Wednesday, August 14, 2013

Did the Stimulus Help? You Decide.

How and when was the money in the 2009 stimulus package spent? How much might have the economy shrunk, or unemployment risen, had the federal government not cut taxes and boosted spending?

Economists and the government itself have been trying to answer these questions for the last four years. The Obama administration built the website to “track the money.” Dozens of economic models were used and studies published, too, some saying the money saved the U.S. from a second Great Depression, others saying the effect was piddling.

But there was a big hole in all of these analyses: Nobody had actual data that tracked the federal outlays or tax cuts.

Enter the Bureau of Economic Analysis. The agency that publishes gross domestic product figures has been quietly releasing the official government data estimating how and when the stimulus money was spent. To the best of my knowledge, nobody has used these figures to construct a counterfactual scenario of how the economy would have done in the absence of the stimulus.

In some sense this task is impossible. Not only do we have parameter uncertainty -- in that we don't know how much "bang for the buck" in terms of GDP we get from, say, increasing the generosity of food stamps -- but also we have model uncertainty. We don't know how to project the path of a recession and the recovery.

What we can do, though, is construct a counterfactual in a very basic accounting-math sense. Assume a few multipliers and deduct the spending, transfers, and tax breaks from GDP, assume the parameter that relates changes in GDP to changes in the unemployment rate from Okun's law, and you have it. We can disagree as to whether the dynamic effects of a recession -- for instance, would a deeper recession have allowed for a sharper bounce-back? or would a deeper recession have spiraled? -- but this is a good baseline for the debate.

Here's where I come in. I've built a program in Excel that allows you to do just that, estimate a counterfactual for GDP and unemployment. Plus, I've provided you with Mark Zandi's estimates of the multipliers, as well as lower- and upper-bound estimates of the multipliers that are reasonable given slack demand. And it makes you nice graphs!

Here's a screenshot, and you can download the program here. Click the image to enlarge:

The program will always be available under the "Data" tab, also. The way I've set it up is to user-friendly: All you do is enter your estimates on the front spreadsheet, and graphs pop up. The other sheets do the number-crunching in the background.

I've been trying to construct counterfactuals for a while -- see this earlier post -- but this is by far the most detailed attempt I've ever made.


  1. There is a disagreement about what the various multipliers could be (and I didn't want to download the file on a work computer, especially since I couldn't tell what the file extension was), but you have a bigger problem with this analysis. Your assumption results in real GDP declining for several additional quarters. However, you seemingly made no additional assumption about what the Fed would do in the event of no fiscal stimulus. Taking no additional assumptions implies that the Fed would have acted the same way, regardless of whether there was fiscal stimulus or not. In addition, many of the financial companies that received bailouts would have become very close to failing (Citigroup shares had fallen below $1 in March 2009). Continued weakness in GDP growth would have kept these shares depressed and possibly required a second bailout, or some other radical policy. As such, it is reasonable to believe that in the absence of fiscal stimulus, the Fed would likely have acted much more quickly in terms of expanding QE or other policies. This would have put upward pressure on GDP growth, relative to your estimates.

    1. Agreed -- this is why I make the huge caveat about "model uncertainty" in the piece.