OK, how does this reasoning sound...
Suppose we assume the company gives new employees 20 hours each in February and May. * To a CANDIDATE, 2 February hours are worth 1 May hour * To a FELLOW, 1.2 February hours are worth 1 May hour That is, fellows have a preference for May over February, but it is not as strong as candidates preferences.
The only way a trade can occur is for a candidate to give up X February hours (where X is bigger than 1.2 but smaller than 2) for 1 May hour  no fellow will agree to the reverse trade, because it decreases the number of hours worked. * The candidate gains 2  X study hours (in units of February hours) * The fellow gains X  1.2 work hours (in units of February hours)
Now suppose we require approval by a committee. * Candidates on the committee will not approve trades in which the fellow gains more than, say, 0.6 hours  that is, candidates will approve only exchange rates LOWER than 1.8 February to 1 May. * Fellows on the committee will not approve trades in which the candidate gains more than, say, 0.6 hours  that is, fellows will approve only exchange rates HIGHER than 1.4 February to 1 May.
Suppose, on average, that candidates will have the following distribution of trades when there are no transaction costs: I) 5 hours are traded at 1.9 to 1 (0.1 to candidate, 0.7 to fellow) II) 5 hours are traded at 1.6 to 1 (0.4 to candidate, 0.4 to fellow) III) 5 hours are traded at 1.3 to 1 (0.7 to candidate, 0.1 to fellow) IV) 5 hours are not traded (remain in February) As a result, the average candidate will end up with 29.6 May hours and 5 February hours, for a total of 34.6 hours. The candidate gains the equivalent of 4.2 hours of February study time while the fellows gain 3.5 hours of February work time.
Now if trades are approved by a committee of only fellows, III will not occur. The candidate ends up with 25.8 May hours and 10 February hours, for a total of 35.8 hours. The candidate gains 1.5 hours and the fellow gains 3.1 hours.
If trades are approved by a committee of only candidates, I will not occur. The candidate ends up with 27 May hours and 10 February hours, for a total of 37 hours. The candidate gains 3.9 hours and the fellow gains 1.6 hours.
If trades are approved by a committee of BOTH, I and III will not occur. The candidate ends up with 23.1 May hours and 15 February hours, for a total of 38.1 hours. The candidate gains 1.3 hours and the fellow gains 1.3 hours.
In this case, having a committee of CANDIDATES is: * Good for the candidates because they have the biggest gain from trade (3.9 > 1.5 > 1.3) * Good for the fellows because they can offer the highest number of total study hours and with it negotiate a lower salary.
