Contracting to Dis-Incentivize

Contracting to Dis-Incentivize

Desmond (Ho-Fu) Lo Assistant Professor of Marketing

Leavey School of Business Santa Clara University

500 El Camino Road, 221K Lucas Hall Santa Clara, CA 95053

Phone: 408-554-4716 Email: hlo@scu.edu

Giorgio Zanarone

Associate Professor of Business Organization Colegio Universitario de Estudios Financieros

Leonardo Prieto Castro 2 28040 Madrid, Spain Phone: (34) 914-480-892 Email: gzanarone@cunef.edu

and Mrinal Ghosh Professor of Marketing Eller College of Management

University of Arizona

1130 E Helen Street, 320P McClelland Hall Tucson, AZ 85721

Phone: 520-626-7353 Email: mghosh@email.arizona.edu

February 25, 2017 ___________________________________________

This study received financial support from the Spanish Ministry of Economy and Competitiveness through grant ECO2014-57131-R. We thank Robert Gibbons, Kristina McElheran, Joanne Oxley, and participants at the Society for Institutional and Organizational Economics Conference, the Institute for the Study of Business Markets Conference, and the International Conference on Contracts, Procurement, and Public-Private Arrangements for comments. The usual disclaimer applies.

1

Contracting to Dis-Incentivize

Research in collaborative ties in business markets has pre-dominantly focused on studying how governance forms balance the potential gains against transaction hazards within that relationship.

Using an incomplete contracting approach, we examine OEM-component supplier relations and show how the buyer – the OEM – trades off gains obtained from within the supplier relationship against protecting its resources that pre-exist outside the relationship. Adapting a recent

theoretical model to the context of industrial markets, we show that OEMs with high level of pre-existing resources choose closed-price contracts over open-price contracts to dis-incentivize suppliers from investing in capabilities that may facilitate the appropriation of the OEM’s resources. Consistent with this model, but not with alternative theories, our data on component procurement contracts show that OEMs tend to use closed-price contracts when their pre-existing resources are high. The use of closed-price contracts reduces both the supplier’s dedicated

investment and the OEM’s profitability and end-product gains. Our work provides evidence on how parties, cognizant of the “dark side” of relationships, strategically trade off safeguarding pre-existing resources against creating value in inter-organizational relationships.

Keywords: Contract, Price Format, Governance, Firm Resources and Capabilities, Transaction Cost Economics, Resource-Based View, Specific Investments

JEL codes: D23; L14; L22; M21; M31

2

INTRODUCTION

In business market, supply chain, and distribution channel settings, firms often enter into agreements with value-chain partners to offer products and services that are sources of

competitive advantage (e.g., Mowry, Oxley, and Silverman 1996; Jap 1999). These value-added products and services are created through dedicated investments and efforts (e.g., Heide and John 1990; Ghosh and John 2005; Lee 2011) that frequently leverage the partner firm’s technological, engineering, product development, customer equity, and channel capabilities (e.g., Moorman and Slotegraaf 1999; Lo, Frias, and Ghosh 2012). A vast body of literature, both theoretical (e.g., Williamson 1979; Grossman and Hart 1986; Wernerfelt 1997) and empirical (e.g., Heide and John 1990; Houston and Johnson 2000; Jap 1999) has focused on formal contractual and

informal mechanisms to incentivize parties to undertake these value-enhancing investments and activities.

The dominant rationale for contractual safeguards proposed in the literature on

organizational design and governance has been to protect the productive dedicated investments of the investing party from appropriation by its counterparty (e.g., Williamson 1979). What has been ignored in this literature, however, is that dedicated investments can also enhance the investing party’s ability to appropriate its counterparty’s pre-existing resources. Our study explores the implications of this potential “dark-side” motive by testing an integrated governance model that explicates the links between three factors: firm-specific pre-existing resources,

contract design, and within-relationship investments and outcomes. In our context of supplier-

OEM relations, the OEM’s dilemma is that while the supplier’s investment creates value in the

relationship, they may also enable the supplier to build up its capability to exploit the OEM’s

resources for its own private benefit. We investigate how far-sighted OEMs (original-equipment-

3

manufacturers) seek to design contracts – in particular, price formats – that lower the exposure of their pre-existing resources through curbing a supplier’s dedicated investments, which in turn adversely impact value creation within the relationship. In effect, we posit that, to protect pre- existing assets by sacrificing value creation, OEMs are willing to purposefully dis-incentivize their supplier’s investments, suggesting a tantalizing strategizing-economizing trade-off in collaborative ties.

“Dark-side” motives in business markets could take many forms. In a supplier-OEM (original-equipment-manufacturer) context, an opportunistic vendor might make investments to privately benefit from exploiting the OEM’s resources through overt or covert rent-seeking behavior or outright appropriation (Wathne and Heide 2000; Jap and Anderson 2003; Mohr and Sengupta 2002). For instance, suppliers or partners could use their acquired capabilities to shade on quality (Anderson and Jap 2005), exploit the spillover from innovations (Dutta and Weiss 1997), or move up the value chain and sell competing products – as is evidenced from the experience of many European and Japanese automobile manufacturers in the Chinese markets (e.g., the Audi A4 versus the YEMA F-16), and in the computer hardware (Arruñada and

Vázquez 2006) and wireless phone (The Economist 2012; Alcacer and Oxley 2014; Wan and Wu 2016) industries. Such “guileful” behavior whereby suppliers aggressively learn and build

capabilities at the expense of their partner during collaborations (e.g., Jap and Anderson 2003;

Arruñada and Vázquez 2006; Wan and Wu 2016) is consistent with the classic behavioral

assumption on opportunism (Williamson 1979). In turn, the focal firm (e.g., an OEM) with

foresight on the potential harm, could adopt mechanisms to limit such behavior (Wathne and

Heide 2000), to the extent that they find it optimal to safeguard their pre-existing resources from

4

being appropriated, by dis-incentivizing value-enhancing investments and activities by its counter-party (e.g., the supplier).

To generate testable propositions, we first adapt the general model on governing knowledge investment provided by Zanarone, Lo, and Madsen (2015; “ZLM” hereafter) to our context of industrial procurement arrangements. Specifically, in our context (and consistent with ZLM’s general model), an OEM possessing a unique set of pre-existing resources seeks a

potentially value-generating collaboration with a component supplier. To govern this transaction, the OEM has to choose between either a closed-price contract where the division of trade

surplus is agreed upon ex ante before the supplier undertakes its value generating investments, or an open-price contract where prices are negotiated ex post after these investments are

undertaken. The key testable prediction is that OEMs possessing high levels of pre-existing resources would find it optimal to use a closed-price over an open-price contract to safeguard their resources from appropriation. This is because a closed-price contract fixes the supplier’s payoff ex ante and leaves the OEM as the residual claimant of the realized surplus. The close price then dis-incentivizes the supplier from undertaking investments and effort, part of which could permit it to opportunistically appropriate the OEM’s resources ex post. Simultaneously, however, by dis-incentivizing the supplier’s investments, the closed-price contract lowers the value-add in the relationship, compared to open-price contracts.

