Thursday, June 19, 2014

Keynes's Not-So-General Theory and the Supposed Impotence of Monetary Policy

In 1935, John Maynard Keynes wrote to George Bernard Shaw:

I believe myself to be writing a book on economic theory which will largely revolutionize—not, I suppose, at once but in the course of the next ten years—the way the world thinks about economic problems.”

After he published The General Theory, Keynes’s formulation of economics was received as though it was revolutionary, especially by his younger followers. Many older economists were not quick to embrace Keynes’s doctrine. As David Laidler points out, “Pigou and Knight in particular, were scornful of his claims to novelty (Fabricating the Keynesian Revolution, 21).” In The General Theory Keynes draws upon arguments from both his contemporaries and past economists, but especially in the case of his contemporaries, he typically fails to cite them. So what did Keynes actually contribute to economic theory? His main contribution was to call attention the need for economic analysis where the macro-economy fails to reach an equilibrium, but this contribution is obscured by a framing of the argument that ignored the economic significance of institutional collapse and his denial of the ability of monetary policy on its own to aid the process of recovery.

In the opening chapter of The General Theory, Keynes immediately clarifies his stance and his goals. “The postulates of the classical theory,” Keynes writes, “are applicable to a special case only and not to the general case (3).” The particular case, according to Keynes, is the case of full employment and the general case includes all states where the economy operates below full employment. He builds his theory with the belief that the economy does not typically operate at full employment, but rather “without any marked tendency either towards recovery or toward complete collapse (249).” If both of these claims are true, then in most circumstances the classical model is inadequate to employ in analysis. For the sake of remaining concise, I shall only briefly state that this proposition is untrue. Empirical investigation shows that the economy tends to move toward the long-run outcome predicted by the classical model (Kehoe and Prescott 2007). Only in the case of a general fall in prices and sticky wages is there a shortfall in demand where the economy operates below its potential (Galloway and Vedder, 89-97; Leijonhufvud, 49-50).
It appears that Keynes’s theory is the “special case”. Not only is it special, it is so particular as to call into question its applicability altogether. That is, Keynes questions the efficacy of monetary policy and its ability to return aggregate demand to its potential. In order for his theory to be useful, it needs to be better than just a second best option, which, if monetary policy is effective, is the ranking to which the theory must be relegated. As Hawtrey explained in a paper critiquing the support of Keynes and others for increased capital outlays as a remedy for depression,

Currency depreciation is far the most satisfactory measure of revival. Not only is it better balanced, but it is quicker and easier to bring about. I have already pointed out that a capital programme regarded as a measure for breaking the vicious circle of depression is likely to be too slow and too gradual to be successful, and I have suggested that, when cheap money fails to bring about a prompt revival, there is more to be hoped from an open market policy, the purchase of securities by the central bank. I should be inclined to leave the question at that, confident that a sufficient purchase of securities would overcome any depression however severe. For whereas cheap money reaches a limit when the rate of interest approaches zero, the purchases of securities can be increased indefinitely.

. . . The capital programme has the grave disadvantage of coming into operation tardily and gradually. Nor is it possible to say how great a programme will is needed to resolve the deadlock or whether any practicable programme will be great enough. If a capital programme were the only means of resolving the deadlock, we should have to make the best of it, but I believe that there are good reasons for supposing that a sufficiently liberal measure of open market purchases by the central bank would be bound to achieve this object.

. . . Since a programme of capital outlay offers so limited and doubtful a contribution towards revival, I think it is regrettable that excessive prominence is given to it by economists. (456-58)

The need for capital outlays is contingent on Keynes’s claim that the price level will not respond to an increase in the money supply when the economy is at less-than-full employment because he proposes that the price level is primarily a function of wages. If interest rates are too low to encourage investment, entrepreneurs will not invest, and therefore, output will remain stagnant.

The acuteness and the peculiarity of our contemporary problem arises, therefore, out of the possibility that the average rate of interest which will allow a reasonable average level of employment is one so unacceptable to wealth-owners that it cannot be readily established merely by manipulating the quantity of money.

. . . But the most stable, and the least easily shifted, element in our contemporary economy has been hitherto, and may prove to be in future, the minimum rate of interest acceptable to the generality of wealth-owners. If a tolerable level of employment requires a rate of interest much below the average rates which ruled in the nineteenth century, it is most doubtful whether it can be achieved merely by manipulating the quantity of money. (308-9)

As Keynes links changes in the price level with changes in employment, this is his subtle way of saying that an increase in money will not lead to an increase in investment as holders of the new money will not lend it out. As mentioned in my last post on The General Theory, tremendous deflation occurred in England, Keynes home country, between 1929 and 1931. This continued in gold standard countries generally, including the U.S., until 1933. During this period of deflation, we can expect that the [hypothetical] equilibrium nominal rate of interest was negative for an extended period of time. Remember that,

i = π + r

Ex post real rates for this period are in the double digits during some years! (For example, see Thayer Watkins calculations for the U.S. here) The dramatic fall of in investment during this time period suggests that this was out of equilibrium play.

