Nate Silver, the respected blogger at 538.com, has taken issue with my study of how stimulus funds have been disbursed. This is a good thing, because we really need to have more discussions to determine where the stimulus dollars are going and why they might be headed in particular directions. In fact, this is the reason why all my data is up at the Mercatus Center website for everyone to see and discuss. This is also why I have detailed my methodology in the paper. This report is part of a series that will come out every quarter as more data becomes available, and it is a work in process.
Mr. Silver characterizes the findings in my study by saying, “My bet is that this is all a bunch of noise resulting from an incomplete — and possibly deliberately biased — research design.”
However, there is more to my analysis than Mr. Silver’s post suggests.
1) I agree with Mr. Silver that checking for urban/rural populations and race may be a good idea, and I’d like to re-run the regressions per his specifications. I will gladly give him the Stata printouts when I am done.
2) I will also check for state capitals. While is no doubt that since the reporting only includes primary and sub recipients, it might be the case that money is being disbursed from the capitals. However, after skimming government documents about how the money is allocated there is no clear evidence that this is the case. I will look into it with Mr. Silver’s comments in mind.
I worked within the confines of $18 million Recovery.gov website, a website that we were promised would allow us to track the money to the last cent. Obviously, that is not the case. The money trail ends at the level reported, and from the website one cannot tell where the money went next.
As for this point alone being evidence of a lack of political bias, I would like to quote Mr. Silver’s own words: “By the way — if you throw out the districts that are home to state capitals, those which elected Democratic members to Congress still rank higher, receiving 31 percent more stimulus funds, on average, than those which elected Republicans. So, perhaps there is hope for her analysis yet.”
So even after I use his methodology I will find that Democratic districts, other than state capital ones, are getting 30 percent more than Republican ones. That does seem like a possible political bias to me, which would be worth looking into. I will, however, definitely look if the coefficient remains significant after I remove state capitals..
How much of a bias? I don’t know. Let’s not forget that my take on the data has always been the following: The regression analysis shows that district’s party representation matters. However, I cannot say how much it matters compared to other factors (such as the formula used by different agencies). I said it loud and clear each time I presented my findings. Indeed, I explained it in plain language to Chairman Oberstar last Friday when I testified before his committee.
If it is not possible to nail down the precise amount that party affiliation matters, does anyone truly want to argue that there are no political factors influencing this stimulus or stimuli in the past (whether put into place by Republicans or Democrats)? There is a lot of literature in economic-history journals on similar patterns in New Deal spending, and it consistently shows that New Deal spending correlated rather strongly and negatively with the margin of votes in the previous election. Areas where Roosevelt won by a little got more New Deal bucks than ones where he won by a lot. (I was directed to one article in particular by a reader this morning, and it is worth looking into: Price V. Fishback, Shawn Kantor, and John Joseph Wallis’s “Can the New Deal’s Three Rs Be Rehabilitated?: A program-by-program, county-by-county analysis.” Explorations in Economic History 40 (2003), pp. 278-307.)
I am confident that a similar pattern can be found with President Bush’s stimuli, which, by the way, I was publicly and consistently against.
3) The unemployment data for the regressions has in fact been used by congressional district, not by MSA. The confusion comes from the fact that the Excel file on the website includes unemployment by MSA; however, for the actual regressions, congressional district data was used. I would be happy to send him the Stata file if he’d like it (or anyone else, for that matter).
With that in mind, it is important to underline an important finding in my report. Controlling for the percentage of the district (not MSA) employed in the construction industry, which is often used as a proxy for the district’s vulnerability to recession, we find no statistical correlation between all relevant unemployment indicators and the allocation of funds. This suggests that unemployment, at least thus far, has not been not the factor leading the awards. This is a counterintuitive finding for a stimulus that claimed to help with job creation.
4) The issue with including many demographic variables is that these variables (particularly income variables) are likely to be highly correlated and therefore to introduce bias. However, many demographic variables were tested and found to have no effect on stimulus funds. I should have made this clear in the methodology and I will make sure to do it from now on. Again, I will gladly provide him Stata printouts.
Mr. Silver suggests that we use poverty rate, but my suspicion is that using median income (as we used) is not fundamentally different from using poverty rate as an indicator.
5) As for the suggestion that we have deliberately biased the research design, I challenge Mr. Silver to find evidence for this anywhere in the methodology. That claim is as inflammatory as it is unsubstantiated.
This is not to say that I do not begin my research from a particular starting point. Readers who have taken interest in my work before Barack Obama became president will note that I was a consistent and highly vocal critic of his predecessor when George W. Bush’s policies confounded widely understood economic truisms. Throughout his years in office, I published many articles about Bush’s profligate spending and was the first researcher to project that he was increasing spending faster than Lyndon Baines Johnson; similarly, I am known for having decried Bush’s regulatory excess.
However, my predisposition toward limited government and sound fiscal policy hardly means that I rig my data or designs. Rather, it simply means that I am particularly skeptical when anyone claims that politicians (of all parties) do not programmatically seek to advantage their allies while punishing their adversaries. That was a useful guiding assumption under George W. Bush and, under the current administration, no less so.