Friday, November 8, 2013

New vs. Old Keynesian Stimulus

While fiddling with a recent paper, "The New-Keynesian Liquidity Trap" (blog post), a simple insight dawned on me on the utter and fundamental difference between New-Keynesian and Old-Keynesian models of stimulus.

Old-Keynesian. The "Keynesian cross" is the most basic mechanism. (If you are worried that I'm making this up, see Greg Mankiw's Macroeconomics, p. 308 eighth edition, "Fiscal policy and the multiplier: Government Purchases.")

Consumption follows a "consumption function." If people get more income Y, they consume more C

C = a + m Y.

Output Y is determined by consumption C investment I and government spending G

Y = C+ I + G.

Put the two together and equilibrium output is

Y = a + mY + I + G
Y = (a + I + G)/(1-m).

So, if the marginal propensity to consume m=0.6, then each dollar of government spending G generates not just one dollar of output Y (first equation), but $2.5 dollars of additional output.

This model captures a satisfying story. More government spending, even if on completely useless projects, "puts money in people's pockets." Those people in turn go out and spend, providing more income for others, who go out and spend, and so on. We pull ourselves up by our bootstraps. Saving is the enemy, as it lowers the marginal propensity to consume and reduces this multiplier.

New-Keynesian. The heart of the New-Keynesian model is a completely different view of consumption. In its simplest version

Here consumption C, relative to trend, equals the sum of all future real interest rates i less inflation π i.e. all future real interest rates. The parameter σ measures how resistant people are to consuming less today and more tomorrow when offered a higher interest rate.

(This is just the integrated version of the standard first order condition, in discrete time
People in this model think about the future when deciding how much to consume and allocate consumption today vs. tomorrow looking at the real interest rate. I've simplified a lot, leaving out trends, the level and variation of the "natural rate" and so on.)

In this model too, totally wasted government spending can raise consumption and hence output, but by a radically different mechanism.  Government spending raises inflation π . (How is not important here, that's in the Phillips curve.) Holding nominal interest rates i fixed, either at the zero bound or with Fed cooperation, more inflation π means lower real interest rates. It induces consumers to spend their money today rather than in the future, before that money loses value.

Now, lowering consumption growth is normally a bad thing. But new-Keynesian modelers assume that the economy reverts to trend, so lowering growth rates is good, and raises the level of consumption today with no ill effects tomorrow. (More in a previous post here)

Comparing stories

This new-Keynesian model is an utterly and completely different mechanism and story. The heart of the New-Keynesian model is Milton Friedman's permanent income theory of consumption, against which old-Keynesians fought so long and hard! Actually, it's more radical than Friedman: The marginal propensity to consume is exactly and precisely zero in the new-Keynesian model.  There is no income at all on the right hand side. Why? By holding expected future consumption constant, i.e. by assuming the economy reverts to trend and no more, there is no such thing as a permanent increase in consumption.

The old-Keynesian model is driven completely by an income effect with no substitution effect. Consumers don't think about today vs. the future at all. The new-Keynesian model based on the intertemporal substitution effect with no income effect at all.

Models and stories

Now, why is Grumpy grumpy?

Many Keynesian commentators have been arguing for much more stimulus.  They like to write the nice story, how we put money in people's pockets, and then they go and spend, and that puts more money in other people's pockets, and so on.

But, alas, the old-Keynesian model of that story is wrong. It's just not economics. A 40 year quest for "microfoundations" came up with nothing. How many Nobel prizes have they given for demolishing the old-Keynesian model? At least Friedman, Lucas, Prescott, Kydland, Sargent and Sims. Since about 1980, if you send a paper with this model to any half respectable journal, they will reject it instantly.

But people love the story. Policy makers love the story.  Most of Washington loves the story. Most of Washington policy analysis uses Keynesian models or Keynesian thinking. This is really curious. Our whole policy establishment uses a model that cannot be published in a peer-reviewed journal. Imagine if the climate scientists were telling us to spend a trillion dollars on carbon dioxide mitigation -- but they had not been able to publish any of their models in peer-reviewed journals for 35 years.

What to do? Part of the fashion is to say that all of academic economics is nuts and just abandoned the eternal verities of Keynes 35 years ago, even if nobody ever really did get the foundations right. But they know that such anti-intellectualism is not totally convincing, so it's also fashionable to use new-Keynesian models as holy water. Something like "well, I didn't read all the equations, but Woodford's book sprinkles all the right Lucas-Sargent-Prescott holy water on it and makes this all respectable again." Cognitive dissonance allows one to make these contradictory arguments simultaneously.

Except new-Keynesian economics does no such thing, as I think this example makes clear. If you want to use new-Keynesian models to defend stimulus, do it forthrightly: "The government should spend money, even if on totally wasted projects, because that will cause inflation, inflation will lower real interest rates, lower real interest rates will induce people to consume today rather than tomorrow, we believe tomorrow's consumption will revert to trend anyway, so this step will increase demand. We disclaim any income-based "multiplier," sorry, our new models have no such effect, and we'll stand up in public and tell any politician who uses this argument that it's wrong."

That, at least, would be honest. If not particularly effective!

You may disagree with all of this, but that reinforces another important lesson. In macroeconomics, the step of crafting a story from the equations, figuring out what our little quantitative parables mean for policy, and understanding and explaining the mechanisms, is really hard, even when the equations are very simple. And it's important. Nobody trusts black boxes. The Chicago-Minnesota equilibrium school never really got people to understand what was in the black box and trust the answers. The DSGE new Keynesian black box has some very unexpected stories in it, and is very very far from providing justification for old-Keynesian intuition.



83 comments:

  1. John Cochrane notes that as far as the mechanisms go, the old Keynesian and new Keynesian models share practically none of the mechanisms that cause government spending to stimulate. I think he's being uncharacteristically charitable in this blog post. The obvious first guess why these old and new Keynesian models share part of the name and are advocated by the same people is that those people started with their desired policy recommendation (moar moar government spending) and then figured out some models to support that policy recommendation.

    The policy recommendation is the exogenous variable in those people's "research!"

    ReplyDelete
    Replies
    1. Which, of course is completely untrue - Ha! - of the Chicago/Minnesota school. (Sorry couldn't quite stifle my laughter till the end of the post.

