The front page of the WSJ's Weekend Journal section addresses this question today (subscription required). Darren Everton does a nice job observing that
Baseball managers are tough to grade. Compared with football, basketball and hockey, where coaches are often hired because of their offensive or defensive acumen, a manager's job involves relatively little strategic philosophy. Even the tactical decisions managers make, from how to set a batting lineup to when to take out a starting pitcher, are based on a careworn historical consensus shared by almost all managers.
The article mentions various assessment techniques and compiles a table that aggregates three methods for rating managers, i) performance in close games, ii) performance based on number of runs scored and allowed, and iii) how players perform under different managers. Torre comes out 17th. My issue with method (i) is whether it measures incremental manager value or just realized values of the "error term." Method (ii) assesses "technical efficiency" given scoring but does not include contributions to runs scored or allowed. In this respect, Method (iii) complements Method (ii). Interestingly, Torre does relatively well in Method (ii) but very poorly in Method (iii).
In the end, I tend to line up with the sports economist, J.C. Bradbury, who is quoted:
"I think managers are a bit overrated in terms of the impact that they have on their players," To make a team better, he says, "get better players."
I'm studying the effects of managers across different sports vis-a-vis general managers. My preliminary results suggest J.C.'s point is especially applicable to baseball.