relationships, reputation capital and formal contracts in procurement: evidence from it outsourcing

35
Electronic copy available at: http://ssrn.com/abstract=1975215 Electronic copy available at: http://ssrn.com/abstract=1975215 1 Relationships, Reputation Capital and Formal Contracts in Procurement: Evidence from IT Outsourcing Lan Shi Anjana Susarla Department of Economics, University of Washington Tepper School of Business, Carnegie Mellon University December 19, 2011 Abstract Using a novel data set on information technology outsourcing contracts, we examine the tradeoff between providing incentives in controlling costs and lower renegotiation costs, and how relational contracts affect the formal contracts. We find that i) ) complex projects are positively associated with the use cost-plus contracts; ii) a service provider with high reputation capital in fair bargaining (cost-containment) is more likely to be awarded a fixed-price (cost-plus) contract; iii) parties with potential future or contemporaneous businesses are more likely to use cost-plus contracts. Keywords: Information technology outsourcing, fixed price contract, cost plus contract, incentives, hold up, renegotiation, relational contract, procurement contracts, self-enforcing, reputation capital JEL Codes: D2, D86, L24, M15 An extensive on procurement prescribes that contracts should be structured to safeguard the interests of parties involved 1 . However, several aspects of business interactions are not easily measurable or verifiable by a third party, limiting the effectiveness of formal contracts. We examine theories of relational contracting in the context of information technology (IT) outsourcing, which is defined as the sourcing of information systems that confer managerial advantages. With the digitization of virtually every activity undertaken by firms, IT outsourcing can encompass a range of processes and functions including complex business functions. The intangibility and lack of standardization of IT services creates considerable challenges in contracting for IT services. Given the interlinkages between IT and other functions performed in an organization, it could be difficult to assign performance benchmarks or responsibility to the vendor (Whang 1992) 2 . Banerjee and Duflo (2000) observe the prevalence of repeated relationships and the 1 Transaction cost economics (e.g., Williamson 1975) highlight the hazards of ex-post costly renegotiation while the property rights tradition (e.g., Grossman and Hart 1986), emphasizes the effect of (efficient) renegotiation in distorting (unverifiable) ex- ante relation-specific investment. 2 While companies typically engage in reference checking before entering into a relationship with a provider, in the face of confidentiality agreements, and given the complexities of the business and technology environment, it is difficult to obtain reliable information on the underlying causes of performance issues in any given arrangement (Gurbaxani, 2007).

Upload: michiganstate

Post on 11-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Electronic copy available at: http://ssrn.com/abstract=1975215Electronic copy available at: http://ssrn.com/abstract=1975215

1

Relationships, Reputation Capital and Formal Contracts in Procurement: Evidence from IT Outsourcing

Lan Shi Anjana Susarla

Department of Economics, University of Washington

Tepper School of Business, Carnegie Mellon University

December 19, 2011

Abstract

Using a novel data set on information technology outsourcing contracts, we examine the tradeoff between

providing incentives in controlling costs and lower renegotiation costs, and how relational contracts affect

the formal contracts. We find that i) ) complex projects are positively associated with the use cost-plus

contracts; ii) a service provider with high reputation capital in fair bargaining (cost-containment) is more

likely to be awarded a fixed-price (cost-plus) contract; iii) parties with potential future or

contemporaneous businesses are more likely to use cost-plus contracts.

Keywords: Information technology outsourcing, fixed price contract, cost plus contract, incentives, hold

up, renegotiation, relational contract, procurement contracts, self-enforcing, reputation capital

JEL Codes: D2, D86, L24, M15

An extensive on procurement prescribes that contracts should be structured to safeguard the interests of

parties involved1. However, several aspects of business interactions are not easily measurable or

verifiable by a third party, limiting the effectiveness of formal contracts. We examine theories of

relational contracting in the context of information technology (IT) outsourcing, which is defined as the

sourcing of information systems that confer managerial advantages. With the digitization of virtually

every activity undertaken by firms, IT outsourcing can encompass a range of processes and functions

including complex business functions.

The intangibility and lack of standardization of IT services creates considerable challenges in

contracting for IT services. Given the interlinkages between IT and other functions performed in an

organization, it could be difficult to assign performance benchmarks or responsibility to the vendor

(Whang 1992)2. Banerjee and Duflo (2000) observe the prevalence of repeated relationships and the

1 Transaction cost economics (e.g., Williamson 1975) highlight the hazards of ex-post costly renegotiation while the property rights tradition (e.g., Grossman and Hart 1986), emphasizes the effect of (efficient) renegotiation in distorting (unverifiable) ex-ante relation-specific investment. 2While companies typically engage in reference checking before entering into a relationship with a provider, in the face of

confidentiality agreements, and given the complexities of the business and technology environment, it is difficult to obtain reliable information on the underlying causes of performance issues in any given arrangement (Gurbaxani, 2007).

Electronic copy available at: http://ssrn.com/abstract=1975215Electronic copy available at: http://ssrn.com/abstract=1975215

2

complexity of IT that makes it difficult to write and enforce formal contracts. This suggests that relational

contracts could provide an alternate mechanism to uphold contractual obligations that are difficult to

specify upfront and hard to verify ex post.

The two most prevalent forms of contracts for outsourcing IT services are fixed price (FP) and

cost-plus (CP) contracts3. As analyzed by Bajari and Tadelis (2001), spot fixed price contracts carry with

them the transaction (renegotiation) costs associated with a change order. A further weakness is that FP

contracts do not encourage information exchange crucial for a successful partnership between the client

and the vendor (Goldberg 1977). A vendor who expects to profit from change orders can withhold

valuable knowledge of system requirements and spend insufficient effort in defining the system upfront in

order to increase his chance of obtaining a change order.4 While cost-plus contracts enable ex post

adaptation and foster information exchange, vendors have little incentives to keep costs low.

We consider a distinct form of rent seeking activities in the context of IT outsourcing. In

renegotiation, a vendor could divert effort into making the system opaque to an alternate vendor, using

proprietary technology that raises the cost of modification for an alternative vendor, or shirk in

transferring know-how and knowledge of the system architecture to an alternative vendor, increasing the

latter’s integration and compatibility costs. Building upon Baker, Gibbons, and Murphy (2002), we

consider rent seeking in the form of individually profitable but total-surplus-depleting behavior.

From our stylized model of IT outsourcing we develop two propositions. First, complexity of the

project favors cost-plus contracts. Two, the greater the potential for rent seeking, the less attractive is a

fixed price compared with a cost-plus contract. The novel element of this paper is to introduce relational

contracting. While companies typically engage in reference checking before entering into a relationship

with a provider, given the nature of confidentiality agreements and the complexities of the business and

technology environment, it is difficult to obtain reliable information on the underlying causes of

performance issues in any given arrangements (Gurbaxani, 2007). The possibility of a future horizon

provides a mechanism to reward good behavior while sanctioning deviations, facilitating cooperative

outcomes that may be infeasible in one-shot transaction. Implicit agreements whereby the vendor exerts

3 Fixed priced contracts specify a price for a detailed project specification delineated upfront when the contract is signed. When modifications to the initial specification are needed, parties have to draft a change order and negotiate an amendment with the new set of services and pricing clauses. By contrast, in a cost-plus contract, a project can begin without a detailed initial specification, and a vendor is compensated for the costs incurred, at a pre-specified hourly rate, plus a stipulated fee. 4 Bajari and Tadelis (2001) model such inefficiency as the loss arising from information asymmetry between vendors and clients.

3

effort in keeping realized costs low may mitigate some of the weaknesses of cost-plus contracts, and

implicit agreements that reward fair bargaining during renegotiations may mitigate the inflexibility of

fixed price contracts, i.e., relational contracting improves efficiency. However, relational contracts need

to be self-enforcing since vendors’ cost-controlling and fair bargaining behaviors are observable to the

transacting parties but are not verifiable by the third party.

Corts (forthcoming) builds on Klein (1997) that formal contracts affect the self-enforcement

range required to sustain a relational contract; Corts (forthcoming) predicts that a (formal) fixed price

contract with implicit (or relational) contracting makes parties more likely to renege since the gain from

reneging is more easily greater than the long-term gain from cooperative behaviors. We test the prediction

that cost-plus contracts are superior in facilitating relational contracting in the setting of IT outsourcing.

