etc.) It was a beautiful piece of jiu-jitsu, but it took 15 years of super high inflation to figure out.

  17. Why would reducing the IROR help the real economy? I would guess all it would do is push banks into other short end instruments that they regard as substitutes for reserves (e.g. T-Bills).

  18. Anonymous,

    "Why would reducing the IROR help the real economy? I would guess all it would do is push banks into other short end instruments that they regard as substitutes for reserves (e.g. T-Bills)."

    I assume banks see some benefit in holding reserves rather than "other short end instruments." Yes, the marginal benefit is probably small, but if you eliminate it, banks will find it attractive to make a few more (?) loans than they would if they continued to earn interest on reserves.

  19. "I assume banks see some benefit in holding reserves rather than "other short end instruments.""

    Yesterday afternoon, the 2y note was 21bp and the IROR was 25bp, so for sure there is "some benefit"!

  20. "Why would reducing the IROR help the real economy?"

    I'm not suggesting it would. I think the Fed should focus on inflation control.

  21. "Yesterday afternoon, the 2y note was 21bp and the IROR was 25bp, so for sure there is "some benefit"!"

    No, you have it backward. Financial institutions are willing to hold other assets with lower rates of return than reserves. Reserves actually have negative benefits relative to T-bills, for example.

  22. from anon 329: I'd appreciate if you elaborate on what you mean when you write "Reserves actually have negative benefits relative to T-bills".

  23. The interest rate on reserves is 0.25%. I looked up the current interest rate on a 3-month T-bill, and that is essentially zero. Why would banks hold T-bills if they can hold reserves and earn a higher rate of return. Yet banks do in fact hold T-bills. Thus, there must be some sense in which the T-bills provide services to banks that the reserves do not. At the margin, reserves are not currently useful for anything other than sitting overnight and earning 0.25%. Some reserves are needed to make interbank transactions during the day, but that amount is very small - a tiny quantity of reserves can support an extremely large quantity of intraday transactions. But T-bills can also be exchanged in large financial transactions, more financial market participants hold them than have reserve accounts, and T-bills are very useful as collateral in overnight repo transactions.

  24. thanks. i'm aware of all that. i'm still curious as to why banks would hold t-bills rather than reserves. as i see it, the most obvious advantage of t-bills relative to reserves is that they provide liquidity on-demand, i.e. you can trade t-bills 24h/day, whereas reserves presumably have to be kept overnight and cannot be withdrawn 'til the next morning. is that correct?

  25. No, as far as I know, I can't find a place to trade a T-bill in the US in the middle of the night, so that's certainly not an advantage for a T-bill.

  26. as treasurys are traded in european and asian hours you must be looking in the wrong places. try calling your favorite primary dealer.

    back to the question at hand - why do banks hold bills rather than reserves?

    all very well to say "T-bills provide services to banks that the reserves do not", but what services are they?

    you mentioned "T-bills can also be exchanged in large financial transactions", but reserves are just cash, which can also be used in large financial transactions.

    you also mentioned that "T-bills are very useful as collateral in overnight repo transactions". sure. but why buy bills yielding 1/2bp (3mo) and then repo them at 20bp (current O/N GC)?