August 10, 2010
Systematically Wrong

It’s a curious thing when a commentator can’t seem to get anything right. Such was the case with David Brooks in a column several weeks ago. One can only take it as testimony to the mind-warping powers of ideology. Take this paragraph:

The Demand Siders don’t have a good explanation for the past two years. There is no way to know for sure how well the last stimulus worked because we don’t know what would have happened without it. But it is certainly true that the fiscal spigots have been wide open. The U.S. and most other countries have run up huge, historic deficits. And while this has helped save public-sector jobs, we certainly haven’t seen much private-sector job growth. It could be that government spending is a weak lever to counter economic cycles. Maybe monetary policy is the only strong tool we have.

Nearly every sentence in this paragraph is false or misleading. Let’s take it sentence by sentence:

1. “The Demand Siders don’t have a good explanation for the past two years.” The “Demand Siders” actually have the only explanation for the past two years. The bursting of the housing bubble and ensuing stock market collapse together eliminated about $13 trillion of wealth. These losses, through a thing called the wealth effect, reduced consumer demand by an estimated $500 billion per annum. The bursting of the housing bubble also caused residential construction to contract, which is estimated to have cost the economy another $500 billion in annual consumer demand. And then there was the bursting of a bubble in nonresidential real estate, which caused nonresidential construction to plunge, costing the economy another $150 billion in demand. The total loss in annual demand is thus perhaps $1.2 trillion. The effect of a demand shock (as some may recall from introductory economics) is downward pressure on output and prices (in contrast to a supply shock, which would put upward pressure on prices), which is what we’ve seen. (Numbers from Dean Baker.)

All evidence appears to be that, in depressed, deflationary circumstances, fiscal stimulus is effective whenever, and to the extent that, it’s been tried. In a statistical study of fiscal expansions during the 1930s, for example, Almunia, Bénétrix., Eichengreen, O.Rourke. and Rua found that “fiscal policy, where applied, worked extremely well…, whether because spending from other sources was limited by uncertainty and liquidity constraints, or because with interest rates close to the zero bound there was little crowding out of private spending.” Also, Adam Posen, who’s studied Japan’s “lost decade” of the 1990s and early 2000s extensively, observes, “Fiscal policy works when it is tried,” but in the case of Japan successful stimulus has repeatedly been thwarted by policy error—fiscal contractions that have undone the positive effects of stimulus. Posen writes that “when the Japanese government paid for fiscal stimulus in 1995, it got economic growth,” but “when it mistakenly pursued fiscal austerity in most of the remainder of the 1992-97 period, it got economic contraction”—a pattern repeated in the early 2000s.

And consider present-day experience in many Asian countries. David Pilling of the Financial Times writes:

However much Asians trumpet the value of parsimony, their governments have been as bold as any in opening the fiscal sluices. One reason is the bitter memory of the 1997 Asian financial crisis when the International Monetary Fund imposed fiscal austerity on several Asian countries. Those measures are now almost universally seen as a blunder that unnecessarily exacerbated economic misery… Unlike in the west, there is little debate in Asia about how well the stimulus worked. It has been spectacular. Asian output is well above pre-crisis levels. HSBC is predicting growth for Asia ex-Japan of 8.6 per cent this year. Rather than contemplating more stimulus, authorities are trying to cool things down.

I’ve shown the following graphs from Barry Eichengreen and Kevin H. O’Rourke in a previous posting, but let’s look at them again:

World Industrial Production

World Trade

Is it really hard to see that the current downturn was probably as bad as the Great Depression? That the reason we aren’t now in a second Great Depression is government actions—bank bailouts, but also demand-side stimuli such as expansionary fiscal and monetary policy—undertaken on an unprecedented scale in many countries? And that ending stimulus now, before recovery is assured, is a very bad idea?

Now, no one doing the math expected a $787 billion stimulus package over two years to fill the output gap. Let’s run through the math. This is very back-of-the-envelope, but back-of-the-envelope is appropriate here since we’re dealing with general magnitudes.