Our stylized model of IT outsourcing also predicts that a history of past interactions might

provide experience and understanding that enables better information exchange and mitigates ex post

transaction costs. In particular, a vendor with greater reputation in cost-containment is more likely given

cost-plus contract and a vendor with high reputation in accommodating changes is more likely given fixed

price contracts. Finally, we examine whether what matters is the reputation in a bilateral context or

reputation in the market (i.e., with other market agents).

One of the challenges in testing predictions of relational contracting is that of obtaining measures

of implicit or relational interactions. The digitization of records of the U.S. Securities and Exchange

Commission (SEC) records provides a novel opportunity to collect detailed data on bilateral interaction.

The data for this study is drawn from 10-K, 8Q and 10Q filings that are part of the public disclosure

filings of the U.S. SEC that mandates that firms disclose material contracts, defined as a contract with a

“substantial likelihood that a reasonable shareholder would consider it important in making an investment

decision”. Even with digital archives, as we describe in the appendix, the process of sorting through

filings from the SEC archives is extremely time consuming, which makes it difficult for buyers to access

the same type of data that we have. In other words, it is fairly difficult for a buyer to obtain information

from prior clients and SEC filings when choosing a vendor. We complement the SEC data with

information on the past relations between the two parties and the service provider’s performance with

collected through trade and business news, firm-level data, and industry reports. In addition, we obtain

proxies for the experience of a particular seller (buyer) with other buyers (sellers) in the market for

4

outsourced IT services to examine whether experience or reputation with the same partner has similar

impacts as the reputation across partners.

The research design is as follows. We first conduct static analysis. Since the complexity of IT

services could encompass several dimensions, we conduct a principal component analysis of various task

dimensions such as business process changes, range of services outsourced, extent to which the contract

involves new systems development etc. to obtain an aggregated measure of complexity. We also

distinguish between the complexity of a service and a vendor’s rent seeking potential by conducting

principal component analysis to obtain an aggregate measure of the rent seeking potential of a vendor.

Our analysis of spot contract choices is consistent with that of prior studies such as Crocker and Reynolds

(1993) that found that cost-plus contracts are preferred for more complex services, and Levin and Tadelis

(2010) that found that more complex services are more likely done by employees than outsourced.

Our main interest is introducing relational contracting and its interaction with the formal

contracts. We measured relational contracting using the presence of clauses to extend contracts. The

identification challenge is that such clauses are a choice variable by the parties, therefore might be

endogenous to the dependent variable. There could also be an omitted variables problem since parties’

decision to include a clause to extend contracts could be correlated with other facts that may also impact

the formal contract structure. We address this issue in the following manner. First we build upon

practitioner insight to consider the role of arbitration clauses, exclusivity provisions, product restrictions,

insurance terms and re-pricing provisions as instruments that impact the likelihood of future interaction

while not affecting the form of contracts. Second, we include as controls the presence of exit clauses and

actions taken by a client in reaching the desired goals from outsourcing to address potential omitted

variables issues. We expand on this in Section 3.5.

In analyzing the impact of implicit contracts on the form of contracts, we find that the prospect of

future interactions favors cost-plus contracts. A vendor with a reputation in fair bargaining (cost-

containment) is more likely to be awarded a fixed price (cost-plus) contract, and the effect is stronger if

the parties expect to transact again. We also find that that the possibility of future interaction with the

same partner matters more than generalized reputation or experience with other market agents.

Our paper makes several contributions to the literature. While prior studies such as Banerjee and

Duflo (2000), Corts and Singh (2004), Ryall and Sampson (2009) and Kalnins and Mayer (2004) focus

primarily on the impact of past relationships on the form of contracts, we highlight the potential for

5

continuing the relationship, which has received relatively limited empirical scrutiny. Gil and Marion

(forthcoming) is an exception; they find that the effect of a supplier’s reputation is stronger when the

future value of ongoing relationships is greater. We also have detailed measures of bilateral interaction,

such as the past and present interactions as well as interactions with other market agents, which allow us

to distinguish between the depth of relationships between a particular client-vendor pair and general

market experience. Thus, we distinguish between the depth of relationships between a particular client-

vendor pair and general market experience. We also gather rich detail about the task characteristics and

contractual contingencies.

The rest of the paper is organized as follows. Section 2 outlines the institutional context and

presents a model and derives predictions. Section 3 explains the data and variable measurements, while

Section 4 provides results. Section 5 concludes.

2. Contracting for Outsourced IT Services

Complex IT outsourcing involves exchange of asset-specific investments, which can involve considerable

intangible and human capital. IT outsourcing service is likely to be incomplete in initial specification for

several reasons. IT services are interlinked with the organization of business processes, which is

idiosyncratic to individual buyer organizations (Bresnahan et al. 2002). An IT service provider’s role then

goes substantially beyond that of supplying a commodity delivered to conform to a standardized set of

specifications. For instance, outsourcing the construction of a data warehouse does not imply that the

provider supplies a repository for the storage of data, but that a data warehouse should lead to streamlined

decision-making and improved customer relationships for the client. Similarly, an IT service vendor

contracted to deliver an IT system to fulfill accounting functions is not simply enabling labor substitution

but participating in organizational restructuring to eliminate redundant processes and to enhance the

effectiveness of internal operations. That is, the role of the IT service provider is to envisage a blueprint

for organizational and process changes. Indeed, buyers often engage in outsourcing of complex IT

services to achieve substantial re-organization and process improvements (Linder 2004), which could

make it difficult to describe the required stream of services at the outset in a formal contract. Thus, the

demands for IT services can be extremely firm-specific and need to consider the underlying business

needs, the firms’ interactions with external partners such as customers and suppliers and the overall

industry context. The pace of technological change creates a need for newer features and compatibility

requirements by the time the project is underway, increasing the incompleteness of the initial

6

specification. Further, the industry and regulatory environment that a buyer is operating in and the

business needs required by a buyer are constantly subject to change. Below are theoretical discussions

that generate testable predictions.

2.1 Production Technology

For set-up, we borrow from Bajari and Tadelis (2001). The initial contract is likely incomplete. A vendor

is chosen to undertake this project, and a contract is signed before the production begins. The payment

method is either a fixed price or a cost-plus contract. In fixed price contracts, if there is a need for

modification to the original scope, the buyer either asks the vendor or an alternative vendor to do it. The

payment for the amendment needs to (re)negotiated, modeled as follows.

We assume that the vendor’s cost of doing the modification is , an alternative vendor’s cost of

doing the modification is where i=H, L, and both the vendor and the buyer know and . We use

Nash bargaining solution (with equal bargaining power) to arrive at the price, i.e., the re-negotiation

during the modification stage results in a price for modificationto be equal to .5

The incumbent service provider with accumulated knowledge of a buyer's needs commands a

unique position to engage in rent seeking behavior, . For instance, a vendor can deliberately make the

system opaque by documenting poorly so that it become difficult for other vendors to undertake

modifications. Alternatively, a vendor could use proprietary technologies, increasing the compatibility

costs of other vendors. Vendors could also shirk in transferring know-how and knowledge of the system

architecture to a potential new vendor, increasing the latter’s integration costs. In particular, we model

that the vendor's rent seeking effort ( ) affects the probability that other vendors’ cost of modification

being high: prob( = )= and prob( = )=1 , where is the effectiveness of the

vendor's rent seeking effort in affecting the probability of happening.

Inefficiency is thus modeled differently than Bajari and Tadelis (2001) where the inefficiency

arises from the fact that the incumbent vendor possesses information on modification costs that the buyer

does not have. Here, it is our maintained assumption that the buyer has the information on the costs, but

we introduce the vendor's actions that enhance private interest yet destroy total surplus.

In cost-plus contracts the buyer reimburses the service provider for costs. Given the nature of a

cost-plus contract, the buyer's requests for modification do not involve a renegotiation of the price.

5 That is, the solution comes from

7

Meanwhile, since costs are reimbursed, service provider has little incentives to control costs (Bajari and

Tadelis, 2001).

2.2 Spot Fixed Price Contract vs. Spot Cost-plus Contract

In the Appendix of our online working paper version, we examine when fixed price contract is the

preferred contract. We first examine the vendor's effort choices facing contracts and then the total surplus

and derive the best contract. The trade-off is that spot fixed price contracts offer strong incentives for

cost-containment, but have high renegotiation costs in the form the vendor's engaging in wasteful actions

to raise its profit, and spot cost-plus contracts are poor at providing vendors incentives to control costs but

are free from renegotiation costs. The predictions from the static analysis are:

Proposition 1: Fixed price contracts become less attractive compared with cost-plus contracts

when the project is more complex.

