Was “fiscal stimulus” the central idea that saved us from a second great depression, with our only regret that is was not bigger? Or is this an old fallacy, tried and failed, and ready finally for the ash-heap of history?

Let’s be clear what this issue is not about.  Governments should run deficits in recessions, and pay off the resulting debt in good times. Tax revenues fall temporarily in recessions. Governments should borrow (or dip into savings), to keep spending relatively steady. Moreover, many of the things government spends money on, like helping the unfortunate, naturally rise in recessions, justifying even larger deficits.  Recessions are also a good time to build needed infrastructure or engage in other good investments, properly funded by borrowing.  For all these reasons, it is good economics to see deficits in recessions – and surpluses in booms.

We can argue whether the overall level of spending is too high; whether particular kinds of recession-related spending are useful or not; whether particular infrastructure really is needed; and we certainly face a structural deficit problem. But those are not the issue either.

The stimulus question is whether, beyond all this, the government should intentionally borrow and spend, or even give money away, on the belief that each dollar so borrowed and spent will raise output by $1.50, and therefore lower unemployment. This, and only this, is the Keynesian “multiplier” argument, and the true meaning of “fiscal stimulus.”

Before we spend a trillion dollars or so, it’s important to understand how it’s supposed to work.  Spending supported by taxes pretty obviously won’t work:  If the government taxes A by $1 and gives the money to B, B can spend $1 more. But A spends $1 less and we are not collectively any better off2. 

“Stimulus” supposes that if the government borrows $1 from A and gives it to B we get a fundamentally different result, and we all are $1.50 better off.

But here’s the catch: to borrow today, the government must raise taxes tomorrow to repay that debt. If we borrow $1 from A, but tell him his taxes will be $1 higher (with interest) tomorrow, he reduces spending exactly as if we had taxed him today! If we tell both A and B that C (“the rich”) will pay the taxes, C will spend $1 less today.

Worse, C will work less hard, hire a bunch of lawyers, lobby for loopholes, or move to Switzerland. A will hire a lobbyist to get more stimulus. All this is wasted effort, so we’re worse off than before! The question for the “multiplier” is not whether it is greater than one, it’s how on earth it can be greater than zero? (Conversely, so far my arguments for the ineffectiveness of spending apply equally to tax cuts. But tax cuts can cut rates, which improves incentives.)

These statements are a theorem not a theory. I’m explaining (in very simple terms) Robert Barro’s (1974) famous “Ricardian Equivalence” theorem. “Theorem” means that if a bunch of assumptions, then borrowing has exactly the same effect as taxing. That doesn’t mean it’s true of the world, but it means that if you want to defend stimulus, you have to tell us which of the “ifs” you disagree with. That discipline changes everything.

Thoughtful stimulus advocates respond. Well, maybe people don’t notice future taxes. Does the man or woman on the street really understand that more spending today means more taxes tomorrow?

That’s an interesting position, but at this point, most of the battle is lost. Stimulus is no longer an “always and everywhere” law, it’s at best a “if people don’t notice that deficits today mean taxes tomorrow” idea. This qualification has deep implications.

First, it means that a “stimulus” policy can only work by fooling people. Is wise policy really predicated on fooling people? Also, people are unlikely to be fooled over and over again. If that’s how stimulus works, you can’t use it too often.

Second, it means that stimulus will work sometimes and not other times. Are American voters right now really unaware that larger deficits mean higher future taxes? Or is the zeitgeist of the moment exactly the opposite: Americans are positively aghast at the future taxes they think they’ll be paying? If you think people can be “irrational” they can be irrational in both directions. They can pay too much attention to future taxes corresponding to current deficits, and stimulus can have a negative effect!  When I compare tea party rhetoric to the actual reforms needed to cure America’s deficits, I think there’s a good chance we’re in this range.

