Two prominent economists, Russ Roberts from the Hoover Institute, and Noah Smith from Stony Brook University, partook in an insightful conversation on Roberts’ podcast EconTalk a few weeks ago. Though they covered many interesting topics, most provoking to me was their discussion regarding whether or not economics is a science (Found from approximately minute 14 to minute
Roberts, a strong neoclassical liberal economist, fell strongly on the side that economics is not a science arguing that “economics is not much of a science in the way that people in everyday English think about it” because “it doesn’t produce reliable empirical relationships that can be used as the basis for economic policy.” Smith, who considers himself a pragmatist in that he is not ideological and makes economic judgements on a case by case basis rather than an adherence to one or multiple schools of thought, argued that economics is a science.
Smith’s argument revolved around the so called “credibility revolution” which is the thought that improved econometric techniques and better research design can take naturally occurring events in the economy and identify what caused them to occur, thereby establishing a cause and effect relationship between economic policy and the outcomes that policy produces.
An example of a “natural experiment” as they are referred to by economists, is implementing a higher minimum wage in one state (Washington for example) while leaving the current minimum wage in another state (Idaho) at its current level and analyzing the effect the policy change has on a Washington town that borders Idaho in comparison to a nearby town in Idaho. An example would be Pullman, WA and Moscow ID, separated by 9 miles.
Though not completely sold by the “credibility revolution” himself, Smith argued that because they can be used to identify causal relationships between policy and economic effects, natural experiments are turning economics into a science in that it can be studied like a natural science.
The idea is that just like a chemist can study the effect of mixing two chemicals together, an economist can study the effect of a progressive tax on the Missouri side of Kansas City versus a flat tax on the Kansas side and use that result to make broader claims about the effects of progressive and flat taxes on the economies of cities. Though this may sound good, I find it hard to imagine that any economist could claim such a project is as cut and dry as it would be in a natural science.
Roberts’ own retort is that economists have strong ideologies and it is very difficult if not impossible to separate one’s priors from empirical data, i.e. economists are prone to confirmation bias. That is, it is a hard ask of any economist not to judge data in such a way that it confirms her general economic theories and beliefs because ideologies are so intertwined with macroeconomics that all data is subject to interpretation – an interesting perspective in its own right.
Personally, I side more with Roberts’ argument. I believe it is hard to label economics as a science because the connections modern econometrics and empiricism can make between economic policies and their outcomes are always up for debate. While a study in a natural science often leads to a conclusive result, there are few studies in economics that can conclusively convince economists that their priors are wrong.
So even though natural experiments can be analyzed and studied, they fail to change economists’ wide ranging views on macro issues. And like I already said, this is unlike a natural science where there are many examples of indisputable results, or at least results that are so close to indisputable that they are able to convince practically everyone. For these reasons I am hard pressed to go so far as to call economics a science.
This is an ongoing debate among economists and it will continue to be as economics continues to shift from a more theoretical practice to an empirical one. Tracking the debate will surely prove to be very interesting.