The intuition is that when the project is more complex, it is very hard to anticipate all possible

states hence there is a greater probability that the contract needs to be modified and renegotiated. Because

of the renegotiation costs, fixed price contracts become less attractive. From the above discussion, we also

have

Proposition 2: Fixed price contracts become less attractive compared with cost-plus contracts

when the potential for the vendor’s rent seeking and holding up of the buyer is greater.

While we derive similar predictions on spot contracting as Bajari and Tadelis (2001), this paper's

main contribution lies in introducing relational contracts, which we discuss below.

2.3 Relational Contracting

Practitioners emphasize the role of collaborative or cooperative relationships in the relation between IT

service providers and buyers. The various forms of inefficiency can be overcome if the two parties

cooperate. In the case of fixed price contracts, when the need for modification arises, it is more efficient

that the vendor accommodates the change request without wasting resources to seek rent. That way, the

buyer will incur less cost while the vendor makes reasonable profit. Similarly in cost-plus contracts, the

vendor can expend effort to contain cost, which results in lower costs and greater total surplus, part of

which the buyer can use to reward the vendor's effort.

Note that these efficiency-enhancing actions – accommodating changes cooperatively and cost

containment effort – are hard to contract on, yet they are likely observable between the parties. Without a

8

third party enforcing a formal contract, though, these implicit contracts to implement cooperative

behavior have to be self-enforcing.

Fixed price contracts with implicit contracts between the parties is an exchange relation where

two parties have an implicit agreement that the vendor accommodate the buyer's needs for changes at

reasonable costs, and the buyer rewards the vendor for that, and the legal contract uses fixed price

contracts. Cost-plus contracts with implicit contracts between the parties are an exchange where the

buyer and the vendor form an implicit agreement that the vendor will contain costs and the buyer rewards

the service provider's cost-containing effort, and the legal contract uses cost-plus contracts. If either party

reneges, the implicit agreements break down, the parties switch back to non-cooperative state and the

transaction is governed by spot contracts. Under either scheme, the element that prevents them from

reneging is the prospect of greater payoff under a cooperative relation relative to a spot transaction.

Corts (forthcoming) offers a theoretical prediction on what contract better facilitates self-

enforcement of relational contracts. When cost-plus contracts are used in an implicit contract relation,

with future business at stake, the vendor knows that if it does not perform well in controlling costs, it will

not get the future business from the buyer and the profit from a cooperative relation. When fixed price

contracts are used in an implicit contract relation, when a reneging opportunity presents itself, the

deterrent is the prospect of future payoff, which is the efficiency gain (from the absence of rent seeking),

but the (future) gains happen with a probability less than 1 since modification needs are not certain.

Therefore, the vendor is more likely to renege in an implicit contract relation using fixed price contracts.

This explains why cost-plus contracts are superior to fixed price contracts in facilitating self-enforcement

of relational contracts. Thus our third testable prediction:

Proposition 3: Relational contracts favor the use of cost-plus contracts.

2.4 The Effect of Reputation Capital on the Form of Contract

The above discussion focuses on the self-enforcement aspect of relational contracts in IT outsourcing

once the implicit agreement is formed. Put it in another way, a relational contract assumes that the two

parties will start by behaving cooperatively and then terminate when either one reneges.

Suppose sellers have different stock of reputation capital with the buyer, we have the following

prediction: Sellers with higher reputation capital in fairly accommodating changes are more likely to be

awarded fixed price contract. The reason is that an act of holding up the buyer will destroy the seller's

9

reputation capital and the future surplus, and this is a greater loss for sellers with an existing high stock of

reputation capital.

The case for reputation in controlling costs is similar. A vendor with a high reputation capital in

cost-containment has greater capability in controlling cost, and thus has higher likelihood of finishing the

project with low costs. We thus predict that a vendor with a past history of good performance in

containing costs is more likely to be awarded a cost-plus contract.

Note that these discussions center on the role of bilateral reputation instead of the parties’

reputation in the market. We conceptualize the reputation as bilateral because, under many circumstances,

the nuances of a relationship and what transpired within a contract are not observable by parties outside

the relationship. Sections 4.4 and 4.5 explore the distinct differences between the role of bilateral

reputation and general market reputation on formal contracts6.

3. Data and Measures

The data used in this study is drawn from public filings of companies. The U.S. Securities and Exchange

Commission (SEC) mandates that firms disclose material contracts, defined as a contract with a

“substantial likelihood that a reasonable shareholder would consider it important in making an investment

decision” as a part of either their annual reports in 10-K, current reports in 8Q and quarterly reports in

10Q filings. Firm and vendor specific information for each contract was supplemented with data from

trade and industry databases such as the One Source Online Business Information database and the

Hoovers database.

The SEC filings provide information about prior contracts, if any, between the vendor and client

as well as the client’s assessment about the prior performance of the vendor in providing the outsourced

functions (provided in the financial statements). We supplemented this data with data from a number of

publicly available databases that aggregate news and press releases such as Dow Jones Interactive,

Factiva database and industry reports and trade and business press that reports on outsourcing deals.

Whenever possible, we verified this data by examining press releases from either clients or vendors as

well as press releases posted on the archived websites of vendors and clients. Typically, a contract

between a client and a vendor consists of a formal contract, that identifies responsibilities both parties and

a payment schedule, and a statement of work (SOW) that focuses on the technical details of the system.

6 The general market reputation is what Banejee and Duflo (2000) used for their reputation variable.

10

Unless otherwise noted, most of our variables are indicator variables that take the value of 0 or 1. In table

4, we show the details of how we measure each of the variables we use in our analyses. In the remainder

of this section, we describe how the variables relate to the concepts of interest.

3.1 Formal Payment Structure

The primary dependent variable we examine is the contract type, i.e., fixed price or cost-plus

contracts, coded from the contract obtained from the SEC filings. Fixed price contracts specify a defined

payment schedule for services specified in the contract7 while cost-plus contracts pay the vendor a

markup based on realized costs, including compensating the vendor by costs incurred when buyers

require additional services that are not defined in the original agreement. Cost-plus agreements are

typically referred to as “time and materials” contracts in this industry. While cost-plus contracts bill the

vendor for incurred costs, in practice both parties expend effort in scheduling, making it difficult for the

vendor to manipulate bids when modifications are desired.

3.2 Characteristics of Parties

Size and Bargaining Power: Client Size and Vendor Size were obtained from the Hoovers database.

According to risk-sharing theories, contract for a larger vendor is more likely to be a FP contract since the

vendor can better bear the risk. It is also possible that the size of a vendor also signals perceived

trustworthiness and competence, which also affects contract form. We also measured size of vendors and

clients by examining whether they are Fortune 1000 or Fortune 500 companies. A measure of Industry

Accreditation of the vendor was coded from firms such as Gartner and Forrester that publish quarterly

rankings of the vendor. A measure of Client Market Power was assessed based on SEC filings, and

measures whether the client accounts for more than 10% of a vendor’s revenue (SEC mandates that firms

disclose this information). Finally, we measured whether the vendor is a publicly traded company since

access to capital markets might impact the risk preferences and therefore the choice of contracts.

3.3 Task Characteristics

3.3.1 Complexity of Service and the Potential for Rent Seeking

We coded a number of measures to assess the complexity of the contracted task. First, we examined the

business objectives of the underlying task. Industry parlance refers to contracts as transformational when

7 Fixed price contracts may include indices for inflation and cost of living adjustments to reflect changes in prices of inputs such as hardware and software and costs of hardware upgrades. Invoices by the vendor typically include a description of the type of service rendered, a brief description of the out of pocket costs, the billing period, and other information requested by the client.

11

the business objective is significant reengineering of processes and job functions (Linder 2004) in the

client organization, where the ambiguity in specifying contract outcomes increases contract

incompleteness. For instance, one of the contracts specifies that the vendor, EDS, “conduct a

comprehensive assessment of the client’s information technology systems in light of the business priorities

and competitive market conditions and growth requirements” before creating a technology plan and

implement the proposed solutions over a period of time. We therefore coded a measure of process

restructuring in the contract. Second, we coded a measure indicating that the purpose of the contract

involves significant new systems development. As explained earlier, new systems development involves

considerable complexity in terms of re-organizing business functions and organizational transformation,

making it difficult to envisage benchmarks and performance standards. Third, we coded a measure of new

technological standards used in the contract. Projects that rely on newer technologies and process

architectures are more complex than ones that rely on familiar and established technologies. Fourth, we

classify services delivered along a typology of 14 different types of sub-services that constitutes IT

outsourcing as specified by management literature (e.g., Lee et al. 2004). We coded a measure of breadth

of activities included in the contract as a summation of the number of sub-services delivered by the

vendor. Contracts with a greater breadth of activities may be more complex in terms of greater challenges

in planning deliverables and executing milestones. Table 2 provides an exhaustive description of activities

included in the contract. Finally, the contractual detail in terms of the number of pages of the contract

provides an indication of the underlying complexity.

