Russ Roberts’ Congressional Testimony on the Stimulus

Russ Roberts testified before Congress on the effects of the economic stimulus package.

… Over the last two years, the American Recovery and Reinvestment Act of 2009 has injected over half a trillion dollars into the US economy in hopes of spurring recovery and creating jobs.

The results have been deeply disappointing. Job growth has been anemic while our deficit has grown, limiting our future policy options. Fourteen million workers are unemployed. The unemployment rate among African Americans is over 15%. This is an American tragedy.

What went wrong? Why were the predictions so inaccurate?

There have been two explanations. One is that the economy was in worse shape than we realized. The only evidence for this claim is circular—the standard Keynesian models under-predicted unemployment.

I agree with this, and it’s why Krugman’s repeated claims that there wasn’t enough stimulus really bug me. Krugman acts as though there’s no empirical doubt about the effects of fiscal stimulus, even though there’s no empirical proof fiscal stimulus works. He treats the failure not as any possible evidence against fiscal policy but as definitive proof of insufficient fiscal policy.

I prefer a simpler explanation: the models that justified the stimulus package were flawed. Those models were broadly based on the Keynesian notion that the road to recovery depends on spending. In the Keynesian worldview, all spending stimulates. Somehow, subsidizing university budgets in the Midwest or paying teachers in West Virginia helps unemployed carpenters in Nevada. That may be good politics. It’s lousy economics.

This part gives me pause. Giving more money to teachers in West Virginia allows them to either spend or invest more, and that ultimately helps somebody somewhere. My concern with fiscal policy is Friedman’s, that it doesn’t actually add to spending and investment, but just displaces private spending and investment. But if in fact it is adding spending and investment, then paying teachers here helps someone else besides them, even if it’s not specifically Nevada carpenters.

This isn’t the first time the Keynesian worldview was wildly inaccurate in predicting the impact of changes in government spending. Look at the beginning and end of WWII.

Keynesians frequently argue that the military spending on WWII ended the Great Depression.

Certainly unemployment fell to nearly zero because of the war. But did the war create an economic boom? There was a boom for the industries related to the war. But there was little prosperity for the rest of the country. The war was a time of austerity. Government spending didn’t have a multiplier effect on private output. It came at the expense of private output.

What about the end of the war, when government spending plummeted?

Paul Samuelson, a prominent Keynesian, warned in 1943 that when the war ended, the decrease in spending combined with the surge of returning soldiers to the labor force would lead to “the greatest period of unemployment and industrial dislocation which any economy has ever faced.” He was not alone. Many economists predicted disaster.

What happened? Government spending went form 40% of the economy to less than 15%. And prosperity returned to America. Unemployment stayed under 4% between 1945 and 1948. There was a short and mild recession in 1945—while the war was still going on. But the economy boomed when government spending shrank and price controls were removed.

I am constantly mystified by the folks who think WWII ended the Depression. We took about ten million men who were unemployed and put them in uniform, then restructured the economy almost completely on command-and-control lines. A necessary task, I would agree, but not exactly an economic recovery–certainly not one anyone would like to emulate as a solution to our current slump, especially with a return of ration cards for gasoline, eggs, shoes, etc.

We are told that the failure of the current stimulus proves it simply wasn’t big enough to get the job done. But it is equally plausible that the opposite is true—that government intervention in the economy prevented the recovery.

I am sympathetic to the politicians, particularly presidents, who face tremendous pressure to be seen to be doing something. After all, look at George H. W. Bush–the economy was in fact recovering by the election of 1992, but he was crucified for appearing to not be doing enough to fix the economy (mea culpa, I voted against him for that reason). But I am not sympathetic to people who don’t fact that pressure but still insist that the government had to act, that it was with certainty an absolutely necessary thing to do to save the economy. Really? And how do we know that? What’s the evidence for it? As we saw a couple days ago, leading Keynesian Paul Krugman says we’ve never actually tried real Keynesian stimulus. So without real testing, how can we be so sure it’s necessary?

The truth is that our knowledge of the complex system called the economy is woefully inadequate and may always remain that way. We ask too much of economics.

Economics is an explanatory science, not a predictive one, just like evolutionary biology. For some reason, people think that’s ok with biology, but mock economics for it.

Even our best attempts to measure the job impact of the stimulus spending make this clear. In November of 2010, the CBO estimated that the stimulus had created between 1.4 and 3.6 million jobs. Not a very precise estimate.

But even this estimate was more of a guess than an estimate. The CBO estimates didn’t use any of the actual employment numbers after the stimulus was passed. Instead the CBO based its “estimates” on pre-stimulus relationships between government spending and employment, relationships that failed to predict the magnitude of our current problems.

The CBO’s results and those of other forecasters using multi-equation models of the economy are not science but pseudo science–what the economist F.A. Hayek called scientism—the use of the tools and language of science in unscientific ways.

I’m not an anti-quant. In fact as a political scientist I regularly find myself defending quantitative work from colleagues who despise it and condemn its value (always without understanding it, of course). But I know just enough mathematics myself to know that simply attaching values to variables doesn’t create empirical accuracy. I’d be appalled if the economics profession quit doing quantitative work, but before we commit to major, and costly, public policies on the basis of quantitative analyses, we really ought to have some certainty that that those analyses have substantial external validity.

So where does that leave us?

Let’s get back to basic truths.

When you’re in a hole, stop digging. Stop running deficits of over 1.5 trillion dollars. Act like grownups and get your fiscal house in order. Stop spending 25% of what we produce. Stop wasting my money and giving it to your friends. Stop passing legislation that makes it hard to figure out what the rules of the game are going to be. Get out of the way. Make government smaller and give us a chance to do what comes naturally—seeking ways to make profit, avoid loss and work together. That is the only sustainable path to prosperity.

I very uneasily argued against the fiscal stimulus approach. Uneasily because I wasn’t that certain in my opposition–based on my knowledge of economics I doubted the value of fiscal stimulus and was concerned about vastly increasing government spending and borrowing, but I, too, felt the pressure against just doing nothing. In fact I was firmly convinced that doing nothing would work, given time. But how much time?

But both the Keynesians and anti-Keynesians now agree the fiscal stimulus didn’t do diddly. They Keynesians say it’s because it wasn’t large enough, and Krugman makes this somewhat plausible by noting that state and municipal spending dropped so much that the federal spending only compensated for it, without actually increasing total spending (although he usefully doesn’t count the tax cuts as part of the stimulus, which sounds a lot like cooking the numbers to me). Given the exorbitant cost of running a fiscal stimulus program large enough to persuade Keynesians it’s a true stimulus, and the already unsustainable deficits we’re running, it sounds like a tremendously risky experiment. And as Scott Sumner points out (same link as above), the government has had trouble spending the money from the first stimulus package (in fact there’s a good chunk that’s not yet been spent at all), so adding even more to that package probably wouldn’t do much to help (how much unspent money do you have to add to prior unspent money to create a stimulus?)