      Delete
    2. I agree, the "Keynesians" are not the only people suffering from "exogenous policy recommendations." It plagues, to varying degrees, most non-experimental sciences that have any policy relevance. I do however find the hodgepodge of arguments for fiscal stimulus a particularly egregious example of this curse.

      Delete
  2. “But people love the story. Policy makers love the story. Most of Washington loves the story. Most of Washington policy analysis uses Keynesian models or Keynesian thinking. This is really curious. Our whole policy establishment uses a model that cannot be published in a peer-reviewed journal.” - Dr. Cochrane


    Maybe curious, maybe not. What if Keynesianism is a facilitator of politicos, or became a facilitator of politicos? Is Keynesianism, right or wrong, merely the excuse for other people [politicos] spending other people’s money [taxpayers] on other people [political constituency building exercise]?

    The government is simply a mechanism which has the power to take from some to give to others. It is a way in which some people can spend other peoples' money for the benefit of a third party - and not so incidentally themselves".

    The Invisible Hand in Economics and Politics, Milton Friedman, Institute of Southeast Asian Studies, 1981, p11.


    -And/or-


    “Keynes was exceedingly effective in persuading a broad group—economists, policymakers, government officials, and interested citizens—of the two concepts implicit in his letter to Hayek: first, the public interest concept of government; second, the benevolent dictatorship concept that all will be well if only good men are in power. Clearly, Keynes’s agreement with “virtually the whole” of the Road to Serfdom did not extend to the chapter titled “Why the Worst Get on Top.”

    Keynes believed that economists (and others) could best contribute to the improvement of society by investigating how to manipulate the levers actually or potentially under control of the political authorities so as to achieve desirable ends, and then persuading benevolent civil servants and elected officials to follow their advice. The role of voters is to elect persons with the right moral values to office and then let them run the country.”

    - Milton Friedman, Richmond Federal Reserve Economic Quarterly, volume 83/2 Spring 1997.

    http://www.richmondfed.org/publications/research/economic_quarterly/1997/spring/pdf/friedman.pdf

    ReplyDelete
  3. Shouldn't the first equation (consumption as integral of real rates) have some kind of discounting? Unless real rates go to zero, this would imply consumption is infinite.

    More substantial comment - there are more equations to basic New-Keynesian model than just consumer's FOC. After all, budget constraints must hold, so consumers can consume more only if they can spend more - and they have higher incomes because producers must hire more labor to satisfy increased demand. This part of story seems more in line with Old-keynesian logic. Of course, in the model everything happens simultaneously, so it's hard to tell a single simple story about what's going on (at least for me).

    ReplyDelete
  4. But for fiscal stimulus to increase inflation, don't you need the Old Keynesian multiplier story? I'm thinking of this portion of your old essay of Fiscal Fallacies.

    "Well, suppose the Government could borrow money from people or banks who are pathologically sitting on cash, but are willing to take Treasury debt instead. Suppose the government could direct that money to people who are willing to keep spending it on consumption or lend it to companies who will spend it on investment goods. Then overall demand for goods and services could increase, as overall demand for money decreases."
    http://faculty.chicagobooth.edu/john.cochrane/research/papers/fiscal2.htm

    Here you are taking money from people that won't spend it and giving it to people that will. If this is not the story that New Keynesians use, why do they think that government spending will lead to inflation?

    ReplyDelete
  5. John, I went to University of Chicago where I learned about a model which says that markets are not predictable in the short term. Since then, for 8 years I have been working in quant finance predicting markets in the short term. So maybe, it's a similar situation.

    ReplyDelete
  6. Don't the coordination games that Russell Cooper and John Bryant talk about give some game theoretic microfoundations for Keynesian cross models ?

    ReplyDelete
    Replies
    1. If, 80 years later, you're still looking for "microfoundations," that suggests a bit of a wild goose chase.

      Delete
    2. That seems a rather narrow-minded approach to the philosophy of science (economics). I guess if the physicists don't tie up this whole Standard Model thing in the next 40 years then it was all just a wild goose chase. That, or maybe, things are just harder than we think they are.

      Delete
    3. Jacks,

      I would say that the two situations aren't really equivalent. The main barrier to tying up the Standard Model, or disproving it, is a matter of having the technology to do so.

      Delete
  7. a few of the above comments seem to be saying this more explicitly, but the balance sheet recession model wherein for instance Krugman makes an exogenous shock to credit worthiness, certainly seems to lead to non-ricardian effects...and diverges from the Woodford paradigm of effects...really not sure whether that is old Keynesian though...

    ReplyDelete
  8. "The government should spend money, even if on totally wasted projects, because that will cause inflation, inflation will lower real interest rates, lower real interest rates will induce people to consume today rather than tomorrow, we believe tomorrow's consumption will revert to trend anyway, so this step will increase demand."

    Even this is flawed because inflation makes market interest rates move up. Of course the real interest rate on reserves is lower but no one can borrow off the central bank.

    ReplyDelete
  9. "People in this model think about the future when deciding how much to consume and allocate consumption today vs. tomorrow looking at the real interest rate." Except that there's a large population of "hand-to-mouth" households for whom this trade-off does not even register. Do you really believe the MPC of this group is zero?

    ReplyDelete
  10. John,

    "The parameter σ measures how resistant people are to consuming less today and more tomorrow when offered a higher interest rate."

    This is way oversimplified. It forgets that someone else had to borrow to spend today so that another person could lend to obtain interest to spend tomorrow. The New Keynesian model is incomplete. It includes interest rates but excludes debt levels.

    ReplyDelete
  11. John Cochrane: "The marginal propensity to consume is exactly and precisely zero in the new-Keynesian model."

    No it isn't. Consider log-linear utility as the simplest counterexample: U=B0C0+B1C1+...+BtCt+....

    As with any log-linear utility function you can read off the expenditure shares immediately. The fraction of an extra $1 of income spent on C0 is B0/(B0+B1+...+Bt+....). This is typically small but it isn't zero.

