‘Wrong’ economic theories and performativity: reply to Ezra
October 28, 2008
This post started as a reply to a post on OrgTheory, but it got slightly longer and raised some interesting issues, so I thought that I’d make a post out of it.
Let me give you the context. The issue here is the question of whether or not a ‘wrong’ economic theory can be performed in such a way that it ‘becomes’ accurate. I claimed that Black-Scholes-Merton is an example (in fact, a very good example) for a wrong, but very successful, economic model. Ezra answered that “The inaccuracy of BSM at the outset was not a surprise to anyone because it was not a descriptive theory, but a prescriptive one- a model for what one *should* do. After all, the options market basically did not exist when the theory was developed, so it could not have been intended as description.”
Below is my answer to Ezra
Ezra, I see what you mean now. However, Black-Scholes-Merton is a good example of a wrong model that ‘became accurate’ and that’s for two reasons: I would call them the ‘weak’ reason and the ‘strong’ reason.
First, the ‘weak’ reason. Yes: an organised options market did not exist when the model was published and the assumptions underpinning the model did not exist in the market even when it was established (i.e. no restriction on short selling, no fees on borrowing, continuous trading). So, from this respect you can say that the model, like many other economic models, was talking about a ‘would be’ or a ‘utopian’ market rather than an existing one. That, of course, does not turn the model into a prescriptive model. No one in the Chicago options’ market or at the SEC used the model with the intention to prove that Black, Scholes and Merton were right. They used the model for a variety of reasons, most of which are related to operational efficiency. As the performativity thesis claims, an economic theory becoming accurate is a result of a networked emergence rather than the outcome of specific agents.
Now, for the ‘strong’ reason. The original, theoretically driven Black-Scholes-Merton model was based on a lognormal distribution of the underlying stock (the theory here goes all the way back to Bachelier, tying the movement of stock prices to Brownian motion, etc). Without this assumption at its basis, the model would be not much more than a fancy card trick run on high power computers. But, guess what… Nowadays, virtually no one uses the plain vanilla (but theoretically justified) lognormal distribution in his or her BSM-based applications. Since the crash of 1987, where the Black-Scholes-Merton was not accurate, the ‘engine’ of the model, if you like, was replaced by a variety of different distributions, none of them justified by the theoretical roots that led to Scholes’ and Merton’s Nobel prize. So, again, for a very long time (at least since the early 1990s) the Black-Scholes-Merton model has been ‘wrong’ theoretically, but useful operationally.