Okun’s law says that for every 1 percent increase in unemployment above full employment, GDP will be 2 to 4 percent lower than potential GDP. Let’s take the optimistic end of the range and assume that, for every 1 percent increase in unemployment above full employment, GDP is 2 percent lower than potential GDP. And let’s call full employment 5 percent (the CBO’s estimate of the NAIRU is 5.2 percent). Shortly after the stimulus was enacted, and before stimulus funds hit the economy, consensus forecasts of unemployment worsened to about 10 percent (and of course unemployment eventually reached 10.5 percent). So let’s say unemployment was forecast to be 5 percent above full employment. That means the output gap would be $1.4 trillion (10% x $14 trillion) during the first year of the stimulus and somewhat less than that the following year. To reduce unemployment by 1 percent should then require about $280 billion of additional annual GDP (since 2 percent x $14 trillion = $280 billion). To reduce it by 5 percent and close the output gap should then require about $1.4 trillion of additional annual GDP.

The stimulus bill spent $787 billion over two years. A little less than $100 billion of that was a patch of the AMT, so let’s call it $700 billion. And about $300 billion was offset by cutbacks and tax increases at the state and local level, so the net stimulus was about $400 billion. About two-thirds of the stimulus was spending and about one-third of it was tax cuts. Spending has higher multipliers than tax cuts; so let’s say the “spending multiplier” was 1.5 and the “tax cut multiplier” was 1. And let’s assume that half the money was spent the first year and half the second year. The addition to annual GDP over what it would otherwise have been would then be approximately:

(1.5 x $133 billion) + (1 x $67 billion) = $267 billion

If we get a 1 percent reduction in unemployment for every $280 billion of additional annual GDP, we get an approximately 0.95 percent reduction in unemployment after one year and about 1.9 percent after two years. If we factor in growth of the labor force, we might wind up with a reduction in unemployment similar to the estimates of Moody’s Analytics and the CBO for the fourth quarter of 2010, diagrammed below.

So the stimulus, according to this analysis, shouldn’t have been enough.

What would have been enough? My back-of-the-envelope math says $1.4 trillion (assuming no AMT fix). Taking out the $300 billion offset for cutbacks and tax increases at the state and local level, the net stimulus would be $1.1 trillion. The addition to annual GDP over baseline would then be approximately:

(1.5 x $367 billion) + (1 x $183 billion) = $733 billion

The unemployment reduction over two years should then be a little over 5 percent ($1.47 trillion ÷ $280 billion ≈ 5).

Interestingly, $1.4 trillion is the number reportedly proposed by the Treasury Department and very close to the one suggested by Christina Romer ($1.2 trillion) when the stimulus proposal was being formulated. The watering-down of these proposals by political considerations may be one of the tragedies of our age. (See Ray Lizza’s New Yorker piece; also this.)

These particular figures are not, of course, to be taken too seriously. The point is simply to illustrate that, using a textbook rule-of-thumb like Okun’s law, we can see that the stimulus ought to have been closer to the ballpark of $1.4 trillion than that of $787 billion. The stimulus, though obviously helpful (even a godsend considering the alternative), should have been insufficient to close the output gap and return the economy to full employment.

So the “demand siders” have a full—indeed the only plausible—explanation for the last two years. But that’s enough for sentence 1. Let’s go on to sentence 2.

2. “There is no way to know for sure how well the last stimulus worked because we don’t know what would have happened without it.” There’s no way to know for sure, in the strictest sense, how well the last stimulus worked. This is quite an uninformative statement. Obviously what we have are estimates, plenty of them, from reputable sources. For example, there are these from Moody’s Analytics and the CBO:

Macroeconomic Advisers and IHS Global Insight (and thus all three major macroeconomic forecasting firms) have produced similar estimates.

If Brooks or anyone knows reasons why the providers these estimates would be biased, they should explain. Republicans speak favorably of the CBO whenever its analyses support their policies. The head of Moody’s Analytics is former McCain adviser Mark Zandi.