    ReplyDelete
  12. The implicit assumption in John's new keyenesian consumption equation is that the output gap eventually converges to zero, which holds if inflation eventually converges to some long run inflation target. This is coherent with a finite current deviation of consumption from its long run path because the sum of interest rate deviations converges, hence no infinite consumption.
    The workhorse new keynesian fiscal policy models (IMF's GIMF model for example) include significant "old keynesian effects" by having a subset of households who are very impatient and like to consume all their income. These hand to mouth households give some of the old income multiplier story, though their estimated share in consumption is usually less than 40%.

    I think the new keynesian expected inflation channel on consumption was also present in older keynesian analysis that took into account inflation expectations (not what you learn in macro 101 though). I believe it was called the Mundell-Tobin effect, and the general idea was similar though expressed differently: higher future expected prices would induce people to buy more today, and vice versa expected deflation would induce people to delay consumption today.
    Finally you can get big fiscal spending multipliers in a neoclasical model with flexible prices and wages if there are large costs preventing shifting workers between different production sectors. Roughly speaking, even with flexible prices, reductions in government spending will have big negative effects on output if it's hard for the private sector to take on the newly fired government workers, because of retraining costs or other forms of mismatch between the skills of government workers and private sector labour demand.
    I think Beaudry and Portier have a story like this in this paper,
    http://www.economics.ubc.ca/files/2013/05/pdf_paper_paul-beaudry-understanding-non-businesscyc.pdf
    Needless to say, this effect can be quite assymetric. I think it's more relevant for fiscal contration than for fiscal expansion.

    ReplyDelete
  13. I love your last paragraph!

    You are basically right about Old and New Keynesian models having totally different stories of how fiscal policy works. I would say it (and have been saying it) a bit differently:

    In Old Keynesian models (with an Old Keynesian IS curve), an increase in the level of G increases the natural rate of interest.

    In New Keynesian models (with a New Keynesian IS curve), a DECREASE in the expected GROWTH RATE of G increases the natural rate of interest.

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/03/fiscal-policy-with-old-and-new-keynesian-is-curves.html

    So if you want fiscal stimulus in an NK model, you want expectations of DECLINING G going forward, not HIGH G now, like in an OK model.

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/06/new-keynesian-countercyclical-fiscal-policy-isnt-what-you-probably-think-it-is.html

    BTW, you don't need to assume that G is totally wasteful. A sufficient condition is that U(C,G) is separable between C and G, so that a change in G does not affect the marginal utility of C.

    ReplyDelete
  14. Very fair points. But this raises another important question then: the empirical literature, although very divided, does find evidence that the fiscal multiplier is, at the very least, positive (I'll completely abstain from whether it's above or below 1, that's besides the point).

    In that case, what is the transmission mechanism through which government spending works in the economy? Just pure crowding out effects?

    ReplyDelete
    Replies
    1. Today, we're not talking about what's right or wrong, just pointing out that two different models have vastly different mechanisms. Both models produce multipliers, so the empirical literature you quote won't distinguish them.

      Delete
  15. You claim that old Keynesian theory is wrong and that many Nobel prizes have been awarded for demolishing it. Now, why is it wrong? It might not be perfect, but what is the fundamental reason for it being less accurate than the new theory? The term in the new theory equation (rate variation) is actually very important. How can I make a right decision if the uncertainty of future interest rates is high?

    ReplyDelete
  16. Neo-Keynesian and New-Keynesian are something different
    http://en.wikipedia.org/wiki/Neo-Keynesian_economics
    http://en.wikipedia.org/wiki/New_Keynesian_economics

    BTW, I do not think that any serious Keynesian defends "what you spend does not matter, just spend" thing. Money shall be spent on high multiplier projects that will not create crowding-out effect such as infrastructure spending..

    ReplyDelete
  17. per climate models: did you know many of them have 70's era large scale macro models underlying them?

    ReplyDelete
    Replies
    1. The abject failures of large-scale keynesian models certainly fuels a certain reserve among economists when contemplating climate models. But I though climate models were built on atmospheric physics, whose structural equations are at least understood, so the problem is nonlinear dynamics. Where do the large scale macro models fit in? Are they trying to forecast energy usage and co2 production with some sort of "economic" model feeding the climate part? That would be rather hilarious as long term growth is the economic issue, and the large scale models weren't even designed to think about those issues.

      Delete
    2. Yup, they see the economy as part of the carbon cycle, which it clearly is, so it's modeled just as the clouds are, another part of the feedback loop ...see Sokolov's work on the MIT Integrated Global System Modeling framework...
      http://globalchange.mit.edu/research/IGSM

      Delete
  18. "But people love the story. Policy makers love the story. Most of Washington loves the story. Most of Washington policy analysis uses Keynesian models or Keynesian thinking. This is really curious. Our whole policy establishment uses a model that cannot be published in a peer-reviewed journal."

    Perhaps this is because we still teach this stuff to undergrads.

    ReplyDelete
    Replies
    1. Less and less, actually. Most of the current crop of policy people were undergrads in the 1970s, when this stuff was still taught. Nobody has taught ISLM in graduate programs since about 1980 as anything but history of thought, and it's fading out of undergraduate and MBA education as well, especially at the top schools.

      Two verities: old ideas usually fade away with the people who hold them. And something I learned from my flight instructor training: the "law of primacy." If a student learns something wrong the first time, they will keep that wrong idea seemingly forever, and it will be very hard to eradicate.

      Delete
  19. Not an economist, so I can't tell if this is right: in the NK model, you say that the additional government spending lowers consumption growth. I can understand that. But if you believe that additional consumption now will also stimulate production, and economic growth overall, then doesn't that cancel out the effect? By changing the expected permanent income? I don't know if that makes sense.

    And on the policy angle, it seems relevant to point out that the loudest "Keynesian" voice - Krugman - has written a lot about how more inflation would be good. I don't know what the rest of the policy debate looks like, but it's not true that all of the Keynesian side are denying the need for inflation.

