Friday, September 9, 2011

Evans Speaks

If you think the FOMC is confused, this speech by Charles Evans, Chicago Fed President, might tell you why.

First, Evans wants to take the Fed's dual mandate very seriously - never a good idea. The so-called Humphrey-Hawkins Act of 1978 mandates, in somewhat vague language, that the Fed foster price stability and also pursue policies that promote growth and employment. There is a problem, though. It is essentially universally-accepted among economists that a central bank can control the rate of inflation under most circumstances, though there may be particular circumstances, such as exist currently, under which that is not entirely true. It is also widely-accepted that money is neutral in the long run, and there is much disagreement about the nature and quantitative significance of short-run nonneutralities of money. Thus, any group of economists and non-economists finding themselves sitting in an FOMC meeting will have a hard time interpreting what the dual mandate dictates they should do, and they certainly will not agree on what the Humphrey-Hawkins prescription is.

The Fed has found creative ways of getting around this problem however. First, in speeches by Fed officials and in FOMC statements, lip service is paid to the dual mandate, but the Fed could actually be more-or-less ignoring the real side of the economy. Second, the Fed often uses Phillips curve language, in spite of the fact that the Phillips curve is a problematic object with a sordid history, and output gaps and unemployment rates are of demonstrably little use in forecasting inflation. Why does the Fed do this? Because this is a convenient way to get agreement among FOMC members - if real GDP growth is expected to be high (low) then the Phillips curve tells us that inflation will be high (low), and all the FOMC members vote to tighten (ease), whether they are Keynesians or not. Third, it can be convenient for Fed officials to speak in public about how a low and stable inflation rate actually fosters economic growth. By this logic, in fulfilling one part of the mandate, the Fed can fulfill both, and kill two birds with one stone. The logic is also correct, though we could argue about the quantitative effect of inflation on economic growth.

None of this namby-pamby vague central-bank-speak for Evans, though. He wants to interpret the dual mandate in terms of hard numbers. According to him, the "natural rate" of unemployment is 6%, and price stability means an inflation rate of 2%, so the bliss point for the US economy is a 6% rate of unemployment and a 2% rate of inflation, and the Fed's performance should be measured in terms of a quadratic loss function. Why? Mike Woodford told him it was OK.

What should the Fed do under the current circumstances? Evans anticipates that the inflation rate will fall below 2% and the unemployment rate is of course well above 6%, so the choice is clear for him: there should be more monetary accommodation. Implicit in this argument of course is the Phillips curve - more monetary accommodation implies more inflation and less unemployment. But how much accommodation? For Evans, this is just a matter of what coefficients go into the quadratic loss function. He clearly puts a low weight on the losses from high inflation and a high weight on the losses from high unemployment, so he is willing to bear a much higher inflation rate so as to bring unemployment down.

There are three problems here.

1. Evans is forgetting the lessons of the 1970s. What Evans is proposing is a change in the policy rule - a change in how the state of the economy maps into actions by the Fed. What economists understand today that they did not in 1975, is that commitment by the Fed to a policy rule is critical for its success in fulfilling its mandate. Once the public understands that the Fed intends to exploit a short-run Phillips curve relationship (and the problem is worse if the short-run inflation/unemployment tradeoff in fact does not exist), then all bets are off. High inflation can become well-entrenched and we have to go through an episode like the policy-induced "Volcker recession," followed by a long period where the Fed re-establishes its credibility. This is exactly the logic, I think, behind Kocherlakota's dissent at the last FOMC meeting. In the 1970s, the Fed was dominated by many well-intentioned people much like Charles Evans, and they got us into trouble.

