There is more to contracts than incompleteness: a review and assessment of empirical research on inter-firm contract design

The meaning of this expression will be made explicit in Sect. 2.

Throughout his exposition Stinchcombe referred to ‘hierarchies’. We assume that he borrowed the term from TCE itself, without implying that the organizations that are substituted by contracts necessarily score high on hierarchical intensity. For this reason we prefer to use the terms ‘integrated structures’ or ‘organizations’.

Indeed, this requirement wipes away the bulk of the TCE-inspired empirical literature on contracting and restricts the target population to a few dozen articles.

For clarity’s sake, this means neglecting essentially those studies in the specialized literature on franchising that have investigated the antecedents of variables like the level of the ‘initial fee’, ‘royalty rate’, etc. The interested reader may refer to Lafontaine and Slade (1998) for an excellent review of the empirical literature on franchising.

We neglect also some specialized literature, like that on public debt and agricultural contracts.

Readers interested in this kind of studies may refer to a paper by Keser and Willinger (2002).

“Previous literature [focused] only on the strictly ‘monetary’ aspects of the contracts” (Arruñada et al. 2001: 257). “Empirical transaction-cost research on contract design has looked primarily at three types of provisions: incentive provisions, pricing structures and price adjustment methods” (Masten and Saussier 2002: 285).

Hubbard and Weiner (1986) have also interpreted take-or-pay provisions as efficient responses to the need for adjustment in long-term contracts. DeCanio and Frech (1993) show how an efficiency interpretation of take-or-pay provisions in natural gas supply is more convincing than alternative arguments based on market-power, and provide an estimation of the efficiency gains entailed by vertical contracts with minimum bill provisions.

We shall discuss the implications of this study for contract ‘completeness’ in Sect. 5.4.

This study also found that contracts associated with lower incentive intensity tend to be chosen as prior relationships between the parties (measured at the site level) increase.

In Saussier’s (2000) reading of this article Crocker and Reynold’s decision to focus on the probability of each contracting party to behave opportunistically was due to data limitations that did not allow measuring asset specificity.

“For cost reductions in operating and maintenance activities, which are difficult to measure with RIB’s [company name] cost data, the partners in good faith simply agreed to ‘negotiate a reasonable estimate’ of the savings, to come to a fair division of the alliance’s financial benefits” (Dekker 2004).

Strictly speaking the data analyzed are not exclusively contract clauses since the variables are coded from information collected by a specialized industry analyst that relies on a variety of sources, besides contracts.

Given the puzzling nature of these results, it is a bit unfortunate that the authors did not discuss in detail the issue of endogeneity since it is perfectly conceivable that the financial strength of the R&D firm is affected by the number of patents it holds, the proxy for project maturity.

Clearly only some of them relate to decision-making.

Consistent with the predictions of agency theory (Holmstrom 1979), the paper also found that the pay-performance sensitivity of entrepreneur’s remuneration decreases as asymmetric information about venture quality declines.

In this paper subjectivity in the measurement of this and other independent variables clearly could be an issue. To circumvent this problem the authors supply readers almost literally with each sentence in the investment analyses documents that relate to the focal independent variable, and the way it was coded.

“Higher internal risk is associated with more VC control, more contingent compensation to the entrepreneur, and more contingent financing in a given round (...) Overall, we interpret these results as very positive for the agency theories (...) External uncertainty is also related to many contractual features. Like internal risk, higher external risk is associated with more VC control and more contingent compensation (...) with increases in VC liquidation rights (...) These findings are highly inconsistent with optimal risk sharing between risk-averse entrepreneurs and risk-neutral investors” (Kaplan and Strömberg 2004: 2199).

“Execution risk is significantly positively related to founder time vesting provisions and negatively related to contingent compensation and VC liquidation rights” (Kaplan and Strömberg 2004: 2200).

Just to mention a few, the manufacturer has the authority to decide the sales targets, the size and décor of the show room, to set the maximum authorized price, etc.

Here we are using the term ‘fiat’ simply in the sense of a right to make decisions, even against the will of the counterparty. Following Williamson (1991) it could be argued that in a contractual relationship such a right is qualitatively different from that of an internal organization, since “courts will refuse to hear disputes between one internal division and another” over technical issues (Williamson 1991: 274). However, such a difference is no longer clear if the parties wave their rights—as they often do (Ryall and Sampson 2003: 14; Grandori and Furlotti 2007: 29)—to bring disputes to courts.

Arruñada et al. (2001) also consider monitoring rights. We shall treat monitoring as a separate dimension and report their findings later.

It must be noticed that the contract the authors focus upon only improves the payoff of the financing firm, not the overall surplus. Therefore the allocation of property rights it establishes is profit-maximizing for the financing firm only if it is assumed that the R&D firm is financially constrained, hence unable to compensate the financier for agreeing to a different arrangement.