    ReplyDelete
  20. Perhaps the problem is precisely that New Keynesian put too much faith in the permanent income hypothesis --Campbell and Mankiw (1991) a found the data is consistent with about half of the economy operating under a liquidity constraint. (Permanent Income, Current Income and Consumption http://www.nber.org/papers/w2436)

    From the abstract: ;
    This paper reexamines the consistency of the permanent income hypothesis with aggregate, post-war, United States data. The permanent income hypothesis is nested within a more general model in which a fraction of income accrues to individuals who consume their current income rather than their permanent income. This fraction is estimated to be 40 or 50 percent, indicating a substantial departure from the permanent income hypothesis. This finding is robust to various statistical problems that have plagued previous work, such as time aggregation, and cannot be easily explained by appealing to changes in the real interest rate or to non-separabilities in the utility function.

    ReplyDelete
    Replies
    1. Interestingly, Cochrane has a response to that paper, which does raise some valid concerns.

      Delete
  21. I know it's hard, but let's try to stick to the point .. not which model is right (both are wrong, and both are horribly simplified here), not why you think stimulus or Keynesian models might work, not what bells and whistles we can add to the models, not what Krugman is writing these days and so on... Today's topic is the comparison between the central mechanisms of two very stripped down versions of two classes of models.

    ReplyDelete
    Replies
    1. But in a New-Keynesian model with borrowing constraints, wouldn't the transmission mechanism be similar to that in the old-Keynesian model? Isn't this what Krugman and Eggertsson show? I am not saying that I think their model is a good one. However, it does seem to me that if, because of liquidity constraints, the consumption path is too steep relative to the optimal path, the New-Keynesian mechanism does seem to resemble that in old Keynesian models.

      Delete
  22. To begin with, old Keynesians reject the entire exercise of microfoundations. Long before Lucas sounded the call to battle, Keynes had already nailed his colours to the mast with his "animal spirits". Old Keynesians believe agents behave irrationally and inconsistently. To them, the whole exercise of microfounding models in terms of rational and consistent agent behaviour is misguided from the start. If modern journals do not accept non-microfounded old Keynesian arguments, that is the fault of the journals, not the arguments, as far as they are concerned.

    Some of these old Keynesians took up new Keynesian theory purely as a necessary evil -- a forced sop to the misguided intellectual fashion of microfoundations. They never really believed the new theory -- it was at best a promising line of enquiry, at worst a Trojan Horse for their real old Keynesian ideas. Now that the Great Recession and accompanying embarrassment for mainstream macro has given them an opening, many of them are showing their true colours. There has been furious attack against microfoundations, the Lucas critique, rational expectations and so on from prominent Keynesians. The new Keynesians were old Keynesians in disguise all along, and now they throw off their masks and agitate for macro to be rolled back to where the General Theory would be respectable in journals again.

    They are not trying to hide old Keynesianism under a respectable veneer of new Keynesianism. They are trying to completely abolish what being "respectable" means in macro, and bring old Keynesianism back in toto.

    ReplyDelete
    Replies
    1. One could always prefer a different set of microfoundations. But "animal spirits" by itself is not very predictive. Behavioral economists (who are actually psychologists, for the most part) have done good work going more-micro-than-thou, but I don't think it's really been integrated into macro.

      Delete
  23. how can you discard old and new keynesian models, when on the first place Lucas-Sargent-Prescott models cannot even explain the crises in the first place ant its endurance, not even capable of starting any discussion in economics for the past five years about the mess we are in. keynesian models at least give a framework a starting point. blaming someone when you don't' have anything on the other hand is just mean, or maybe grumpy for a lack of a better word.

    ReplyDelete
  24. This paper might be relevant:
    http://research.stlouisfed.org/wp/2013/2013-026.pdf

    ReplyDelete
    Replies
    1. This is the Duopor and Li paper noticing that stimulus in new Keynesian models must operate through inflation, and the inflation is missing. Blog post here
      http://johnhcochrane.blogspot.com/2013/10/dupor-and-li-on-missing-inflation-in.html
      Bill has written a number of good papers evaluating the ARRA.

      Delete
  25. " tomorrow's consumption will revert to trend anyway"

    The core assumption underlying all Keynesian models, new or old, is that when the economy plunges like it did between 2007 and 2009, it is not due a rational revaluation of future economic prospects, but due to an irrational loss of confidence, desire to invest, "animal spirits" or whatever - in other words, the true productive capacity of the economy is no different, but investors' perception of it irrationally changes. Hence, we can assume that it will "revert to trend" and there is no reason to wait N years and do nothing while it does - government should step in and smooth out the cycle. Your opposition to this is basically that you do not believe investors/business en masse can be irrational. I would like to understand what this opposition is based on, other then a purely ideological view.

    ReplyDelete
    Replies
    1. No "opposition," I'm just explaining how the model works. To untutored eyes, the idea that higher growth is bad is otherwise perplexing, and the absence of an income effects, permanent or otherwise, is perplexing. Anchoring future consumption explains those otherwise perplexing features of the model.

      Delete
  26. But that's the key, isn't it? if you believe that US economy, as measured say by the market capitalization of S&P 500 or GDP or other ways, became in aggregate, worth 50% less in 2008/2009 than 2008 and this change in valuation is rational and efficient, then there is no way short-term government intervention can help, it can't change this fact. But if you do not, then you have to make assumptions about multipliers, mean reversion of consumption, etc. But this philosophical difference is at the root of the issue.

    ReplyDelete
  27. I know this post is about Keynes and Krugman and all, but I would like to digress: Europe is not doing great (I think we ALL agree that the the tiny growth in the last quarter is the result of a free falling that eventually has to decelerate, and not that they are doing something that works).
    Right or wrong, Keynesians are trying to fix the situation there. If their models are wrong, which models are right, and what they say that must be done? If they say that it is what Olli Rehn is doing, maybe they are not so right after all. If it's another answer, I'd like to know.
    An elegant model is pretty and all, but if doesn't help people with, you know, real problems, it's not really useful, is it?

    ReplyDelete
    Replies
    1. This is so wrong in so many ways. What, only Keynesians have a heart, and nobody else "cares?" How about caring enough to advocate policies that are logically coherent and might actually work, not advocating that, say, France, with 56% of GDP spent by government, deliberately spend more, even if wasted, to chase a will-o-wisp of "stimulus" rather than actually fix its broken economy? That's "helping?" To help people with real problems, you need models that work! Doctors that bled people thought they were "helping."

