Thursday, July 29, 2010

Bullard and Deflation

The St. Louis Fed President, Jim Bullard, released a paper today that he has written on the possibility of a Japanese-style deflation occurring in the US in the near future. This is mainly a survey of a body of macroeconomic academic literature, and an interpretation of that literature in terms of the current monetary policy debate. This literature is fairly technical, even for your average macro nerd, so I'll try to give you a brief summary.

The departure point for Bullard's paper is published work by Benhabib et. al. on "The Perils of Taylor Rules" from 2001. A Taylor rule (discussed here) is a policy rule for a central bank that dictates how the central bank's interest rate target should be set in response to the inflation rate and the "output gap" (the deviation of real GDP from its desired level). Benhabib et. al. argued that, if a central bank follows a Taylor rule, then it is possible that the economy ends up in a permanent state with deflation and a zero nominal interest rate.

Why should we be concerned about this? As Bullard argues, much macreconomic analysis and policy discussion is carried on in the context of Taylor rules. Further, the Fed's actual behavior seems to conform to a Taylor rule. In the current context, according to Bullard, the US economy is getting dangerously close to the a state where the nominal interest rate is zero for an extended extended period, and could slip into an extended period of deflation. Why is this bad? Bullard says:
Perhaps the most important consideration is that in the unintended steady state the policymaker loses all ability to respond to incoming shocks by adjusting interest rates— ordinary stabilization policy is lost, and possibly for quite a long time. In addition, the conventional wisdom is that Japan has suffered through a “lost decade”partially attributable to the fact that the economy has been stuck in the deflationary, low nominal interest rate steady state illustrated in Figure 1.
So the idea is that we don't want to be in the zero-nominal-interest-rate world, as then the Fed has nothing to do, and since Japan suffered a lost decade, in part associated with some deflation, we want to avoid looking like that, even if we don't understand what deflation had to do with the decade getting lost.

Bullard suggests two remedies for the problem he lays out, which are (i) The FOMC should tone down this language:
The Committee will maintain the target range for the federal funds rate at 0 to 1/4 percent and continues to anticipate that economic conditions, including low rates of resource utilization, subdued inflation trends, and stable inflation expectations, are likely to warrant exceptionally low levels of the federal funds rate for an extended period.
The suggestion seems to be that there should be a commitment to raise rates sometime in the future. (ii) Bullard says:
Under current policy in the U.S., the reaction to a negative shock is perceived to be a promise to stay low for longer, which may be counterproductive because it may encourage a permanent, low nominal interest rate outcome. A better policy response to a negative shock is to expand the quantitative easing program through the purchase of Treasury securities.

My comments are as follows:

Taylor Rules
Bullard says:
Active Taylor-type rules are so commonplace in present day monetary policy discussions that they have ceased to be controversial.
I'm not sure this is true or not. In any case, Taylor rules should be controversial. Take the 2001 Benhabib et. al. paper for example. The authors basically drop a Taylor-rule central bank into two different example economies. The first one is a Sidrauski-type money-in-the-utility function economy with a fixed endowment of goods. We know exactly what an optimal policy is in this economy - it's a Friedman rule with a zero nominal interest rate and deflation at the rate of time preference. There is nothing bad about the "bad" steady state here - it's the "good" steady state that is actually suboptimal. The other example is a sticky price economy with monopolistic competition. In this economy an optimal monetary policy involves having a constant inflation rate that trades off three distortions: the sticky price distortion, the Friedman rule (positive nominal interest rate) distortion, and the monopolistic competition distortion. In neither example is the Taylor rule an optimal policy rule that the central bank would choose to follow.

Any discussion of policy choices by a central bank needs to be conducted in models where the policy rules chosen by the policymakers make sense in terms of the information they have, their constraints, and their objectives.

Deflation and Japan
Most of what I have read recently on the potential for deflation in the United States currently and how this relates to the Japanese "lost decade" goes as follows. Problems in Japan started with a collapse in asset prices and an ensuing financial crisis. There was also a deflation and very low growth in GDP. In the US we have seen the collapse in asset prices, and the financial crisis. If we allow deflation, then surely we will have a lost decade of weak GDP growth.

I have not done research on Japan's lost decade, but what I know about it tells me that Japanese experience in the 1990s does not look much like recent US experience. The lost decade in Japan seems to have been about repeated financial crises due to an inability of policymakers and regulators to deal with problems in the financial industry. While the US potentially has future financial problems lurking (Fannie Mae, Freddie Mac, and unresolved moral hazard problems), problems in our financial industry were, for the most part, dealt with swiftly.

