Fama's Fallacy: Hoisted from Ten Years Ago
I was profoundly embarrassed by and ashamed of the Swedish Nobel Committee and of being an economist when they awarded the Nobel Prize to Eugene Fama.
You see, the economists who cheerled for the Trump-McConnell-Ryan tax cut and claimed it would rapidly and permanently boost annual investment in America by 800 billion had arguments—bad arguments. The economists who condemned Benanke's quantitative easing and claimed it would soon lead to high inflation and a debased dollar had arguments—bad arguments. I do not think any of them made those bad arguments in good faith: the failure of those in either group to acknowledge that they got a big one wrong and to engage in Bayeisan updating is interesting: that silence speaks volumes.
But Fama and the others who claimed a decade ago that, while private decision to spend more boosted employment and production, public decisions to spend more—fiscal stimulus—not only would not, but could not possibly ever boost employment and production... they had no argument at all.
What do I mean? This. This is why I am embarrassed and ashamed: Hoisted from Ten Years Ago: Fama's Fallacy, Take I: Eugene Fama Rederives the "Treasury View": A Guestpost from Montagu Norman, former Governor of the Bank of England:
Back in the 1920s and 1930s—in the days that overly-clever bisexual academic dilettante John Maynard Keynes was trying to persuade us that if only we got the government to spend more money the unemployment rate might go down—by far the silliest argument against his position was the one put forward by the staff of the Chancellor of the Exchequer: the so-called "Treasury View."
The Treasury View was that nothing could boost employment: not government spending, not tax cuts, not private business decisions to expand their capacity, not irrational exuberance on the part of entrepreneurs—for the level of output was what it was and the unemployment rate was what it was and no fiscal policies or private investment decisions could change it, for all they could do was move resources from one use to another without affecting the total flow of economic activity.
Back on Christmas Eve Paul Krugman whacked Caroline Baum of Bloomberg on the nose for rediscovering the Treasury View. Now Eugene Fama of the University of Chicago has rederived it from scratch (apparently without knowing anything of its history), claiming that the savings-investment national income identity proves that fiscal policy cannot have any effect on output and employment.
This is a howler of such magnitude that it has pulled me from my grave to speak—because we went over and over this in the 1920s starting with R.G. Hawtrey (1925), "Public Expenditure and the Demand for Labour," Economica 5, pp. 38-48, and with F.W. Leith-Ross's various Treasury memos to P.J. Grigg, and thrashed this out to a conclusion that Fama appears not to know. It is very strange: the argument Fama wants to make—that government deficits completely crowd out private investment so that fiscal policy has no effect on output or employment—is, depending on circumstances, sometimes true and usually false, depending on circumstances. But the premise from which Fama attempts to derive complete crowding-out is the savings-investment accounting identity in the National Income and Product Accounts—and an accounting identity is something that must be true by construction, no matter what. The fact that savings equals investment in the NIPA is logically independent of whether the complete crowding-out doctrine is true or false.
Here is Fama:
Bailouts and Stimulus Plans: There is an identity in macroeconomics... private investment [PI] must equal the sum of private savings [PS], corporate savings (retained earnings) [CS], and government savings [GS]....
(1) PI = PS + CS + GS....
The problem is simple: bailouts and stimulus plans are funded by issuing more government debt.... The added debt absorbs savings that would otherwise go to private investment.... [S]timulus plans do not add to current resources in use. They just move resources from one use to another.... I come back to these fundamental points several times below....
The Sad Logic of a Fiscal Stimulus: In a "fiscal stimulus," the government borrows and spends the money on investment projects or gives it away as transfer payments to people or states. The hope is that government spending will put people to work.... Unfortunately, there is a fly in the ointment.... [G]overnment infrastructure investments must be financed — more government debt. The new government debt absorbs private and corporate savings, which means private investment goes down by the same amount....