It’s true some analyses suggest little or no positive impact from the stimulus. But these are a minority. The most prominent of them is perhaps Cogan, Cwik, Taylor, and Wieland, which estimates multipliers about a sixth as large as those of the above sources. But the reason for the difference appears to be the parameterization. In particular, fiscal expansions don’t happen in a vacuum, but evoke a monetary policy response, since the Fed at some point will have to raise interest rates to stem inflation. The quicker rates rise after a fiscal expansion, the larger the crowding-out effect will be, and the smaller the positive effect of the stimulus on output and employment. So a lot depends on how monetary policy is modeled. Moody’s, the CBO, Macroeconomic Advisers, and Romer/Bernstein all assumed relatively easy monetary policy over an extended period. Macroeconomic Advisers, for example, tells us:

Currently the nominal federal funds rate is pinned at nearly zero, and our favorite monetary rule suggests that it should be negative. Hence, when we prepared our analysis of ARRA, we explicitly assumed the Fed would not raise nominal interest rates in response to the stimulus because it wanted to encourage the largest possible impact on the economy. Furthermore, since the stimulus reduced slack in the economy and so put upward pressure on inflation (or prevented further disinflation), the real interest rate fell in our simulations. This is opposite the “normal” response of real rates in most reduced form models with an implicit (or explicit) monetary rule, and works to magnify the multiplier effects — as intended by the Fed.

In other words, because (a) the “correct” federal funds rate is negative, so that the federal funds rate will not likely rise for a considerable time, and (b) a fiscal expansion puts upward pressure on prices, the real interest rate

r = i - π   (real rate of interest = nominal rate of interest - inflation rate)

is destined to fall (or at least be lower than it otherwise would be). This creates a window in which a fiscal expansion is likely to be especially effective, i.e., with minimal crowding out.

By contrast Cogan, Cwik, Taylor, and Wieland tell us, “our analysis allows for the return to a stabilizing monetary policy after one or two years.” With possible deflation now on the horizon and the Fed making noises about engaging in unconventional monetary policy, it seems clear which parameterization was the more reasonable.

3. “But it is certainly true that the fiscal spigots have been wide open.” Not sure what Brooks means by “wide open.” If he means opened up to some degree, he’s right. There has been more federal spending. But the spigots haven’t been wide open if by “wide open” Brooks means unconstrained. The fiscal stimulus may or may not have been all that Obama could have gotten from Congress, but mathematically, as shown above, it wasn’t enough to close the output gap.

4. “The U.S. and most other countries have run up huge, historic deficits.” It’s true that the U.S. has run up huge deficits, but Brooks makes it appear that they’re entirely the result of the stimulus, which is anything but true. Much bigger contributors to the deficits are the Bush tax cuts (which Brooks supported) and the recession itself (i.e., the loss of tax revenues and automatic increases in, e.g., unemployment insurance and food stamps). Besides which, the whole point of the stimulus is to inject demand into a demand-starved economy. That by definition means deficit spending. Since stimulus is temporary, it doesn’t add significantly to long-term debt (unlike the Bush tax cuts that Brooks supported). Indeed by stimulating growth it improves long-run budgetary prospects over what they would otherwise have been.

5. “It could be that government spending is a weak lever to counter economic cycles. Maybe monetary policy is the only strong tool we have.” Most economists, including most “demand siders,” would agree that monetary policy is the best stabilization tool we have. Unless monetary policy is at the zero lower bound, in which case it’s largely ineffective. There’s nothing now to be done about short-term interest rates. They’re zero. And the Fed’s capacity to influence long-term rates appears to be minimal (see Krugman). Really this is the whole point behind arguments for stimulus. We’re in a liquidity trap, i.e., incomes and wealth are low, so demand is low and investment is unprofitable. That means that almost however much money the Fed puts into the economy (and the economy is now awash in cash), it doesn’t get used productively—a situation that will continue until demand recovers.

Is it really that hard to grasp that people and businesses don’t necessarily spend their money when (a) households are trying to recoup their lost wealth and (b) there are relatively few profitable investment opportunities?