Third, if this is the reason that stimulus works, then the current policy attempt, consisting of  stimulus now, but strong promises to address the deficit in the future, can have no effect whatsoever. If you think stimulus works by fooling people to ignore future tax hikes or spending cuts, then loudly announcing such tax hikes and spending cuts must undermine stimulus!  Augustinian policy, “give me chastity, but not yet,” will not work. Casanova is needed.

Well, maybe some other Barro assumption is wrong. Yes, there are many. (“Liquidity constraints” are a common complaint, keeping people from acting based on their estimation of the future.) But if you take any of them seriously, the case for stimulus becomes similarly circumscribed. Each specifies a channel, a “friction,” something fundamentally wrong with the economy that matters some times more or less than others, that restricts what kinds of stimulus will work, and that can be independently checked.  And in many cases, these “frictions” that falsify Barro’s theorem suggest much better direct remedies, rather than exploitation by fiscal stimulus.

Relaxing Barro’s assumptions can also lead to negative multipliers. For example, Barro assumed perfectly-efficient “lump-sum” taxes. In fact, we have proportional taxes with lots of loopholes. In the real economy, raising tax rates is an inefficient process, as is spending money. Recognizing this fact leads to my guess of a negative multiplier.  

So the biggest impact of Barro’s theorem is not whether it is “right” or “wrong” as a description of the world. The biggest impact is that, if you are at all intellectually honest, it forces you to deal with why it is wrong.  Many proponents do not do this; they just cite one assumption they don’t like, on the basis of intuition rather than real evidence, and go back to simplistic always-and-everywhere mulitipliers.  

There is a deeper problem with stimulus. Even if nobody notices future taxes, A was going to do something with the money.  Suppose, for example, A was a small business owner, and he was going to buy a forklift3. The government borrows the money instead, and gives it to B who buys a car.  Now the composition of spending has changed towards more “consumption.” But does the economy really care if B buys a car rather than A buying a forklift?  Barro’s theorem gives conditions in which nothing changes, including the split between consumption and investment. But his real point is deeper: Borrowing does not alter the “intertemporal budget constraint,” society’s overall wealth.

These two stories capture the central logical errors of Keynesian economics, and central advance of “equilibrium” or “inetertemporal” thinking that destroyed it and revolutionized macroeconomics in the 1970s. Some other big names in this effort are Friedman (1957), Lucas (1975), with Sargent (1979), Kydland, Prescott (1982). They pointed out two big mistakes in Keynesian economics:

First, Keynesian economics treats each moment in time in isolation. People’s consumption depends on their current income, not their future prospects. Investment decisions depend on current sales and interest rates, not whether companies expect future sales to be any good.  Modern macroeconomics extends across time. It recognizes that what people expect of the future is central to how they behave now.   Now, maybe people don’t “perfectly” or “rationally” evaluate the future. But that’s a far cry from saying they don’t consider the future at all!  And budget constraints – the fact that debt today must be paid off – are independent of your feelings.

Second, the “plans” of Keynesian economics4 ; how much we suppose people want to consume, invest, etc.; don’t automatically add up to their income, unlike the “demands” of regular economics  that must do so.  Keynesian economics ignores budget realities at each moment in time as well as the “intertemporal budget constraint” emphasized by Barro.

For these and other reasons, Keynesian ISLM models have not been taught in any serious graduate school since at least 1980, except as interesting fallacies or history of thought. I include my own graduate education at very liberal Berkeley starting in 1979. Even sympathetic textbooks, like David Romer’s Advanced Macroeconomics, cannot bring themselves to integrate Keynesian thinking into modern macro.  The “new Keynesian” economics, epitomized by Mike Woodford’s Interest and Prices has nothing to do with standard Keynesian thinking5 . Not a single policy simulation from a Keynesian model has appeared in any respectable academic journal since 1980. Not one. The whole business was simply discredited as being logically incoherent 30 years ago.