Suppose the stimulus plan takes the form of lower taxes... we can't get something for nothing this way either... lower tax receipts must be financed dollar for dollar by more government borrowing. The government gives with one hand but takes them back with the other, with no net effect on current incomes...
Fama's reasoning is that fiscal policies don't change private saving, but fiscal policies do change the government deficit, thus investment must change in an amount equal and opposite to the change in the government deficit. Fama's reasoning is dead wrong. Fama's reasoning is dead wrong for an elementary reason. The accounting identity that savings are equal to investment is true only under a particular definition of investment—one that counts unwanted growth in inventories as part of investment—and under a particular valuation of unexpected inventory accumulation—that which values unwanted inventory accumulation at its cost.
In general, the value of unwanted inventory accumulation can't be equal to its cost—the inventory accumulation is unwanted and unexpected, meaning that they tried to sell it at a normal price and failed, and it is now sitting in a corner of a warehouse somewhere. When Fama writes "bailouts and stimulus plans... absorbs savings that would otherwise go to private investment" he does not think that the rise in public spending is truly useful stuff while the fall in private investment is a decline in unwanted inventory accumulation—a decline in the amount of stuff made at high cost that firms could not sell and then must mark down in value.
This matters a lot because whenever unwanted inventories accumulate the next thing that happens is that incomes and savings drop. (i) NIPA-defined investment is equal to (ii) private savings minus the (iii) government deficit, so if (iii) changes and (ii) doesn't then (i) must change. But if that change in NIPA-defined investment is driven by unwanted inventory accumulation or unexpected inventory declines then private savings do change, and do change quickly and substantially.
Let's tell the story of how:
Suppose that it is Friday, January 2, 2009, and all of a sudden the federal government borrows some money—reducing savings—and buys some extra stuff. Savings is still equal to investment on January 2: savings went down because the government ran a bigger deficit but investment also went down because firms sold extra and so their inventories dropped.
What happens on Monday, January 5? Over the weekend the firms mark the value of the goods in their remaining inventory up: inventories are now scarce. They revisit their production plans. Sunday night they call some extra workers and tell them to show up on Monday—that they are expanding production because they are now short of inventories. So when Monday rolls around more people are at work. Thus incomes are higher on Monday than they were on Friday. And in all likelihood savings will be higher as well, for consumers on Monday probably won't raise their consumption spending by as much as their incomes rose. Maybe on Monday purchases will be back in balance with production, and there will be no more unwanted inventory changes. Maybe it will take until Monday January 12 before the change in inventories is back to its desired level. Maybe it will take until the third quarter of 2009, or perhaps 2010. But when the change in inventories does come back to its wanted level, production, employment, income, savings, and investment will all be higher than they were on January 1: the stimulus will have worked.
Yet at every point—on every single day—savings are equal to investment according to the accounting conventions of the National Income and Product Accounts. Fama's premise holds. His conclusion—that stimulus programs cannot work—doesn't. How can this be? The reason is that his conclusion has nothing at all to do with his premise. Whether there is complete crowding-out depends on circumstances—on how much of offsetting investment changes are unwanted and unexpected changes in inventories, and what the consequences of those unwanted and unexpected changes in inventories are for private savings. But whether there is complete crowding-out or not, savings always equals investment in the NIPA framework by construction, by definition.
Thus Fama's claim that "stimulus spending must be financed which means it displaces other current uses of the same funds..." rests on Fama's implicitly making one of two assumptions: either that stimulus spending does not lead to any surprise reduction in inventories, or that a surprise fall in inventories does not lead to any change in the flow of saving. Make either of these assumptions, and Fama's argument goes through—but it is those ancillary assumptions that Fama does not explicitly own up to that drive his conclusion, not his stated premise of the truth of the NIPA savings-investment identity.
But why should you make either assumption? Why would you ever assume that there can't be unwanted growth in inventories? Why would you ever assume that household incomes and saving do not change whenever firms' stocks of unwanted inventories grow ever larger?