One issue to consider is whether the vendor’s potential for rent seeking and holdup is distinct

from that of the complexity of the service description itself. When building an IT system using the

vendor’s proprietary technologies and platforms, the client benefits from the dedicated technological

expertise of a vendor but at the same time is liable to be held up during modifications. We consider three

different dimensions that denote the potential for a vendor to extract rents from the clients. First, we

consider whether the task description depends upon the vendor's proprietary platform (e.g., Kalnins and

Mayer 2004). Second, we examine whether the outsourced task relies upon the use of a vendor's

proprietary technology (Mayer and Nickerson, 2005). Third, we consider whether the task involves

licensing of technologies to the client and create a variable standards licensing that takes the value of 1 if

the vendor employs standards that are owned by the vendor and need to be licensed to the client for use in

the contracted task. Lastly, it is difficult for vendors to hold up the client for projects where the process is

12

well understood by the client. We therefore coded a measure of process maturity based on service

descriptions that denotes how well the client understands the technology and systems architecture used by

the vendor (based on the descriptions in the contract that indicate that the client has prior experience of

the technology and systems being deployed in the outsourcing initiative), and the stability of the

underlying technology used for the contracted task.

3.3.2 Principal Component Analysis of Complexity and Rent Seeking Potential

We conducted principal component analysis to get composite measures of complexity and the

potential for rent seeking by a vendor. We extracted two principal components from this analysis. The

first principal component explained 45% of the variation in our five variables, and the two principal

components cumulatively explained 77% of the variation. The first principal component captures the

impact of complexity of the services, which we label as task complexity, and the second principal

component, which we label the vendor’s rent seeking potential, captures the impact of factors such as the

use of a vendor’s proprietary platform, intellectual property protection clauses etc. that enable a vendor to

extract greater rents from the client. The contract documents also provide us with the contract duration

and contract value. Table 3 presents the principal component analysis.

3.4 Contractual Contingencies and Monitoring Terms

Besides pricing, IT outsourcing contracts contain considerable detail on project management and

performance measurement. We therefore coded whether the contract contains (i) performance milestones

that are tied to specific outcomes (e.g., Lichtenstein 2004), (ii) contract clauses facilitating monitoring,

such as audit rights and inspections, and (iii) service level agreements that guarantee a level of services to

be provided by the vendor. We control for these variables since measures of monitoring by the client

could impact the possible rent seeking by vendors and thereby the form of contract.

Contracts can stipulate channels of communication that help inter-firm interactions and lessen

some of the difficulties in contracting, and therefore affect the form of the contract. For instance, a

contract between Coors and EDS states that each will designate an individual as its project executive,

stipulating that such individuals have day-to-day authority of handling project and contract management.

The purpose of creating such roles and responsibilities is to designate a single point contact of

accountability that is authorized to act as the primary contact for each company. Contracts also stipulate

the frequency of meetings between such key personnel representing the client and the vendor. For

example a clause states: “meetings will be held to discuss daily performance and planned or anticipated

13

activities and changes that might adversely affect performance.” Therefore we coded two measures

denoting whether (a) contracts include processes for inter-firm communication and/or (b) designate clear

roles and responsibilities enabling joint management of outsourced tasks.

3.5 Relationships and Reputation Capital

3.5.1 Expectation of Future Interaction

The possibility of future interactions introduces a potential new mechanism to reward cooperative

behavior of transacting parties, i.e., it makes relational contracts feasible. Contracts can include a

provision to extend the outsourcing arrangement with minimal renegotiation costs at the end of the

contracting horizon and to specify a future horizon of interaction. Such provisions provide an indication

of parties’ expectations of future interaction after the current contract is completed. We also examined

press releases from buyers and news reports in the trade and business press suggesting that parties expect

to continue their contractual relationship into the future. The parties’ expectation of future interaction

provides an indication of the continuation value of relationships, and thus the value of the relational

contract.

3.5.2 Instruments and Controls for the Prospect of Future Interaction

As we explain in the introduction, addressing the relationship between the prospect of future interactions

proxied by the extensibility clauses and formal contracts carries with it a potential endogeneity problem.

We therefore incorporated several measures to address this issue.

Practitioners suggest that parties should include provisions for dispute resolution and price

adjustments that enable smooth resolution to contentious issues ex post. In particular, we consider the role

of arbitration clauses whereby parties have an agreed upon framework for resolving disputes (Overby

2011), product restrictions that clarify the manner in which parties can re-use the technologies and

knowledge generated in the contracting initiative, exclusivity provisions that restrict vendors from

deploying services developed for a particular client for the client’s competitors, and insurance terms for a

vendor to protect against major schedule slippages that are outside the vendor’s control. Mayer and

Argyres (2004) observed that when outsourcing IT services, any disputes over project costs or guidelines

for specification of requirements could seriously disturb not only the project schedule but also erode the

value from the contract. Re-pricing provisions that provide an index for re-pricing of services similarly

make it easier to extend a contract, since parties can avoid being locked into higher rates. We create

indicator variables for all of these factors and use them as instrument variables. We argue that these

14

variables affect the prospect of future interactions but not the form of contract directly. We expand on this

in Section 4.3.2.

The other issue is that of omitted variables. It could be possible that, instead of using the

extensibility clause to promise a future horizon where good performance by a vendor leads to continued

business, clients may intentionally break up the whole business into several parts and use extensibility

clause to continue business only if the vendor’s performance is satisfactory. In examining the presence of

future interaction, we include as controls the presence of termination for convenience clauses that

facilitate smooth exit and therefore pose lower risk to the buyer in offering an extensibility clause.

Another issue to examine is whether parties can take actions that are crucial to the success of desired

targets from outsourcing. The actions taken by both parties to increase the likelihood of such outcomes

are likely to be correlated with the use of future interaction clauses. We therefore control for clauses

detailing a client’s participation in setting standards for delivery time and quality norms as evidence of

actions taken by a client in reaching the desired goals from outsourcing. Finally, we included a number of

controls for size and bargaining power of clients and vendors, which could influence their willingness to

transact in the future, and thus the expectations of future interaction.

3.5.3 Relationship Capital from Past Interactions

Prior Relationships: Information on the history of interaction between parties was collected from the

years 1992 to 2005. The SEC filings and the data from public databases provide information on whether

the vendors and clients have contracted with each other before. The prior contract filings with SEC also

provide us with a data source for the contract type in the prior relationship. Based on the relationship

history, we first categorize whether parties had prior relationships with each other. For parties with a

history of interaction, a detailed examination of public data sources, financial statements and documents

filed with the SEC provide evidence of fair bargaining and a vendor’s performance in keeping costs low.

Fair Bargaining: When a publicly filed contract is renegotiated, parties are required to file the

amendments with the SEC. We therefore examine amendments in a previous contract as a proxy for

amicable bargaining between parties. We create an indicator variable that takes the value of 1if there are

amicable amendment agreements filed for an earlier contract between parties, from data obtained from the

15

SEC’s EDGAR database as well as press releases by client firms.8 As discussed earlier, the purpose of a

modification is to add richer detail to the initial service description, such as including service scope and

requirements not envisaged in the original agreement, as the excerpt from an amendment illustrates:

“(Vendor) will deliver to (Client) software in accordance with acceptance criteria that will be defined as

part of the Work Authorization ("WA") exercise. (Vendor) will deliver this software no later than (date

specified in the amendment).”

An additional issue we need to consider is whether there could be smaller changes needed that

parties can amicably resolve without a formal amendment. Since overruns are fairly frequent in IT

outsourcing, we examine the annual filings of the vendor company to examine whether there are any

references to contractual overruns. The trade and industry press reports on the progress of complex

outsourcing deals including the ones we have in our sample. Therefore we scrutinized such reports as an

additional method of gathering data on amicable amendments in a prior agreement. For instance, a trade

report measured that parties amicably resolved ex post changes without a formal amendment.

