The typical analysis applying the prisoners' dilemma to cartel theory concludes that as the number of players increases, so does the incentive to cheat, especially when there are weak or ineffective additional enforcement mechanisms. Paul Kedroski of Infectious Greed (and PhD from UWO's bizskool) provides a telling example:
OPEC had announced a cut of 1.2 million b/d in October, and another 500,000 b/d production cut is scheduled to take place in February. Our analysis suggests that only about 65% of the agreed October cuts have been implemented.
By not cutting back on production by the agreed amount, each of the players (producers) is cheating on the agreement. If all the other producers cut back on production, and one doesn't, then that one producer can sell more at a higher price. But of course all the producers either think that way or expect the others to think that way with the result that all or nearly all of them cheat on the agreement and produce more than the agreed upon amount.
But this is a repeated game, and surely each of the players has learned to anticipate this response from the other players. If so, then it makes sense to state a nominal amount that production will be reduced with the full expectation that no one will abide by the agreement. The only scope left for cheating on the cartel, then is by how much one does
not stick to the nominal agreement, and that uncertainty gives rise to more incentives to cheat.
This is surely not a stable cartel in the long run without additional enforcement mechanisms.