We then test these predictions using data from 155 procurement contracts between OEMs

and their component suppliers for engineered components that are incorporated into the OEM’s

end product. This context is well suited for the theoretical model for a variety of reasons. First,

many OEMs of industrial products source components and sub-systems through non-integrated

supply chains; hence, procurement contracts are a very prevalent form of governance in such

5

contexts. Second, these OEMs frequently require the suppliers to design and engineer their components and sub-systems not only to the OEM’s specific product architecture but also to provide improved functionality and enhance the performance of the OEM’s end product or system. As such, suppliers have to undertake substantial level of development and engineering effort and investments. Third, these OEMs themselves, by virtue of the engineering and technological capabilities of their systems, are endowed with varying levels of reputation and

“presence” in their down-stream customer markets. As per theory, it is these pre-existing OEM- specific resources that compels the OEM to seek out contractual safeguards, in the form of a closed-price contract which in turn would impact the supplier’s investment effort and

subsequently impact the value-add in the relationship. Finally, although the industrial firms in our sample all engage in procuring engineered components, they show considerable variation in both price formats and transaction characteristics.

Using a two-step endogenous switching regression technique (Maddala 1983;

Wooldridge 2010) that examines the determinants of price format and its effect on supplier’s investment and value-add to OEMs, we find that, in line with our hypotheses, component procurement contracts are more likely to be closed prices when the OEM’s strength in its end- product market is high. Our analysis also shows that the component supplier’s dedicated investment is significantly lower under closed-price than under open-price contracts, and that closed price contracts lead to lower OEM profitability and end-product enhancement.

Taken together, our evidence showcases the implications of potential dark side on inter-

firm contracts and investment decisions and how contracts are used to strategically balance the

protection of firm-specific, pre-existing resources against the creation of value within the

relationship. By employing an incomplete contracting approach, we contribute to governance

6

studies in Marketing in that our empirical analysis shows an exact mechanism underlying the link between pricing formats (governance), investment dis-incentives, and within-relationship value-add. Our evidence contrasts existing work’s sole focus on how market uncertainties lead firms to use price formats as a means to motivate investment and post-contractual adaptation (e.g., Goldberg and Erickson 1987; Ghosh and John 2005; Lo et al. 2012). Equally important, as we explain in the next section, our finding cannot be explained by other prominent governance theories including the property rights theory (Grossman and Hart 1986; Hart and Moore 1988), the classic hold-up model of transaction cost economics (Williamson 1979), or the multi-task agency models (Holmstrom and Milgrom 1991).

LITERATURE REVIEW

Below, we provide a brief review of the relevant literature in inter-firm governance and contract design and situate our research with respect to these streams of work.

On the one hand, the resource-based view (RBV) of the firm (e.g.Wernerfelt 1984) and its knowledge-based view counterpart (e.g., Grant 1996) elaborate on how firm-specific

resources create value and are sources of competitive advantage (e.g., Barney 1991; Slotegraaf et

al. 2003). Firms leverage their resources profiles and enter into collaborative ties with other firms

to generate additional value through dedicated investments (e.g., Madhok and Tallman 1998). On

the other hand, governance theories – in particular, transaction cost economics (Williamson

1979), property-rights theory (Grossman and Hart 1986; Hart and Moore 1988), and adjustment

cost theory (Wernerfelt 1997) – have predominantly focused on designing governance structures

to mitigate either the hold-up or the adaptation problem that arise from uncertainties within

collaborative ties. A long line of empirical research has used insights from these theoretical

approaches to show how governance arrangements, both formal (e.g., Goldberg and Erickson

7

1987; Crocker and Masten 1991; Crocker and Reynolds 1993) and informal (e.g., Heide and John 1990; Anderson and Weitz 1992), are used to tackle the safeguarding and adaptation problems by incentivizing dedicated investments within the relationship. For instance, as

prescribed in the standard incomplete contracting argument, Crocker and Reynolds (1993) show that flexible pricing arrangements encourage adaptation to uncertainties and motivate dedicated investments. These theories, however, have not considered the role of pre-existing resources (firm heterogeneity, in general) that are independent of the ex post adaptation issue in the design of governance mechanisms.

There is an increasing realization that firms seeking collaborative ties are not atomistic entities but rather possess pre-existing resources (e.g., Madhok 2002). Those resources should motivate the design of inter-firm governance structures because they can be at risk of hold-up in close relations (Helper and Levine 1992; Klein 1996; Ghosh and John 1999; ZLM). These are important and practical concerns because as Arruñada and Vázquez (2006), Alcacer and Oxley (2014), and Wan and Wu (2016) show, a supplier can acquire capabilities in collaborative relations to appropriate the counter-party’s resources, for example, by directly the downstream product market or developing products for the buyer’s competitors. Notice that, in the classic hold-up problem, it is the investing party, rather than the party benefiting from the investment, that at risk of appropriation (Williamson 1979).

Our empirical framework builds on these recent developments that integrate the resource- based and governance perspectives (ZLM). We explicitly test not only the role of pre-existing resources into the design of governance structures (i.e., price formats), but also its subsequent impact on the focal dyad’s ability to create value. Our evidence shows how OEM’s non-

contractible, pre-existing resources get entwined with supplier’s dedicated investments and how

8

price formats simultaneously balance safeguarding and productivity concerns within the relationship (see also Klein 1996).

Extant empirical work that is consistent with the integrated RBV and governance perspective sheds light on certain aspects of these models but has not coherently examined the three key inter-connected factors: the firm-specific pre-existing resources, inter-firm contract design, and within-relationship investment and outcomes. For instance, while Bensaou and Anderson (1999) and Ghosh and John (2005) take dedicated investment as an outcome variable, they do not directly link the firm’s pre-existing resources to contract design. Likewise, whereas Lo et al. (2012) show that high level of pre-existing resources in contracts for branded

components are associated with fixed-price contracts, they treat dedicated investment as exogenous and hence cannot link the safeguarding of pre-existing resources (through closed- price contracts) with within-relationship value creating investment activities.