Someone will have to be more explicit about what caused the deflation in Japan, how that was connected to poor macroeconomic performance, and what that has to do with our current predicament in the United States. I'm not convinced.

Quantitative Easing
Bullard tells us that if we get a "negative shock" in the future, the Fed should buy long-term Treasury securities, as he feels this is effective, both in reducing long-term bond yields, and in increasing the inflation rate and inflationary expectations. He cites evidence that quantitative easing in the US and the UK had these effects. My response to this is a series of questions:

1. What negative shocks does Jim have in mind? Lower-than-expected inflation? A low GDP report? More bad news on sovereign debt in Europe? Is the purchase of long-term Treasuries the solution to every "bad" shock on the horizon? Why?

2. Does this mean the Fed is now interested in fine-tuning?

3. What theory explains the potential ability of the Fed to manipulate the term structure of interest rates by swapping interest-bearing reserves for long-term Treasuries? I need some structural empirical evidence - not just correlations.

4. If the Fed has the ability to lower long-term interest rates, and this is a good thing, why does the Treasury issue long-maturity securities if this only serves to crowd out private investment?

5. The Fed can get more inflation (though not much more) by reducing the interest rate on reserves to zero. If the Fed indeed wants more inflation, why doesn't it do that?


  1. "The Fed can get more inflation (though not much more) by reducing the interest rate on reserves to zero."

    You weren't confident that this would release the whole about one trillion of excess reserves into the economy. I'm still not sure why. Why would banks leave one trillion in excess reserves paying zero interest when they could put all of that money in the government bills money market and get positive interest(convenience/liquidity is one incentive, but you discounted that)? Even a tiny bit of interest on a trillion dollars is a lot of money . Plus, historically banks never carried any excess reserves when the interest on them was zero.

    So, I don't see this. But if it were true why not go to negative interest like Sumner suggests to get the entire one trillion disgorged into the economy? The fiscal stimulus was only 800 billion and it had a big effect, so it seems one trillion of reserves entering the economy should have a sizable effect.

  2. I'm not confident it will release the whole $1T into the economy either, Richard.

    In recent (21st century, that is)history, banks have carried $1-2B in Excess Reserves. The asymmetrical nature of a calculation error leaves it best to err on the side of holding a bit of an excess.

    The other reason to hold a bit of an excess now is that there are at least $550B--I think there are more--in QE trades that will be unwound at some point. Right now, no one is pushing for them to be reversed: the Fed wants to get the value the banks claimed the paper had/has, and the banks both know better than to pay that and don't want to remark their books at the fair, as opposed to "market," price.

    The $800B was only the tip of the stimulus--it's what you have if you ignore QE in its various forms.

    Don't get me wrong--paying interest on Excess Reserves in an environment where r=0 is insane, and it should be stopped. But I'll give you odds that half to two-thirds of that $1T suddenly finds its way into other non-circulating areas.

  3. Regarding 2, what fine tuning? The Fed has an inflation target (semi-officially). When its forecasts suggest that current policy will hit significantly below target, then it is appropriate for the Fed to take actions that it expects to increase the inflation rate. Bond purchases are (arguably) such an option. That's coarse tuning, not fine tuning. What should guide monetary policy if not a target?

    Regarding 3, the theory seems pretty obvious to me. All you need are heterogeneous investors -- for example, investors with different degrees of risk aversion. Some investors are on the margin at the current long-term interest rate, just barely willing to accept the additional duration risk. When the Fed buys, those investors sell and move into shorter-term securities. Other investors who are less risk averse are willing to accept a lower long-term interest rate, so they continue holding bonds, and the yield goes down. (Alternatively, you could model investors with different time horizons, some of whom prefer longer maturities; but it doesn't really matter: almost any kind of heterogeneity will give the same result, that the demand curve for bonds slopes downward.)

    The evidence is that, on occasions when the Fed has announced asset purchases, yields on longer-term bonds have fallen. You might argue that deflationary or low-growth expectations cause both the asset purchases and the yield declines, but that argument ignores the lag in the Fed's response. By the time the Fed announces an asset purchase, the expectations that led the Fed to make that decision should already be reflected in market yields. Perhaps, if you think the Fed has better information than the market, you could argue that the asset purchase announcements are a signal of deflation or low growth, but I don't find that story very plausible. By far the most plausible explanation for the announcement effects is that the demand curve for bonds slopes downward.