The answer is that you never would—but that Fama does not know enough national income accounting to know that that he is making these two ancillary assumptions. He does not understand the identity he deploys as equation (1). He thinks that "investment" means "growth in the value of the capital stock." He simply does not understand what the NIPA investment concept is, or that what he thinks of as "investment" is not in general equal to savings.
All of this is part of the undergraduate sophomore economics curriculum. It is gone over again very quickly in graduate school—for example, David Romer (2006), Advanced Macroeconomics 3e, p. 224:
If one treats goods that a firm produces and then holds as inventories as purchased by the firm, then all output is purchased by someone. Thus actual expenditure equals the economy's output, Y. In equilibrium, planned and actual expenditure must be equal. If planned expenditure falls short of actual expenditure, for example, firms are accumulating unwanted inventories; they will respond by cutting their production...
These mistakes are, literally, elementary ones.
They were elementary when R.G. Hawtrey and the other staffers of the British Treasury made them in the 1920s.
They carry the implication not just that government cannot stimulate or depress the economy, but that no set of private investment or savings decisions can stimulate or depress the economy either, and thus that there can be no business cycle fluctuations from any source whatsoever—because every action that shifts savings or investment simply moves resources from one use to another.
What is extraordinary is that these mistakes are being rederived today, at the end of the 2000s—without any consciousness of their past or of the refutations of them made by past theory and history.
I think it is time to draw a line in the sand: no more economists who know nothing about the economic history of the world or the history of economic thought.
I, the ghost of Montagu Norman, have risen from my grave to say this.
Jeebus save us...
Fama's Fallacy II: Predecessors: Eugene Fama's predecessors in error. The "Treasury View." From G.C. Peden (2004), Keynes and His Critics, p. 80:
F.W. Leith-Ross to Sir Richard Hopkins and P.J. Grigg, 3 April 1929: Before the government can give increased employment it must obtain resources.... Unless the government is prepared to... bring about an inflation... [it] can only obtain [resources] by taxation or borrowing.... The proposal that we are examining is that all the money required is to be borrowed.... When the Government borrows, it enters the money market as a competitor with all other enterprises.... The resources from which the government must draw... are the savings of the people.... But it is precisely on these that industry relies on.... The competition of the Government with private traders by means of large Government loans would not (apart from inflation) increase the resources available for the employment of labour. It would only mean that a portion of these resources would be directed by the Government instead of being directed by private persons...
Fama, actually, is much worse than the British Treasury economists of the 1920s. They acknowledged that monetary policy could affect the level of employment--could do more than shift resources from one use to another. Fama's argument based on his misinterpretation of the NIPA savings-investment identity has the implication that monetary policy cannot affect the unemployment rate either.
See R.G. Hawtrey (1925), "Public Expenditure and the Demand for Labour," Economica 5, pp. 38-48...
Fama's Fallacy, Take III: "... >...Ummm... Greg Mankiw writes:
Fama on Fiscal Stimulus: Eugene Fama is a stimulus skeptic: In fact, he is even more skeptical than I am. I am willing to concede that many Keynesian effects work in the short run, although I prefer monetary policy to fiscal policy and, within fiscal policy, I prefer the use of tax instruments to government spending as a tool for short-run demand management. By contrast, I read Fama's article as a largely wholesale endorsement of the classical model with complete crowding out...
No, Greg. It's not an endorsement of any model. It's just a mistake. Fama mistakes the NIPA savings-investment accounting identity for a behavioral relationship that constrains the behavior of investment: when the government deficit goes up, Fama says, private investment must go down by the same amount.
The complete crowding-out argument is different: it is that the velocity of money is completely interest-inelastic, so when government purchases rise that triggers a rise in interest rates large enough to discourage investment and exports, and the sum of the discouragement equals the rise in government purchases. It is something that happens in equilibrium because prices move to make it so. In Fama there are no prices moving. There is just an accounting identity.
