New Keynesian Vs Keynes Challenge II

Eminent New Keynesian Economist Susanto Basu commented on a post challenging people to come up with an empirically useful idea in macroeconomics not to be found in “The General Theory of Employment Interest and Money.” Here I will just note again how incredibly polite New Keynesians tend to be. bold numbers added for future reference

This is a great challenge. I am answering off the top of my head, and will probably think of things I want to add later. Empirical findings that seem hard to explain without forward-looking expectations: [1] Mankiw, Miron and Weil on interest rates and the founding of the Fed (AER, 198?), [2] Sargent on the quick remonetization of economies ending hyperinflation, and probably most importantly [3] the empirical work in Milton Friedman’s Theory of the Consumption Function. Apropos that last, since a lot of what you say is about consumption, the recent micro research on consumption is much more nuanced than your portrayal. One of the best recent papers is Hsieh (2003, AER), which shows that the PIH works very well for large, predictable income changes but *among the same people* quite badly for small and irregular changes. This finding points to the need for refinements of models based on forward-looking expectations (like costly information processing) rather than their wholesale removal. Cochrane (1989, AER) is insightful on this point. Apropos aggregate data and excess sensitivity, I would point to my paper with Miles Kimball (, which tries to challenge the interpretation of the famous Campbell-Mankiw papers. Shifting from data to models qua insightful stories, I think it’s pretty clear that Keynes makes sense only with substantial nominal rigidities. Now I stand foursquare for the importance of sluggish nominal wage and price adjustment in lots of circumstances and for long periods of time, but even I have a hard time saying that decade-long events like the Great Depression or Japan in the 1990s are due to price inflexibility over such long periods of time. (And yes, I know Tobin/DeLong-Summers on destabilizing price flexibility. But still.) For something on that scale, I find multiple-equilibrium models like Diamond (1982) or Benhabib-Farmer (1994) much more appealing intuitively. [4] But I don’t think one can tell those stories without rational expectations–at least not easily. Benhabib-Farmer show conditions under which the steady state becomes a sink rather than being saddlepoint-stable in the forward dynamics. But my guess is that if you put in backward-looking expectations, the sink would turn into a source, and the multiple equilibria would disappear. This is all about intuition and telling just-so stories with models rather than anything with clear empirical support, but I would say that’s also the appeal of Keynes. Thanks for making me come up with reasons why I think modern macro isn’t all a waste! I look forward to your thoughts. Susanto Basu

I answer the answer to the challenge after the jump.

[1] I haven’t read “Mankiw, Miron and Weil on interest rates and the founding of the Fed (AER, 198?)”. I will just note that I aaked for a prediction and this paper was written after the founding of the Fed.

[2] “Sargent on the quick remonetization of economies ending hyperinflation.” This time I won’t just note that it isn’t a prediction, because it is in “The General Theory …”

2) Sargent (relying on the hard boring archival research of economic historians) noted that at the end of four hyperinflations the Phillips curve was vertical. Keynes relying on casual empiricism said the same thing (not similar the same). The following passage is incomprehensible without context (and uncharacteristically turgid in context) but it is Sargent’s observation.

The General Theory … chapter 21 section 6

“e without suffix ( = Mdp/pdM) stands for the apex of this pyramid and measures the response of money-prices to changes in the quantity of money.


“But in general e is not unity; and it is, perhaps, safe to make the generalisation that on plausible assumptions relating to the real world, and excluding the case of a “flight from the currency” in which ed and ew become large, e is, as a rule, less than unity.”

In English, the claim is that in the case of hyperinflation “flight from the currency” an increase in the money causes a proportional increase in the price level so no effect on output or employmen (OK it is possible that he thought e would be greater than 1 in case of hyperinflation so increased prices would cause reduced real balances as claimed (with data) by Cagan). Having been alive in 1923 Keynes had noted that the end of the German hyperinflation was not accompanied by a sharp decline in employment. I think Sargent set up and knocked down a straw man. He certainly didn’t cite Keynes or any old Keynesian claiming that there was a sloping Phillips curve in e.g. Germany in the 20s.

[4] Not in order. Benhabib and Farmer have interesting models. In fact it is much much easier to get models in which sunspots matter if one doesn’t assume rational expectations. The hard part in these models is to find a way that an asset price (or shadow price) can be different from the steady state (or balanced growth — that is steady state for the properly scaled variables) without exploding. This is hard if an asset value equation is imposed — that is hard exactly if one assumes rational expectations. If one allows irrational exuberance and/or panics everything is much too easy — trivally easy in fact. That’s why the papers always impose rationality. I think that Benhabib and Farmer are on to something, but that is a hope not a confirmed prediction.

As an aside, I note that there is a quite different way to get stability and indeterminacy. Different enough to be published anyway. See

Alessandra Pelloni and Robert Waldmann (1998) “Stability Properties of a growth Model,” Economics Letters, vol. 61 pp 55-60.