2. What is the economic inefficiency that Evans thinks he is trying to correct? By efficient, I think we mean a state of affairs that is optimal from the point of view of a policymaker - there is nothing that a policymaker can do given that state of affairs to increase aggregate economic welfare. Evans seems convinced that the state we are in is not efficient. In this quote, he comments on "the story" that interprets our bad state of affairs as intractable, from a policymaker's point of view:
I suppose it is natural to believe that some elements of the story are true. But for me, the evidence for this is minimal, and the implications for productive capacity are exceedingly pessimistic. And even if it is true, the market mechanism should cause wages and prices to adjust in order to reemploy unused resources. For example, there should be some lower real wage that would make it profitable for firms to fund the necessary on-the-job training for workers who need some modest acquisition of skills. According to this pessimistic hypothesis, something is preventing the market’s pricing mechanism from achieving such results within a satisfactory time frame.
This seems confused. What he seems to be saying is that theories that attribute a rise in the unemployment rate to frictions associated with sectoral reallocation must rely on price or wage distortions. However, the sectoral reallocation frictions that typically come into play in these discussions involve the time and effort associated with acquiring sector-specific human capital, information frictions, and the costs of moving labor across geographical regions. One would think that wage and price distortions might be key to Evans's argument - he's clearly a hardcore Keynesian, and one would not think he would be appealing to wage and price flexibility to shoot down the alternative case.

So what is our key macroeconomic problem, from Evans's point of view?
...I think the evidence favors the belief that aggregate demand is simply much too low today.
Arrrgghhh. If all economists could take a pledge never to use the words "aggregate demand" again, the world would be a better place. What Evans is saying is that he does not know what is going on. Aggregate demand is Keynesian language. When the language is used, what it means is that there is an inefficiency that the monetary and/or fiscal authority might be able to correct. In Keynesian theory, the inefficiency can come from two sources: (i) sticky wages and prices or (ii) multiple equilibria. The specifics of the inefficiency actually matters for what the optimal policy response is. Which prices are sticky and which are not? Are the prices sticky, are the wages sticky, or both? If the problem is not sticky wages and prices but the fact that we are just in a bad equilibrium, the solution to getting to the good equilibrium might be quite different than solving the price/wage distortion problem.

Further, Evans tells us about Reinhart and Rogoff, the debt overhang, and how this prolongs our economic recovery. How does he know that the "unused resources" he is seeing are not unused because of the financial problems created by the recent crisis? Debt overhang in the economy may create conditions under which those resources will go unused, no matter what the Fed does. Did debt overhang actually go into Evans's 6% "natural rate of unemployment" calculation?

3. How can the Fed actually be more accommodative under current conditions? As I have discussed before, the Fed has only one policy instrument given the large quantity of excess reserves in the financial system: the interest rate on reserves (IROR). Quantitative easing, or changes in the maturity structure of the assets on the Fed's balance sheet will accomplish nothing. Thus, to be more accommodative through current actions is impossible, unless the IROR goes to 0%. Bernanke told us a year ago that this was not on the table, but maybe he has changed his mind. In any case, setting the IROR at 0% will not change anything much. However, the Fed can change its statements about the future path of the IROR, and that can matter.

Evans suggest three types of "forward guidance," that the Fed could contemplate. The first is a policy rule explicitly contingent on the unemployment rate. The second is contingent price-level targeting, and the third is nominal GDP targeting. The first policy is quite ill-advised, partly for reasons discussed above, but in particular because the "natural rate of unemployment," whatever it is (there are many definitions) is a moving object, and it moves in unpredictable ways. The other two proposals actually do not imply anything especially new about how we formulate policy, as you ultimately have to reformulate those things in terms of a rule for the policy interest rate.

Some of what Evans contemplates would open up the possibility of a future with high and sustained inflation. Evans should think carefully about which he prefers - some heat from Krugmaniacs and the unemployed about unused resources, or a lot of heat from everyone about the high rate of inflation.


  1. Quasi-monetarists seem to be in agreement with Evans. They might argue that "high and sustained inflation" is impossible under an NGDP or price level target: the Fed would simply tighten to prevent it. Kocherlakota reminded us in his speech on Monday that tightening under a large output gap can be quite painful. He envisioned a situation in which "temporary" high inflation leads to an un-anchoring of l.t. inflation expectations, which leads to accelerating inflation, forced tightening, and even higher unemployment than initially. The potential benefits of stimulus seem to get a lot of attention; this risk seems to be seldom discussed.

