Friday, October 1, 2010

Is There a Central Bank Intervention Rosengren Doesn't Like?

I feel like I am doing a series on speeches by Fed Presidents. I don't know why, but these things never fail to interest me. Here is Eric Rosengren, speaking to a group of New York forecasters about current Fed policy issues.

Hiding in here is a good lesson for anyone who thinks that your graduate school education is an indoctrination. For example, some people seem to think that MIT (in the old days anyway) sends you off to the far left, and Chicago sends you off to the far right (again, in the old days). Eric Rosengren and I entered the PhD program at the University of Wisconsin-Madison in 1980, and we were both teaching assistants in Don Hester's undergraduate money and banking course. The funny thing is that one of the responsibilities of Don's TAs was to run a computer-simulated banking game. The students ran banks, and the TA played the role of central banker. Apparently Eric took this job very seriously, as this ended up being his chosen career. Another one of our classmates, who had shown up the previous year on campus was Jeff Lacker, now also a Fed President (Richmond). Eric and Jeff could not have more different ideas about how to run a central bank. While Eric is very much into intervention and guiding the economy, Jeff is very much into market solutions, and is highly skeptical of intervention. Obviously these guys are independent thinkers.

In his speech, Eric makes clear at the outset what he has in mind:
In short, policymaking has been effective at stabilizing the economy, but the recovery remains far too weak to restore what we consider full employment with the speed I would like to see.
Clearly, his primary focus is on the real side of the economy, not inflation (though more about that later), he thinks he knows how aggregate empoloyment should be behaving, and he thinks monetary policy can help in making it behave better than it is.

Eric is dismissive of the the notion that sectoral reallocation (which I have discussed here and here) is an important phenomenon that we should be thinking about. He says:
To me, this does not suggest that the driver is structural change in the economy increasing job mismatches – although no doubt some of that exists – but instead I see here a widespread decline in demand across most industries.
Eric's arguments are the usual ones coming from Old Keynesians, i.e. production and employment dropped across all sectors during the recession; sectoral reallocation is no more important and probably less important in the current recession than in other recessions; the currently low level of employment and high unemployment rate are due to "deficient demand." If this Econ-101 aggregate-demand-management vision of the world were correct, then the unemployment rate in Canada should be about 11.5%, instead of 8.1%, as it is now. What Eric and others are ignoring are the peculiarities of this recession related to the housing sector, which we have to trace back to the year 2000, and perhaps earlier. The incentive-problem-induced boom in housing construction from 2000 to 2006 distorted the allocation of resources in the U.S. economy, and the subsequent housing bust created a need to reverse that reallocation. Housing construction will not be recovering any time soon, and that is a key anomalous feature of the current recession - housing construction typically leads the recovery.

What is surprising about Eric's views is that, given those views, it is likely that he was one of the architects of the Fed's purchase of more than $1 trillion in mortgage-backed securities (MBS). This is a sectoral intervention if there ever was one. If sectoral reallocation is unimportant in the current recession, why did the Fed need to buy MBS, in the process reallocating credit, and resources, to the housing sector? Why didn't the Fed just buy $1 trillion in long-term Treasuries instead, which would have at least been sector-neutral, in some sense. Note further that the Fed's MBS purchases worked against the sectoral reallocation that was taking place, and that needed to take place.

Eric goes on to discuss policy options. Here, he starts by taking an approach that appears to be common currency in parts of the Federal Reserve System.
A simple Taylor Rule calculation, shown in Figure 10, relates movements in the federal funds rate to deviations in inflation and unemployment from their targets. The far right section of the chart highlights that given the very low level of inflation and the very high level of the unemployment rate, the Fed would have continued to reduce the funds rate over the past two years, but for the zero lower bound.
Figure 10 is a picture I have seen before. The idea is that you fit a Taylor rule to the data (presumably linear in this case). Out-of-sample, this Taylor rule predicts a nominal interest rate of about -6% or -7% currently. Then, the typical argument seems to be that, if the Fed were doing what it was doing before (which we all know was the "correct" policy), it is going to be constrained by the zero lower bound on nominal interest rates from doing the right thing, which justifies doing other stuff. That other stuff includes quantitative easing (QE) - purchases of long-maturity Treasuries by the Fed.

Now, implicit in this idea is that the Fed should be claiming property rights over both the unemployment rate and the inflation rate. The Taylor rule policy that it was apparently pursuing in the past is claimed (in particular by John Taylor, see this) to have been a great success, witness the "Great Moderation," for example (the period after the 81/82 recession and before the financial crisis). Of course, if the Fed has property rights over the unemployment rate and the inflation rate, and the Taylor-rule policy was such a great success, what was the financial crisis and the current recession all about? Eric, and others at the Fed, want to claim that the financial crisis was a triumph of central banking, but if we take Eric's views seriously, this episode should be considered a disaster.

