Comparing Klein et al. (1978) to formal models


Taking the example of the printing press and the publisher in Klein, Crawford & Alchian (1978, pp. 298-302—focus mainly on the ex post problem on p. 298), try to reformulate the problem on the terms of Tirole’s Chapter on the Theory of the Firm (1988, pp. 21-34). You are advised to pay careful attention to the explicit and implicit assumptions in both texts. In particular, you should begin by trying to think which of the Tirole’s models is closer to KCA and, moreover, which aspects of such a model substantially depart from the main problem that worries KCA.


In terms of Tirole, the situation might be seen as one of bargaining constrained by an incomplete contract. However this incomplete contract has quite different features from that assumed by the incomplete contracting literature. Specially, it is assumed that there are significant enforcement and bargaining costs, and the minimization of these costs is seen by KCA as the force driving economic organization—in this case, decisions about vertical integration. These costs make quite trivial the information asymmetries about value and cost which perhaps because of their mathematical tractability make the backbone of the formal analysis. Students tend to focus wrongly on different varieties of them. Wrongly because the result in KCA is driven by bargaining and enforcement costs which do not fundamentally depend on such asymmetries.

In that part of the KCA paper:

(1)  The parties have agreed to a fixed price contract. However, this contract is costly to enforce, quite different from both the complete or the incomplete contracts modeled by Tirole. This makes irrelevant the distinction between the bargaining and contracting worlds of Tirole.

(2)  Bargaining is costly by itself, meaning that gains from trade are costly to divide, whatever the parties’ information sets.

(3)  Gains from trade do exist for sure (c < v), the cost c and the value v are not stochastic (c = 2,500, v = 5,500 + d) and both, the cost c and the value v, may be thought as common knowledge, because both parties are imagined as expropriating the whole value of the other party’s expropriable quasi-rents (this is not necessary for the argument, however). KCA talk about quasi-rents, which are represented in Tirole by the difference between the value v and the cost c, but this simplification does not affect the conclusions.

We know that, under Tirole’s assumptions, when gains from trade do exist for sure a rigid norm contract specifying at period 1 the price and the volume of the trade to take place at moment 2 is efficient even if there is bilateral asymmetric information about the cost c and the value v. It will be efficient a fortiori under symmetric information. However this result needs costless enforcement.
The main difference comes from point (1): contracts in KCA are costly to enforce. Initially, we might be tempted to think that the KCA problem is not about bargaining, because there is a contract. However, given that in KCA the contract is costly to enforce, we go back to bargaining. Given that parties have to bargain, the bargaining setup becomes relevant. Here, both texts depart substantially.

Thus, about the ex post division of the gains from trade the assumptions are very different:

a) “This more subtle form of opportunistic behavior [ex post opportunism] is likely to result in a loss of efficiency and not just a wealth-distribution effect” (KCA, 1978, 301).

b) “More generally, bargaining under symmetric information is efficient, so the issue of an inefficient ex post volume of trade in a bargaining situation does not arise. (This is a version of the Coase [1960] theorem.)” (Tirole, 1988, p. 22). This efficiency of bargaining refers really to the trade consequences (trade iff c = v) of the result of a bargaining process which is in fact assumed (gains being evenly divided, according to a Nash solution: “Suppose that ex post, the parties’ bargaining leads to the Nash solution: They split evenly any gain form trade” [1988, pp. 24-5; and on similar terms twice on p. 30]).
In fact, the parties are assumed not only to divide gains evenly but also to divide them without incurring any cost in the very process of dividing them. The cost, if any, is related to consequences which are external to the bargaining process (no trade when c < v, underinvestment).
This Table summarizes the differences between KCA and the transaction costs literature, on one side, and the contracting models condensed in Tirole, on the other:

Complete contracts

Incomplete contracts


investment, I

observable, but not verifiable;
therefore, non contractible

Costly to


Value, v,
for the buyer

Different assumptions about
observability and verifiability

observe and verify.
Hence, costly

Cost, c,
for the seller

provide different models with
conflicting results

 to contract

Ex ante trans-
action costs

Contractual completion




Ex post trans-

Contractual enforcement




action costs

Contractual bargaining

costless & well
defined ex ante

costless & well
defined ex ante

costly &

Go back