Krugman writes today:

The biggest problem facing the Obama plan, however, is likely to be the demand of many politicians for proof that the benefits of the proposed public spending justify its costs — a burden of proof never imposed on proposals for tax cuts.

This is a problem with which Keynes was familiar: giving money away, he pointed out, tends to be met with fewer objections than plans for public investment “which, because they are not wholly wasteful, tend to be judged on strict ‘business’ principles.” What gets lost in such discussions is the key argument for economic stimulus — namely, that under current conditions, a surge in public spending would employ Americans who would otherwise be unemployed and money that would otherwise be sitting idle, and put both to work producing something useful.

Ffirst let us note that Krugman is implicitly backing wasteful government infrastructure spending. Krugman is arguing for the abandonment of any “proof that the benefits of the proposed public spending justify its costs” and instead wants stimulus decisions made solely on whether it “would employ Americans who would otherwise be unemployed.”

On this point Harvard professor Greg Mankiw writes:

In thinking through the fiscal policy options and their implications, it might be useful to compare a few hypothetical, fanciful scenarios. Suppose that the federal government borrows some money and then…

Case A: uses the money to give a lump-sum payment (such as a tax rebate) to Joe Average, who chooses to spend his free time sitting at home watching Mork and Mindy reruns.

Case B: uses the money to hire Joe to sit at home and watch Mork and Mindy reruns.

Case C: uses the money to hire Joe to sit at home and watch Family Feud reruns, which Joe does not enjoy quite as much as Mork and Mindy.

In all the cases, Joe will spend some of the money he gets on consumer goods and services, leading to a Keynesian multiplier. But those knock-on effects are the same in the three cases, so we can put those aside for now.

Let’s begin by comparing cases A and B. These two scenarios are identical in terms of final allocations and economic welfare. Joe is doing the same thing, and all the money flows are the same. But note that the macroeconomic statistics would be different. In Case B, Joe is employed producing a government service. If we used standard data to compare Case B with Case A, Case B would show more hours worked and a higher Gross Domestic Product.

Now look at Case C. It has the same employment and GDP as Case B, but welfare is strictly lower. Joe is, after all, less happy watching Family Feud. Comparing Case C with Case A, therefore, we see greater employment, greater GDP, and lower welfare.

Usually, GDP is a reasonable proxy for economic well-being, so more is better, but that is not true in this example. Part of the problem here is that GDP includes government purchases at cost. If the government hires people to produce stuff that is worthless, that stuff is included in GDP just as much as if the government buys something valuable. When calculating GDP, the national income accountants do not pass judgment on the social utility of government spending. Anyone concerned with economic well-being has to go beyond thinking about GDP.

The moral of the story: If the government spends a fiscal stimulus package on goods and services without much public value (as in Case C), it could well stimulate the economy as measured by macroeconomic aggregates but leave the participants in the economy worse off (compared with a feasible alternative, Case A). Avoiding this trap requires that the government spend taxpayers dollars only those items that pass a strict cost-benefit test. That is hard to do quickly. Willy-nilly spending is a good way to stimulate the economy only if the outcome is judged by the wrong metric.