Lutz Hendricks - UNC - Department of Economics

Specialization (or the Lack Thereof)

Researchers in other disciplines are highly specialized. Economists are not. This is inefficient.

In Biology, someone may spend 20 years studying Red Ants in East Africa. It would by unthinkable for that person to suddenly write a paper on Emperor Penguins.

In economics, the analogous behavior is entirely common. People write on topics they don’t know anything about. And nobody thinks this is odd.

As a referee, I commonly encounter authors that appear not to understand or not to be familiar with the literature. This is one cost of the lack of specialization.

Another cost is duplication of effort. If an author wants to work with a particular dataset, he/she typically just figures out how that dataset “works.” This leads to mistakes and it wastes time. We need authors that specialize in specific datasets.

Similarly, very few economists know how to write reasonable computer code. Having looked at quite a few examples of programs underlying published papers, I find:

  1. Spagetti code: single functions with over 1,000 lines of non-trivial code.
  2. No evidence that the code has been tested.
  3. Globals are used to set parameters.
  4. No easy way of swapping out model elements, even though a typical paper solves many different versions of a model.
  5. Little useful documentation.
  6. Essentially no general purpose code is used.
  7. Essentially no code is used that was written by someone else.

In other words, most economists violate all the guidelines one would typically learn in Programming 101. Most code that is written is likely wrong. We need authors who specialize in programming.

Consider again how other disciplines work. As an extreme example, experimental physics employs a wide variety of specialists (engineers, programmers, machine shop workers, etc). This makes it possible to run very complex projects, such as particle beam colliders. Economists do nothing of remotely similar complexity.

We need to specialize.