Does Government Spending Boost the Economy?

A recent article in Business Insider by Jim Edwards offers putative “Proof That Government Spending Cuts Hurt Economic Growth.” He even goes so far as to claim that “war is good (economically).” In this article, I’ll explain what’s wrong with this popular and age-old fallacy.

First, I want to point out something quite amusing. Edwards relies on a Financial Times story that presents a series of charts produced with data from the Bureau of Economic Analysis (BEA). Here is one of the charts, along with Edwards’s description:

This chart, from the FT’s Matthew Klein based on data from the BEA, seems to show that government has a pretty straightforward effect on GDP. When spending goes up, it adds to economic growth. When it goes down, it subtracts from it and hobbles the economy:

Edwards seems to think that the above chart shows at least a correlation between government spending and economic growth. After all, he wrote that the BEA chart “seems to show that government has a pretty straightforward effect on GDP.” But as Scott Sumner pointed out in amusement when he saw the article, the chart does nothing of the kind.

Look carefully at the legend. The various colored rectangles are different components of government spending. Specifically, the rectangles indicate how the change in each component — positive or negative — relates to the change in overall GDP. The black line is not GDP growth, but is instead the sum of the various components of government spending. In short, Matt Klein at the FT is telling us that if we take the BEA’s word for how much each component of government spending contributed to GDP growth in each quarter, then we can stack those numbers on top of each other and even add them up! Contrary to Edwards, the FT chart doesn’t “show” anything at all, except that the BEA each quarter announces how much various components of government spending contributed to, or subtracted from, GDP growth.

But let’s move past Edwards’s hilarious misinterpretation of the chart and get to the more fundamental issue. The problem with these ostensibly scientific and empirical measurements is that GDP itself is definedto include government spending. As they teach in any introductory macro class, the expenditure-based formula for GDP is

GDP = C + I + G + NX,

where C and I are private consumption and investment, G is government spending, and NX is net exports (gross exports minus gross imports).

Now we see the problem. Even if we set aside the serious theoretical and practical difficulties with the aggregation necessary to estimate these figures, we are still stuck with the fact that the above formula is an accounting tautology, not an economic theory. Yes, other things equal, an increase in government spendingG on the right-hand side will make GDP on the left-hand side increase dollar for dollar. The whole argument, however, centers on whether other things will remain equal.

For example, in a depressed economy with excess capacity, the typical Keynesian will say that an increase inG will cause private consumption and investment to increase also, so that a dollar of extra government spending will cause GDP to rise by more than a dollar — the famous Keynesian multiplier.

In contrast, the typical Austrian- or Chicago-school economist will say that an increase in G will tend to make private-sector spending fall by a greater amount, so that a dollar of extra government spending will cause GDP to fall. (We could get the confident support of free-market economists for this conclusion if we stipulate that the extra government spending is financed through higher taxes, which destroy more private after-tax income than they raise in extra revenue.)

Moreover, even if “total GDP” rises somewhat because of an increase in government spending, that wouldn’t be a good thing, because $10 million spent by politicians is not nearly as likely to channel resources to valuable uses as $10 million spent by private investors.

After this discussion, we can see why pretty charts from the FT showcasing government spending’s “contribution to GDP growth” quarter by quarter don’t really mean anything. It’s the same for the ex post “empirical” analyses that concluded that the Obama stimulus package “saved or created” such-and-such million jobs. The underlying models that generate these estimates assume a Keynesian world, and thus cannot test whether the Keynesian model is correct.

The critical yet missing piece of information in these analyses is the counterfactual, to know what the size of the economy and level of employment would have been in the alternate universe where government spending had taken a different course. From a naïve, “let the facts speak for themselves” perspective, the Obama stimulus package clearly hurt the economy. Remember that unemployment shot up higher with the stimulus than the Obama team warned people would occur without the stimulus.

The exact opposite happened with the so-called sequester. For example, the firm Macroeconomic Advisers, using a Keynesian model, predicted that the spending cuts would knock 1.3 percentage points off of second quarter 2013 growth, and 0.6 percentage points off of third quarter 2013 growth. Here’s what really happened:

It’s the mirror image of the Keynesians’ stimulus blunder. The economy grew faster with the sequester than the Keynesians said would occur without the “drag” of the spending cuts. In the case of the Obama stimulus, their excuse was, “Wow, the economy was worse than we realized, good thing we got that deficit spending in there, inadequate though it was.” In the case of the sequester, their response would have to be, “How about that, the economy was stronger than any of us realized. We dodged a bullet, since the sequester dragged down growth so much.”

In summary, we shouldn’t trust empirical “proof” that government spending boosts the economy, when the alleged evidence so often rests on a model that assumes as true the very issue under dispute. It is particularly absurd to argue that government spending on war makes us richer, because war doesn’t merely deploy scarce resources into unproductive lines — it actually destroys both equipment and workers.