  2. Last year you mentioned the possibility of negative rates on reserves. Maybe not a good idea, but it is possible. (although, as you said, the Federal Reserve Act might need to be changed)

  3. "Quasi-monetarists seem to be in agreement with Evans. They might argue that "high and sustained inflation" is impossible under an NGDP or price level target: the Fed would simply tighten to prevent it."

    True. However, a problem with what I think the quasi-monetarists are saying is that they seem to think that announcing, say, an NGDP target can increase the inflation rate for sure. I don't think that's correct. It's possible that you can't operationally guarantee a higher inflation rate, given where we are now, i.e. you may not be able to hit the NGDP target from below, even if you wanted to.

  4. David,

    Exactly. You would need Congress to act on that.

  5. 1. Common: Aggregate demand is a well defined object in all modern DSGE models. It referes to -- simply put -- the number of goods that are bougt in the aggregate from firms. Take Christiano, Eichenbaum, and Evans, for example. It is a well defined object there.
    2. When the economcy is subject to a temporary shock -- so the zero bound is binding -- there is a well defined temporary tradeoff between inflation and output. The 1970s literature is about there is no permanent tradeoff, not about how you should react to shocks.
    Read the literature, for christ sake, perhaps starting with Christiano, Eichenbaum and Evans, and then you could even veture into Christiano, Eichenbaum and Rebello, where the zero bound is taken into account, -- both papers are in the JPE which I'm sure you can access on you computer ---and I can assure you that i) there is a well defined aggregate demadn and ii) there is a well defined temporary tradeoff which the Fed is legally mandated to take into account (at least if you believe this model, and say, the last 20 years of macro-research).

  6. The state of "aggregate demand" does not describe the inefficiency. Why do you even want to use that language, for Christ's sake? Then you get a load of nonsense coming from people who think that "insufficient aggregate demand" is something you can see by looking out the window.

    "When the economcy is subject to a temporary shock -- so the zero bound is binding -- there is a well defined temporary tradeoff between inflation and output."

    How can there be a tradeoff at the zero lower bound? At that point monetary policy doesn't work in your model, right?

    " least if you believe this model..."

    Maybe not. NK is not the only development in the last 20 years.

  7. Stephen,

    How can you say Evans is forgetting the lessons of the 1970s? He is proposing level targeting, an explicit rules-based approach that actually better anchors long-term inflation than does a simple inflation target. Based on the form it took, such a rule would allow rapid catch-up inflation or ngdp growth to the level target. Thereafter it would grow the target variable at a normal pace (e.g. 2% price level growth or 5% ngdp level growth.)

    Such an approach would also imply a much lower, but conditional path of future real interest rates than is currently perceived by the market, so no it would be fundamentally different. Similarly, it would also imply a credible commitment to higher inflation until the level target was hit. As FDR showed in 1933, shaping expectations of the future path of monetary policy can pack a real economic punch.

    Charles Evans is not the only economist promoting a price level target. So is Michael Woodford. Are you going to claim he is confused too?

  8. "Such an approach would also imply a much lower, but conditional path of future real interest rates than is currently perceived by the market, so no it would be fundamentally different. Similarly, it would also imply a credible commitment to higher inflation until the level target was hit."

    1. This lowers real rates? Permanently? How does that happen?
    2. It's not a credible commitment to higher inflation, as the Fed actually has to execute asset trades and settings for the policy rate that deliver the result.

  9. Addendum to mouthy NK anonymous:

    In standard NK models, with monopolistic competition, output is indeed demand-determined, by assumption. Thus, aggregate demand and aggregate output are synonymous. So, if there is a state of insufficient aggregate demand, there is too little aggregate output. How do you find that helpful?

