The White House recently released the Economic Report of the President. In a post at the White House blog, Christina Romer brags that the stimulus legislation was a big success.
This Act is the great unsung hero of the past year. It has provided a tax cut to 95 percent of America’s working families and thousands of small businesses. It has meant the difference between hanging on and destitution for millions of unemployed workers who had exhausted their conventional unemployment insurance benefits. It has kept hundreds of thousands of teachers, police, and firefighters employed by helping to fill the yawning hole in state and local budgets. And, it has made crucial long-run investments in our country’s infrastructure and jump-started the transition to the clean energy economy. All told, the Recovery Act has saved or created some 1½ to 2 million jobs so far, and is on track to have raised employment relative to what it otherwise would have been by 3.5 million by the end of this year.
Let's set aside some of the disingenuous components of her post, such as categorizing income redistribution as tax relief, and focus on her claim that the legislation created at least 1.5 million new jobs when total employment has dropped by 3 million. Romer is not bad at math. Instead, she is saying that the economy would have lost 4.5 million jobs if it were not for the $787 billion increase in government spending. This what-might-have-been analysis is completely legitimate, assuming that there is good theory and evidence to back the assertion. Unfortunately (at least for the White House's credibility), Ms. Romer and another colleague last year prepared a supposedly rigorous what-might-have-been report, where they estimated that the so-called stimulus would keep the unemployment rate at 8 percent and that failure to increase the burden of government spending would drive the unemployment rate to 9 percent. Yet as this chart from their paper indicates, when we add in the data for what actually has happened, in turns out that bigger government is not only theoretically misguided, but it also doesn't work in the real world.