Our data identify the distinctive role of a firm’s pre-existing and non-contractible resource set on contract design, apart from conventional transactional attributes such as

complexity and uncertainty and governance mechanisms such as monitoring and decision-right allocation. We provide systematic evidence on how firms possessing high levels of such resources are mindful of safeguarding those using close prices to dis-incentivize partner’s investment by sacrificing value creation. This dis-incentivizing effect of contracts suggests a weakness of firms who possess large pre-existing resources: it forces them to choose less value- enhancing contractual arrangements. Therefore, we add to the growing literature on the

implications of the “dark side” of relationships (e.g., Jap and Anderson 2003; Arruñada and Vázquez 2006; Wang, Kayande, and Jap 2010). By incorporating firm-specific heterogeneity

1 Repeated interactions and non-disclosure policy may help to reduce opportunistic behavior in our context.

Nevertheless, ZLM show that even when these two mechanisms are available, closed-price contracts are still effective to dis-incentivize dedicated investment and capability build-up to safeguard appropriable resources.

9

which is the hallmark of the RBV literature into the incomplete contracting framework, our research contributes to the study of collaborative ties from the generic theory of the firm simply based on transaction attributes toward a specific theory of a firm based on firm-specific

resources (Williamson 1999; Madhok 2002) where the transactional attributes (e.g., dedicated investments) are endogenously chosen by firms. This also helps us to find empirical evidence in OEM-supplier relations that cannot be reconciled conventional theories on governance design.

Specifically, in the standard transaction cost economics model of holdup and post-

contractual opportunism (e.g., Williamson 1979), dedicated investments made by the supplier are

assumed to only have productive value; hence, using closed-price contracts that reduce the

supplier’s investment would not be in the interest of an OEM-supplier collaboration. In the

property rights models of incomplete contracts (e.g., Grossman and Hart 1986; Hart and Moore

1988), both the OEM and the supplier may invest in the relationship. A closed-price contract that

incentivizes the OEM at the cost of dis-incentivizing the supplier may be optimal if the OEM’s

pre-existing resources are high and thus its investment is more productive. However, these

models would also predict that under a contract that optimizes between both parties’ productive

investments, value-add to the OEM should be enhanced; yet our results show otherwise. Finally,

in a multi-task agency model (e.g., Holmstrom and Milgrom 1991; Slade 1996), it might be

optimal to choose a closed-price contract that reduces the supplier’s margin-enhancing effort,

such as component customization, but increases his cost-reducing effort, provided that the latter

is more important for the OEM (Ghosh and John 2005). However, inconsistent with the multi-

tasking logic, our data show that closed-price contracts are chosen when the OEM has high

levels of pre-existing resources and hence stands to debase quality or performance that results

from reduced investments in customizing the component.

10

MODEL AND HYPOTHESES

In this section, we adapt the ZLM model to develop three testable propositions.

This model helps to illustrate the inter-linkage of pre-existing resources, price format, and investment and value-add outcomes in the context of OEM-supplier relations. While not necessarily being exhaustive to all possible explanations, this model provides a theoretical foundation for our empirical analysis by delineating some of the forces, ceteris paribus, underplay between contracting parties.

Consider an OEM (M) who seeks to procure a component from a supplier (S). This component is integrated into M’s system or end product and sold downstream to customers.

Following standard convention in models of incomplete contracts (e.g., Grossman and Hart 1986; Hart and Moore 1988), in conjunction with the fact that these are sizable corporations in our samples, we assume that both M and S are risk-neutral. To focus on the dis-incentivizing role of price formats, we abstract from uncertainty by assuming that the environment is deterministic – i.e., there are no market fluctuations or technological shocks.

The supplier S undertakes an investment, 𝑎, where 𝑎 ≥ 0. One may interpret it as S’s adaptation effort to understand M’s technology and the needs of its customers, and/or as S’s dedicated investment into building capabilities that will improve the fit of its component with M’s end product. S’s investment is costly to S, and its cost corresponds to the investment level.

We assume the investment cost exactly equals 𝑎. Let this investment generate a value 𝑞(𝑎) for M, where 𝑞

> 0. We assume that 𝑞(∙) is concave (𝑞

< 0) and 𝑞(0) = 0. Simultaneously,

2 This section adapts from the non-renegotiable contract case (i.e., Case 1) in ZLM. We further assume that supplier’s activities in dedicated investments and adaptation effort mostly benefit the OEM through a better OEM system-component fit, and the supplier shares the division of surplus through the pricing contract. Our assumption of the nature of supplier’s investment is legitimized by our measures on supplier investment and adaption effort. See data descriptions in the Method section.

3 Nevertheless, to isolate the effects of our focal interests, we do control for environmental uncertainty in our empirical analysis. See the sections on Method and Results.

11

S’s investment also enables S to cause harm to M, for instance, by appropriating M’s pre- existing resources as suggested by anecdotes mentioned in the introduction by various scholars.

We denote S’s private benefit from acting opportunistically as 𝜋(𝑎) , and M’s loss from S’s opportunism as 𝛽(𝑎)𝜔, where 𝜔 is the value of M’s pre-existing resources. In general, 𝜔 can be interpreted as the value of M’s product-design and customer equity that pre-exist at the time M seeks procuring the component from S. To model the scenarios identified by anecdotal evidence where the higher the value M’s resources, the larger the loss it could suffer due to S’s potential opportunism (e.g., Hamel 1989; Arruñada and Vázquez 2006; Alcacer and Oxley 2014), we assume a multiplicative term 𝛽(𝑎)𝜔 for the potential harm that S could inflict on M.

We further assume that 𝜋

> 0 and 𝛽

> 0, 𝜋(∙) and 𝛽(∙) are concave (𝜋

< 0, 𝛽

< 0), 𝜋

− 𝛽

𝜔 < 0, 𝜋(0) = 𝛽(0) = 0, and that 𝑞

(0) is large enough for S to choose a positive level of the investment 𝑎 in all scenarios analyzed in the paper.

Timing. Figure 1 shows the timeline of events in our model. At stage 1, M and S choose the price terms that will govern their relationship. In particular, they choose the price format for the

component being procured and agree on whether the final price should be specified upfront (i.e., a fixed-price contract) or agreed upon ex post (i.e., an open-price contract). At stage 2, S chooses its level of investment – 𝑎, that is dedicated to M. At stage 3, the contract terms are executed, unless they are renegotiated. Moreover, this investment 𝑎 also provides S ex post capability to

4

) and 𝑞(∙), we would obtain qualitatively identical results provided that the supplier’s potential to harm the buyer through opportunistic behavior is big relative to the potential benefit to the relationship. This condition matches the anecdotal evidence from Hamel et al. (1989) and Arruñada and Vazquez (2006). Further, for the sake of parsimony, we use a single-period game in our model. Nonetheless, our results would extend to a model in which many suppliers interact with many OEM manufacturers over an infinite sequence of periods, with each supplier being matched to one (potentially different) supplier in each period. ZLM has shown these generalizations and extensions in their model, which could be adapted to our context as well.

