By Zachary Hayward
Social science has a problem. Over the last 10 years, there has been an increasing focus on the replicability of social science research. The discipline of economics in particular, has come under scrutiny after the financial crisis. The main challenge to its credibility being that the dismal science is “unscientific” and therefore, in some sense, unworthy of guiding policy.
When we think about what counts as science, most people immediately start with disciplines like physics and chemistry. They see these as highly rigorous disciplines searching for fundamental laws. Their primary methods of inquiry being falsification of testable hypotheses. This methodology is in fact what Karl Popper uses to demarcate science and pseudo-science.
And this, is where critics of economics find their first target. They claim that despite the efforts of empirical researchers over the last few decades to introduce more rigorous experimental techniques, economics can never satisfy this key criterion of providing testable hypotheses. That the discipline can simply explain away falsifying results by appeals confounding factors. A hallmark of pseudo-science. However, it should be noted that this is a possibility for all empirical science and to even consider empirical evidence in the first place, there need to be certain background auxiliary assumptions which may or may not be testable themselves. This is not to say that empirical economics research is in the same category of rigor as physics experiments where the auxiliary assumptions can be effectively minimized. However, it does seem to imply that experiments in economics are not inherently different. They do, however, need to be viewed as a more difficult undertaking with far less certainty involved.
The fact that any economic experiment has an inherently more complex causal field of background assumptions also implies that any result is more fragile, and the conclusions less strident. Hence the absence of true economic laws of gravity. It is probably important to note that much of the discussion about economic methodology gets stuck in the quagmire of whether it is too mathematical. This is particularly true of macroeconomics. I think this is a mistake. It hides the very real problems with some economic models (particularly where a non-obvious conclusion truly is not testable) and too readily concedes the very thing that allows us to see why our predictions fail.
Dani Rodrik argues in his book Economics Rules, that we should see the entire discipline as a collection of useful models tailored specifically for relevant contexts. He argues that the major differences with the physical sciences are in degree rather than in type. If anyone doubted the importance of context, they should ask Long-Term Capital Management about it.
Whether or not economics is truly scientific is not so much the point, it’s whether we can trust its conclusions. Unfortunately, without a credible methodology, policy recommendations are nothing but ideological statements. It is critical that the discipline take more notice of how scientific enquiry should be conducted and the inherent limits that this places on our conclusions.