  10. you are trivializing nk economics. read the early literature, say blanchard and kiyotaki, ball and romer, and you find pleanty of stuff on demand externalities, strategic complementaries etc. the modern literature assumes you have read this stuff, which is clearly an unrealistic assumption, as it happens
    regards, a mouthy nk

  11. Retort to mouthy anonymous:

    Again, too little "aggregate demand" is the same as too little "aggregate supply", because supply = demand. Go back to undergrad, little boy, and leave hard questions to smarter people.

  12. Things are set up for an interesting FOMC meeting. Evans piece was well-written, but as Stephen points out, he glossed over the issue of how much extra employment he thinks he can get short and long term in exchange for a bit more inflation. But at least Evans is honest. A lot of people on the FOMC will be pushing for more accommodation without changing the inflation target. To his credit Evans explicitly acknowledges that the only way to loosen policy (ie lower real rates) at the zero lower bound is to raise inflation. From the perspective of Fed credibility it's better to have 3% inflation with an explicit 3% inflation target than to have 3% inflation with the old 2% or a bit lower target.

  13. "High inflation can become well-entrenched and we have to go through an episode like the policy-induced "Volcker recession,""

    So you're admitting that the Fed has the ability to create a major recession, the worst at that time since the Great Depression, to destroy a huge number of jobs.

    But if this is true, if they can destroy a huge number of jobs, then why can't they create a huge number of jobs? perhaps by just doing the same things in reverse.

  14. "Quantitative easing, or changes in the maturity structure of the assets on the Fed's balance sheet will accomplish nothing."

    So, QE-infinity, the theoretical one where they buy up all the assets/debt in the entire country, will have no effect, correct?


  15. "you are trivializing nk economics. read the early literature, say blanchard and kiyotaki, ball and romer, and you find pleanty of stuff on demand externalities, strategic complementaries etc. the modern literature assumes you have read this stuff, which is clearly an unrealistic assumption, as it happens"

    We could think of blanchard/kiyotaki, ball/romer, etc. as related to Woodford, in the same sense as the General Theory, Hicks, Diamond, Benhabib, might be related to Woodford. You can find some things they have in common, but a lot that is different. Woodford is what he is - the inefficiency is due to the fact that the relative prices are wrong, and a distortion from monopoly power (which is not important for the main idea - we don't think that is something monetary policy is there to correct). Woodford is not about demand externalities and strategic complementarities, which is the stuff of Bryant/Diamond/Cooper-John for example.

  16. Nick Rowe on when aggregate demand determines output, and when aggregate supply does.

    Karl Smith has been arguing that our output doesn't look bad compared to the tech boom, but our employment is much worse. I suppose the way to reconcile that is to say that with higher labor productivity any given employment rate corresponds to more output than it did before.

  17. Not much to say since I think you covered it well. But, I find it mildly amusing that when things get a little confusing, people seem to want to go back to what they first learned (namely AS-AD).

    I found the debt overhang bit interesting. He says that "debt burdens are key contributors to the headwinds I discussed earlier …," but what makes him believe monetary policy is the best way to correct debt problems? Surely fiscal policy could do a much better job? For example, his own colleague Kocherlakota once argued that the government should issue real bonds to finance a program where they pay off banks to do significant mortgage reductions. Whether or not that's a good solution, it would certainty work better than trying to use monetary policy to correct a debt overhang problem. Now, of course, the FOMC couldn't do this policy, but surely the FOMC could use their influence to try and nudge congress to a fiscal solution to debt overhang if they believe it's such a constraint on the economy.

    Also, Stephen, you talk a lot about New Keynesian frictions, I was wondering if you had seen Bai, Rios-Rull, and Storesletten's recent paper where they generate "aggregate demand shocks" via a search friction in shopping for goods and investment, rather than relying on sticky prices? I'm still trying to wrap my head around the idea of their shopping friction, but I find the idea as an interesting alternative to the typical New Keynesian friction that allows output to be "demand-driven." The model doesn't have money in it, but I think if it did higher future expected inflation could increase output, since expectations of their money being less valuable in the future should cause search effort for current consumption to rise in their model. Whether that's good for welfare or not would be a different question - of course, I'm really speculating here also.