How can demand recover? One way would be to leave the economy to its own devices (a “liquidationist” approach). After a long, savage depression during which vast amounts of capacity (including labor, i.e., lives) are wasted, wealth and incomes should start to recover. But luckily we have a better way. When demand is depressed, the government can step in and bolster demand through temporary deficit spending. It doesn’t adversely impact long-terms debt, since the spending is temporary, and in any case stimulating growth now makes future revenues higher than they would otherwise be. It’s not inflationary since, with demand depressed, pressure on prices is downward. There’s minimal crowding out of investment since in a liquidity trap there’s little competition for loanable funds. And the historical and contemporary evidence is that stimulus works when, and to the extent that, it’s tried.

Further down the column Brooks writes:

These days, debt-fueled government spending doesn’t increase confidence. It destroys it. Only 6 percent of Americans believe the last stimulus created jobs, according to a New York Times/CBS News survey. Consumers are recovering from a debt-fueled bubble and have a moral aversion to more debt.

(People, of course, believe the stimulus failed because unemployment is nearly 10 percent. And they don’t know that absent the stimulus unemployment would almost certainly be much worse.)

Brooks oddly conflates government debt and household debt. It’s all just “debt.” And Americans now have a “moral aversion” to it and therefore believe spending in general should stop. It’s unclear when Americans got this religion. They showed no “moral aversion” to debt during the housing bubble, or, as far as I can tell, ever:

And somehow now, in a depressed economy, when deficit spending is appropriate (for once), they suddenly have a “moral aversion” to it? To the extent that Americans have a “moral aversion” to deficit spending, I suspect it’s because commentators and politicians (many of whom supported the policies that caused the debt problem) keep nattering on and on about it, and falsely imply that temporary stimulus is a major contributor to long-term U.S. debt.

But actually debt and deficits aren’t major concerns of Americans. We can look at polls:

(See also this.)

The fact that Americans are now deleveraging has an obvious explanation: they have no choice, moral or otherwise:

Brooks is concerned about confidence which he claims government debt destroys. And what about mass unemployment? Does that affect confidence? Most Americans don’t understand that there’s a short-term tradeoff between deficit spending and unemployment. But if given their choice between policies that lower deficits and policies that lower unemployment, which would they choose? I wonder.

Still further down the column, Brooks writes:

You can’t read models, but you do talk to entrepreneurs in Racine and Yakima. Higher deficits will make them more insecure and more risk-averse, not less. They’re afraid of a fiscal crisis. They’re afraid of future tax increases. They don’t believe government-stimulated growth is real and lasting. Maybe they are wrong to feel this way, but they do. And they are the ones who invest and hire, not the theorists.

Yes, you can talk to entrepreneurs. And here’s what they say:

The National Federation of Independent Business (NFIB) has regularly surveyed small business opinion since 1973. The above statistics are from its June report. Here are the same statistics over a longer period:

Brooks writes as if he’s surveyed business opinion and is reporting his findings: that their main concerns are government debt and taxes. Obviously he’s done nothing of the kind. Taxes are a perennial concern of businesses, to be sure, though they appear to be less of a concern now than they’ve been over most of the last 25 years. And currently they’re superceded by sales, as one would expect when consumer demand has plummeted. Government debt, on the other hand, seems not to be on the radar at all.

Though businesses have their special interests, the bottom line is clear. The report comments:

Yet Brooks is adamant that businesses are concerned about government debt. Why? Does he so strongly believe that they must feel this way that he went ahead and asserted it without bothering to check whether it was true?  Curious, in any case, that entrepreneurs would be so concerned about government debt when those with most at stake in U.S. government solvency, bond traders, aren’t. The ten-year Treasury rate is now (August 10) 2.78 percent. Bondholders are demanding a 2.78 percent return to lend money to the U.S. government for ten years!

In an apparent attempt to seem empirical, Brooks cites a study by Lauren Cohen, Joshua Coval and Christopher Malloy of Harvard that found that increased government spending in certain congressional districts dampened corporate sector investment and employment. Brooks, invoking his theme of epistemic modesty, says, “You wish somebody could explain that one to you before you pass on more debt burdens to your grandchildren.” There are three problems with Brooks’s reference to the study. First, the study is at the local (congressional district) level, not the national level, which is the level relevant to the stimulus debate. Second, it uses data from the non-liquidity-trap years of 1968-2008 and thus doesn’t pertain to conditions relevant to stimulus in a depressed economy, which is the question we now face. And third, and most importantly, what the study actually finds is consistent with “demand siders’” views about stimulus. The authors write:

...the coefficient on firms in states with low unemployment is -0.016. For firms in states with high unemployment, the coefficient is 0.024 larger, which is sufficient to reverse the effect entirely (even considering the main effect of High Unemployment itself). This result can be interpreted as providing evidence consistent with the view that government stimulus crowds out private sector employment when the economy has little slack in the labor market, but does not when the economy is experiencing significant slack in the labor market” (italics added). 