When the government deficit goes up, private savings could go up by more—and private investment could increase. Private savings could go up by less—and private investment would fall by less than the rise in the government deficit. Private savings could remain unchanged. Or private savings could fall. Determining which of these is most likely to happen would require a model of the economy of some sort—and Fama does not have one: all he has is an accounting identity that he does not understand...
Fama's Fallacy IV: The Decline of Chicago: Note to Self: How to make the "crowding out" argument the intellectually coherent way.
Milton Friedman (1972), "Comment on the Critics," Journal of Political Economy 80:5 (September-October), pp. 914-5 http://www.jstor.org/stable/pdfplus/1830418.pdf: "'I do not share the widespread view that a tax increase which is not matched by higher government spending will necessarily have a strong braking effect on the economy...
...True, higher taxes would leave taxpayers less to spend. But this is only part of the story. If government spending were unchanged, more of it would now be financed by the higher taxes, and the government would have to borrow less. The individuals, banks, corporations or other lenders from whom the government would have borrowed now have more left to spend or to lend-and this extra amount is precisely equal to the reduction in the amount available to them and others as taxpayers. If they spend it themselves, this directly offsets any reduction in spending by taxpayers. If they lend it to business enterprises or private individuals—as they can by accepting a lower interest rate for the loans the resulting increase in business investment, expenditures on residential building and so on indirectly offsets any reduction in spending by taxpayers.
To find any net effect on private spending, one must look farther beneath the surface. Lower interest rates make it less expensive for people to hold cash. Hence, some of the funds not borrowed by the Federal government may be added to idle cash balances rather than spent or loaned. In addition, it takes time for borrowers and lenders to adjust to reduced government borrowing. However, any net decrease in spending from these sources is certain to be temporary and likely to be minor.
To have a significant impact on the economy, a tax increase must somehow affect monetary policy—the quantity of money and its rate of growth. (Newsweek, January 23, 1967, p. 86)....
Why "certain to be temporary"? Because the leftward shift in the IS curve is a once-for-all shift.... Put in monetarist terms, the lowered interest rate resulting from the federal government's absorbing a smaller share of annual savings will reduce velocity; the transition to the lower velocity reduces spending for a given money stock....
Why "likely to be minor"? Because the monetarist view is that "saving" and "investment" have to be interpreted much more broadly... that the categories of spending affected by changes in interest rates are far broader than the business capital formation, housing construction, and inventory accumulation to which the neo- Keynesians tend to restrict "investment." Hence, even a fairly substantial tax increase will produce only a minor shift in the IS curve....
Of course, the terms "temporary" and "minor" are highly imprecise. We get closer to a rigorous statement by comparing the changes resulting from a reduced or increased deficit without any change in monetary growth with those that result when a change in the deficit is matched by a dollar-for-dollar change in monetary growth.... [A] deficit financed by borrowing... [is] a once-for-all shift to the right in the IS curve, a higher interest rate, a higher velocity, and a higher level of spending for a given monetary growth path.... [F]inancing the deficit by creating money... shifts the LM curve to the right.... But this is not a once-for-all shift. So long as the deficit continues, and continues to be financed by creating money, the nominal money stock continues to grow and the LM curve (at initial prices) continues to move to the right. Is there any doubt that this effect must swamp the effect of the once-for-all shift of the IS curve?...
We may put this point differently. Assume a one-year increase in the deficit, with the budget then returning to its initial position. If this is financed by borrowing from the public with no change in monetary growth, then, in the most rigid Keynesian system, the IS curve moves to the right and then back again; real and nominal income rise for one year, then return to their initial values. If the one-year increase in the deficit is financed by creating money, the LM curve moves to the right as well, and stays there after the IS curve returns to its initial position. If prices remain constant, real and nominal income stay at a higher level indefinitely. If, as is more reasonable, prices ultimately rise, real income may return to its initial level, but nominal income will stay at a higher level indefinitely. Surely, to paraphrase a remark of Tobin's in another connection, the monetary effect is "alchemy of a much deeper significance" than the fiscal effect...