Fair Cost Performance: To obtain data on vendors’ perceived behavior in keeping realized costs

low, we relied upon an additional source of data, the annual financial reports and investor statements

released by clients and vendors. Vendors regularly highlight such favorable performance in their press

releases and such information is corroborated by the investor briefings and annual reports, disclosed as

part of 10-K filings where clients report assessments of vendors’ performance9. In obtaining this measure,

we need to distinguish fair cost performance by a vendor from the expected performance that is

endogenously related to the decision to outsource. Therefore we looked at the start date and duration of

the prior contract between parties. A press release or financial statement that dates to the start (or

beginning stages) of a contract is likely to mention expected performance while press releases or

statements that are near the end of a prior contract or post contract execution refer to realized costs.

Contemporaneous Relationships: The public databases provide a rich source of data to examine

whether parties have other ongoing relationships, such as marketing alliances, business partnerships, and

strategic relationships in tandem with the outsourcing initiative, which formed the basis of coding an

indicator variable for contemporaneous relationships between clients and vendors.

8 This variable takes the value of zero for parties that do not have a history of amending a contract since they do not have a chance to learn whether the vendor engages in fair bargaining. 9 We coded this measure only if we were able to match details from both vendors and clients.

16

Prior Contract Termination: When contracts are terminated early, clients (and usually vendors)

disclose such terminations in their press filings and investor briefings, which provided us with the data to

code contract cancellations in a prior relationship.

3.5.4 Market History

Data from public databases and the trade and industry press provides a source to obtain details about the

other contracts signed by clients. We therefore examined a five-year horizon preceding the contract

signing date as well as a five-year horizon following the date of contract signing to examine other

outsourcing engagements entered into by each client firm. For each contract in our sample, we examine

whether the client has signed contracts similar in scale (dollar value) and scope (the set of services

outsourced). This would give us an idea of how easy it is to switch providers and an indication of the

buyer’s experience with outsourcing. A measure of interactions between a particular client and its

alternate vendors provides a proxy for the ease of switching a particular vendor. Further, the ability to

contract with alternate vendors implies that the client may place a lower value on continued interaction

with a particular vendor.

To obtain the market history of each vendor, we examined whether vendors had signed contracts

with buyers in the same industry in a five-year horizon preceding and succeeding the date of contract

signing. The market experience of a vendor in the buyer’s industry provides an indication of how well the

vendor understands a client’s business needs as well as providing an indication of the rent seeking

potential of a vendor. We also considered mergers or acquisition activity in the vendor company, since

such activities could disturb the continuity in the interactions between the seller and the buyer, and

thereby diminish the value of future interactions. We therefore obtained a proxy for merger and

acquisition activity of vendors by examining the Hoovers and Compustat databases.

4. Econometric Approach and Results

4.1 Descriptive Statistics

Contracts in our sample were written during the period from 1998 to 2005. A majority of firms

only have one contract in the sample while the remaining have about two to five contracts. The contracts

in our sample represent a broad spectrum of outsourcing activities ranging from a comprehensive

information technology services agreement involving parties such as General Motors and Electronic Data

Systems (EDS) to smaller contracts for fairly standard data processing services such as that involving a

small California-based bank. Large clients in our sample include several Fortune 500 firms such as Coors

17

and Goldman Sachs. Large vendors in our sample include EDS, IBM and Computer Sciences

Corporation.

Table 5 provides the descriptive statistics of contract terms. The sample exhibits considerable

variation along the types of vendors and clients as well as the type of services that are being delivered in

the outsourcing agreement. About 60% of our sample contains fixed price contracts. On average, the

value of a contract is $49 million and lasts 47 months. An average client in the sample has 13444

employees while the average vendor has 12680 employees. Our sample of vendors contains large

companies such as IBM, which accounts for the fact that the average vendor size is comparable to the

average client size. Thirty-three percent of the vendors enjoy a high industry reputation. Prior

relationships are fairly common (49 percent of contracts). In 21 percent of the contracts the buyer and

seller had a successful (thus amicable) amendment to their previous contracts. In 20 percent of the

contracts the buyer is satisfied with the seller’s performance in reducing costs in their previous contract.

In about 10 percent of the contracts the buyer and the seller cancelled their previous contract. In about 53

percent of the contracts, the parties expect to transact with each other in the future.

Table 6 presents several cross-tabulations of contract types as a function of task characteristics,

future interaction potential, and bilateral reputation.10 The raw data shows that contracts with a prior

relationship between parties and those where the vendor performed favorably in cost containment in an

earlier contract are more likely to be cost-plus contacts; contracts where the vendor successfully reached

an amendment agreement with the buyer in an earlier contract are more likely to be cost-plus contracts. At

first sight this finding contradicts the predictions from the analysis in this model, but as we will show

later, once we control for other relationship variables, past good performance in accommodating changes

is positively associated with fixed price contracts. We also find that contracts with prior relationships are

more likely to be longer, larger valued and more complex, all of which could impact the contract type.

We thus control for these variables in the contract type regressions.

4.2 Trade-off in Spot Contracts

Our empirical identification strategy relies on cross-sectional variation in IT contracting. We

estimate a binary choice model of contracting to test the predictions from the theoretical discussion. The

variation in the type of services provided as well as the degree of past and future interactions between

10 Contract renegotiation in a past transaction is fairly common. In the earlier relationships, we observe that 67% of contracts that were amended were fixed price contracts.

18

clients and vendors provides a rich source of heterogeneity at the contract level. Our empirical approach

hinges upon a detailed understanding of the outsourcing and information technology development

context. We adopt an econometric specification of the form:

Pr

where i refers to the client and j to the vendor, ijy is an indicator variable that takes the value of 1 if the

contract between the client i and the vendor j is a fixed price contract, V is a vector of vendor-specific

characteristics, C is a vector of client characteristics and ijX is a vector of project characteristics such as

the project complexity, the maturity of the underlying technology and business processes, the project

duration and value, the communication between clients and vendors and the extent of monitoring

undertaken by the client. This specification ignores the impact of past or future relationships on contract

choices.

Our results in column 1 of Table 7 are consistent with prior literature that has examined spot

contract forms. We find that the principal component measuring task complexity is positively associated

with cost-plus contracts, confirming the predictions of theoretical models such as Bajari and Tadelis

(2001) and empirical literature (e.g., Crocker and Reynolds 1993). Fixed price contracts are plagued by ex

post transaction costs when the vendor can take advantage of the need for renegotiation and raise

individual payoffs by engaging in wasteful actions that raise her bargaining power and rents that accrue

from renegotiation.

Other coefficients are reasonably estimated. Larger buyers are positively associated with fixed

price contract, contrary to prediction from risk-based theories, yet consistent with theories based on

bargaining power. Larger suppliers are positively associated with cost-plus contract, also contrary to

prediction of risk-based theory, yet consistent with a story that larger supplier have more bargaining

power. Another possibility is that larger suppliers are perceived to be more trustworthy, and thereby more

likely to be awarded a cost-plus contract. Frequent communication between the vendor the client is

positively associated with cost-plus contracts, consistent with the need for cost-control in cost-plus

contract and also the information-facilitating effect of a cost-plus contract.

4.3 Impact of Relational Contracts on the Form of Contracts

We test the prediction that cost-plus contract is superior in self-enforcing relational contract by examining

how the structure of the formal contract varies when there is a possibility that parties would transact with

19

each other in the future. Column 2 in Table 7 introduces the indicator variable for potential future

business measured by the presence of clauses to extend contracts. The coefficient is significant and

negative.

4.3.1 Validity of Using Extensibility Clause to Measure Future Interaction

One concern is that buyers use extensibility clause to intentionally divide up a whole contract;

that way, the buyer can dump the vendor if it does not perform (or continue using it if it performs well.)

That is, the extensibility clause, compared with the absence of it, indicates greater uncertainty of future

business rather than more certainty.

Second, buyers and sellers engage in extensive discussions in outlining the sourcing requirements

and in negotiating contract terms prior to signing a contract. The process by which parties decide to enter

into a contracting agreement and define the terms of the exchange is unobservable to the econometrician.

It is possible that contracts may be more likely to promise future interaction when parties engage in

trustworthy behavior in negotiations prior to the start of the contract and laid the foundation to manage

the relationship in a cooperative manner, and such conditions could be characteristic of cost-plus

contracts.