I do not think that the Fed's behavior had much to do with the causes of the financial crisis. I know that some people think that low interest rates during Greenspan's tenure contributed to our financial problems, but I don't put much stock in those arguments. The danger in having the Fed claim too much credit for economic successes, particularly on the real side of the economy, and in attributing too much power to the Fed over real activity, is the property rights problem. Congress, and the public at large, come to believe that the Fed owns the real economy, and is responsible, not only for the real successes, but for the real failures as well. This is a dangerous threat to the Fed's independence.

Eric goes on to discuss quantitative easing (QE): the purchase of long-maturity Treasuries by the Fed. This is the most commonly discussed potential intervention the Fed could undertake in the near future. Eric says:
Now some concerns: While purchases of Treasury securities have the advantage of not directly “allocating credit” to a particular industry, they have the disadvantage of only indirectly affecting the private borrowing rates that more directly affect private investment spending. In addition, Treasury purchases raise for some a concern that the Fed intends to monetize the federal debt, using monetary policy to accommodate the financing of fiscal policy. I can assure you that we have no desire or intention whatsoever to do so.
So, a QE program involves changing the structure of the outstanding debt of the consolidated governnment (Fed and Treasury). The Fed is swapping outside money for long Treasuries, which is literally monetizing the debt - shortening the average maturity of the debt, and making it more liquid. How can you say you have no intention of doing something you are literally doing?

It is looking increasing likely, given the public statements of various Fed officials, that, barring some surprises in new data coming in, the FOMC will decide on some change in their strategy - likely some QE - at its next meeting. Likely the Keynesians and non-Keynesians can agree, based particularly on the inflation data, that more expansion is needed. In spite of what Eric thinks (i.e. the key risk is deflation), there is still inflation risk associated with the large quantity reserves in the system and the maturity structure of the Fed's balance sheet. This will be exacerbated by any further QE.


  1. How can you say you have no intention of doing something you are literally doing?

    I find nothing in his speech indicating an immediate intention to monetize the debt. I see only a careful consideration of the available options while fully acknowledging both the advantages and the risks of each.

    Now, implicit in this idea is that the Fed should be claiming property rights over both the unemployment rate and the inflation rate.

    I don't know what you mean by "claiming property rights", but certainly both are within the Fed's mission.

  2. 1. "How can you say you have no intention of doing something you are literally doing?"

    You misunderstood me. I know he's not committing himself to QE. The point is that QE does, in fact, amount to monetizing the debt. You can't say you have no intention of monetizing the debt when that is exactly what you are considering.

    2. The Humphrey Hawkins Act is a funny thing. There is language in there about full employment, and you'll see Fed officials dancing around that when they speak in public. They have to acknowledge it, or they'll get in trouble. However, the Act is so vague that you could drive a truck through it. The Fed could actually just be targeting inflation, and speak as if they care about unemployment. Eric wants to give Humphrey Hawkins a strong Keynesian interpretation. How you speak to the issues determines property rights.

  3. We have less new firm creation partly (as Dudley pointed out yesterday) because small business credit is normally secured by property, and since house prices are low, would-be entrepreneurs lack collateral. The lack of new firm creation shows up as widespread net job loss. Some point to these widespread losses and argue that they are proof of an AD shortfall. In reality, it ties back to the drop in home prices and housing leverage.

    If I were to ask the Fed to do one thing, it would be to tell us a narrative of how AD stimulus actually feeds through the economy, sector by sector. So far we have, "interest rates will fall by some tens of basis points if we do QE." And how would slightly lower long term rates actually put us back on "trend growth"? What are the risks to de facto deficit monetization; and to a too-lose policy for the dollar bloc (i.e. China and others)? A thorough exploration of that question would be useful.

  4. anon1:

    Yes, the collateral effect is important, and I agree on the other stuff.

  5. The mortgage debt purchased by the Fed is guaranteed by the US government. In that case, what is the difference between the Fed purchasing long term US treasury debt and mortgage debt? In your post you indicate there is a sharp difference.

  6. While the Fed was acquiring its large stock of MBS, essentially all new mortgage debt in the US was being routed through Fannie Mae and Freddie Mac, and it all ended up on the asset side of the Fed's balance sheet. Thus, the terms on which the GSEs and the Fed jointly acquired the mortgage debt determined mortgage interest rates. It's different if the Fed buys long Treasuries - the Fed would not be monopolizing that market.

  7. Rosengren's evidence that employment losses in the wake of the 2008 financial market turmoil were broad-based across sectors seems to contradict the proposition that we are now primarily experiencing a 'labor reallocation shock'. You dismiss Rosengren's evidence, but without explanation (I don't understand how you came up with the numbers in your Canadian example). It seems to me that the reallocation shock view implies employment should be dropping in housing construction and expanding elsewhere. Where's the sector that's expanding?

