What is the point of the General Theory?

I plan to reread parts of the General Theory in the near future, but before doing so I thought I’d sketch out why I’ve never been able to take the book seriously (unlike the Tract on Monetary Reform, which is a fine book.)  It seems clear to me that in the GT Keynes was focusing almost entirely on aggregate demand.  That is, he was primarily concerned with nominal income determination, not how nominal income gets partitioned between prices and real output.  Unfortunately, he doesn’t seem to have any model of nominal income.  If you search the GT for an explanation of why U.S. nominal GDP is roughly $15 trillion, rather than $15 billion or $15 quadrillion, you won’t find any explanation beyond a perfunctory nod to the long run role of the classical model (MV=PY.)

I presume that Keynes or his supporters would argue that the classical model determines the long run, whereas the model in the GT determines short run changes in nominal income.  But even if this were true it wouldn’t mean the GT had any sort of coherent theory of nominal income, rather it would mean that the GT was incoherent in a different way.

When I hear people discuss the long run I am sometimes reminded of students who mistakenly assume the term ‘long run’ means ‘the distant future’ and ‘short run’ means ‘the present or near future.’  But that is not at all what these terms mean.  The present, right now, is the “long run” for policies instituted years ago.  So if Keynes believed that in the long run nominal income is determined by monetary factors, then he should have explained current movements in nominal income in terms of past movements in monetary policy.  Of course that is not what the GT does.

Let’s continue to assume that Keynes had some sort of long run monetary model of nominal income determination–i.e. that the GT is not completely empty.  Can we still make sense out of the other components of the model, the MPS, the paradox of thrift, the liquidity trap, etc?  It’s hard to see how.  Since Keynes saw a lack of AD as the key problem of his time, he should have recommended a monetary policy aimed at making sure nominal income continued to grow at a rate consistent with something close to full employment.  Under that sort of policy, however, the rest of the GT would make no sense.  Changes in the MPS, investment shocks, fiscal policy, etc, would be fully offset by the optimal monetary policy response to those shocks, leaving expected nominal income growth unchanged.  Obviously Keynes thought non-monetary shocks were very important, so presumably he did not believe that their impact would be fully sterilized under an optimal monetary regime.  Indeed, he never even suggested such an option was feasible, even in theory.  Why not?  The only explanation I can think of is that he saw monetary policy as being constrained in some fashion.

Elsewhere I have argued that Keynes never really developed an adequate model of what those constraints were.  At times he hints at the famous “liquidity trap” concept.  But if monetary policy has a long run effect on nominal spending, then in principle there should be no “trap”–as monetary expansion would raise the expected rate of inflation, and depress real interest rates at the zero bound.  In the GT he only suggests one example of extreme liquidity preference–the 1932 Fed open market purchases.  In this case, however, the problem wasn’t a liquidity trap, it was the constraints of the international gold standard.

Alan Meltzer wondered why if Keynes was so concerned about liquidity traps he didn’t recommend a significantly positive trend rate of inflation.  The answer is that Keynes never envisioned such a monetary regime as being possible.  And even if he had, he did not believe in the Fisher effect (except under hyperinflation) and thus would not have seen Meltzer’s proposal as accomplishing anything.  Keynes lived in a commodity money world and always opposed the adoption of the sort of completely unanchored fiat regime that is required to generate positive long run inflation.

Both Hicks and Friedman saw extreme liquidity preference as the key to the GT, indeed Hicks saw it as Keynes’ only real contribution.  It is true that shifts in investment spending or in the MPS or in the budget deficit can impact velocity, and thus nominal income, but Hicks said this was already common knowledge in Cambridge.  Without extreme liquidity preference, the optimal response is merely to move M to offset V.  Keynes’ key insight, his only deep theoretical insight, was to argue that this wasn’t possible.  Everything else was putting commonly understood concepts into a different language.

Krugman recognized that Keynes’ liquidity trap rested on a foundation of sand and attempted to build a model of “expectations traps” that was consistent with rational expectations.  I have doubts about this model, but even if one accepts Krugman’s argument it doesn’t really help the GT very much.  Krugman argued that temporary increases in the money supply will be hoarded, leaving AD almost unchanged.  But this would apply even more strongly to budget deficits, which unlike money supply increases, must be temporary.

If a transitory budget deficit will have no long run impact on nominal spending (holding money constant) then its effect on current spending will be even weaker than otherwise.  There is a reason why modern graduate macro texts place so little emphasis on the ideas that Keynes developed in the GT, they are very hard to justify in a model with rational expectations.  What puzzles me is why concepts such as the MPC, the multiplier, the paradox of thrift, and fiscal stimulus have recently become so widely debated among economists.  Do these concepts help us understand movements in nominal spending?  And if so, what is the model that justifies that view?