Third, IT vendors face fluctuating revenues from the stream of contracting arrangements they

enter into, and this uncertainty in revenue also translates to greater uncertainty to the seller in planning for

personnel and inputs to the production process. The possibility of future interaction offers some risk

mitigation to a seller in assuring a stream of future revenues, i.e., the extensibility clause might be used to

mitigate risk in revenue.

To address the first concern, we always include the contract duration and contract amount in our

main regressions. That is, if the above concern is valid, using extensibility clause should be correlated

with shorter duration and lower amount than otherwise. Therefore, including contract duration and

amount should capture the part that extensibility clause is used for the purpose of dividing a whole

project. Put it another away, by including the contract duration and amount in the regression, we

investigate, between two contracts of the same duration and amount, what is the effect of having

extensibility clause on the contract form.

To address the second concern, we include a series of variables that capture the possible

reputation level between the contracting parties that might affect both the use of extensibility clause and

the pricing scheme: vendor is a Fortune 1000 firm, client is a Fortune 1000 firm, vendor is a publicly

20

traded firm, and vendor has industry accreditation, ln(vendor size), and ln(client size). These variables

capture the trustworthiness of the parties in various ways. In the following Section 4.4, we further include

the past reputation capitals between the vendor and the client in the regressions, to capture the reputation

capital between the two parties.

To address the third concern that parties might use extensibility clause to reduce the risk in

revenue, we note that the variable ln(vendor size) already captures the vendor’s need for extensibility

clause to reduce revenue risk.

4.3.2. Potential Endogeneity of the Future Interaction Variable

Finally, there is the concern that the cost-plus contracts are well-defined payment methods that

make it easier for the transacting parties in extending their outsourcing relation, i.e., cost-plus contracts

make including the extensibility clause less costly, thus more likely. Building on practitioner insight, we

address this potential endogeneity problem by conducting instrument variable analyses, i.e., we employ

instruments that affect the use of extensibility clause, and do not directly affect the use of contract form.

Specifically, we estimate simultaneous equations of the below form

Pr ,

and

Pr , ,

where Pr represents probit analysis, is an indicator variable for using extensibility clause,

include the client-specific variables and include the vendor-specific variables as discussed above,

are the instrument variables for the Future variable, and are the control variables for the use of

extensibility clauses that might also affect the contract type.

We use the following five variables for our instrument variables: arbitration clauses, exclusivity

provisions, product restrictions, insurance terms, and re-pricing provisions. Arbitration clauses

whereby parties have an agreed upon framework for resolving disputes, performance restrictions that

clarify the manner in which parties can adjust performance terms ex post (Overby 2011), exclusivity

provisions that restrict vendors from deploying services developed for a particular client for the client’s

competitors, insurance terms that protect vendors against major schedule slippages that are outside the

vendor’s control, and re-pricing provisions that provide an index for re-pricing of services all make it

easier to extend a contract, without affecting the pricing scheme directly.

21

For variables that matter for future interaction, ,i.e., variables that affect the use of

extensibility clauses, we include as controls termination for convenience clauses that facilitate smooth

exit and therefore pose lower risk to the buyer in offering an extensibility clause. We also note that

offering extensibility clauses is less costly or more valuable if the current relation has high value.

Therefore actions that are crucial to the success of desired targets from outsourcing raise the value of

keeping the relation. That is, we expect that the actions taken by both parties to increase the likelihood of

such outcomes are likely to be correlated with the use of future interaction clauses. We therefore control

for clauses detailing a client’s participation in setting standards for delivery time and quality as evidence

of actions taken by a client in reaching the desired goals from outsourcing.

We conduct a recursive, simultaneous bivariate probit estimation that considers whether the

possibility of future interaction is endogenous to the structure of the formal contract. This specification

allows us to simultaneously examine the likelihood that contracts contain clauses to extend and factors

that impact the pricing structure of formal contracts. The likelihood function of a recursive simultaneous-

equations probit model is similar to that of a bivariate probit model (Greene 2003). We perform a

likelihood ratio test of the correlation coefficient of the residuals in recursive-simultaneous bivariate

probit specification and that of a bivariate probit model that ignores the association between these two

choices. We found that the likelihood ratio is significant, thus we rule out the null hypotheses that the

contract type and the expectation of future interaction are independent. In other words, we do find some

evidence that the presence of a future horizon is not completely exogenous to the structure of the formal

contract. Instrument variables for the prospect of future interactions are thus needed.

Table 8 shows the results from the first-stage regression. We include three sets of control

variables: exit clauses; parties’ future responsibilities; and size and bargaining power. The instrumental

variables included are arbitration clauses, product restrictions, exclusivity provisions, insurance terms,

and re-pricing provisions as described above.

Among the instrumental variables, we find that re-pricing provisions, product restrictions, and

insurance terms appear to have the greatest impact on the use of extensibility clauses. Among the control

variables, some size and bargaining power variables matter for the use of extensibility clauses, and exit

clauses are negatively associated with the use of extensibility clauses.

Table 9 reports the estimated coefficients from the 1st and the 2nd stage regressions. The even

columns report the results from stage 1 regressions and the odd columns report the results form stage 2

22

regression. In column 1 where the set of basic control variables are included, we find that the presence of

extensibility clause is significantly associated with a cost-plus contract.

Column 3 introduces indicator variables for joint management and channels of communication in

the second stage regression. We find that coefficients on our variables of interest change little, joint

management is positively associated with the fixed price contracts, and having communication channels

stipulated is positively associated with cost-plus contracts. Column 5 further introduces indicator

variables for performance milestones, audit rights and inspections, and service level agreements. We find

that the coefficients on the future interaction variable are little changed, and audit rights appear to be

positively associated with cost-plus contracts.

Comparing the static probit analysis with the bivariate probit analysis with the future interaction

variable reveals interesting differences between the roles of complexity and the need for adaptation with

the promise of future interaction. We find that the magnitude of the coefficient on the complexity variable

is -0.174 in column 1 of Table 7 and it becomes -0.116 in column 1 of Table 9. The reduced effect of

complexity on the use of fixed price contract once relational contracting is in play suggests that when

implicit contracts are fostered through the promise of relationship continuity, it is not necessary for clients

and vendors to incur costs in delineating contingencies in contracts, i.e., the role of complexity in

determining the type of contract is reduced.

4.4 Bilateral Reputation and the Form of Contract

We further examine the impact of past reputation capital on the form of contracts. Results are in columns

1 and 2 of Table 10. All specifications include the complexity principal component, rent seeking principal

component, log(contract value), the expectations of future interaction (instrumented by instrument

variables.) In column 1, we include the overall reputation capital measured by prior relationship, and also

include the indicator variable for contemporaneous relationship. The coefficient on the prior relationship

variable is negative, indicating that the impact of prior relationships makes a cost-plus contract more

likely, consistent with prior findings, such as Banerjee and Duflo (2000), Corts and Singh (2004) and

Kalnins and Mayer (2004). The impact of prior interaction on formal contracts is likely due to the

vendor’s acquiring experience of a particular buyer as well as enhanced information exchange between

parties through repeated interaction. Williamson (1979) posits that a “specialized language develops as

experience accumulates,” enabling parties to communicate with each other better and realize gains from

exchange.

23

We distinguish between specific measures of inter-firm reputation capital accrued through prior

interactions and the more general measure of prior history of interactions. . We therefore further include

the reputation in fair bargaining, fair cost performance, and prior cancellation. Since the indicator variable

of prior relationship is included in the regression, the coefficient on fair cost (bargaining) performance

captures the effect of the reputation capital on cost (bargaining) performance on the form of contracts

among transacting relationships that have transacted before. The results are in column 2. The coefficient

on the fair bargaining reputation is positive, suggesting that for vendors that have high reputation capital

in fair bargaining and thus less evidence of rent seeking, fixed price contracts are more likely to be used

when parties transact again. The estimated coefficient suggests that for two projects that are of equal

complexity and scope, experience of successful amendments in a past transaction makes it 46% more

likely to use a fixed price contract than a transaction where both parties do not have a prior history of

successful amendments. The coefficient on the vendor’s reputation in keeping costs low is negative,

suggesting that when there is a history of good performance by the vendor in terms of realized costs,

parties are 50 percent more likely to use cost-plus contracts compared to transactions where parties do not

have such a relationship history.

