Tag Archives: models

Models, Counterfactuals, and How I Know the Stimulus Worked

Counterfactuals are a bitch.

A counterfactual is a “what-if?” Asking, “How would things be if X had happened instead of Y?” Essentially, counterfactuals are alternative realities.

Economists often deal with counterfactuals. Some economists ask counterfactual questions like “What if the South had won the Civil War?” or “What if the U.S. had led the global monetary system after World War I?”

These may sound like pointless academic exercises, but sometimes there’s no better way to consider the impact of an economic policy than a good counterfactual thought experiment. After all, it’s not like we can run random experiments on an economy. “Let’s raise interest rates and see if millions of people lose their jobs.” That sort of thing doesn’t go over too well with the public, particularly in an election year.

But counterfactuals are tricky. Who can know how things would be different today if Nixon hadn’t imposed price controls, or if the U.S. hadn’t toyed with the Phillips Curve in the 60s? The world’s a complicated place – butterfly wings and all that. Tossing different pebbles would create very different ripples.

And yet counterfactuals are the name of the game in real-world economics. Everyone wants to know how things would be if we’d done something different. We have to ask what would’ve happened if we’d zigged instead of zagged.


The fiscal stimulus is a case in point. In this election season, pundits are dissecting and critiquing the $840 billion American Recovery and Reinvestment Act (ARRA), passed in 2009. “What did we get for our money?” people ask. “The unemployment rate is still over 8 percent.”

Most economists respond that it would have been worse without it.

As the New York Times said recently, ” … the American Recovery and Reinvestment Act is responsible for saving and creating 2.5 million jobs. The majority of economists agree that it helped the economy grow by as much as 3.8 percent, and kept the unemployment rate from reaching 12 percent.”

And I believe it. People don’t seem to understand the severity of the current downturn. The number of jobs (total nonfarm employment) fell by nearly 9 million from January 2008 to February 2010. Up until then, the most jobs that had been lost in a recession since World War II was less than 3 million. Here’s a comparison of the total jobs lost in U.S. recessions since 1949.

Sure, the economy’s still not doing well. The unemployment rate is much higher than anyone wants it to be. But the reality is that it would’ve been even worse without fiscal stimulus.

How do I know this? How can economists be sure that things would have been worse without the fiscal stimulus of the ARRA?

One word: Models.

Economists live and die by models. A model is a simplified representation of the real world. Economics models usually take the form of a bunch of mathematical equations, or an Excel spreadsheet, or a set of computer programs.

Whatever the form, economists use models to estimate how the real world works. And, ideally, to predict how the real world would respond under different situations – including different policy options. Models basically let economists run experiments on the economy without anyone having to really lose their jobs.

Models allow economists to examine counterfactuals – to ask, “What would have happened if … ?”

NPR recently had an article about Mark Zandi of Moody’s Analytics and his macroeconomic models. This article gives an insider’s view of how economists use economic models.

No model can predict the future with certainty. But good economists calibrate and verify their models against real-world data, to the point where they can do an amazing job of predicting how economies would react to different policies. And if you ask enough good economists what their models tell them, you’re going to get a good idea of how a real economy would react.

So I have little doubt that the fiscal stimulus of the ARRA has helped the U.S. economy substantially. The fact is, things would have been much worse had it not been for the stimulus.