Surfing the net the other day, I ran across a working paper concerning the decision to steal second base by Ashgar Zardkoohi, an economist in the management school at Texas A&M, and several colleagues. While I don't know Ashgar personally, our networks of associates have overlapped, so I decided to take a look at the paper. Using over 200,000 decisions in MLB between 1985 and 1992, the authors estimate the likelihood of an steal attempt based on game situation (lead, pitcher characteristics, count, league, ...). Their objective is to determine whether managers select stealing attempts (as if) taking into the full range of likely benefits or costs (expected utility) or whether certain biases appear whereby managers appear to use rules-of-thumb that limit their decisions based on the context of the situation or their biases (prospect theory). They find two main results suggesting the latter:
- Teams that are losing are less likely to steal than are teams that are tied, and teams that are winning are significantly more likely to steal than are teams that are tied.
- Managers become less likely to have players steal as the game progresses.
I'll leave it to interested readers to look at the paper for detailed explanations of why these support the idea that managers employ rules-of-thumb. In short, both indicate that something in addition to the likelihood of influencing the game outcome is at work. Managers appear to eliminate or alter some choices that, if considering the real probabilities, would be better.
The study offers insight into the value of sports economics. Most experiments comparing different models of decision making utilize contrived settings that always raise the specter of departures from decisions in "real" situations. The authors explicitly pursue this angle of offering a "field" comparison.
As for the findings, some proponents and critics of "behavioral" methods stand ready either to dismiss such results out of hand or use them as supposed evidence of the complete failure of economics. I don't see it that way. Economic models are very useful, but exactly how people deal with uncertain situations and process available information is an area of open study. "Why Good Refs Make Bad Calls" highlights the work of Roy Radner (economist) working on the more theoretical end and Max Bazerman (psychologist) working more from the experimental-empirical side on subject of information processing. More recent developments in prospect theory have shown how it can be consistent with basic and tenets of choice in economics that are crucial and universal. What is needed over time from the experimental-empirical side is some sorting through the long list of "biases" and "effects" to find ones that are hold across most any situation. Otherwise, one is left with only a "theories" of decision making applicable in one context but not another. That's not useful for much other than seeking grant support for the latest experiment.