12

appropriate M’s pre-existing resources 𝜔. Finally, at stage 4, the payoffs are realized for M and S.

< INSERT FIGURE 1 ABOUT HERE >

Informational assumptions. Consistent with standard incomplete contracting models (e.g., Hart and Moore 1988; Aghion and Tirole 1997) and anecdotal evidence from industrial markets (e.g., Hamel 1989; Arruñada and Vázquez 2006), we assume that S’s ex ante investment 𝑎, the value of M’s resources 𝜔, and outcomes 𝑞(𝑎), 𝜋(𝑎), and 𝛽(𝑎)𝜔, are all observable to M and S but non-verifiable. Therefore, no contracts contingent on these variables can be enforced by third parties such as courts. We also assume that S’s ex post decision to potentially act in an

opportunistic manner cannot be prevented via explicit contractual clauses. This is justified by the fact that while non-compete covenants and trade secret provisions may somewhat reduce S’s ability to use its acquired knowledge and capability to compete with M in the short run, they often have limited duration and enforceability (e.g., Liebeskind 1996, 1997; Pooley, 1997; Arora and Merges 2004; Garmaise 2009). In essence, then, S has an opportunity at some point of time to extract private benefits from his built-up capability at M’s expense. Finally, following

standard practice in hold-up models (e.g., Hart and Moore 1988), we assume that whether or not M and S trade the component, and the price at which trade occurs, is verifiable.

First Best

We begin by analyzing the benchmark first-best case where, at stage 2, rather than only look after its own interest, S unselfishly chooses its investment level to maximize the joint surplus given by

𝐽𝑆(𝑎) ≡ 𝑞(𝑎) + 𝜋(𝑎) + [1 − 𝛽(𝑎)]𝜔 − 𝑎.

The first best is given by

13

(1) 𝑎

≡ 𝑎𝑟𝑔𝑚𝑎𝑥

{𝐽𝑆(𝑎)} > 0.

Open-Price Contract

If M and S have agreed to trade but not specified the terms of the trade (i.e. the final price) at stage 1, they will have to negotiate the price at stage 3, after S invests. We refer to this as the

“open-price contract” scenario. The outcome of the negotiation can be arrived at using a Nash bargaining solution and it would be as follows: each party receives its disagreement payoff (i.e., its payoff when M and S do not trade the component) plus one half of the surplus from trade (i.e., of the difference between the joint surplus when M and S trade the component and the surplus when they do not trade it).

Notice that, since S’s opportunistic behavior and the value of M’s resources are non-contractible, S will appropriate M’s resources and receive private benefit 𝜋(∙) regardless of whether the component is traded or not. Hence, M’s and S’s payoffs if the

component is not traded (that is, their “disagreement payoffs”) are [1 − 𝛽(𝑎)]𝜔 and 𝜋(𝑎), respectively. Moreover, the joint surplus obtained from trading the component is simply the difference of what M and S gets from the agreement:

{𝜋(𝑎) + [1 − 𝛽(𝑎)]𝜔 + 𝑞(𝑎)} − {𝜋(𝑎) + [1 − 𝛽(𝑎)]𝜔} = 𝑞(𝑎).

As a result, M’s and S’s post-negotiation payoffs are, respectively:

(2) 𝑈

(𝑎) ≡ [1 − 𝛽(𝑎)]𝜔 +

2

𝑞(𝑎), and

(3) 𝑈

(𝑎) ≡ 𝜋(𝑎) +

2

𝑞(𝑎).

Given its post-negotiation payoff in (3), S’s ex ante investment level, chosen at stage 2, is:

(4) 𝑎

≡ 𝑎𝑟𝑔𝑚𝑎𝑥

{𝑈

(𝑎) − 𝑎} > 0.

5 Asymmetric bargaining power and hence unequal sharing parameters will generate identical qualitative results in comparative statics in terms of the effect of pre-existing resources on contract forms, and the effect of contract forms on investments and outcomes. Technical arguments are available upon request. In our empirical analysis, we do control for bargaining power by using measures of relative OEM-supplier size, number of potential suppliers for the component, and supplier irreplaceability, respectively.

14

Lemma: Relative to the first-best level, under an open-price contract, the level of S’s investment

is inefficiently low (𝑎

< 𝑎

) at low levels of 𝜔 and inefficiently high (𝑎

> 𝑎

) at high levels of 𝜔.

Proof: In Web Appendix WA1.

As in standard holdup models (e.g., Hart and Moore 1988), when the price is open for negotiation, S obtains only half of the value that its component adds to M’s end product.

However, unlike in those models, this weakening in S’s incentives (given that it extracts only half the value) may be overcome by the fact that, under an open-price contract, S’s investment also helps S to develop capabilities that enable appropriation of M’s resource ex post. As such, when the value of M’s resources, 𝜔, and hence the potential ex post gain from S’s appropriation, is large enough, an open price contract will lead S to undertake too much of the investment 𝑎, relative to the first best (𝑎

> 𝑎

).

Closed-Price Contract

Suppose that, at stage 1, M and S instead agree on a closed-price contract that specifies ex ante a price 𝑝 at which the parties will trade the component. At stage 3, given S’s investment level 𝑎, selling the component to M is efficient, so M and S will trade the component at the pre- agreed price, provided that doing so makes them both better off – i.e., 0 ≤ 𝑝 ≤ 𝑞(𝑎). Let 𝑈

(𝑎) ≡ 𝜋(𝑎) + 𝑝 be S’s payoff after it has undertaken the investment. Then, S’s ex ante investment level is given by

(5) 𝑎

≡ 𝑎𝑟𝑔𝑚𝑎𝑥

{𝜋(𝑎) + 𝑝 − 𝑎} > 0.

Anticipating this outcome, M and S will agree at stage 1 on a price 𝑝 consistent with their participation constraints – i.e., such that given 𝑝, M prefers buying from S,

𝑞(𝑎

) + [1 − 𝛽(𝑎

)]𝜔 − 𝑝 > 0,

15 and S prefers selling to M,

𝜋(𝑎

) + 𝑝 − 𝑎

> 0.