      Delete
    2. John, what do you mean "56 per cent spent by government". Are you referring to expenditure on the primary government accounts, and/or expenditure by state institutions and state-affiliated institutions? In some economies the distinction between the private and state sector is a blurred, often meaningless one. In some economies a large defence sector is where innovation takes place and cross subsidises the private sector through technology transfer. In some countries the university sector is completely state owned and is basically a training ground for the private sector workforce. There are countries with large state sectors that perform well (Scandinavian and Asian ones) and others that do not (UK, South America). Also there is no connection between state ownership of industry and performance. The Japanese bullet train was an incredible service before and after it was privatised. British trains were disastrous before and after they were privatised. I also wonder whether a lot of these multiplier effects and transmission mechanisms and whether consumption depends on current or permanent income is context specific and an entirely empirical issue. For example the OK or NK explanation and policy response ("model") may be right for the UK in the 1930s, but not for Japan in the 2010s. Are we basically restricting our analysis to the US post 2008? Arguably what some would call a liquidity trap situation?

      Delete
  28. Spain, Italy, Greece and Portugal are right, and France is wrong? Which one has the most stable forecast of Debt/GDP ratio path? All I see in your post (contrary to Krugman - oh, the blasphemy) is that they need to correct their economies, but not a single piece of advice.
    The answer to the question, btw (Debt/GDP in 2012 and 2018, by the IMF): France (90% and 89%), Greece (156% and 142%), Spain (85% and 105%), Portugal (123% and 116%).
    All the pain that the PIIGS are enduring, are they worth it, looking at these numbers? Really?

    ReplyDelete
  29. Hi John,

    I'm a New Keynesian myself and I also have problems with the proponents of fiscal stimulus. I'm frankly quite skeptical of large fiscal multipliers (the evidence is quite mixed, the proponents like to mention Romer or Blanchard's work but there is also strong work against it, such as that of Barro or Ramey). The main issue in my view is that strong evidence is simply not there (I recall that Paul Krugman admitted this as well in a post in his blog) and it seems a mistake to me to gamble large amounts of tax payers money on faith in Keynesian theory (old or new).

    I think however it is important to remember that:
    a) sometimes strange predictions by models or theories do later turn out to be shown correct (I can think of the case of anti-matter in physics);
    b) there are sound arguments to support the old Keynesian multiplier too. I can think of rule of thumb consumers and credit constrained households. There is evidence that this accounts for a large fraction of consumption movements (I can think of the work of Mankiw for example).

    So my opinion is that we should not put faith on the predictions of old and new Keynesian models with respect to fiscal policy (it can be perfectly the case that the models deliver a reasonable description of the economy in some aspects, like monetary policy, but poor in others, like fiscal policy), not because the theories sounds a bit crazy (after all many of today's accepted theories were once considered absolutely insane - the idea that the earth revolves around the sun; the idea that tiny, unseen to the human eye, organisms cause disease; ...), as I argue in a), or because of the lack of microfoundations, as I argue in b), but because the evidence in its favor is so far lacking.

    Regards,
    Joao

    ReplyDelete
  30. From Dave Burge, the funniest man on Twitter:

    "To help poor children, I am going to launch flaming accordions into the Grand Canyon."
    "That's stupid."
    "WHY DO YOU HATE POOR CHILDREN?"

    ReplyDelete
  31. I found Kevin Murphy's 2009 talk on the effectiveness of stimulus (the link below) the most lucid and persuasive of anything I have read. It was wonderfully accessible to a layperson and seemed to be logically consistent. How does that fit into what you are discussing. His framework seemed to be "Old Keynesian". Is it at odds with "New Keynesian" models?

    link: http://media.chicagobooth.edu/Mediasite6/Play/439a24a984fa449a8833412955afac451d

    ReplyDelete
  32. "A 40 year quest for "microfoundations" came up with nothing." In other words the juju that is macro-economics is baseless.

    "those people started with their desired policy recommendation (moar moar government spending) and then figured out some models to support that policy recommendation." That's probably what Keynes did too, but he was a very clever fellow. He could quick-wittedly knock together some story to justify his instinct, knowing that when his instinct changed he could spatch together a different yarn, and use it to justify a new policy. Unfortunately the long run caught up with him.

    ReplyDelete
  33. The equations in the article add a patina of mathematics, but the equations are not complete specifications for the Income-Consumption macroeconomic models advanced by either school. The 'New Keynesian' model as set out in the article states that current period aggregate consumption C(t) is a function of the sum of future real interest rates from the current period t to Infinity multiplied by the inverse of a scale parameter labelled "sigma". The parameter "sigma" is shown as being invariant with respect to the variable t (time). Now, if the integrand is negative for all time, i.e., the real rate of interest is negative, then the integral is negative; and, unless the parameter sigma is negative, aggregate consumption C at time t, and for all subsequent periods, will be negative. Additionally, if the nominal rate of interest equals the rate of inflation, for all time, then the integral equals zero and aggregate consumption C at time t, and all subsequent periods, will be zero. Now, if the nominal rate is less than inflation by a constant amount d for half the periods between t and Infinity, and greater than inflation by the same constant amount d for the other half of the periods between t and Infinity, the integral will evaluate to zero and aggregate consumption will be zero at time t and in all subsequent periods. The New Keynesian model as set out in the article is misspecified. Having misspecified the fundamental models, it is difficult to discern whether the author's remarks and conclusions are the result of the misspecification or the result of a fundamental disagreement with the assumptions underlying the actual models advanced by the Keynesians, New and Old. This writer is not convinced that the "multiplier" effect as stated by economic theory is real, but that lack of conviction has less to do with theory and more to do with the apparent inability of economists to determine empirically whether the ''multiplier'' effect in fact exists, and if it exists what it's value is reliably. Absent empirical evidence, the best that can be said for ''stimulus'' spending by government is that it brings capital into the economy from outside through the borrowing that government must undertake in order to pay for the ''stimulus''. The ''stimulus'' spending increases the government's budget deficit and raises interest rates; the off-setting ''immunization'' by the central bank acts to ameliorate the rise in interest rates caused by government borrowing by increasing the money supply which in turn raises the prospect of inflation in subsequent periods. Is this the linkage between aggregate consumption on the one hand and inflation and interest rates, on the other hand, embedded in the New Keynesian models? If it is, then it's not clear from the author's depiction of those models.