We find that the coefficient of prior history changes when including other bilateral reputation

measures. Prior relationships predict fixed price contracts when we include the impact of fair bargaining

and fair cost performance in an earlier contract. The measures of fair cost performance and fair bargaining

point to a specific type of bilateral reputation as distinct from the generalized experience that results from

a history of prior interaction. Thus, it is possible that the reputation variable used in previous literature

contains both with the impact of general reputation (extent of prior interaction between parties) and

reputation in fair cost performance.

4.5 Market Reputation

One of the empirical issues is to distinguish between generalized reputation in the market and that of

specific reputation in a bilateral context. Parties could place different value on the continuation of

relationships (future interaction), bilateral reputation from a history of interaction and their market

reputation. We therefore compared a buyer’s contracts in the 5 years preceding and in the 5 years

succeeding the contract date. Since a vendor’s business depends on outsourcing, it is likely that a vendor

would undertake outsourcing transactions throughout. We therefore looked at the seller’s experience in

24

the buyer’s industry in the 5 years preceding and succeeding the contract date. We created 4 indicator

variables.

Column 3 of Table 10 reports results including the indicator variables for buyer’s future

interactions with other sellers and buyer’s past history with other sellers. Column 4 further includes

indicator variables for seller’s future interactions in buyer industry and sellers past interactions in buyer

industry. We find that both vendor’s market reputation and a client’s market history are significantly

associated with fixed price contracts. This could be for two reasons. One, deviations are difficult to assess

outside the contractual relationship, so a vendor’s market reputation could provide an indicator of

capability to fulfill the terms of the contracted task, rather than a perception of trustworthiness. Second, a

seller’s generalized reputation observed in the market does not imply parties can immediately work out

the details of rich information exchange that characterize cost-plus contracts. Chassang (2010)

distinguishes between the ability to maintain cooperation and the challenges involved in building

cooperation. Even when cooperation is feasible, the details of achieving success in an IT outsourcing

initiative could involve ironing out a number of business and operational issues, as well as learning to

transact with each other. The process by which parties learn what is the right method to implement an

informal understanding still needs to be understood by the parties. The coefficient of future interaction in

the contract type regression is lower once we introduce market experience of parties. These results are

consistent with an interpretation that while relational interaction is important, a breadth of interaction with

many sellers is good for buyers, which in fact would lower their incentive to form self-enforcing

relationships with a particular seller and the impact of relational contracts on the contract type is reduced.

4.6 Relationship to the Literature

This paper lies in the intersection of the literature on procurement contracts (Bajari and Tadelis, 2001)

and the literature on relational contracts (Macaulay, 1963; Bull, 1987; Baker, Gibbons, Murphy 2002).

Limited work exists, however, on the interaction between relational contracts and procurement contracts

in outsourcing. Papers by Banerjee and Duflo (2000), Corts and Singh (2004) and Kalnins and Mayer

(2004) examine the impact of past transactions on formal contracts. We differ from these studies by

focusing on relational contracts as well as reputation within the relationship. We also examine a variety

of measures of bilateral reputation (within a relation) and their impact on the form of procurement

contracts. And while Gil and Marion (forthcoming) examine the impact of future interaction on the

contract performance, we focus on its impact on contract forms.

25

5. Conclusions

We examine the interaction between relational contracts and the form of explicit contracts,

namely, fixed price contracts or cost-plus contracts. We show that the form of explicit contract affects the

parties’ reneging temptation on a given relational contract, and hence affects the best relational contract

the parties can sustain. When implicit contracts are fostered through the promise of relationship

continuity, it is not necessary for clients and vendors to incur costs in delineating contingencies in

contracts. Rather, contracts could be initiated with an incomplete specification ex ante, where parties

would leave open the possibility of renegotiating future trading opportunities as contingencies unfold. The

difficulty of writing exhaustive contracts facilitating formal enforcement may then be mitigated by

parties’ ability to engage in implicit contracts.

Our contribution to the literature is threefold. First, we collect original data in an IT outsourcing

setting and distinguish between complexity and rent seeking. Second, we enhance the literature on the

impact of past relationships on the current contract by measuring different dimensions of reputation

capital. Third, we measure relational contracting using the prospect of future business, and show its

impact on the form of contract from the perspective of self-enforcement.

One limitation of our analysis is that we do not observe the dynamics or how the trajectory of a

collaborative relationship between parties is developed when there is an opportunity for future interaction.

The dynamics or the path of how a collaborative relation is formed is an under-explored area. Mayer

(2007) offers a case study of an evolving relationship involving “give and take” between parties, and how

that impacts the design of contract clauses. Chassang (2010) examines how two parties learn to cooperate

when details of cooperation are not common language. Examining the dynamics of bilateral interaction

and reputation building is a fruitful avenue for future research.

Appendix on Sample Construction

The data used in this study is drawn from public filings of companies from the U.S. Securities

and Exchange Commission (SEC). We first identified vendors of outsourced services by examining all

registrants that were classified in the SIC categories 73, which denotes that the registrant provides

computer related services. Details of large clients were identified based on datasets of press releases of

outsourcing announcements compiled by a professional advisory firm and a trade journal that lists

publicly announced outsourcing deals. A total of 1724 vendors and 1024 clients were identified in this

manner. Since contracts classified as material consist of a range of contracts such as asset purchase

26

agreements, license transfers, executive compensation, contracts for non-IT related services etc, we

screened the sample to only include agreements that represent IT outsourcing. We then screened the

sample to include only those contracts where the identities of the vendor and the client were clearly

specified. The data was further screened to remove contracts for outsourcing that also involved inter-firm

agreements involving alliances, joint ventures, mergers or acquisition related agreements that pose

substantially different challenges in contracting such as intellectual property issues and contracting of

research-intensive activities that involve different types of contracting hazards other than the ones we

consider here11. From an overall sample of roughly 3800 contracts deemed ‘material contracts’, the

screening process resulted in 466 contracts. Filings with a substantial amount of the contract details were

missing were removed from the sample12. The end sample contained 149 outsourcing relationships

between a total of 239 clients and vendors with clearly defined service obligations13. We supplemented

this data with data from a number of publicly available databases that aggregate news and press releases

such as Dow Jones Interactive, Factiva database and industry reports and trade and business press that

reports on outsourcing deals as well as industry databases such as the One Source Online Business

Information database and the Hoovers database. Whenever possible, we verified this data by examining

press releases from either clients or vendors as well as press releases posted on the archived websites of

vendors and clients obtained from the Internet archives (www.archive.org). Table 1 provides the details.

Reference

Azoulay, Pierre (2004), "Capturing Knowledge within and across Firm Boundaries: Evidence from

Clinical Development," American Economic Review, 94 (5), 1591-1612.

Bajari, Patrick and Steven Tadelis (2001), "Incentives versus transaction costs: a theory of

procurement contracts," RAND Journal of Economics, 32 (3), 387-407.

Baker, George, Robert Gibbons, and Kevin J. Murphy (2002), "Relational Contracts and the Theory of

Firm," Quarterly Journal of Economics, February, 39-84.

Banerjee, Abhijit and Duflo (2000), "Reputation Effects and the Limits of Contracting: A Study of the

Indian Software Industry," Quarterly Journal of Economics, 115 (3), 989-1017.

11 Azoulay (2004) and Lerner and Malmiendier (2010) examine contracts in such settings. 12 The SEC allows separate filing of portions that are deemed confidential, and such filings are not part of a public disclosure by firms. 13 Contracts with no specified service duration could be little more than asset purchase agreements where there is a transfer of technology developed by one party, rather than outsourcing agreements.

27

Bresnahan, T.F., E. Brynjolfsson, and L. M. Hitt (2002), "Information Technology, Workplace

Organization, and the Demand for Skilled Labor: Firm-Level Evidence," Quarterly Journal of Economics,

117, 339-376.

Bull, Clive (1987), "The Existence of Self-Enforcing Implicit Contracts," Quarterly Journal of

Economics,102, 147-159.

Chassang, Sylvain (2010), "Building Routines: Learning, Cooperation, and the Dynamics of

Incomplete Relational Contracts," American Economic Review, 100 (1), 448-465.

Corts, Kenneth S. (forthcoming), "The Interaction of Implicit and Explicit Contracts in Construction

and Procurement Contracting," forthcoming, Journal of Law, Economics, and Organization.

Corts, Kenneth S. and Jasjit Singh (2004), "The Effect of Repeated Interaction on Contract Choice:

Evidence from Offshore Drilling," Journal of Law, Economics, and Organization, 20 (1), 230-260.