Hypothesis 1 (H1): A closed-price contract induces the supplier to undertake a lower level of

investment than under an open-price contract (𝑎

> 𝑎

).

Proof: In Web Appendix WA1.

Since S’s investment increases M’s value from incorporating the component into the latter’s end product, using the same approach in the proof of Proposition 1, we have the following companion prediction:

Hypothesis 2 (H2): M’s value-add from incorporating S’s component into the end product is

larger under an open-price contract than that under a closed-price contract (𝑞(𝑎

) > 𝑞(𝑎

)).

Proof: In Web Appendix WA1.

The intuition is as follows. A closed-price contract reduces S’s incentives to invest, because the payoff that S obtains from selling the component to M is pre-determined and hence does not depend on the level of such investment. Hence, when choosing its investment level under a closed-price contract, S does not put any weight on the value such investment created for M, 𝑞(𝑎). In contrast, S gets half of 𝑞(𝑎) through bargaining under an open price, which motivates it to invest.

Notice that switching from a closed-price to an open-price contract does not change S’s propensity to be opportunistic: for any given level of investment 𝑎, S receives 𝜋(𝑎) from

appropriating M’s resources regardless of contract form in price. As shown by H

and H

above,

what does change across contract forms is S’s ex ante investment 𝑎, and hence the ex post

realized value creation for M, 𝑞(𝑎), and the size of M’s loss from S’s opportunism, 𝛽(𝑎)𝜔. It

follows, then, that a closed-price contract has both a benefit, given by a lower ex post dilution of

16

M’s resources due to S’s opportunism (𝛽(𝑎

) > 𝛽(𝑎

)), as well as a cost, given by a parallel decrease in the value created within the relationship between M and S (𝑞(𝑎

) > 𝑞(𝑎

)) and in the value obtained by S from exploiting M’s resources (𝜋(𝑎

) > 𝜋(𝑎

)). When the benefit is larger than the cost, a closed-price contract will be more efficient than an open-price contract.

Formally, this will be the case if, and only if, 𝐽𝑆(𝑎

) > 𝐽𝑆(𝑎

), or

(6) [𝛽(𝑎

) − 𝛽(𝑎

)]𝜔 + (𝑎

− 𝑎

) > [𝑞(𝑎

) − 𝑞(𝑎

)] + [𝜋(𝑎

) − 𝜋(𝑎

)].

Said otherwise, a closed-price contract is more efficient – and thus chosen – for the relationship than an open-price contract when the “protection” offered by a closed-price contract grants to M’s resources, measured by [𝛽(𝑎

) − 𝛽(𝑎

)]𝜔, plus the saving in S’s investment cost, 𝑎

− 𝑎

, outweighs the lower level of value creation, measured by [𝑞(𝑎

) − 𝑞(𝑎

)], and the supplier’s reduced private benefit from appropriation of M’s resources, measured by [𝜋(𝑎

) − 𝜋(𝑎

)].

Hypothesis 3 (H3): A closed-price contract is chosen over an open-price contract when the value

of M’s resources (𝜔) is high enough.

Proof: In Web Appendix WA1.

If M’s resources were not exposed to S’s ex post opportunism (𝛽

= 𝜋

= 0), as

normally assumed in the classic holdup and incomplete contracting models (e.g., Hart and Moore

1988), condition (6) would not hold because the investment level 𝑎 under closed-price would be

zero and that under open-price would be lower than the first-best one. Hence, a closed-price

contract would never be efficient because, absent ex post appropriation, S’s investment would

have the sole effect of creating value for M. Since a closed-price contract mutes S’s incentives to

undertake investment 𝑎 ex ante, it would further exacerbate the lack of incentive that S already

faces under an open-price contract (Che and Hausch 1999). By the same token, unlike in our

17

model, in a standard holdup model where S’s investment has the sole effect of creating value for M, it would never be in M’s and S’s interest to choose a contract that reduces the value produced within the relationship. In contrast, under some conditions in our model M and S can choose a contract that reduces the value generated in the particular relationship and it is precisely this aspect that will enable us to empirically distinguish our theoretical explanation for the use of closed-price versus open-price contracts from that provided in the existing transaction cost and property rights literatures.

METHOD Empirical Context and Data Collection Procedure

We test our hypotheses in the context of industrial OEMs procuring from independent suppliers engineered components and sub-systems that are incorporated into their equipment or systems.

We hence focused on OEMs operating in three major industry sectors of the US economy: the non-electrical machinery (SIC 35), electrical and electronic machinery (SIC 36), and

transportation equipment (SIC 37) sectors. We use data obtained by Ghosh and John (2005) for their study; hence, we provide only a brief description of the data here. On-site, in-depth interviews with OEM purchasing managers suggested that our core theoretical concepts were material in these settings. Inputs obtained from these interviews were used to develop a pilot questionnaire that was then administered to purchasing managers at 18 OEMs to verify appropriate wording, response formats, and clarity of the instructions. The final survey instrument was constructed based on the feedback. The unit of analysis for our study is a procurement contract between an OEM and its independent supplier for the supply of a

component, or a set of technologically indivisible components integrated into a sub-system, that

are physically incorporated into the OEM’s end-product. “Independent supplier” in our context

18

means a supplier who is not tied to the OEM by cross-equity holdings; hence, joint ventures and other equity arrangements are excluded from our analysis.

The key informant methodology (Campbell 1955) was used to qualify the informants in the study. These individuals were either purchasing managers or directors in industrial OEMs in three different industry sectors: SIC 35, 36, and 37. Multiple telephone calls, five on average, were used to qualify the informant in each firm. These individuals at the OEM firm were then asked to identify their firm’s most important product-line and to identify a procurement agreement with an independent component supplier under which their firm purchased an

engineered component or sub-system. To encourage participation, these informants were offered a customized report that summarized the relationship profiles in the sample and compared their own relationship with the average profile in the data.

This process yielded a total of 521 informants to whom the questionnaires were mailed.

After using reminder cards and follow-up telephone calls and removing responses because of excessive missing data, we obtained a final sample of 155 responses. The sample size is smaller than the original one in Ghosh and John (2005) since we use a larger set of variables and some of those have missing values.

This response rate of almost 30% is similar to the ones obtained in previous studies in similar industrial settings (e.g., Heide and John 1990). Two items that measured informant involvement in and knowledge about the procurement relationship were used to assess the quality of the key informants. The involvement question, “How involved are you personally in your business unit’s dealings with the supplier?” received an average score of 6.40 (s.d. = 0.66, range = [4, 7]) and the knowledge question, “How knowledgeable are you in general about your firm’s dealings with this supplier?” received an average score of 6.38 (s.d. =

6 The original data set in Ghosh and John (2005) has quite a few missing data on some of the variables uniquely used in our study (amount of supplier investment and norm of flexibility). By dropping firms that have missing data, our usable sample size reduces from 189 to 155.