    ReplyDelete
  34. Prof. Cochrane, if not Mankiw's, what intermediate macroeconomics book should I use?

    ReplyDelete
    Replies
    1. I think Williamson has the most modern treatment. That's the one I am using.

      Delete
  35. Regarding the New-Keynesian IS curve you write:
    "There is no income at all on the right hand side. Why?"
    The answer is that you have stated an equilibrium condition with four edogenous variables. In the genreral equilibrium version, there is a market clearing condition, linking output and consumption. And much more to determine the model's predictions to changes in Y, G and what not.

    So your attempt to make a big distinction between the permanent income hypothesis and the New Keynesians fails somewhat in my view. They start out from the same consumption-Euler equation. In your Friedman story, you solve out for wealth as a right-hand side variable. Well, that can be done in the New-Keynesian version also. But you just stop that particular story at the Euler equation, and treats all endogenous variable as some you can toy around with. This is simply not a fair, nor sound, general equilibrium presentation.

    Kind regards,

    Henrik Jensen
    University of Copenhagen

    ReplyDelete
    Replies
    1. He wasn't making a distinction. "The heart of the New-Keynesian model is Milton Friedman's permanent income theory of consumption".

      Delete
    2. He certainly was here:
      http://johnhcochrane.blogspot.dk/2013/02/three-views-of-consumption-and-slow.html
      where the PIH gets some more kudos - simply because consumption is solved out as a function of exogenous variables. In any case, my point is just that one cannot say much of any theory with just one equation with four endogenous variables. Partial versus general equilibrium is quite important, as I am sure John would agree to.

      Delete
  36. Readers interested in a balanced perspective - among whom we may hope Grumpy himself is included - should see Brad Delong's comprehensive refutation of this post at the Washington Center for Equitable Growth blog:

    http://equitablegrowth.org/2013/11/12/622/oh-dear-megan-mcardle-relies-on-john-cochrane-and-so-goes-badly-astray

    ReplyDelete
    Replies
    1. Brad shows his talent for not reading things before he trashes them. Do you see any sign he read this post and not just Megan's coverage?

      This post is not about which model is right, or what happens if you add constrained agents, bells and whistles to the NK model. I said nothing about either issue.

      Chris Phelan's beautiful comment above applies too

      Delete
    2. Whoa! Now you're saying it "is not about which model is right . . . ," but in your post you said ". . . the old-Keynesian model of that story is wrong."

      And if new-Keynesians don't "forthrightly" accept your caricature
      of their views, they're not being "honest."

      Chris Phelan's comment is an unhelpful caricature, too, one which isn't truly representative of anyone's actual views.

      Delete
    3. Good catch. I'll stand by "the old-Keynesian model of that story is wrong." Wrong as in, doesn't make economic sense, and the post is about that indeed and the curious way that policy uses discredited models.

      I have stripped both models down to essentials, though, and the point of the post is to compare their mechanisms, not to advance a model which I think is "right." Can't do everything in one blog post.

      Delete
    4. Professor,

      You say, "This is really curious. Our whole policy establishment uses a model that cannot be published in a peer-reviewed journal. Imagine if the climate scientists were telling us to spend a trillion dollars on carbon dioxide mitigation -- but they had not been able to publish any of their models in peer-reviewed journals for 35 years."

      And yet you link to your article that references a wealth of articles published in peer reviewed journals that do in fact advocate fiscal policy solutions to the financial crisis.

      Consider Mike Woodford's paper, Simple Analytics of the Government Expenditure Multiplier Published in 2011 in the American Economic Journal: Macroeconomics. Is that not a peer reviewed journal article? Does it not refute your claim that these policy recommendations are based on models that can not be published?

      Delete
    5. Fair enough. Are you saying both models have short-comings? Who disagrees with that? But from here it sure looks like you're taking a swipe at the whole idea of "stimulus" and "multiplier" and can't resist the wasteful-spending-is-good, savings-and-growth-are-bad canards. We've been here before.

      For whatever reason (?) you find the very idea of government spending having a stimulus/multiplier effect so erroneous and repugnant you can't resist taking pot shots at it, even in the face of evidence of its positive results in spurring the economy and alleviating unemployment.

      Delete
    6. The point Mr. Cochrane makes here is completely different. It is something akin to what Nick Rowe says. It basically goes like this:

      New Keynesian and Old Keynesyan models are COMPLETELY different. They tell absolutelly DIFFERENT story.

      Now we have public intelectuals that use New Keynesian models but with old Keynesian "intuition". But that intuition is not supported by the model!!!

      It is like modern physicist calculating what is going on inside the black hole will use Quantum Field theory but he will also assume some results of newtonian physics just to make his model "work" and get rid of some unpleaseant singularities and infinities that would otherwise be implied by his own model.

      This is not how Science is made! Either New Keynesian model is "righ" and they will explain why some "counterintuitive" (relative to Old Keynesian intuition) explanation that arises from the model is not valid - like for instance promising to decline Government Expenditure in the. future.

      Delete
    7. John: As for Woodford's paper please go and read starting page 3 containing sentences like

      "Suppose that the central bank maintains a strict zero inflation target whenever this is possible, and a nominal interest rate of zero whenever defation is unavoidable"

      What folows is a very technical desription how to use Govenrment Expenditure to prevent deflation that monetary policy is supposedly unable to preven. So long story short - it is your classic New Keynesian model where stimulus works by increasing inflation! If you would present this paper to any Keynesian economist in 1960ies he would stare at it unable to follow. It is completely different story compared to what he was thought how government stimulus works.

      PS: By the way am I the only one that finds it hardly believable that any central bank that has enough ink and paper on stock will find itself in a situation where "defation is unavoidable". That is complete BS.

      tl;dr

      Nope, no problem with Woodfor Paper. It is New Keynesian model through-and-through as described by Mr. Cochrane

      Delete
    8. And Woodford is always admirably honest about describing how his models work. Not many people understand him, or prefer to cite him incorrectly.

