Causal Claims Without Identification: A Defense of Institutional Details and Economic Theory

I was listening to a lecture Chad Syverson gave at the Becker Friedman Institute, and I was struck by a particular argument he made in the talk.  Dr. Syverson showed graphically (see below) that the efficiency of American sugar refining steadily increased until 1934 when the Sugar Act was passed. Sugar refining then steadily decreased until the act was repealed in 1974. Since the repeal, there has been a steady increase in efficiency, paralleling the upward trend before the sugar act was originally passed.

Screen Shot 2017-05-20 at 11.21.01 PM

Dr. Syverson used this as support for his argument that government regulation can create incentives that are antithetical to efficiency. He went on to explain that the Sugar Act essentially paid farmers for producing sugar which created an incentive for sugar beet farmers to produce larger sugar beats which, in turn, decreased refining efficiency (for more details listen to minutes 51:00–54:25 of his talk). In effect, he was claiming that the sugar act caused the decrease in sugar refining efficiency that occurred for 40 years while the act was law. When I heard his argument I realized two things: (1) that I was pretty convinced by his causal claim and (2) that Dr. Syverson never presented a well-identified estimate of the causal effect of the Sugar Act on sugar refining efficiency (it is very possible that such an estimate exists, but it is irrelevant to my point). How was this possible? As an aspiring applied microeconomist, I have been trained to be skeptical of all unidentified causal claims, so how could this argument be convincing without treatment and control groups?

The obvious answer is that there are other ways to convince even the most rigorous applied economists that a claim is causal. In this case, Dr. Syverson presented two empirical facts that are consistent with a straightforward economic explanation. For him to be wrong, farmers would need to deviate from profit maximizing behavior, and the timing of these changes in efficiency trends must have been coincidental. The burden of proof would have been on me if I were skeptical of his claim.

This experience was a good reminder that causal claims rely on theoretical explanations and institutional details as well as causally identified coefficient estimates. In fact, sometimes no estimate is needed at all, as in this case, to make a convincing argument.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s