19

0.70, range = [5, 7] suggesting a reasonably high level of understanding the business

relationship. Finally, we also conducted the Armstrong and Overton (1977) non-response test on early versus late responders. We did not detect statistically significant differences on key

demographic variables pertaining to the procurement ties including annual volume of purchase, number of potential suppliers of the focal component, and the proportion of purchase of the component from this supplier.

Measures

We provide below a description of our measures. Table 1 describes the measures and provides the summary statistics. Table 2 shows their pairwise correlations.

<INSERT TABLE 1 & TABLE 2 ABOUT HERE>

Main Variables. Contract form: This measure describes the price format used in the focal contract to procure the engineered component or sub-system. Our measure is adapted from previous work by Crocker and Reynolds (1993), Ghosh and John (2005), and Lo et al. (2012).

Accordingly, we classified closed-price contracts as those agreements in which the OEM and the

supplier agreed to either a fixed price or a price formula that is adjustable but only per some

objective, verifiable criteria that are exogenous to an individual firm’s actions (e.g. based on

inflation in commodity prices, producer price index, etc.). Closed-price contracts hence pre-

determine the division on of trade surplus over the length of the contract. In contrast, we

classified open-price contracts as those that either did not specify a specific price ahead of

shipment, or did specify a price but allowed for negotiated adjustments ex post. Under such

open-price contracts, the distribution of trade surplus is determined ex post. Price format is

coded as a binary variable, with closed-price contracts and open-price contracts being assigned a

value of 1 and 0 respectively.

20

OEM’s strength in its downstream product market: We measured this using a five-item, 7-point Likert scale that measures how much customer value the OEM’s end product commands over competing products and end-product market share and margins. Consistent with our

theoretical construct, this variable (OEM product strength), adopted from Ghosh and John (2005), measures how strong customers perceive the OEM’s product to be, compared to the products of its focal competitors, and hence constitutes a measure of the OEM’s pre-existing resource and capability that is potentially appropriable in a supplier relationship.

Supplier’s dedicated investment: We asked the purchasing manager of the OEM to rate on a six-item, 7-point scale on how extensive the supplier is required to invest in resources, efforts, and training to produce the component that fits the end-product. This measure, Supplier’s dedicated investment, denotes a broad spectrum of tangible and intangible investments

undertaken by the supplier. Notice that a typical dyad in our samples already has had a long experience (over eight years) dealing with each other, so the purchasing manager must have good understandings of its partner’s business including dedicated investment and effort. The purchasing manager was also asked to estimate the total dollar value of the component supplier’s equipment and training expenditures dedicated to facilitating the procurement of the relevant component, choosing from seven rank-ordered intervals (ranging from less than $10,000 to over

$2.5 millions). This rank-ordered variable, Amount of supplier’s investments, is ordinal and acts as an alternative measure of supplier’s dedicated investment.

Value-add to OEM’s end product: To measure the value-add generated within the OEM- supplier relationship, the key informant managers rated on a 7-point scale the perceived

profitability of the end-product under the focal component procurement contract, relative to what

the OEM might have obtained from some other suppliers (OEM profitability). As an alternative

21

measure of value creation to the end-product, respondents also answered on a two-item, 7-point Likert scale on the extent to which the component procured has helped to create differentiation of the end product in terms of customer’s perceived image and competitive advantage (End- product enhancement).

Other Variables. The choice between closed and open price terms as well as the

supplier’s dedicated investments may also depend on variables that are not explicitly included in our theoretical model. First, when the OEM’s ex ante bargaining power (i.e., its bargaining power before entering a relationship with a particular supplier) is high, it may use its bargaining power and seek a closed-price contract to commit the supplier to a fixed and lower price. To control for the OEMs’ bargaining power, we use the total Number of potential suppliers for the component and construct a measure called OEM’s relative size – which is the ratio of the OEM’s to supplier’s sales volume, both in terms of their full portfolio of products. Likewise, the OEM’s ex post bargaining power might be lower if the supplier cannot be replaced easily. As such, the OEM might not be able to implement a closed-price contract, if desired. To control for this, we use Supplier irreplaceability, which measures the number of months the OEM needs to take to replace the current supplier with a new one. We also control for the importance of the component item (Component importance) in our regressions.

Second, the desirability of closed-price contracts, and the level of suppliers’ dedicated investments, may depend on the extent to which price terms are contractible. As such, parties might stipulate closed-price contracts when they perceive that the formal contracts are

enforceable by courts. We measured this using a 7-point item Contract enforceability, which we

expect to be positively correlated with the use of closed-price contracts. Third, several papers

adopting the transaction-cost framework have argued that closed prices are costly to be stipulated

22

and thus less useful in uncertain and complex environments where the terms of trade need to be adapted and modified in the course of the contractual relationship (Crocker and Reynolds 1993;

Bajari and Tadelis 2001; Lo et al. 2012). As such, we include Technological uncertainty (a three- item scale), which measures the unpredictability of the technology involved in the development of the component, and Interface complexity (a single-item scale), which measures the complexity of the interface between the component and the end product.

Fourth, OEMs and suppliers may exhibit behavior that is more cooperative if they have dealt with each other in the past; this familiarity may make the long-term relationship work as a self-enforcing agreement. This may induce the parties to choose more open price terms (e.g., Corts and Singh 2004). To control for such relationship duration, we use Tenure, which

measures the length of the parties’ relationship, in number of years. Cooperative norms have also been shown to be important in industrial contexts (e.g., Macneil 1980; Heide and John 1990;

Heide 1994).We include Norm of long-term relationship and Norm of flexibility, both on a four- item, 7-point Likert scale, to measure, respectively, long-term orientation in the relationship and how flexible the parties are in making adjustments to cope with possible changes in the

environment.

Finally, firms may adopt alternative governance mechanisms in addition to price format.

Our regressions control for three of the commonly used ones. On the one hand, a supplier may be

hesitant to commit dedicated investments due to the classic hold-up concern. However, if the

OEM also makes dedicated investment, that commitment itself would motivate supplier’s

investments (Williamson 1983; Anderson and Weitz 1992; Gundlach et al. 1995). To control for

this, we use a four-item Likert scale to capture the level of OEM’s investment. On the other hand,

to discourage supplier’s opportunistic behavior such as shirking and misrepresentation of

23

information, the OEM would engage monitoring activities and seize more decision rights. As such, we control for the extent of OEM’s monitoring (variable named Monitoring of supplier) across five upstream activities such as manufacturing processes, quality, and technical

specifications. We also include the variable Control of decision rights that measures OEM’s relative control over its supplier on six decisions on their relations such as delivery schedule, pricing, engineering design, and quality control processes.