      Delete
  37. Did I miss something from Friedman's "Positive Economics"? You don't judge a model by its unrealistic assumptions or its ability to mimic people. You judge a model on the basis of how well it predicts. Now I frankly I find the NK story pretty far fetched - except that it is predicting well what is happening in Japan with Abenomics. I also think WWII (and the 60s) are pretty convincing pieces of evidence for old Keynesian predictions too.

    ReplyDelete
  38. Dr. Cochrane,
    Its easy to bash "Old Keynesianism." I don't believe in it. But if Im not mistaken, your original post attacking fiscal stimulus as fiscal fallacy did NOT address New Keynesianism. At all. I commend you now for doing so!

    So where's the in-depth analysis of why New Keynesianism is wrong? :-) Because it seems to me that THAT theory, not the idea of the multiplier, but the idea of increasing inflation expectations to lower real interest rates, makes a lot more sense. Advocates of stimulus have this story to fall back on and they do.


    Just a side note on the 1970's. Yes, the 1970 DID show that there was no LONG RUN trade off between inflation and unemployment, but even Milton Friedman was careful to admit that indeed there was a SHORT term tradeoff. (A tradeoff that we ought to exploit vigorously now) Just look at the evidence when Paul Volcker relented and lowered the punishingly high interest rates he used to create the 1982 recession. If it were true that there is no tradeoff AT ALL between inflation. and unemployment, all the lowering of interest rates would have done is increase inflation but it didn't! It lowered the unemployment rate. The short term tradeoff was reformed.

    Level targeting is a rules based policy that uses this insight. Imagine an economy running smoothly at an inflation rate of 2 percent. Then, a financial crisis hits, Inflation falls into 5% deflation! The country's central bank acts quickly to end the deflation, but not to the previous level of 2%, Instead they rest easy at 1.0%. (thats why your argument that the Philips Curve is screaming supply was so troubling to me). But even IF the CB was more aggressive and pushed inflation to 2%, The NAIRU there would be profoundly misleading, because it would ignore the huge drop in the financial crisis. The proper inflation rate is 7%! (At least for that year) 2% inflation which is the normal trend, plus 5% "make-up" for the crisis. Only then can we determine what the natural unemployment rate is and determine demand from supply. The NAIRU is wrong. it should be the NAPLU (Non-accelerating price level of unemployment) Or, as I would prefer, the NANILU! (Non-accelerating Nominal Income Level of Unemployment)

    ReplyDelete
    Replies
    1. In depth analysis of why New Keynesianism is wrong here
      http://faculty.chicagobooth.edu/john.cochrane/research/papers/cochrane_taylor_rule_JPE_660817.pdf
      and here
      http://faculty.chicagobooth.edu/john.cochrane/research/papers/zero_bound_2.pdf

      Delete
    2. I think you are selling your excellent JPE 2011 paper short here. It is not a paper about "why New Keynesianism is wrong". It is about price determination, and lack thereof, in models with Taylor-type interest rate rules. All formalism on how explosive equilibria cannot be ruled out by reasonable arguments are confined to flex-price models (where prices and inflation are irrelevant, and the latter only appears by insistence on focusing on a Taylor-type rule). When it comes to the New-Keynesian models, you write on p. 593:

      "One might complain that I have not shown the full, nonlinear model in this case, as I did for the frictionless model. This is a valid complaint, especially since output may also explode in the linearized nonlocal equilibria. I do not pursue this question here since I find no claim in any new-Keynesian writing that this route can rule out the nonlocal equilibria."

      Indeed, I might complain, but hope to see the analysis done. I do not think one falsifies an entire theory because one cannot find papers that have formally validated one's own conjecture of what could otherwise "save" it.

      Best, Henrik

      Delete
  39. Just an addendum: The 7% rate is only make up for the financial crisis, 2% is what the CB aims for the next year, and the normal price level increase year to year

    ReplyDelete
  40. To John Cochrane... you said, "If you want to use new-Keynesian models to defend stimulus, do it forthrightly..." I'm baffled that you do not recognize that the Keynesian "multiplier" effect is just another way of describing the govt-spending-causes-higher-inflation effect. One is a FLOW model, the other is a STOCK model. The stories as you've penned them are not *always* how they play out - the outcome of increased govt spending depends on the market the spending fall upon. These dynamics are what those Nobel prizes were awarded for, not for disproving Keynes.

    Finally, Keynes was utterly forthright: even if the Treasury buried money, it would be more beneficial in depression conditions than doing nothing. Krugman and Delong have never disowned this example. Krugman has repeatedly explained that the ramp-up to WWII is the (wasteful) govt spending that finally brought the US out of the Depression, and he even created a new version of it: if govt spending were to increase to fend off a (fake) alien invasion, that would be better than doing nothing, waiting for the economy to heal itself, under these conditions.

    The big thing you're getting wrong is thinking that New Keynesians are trying to secretly disown the "multiplier."

    Note: When I say above that one is a Stock and one is a Flow analysis, resist the urge to simplistically dismiss this because it doesn't jive with whatever equations you wish to relate it to. You can understand this without the math - and, in fact, that math tends to add a layer of difficulty, because...: If you want to understand the micro-foundations of empirical macroeconomic reality, you have to embrace complexity: the system that arises from many agents acting with imperfect competition, information asymmetries, and human (evolutionary) psychology. If you start with the assertion of rational expectations and inter-temporal maximization, you've already missed the boat to understanding.

    Honestly, I'm not sure the paper has been written yet that provides the accurate Stock vs Flow equations, but Krugman hinted at some of this in his 11/7 IMF presentation, and his analysis (re international trade and currency flows, exchange and interest rates, and both public and private debt as an inter-temporal buffer) goes a long way toward providing a micro-foundations model that applies to individual behavior, too. But to see this, you have to think carefully about the analogues of these variables in individual psychology.

    ReplyDelete
    Replies
    1. Guys will you actrually read what Mr. Cochrane wrote in his article? He is not exactly arguing against any particular results of New or Old Keynesian models. He just says that they are different. One example that Nick Rowe uses.