I can promise you that this model has exactly zero influence on my thoughts about the economy. Mathematically it works just as well as the Benhabib Farmber models, but I am absolutely sure that it has nothing to do with reality. Speculation about any possible conflict between co-authors due to this freequently stated certainty would not be appreciated.

[3] “the empirical work in Milton Friedman’s Theory of the Consumption Function. Apropos that last, since a lot of what you say is about consumption, the recent micro research on consumption …”

The PIH. Look I wasn’t always like this. I used to be only 90% alienated from the mainstream. I have said (in lectures) that the PIH is partly true, explains facts which seemed mysterious and is not in “The General Theory …”. In fact the challenge used to include “except for the PIH”. But then I had an alarming thought (see below).

Keynes considered the effect of expected future income on current consumption. Relying as always on casual empiricism (it is hard to run regressions with almost no data and no computer) he guessed that expected future income must be considered to explain individual consumption but also that it is not useful in explaining aggregate consumption.

“The General Theory …” Chapter 8 section 2 point 6

(6) Changes in expectations of the relation between the present and the future level of income. — We must catalogue this factor for the sake of formal completeness. But, whilst it may affect considerably a particular individual’s propensity to consume, it is likely to average out for the community as a whole. Moreover, it is a matter about which there is, as a rule, too much uncertainty for it to exert much influence.

Much evidence that there is something to the PIH (glass half full– there must be at least some liquidity constrained consumers to fit the facts) is based on micro data. This tends to confirm Keynes’s guess as usual. I note that his mention of “uncertainty” seems to appeal to his “Treatise on Probability” which I haven’t read. He does not assume that agents have a subjective probability distribution and maximize expected utility. He considers the possibility that, if something is highly uncertain, people will just put it out of their minds. I think he is right. This is much much further from contemporary practice than just assuming that people don’t have rational expectations so their subjective probability distribution does not correspond to the objective probability distribution conditional on information available to them. It is an explanation (using the word very loosely) of why people might act myopic regarding aggregates even if the data should give them some ability to forecast GDP and the price level.

OK what about reality ? Is there any evidence that forecasting future aggregate income is a useful step in forecasting or fitting aggregate consumption ? Hmmm well first, as predicted by the PIH for some stochastic processes of aggregate income, the coefficient of variation of consumption is lower than that of income. For other stochastic processes, this is a problem for the PIH (excess smoothness). Given the state of the debate (when last I checked long ago) this is about 0 evidence for or against the PIH. There are other explanations. One (very important to old Keynesian models and I admit based in large part on what Keynes wrote) is that consumption is a constant “autonomous consumption” plus a constant times disposable income. This silly model incorrectly implies that the ratio of consumption to income should decline as per capita income increases. Also poor people violate their lifetime budget constraint somehow.

I really don’t think Keynes was that silly. In particular he presented another explanation for the smoothness of consumption ( which implies an interpretation of “autonomous consumption not as an exogenous constant but as a predetermined endogenous state variable)
chapter 8 section 3 paragraph 3

“This is especially the case where we have short periods in view, as in the case of the so-called cyclical fluctuations of employment during which habits, as distinct from more permanent psychological propensities, are not given time enough to adapt themselves to changed objective circumstances. For a man’s habitual standard of life usually has the first claim on his income, and he is apt to save the difference which discovers itself between his actual income and the expense of his habitual standard; or, if he does adjust his expenditure to changes in his income, he will over short periods do so imperfectly. Thus a rising income will often be accompanied by increased saving, and a falling income by decreased saving, on a greater scale at first than subsequently.”

Ah habit formation. Is there really evidence which supports the PIH against the hypothesis of myopia combined with habit formation ? The second is the Marglin model of consumption. It is definitely anticipated by Keynes. The alarming thought is that habit formation explains the cross sectional patterns which are also explained by the PIH (eg the fact that for the same current income African Americans have lower consumption than white Americans). Consumption is not just a function of current income. But does it work better to consider the discounted stream of future income or a geometric average of past income (as would be true in the model of myopia and habit formation). Especially is there macro evidence which distinguishes the two ?

Well the PIH implies Ricardian equivalence and the myopic habit model doesn’t. Uh the evidence is not kind to the PIH on this. OK look across countries there are some countries with persistently rapid growth and others with slow growth. Other things (especially r) equal, the PIH says rapid growth countries should have high C/Y. The myopic habit models say they should have low C/Y. They have low C/Y. OK OK given the correlation of S and I, this could be just low C/Y causes high growth (it is two facts that the variance in savings rates is high compared to the variance of current account surpluses and that the correlation is not one). But the evidence, such as it is, is again unkind to the PIH.

I’d say Keynes on consumption reads a whole lot like an informal summary of recent empirical research on consumption (note also the bit about precautionary savings which I haven’t mentioned yet).

Chapter 9 section 1 point 1

“There are, in general, eight main motives or objects of a subjective character which lead individuals to refrain from sepdning out of their incomes
(i) To build up a reserve against unforeseen contingencies”

I note that Hsieh (2003 AER) tends to confirm Keynes’ guess.