This, of course, is an argument most stimulus advocates would make. Indeed the authors note, “…consistent with Keynes’ view that  crowding out should only occur under conditions of full employment, we find a weaker firm response to spending shocks when state-level employment is at or below its long-term historical average.” What they find, as noted in the above quote, under conditions high unemployment is a reversal of the crowding out effect they find under conditions of greater vibrancy.*

It’s not that the study supports Keynesian stimulus. The data don’t include the depressed economic conditions in which the question of Keynesian stimulus arises, so it doesn’t address the issue. Either Brooks didn’t read the article and/or he’s unfamiliar with the arguments for stimulus. I suspect both.

Modern-day Keynesians argue that we should spend when the economy is depressed and save when it’s healthy. (It’s not very complicated.) That prescription was followed by Keynesian-like economists who held power in the early 1990s. The result, by decade’s end, was surpluses as far as the eye could see, soon to be replaced by deficits as far as the eye could see, thanks to massive tax cuts supported by Brooks. Brooks and other conservatives, by contrast, apparently believe that we should save when the economy is depressed and spend when it’s healthy. This is simply not economics. You will not find this prescription in any macroeconomics textbook, including those by conservative economists. 

Brooks, an all-purpose pundit, must opine about the economy, though he obviously lacks a background in the subject. He finesses the latter fact by fitting economics to his narrative, the narrative that is his livelihood. What is this narrative? It seems to combine two main themes: that Americans are skeptical of big government, and that social reality is too complex for policy makers to adequately grasp and so their grandiose policy schemes tend to fall short. Most Americans lack Ivy League credentials or high IQs. But their skepticism of big government, and of the grand schemes of policy makers, shows a deep native wisdom. (Or at least it might be if they had it.)

So how should government proceed? Well, some modest policy-making is okay. Brooks’s advocacy of that makes him a “moderate” when compared to some of his right-wing brethren. But the third leg of Brooks’s narrative is the market. Policy makers, with their high IQs and impressive academic credentials, tend to be overconfident in their abilities to manage society. So intellectual modesty bids that we not tinker too much with the institution (the market) that is the fount of modernity and that centuries of evidence tell us embodies a collective wisdom vastly greater than that of any individual woman or man. 

The only problem with this is that that’s not what centuries of evidence tell us. Much of the last 300 years has been economically highly unstable, with booms, busts, bubbles, market failures, and depressions the most prevalent manifestations of our dynamic capitalist economy. It’s only in the post-World War II era, as the government has taken a bigger role in the economy, that large-scale instability has been more or less tamed.

One might have thought the third leg of this three-legged stool was knocked out in the fall of 2008. But now that the stimulus and other measures have rescued the economy from the abyss (though not restored it to full health), it becomes possible again to imagine, with appropriate blinders on, that the “market” is the self-correcting marvel people thought it was, that the insights gained over 70 years about the potential positive role government might play in depression-like circumstances don’t apply, and that all that’s needed to bring the economy back to health is some tinkering around the edges to raise confidence. But what Americans neeed isn’t “confidence.” (Confidence to what? Spend what they don’t have? Yes, that would be helpful.) What they need is income. And they’re not going to get it if we blunder into a Japanese-style lost decade. Brooks’s narrative is a force compelling us to do just that.

*The authors point to a likely “crowding out” effect—not crowding out in the traditional sense of increased competition for loanable funds driving up interest rates (since here we’re just talking about reallocations of government spending from one congressional district to another, not an absolute increase in government spending), but the possibility of increased competition for factors of production. One might reasonably expect government spending to entail such increased competition when labor markets are tight.