Fiction: The Obama GDP Counterfactual

Time magazine recently ran an article titled "The Counterfactual President: Obama Averted Disasters, but Getting Credit Is the Hard Part". The article explains how most of the accomplishments of the Obama administration are based on the counterfactual argument that things would have been worse without the policies enacted by the administration.

This particular quote from the article caught my attention since it implies a testable proposition:

The most extreme example, of course, was the $787 billion stimulus package that Obama signed during his first month in office, when the economy was shedding 700,000 jobs a month. The immediate goal was to avoid a depression, and in that sense it was a tremendous success, stopping the hemorrhaging and stabilizing the scariest economic situation since the Great Depression.

The idea that steps taken by the administration prevented the onset of the second Great Depression implies the following causal model.

The Obama administration, O, after it took office in 2009 not only solved the Subprime Mortgage Crisis, SC (which was negatively affecting the state of the economy, S), but also through the stimulus package improved the state of the economy. Since Gross Domestic product, GDP, is dependent on the state of the economy, improvements in the state of the economy should have improved GDP. To be clear, the model above supposedly explains the actual history we saw after the Obama administration came to office in January of 2009.

The counterfactual argument ("Things would have been worse without us") is described in the model above. The Subprime Mortgage Crisis would have continued to have a negative impact on the state of the economy (and thus GDP) to the present without the policy steps taken by the Obama Administration.

The null hypothesis for the two models above would be that the Subprime Mortgage crisis was over in 2008 and the economy was on the road to recovery in any event. The self-loop in the state of the economy, S, is meant to show that the US economy has its own internal feedback dynamic in response to external shocks such as the Subprime Mortgage Crisis, SC.

A complicated way to test the proposition would be to try to directly estimate the effects of the stimulus on the US economy and demonstrate that the effects were positive. This approach has been tried by the Congressional Budget Office (here), the council of Economic Advisors (here) and economists Alan Blinder and Mark Zandi (here). They have all concluded that the effects were positive.

A simpler way to test the counterfactual would be to estimate a state-space model of the US economy (the USL20 model) and use the state of the US economy to predict real GDP up to the last quarter of 2008, right before the start of the Obama administration. Then simulate the USL2008 model from the first quarter of 2009 until the present. In other words, run the model forward (counterfactually) as if the policies of the Obama administration had never been enacted. If the economy goes off the cliff, than there would be evidence for the counterfactual assertions of the Obama administration ("Things would have been worse without us"). If the economy recovers by itself, then there would be evidence for the null hypothesis ("The subprime Mortgage crisis ended in 2008").

The first question should be how good a job did the USL2008 model do of predicting real GDP (GDP2005 or GDP in 2005 dollars) for the late 20th century. The graphic above shows that the model did a very good job.
The USL2008 model can then be run forward to the present (BEA is currently publishing GDP data into the second quarter of 2011). The model indicates that the economy would not have fallen off a cliff and was in fact underperforming. However, this result cannot be used to argue that the Obama administration policies were inhibiting economic growth (the Republican argument), as the next graphic demonstrates.

There is the matter of probability and the associated prediction intervals for the simulation. As shown above, GDP is still within the 98% prediction interval for this model, but it is getting very close to the lower prediction interval (improbable poor performance) in the second quarter of 2011.

The counterfactual simulation, however, does show that the economy would have hit bottom in mid-2009 regardless of Obama administration policies. The poor performance from mid-2010 to the present could be the result of many other factors (e.g., poor performance in the world economy) other than or in addition to policies of the Obama administration. The simulation, however, does not demonstrate that the "stimulus was a tremendous success" as argued by the Time magazine article.