Having included a battery of control variables, we believe our empirical analysis would be able to isolate the impact of OEM resources on price formats and the effect of price formats on supplier’s investment and value-add for the OEM. However, because of lack of data, we did not include a measure of how much harm is actually caused to the OEM’s resources due to the opportunistic use of the acquired knowledge by the supplier. Nevertheless, the supplier’s ex post opportunism (in terms of using the knowledge to develop its own end products or develop

components for a competitor’s product) would after the completion of the existing OEM-supplier relationship (Arruñada and Vázquez 2006; Alcacer and Oxley 2014; Wan and Wu 2016).

Therefore, our OEM manager informants, who were responding to a supplier relationship that they were currently in, might not observe this kind of harm when they responded to our survey instrument. However, we expect these managers to have the foresight that their pre-existing resources are potentially at risk of appropriation in such supplier ties and hence we expect them to choose governance forms that balance these hazards against the value-generating potential of the supplier relationship.

Measure Reliability and Validity

We used confirmatory factor analysis (CFA) to assess the validity of our multi-item measures.

The CFA model included the measures for the OEM’s strength in its downstream end-product

24

market, the norm of flexibility, and technological uncertainty. The CFA model suggested an acceptable model fit (

= 221.34, p < .05; NNFI = 0.952; CFI = 0.968; RMSEA = 0.065). Each item loaded significantly (minimum of 0.66) on each of the hypothesized constructs suggesting good convergent validity. In addition, the average variance extracted (AVE) ranged from 0.64 to 0.77 and we found that the AVE for each construct exceeded the squared inter-construct

correlations suggesting good discriminant validity (Fornell and Larcker 1981). Overall, our analysis provides confidence in our measures and constructs.

Common Method Variance Analysis

Common method variance is always a concern, especially with perceptual measures in survey

data collected from one source. We used a marker variable approach suggested by Lindell and

Whitney (2001) to test for common method variance. Specifically, we used two different

variables – qualification of service provided by the supplier and monitoring of the supplier’s

quality control procedures – each of which is theoretically unrelated to our key dependent

variables. We then estimated the correlations between all our relevant constructs and each of

these variables and found that none of the correlations was significant (p > 0.10). In addition, we

also used the Harmon one-factor test (Harmon 1976) and found that the highest factor accounted

for only 9.03% of the total variance explained. Together, these results suggest that common

method variance is not a concern in our data.

25 RESULTS Estimation Approach

We test the following hypotheses for a collaborating OEM-supplier dyad: (i) closed-price contracts should be used when the OEM’s strength in its downstream product market is high (H

), and (ii) the supplier’s dedicated investment and the value-added to OEM’s end product should both be lower under a closed-price contract than under an open-price contract (H

and H

respectively). Evidence that is jointly consistent with these hypotheses would be contrary to predictions from a standard holdup (Williamson 1979), multi-tasking (Holmstrom and Milgrom 1991), or property rights (Grossman and Hart 1986) models. Recall that in those models,

contracts are always designed to incentivize an optimal level of dedicated investment with the sole purpose of enhancing the value-add within the relationship but without any ex post harm to partner’s resources. Under such circumstances, ceteris paribus, either party would not desire closed-price contracts that reduce the value-added in the relationship.

Since contract form is an endogenous decision variable, H

and H

should not be tested by simply regressing Supplier’s dedicated investment (or its amount) and OEM’s profitability (or end-product enhancement) on Price format; otherwise we would obtain biased and inconsistent estimates (Heckman 1978; Lee 1978; Shaver 1998). Since we have full data on the outcome variables under both price formats, we use the endogenous-switching regression approach rather than the better known sample-selection regressions to correct for the endogeneity of contract choice (Maddala 1983; Wooldridge 2010, pp.948-951). In particular, our empirical model is formulated as a system of the following two equations:

(7)

is a probit model named the “switching equation,” whose dependent variable C

takes value 1 if

26

* 0

Ci 

, and value zero otherwise, and

(8)

,

is an ordered-probit model named the “outcome equation,” whose dependent variable

takes value y

 1 if

, y

2 if

, …, y

h if

, where k

,…, k

are threshold parameters.

In the contract-choice equation (7), C

is the dummy variable for price format (closed- price contract = 1; open-price contract = 0), ω

is a measure of the OEM’s resources (measured by OEM strength in its downstream product market), the vector of regressors z

includes all other variables. In the outcome equation (8) where we use the ordered probit regression, y

is the ordinal variable of suppliers’ decisions on dedicated investment and the two collaboration outcomes for the OEM – OEM’s profitability and end-product enhancement due to the relationship, x

is a vector of regressors that includes all the variables in z

from the contract- choice equation, except for the instrumental variable – Contract enforceability. α, γ, κ, and θ are coefficients to be estimated.

As discussed earlier, enforceability of formal contractual terms is directly related to the choice of price format that is explicit in the contract. However, typical procurement “boilerplate”

contracts in component or sub-system purchases (e.g., Ben-Shahar and White 2006) do not

explicitly specify the levels of supplier decision on dedicated investment and effort. This is

because such investment is of an ongoing nature and thus non-contractible. Likewise, these

contracts almost never specify collaboration outcomes on OEM profitability and end-product

enhancement, despite the fact that they may stipulate some technical requirements in the

27

component or sub-system being procured. These facts make Contract enforceability a valid instrumental variable for Price format.

The two error terms, u

and v

, are assumed to have a bivariate normal distribution, and the level and statistical significance of their correlation coefficient, ρ, indicates whether price format, C

, is endogenous in equation (8). We estimate the endogenous switching regression of (7) and (8) jointly by using the full-information maximum-likelihood (FIML) method specified in Miranda and Rabe-Hesketh (2006). These authors also provided the corresponding algorithm to Stata, the statistics software we used in our empirical analysis.

Estimation Results

We present our empirical results on the choice of price format and its effect on supplier’s dedicated investment in Tables 3 and 4 and on OEM outcomes in Tables 5 and 6.