Now stimulus advocates may say “all academia lost its mind in about 1975.”  Paul Krugman’s New York Times article pretty much took that view. Maybe so.  I think most of academic macroeconomics lost its mind about 1935 and only started to regain it in the 1960s, so it certainly can happen.   But the claim “all academic economics from the last 35 years is wrong” is a far different form of scientific advice to policy-makers than is “sensible well-understood and widely-accepted economics supports my view.”  

I will admit to a bit of disappointment that so many economists revert to archaic Keynesian fallacies when under pressure. I have seen far too many well-trained Chicago Ph.D.s, whose entire professional career consists of exquisite intertemporal equilibrium modeling, fall apart when asked to explain policy, say to a Fed governor. Suddenly output becomes the sum of consumption “demand”, investent “demand” and government “demand,” ignoring that there is a supply in each case, and demand for one thing must come a the expense of another. I have seen the same economists forget that high interest rates are as likely to be a symptom of good times (high marginal product of capital) as a cause of bad times. In this sense, modern economics has indeed failed – we have failed to train our graduate students to really understand, apply and use the tools of modern macroeconomics, and to explain that analysis to slightly older economists trained in the ISLM tradition. But that doesn’t mean that good policy results by reverting to theories that were proved logically incoherent in the 1970s. 

Can’t the facts settle this argument? Alas, not easily – and both policy and economics would be better if they acknowledged this fact.  Some stimulus skeptics say “well, unemployment hit 10% after your stimulus,” and cite CEA chair Christina Romer and Jared Bernstein’s (2009) unlucky forecast that unemployment would stay below 8% if the ARRA “stimulus” package were passed.  This is unfair.  Stimulus advocates counter “without us, unemployment would have hit 15%” (or 20%, or pick your number, these are all made up out of thin air.)  Alas, cause and effect are hard to tease out in economics, because so much else is going on.

Stimulus buffs point to World War II, when the US borrowed and spent a lot, and output boomed. But skeptics point out that “everything else” is hardly kept constant in the middle of the greatest war ever fought, with price controls, production controls, direct command of much of the economy, rationing and a draft.  War economies can produce a lot, for a short while, and that fact is not only due to deficit spending.  Countries like the USSR that try to permanently run war economies don’t do so well.   

Skeptics might point to the Great Depression and Japan’s lost decade, in which stimulus seemed not to do much. But here advocates can point to other things going on, and claim again that things would have been worse without stimulus.

I can’t think of a single example in which a country attained prosperity and sustained growth by large deficit-funded social programs, make-work programs, or salaries and pensions for government workers, which is the issue here and a better precedent than WWII.  European and Latin American sclerosis come more to mind in this context. If stimulus leads to “growth,” why is Greece not superbly rich?

I am also dubious about empirical work in the absence of theory. If you don’t know how stimulus can work, can you productively look for it?  Is this like empirical work on the existence of UFOs?  Seriously, the violations of Barro’s theorem that might make stimulus work at one time or place are surely different than those which make it work in another time or place. Surely empirical evaluation must tie to measurement of which of Barro’s assumptions one feels does not hold.  

But the bottom line is that empirical work is hard.  21st century Economics is a lot like 18th century  medicine. The patient gets sick. The doctors come in, and bleed him, give leeches, and awful “medicines.”  “The Humors [“demand”] are out of whack.” “What humors?” “Oh, shut up and go read Galen [Keynes].” The patient gets worse, “we didn’t do enough,” and they bleed him some more.  He gets better, “we saved him.” Or maybe it was the toast with the strange mold that did it? Telling cause from effect is really hard.

And when you really don’t know,  “the patient is sick, we have to do something” is not wise policy or medicine.

Stimulus has all the telltale signs of bad, crackpot ideas. Already, I’ve mentioned repudiation of literally all macroeconomics taught in every Ph.D. program since 1980. Here are some more  “Bad Science Detectors:”

1. Stimulus supporters never say how it will work, and just cite authority. Ok, there’s this Barro theorem that says stimulus can’t work. Which one of the assumptions did administration economists or popular advocates disagree with? Let’s check. Instead, they simply assert it will work not even mentioning the classic theorem to the contrary.

