Now that March Madness is upon us, allow me to suggest some light reading for those tv timeouts and halftime lulls. It's something I heard about this on NPR during the drive home from work tonight: a report by economist Justin Wolfers on the likelihood of point shaving in NCAA basketball.¹

We know that point shaving exists. Wolfers' paper claims to be able to measure the incidence of this practice and suggests that it is much more prevalent than previously thought. The strongest part of his argument lies in his finding that teams tend to just barely miss covering the spread by a much larger margin than those who just barely succeed in covering the spread only when the point spread is set at a high margin (12 points or higher). If point shaving did not occur, you would expect these results to be evenly split (as tends to occur when point spreads are tight).

It's an interesting argument, but it fails to convince me. From what I understand (I'll have to consult a more knowledgeable gambler to know if I've got this right), bookies in Vegas don't care about the actual outcomes of a game. The spread is set so as to induce an equal amount of betting action on either side. Since heavily favored teams also tend to be popular teams, bookies would need to induce more betting action against these teams by artificially inflating the point spread.

For what looks like a more authoritative opinion, the comments to this Freakonomics blog entry contain some interesting analysis of Wolfers' argument.