The dependent variable is operationalized in two alternative ways. All the operationalizations deliver approximately the same results. The operationalization of the main independent variable (non-contractibility of output) takes advantage of a particular feature of biotechnology research, where it is easy to classify projects according to the fact that a lead product candidate is specifiable or not at the time of the agreement.

These clauses include supplier’s threats, like the right of exploitation of further developments of the contractual project, and buyer’s commitments, like the right of the supplier to be informed about further developments.

‘Free riding hazard’ is a variable capturing the interaction of the brand-name value and the spillover of the effects of franchisee’s improper behavior on the rest of the franchise.

The items considered are ‘financial penalties for underperformance’ and ‘right to terminate for underperformance’.

The authors analyze also the influence of relational mechanisms on other contract terms. We shall present other results from this study in Sects. 4.2.4 and 5.4.

As argued by Klein and Murphy (1988), as long as the marginal return to a franchisee is only a fraction of the total return of an extra sale, the franchisee chooses to provide a lower amount of services than would be optimal from the point of view of the whole franchising network.

The prescription of a specific amount of advertising does not remove the externality, so that the marginal return to the agent of additional expenditure is lower than his marginal cost. Thus if actions could not be observed, the franchisee would still have an incentive to free ride.

We take this article to represent a series of four that Joskow published between 1985 and 1990 on contracts between coal suppliers and electric plants.

The authors discuss at some length an issue of identification (whether contract duration reflects hold-up risk or the fact that it takes longer to complete a novel project) and conclude that upon controlling for task characteristics that may influence project length independently of contracting hazards, contracts for novel technologies are still significantly longer than contracts for more mature ones. However, the authors had to make do with the limited information about task characteristics that is available in their dataset. Hence there is room for future studies employing richer databases to try isolating project effects and contracting effects.

The last of these relationships is fairly easily understandable. The first and the second one warrant a little clarification. As the authors explain, “contingency planning can place limits on how much of the supplier’s proprietary technology must be revealed in the event of changes to the schedule or the addition of new features” and “the parties can outline exactly what access is allowed and what steps will be taken if certain problems occur that may impact the use of the supplier’s proprietary technology” (Mayer and Bercovitz 2003: 14–15). This explanation makes clear that contingency planning, qua planning, not only enhances flexibility but also specifies, and thus constrains, how the parties will respond to certain changes. To the extent to which contingency planning constrains responses, it is a little surprising that it has been found efficient in situations characterized by one type of interdependence that prima facie could be described as ‘reciprocal’. In fact under those conditions organization theory would typically recommend coordination by mutual adjustment, rather than by plan (Thompson 1967). Although the coefficient of task interdependence is significant at a very high confidence levels, we think this is an issue that requires further investigation.

Other findings of Argyres et al. (2007) are mentioned in the Sect. 5.4.

As noise increases, “the difference in the cost of providing the control right for high and low quality firms becomes greater” (Elfenbein and Lerner 2005: 7).

One caveat is in order. We do not claim that unilateral options are suitable to enhance the adaptability of all the types of contracts. At minimum one should be aware that the use of certain unilateral options, like stipulated damages, “requires that most of the uncertainty associated with performance be only on one side of the transaction. If there were uncertainty also on the other side, the penalty stipulated ex-ante could lead to inappropriate incentives ex-post” (Crocker and Masten 1988: 329).

Subject to the disclaimer as per note 35, contingency planning also increases the higher the task interdependency between the parties.

“Actual contracts incorporate few if any explicit contingencies” (Masten 2000: 29).

While the authors interpret their findings as indicative also of complementarity between contractual and social governance, we prefer to say they indicate an impact of the ‘shadow of the past’ on contractual governance. In fact, past alliancing experience is not an element of ‘governance’, susceptible to design. Rather, from a design perspective it can be regarded as a dimension of the transaction.

This result is in contrast with what has been found by Ryall and Sampson (2006). However, it must be noticed that while in Ryall and Sampson (2006) contract detail is a six-values polychotomous variable, in Corts and Singh the parties are faced only with a stark choice between ‘turnkey’ and ‘dayrate’. Thus, parties that opt for more detailed contracts (turnkey) have to accept an accompanying sharp increase in maladaptiveness.

For precision’s sake, Crocker and Reynolds use ‘completeness’ instead of ‘ambiguity’ but the contract characteristic they measure better captures the dimension of ambiguity.

Saussier finds that the dependent variable is positively affected by asset specificity and negatively by uncertainty.

Saussier (2000: 191) made this point, while acknowledging one attempt by Hart and Moore (1999) to develop a theory of ‘partial’ incompleteness.

The actual figure is smaller than it appears. In fact, some of the studies that investigated actual contract content based most of their analyses on readily available variables coded by industry analysts, who did not necessarily have specific theoretical concerns in mind.