11 Responses to “What is the point of the General Theory?”

  1. Gravatar of Bill Woolsey Bill Woolsey
    17. February 2009 at 16:46

    The constraint on monetary policy is that it involves targetting the federal funds rate and then using open market operations with T-bills to hit that target. The target is lowered in order to raise (or prevent a drop) in nominal income. Once the target hits zero, there is nothing more to be done, and nominal income remains too low….. There is nothing more that monetary policy can do.

    Yes, pretty silly, isn’t it.

    OK.. Once the T-bill rate is zero, base money and T-bills are perfect substitutes. So, further open market operations using T-bills have no economic effect. So, quantitative easing using T-bills has no effect one the interest rate on T-bills is zero. (I suppose this really means that once the interest rate on T-bills hits zero, the relevant supply of base money increases to include all the zero-interest T-bills that have yet to be purchased.

    Anyway, the next step is for the Fed to start purchasing longer term and higher risk securities. Well, that means the Fed might take a loss. Can’t do that! (Of course, the Fed is doing it now.)

    As you mention, in a gold standard world, maintaining gold convertibility is an important constraint. If people are willing to hold T-bills rather than gold, budget deficits might reduce gold demand and raise nominal income.

  2. Gravatar of ssumner ssumner
    17. February 2009 at 17:30

    Bill, Thanks for the comment. The problem with the zero interest rate bound is that it only applies to nominal rates, the Fed should be able to drive real interest rates as low as they like, by increasing inflation expectations. Of course a Keynesian could argue that the liquidity trap will last from now until the end of time, but that would mean that the Federal government could finance its entire debt at zero cost. If we take the more realistic assumption that an increase in the money supply will eventually increase the price level, then the Fed can reduce real rates through monetary expansion, even in a liquidity trap.

    I need to post a piece on Krugman’s expectations trap–which might provide an alternative justification for Keynes’ skepticism about monetary policy, but one that I find highly implausible.

  3. Gravatar of Bill Woolsey Bill Woolsey
    20. February 2009 at 03:12

    I think there are two problems with using inflation expectations to manipulate real interest rates.

    The most important one is that creating a credible expectation of inflation requires that the Fed follow through, and the Fed shouldn’t cause inflation. I don’t favor a 2% or 1% trend inflation rate. But, if one has that, then if the real interest rate needs to be more negative than 2%, then the Fed would need to create an expectation, and then follow through and create 3%^ inflation. That is undesirable.

    The second problem is that if the Fed must lower real interest rates to raise demand and so cause inflation, then it cannot credibly cause inflationary expectations. If people believe that the Fed can magically create inflation, then it can make real interest rates what it wants by announcing the inflation it will create. But if real interest rates must fall to get to the inflation, then why would anyone believe what the Fed says?

    The problem remains the contraint that the central bank purchase assets or else lend. (Of course, these amount to the same thing in some fundamental sense.) I don’t think this is insurmountable. But that is the problem.

    Clearly, if the Fed purchased new cars and houses, this would raise the demand for these things. But like heliocopter drops, this isn’t really monetary policy.

    Of course, once the price level has fallen, then a commitment to getting the price level back to “normal,” would involve reflation–inflation. As a rule, I would support such a policy, so that solves the first problem–but
    only after there has been deflation and recession.

    My view has always been that there is little reason to claim that monetary policy is ineffective until the Fed has purchased all of the T-bills. And then, only after it has gone a good bit of the way towards buying the entire national debt (the long term government bonds) and a variety of short term private debt as well.

    But, I do believe that there must be a plausible market process that leads money creation (including the increase in liabilities or decrease in other assets to the rest of the economy) to an increase in demand. Going straight from money creation to inflationary expections doesn’t make it. And, again, I generally think causing inflation is a bad thing.

  4. Gravatar of ssumner ssumner
    22. February 2009 at 08:15

    Bill, You raise several issues here, and I’d like to begin by differentiating between what you consider to be undesirable during a liquidity trap (creating inflation) and what Keynes claimed was impossible (creating inflation.) In the 1930s (more outside the GT that withing the GT) Keynes often claimed that an expansionary policy could not boost prices. My real interest rate mechanism was meant only to be a counterexample to Keynes’ assertion. As a general rule I don’t favor having the Fed target real interest rates, nor do I think that they should change their inflation target during a liquidity trap. If zero inflation is normally best, they should continue aiming for zero inflation during a liquidity trap (and use futures targeting, or some other unconventional policy.) So we are not far apart on that issue.

    Your second argument says that there is no reason why people would believe that an expansionary policy would raise future prices, and hence lower current real rates. There are two possibilities here. If the money is not expected to raise future prices, the Fed should buy up the entire national debt, which will save the taxpayers the future cost on that debt, without the negative consequences of inflation. But Keynes did not recommend this policy, as I am sure that he did believe that this sort of extreme policy would eventually produce inflation. But if it will eventually produce inflation, it will, ipso facto, immediately reduce real rates when nominal rates are stuck at zero.