    Regarding 4, I am equally puzzled (and have said so on several occasions). Considering that short-term T-bills are now a near-perfect substitute for money, the Treasury could be conducting monetary policy on its own by issuing more (or less) of them in place of longer-term securities. I don't know why the Treasury should pretend that this is not the case.

  4. I made the same point as Andy Harless did in 3, here:

    I wrote:

    ...The Fed tries to buy down 10 year treasury rates (say currently at 3%), by purchasing lots of 10 year treasuries. So their rate (yield) then starts to drop. So, some people start saying, hey, it's not a good deal anymore, so I'll "arbitrage" by selling 10 year treasuries, and instead engaging in a plan to buy 3-month treasuries for the next 40 three month periods...

    ...yes, some people will engage in the above "arbitrage", but others will say it's not worth the risk. Even if the rate on ten-year treasuries drops a half point to 2.5%, it's still worth it to me to have that ten-year lock – which no other investment, combination of investments, synthetic plan, nothing else, can provide exactly, with the U.S. Government guarantee of payment.

    Thus, if the Fed wants to push the ten-year treasury rate down to 2.5% then all it has to do is buy out every investor whose reservation rate is over 2.5%. The remaining investors (and there will be some, who care enough about getting the ten year lock and know that they can't get it exactly with any substitute) will hold at an equilibrium rate of 2.5%.

  5. "Why would banks leave one trillion in excess reserves paying zero interest when they could put all of that money in the government bills money market and get positive interest(convenience/liquidity is one incentive, but you discounted that)? Even a tiny bit of interest on a trillion dollars is a lot of money . Plus, historically banks never carried any excess reserves when the interest on them was zero."

    Exactly. If the IROR goes to zero, in the attempt to substitute out of reserves into T-bills and lending on the fed funds market, the T-bill rate and fed funds rate are also driven to zero. Ultimately, though, short-term risk-free assets will have become less attractive relative to other assets, and banks will want to hold less of them, including reserves. However, given that the banks are currently willing to hold $1 trillion in reserves at 0.25%, my guess is they will hold marginally less than that at zero.

  6. And what if the Fed were to complement the drop in the IROR with a commitment to an explicit NGDP target, growing at x% per year from 2008?

  7. Two questions, if I may:
    1) sticky prices/monopolistic competition - this sounds like an NNS/NK model, so am I correct in assuming the policy prescription is inflation targeting?
    2) "problems in our financial industry were, for the most part, dealt with swiftly." - why do you say this? There is a significant amount of regulatory forbearance going on in the US financial system (both avoiding writedowns and mismarking of assets) which would suggest problems have not been dealt with, swiftly or otherwise.

  8. MW:

    1. The second example that Benhabib et. al. deal with is a model developed by Rotemberg. Optimal monetary policy in this model can be described in terms of a growth rate for the money stock. In the model, "money" is the stuff that goes in the cash-in-advance constraint. You could also describe the policy in terms of an optimal inflation rate, or in terms of a nominal interest rate - it doesn't really matter.

    2. I intended "dealt with swiftly" only in the relative sense (compared to the Japanese experience). As I said, there are plenty of unresolved problems lurking, and it's not like we did a perfect job, by any means.

  9. Thanks Stephen.

    On 5, the answer you might get from the Fed is that it would cause problems for money market funds, who are already struggling to earn returns that are greater than the fees they charge (a problem worsened by recent regulatory reform).

    On 4, the DMO's objective is "the lowest cost of financing over time". Presumably they consider the current issuance schedule consistent with that.

  10. 1. But why are we concerned with the money market funds? Is there some critical intermediation being done there that other intermediaries could not quickly replace if there are massive withdrawals from the funds and they have to liquidate?
    2. If that is the objective, it seems they could have financed at a much lower cost in the past by rolling over short-term maturity debt rather than issuing long-maturity Treasuries. The yield curve is typically (though not always) upward-sloping.

  11. On 1. I would guess money would go from money funds to bank deposits. Would banks buy the same assets that money funds do? I doubt it. See today's WSJ as it contains a piece on US money funds' holdings.

    On 2. Understood, though the obvious criticism is that creates significant rollover risk.

  12. On 2. Yes. Once the market starts to doubt a country's ability to roll over the short-term debt, it is in trouble, and the only alternative is printing money (unless you are Greece - then it's more complicated).