    Here is a link if you care to take a quick look:

  18. TGGP,

    Yes, that's the big puzzle. GDP has certainly been growing more slowly than it usually does coming out of a recession, but the big story is how much employment fell, and the fact that it is not come back much relative to GDP. Very different behavior from Canada, for example, where the GDP behavior is not so different.


    Yes, exactly. If debt is the big problem, we need to think about that carefully, and ask whether some fiscal or monetary intervention might actually be able to correct it.

    I saw Victor present the paper with Bai/Storesletten. That certainly seemed interesting, though I'm not sure if there was anything in it we could apply to our current predicament.

  19. Stephen, you asked the following questions:

    1. This lowers real rates? Permanently? How does that happen?
    2. It's not a credible commitment to higher inflation, as the Fed actually has to execute asset trades and settings for the policy rate that deliver the result.

    Let me respond in terms of a price level target.

    On (1), I am not claiming real rates will be permanently lower, but that they will be lower during the period of higher-than-average "catch-up" inflation that returns the price level to trend under such level targeting. If the Fed says it is committed to doing this ala Swiss National Bank style (i.e. is willing to buy unlimited amount of securities) it will signal to the public higher expected inflation until the price level target is hit. That will temporarily lower real interest rates (below the natural rate level and provide stimulus).

    On (2), I think you are invoking the "Fed is not the only financial intermediary so it's actions are ineffective" argument here. If so, my reply is twofold. First, your argument assumes the private financial intermediaries will perfectly offset any financial intermediation by the Fed. Why assume that will necessarily happen? That seems like an awfully strong assumption. Any evidence to back it up?

    Second, this focus on financial intermediation ignores other channels through which the Fed can work. In particular, I am thinking of a portfolio adjustment-type channel where the non-bank public adjusts its portfolio of assets as (1) the Fed through its purchase of more and more long-term treasuries causes private, non-bank portfolios to be over weighted with money assets and(2)expectations of higher inflation from the price level target also affect the relative returns of holding safe money assets and cause a rebalancing. This channel would work even if the the financial intermediation story you tell were to be at play.

    Again, I ask you to consider the 1933-1936 period of unconventional monetary easing. It worked in far more dire economic circumstances than now.

  20. 1. One can argue that in a balance sheet recession a little unanticipated inflation would do more good than harm.

    2. I think it's clear he refers to wage stickiness, which is consistent with tolerating or even aiming for higher inflation. Why wait for the long run money neutrality to present itself when you can inflate wage rigidities away?.

    3. First of all, the Fed could and should introduce negative interest on reserves. Then the banks would be forced to either lend the money or to spend substantial amounts on the security personnel, equipment and storage space they would require to safely store billions of 100-dollar bills (a creative fed could even refuse to give them anything else then 1 dollar bills if they chose to redeem their deposits rather than pay interest), either of which would be monetary stimulous. We did try a negative interest rate policy in Sweden and from what I'm told the Riksbank is very happy with the results. Even if we accept that this is the fed's only tool, it seems strong enough to me.

  21. 1. Several questions here: (i) What exactly do you think a "balance sheet recession" is? Is the problem too much private debt? If so, one person's liability is another person's asset. Do we have a suboptimal distribution of wealth, and is the way to correct it then through an unanticipated inflation that redistributes wealth in just the right way? Couldn't we do a better job of redistribution with fiscal policy? Is the problem to do with costly defaults? Again, couldn't we correct the problem with fiscal policy? What about moral hazard? Don't we have to worry about that?

    2. Where is the wage stickiness? Do you, for example, think that the wages of construction workers are inefficiently high, and that is why they are unemployed?

    3. It would be nice if the Fed could do this, but the law does not permit it. The law that Congress wrote permits the Fed to pay interest on reserve accounts (and not to GSEs), but does not permit the Fed to tax a reserve account.