Deflation during these years was the result of a collapse of the banking system in the U.S. and of the international gold standard. Between 1929 and 1931, U.S. had experienced a tremendous increase in demand for money. This had made the Depression, to that point, one of the worst on record. Low levels of output in combination with a fragile unit-banking system that struggled to remain solvent prevented recovery. Between May 1931 and March 1933, a series of banking panics led to an increase in cash balances for a fearful public, and therefore, a continuation of the contraction of the money stock (Friedman and Schwartz, 308-315). Unit banking in the U.S. prevented the spread of liquidity which would have likely prevented or slowed the process – banking panics were prominent in the U.S. during this period, a problem not experienced by countries lacking this restriction.

Furthermore, some central banks had begun hoarding gold at the end of the 1920s and continued this practice into the 1930s. The prime offenders were the Bank of France and the Federal Reserve. The bank of France increased its holdings from 7 percent to 27 percent of the world’s total gold reserves (Board of Governors 1943, 544-55). In the U.S. gold holdings shrank only slightly as a proportion of the world’s gold reserves as board members at the Federal Reserve refused to adopt a policy of easy money until February 1932. Even then, they did so timidly until prodded by congress in the following months. By this time, the collapse of the banking system in the U.S. was already under way. 

If there were bottlenecks in production that resulted from interest rates failing to allocate resources across time, the demand deficiencies were the fault of bad monetary policy. Excessive deflation was the result of gold hoarding and tight monetary policy more generally. This being the case, fiscal policy is an unnecessary band-aid if the policy goal is to offset dramatic falls in aggregate demand. Aggressive monetary policy would have done just fine to offset the deflation, as is evidenced by the end of the first phase of the Great Depression in 1933 when FDR devalued the dollar.


  1. James, I think I've encountered your blog before. Do you sometimes comment at David Glasner's blog or at JP Koning's? Do you consider yourself to be an Austrian economist? I took a brief look around... I liked the bit about free will perhaps being an illusion in one of your other posts (something I've thought about many times).

    I'm an amateur and much of this goes over my head, but I think I understood your recent comment at Sumner's. I'm curious how he'll respond.

    I also took a glance at your post about RE and the EMH and incorporating those ideas into Austrian econ. I've never given much thought to Austrian econ, but that did look interesting.

    I've been intrigued recently by another econ theory that is completely different than anything else I've seen (granted I haven't seen much). His formulation does not rely on microfoundations, or expectations models. A version of the EMH results from his framework though, as do supply and demand curves, and several other more mainstream results. His approach supposes that we may not need to model some of the details of individual agents in order to have a useful theory, analogous to the way the principal of "entropy maximization" can sometimes be used to analyze a complex system w/o necessarily delving into detailed models of the constituent parts. He freely admits this idea might be completely wrong wrt macro, but nonetheless I think his model of the price level (for instance) does seem to have a decent match with a lot of empirical data. Here's the "hard core" of the theory (it's short!):

    Here's one of his many price level plots (he has them for different countries and different time periods):

    Have you ever seen an approach like that before?

    On a completely different note, are you familiar with Mike Freimuth? What is your view of this for example? (from his blog):

    I thought this was an interesting post commenting on both Mike Sproul and Sumner:

  2. I comment on Glasner's blog sometimes. I posted about Keynes and Hawtrey a while back and linked to it in his comments. You might remember me from that.

    Two thoughts on the micro-macro post at information Transfer Economics:

    1) Macro does not need micro foundations if the model falls within the positivist paradigm. I.e., what does the model predict? A model that lacks micro foundations is legitimate inasmuch as it is subject to that standard.

    2) There is a lot of room to expand on modern macro by moving away from the pure macro in search of micro-foundations. In other words, agent based modeling, systems theory, and the likes, have a lot to offer macro theorizing.

    I like to critique version 1 of macro, but not because I don't think it is useful. If you have read my paper on the gold standard (currently under review at the Financial History Review), you will see that I use version 1) of macro aided by a robust narrative.

    I wish I had time to comment on your other questions, but I'm short on time right now.


    1. Yes, exactly (re: you old post)... I left multiple comments there didn't I. Ha!

      Also those are an interesting two thoughts you have there. Thanks for your thoughts.

      Jason (the blog I link to above) solves the equation he sets up in the "hardcore" post in a couple of different ways (he calls them the endogenous and exogenous solutions). The endogenous solution looks like this:

      log P ~ (1/kappa - 1) * (log M)

      With kappa = log M / log NGDP, and "~" is read as "goes as."

      kappa can take on a lot of different values, but it tends to stay between 1/2 and 1. When kappa = 1/2, then the QTM applies to the economy in question. When kappa = 1, then P is insensitive to M.

      M is taken to be the currency component of the base (MB) here, sometimes dubbed "M0."

    2. Here's a good (brand new) example of the above:

  3. Interesting post here James. I love these ones touching on the history.

  4. Tom,

    I'm glad that you are enjoying them. So much of this history goes unnoticed. It's worth digging up and presenting concisely so that non-experts can can have an accurate narrative.