2. They don’t take their own ideas seriously. Here are some examples.

a) If you really believe Keynesian Stimulus, you think Bernie Medoff is a hero. Seriously. He took money from people who were saving it, and gave it to people who were going to consume it. In return he gave the savers worthless promises that look a lot like government debt.

b) If you really believe in Keynesian Stimulus, then you don’t care if the money is spent properly or simply stolen. In fact, it would be better if it were stolen: thieves have a notoriously high propensity to consume, and their “spending” doesn’t wait for environmental impact statements unlike the unfortunate “shovel-ready” projects.   Keynesian Stimulus believers should advocate leaving the money around with the doors open.

c) Keynesian stimulus has nothing to do with “creating” or “saving jobs” directly, “green” industrial policy, or “infrastructure.”  If they believe their model, they should say loudly that it doesn’t matter how the money is spent. (And even here, there seems to be a view that the electorate can’t divide. Even if you believe that $800 billion saved something like 2 million jobs, that’s $400,000 per job!)

d) Berstein and Romer’s CEA report on the stimulus famously used a multiplier of 1.5 to evaluate the effects of the stimulus. They took this multiplier from models (p.12). But the multiplier is baked in to these models as an assumption. They might as well have just said “we assume a multiplier of 1.5.”

More deeply, why use the multiplier from the model, and not the model itself? These “models” are after all, full-blown Keynesian models designed purposely for policy evaluation. They have been refined continuously for 40 years, and they epitomize the best that Keyesian thinking can do. So if you believe in Keynesian stimulus, why use the multiplier and not the model?

The answer, of course, is that they would have been laughed at – nobody has believed the policy predictions of large Keynesian models since Bob Lucas (1975) destroyed them.  But how is it that one multiplier from the model still is a valid answer to the “what if” question, when the whole model is ludicrously flawed? If you believe the Keynesian model, let’s see its full predictions. If you don’t believe it, why do you believe its multiplier?

3.  More “Bad Science” detectors. One should always distrust advocates that have one fix (“more stimulus”) for everything, like the proverbial two-year-old with a hammer to whom everything looks like a nail. It sounds a lot like the old patent-medicine salesmen: “Recession got you down? Take some fiscal stimulus. Oh, a banking crisis is causing problems? Why it’s fiscal stimulus you need, son! Sclerotic labor markets? No problem, some fiscal stimulus will perk you right up. Nasty trade dispute? Why old Dr. Paul’s fiscal stimulus will perk you right up! Foreclosures giving a you headache? Why fiscal stimulus will solve that any day.”

A modern economy is a lot more complex than a car, and the answer to car troubles is not always “more gas.” If you see a solution being advocated for every problem, you begin to wonder how serious is the analysis that it solves any problem.

4. Finally, the desperation of recent arguments Keynesian stimulus advocates is a good indication of empty ideas.

Krugman and DeLong are blogging the idea that there is no model under which fiscal stimulus does not work. Well, how about Barro (1974) ? How about Kydland and Prescott (1981)? How about Baxter and King (1993)? These certainly are models, and classic models to boot. They are logically coherent – the “then” follows from the “if” and all the budget constraints and equilibrium conditions are right. The “if” might not describe the real world, so they might be wrong models. But they are models, they exist. And they are hardly obscure. They have been on every first year Ph.D. reading list since 1980, even at Princeton and Berkeley. The hard part has been to find on graduate reading lists any model in which stimulus does work.

How can one digest Krugman and DeLong’s “no model” assertions in light of this fact? There are only two logical possibilities: a) They don’t know about these classic works. b) They are deliberately lying to slander their opponents and mislead their readers.  As far as the strength of their argument, it does not really matter much which it is.