  22. I haven't read R&R's "This Time It's Different", which seems to be the main source of the balance sheet view. But I can come up with a reason why your (initially persuasive) net assets argument could fall short. Let's say that the more money you have the more likely you are to spend. People borrowed expecting their homes to be worth a lot, it turns out they aren't and now they owe money to their banks. They tried their own debts like a cash constraint and so they try to deleverage by spending less and slowly paying off their mortgages. The banks on the other hand think a lot of borrowers aren't going to pay off their mortgages. So the on-paper value of their assets is not regarded as real wealth until its actually paid off, and they become stingy as well.

    As for sticky wages, I would suggest all the people whose wages/benefits haven't decreased along with GDP are overpaid. I expect public employees to be overrepresented in that category.

  23. A better credit story might have to do with collateral, and there are some models of that - e.g. Kiyotaki and Moore. You need safe collateral to support borrowing and lending, the value of collateral (principally housing) has dropped. The expansion is held back by that. Until the value of real estate, in particular, goes up, the recovery will be sluggish.

  24. If there was a general price increase, which included the price of real estate, would that work? Or does the real/relative value need to increase in such models?

  25. 1. Well, I am not advocating that the Fed decide to redistribute income and increase the efficiency of wealth allocation. I'm not saying the Fed should take over the role of the US government in deciding what happens to balance sheets in the economy. My point is simply that while in the 70s expansionary monetary policy would be problematic indeed, today a little more inflation isn't. This doesn't mean the Fed should aim at higher inflation, it just means that IF such a byproduct were to present itself, it would be nowhere close to the 70s disaster.

    2. I don't have the time to read Evans' speech right now -sorry- but I'll reluctantly say yes. Contracts take time to be renegotiated and there is plenty of empirical evidence that workers (especially construction workers) don't really think of money as neutral - particularly in the short term. If a housing bubble made a construction worker earn 20 bucks rather than 15 where labor markets would clear and right now, assuming he won't go back to work for 15 for whatever reason, then inflation could simply give him a paycheck for 20 and the purchasing power of 15. Then we'd get to full employment faster. If I understand Evans correctly, this is what he's trying to say.

    3. OK, I had no idea that the Fed was banned from charging interest, but you have to admit it sounds absurd that Bank of NY can pay negative interest on deposits and the Fed can't. Maybe it's because i've lived in Sweden for too long but it just sounds insane that a private bank can screw its depositors but a public one can't.

    As for the "ballance sheet view", no, it's not really R&R. I mostly think of bank balance sheets in terms of a simple CAPM-style model where banks hold efficient portfolios of loans and where a destruction of AAA assets makes them hold cash instead of those CDOs that weren't as safe as we thought they were. They of course are assumed to try to maximize return given some risk tolerance. So lowering the real interest rate either by inflation or by introducing negative interest rates makes the banks take more risk by altering the optimal composition of the efficient portfolio so that it will include more risky assets and less cash. This is basically just Sharpe/Markowitz portfolio theory, estimated using portfolios of bonds rather than stocks, not a model based on some kind of historical assumptions on who borrowed how much.

  26. Wonks,

    If it's the value of collateral, you want a relative price increase. Another type of effect, though, would have to do with the deadweight losses from default - i.e. default is costly in terms of seizing collateral, or sorting out the losses among multiple creditors. With debt contracts written in nominal terms, unanticipated inflation decreases defaults and therefore saves resources. The cost of an intervention of that type - inflation to effectively forgive debts - is that people start to anticipate this as part of government behavior. Lenders and borrowers start to take this ex post intervention into account when they write debt contracts. On the borrower's side, it creates a moral hazard problem.

  27. "If a housing bubble made a construction worker earn 20 bucks rather than 15 where labor markets would clear and right now, assuming he won't go back to work for 15 for whatever reason, then inflation could simply give him a paycheck for 20 and the purchasing power of 15. Then we'd get to full employment faster."