  8. "Where's the sector that's expanding?"

    It's Education and Health Services. Note that total services employment includes real estate agents - obviously that's related to the housing sector.

    Read these:

  9. You mean, the rise in unemployment reflects that people lost interest in housing construction, manufacturing, retail trade, wholesale trade, transportation, information, financial activities, professional & business services (these are the sectors that contracted, according to Rosengren's figure 3) and wanted instead to have more Education and Health Services? Is that what you meant to say in your post? But, if people really had become so interested in Education and Health services, then why did the share of employment in that sector remain stuck at 20% from August 2009 to August 2010? The share of employment should have increased, no? (For the data, see BLS Table B.6,

  10. 1. When you say "lost interest," it seems you think I'm saying this is all demand-driven. Actually, you can think of this as some combination of long-run shifts in preferences and technology, along with incentive problems in financial markets, and other financial factors at work.

    2. I'm looking at this picture:

  11. So, we can forget the idea that the trigger for recent events is a sectoral shock, right? Incentive problems in financial markets are what we should be thinking about. Is that it? Gertler-Kiyotaki and Gertler-Karadi are the only two papers I'm aware of that offer a model which stresses incentive problems in financial markets as a cause of the recent turmoil. Their model suggests the government ought to do direct lending to borrowers (e.g., Fed purchases of mortgages) and it also suggests that inadequacy of aggregate demand is a problem now. If you think that incentive problems in financial markets are the problem now and you do not like the Gertler analysis (which rationalizes much of what the Fed is doing), then what is your model?

  12. "So, we can forget the idea that the trigger for recent events is a sectoral shock, right?"

    No, it's all tied up together. People have only started on the labor market sectoral reallocation models. Mortensen-Pissarides embeds everything in the matching function, which does not help. Shimer has a stylized mismatch model, which is a start. In the old days, Long/Plosser, Lilien, and Rogerson, among others, thought about sectoral activity in macroeconomics. On the "financial frictions" side, the Gertler/Kiyotaki and Gertler/Kiradi models are interesting, though the contracts are ad-hoc, and those papers really are not about the incentive problems. This paper of mine is addressed to various aspects of monetary policy and financial frictions associated with the crisis:

    It does not deal with the initiating incentive problems, which I think are important. There is not a complete model out there that puts together all of the ideas that I am throwing at you. There is a lot we don't understand. I'm assuming that you are arguing about this because you are in the "deficient demand" camp, as you seem to want to call everything in sight "aggregate demand." Are sticky wages and prices at the heart of the financial crisis and the recession? If so, how do particular fiscal and monetary policy interventions work to cure the inefficiencies you see?

  13. Like many others, I'm struggling to understand what happened post 2008QIII. I would say you're right that I'm sympathetic to the sticky wage-sticky price models, because it's my impression they are the only ones that can replicate the monetary transmission mechanism laid out by Friedman in his presidential address (not to mention Hume's 1750 piece, 'On Money'): a monetary expansion generates almost no change in prices for a long time and a relatively large change in real quantities. The sticky wage story is also capable of accounting for the observation that employment comoves across consumption and investment sectors over business cycles (Benhabib-Rogerson-Wright have a simple proof that comovement is impossible in the RBC model). Finally, there's the Smets-Wouters demonstration that sticky price-sticky wage models forecast as well or better than atheoretical forecasting models.
    Still, I'm willing to abandon the model in a heart-beat when a better one comes along. Shimer's sticky wage and reallocation shock story seems interesting, though I think he needs to locate the triggering shock in the financial sector because otherwise he grossly understates the recent fall in investment. Once he does that, I don't think it would be a reallocation story anymore (which is good, because the sectoral data don't support the notion that there is a lot of sectoral reallocation going on in any business cycle, including the current one). I suspect that with this change Shimer's model will not rationalize the policy prescriptions that emerge from the sticky wage-sticky price stories, and this makes things interesting.
    I just glanced at your paper and I'm looking forward to studying it.

  14. Fisher Black also believed that sectoral shocks lay at the heart of the business cycle, as he argued in his book Exploring Business Cycles (1995). He was sympathetic to the real business cycle approach, but believed the one-good model was far too restrictive and obscured the real process. He did not believe money played any role. (Why does anonymous above believe that it is required that a theory match Friedman's postulated theory?? Bizarre.)

