Who Cares About the Evidence?

-- Posted by Neil H. Buchanan

My new Verdict column, published today, confronts the intensifying calls from conservative politicians and their enablers to reduce or eliminate taxes on businesses and the wealthy. Much of the column covers standard distributional analysis. I also discuss the common analytical flaw of focusing only on the effects of taxes on those who pay, while ignoring how the tax revenues are spent.

To put the latter point more bluntly than I did in the column, it is always possible to tell a story about why some tax has -- or could have -- an effect that we might not like (such as creating perverse incentives, reducing some kind of spending or investing, and so on). If we were to limit ourselves to taxes that have no possible behavioral consequence that we might regret, then there would be no acceptable choices. While radical anti-government ideologues might applaud that result during a Republican presidential debate (between cheering for executions, and calling for allowing uninsured coma patients to die), even the watchman state must collect taxes from some source. "This tax might have a bad effect," therefore, is not a serious argument against that tax.

That might seem like an obvious point, and it ought to be. As we know, however, modern politics has become overwhelmed with arguments that ought to embarrass everyone involved. Another obvious point is that evidence ought to matter. From climate change to the Confidence Fairy, however, it is obvious that wishful thinking has replaced assessment of the evidence in too many people's minds.

It is not only the general public and politicians, however, who ignore evidence. A central point in my Verdict column is that there is no convincing empirical support for the claims that taxes (at levels seen in the United States and even Europe) actually harm the economy. A recent news analysis in the NYT quoted a prominent public finance economist as saying that the evidence that capital gains taxes harm economic growth is "murky, at best." In less guarded terms, the economist was saying that we have tried and tried to find the negative effect of capital gains taxes on investment that standard economic theory assures us must be there, but the evidence just will not cooperate. A few studies succeed, but most do not. The same is true, in fact, of other areas of tax analysis as well. Standard economic theory tells us that there is a bad effect, but the evidence is, shall we say, less than clear-cut.

This is significant not just because of the public policy debate, but because of what it says about academic discourse. Even though the "taxes harm growth" hypothesis is supported by murky evidence, at best, tax scholarship and academic discussions typically proceed as if the evidence supported the theory. Why would we do that?

One answer is that responsible academic discussion must include variables that could matter, in order to be complete. Including a variable in an economic model (or even in informal discussions of models) presumes the possibility that the coefficient on any given variable could be zero, in which case there is nothing lost by allowing ex ante for the possibility that the coefficient is theoretically likely to be negative.

As far as it goes, that is a nice explanation. The problem is that it gives us no way to explain why other variables with no empirical significance are never included in the discussion. After decades of empirical failure, we continue to hold out the possibility that certain variables will stop disappointing us. Apparently, we as a group are deeply convinced that taxes must have bad effects, but we simply have not looked hard enough. It also does not explain why so many scholars make policy prescriptions on the basis of the theory and not the evidence.

This phenomenon is hardly limited to taxes. In the 1990's, there was an enormous brouhaha over a famous study of the effects of the minimum wage on low-wage employment. The widely accepted theory was that increases in the minimum wage reduce employment. An innovative study then found that there was no such negative effect, and there might even be a positive one. Perhaps more important as a scientific matter, the authors of the new study also published an article analyzing all previous studies on the subject, finding not only that the evidence was murky, at best, but that the previous studies, when looked at in the aggregate (meta-study), violated a key property of statistical analysis.

The minimum wage story, of course, is now a matter of scientific pride. Even though there are those (even inside academia) who continue to insist that minimum wages harm employment, the combination of new evidence and assessment of old evidence changed the scholarly conversation. The effect of minimum wages is now considered an open, context-specific matter.

Beyond that one outstanding example, however, economic debates often seem to proceed more on the basis of received theoretical wisdom than assessment of the evidence. This is true even when, as in the example above regarding capital gains taxes, the disappointing assessment of the evidence is provided by a member in good standing of the economics tribe.

For example, standard macroeconomic models are nearly always specified to include a negative effect of interest rates on business investment. The theory is that increased borrowing costs will make it more expensive for businesses to buy new capital, which is a simple application of the standard downward-sloping demand curve. The problem is that the evidence in support of that proposition does not even qualify as murky. It is just not there. One of the top macro guys at MIT wrote an assessement of the evdience in the 1980's, in which he noted drily that it takes "more than the usual amount of econometric ingenuity" to "force" an empirical study to show the desired negative relationship. Nothing in the ensuing years would change that overall assessment of the evidence. Again, however, the norm is to talk as if this negative effect is a real-world phenomenon.

In the legal academy, tax scholars also tend to talk as if the evidence relevant to our areas of inquiry is not murky. The typical move is to simply state as a known fact that taxes harm growth, or that they reduce investment, or whatever. Given that legal scholars are allowed (and in many cases encouraged) to be multidisciplinary, and that they do not have to worry about whether they are offending a dearly-held assumption of another academic field, why would they follow economists in failing to notice or acknowledge the lack of empirical support for such important propositions?

My best guess is that, for all the arrogance that legal scholars bring to the table (and we do bring it!), there is an underlying level of professional modesty and doubt that creates a strong tendency to defer to economists on these matters. Although this is true in other fields of law as well, I suspect that it is especially strong in tax law. The academic discourse in tax law is driven by vintage economic theories, and legal scholars in tax, I think, tend to believe that economists really have the answers. This is unfortunate. While it is true that one needs to have a decent grasp of key concepts of public finance economics to understand tax analysis, that is a far cry from saying that the economists have all the answers.

In any event, the academic discourse in tax is truly odd. The evidence about some of the core questions under study should give anyone serious pause about believing the conventional wisdom. Yet economists -- for a combination of reasons -- continue to act as if the conventional wisdom works, while legal scholars too often act as if the economic theory, and not the econometric evidence, is the path to the truth. This is far better than simply ignoring evidence for ideological advantage, as we see among politicians. It is still, however, quite damaging to our understanding of how to improve matters.