^{**} Massachusetts Institute of Technology

[1] For instance, it establishes an interorganizational
information system with *n* < *N* suppliers. We assume that the
choice of *n* cannot be changed until after period 1.

[2] Since returns to scale are assumed constant, the buyer
will always order from the most efficient supplier--the one with the highest
[[epsilon]]*i*. We assume that supplier efficiency is discovered only
after a relationship has been established with that supplier, and is ex post
observable by the buyer firm and all *n* suppliers.

[3] Formally we assume that *, *, and .

[4] In the parametric analysis, we follow Shapley [25] and assume that each firm will receive an amount equal to the value of each potential coalition less its value without the firm, multiplied by the probability that the firm will be in any given coalition. The exact rule for division of the surplus will generally have no qualitative effect on our results as long as each firm's share of output is positively correlated with its access to essential assets via coalitions with other parties.

[5] This is because adding suppliers increases the buyer's bargaining power by making it easier to threaten to shift to an alternative source. See [3] for a formal derivation of this result.