The Myth of the Multiplier

Why the stimulus package hasn't reduced unemployment

Give us money, and we'll give you jobs. That was the promise President Barack Obama made when he asked Congress for a $789 billion stimulus bill. The cash, he said, would create millions of jobs during the next two years. Without the stimulus, the administration warned in a January report by economic advisers Christina Romer and Jared Bernstein, unemployment by the end of 2010 would reach as high as 9 percent.

Well, Obama got his money. Since then, the economy has shed more than 2 million jobs and the unemployment rate has climbed to 9.4 percent. By May 2009, the Council of Economic Advisers (CEA) had changed its message. Now the stimulus would "save or create" 3.5 million jobs by the end of 2010. 

Measuring total jobs "saved" by a piece of legislation is as difficult as measuring total crimes prevented by police patrols. That's why no agency—not the Labor Department, not the Treasury, not the Bureau of Labor Statistics— actually calculates "jobs saved." As the University of Chicago economist Steven J. Davis told the Associated Press, using saved jobs as a yardstick "was a clever political gimmick to make it even harder to determine whether this policy has any effect."

A look at the CEA's job creation model undercuts its promises even more. The model's calculation of saved or created jobs is based on a macroeconomic estimate, not on actual data. According to the authors, the estimate rests on a "rough correspondence over history" that indicates a 1 percent increase in gross domestic product (GDP) represents an increase of 1 million jobs. They might as well have said the estimate was picked at random.

How did they come up with the 1 percent figure? Since government spending is increasing, and since such spending is a component of GDP, they assumed GDP would grow whether or not the spending produced real growth in the economy. This is akin to assuming I will have a baby in nine months whether or not I am pregnant.

The May report concedes that while the CEA will attempt to measure job creation through data collected from stimulus recipients, the results will contain errors and inconsistencies. "Because of these limitations," it warns, "the reported jobs numbers will need to [be] used with caution and as part of a more complex estimation strategy."

Since then, Romer has told CNBC she couldn't say for sure how many jobs would be created, since we can't know what would have happened without the stimulus. But didn't her report pro-ject what would happen if the stimulus wasn't passed? Wasn't the 3.5 million number supposed to be the difference between employment with the stimulus and employment without it?

The confusion flows from the faulty theory underlying the stimulus bill. In Keynesian thought, a decline in demand causes a decline in spending; since one person's spending is someone else's income, a fall in demand makes a nation poorer. As a poorer nation cuts back on spending, it sets off another wave of declining income. So any big shock to consumer spending or business confidence can set off waves of job losses and layoffs, as fewer goods are demanded and more workers become useless.

Under this logic, one possible remedy is for public spending to take the place of private spending. As government increases its spending, the money creates new employment. That, in turn, spurs those new workers to consume more and prompts businesses to buy more machines and equipment to meet the government-induced demand. Economists call this increase in aggregate income the "multiplier" effect. One dollar of government spending, the theory goes, ends up creating more than a dollar of new income. It's a rare free lunch.

As appealing as the Keynesian story sounds, many economists have long doubted it. In 1991, looking across 100 countries, Robert Barro of Harvard presented historical evidence that high government spending actually hurts economies in the long run by crowding out private spending and shifting resources to the uses preferred by politicians rather than consumers. For a dollar of government spending, we end up seeing less than a dollar of growth. Can long-term poison be short-term medicine?

Even in the short run, if there's a big decline in the demand for workers, why should that alone cause mass unemployment? If all those workers really want to work, why won't wages just fall until all the workers have jobs? That's how markets end a glut, whether it's a glut of employees or a glut of blue jeans: with lower prices. If recessions really are caused by a fall in demand (and nothing else), why don't wages fall enough to keep people from losing their jobs? 

It's because wages are sticky, Keynesians argue. Wages and salaries don't change on a daily basis the way stock prices and gas prices do, so if a company hits a sales slump, salespeople might earn fewer commissions, but the vast majority of workers don't get a pay cut. There's something about the market for workers that keeps businesses from cutting wages in a slump. As long as wages are sticky, in the wake of a nationwide collapse in sales, entrepreneurs will start firing people.

If a decline in demand means mass firing, a rise in demand can mean mass hiring. Even if government spending is inefficient, pork-laden, and financed by future tax increases, the theory goes, it can still create some real jobs, some real output, in both the public and private sectors. 

So what do the data say? There aren't many studies of the issue. But two stand out: Robert Barro's work and research by Valerie Ramey, an economist at the University of California–San Diego, on how military spending influences GDP. Both studies found that government spending crowds out the private sector, at least a little. And both found multipliers close to one: Barro's estimate is 0.8, while Ramey's estimate is 1.2. This means that every dollar of government spending produces either less than a dollar of economic growth or just a little over a dollar. That's quite different from the administration's favored multiplier of four. What's more, Ramey also found evidence that consumer and business spending actually decline after an increase in government purchases.

Why this crowding out of private spending? Government spending comes from three sources: debt, new money, or taxes. In other words, the government can't inject money into the economy without first taking money out of the economy.

Take taxation: Taxes simply transfer resources from consumers to government, displacing private spending and investment. Families whose taxes have increased will have less money to spend on themselves. They are poorer and will consume less. They also save less money, which in turn reduces the resources available for lending.

In addition, higher taxation encourages people to change their behavior to avoid taxes. They might switch their efforts to nontaxed activities, such as household production, or to the untaxed underground economy. Economists call this a deadweight loss, because people give up the taxed activity or good they prefer.

There are high costs to the other options as well. If the government borrows money, that leaves less capital for the private sector to borrow for its own consumption. If the government prints new money, it will create inflation, which reduces the value of the money we own and decreases everyone's purchasing power.

Overall, government spending doesn't boost national income or standard of living. It merely redistributes it—minus the share it spends on the bureaucracy that collects and spends our tax dollars. The pie is sliced differently, but it's not any bigger. In fact, it's smaller.

Contributing Editor Veronique de Rugy ([e-mail]) is a senior research fellow at the Mercatus Center at George Mason University.