Results on price format and supplier’s investment. The first two columns in Table 3 show Model 1’s switching equation in which we investigate the determinants of price format and its outcome equation in which we look at how price format affects the level of Supplier’s dedicated investment. These results are obtained from the joint estimation of the two equations in the endogenous-switching regression specified in (7) and (8). Similarly for the last two columns denoted under Model 2; but there we further include a set of the control variables on alternative governance mechanisms: Monitoring of supplier, Control of decision rights, and OEM’s

investment.

<INSERT TABLE 3 ABOUT HERE>

7

28

Results across the two models are qualitatively similar. Importantly, the contract-form equations shown in columns (1) and (2) find strong support for H

that hypothesizes that the usage of closed prices (versus open prices) increases in the level of OEM resources. In both models, OEM product strength is positively associated with the adoption of closed-price

contracts (column 1: γ = 0.50, and column 2: γ = 0.46), suggesting that closed-price contracts are chosen when OEMs bring high value pre-existing resources into the relationship.

On other variables that may determine price formats, we find that high Technological uncertainty is associated with the usage of open-price contracts. When technology related to the component is highly unpredictable and fast evolving, parties may not be able to specify technical features of the component ex ante in the contracting stage; therefore, it is more difficult to

stipulate the division of the trade surplus in a closed-price contract. As one would also expect, open-price contracts are more likely when Norm of flexibility in the relationship is high.

However, the usage of closed price contracts is increasing in Norm of long-term relationship.

This result appears to be inconsistent with what was suggested in earlier empirical work on long- term pricing (e.g., Crocker and Masten 1991; Crocker and Reynolds 1993). However, recent studies have found that contracting parties may still opt for fixed prices for individual

transactions in the short run when frequent interactions are involved in the future (Lafontaine and

Masten 2002; Masten 2009). This is because a stream of fixed prices repeatedly stipulated at any

given time provides many opportunities for the parties to compensate or return favors carried

over from the past. Also, notice that our result on the norm of long-term orientation is obtained

after accounting for many other effects. The pairwise correlation between Norm of long-term

relationship and Price format is -0.03.

29

As one would expect, closed-price contracts are more likely when Contractual

enforceability is high (in both models, α = 0.16 with p-values = 0.11). Likewise, high Supplier irreplaceability has a negative association with closed-price contracts. This is not surprising either: when suppliers are harder to be replaced, the OEM’s has less bargaining power to demand closed-price contracts. Effects of other variables on price format, including years of dealing, monitoring, interface complexity, OEM investment, control of decision rights, and industry fixed effects, are not statistically significant.

Results on outcome equation specified in (8) are shown in columns (1') and (2'), with Supplier’s dedicated investment as the dependent variable. Our data provide strong support for H

. Specifically, we find that Supplier’s dedicated investment is significantly lower under closed- price contracts than under open-price contracts (column 1': θ = -, and column 2': θ = -).

This is consistent with our core proposition that closed-price contracts dis-incentivize supplier’s investment.

Additional insights can be gleaned from the results in those two columns, after taking

into account the dis-incentivizing effect of contracts. Specifically we find that, OEM product

strength and its relative size to the supplier are both positively correlated with Supplier’s

dedicated investment. These capture the scale effect of the OEM. Supplier’s investment also

increases in the complexity of the interface between the component and the end product and in

Tenure which may imply a more trustworthy and relational dyad. On the other hand, highly

unpredictable technologies would make planning and executing of investment plans more

difficult. This may explain the negative coefficient of Technological uncertainty. Moreover, the

fewer number of available suppliers and the harder it is for the OEM to replace the supplier, the

higher the dedicated investment made by the supplier, although the positive result of Supplier

30

irreplaceability is weakened in column (2') where we control for additional governance mechanisms.

Indeed, governance mechanisms other than price format matter in suppliers’ decisions on investing in the relationship. Supplier’s dedicated investment increases in both cooperative norms of flexibility and long-term orientation. OEM’s commitment in terms of its own dedicated investment has a positive impact on supplier investment, as one would expect. Nonetheless, when the OEM conducts extensive monitoring on the supplier or controls more decision rights in the relationship, the supplier becomes more reluctant to invest. Finally, suppliers in SICs 35 and 36 have a higher baseline investment than those in SIC 37.

The bottom of the table provides the Wald statistics, which suggest good overall fit for these set of regressions. Furthermore, the statistics of the endogeneity test on Price format reject the null hypothesis that the variable is exogenous. This gives us confidence of adopting the two- stage endogenous-switching regressions to correct for potential bias.

Our second set of results in Table 4 shows how the alternative dependent variable Amount of supplier investment is affected by price format, which in turn depends on OEM’s product strength. The results in columns (1) and (2) are similar to those in the contract-form equations in Table 3. Again and supporting H

, OEM product strength leads to the usage of closed-price contracts (column 1: γ = 0.41, and column 2: γ = 0.49). Closed prices also increase in the degree of Contract enforceability (α = 0.21 in column 1 and α = 0.18 in column 2).

Moreover, like the previous set of results, higher technological uncertainty and partnership that is more flexible are positively associated with open-price contracts.

<INSERT TABLE 4 ABOUT HERE>

31

Outcome equations in columns (1') and (2') show that the amount of supplier investment is decreasing in closed-price contracts (θ = -1.04 in column1' and θ = -1.42 in column 2'). These results again strongly support H

. Furthermore, and similar to previous results in Table 3, OEM strength in its downstream product market, interface complexity, years of relationship, and supplier’s irreplaceability are all positively associated with supplier’s investment in dollar amount. Lastly, both cooperative norms lead to higher investment levels, and negative associations are observed for OEM’s control of decisions rights and monitoring activities.

To summarize, results in Tables 3 and 4 are consistent with the logic of designing

contracts to mitigate the “dark side” of relationships. Specifically, closed-price contracts are used to dis-incentivize supplier’s dedicated investment and effort, especially when OEM’s have high- value pre-existing resources at stake.

Results on price format and value-add for the OEM. Tables 5 and 6 respectively show results on using OEM’s profitability and end-product enhancement as the dependent variables in the outcome equations. In the two tables, the first-stage switching regressions on the

determinants of price format in all the four models collaborate to those obtained in the previous set of results. We would like to just point out that OEM product strength again leads to the usage of closed-price contracts, supporting H

.

<INSERT TABLES 5 & 6 ABOUT HERE>

The outcome equations in both models in Table 5 support H

as well (θ = -1.17 in

columns 1' and 2'): closed-price contracts generate lower value in terms of OEM’s perceived

profitability obtained in the relationship. This result is generally consistent with the rationale of

the dis-incentivizing effect of closed-price contracts on value creation when the OEM’s pre-

existing resources may be subject to potential guileful behavior by the supplier.