Is Economics Autistic? Can it be Reformed?

Before we begin, I would just like to state for the record that I'm wholly against using the term "autistic" in a pejorative sense. However, one of the largest groups seeking to reform the study of economics describes itself as Post Autistic Economics (PAE), so I will follow their language.

The PAE movement, in my view, often goes too far in their criticism of modern economics. I do, however, concede that they have a point; economics as it is conducted now does have "autistic" (as they call it) tendencies. Research in economics unfortunately often is set-up as follows:

• Think about a 'problem' and create a number of 'stylized facts' which you believe accurately describe the phenomenon you have chosen to study.

• Create a number of assumptions based on these stylized facts. Ensure that they are mathematically tractable, that is, they have certain mathematical properties that make them easy to use (continuous, twice differentiable, etc.)

• Using a fixed point theorem or other mathematical tools, find an 'equilibrium' to the problem; at a minimum at least prove that one must exist. Bonus points if you can prove that the equilibrium is unique.

It's readily apparent to a non-economist that this approach has a number of flaws. If I want to study, say, how prices are set in the car rental industry, an obvious first step would be to learn more about that particular industry. One could go and talk to managers of car rental dealerships, discuss strategy with high-level executives such as CEOs and VPs of marketing, and read books on the history of the industry. However, too often in economics the starting point is to automatically make a number of assumptions such as: prices are set to maximize revenue, every firm in the market is a rational profit maximizer and knows that the other firms are, each firm faces a constant marginal cost c for each car rented, and so on. However, in the "real world" these are not often the case. Often the guy(s) who sets the prices of vehicles isn't trying to maximize the firm's profit. Maybe he's just trying to keep his job, so he's trying to minimize the chance the firm shows a loss (which is much different than trying to maximize expected profits). Perhaps he holds a number of stock options and his decision making criteria is to maximize the current stock price, even if it's at the expense of future profits. Maybe the opposing firms have committed themselves to a fixed-pricing strategy, so they cannot price 'rationally' as market conditions or opponent firm-pricing strategies change. Without investigating the problem, one can never know if one's assumptions are plausible.

Mathematical tractability also has a number of flaws. I can't tell you how many models I've seen with assumptions such as: "Demand for the product is one unit, unless the price is above p(bar), in which case the demand is zero". I've never seen a market that works like that in the "real world" and I doubt that I ever will. However those assumptions are often required in order to 'prove' a result; a more realistic price elasticity of demand would likely not be mathematically tractable.

There's nothing wrong with trying to offer a mathematic proof to a model, so long as the assumptions make sense. If the assumptions do not make sense, however, all you've proven is that a result holds in a world that does not exist. Sure, it might be fun to know that if the President of the United States is a bug-eyed alien from the planet Neptune that interest rates will be 6%, but since he's (likely) not, what, if anything have we really learned? Bad assumptions lead to irrelevant results, as emphasized by the phrase "garbage in, garbage out".

If we want economics to be more useful, that is to create more models which accurately describe "real world" phenomena, then we must use more accurate assumptions. There are quite a few costs in doing so, however. First economists would need to spend more time on the "front-end" of their research, that is, they would need to spend more time out in the field in order to create a set of assumptions which more adequately describe reality. With these "improved" assumptions, however, we will likely lose mathematic tractability, that is "proofs" become impossible and we are forced to resort to other methods, which lead to results we cannot be as certain about. Economics needs to become less enamored with proofs and more enamored with realism. Can this happen? I see it as unlikely - the researchers who do well in the current environment tend to end up as editors of the top journals. Unless the environment somehow spontaneously evolves, I do not see a situation where researchers who emphasize realism over mathematical certainty will wind up having control over the top journals.

You can send your thoughts to the author by using the feedback form.