For Friedman, the NIPA savings-investment identity is the prelude to the analysis: the meat of the analysis involves going deeper by:
- arguing that savings and income levels will adjust so that the economy will quickly move to a point at which unwanted inventory accumulation is zero (that's the "IS curve").
- analyzing the combination of possible values for interest rates and output levels at which unwanted accumulation is zero (that's the shape and position of the "IS curve").
- assessing how the changing financial asset supplies and demands in the economy pick out a particular point on the IS curve (that's the "LM curve").
For Fama, the NIPA savings-investment identity is the completion of the analysis—hence he gets driven to the conclusion that not just fiscal policy via the government deficit but monetary policy via open market operations has no effect on employment and output as well.
This makes me think I should finish writing up one of the talks that I gave in Singapore—the point of which was that Chicago economists today are profoundly ignorant of what the Chicago School of economics—the school of Friedman and Stigler—believes...
Fama's Fallacy V: Are There Ever Any Wrong Answers in Economics?: Montagu Norman here, back from my grave once again. This time it is Greg Mankiw whose words have summoned me...
One thing that used to give me nightmares—and that provoked several of my nervous breakdowns—was how you could never get any economist (except for John Maynard Keynes) to take a definite position. They were always "on the one hand—on the other hand." This was what led Harry Truman in later days to wish for a one-handed economist, a wish that has never been fulfilled—there is in fact a picture of Barack Obama's economic advisor Christina Duckworth Romer in Time (or is it Newsweek?) showing her with four hands...
The "on the one hand—on the other hand" nature of discourse raises the question of whether in economics—a "science" where there is enormous intellectual and ideological and political disagreement about how the world works—there can ever be any wrong answers?. I believe that there can be wrong answers in economics, because examinations in economics tend to take a particular form: instead of asking (i) "do expansionary fiscal policies increase output and employment?" we ask (ii) "in models where there are idle resources and high unemployment, do expansionary fiscal policies increase output and employment?" (ii) is a question about a particular class of models of the economy, and so has a definite right answer—"yes, in that class of models they do"—and a definite wrong answer—"no, in that class of models they don't."
Eugene Fama claimed that "when there are idle resources—unemployment" expansionary fiscal policies had no effect in models in which the NIPA savings-investment identity:
investment = (private savings) - (government deficit)
held.
Now the NIPA savings-investment identity holds in all models—it is, after all, an identity, true by definition and construction. And every single model that has been built in which there is a possibility of high unemployment and idle resources is a model in which fiscal policy works because increases in government spending lead to unexpected declines in inventories and unexpected declines in inventories lead to firms to expand production, which leads to increases in income and saving.
I would, therefore, say that Fama's claim is "wrong". Not only does it not hold in all models in the class, it does not hold in any models in the class.
Greg Mankiw disagrees:
Greg Mankiw's Blog: Fama's arguments make sense in the context of the classical model... presented in Chapter 3 of my intermediate macro textbook.... I would go on to the Keynesian model.... But whether one leaves the classical model behind to embrace the Keynesian model is a judgment call...
Mankiw thinks that Fama is not wrong but is, rather, making a "judgment call."
But Mankiw writes in his chapter 3 that the classical model "assume[s] that the labor force is fully employed." And so Greg gets himself into Cretan Liars' Paradox territory here: Fama says that there is high unemployment and idle resources, while Mankiw says that Fama is not wrong because he makes sense as long as the labor force is fully employed and there are no idle resources.
Is Mankiw's answer here a "wrong" answer, or is he too making a "judgment call"? I seek an empirical test. I seek a Harvard undergraduate to take Greg Mankiw's course this spring, to write the following in an appropriate place:
the classical model of chapter 3 shows us that expansionary fiscal policies have no effect on output even where there are idle resources—unemployment.
and to report back on the reaction of the course instructors...
#shouldread
