Tim Besley and Andrew Scott Have a Very Good Idea for Britain
Ten Pieces Worth Reading, Mostly Economics: February 27, 2010

Project Syndicate: A Stimulus Opponent Who Can Actually Find His ---!

Stimulus Arithmetic

In the Wall Street Journal on February 23, Robert Barro made an intelligent argument against the American fiscal stimulus—the ARRA. It is a great relief after wading through the works of those who claim, one way or another, that the basic principles of economics set out by Say and Bastiat make it impossible for government decisions to spend. (They would, if they thought about it for even a minute, realize that their arguments also entail the conclusion that nobody else's decisions to spend can alter the flow of economic activity either, and hence that recessions simply do not happen. Which raises the question of why the unemployment rate has risen from 5% to 10%, but I digress.) Barro is, thankfully, not one of that crew. So my first desire is to wind up Barro and turn him loose to deal with those who stopped reading the economics literature before 1890, and not only never understood Milton Friedman[1] but never understood Friedman’s predecessors like Irving Fisher and Knut Wicksell.

However, I think Barro has gotten some key things wrong. First, I believe that Barro makes an error of logic in assessing how his own view of the situation applies to the question of what macroeconomic policy should be.

Barro writes, in “The Stimulus Evidence One Year On,”[2] that he:

estimate[s] a spending multiplier of around 0.4 within the same year and about 0.6 over two years.... [T]he [tax] multiplier is around minus 1.1.... [Thus] GDP would be higher [because of the ARRA stimulus] than otherwise by $120 billion in 2009 and $180 billion in 2010...

and by $60 billion in 2011.

That is roughly 1.3 million more people employed in America in 2009, 1.9 million more people employed in 2010, and 0.7 million people employed in 2011. Suppose that what the government spent money on is worth to us on average 2/3 of what private-sector spending is worth. Then according to Barro we spent $600 billion on the ARRA and got $400+$120+$180+$60 billion = $760 billion worth of goods and services in return, for a net social profit of $160 billion. And it is not as though that $160 billion would have been offset by a loss of leisure time on the part of those who are not but would have been employed. The cyclically-unemployed do not place a high value on their lost leisure.

Only if you think there are additional large costs lurking down the road—that the ARRA has destabilized price expectations and set in motion a damaging and destructive spiral of deflation, or that the ARRA has used up America's debt capacity and so interest rates have spiked and amortizing the debt will be very costly—does the social profit turn negative. And neither of those things have happened. As the London Economist’s “Demcracy in America” correspondent wrote on Wednesday:

You can argue that private actors will spend the money in ways that generate more employment than government.... It's pretty hard to make this argument at a time when banks are not lending... because they see few promising opportunities.... You can argue that government programmes generally take years to get underway, and by the time the spending gets going, the economy will already be in recovery... this argument looks much weaker now than it did a year ago: the recession has been far deeper than expected, we seem to be having a jobless recovery, and unemployment looks set to stay above the 5% “full employment” level for many years...

The long-term nominal and real Treasury rates continue to be absurdly low, so much so that I rub my eyes whenever I see them. Just today the Treasury auctioned $32 billion more of seven-year notes at a yield of 3.08% per year. And the market forecast of inflation—the spread between TIPS and normal Treasuries—remains extremely well behaved. As Martin Feldstein wrote yesterday, the market continues to expect consumer price inflation to average only 2.5% per year over the next several decades.[3]

Thus I really do not understand the logic behind Barro’s last paragraph in his op-ed, in which he claims:

The fiscal stimulus package of 2009 was a mistake. It follows that an additional stimulus package in 2010 would be another mistake...

Second, I think Barro makes an error of analysis in reading the current situation. I think Stanford’s Bob Hall[4] has a better read on what is going on with respect to the current sizes of multipliers:

With allowance for other factors holding back GDP growth during those [years of total] wars, the multiplier linking government purchases to GDP may be in the range of 0.7 to 1.0... but higher values are not ruled out.... Multipliers are higher—perhaps around 1.7—when the nominal interest rate is at its lower bound of zero, as it was during 2009...

The problem, I think, is that Barro tries to use the total war years of the twentieth century to

realistically evaluate the stimulus... the main results come from fluctuations in defense outlays associated with major wars such as World War I, World War II and the Korean War...


the defense-spending multiplier can be precisely estimated...

This suffers from the standard economist's problem of looking for one's lost keys under the lamppost because the light is better there. Yes, the total war defense-spending multiplier can be relatively precisely estimated. But we are not interested in what the multiplier is when the unemployment rate is 3%, we are interested in what the multiplier is when the unemployment rate is 10%. And we are not interested in what the multiplier is when the government is taking all kinds of other steps to diminish consumption and boost private savings via rationing and patriotism-based bond drives, we are interested in what the multiplier is under more normal conditions.

Third, Barro characterizes the stimulus bill as a two-year $600 billion increase in government purchases. But about half of the stimulus money spent to date is on the tax and transfer side, and about a quarter is direct aid to states which enables them not to raise taxes during this recession. It seems to me that Barro should be weighted-averaging his spending multiplier of 0.6 and his tax multiplier of 1.1 to get an ARRA multiplier of 0.9—in which case our social profit is not $160 billion but rather $340 billion, and we should certainly do this again, and again, and again. (Until, that is, there are signs that additional stimulus may start to threaten price-level or debt-management stability, or until unemployment falls far enough to make Barro’s multipliers overestimates.)

Fourth, Barro complains that because Christina Romer has “not [carried out] serious scientific research... on spending multipliers...” he “cannot understand her rationale for assuming values well above one...” To say that policymakers should rely only on their own personal research to formulate policy seems to me simply bizarre.

Fifth, Barro assumes that spending in 2009-2010 is then offset by levying equal taxes in 2011-2012, claiming that “the timing of future taxes does not matter.” But it does. It matters very much. At the moment the U.S. Treasury can borrow at a real interest rate of zero for five years—and shove all of five-year inflation risk onto the lender at that. Time preference means that the $600 billion addition to the debt today that Barro sees as the cost of stimulus is—because of the easy terms on which the Treasury can finance things right now—not nearly as burdensome as a demand to pay $600 billion now would be. We get the goods now. But the costs are delayed until later, when we will in all likelihood be richer and feel the costs less severely.

Sixth, when future taxes will be levied to amortize the added debt induced by the stimulus, they will be levied at a time at which nominal interest rates will not be stuck at their current floor of zero. The Federal Reserve will then be able to ease monetary policy—reduce interest rates—to offset the fiscal drag. There will be no lost production and employment through Keynesian channels. So I do not see why Barro believes that, although the stimulus boosts employment now, amortizing the stimulus must inevitably reduce employment at some point in the future.

[1] Who argued that the stimulative effects of expansionary fiscal policy were, as long as interest rates were at more-or-less normal levels, “certain to be temporary and likely to be minor”—but who agreed that the stimulative effects on production and employment were there.

[2] http://online.wsj.com/article/SB10001424052748704751304575079260144504040.html

[3] http://www.project-syndicate.org/commentary/feldstein20/English

[4] http://muse.jhu.edu/journals/brookings_papers_on_economic_activity/summary/v2009/2009.2.hall.html

February 25, 2010: 1507 words