russ roberts vs. expected utility

If you listen to recent Econ Talk episodes, you will notice that the host, economist Russ Roberts, will express dismay at some standard tools of economic modelling. Specifically, he really finds it troubling that a lot of economic analysis models behavior using expected utility – you assign a number to an outcome, guess at the probability it will happen and then take the average of utility of X and not X. He is also skeptical, or cautious, about econometrics as a whole. It’s not that he hates statistics, but rather finds that a lot of the modelling relies on dubious assumptions and models are only as good as the data.

Here, I want to dig into two of the reasons that Roberts doesn’t like utility theory. First, he has said that by focusing on the mean, you may not really appreciate the importance of various outcomes. For example, when I drive a car, the average is that I am ok. But once in a while, I get into a serious accident. Yet, the average is near zero death for driving a car. There’s something weird about that. Second, Roberts seems to think that economics over the course of its history has shifted from thinking about economic institutions and behaviors to a form of engineering. By reducing economic behavior to expected utility, you miss some big stuff. In a recent podcast, he offered a great example: economics can be either about how many widgets you make, but it’s also important to know why the market for widgets exists in the first place or why a company would want to be in the market for widgets. 

My responses: On the first issue, I think I am ok with expected utility as long as you stick to situations with “well behaved” utility functions. What would count as “well behaved?” Maybe situations where outcomes vary smoothly and rare events don’t have super high or low utility. In statistics, we would say that utility functions have to be a finite mean (when you add it all up to take the average, you get a finite number). I think that would apply to most daily situations, but I can imagine important situations where it doesn’t hold (see Monday’s post). On the second issue, you are now really hitting a truly deep point. What is economics supposed to be? The modern answer is, roughly, “applied decision theory plus regression models.” Of course, there are heterodox economists who might dispute this, like Austrians, but Roberts, I think, hasn’t endorsed Austrianism or another form of heterodoxy. At best, you might say that he’s a Chicago guy who is a “neo-Smithian,” which a is good way to describe his recent book on Adam Smith’s Theory of Moral Sentiments. I’d also humbly suggest that he has an inner economic sociologist who wants to come out.

+++++
50+ chapters of grad skool advice goodness: Grad Skool Rulz ($4.44 – cheap!!!!)
Intro to sociology for just $1 per chapter – INSANE BARGAIN!!!!!
A theory book you can understand!!! Theory for the Working Sociologist (discount code: ROJAS – 30% off!!)
The rise of Black Studies:  From Black Power to Black Studies 
Did Obama tank the antiwar movement? Party in the Street
Read Contexts Magazine– It’s Awesome!!!!