    This statement is both insane and depressing. Did you learn nothing in your economics classes about the difference between nominal and real? The construction worker won't care if you give him $20 with the purchasing power of $15, he'll care only about the $15.

  28. Yeah, I know the distinction between real/nominal (or at least I have managed to pretend to know it well enough to trade fixed income for 3 years and not get fired). But the decision is not made by me. It is made by construction workers, a large part of whom do not understand the distinction until they have seen its effects on their purchasing power. But by then, it's too late, they have already signed their "real 15$" contracts, they are off welfare. Please read animal spirits to get the same argument by a guy with a nobel prize and a guy about to get one.

    You could counter that:
    A. No, construction workers have rational expectations and before they negotiate wages they open an excel sheet and estimate the impact of monetary policy on their purchasing power and adjust wages accordingly. But you would have been wrong. But even if you were right (you aren't) inflation would still lower the real value of their mortgage payment.

    B. This is a slippery slope, a few points lower unemployment rate is not worth the risk of turning into Weimar etc etc. And then you would maybe have had a point, but you didn't argue this, did you?

  29. L,

    That's interesting how you think good policy is tricking people into doing things they will regret later once they figures things out. Akerlof and Shiller's book I would characterize as being on the wacky fringe of economics. Nobel prizes are no guarantee that people will not write nonsense, as should be clear.

  30. A major feature of the New Keynesian model is the possibility that aggregate demand can go awry. I don't think this possibility is as far fetched or incoherent as Steve thinks it is. The best way to make the case for the notion that an economy might have an 'aggregate demand problem' is to begin with the real business cycle model where allocations are efficient, and there is never an aggregate demand problem. Suppose there is a bad intertemporal shock, one that behaves like an increase in the tax rate on the return to capital (this could be a reduced form representation for various types of financial friction shocks). It is useful to think about how the allocations in an RBC model respond to such a shock. The bad intertemporal shock obviously discourages investment. If a component of the demand for goods goes down and no other component goes up, then the national income identity says there must be a fall in economic activity. In principle, this could result in economic waste, as it implies unused capital and other resources. In the RBC model, the price of the good whose demand has declined (current output) falls relative to the price of other goods (future output). That is, the real rate of interest falls. This relative price move causes other sources of demand to increase, such as demand for consumption by households and it even partially reverses the negative impact of the intertemporal shock on investment. This increase in demand moderates the fall in output that would happen if only the first-round fall in investment occurred and nothing else happened. The decline in the real interest rate also leads to an increase in the consumption of leisure, so output does fall, though by less than it would if the initial decline in investment were the end of the story. So, the heart of the RBC analysis of the economic response to an intertemporal shock lies in the response of the real rate of interest. The efficient response that occures in the RBC model presumes an appropriate reaction in the real rate of interest. Now consider the effect of the decline in investment in the NK model. In that model, the real interest rate is the nominal interest rate divided by the anticipated rate of inflation. The real rate in the NK model is not determined exclusively by market forces, as it is in the RBC model. Instead, anticipated inflation is somewhat inflexible because of price and wage frictions. And, the nominal rate of interest is determined by the actions of the central bank. Unless the central bank sets the nominal interest rate so that the drop in the real rate of interest resembles the drop in the RBC model, we don't expect the demand for current goods and services to be the 'correct' one, and this is how aggregate demand might go awry in that model. The most dramatic example of this occurs when the interest rate is already at zero, so that the central bank has no power to reduce the nominal rate of interest. In this case, we expect there to be 'insufficient aggregate demand', and - by the national income identity - to be an inefficiently large recession. This logic suggests that and aggregate demand 'problem' might exist even when the interest rate is not at zero, as long as monetary policy is not conducted properly. Bottom line: if the real interest rate is determined by factors other than market forces, then it is possible for aggregate demand to be 'wrong', and for allocations to be inefficient. The NK model provides one set of reasons why this might be so.