    I would also emphasize that the one occasion when even you would support inflation, when some reflation was desirable after a sharp fall in the price level, was exactly the situation in effect in the 1930s when Keynes was putting forward his liquidity trap model. So I still think that Keynes did not really think through the implications of his hypothesis in any sort of systematic fashion. BTW, I have an article on this subject in the 1999 Economic Inquiry.

    I also still think that the excess cash balance mechanism is a plausible transmission mechanism for anything other than exchanging reserves for zero interest T-bills. And here I think we agree, as you also mention non-conventional open market policies might be effective. The only slight difference is that I think that even T-bill purchases can be effective if expected to be permanent (if T-bill yields are expected to eventually rise above zero.) This is the issue Krugman addresses in his 1998 “expectations trap” article.

    Thanks for the comments, I just posted a long piece on liquidity traps.

  5. Gravatar of Barry Ickes Barry Ickes
    25. February 2009 at 06:32

    You do not explain why government spending on unemployed workers will be hoarded. You take an argument that a temporary tax cut will be hoarded to mean that any fiscal policy is less effective than monetary policy. But that is not obvious. If you do not consider involuntary unemployment there is no reason to read Keynes. If you view all economics through the lens of Arrow Debreu you do not need macroeconomics anyway, nor can you really do it.
    You cannot talk about policy in a crisis without asking what is that causes the crisis. That is why you seem unable to understand the point of the GT.

  6. Gravatar of Floccina Floccina
    25. February 2009 at 08:29

    As a general rule I don’t favor having the Fed target real interest rates, nor do I think that they should change their inflation target during a liquidity trap. If zero inflation is normally best, they should continue aiming for zero inflation during a liquidity trap (and use futures targeting, or some other unconventional policy.) So we are not far apart on that issue.

    What about George Selgin’s idea that we should not target CPI inflation but a level agrigate spending?

  7. Gravatar of Un nouveau blog de macro monétaire bigrement intéressant « Rationalité Limitée Un nouveau blog de macro monétaire bigrement intéressant « Rationalité Limitée
    25. February 2009 at 08:56

    […] lire notamment un billet sur la Théorie générale de Keynes qui tranche violemment avec la Keynes-mania ambiente où Sumner démonte notamment de manière […]

  8. Gravatar of Jon Jon
    25. February 2009 at 09:42

    Bill Woolsey writes:
    “The constraint on monetary policy is that it involves targetting the federal funds rate and then using open market operations with T-bills to hit that target.”

    This really gets everything backwards. The short-term rate has a very small impact on long-rates and very little economic impact onto itself; the rate target is a _discipline_ not a goal. Its purpose is to discipline the rate of monetization. As a side-effect is also manipulates the short-long spread and thus bank earnings, but this is a tiny effect on GDP (but a big effect on bank health).

  9. Gravatar of ssumner ssumner
    25. February 2009 at 18:25

    Barry, A short post will always seem too simplistic, but I have a long article on Keynes in Economic Inquiry (1999) that explains my views more fully. You misunderstand where I am coming from, however. I certainly do not accept the new classical, market-clearing, Arrow-Debreu approach to business cycles. I believe (like most Keynesians) that sticky wages and prices are the key imperfection that causes nominal disturbances to have real effects. I just don’t think that Keynes has a good model of nominal disturbances in the GT. I think he does have a good nominal model in the Tract, as he was dealing (implicitly) with a fiat money world. The Tract (1923) is a response to hyperinflation in the early 1920s–and hyperinflation tends to cause even Keynesians to think like monetarists. Oddly, the GT is a gold standard model, made by an economist who doesn’t understand that he constructed a gold standard model. This is discussed more fully in my 1999 article. Just as fish never notice they are wet, Keynes never noticed that the gold standard subtly influenced almost everything he did in macroeconomics.

    Floccina, Like George, I prefer nominal GDP targeting. But I prefer a positive rate–at least 3% growth, preferably 4% or 5%. I believe George prefers stable NGDP.

    Jon, I think you might have misinterpreted what Bill was trying to say, but I do agree with your general point. When the Fed changes its fed funds target, it may or may not have a big effect on the economy. But if it does have a big effect, it is not because short term rates changed slightly, but because of the change in the money supply (and thus future expected nominal GDP) that was necessary to effect that change in short term rates. My “Rational expectations” post has more to say on this issue.

  10. Gravatar of Green Monkeys Green Monkeys
    8. April 2009 at 07:59

    you should keep increasing your readers i think a lot of people would like the content of your blog.

  11. Gravatar of The ‘Sumner Critique’, or Why Not to Ignore Keynes « Unlearning Economics The ‘Sumner Critique’, or Why Not to Ignore Keynes « Unlearning Economics
    31. May 2012 at 06:36

    […] their time – virtually every macroeconomic insight is already in The General Theory. Sumner says he has ‘never been able to take the book seriously’. Maybe he just needs to read it […]

Leave a Reply