      In new Keynesian Model all that matters is change in Growth Rate of G - not the level. So what can you do to solve the situation? You can do ether this:

      1) Increase G now while holding future G constant

      or

      2) Keep G as it is today and LOWEr future G

      This is what is implied by New Keynesian model. But please provide me any link where Brad DeLong or Krugman propose anything close to 2). They don't. Why? Because only first thing works for OLD Keynesian model.

      But this is not right. Then you are as Old Keynesian as it gets. You cannot be selectively blind and completely ignore results of YOUR OWN MODEL just because some completely different model of how economy works is also aligned with one result.

      And Mr. Cochrane is right - no serious peer rewieved Journal publishes Old Keynesian models right now - and for a reason!. Then why use their intuition? It does not make sense. To say the least it is not intellectually honest.

      Delete
    2. by your and Mr. Cochrane's analysis, QE doesn't work - so why do both bond and stock markets clearly empirically respond to QE announcements ?

      Delete
  41. Complicated models similar to physicist trying to explain light energy. All over the place trying to make the old mechanics work. Then came nice and simple E=MC2. Copernicus similar, and ofcourse Keynes the same in Econ. Lower zero bound and liquidity trap means there is no shifting outward of the PPF curve. We rob Peter to pay Paul. Keynes efforts to shift out curve so more benefit. FDR showed reality that some toes get stepped on with TVA. And yes inflation will occur as idle labor enters market and will harm savings. Ancient capital will be replaced perhaps prematurely with innovation supported by WE THE PEOPLE. But being a sleep at the wheel is no excuse for not retrofiring capital. Savings hurt by inflation is nice problem if you have savings. And besides our economy is quick to create new production when inflation signals demand. The worst inflation I know of in this country was gas in the 70's and we survived. Find the simple solution in Fiscal projects and quit wasting time and confusing the debate over these irrelevant models.

    ReplyDelete
  42. Prof Cochrane,

    Although you have stated that you wanted this to be about the basic models of old vs new Keynesianism. The only "Keynesians" most of your readership seem to be familiar with are krugman and delong. That being said Krugman is wrong when he accuses you of not being familiar with the Woodford/eggertson literature...clearly he doesn't read your papers...and I frankly don't believe he understands the new-keynesian stuff ...

    what i mean to say is that krugman thinks his model is old keynesian and doesn't seem to buy into the interest rate mismatch story all that much ... why he believes his model has anything to do with the ISLM, i am still failing to grasp... but if you are actually talking about Krugman whenever you say keynseians it kind of seems like you are setting up a straw man

    ReplyDelete
  43. Professor Cochrane,

    That was an amusing post.

    However, you are wrong. First, new Keynesians acknowledge that many consumers are credit constrained so any model of the reality should have Old Keynesian effects. Second, the Euler equation does NOT determine consumption by itself, but in combination with the inter temporal budget constraint. Therefore, income (and unemployment, transfer policies, asset holdings etc) have an effect on consumption. There are many reasons why Old Keynesian effects should exist in a New Keynesian world.

    ReplyDelete
  44. Dr. Cochrane,

    This is an interesting post. But I can't help but wonder if you've considered the idea that you can, in fact, have a model that is both old and new Keynesian at the same time. Where is the contradiction in assuming an economy with both kinds of consumers depicted in the Old and New Keynesian models? For example, if I'm a credit-constrained consumer, I will probably spend all my additional income, making me fit the old Keynesian model; if I'm an unconstrained consumer, then I'm more likely to fit the intertemporal maximizing world of the New Keynesian model. You'll probably lament about the "microfoundations" of such a model as being difficult, but I really don't see why we can't have both types of consumers in one model. More importantly, doesn't it make sense to give policy advice knowing that both types of consumers exist? It seems to me that's what contemporary Keynesians like Krugman, DeLong, and Woodford are doing--trying to make their models more real-world. I don't see the contradiction in doing that.

    ReplyDelete
    Replies
    1. There are all sorts of hybrid models out there, and the new-Keynesian DSGE literature has all sorts of bells and whistles. With bells and whistles (like some rule of thumb consumers) they quickly become black boxes where the basic mechanisms are obscured. My point here was not to look at the most general model, but to strip each one down to its core elements so we could see the basic forces at work.

      When you add rule of thumb consumers and other bells and whistles to a new-Keynesian model, you do NOT get back to simple old IS LM. You get a model with the core I described above and a lot of additional complex dynamics. As one example, the vast majority of "stimulus" evaluations end up being Ricardian -- stimulus is the same if financed by taxes or by borrowing.

      Delete
  45. You set up a straw man by confusing "Bastard Keynesianism", a weed infesting the US, and the real thing, a British hot house flower cultivated principally in Cambridge by English and Italian gardners, with occasional help from a Polish theorist. The absence of such perspectives from the journals - which you take as significant - tells one nothing about the quality of their perspective; it merely tells one who is making the rules for the inclusion of articles; it also underscores the reality that all the real Keynesians are dead. One might suggest that Joan Robinson hit the nail on the head when she said ‘‘The purpose of studying economics is not to acquire a set of ready-made answers to economic questions, but to learn how to avoid being deceived by economists’. Perhaps you have offered something that overturns Keynes/Kalecki/Robinson & Sraffa; I must have missed it.

    ReplyDelete
  46. "equals the sum of all future real interest rates i less inflation π i.e. all future real interest rates" please replace by "equals the sum of all future nominal interest rates i less inflation π i.e. all future real interest rates"

    ReplyDelete
  47. The high priest of the Catholic church did not teach about Copernicus and his mad theories and 'instantly rejected' its publication, even when the ideas had gained currency around Europe.

    I expect this submission will also be 'instantly rejected'. I might try for an alternate journal.

    ReplyDelete
  48. The dirty little secret is that both sides are guilty of doing this. When the economy goes South, people want the government to DO something, and the only thing anyone can think to do is a Keynesian response

    ReplyDelete
  49. I find it funny to bash Keynesian models, the one that actually have worked relatively well in this slump. Most of the other models has been predicting inflation for years, and for Japan for a decade or more.

    All models are wrong, some just are more useful.

    ReplyDelete
    Replies
    1. Actually the New-Keynesian model has been predicting deflation and strong output growth for 5 years now.

      Delete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.