    If I may quote from Perry Mehrling's paper on Black:

    For Black the central issue is not the
    balance between aggregate output and aggregate consumption but rather the match between the pattern of output and the pattern of demand. Mismatches happen because investment today depends on a forecast of demand and output patterns in the future, perhaps even the distant future. People do the best they can, but they are bound to make mistakes simply because they
    cannot see the future very clearly. Even if, at the level of the economy as a whole, the real rate of interest and the price of risk are both constant (which they probably aren’t but which Black was inclined to assume as a Bayesian prior) mistakes are inevitable concerning the details. The inevitable mismatch is the fundamental source of risk for individual investments. The volatility
    of asset prices comes from the fact that the present value of real assets depends on details about the future, about which details we form expectations that shift as our views of the future shift.

    “Given the volatility of expectations, I’m surprised that asset prices aren’t more volatile thanthey are” (1995a, 59).

    Mismatch, or “uncertainty about whether we will have what we want in the future, and
    about whether we will want what we have” (1995a, 45), is also the fundamental source of risk for the economy as a whole. When the match is good, we have a boom. When the match is bad, we have a recession. Fixing a bad match means shifting the structure of production more into line with the structure of demand, which takes time and resources in an economy where capital is highly specialized and production very roundabout. By the time we discover that our past
    expectations were wrong, the investments have already been made, and it takes time for the new investments implied by the new set of expectations to begin producing to meet the new structure of demand. Cyclically persistent unemployment is the result."

    (Mehrling: Understanding Fisher Black)


  15. For the structural argument skeptic, "anonymous", above:

    The percent of job loss occurring in the non goods-producing sectors is much higher than in previous recessions. These sectors now make up the majority of losses, versus the teens earlier in the decade, and less than 10% in the '91 recession.

    What does the recent service sector experience mean? These sectors were thought to be less cyclical for two reasons: 1) service sector as a % of GDP experienced a secular boom in past several decades; and 2) the sectors was thought to have more flexible wages and less operating leverage. So why are job losses occurring in the service sector now?

    It would be interesting to hear proponents of the "AD shortfall" case explain the sudden cyclicallity of service sector employment. The "collateral shortage" effect is an explanation for the lack of small business net new firm creation, and so one structural explanation related to housing. Another is potential over-investment in consumer discretionary services--a product of the housing/equities wealth effect and its resulting pull-forward of consumption.

    If households need to save more for retirement after incorrectly forecasting future income from housing/financial asset capital gains; then is this a "structural" or "AD" related problem? To say it is "AD" related, is to say the Fed can restore confidence in those same forecasts of future income. Likely or unlikely?

  16. Here's a shot at an AD analysis. Households were looking forward to a rosy retirement and all of a sudden they learned otherwise. In an RBC-type model, the real rate of interest would suddenly drop a lot. Suppose that there is a lower bound of zero on the nominal rate of interest. This is not necessarily a problem because in principle the anticipated inflation rate can be any positive number. But, suppose that people believe what the Fed has been saying for 30 years, that they will not allow a rise in inflation (the Fed has been repeating its insistence that it will not permit a rise in inflation). In this case, anticipated inflation would not go up. So, we'd be in a situation in which the lower bound on the nominal rate of interest becomes a lower bound on the real rate of interest. What will happen? Here is one story. The cut back in consumption that occurs naturally with the lower expected future consumption results in a fall in output (investment doesn't rise to take up the slack because the real rate isn't falling enough). The fall in output, because it reduces pressure on resources, leads to a fall in marginal cost. The marginal cost would, in a flexible price model, cause an instantaneous drop in the price level. Suppose there are some sort of price setting frictions, however, so that the price level takes time to drop. This translates into a drop in expected inflation and a rise in the real rate (remember, the nominal rate cannot drop because we are at zero). The rise in the real rate makes the whole situation worse because it stimulates more saving and discourages invesment. In addition, the fall in the price level transfers resources from creative people (investors) to the others (savers). The economy falls more. Marginal costs fall more, expected deflation falls more, etc. We now have a major problem. The problem is a 'failure of aggregate demand'. You fix this problem by stimulating aggregate demand: investment tax credit, housing subsidies, government spending, commit to instituting a VAT tax starting next year, etc.
    This story does not explain the huge rise in credit spreads in fall 2008. I suspect there were in fact two shocks. The second one is a problem in the financial sector that reduced the amount of intermediation (there are several candidates for this story, but they all have the same effect on the economy). The problems with intermediation account for the fact that the fall in investment was so very massive. Again, it's an AD story.
    This AD story depends on price setting frictions (and the nature of monetary policy, which prevents people from anticipating inflation). The price setting frictions are key to the deflation which triggers the perverse rise in the real rate of interest.
    Interestingly, as prices get more flexible, the deflation spiral only gets worse (unless you go all the way and make prices perfectly flexible, as De Long and Summers pointed out in a couple of papers in the early 1980s').
    So, there's a shot at an AD story.