Similarly, Krugman now writes in his New York Times column that stimulus “wasn’t tried.” After $1.5 billion of deficit per year with $800 labeled “stimulus,” this claim is simply breathtaking. ($1.5 trillion: In the Keynesian model, all deficits count, not just the parts labeled “stimulus.”) It is obviously a feeble attempt to evade the overwhelming judgment of history that it was tried, in spades, threatening a second sovereign default crisis, and came up short.

5.  Most of all, you can tell a bad idea when you can tell the economics is cooked up after the fact to serve a political agenda.  A politician wants to spend a lot of money. “Hey economist, give me some talking points.”  “Sure,” says the economist, “we’ll defend it as fiscal stimulus.” That’s the only view that makes sense to me of the above inconsistencies, and it explains why this old idea is still bandied around in policy circles.

I am here to save good economic ideas, not to argue an alternative political agenda. Maybe it is a good idea to borrow, hire people, (in US, mostly state and local government workers), start a “green” industrial policy, give lump-sum or temporary tax rebates, and so on. I don’t think so, but we can have a good argument.  Just don’t justify it by “stimulus,” the proposition that each dollar so borrowed and spent makes us all $1.50 better off!  Doing so debases economics’ genuine ability to analyze policy and give us a credible menu of cause-and-effect statements! 

Keynes famously said, “Practical men… are usually the slaves of some defunct economist.” He was right (for once). Now he is the defunct economist. But we need not be his slaves!

 

References

Barro, Robert J. 1974,. "Are Government Bonds Net Wealth?" Journal of Political Economy 82 (6): 1095–1117.

Barro, Robert J. , 1981, “Output Effects of Government Purchases,” Journal of Political
Economy, 89: 1115.

Baxter, Marianne and Robert G. King, 1993,  “Fiscal policy in general equilibrium,” The American Economic Review, 83: 315-334.

Christiano, Lawrence, Martin Eichenbaum, and Sergio Rebelo, 2009, “When is the Government Spending Multiplier Large?” Manuscript, Northwestern University

Friedman, Milton, 1957, A Theory of the Consumption Function Princeton: Princeton University Press

Lucas, Robert E. Jr. 1975, “Econometric Policy Evaluation: A Critique,” in K. Brunner and A. Meltzer, eds., The Phillips Curve and Labor Markets, North-Holland.

Lucas, Robert E. Jr, and Thomas J. Sargent, 1979 "After Keynesian Macroeconometrics"  in After the Phillips Curve, Federal Reserve Bank of Boston, Conference Series No. 19: 49-72; reprinted in
the Federal Reserve Bank of Minneapolis Quarterly Review 3 (1979): 1-6

Kydland, Finn and Prescott, Edward C. (1982). “Time to Build and Aggregate Fluctuations” Econometrica  50 (6): 1345–1370.

Romer, Christina and Jared Bernstein (2009), “The Job Impact of the American Recovery and Reinvestment Plan”, Janua

 

Footnotes

1 John H. Cochrane is a professor of Finance at the University of Chicago Booth School of Business http://faculty.chicagobooth.edu/john.cochrane/research/Papers/

2 Yes, I’m aware that old Keynesian models do give a multiplier to tax financed spending. Also, some new Keynesian models such as Christiano, Eichenbaum and Rebelo (2009) predict huge government spending multipliers whether financed by taxes or by borrowing. However, tax-financed spending is usually thought to have a weaker (if any) effect, which is why the current policy debate is only about borrowing to spend.

3 Advocates will go nuts here, and complain that A might be putting money in the bank, and “banks aren’t lending,” or stuffing the money in mattresses. As you can tell, this line of argument leads us into “something’s wrong” with the banking system, and confusion between fiscal and monetary policy.

4 I take the “plans” language from Greg Mankiw’s textbook “Macroeconomics.” 

5 Explaining “new-Keynesian” economics is too big a task for this essay. My paper “Determinacy and Identification with Taylor Rules” and its references are the best place I can recommend for this question.