Crocker, Keith and Kenneth J. Reynolds (1993), "The Efficiency of Incomplete Contracts: an

Empirical Analysis of Air Force Engine Procurement," Rand Journal of Economics, 24 (1), 126-146.

Gil, Ricard and Justin Marion (forthcoming), " Self-Enforcing Agreements and Relational Contracting:

Evidence from California Highway Procurement," Journal of Law, Economics, and Organization.

Goldberg, Victor P. (1977), "Competitive Bidding and the Production of Precontract Information," The

Bell Journal of Economics, 8 (1), 250-261.

Greene, W. H. (2003), Econometric Analysis. Fourth Ed., Prentice Hall: NJ.

Grossman, Sanford, and Oliver Hart (1986), "The Costs and Benefits of Ownership: A Theory of

Vertical and Lateral Ownership," Journal of Political Economy, XCIV, 691-719.

Gurbaxani, Vijay (2007), "Information Systems Outsourcing Contracts: Theory & Evidence, in

Managing in the Information Economy: Current Research," U. Karmarkar and U. Apte (Eds.), New York,

NY, Spring.

Kalnins, Arturs and Kyle Mayer (2004), "Relationships and Hybrid Contracts: An Analysis of Contract

Choice in Information Technology," Journal of Law, Economics, and Organization, 20 (1), 207-229.

Klein, Benjamin, and Kevin M. Murphy (1997), "Vertical Integration as a Self-Enforcing Contractual

Arrangement," American Economic Review, LXXXVII, 415--420.

Lee, J. N., Miranda, S. M., Kim, Y. G. (2004), "IT outsourcing strategies: universalistic, contingency,

and configurational explanations of success," Information Systems Research, 15 (2), 110-31.

28

Lerner, Josh, and Ulrike Malmendier (2010), "Contractibility and the Design of Research Agreements."

American Economic Review, 100 (1), 214-46.

Levin, Jonathan and Steven Tadelis (2010), “Contracting for Government Services: Theory and

Evidence from U.S. Cities," Journal of Industrial Economics, 58(3), 507-541.

Lichtenstein, Y. (2004), "Puzzles in software development contracting," Communications of the ACM,

47 (2), 61-65.

Linder, J.C. (2004), "Transformational Outsourcing," Sloan Management Review, 45 (2), 52-58.

Macaulay, Stewart (1963), "Non Contractual Relations in Business: A Preliminary Study," American

Sociological Review, XXVIII, 55-67.

Mayer, Kyle J. (2007), "How Contracts and Relationships Evolve Over Time: A Case Study of

Software Contracting," Working paper, University of Southern California

Mayer, K. and Argyres, N.S. (2004), “Learning to Contract: Evidence from the Personal Computer

Industry,” Organization Science, 15, 394-410.

Mayer, K.J. & Nickerson, J.A. (2005), “Antecedents and Performance Consequences of Contracting for

Knowledge Workers: Evidence from Information Technology Services,” Organization Science, 16, 225-

242.

Overby, S. (2011), “Outsourcing: How to Avoid Contract Disputes,” CIO Magazine, retrieved online at

http://www.cio.com/article/685011/Outsourcing_How_to_Avoid_Contract_Disputes

Ryall, Michael D. and Rachelle C. Sampson (2009), "Formal Contracts in the Presence of Relational

Enforcement Mechanisms: Evidence from Technology Development Projects," Management Science, 55

(6), 906--925.

Whang, S. (1992), “Contracting for Software Development,” Management Science, 28, 307-324.

Williamson (1975), Markets and Hierarchies: Analysis and Antitrust Implications (New York, NY:

Free Press).

Williamson, Oliver (1979), “Transaction Cost Economics: The Governance of Contractual

Relations,” Journal of Law and Economics, 22, 233–261.

29

Table 1. Sample Construction

Sample Construction from SEC Filings Observations Total number of registrants 1724 Total number of clients 1024 Total number of material contracts filed with the SEC from the above list of registrants (clients and vendors)

3800

Sample after removing all non-IT outsourcing contracts such as asset purchase agreements, compensation, non-IT outsourcing, wage agreements, etc.

466

Sample after removing other types of arrangements that do not constitute outsourcing 173 Removing contracts without detailed information about vendors and clients 169 Less contracts without financials related data and company information (Data obtained from Hoovers and One Source Business Databases)

161

Less contracts without a detailed history of interaction (Contracts cross-validated against data from Public Databases that aggregate news and press releases such as Factiva, ABI Informs Trade and Industry)

149

Table 2. Breadth of Services Performed in the Contract

Type of Service Percentage of Contracts Systems Planning 28.5 Application Analysis and Design 21.9 Application Development 28.8 Systems Integration 13.2 Operations and Maintenance 37.2 Data Center Operations 19.3 Telecommunications Management 11.2 Software and Data Licensing 47.5 Hardware Products 24.2 IT Facilities Management 18.8 Basic Support 60.5 Training and Documentation 32.3 Advanced Support 52.7 E-marketing and e-Advertising 20.6

Table 3. Component Loadings from Principal Component Analysis

Variable Component 1 (Complexity) Component 2 (Vendor’s Rent Seeking Potential) Systems Development 0.37 0.14 Nascent Technology 0.25 0.04 Contractual Detail 0.31 0.15 Service Breadth 0.39 0.20 Process Transformation 0.44 -0.07 Proprietary platform 0.13 0.44 Proprietary technology 0.03 0.46 Standards licensing -0.43 0.28 Technological Maturity -0.28 0.32

30

Table 4. Variable Definitions and Excerpts from Contract Clauses, Press Releases and Financial Statements Variable Data Source Measure Development Examples

Pricing Structure

Fixed Price Contract

Contract document Fixed Price is coded as 1. Buyer shall pay Vendor the agreed upon charge per month as set forth in Schedule C.

Cost-plus is coded as 0. PROVIDING PARTY shall invoice RECEIVING PARTY on a monthly basis for the Corporate Service Fees, plus the Transition Assistance Fees, as calculated in accordance with Section 3.1 and Schedule 1.1(a).

Measures of Future Interaction Potential and Bilateral Reputation

Expectation of future interaction potential

Contract (validated with press releases)

Coded as 1 if the contract indicates that the relationship might continue into the future (and cross checked with press releases and reports indicating parties’ expectation to continue contractual relationship).

...The term of this Agreement will be extended for additional …. periods unless Buyer or Vendor gives notice to the other at least ….months prior to the then-current Termination Date of its intention to allow this Agreement to expire at the end of the Initial Term or then-current Renewal Term.

Prior Relationship Contract Documents and Press Releases

Coded as 1 when parties to a contract have a prior contracting relationship.

Amicable Amendments (proxy for reputation in fair bargaining)

Amicable Amendments in an earlier contract filed with SEC

Coded as 1 when parties to a contract have had amicable amendments in a prior contracting relationship that led to (i) enhancements in service scope and (ii) incorporating service modifications.

The Services and the matters addressed in the (earlier) Agreement including the Transaction Documents and the Supplement and Schedules are superseded and merged into the (current) Agreement including the Transaction Documents and the Supplement and Schedules thereto. Schedule A, Section 1 will be replaced by (Additional service specification and vendor deliverables added)..Acceptance of deliverable at milestone 1 (estimated date)…Acceptance of deliverable at milestone 2 (estimated date)…. Additional services described in the amendments PIN Based Transactions at $*** (Increase of $*** from original agreement). Off Line Debit Transactions at $*** (Increase of $*** from original agreement).

Fair Cost performance by the vendor

10-K statements cross-checked with Press releases

Coded as 1 when the buyer indicates satisfaction (ex post) with the vendor’s performance on cost targets in a prior contracting relationship.

(Vendor) provided (Client) significant cost savings and operational flexibility by consolidating, automating and managing a large portion of its mainframe operating systems and hardware operations…

Prior Contract Termination

Press releases, litigation filings and 10-K

Coded as 1 when a prior contract was terminated prior to the term.

Excerpt: “To enable this function (application development) to be more responsive to the business, (the project) has been transferred back to Client to support high-level design activities”

Complexity and Rent Seeking Potential

Complexity Coded from the Contract Document based on Contract Objectives.

Transformational; Objective is reengineering of processes and job functions

Coded as 1 when the contract objectives are strategic or transformational

Vendor will deliver innovative and emerging ideas and strategies for more effective use of IT and related business transformation, with an objective that these innovative ideas and strategies can more effectively impact the enterprise transformation of the business of the Client

Systems Development Coded as 1 when the contract involves new systems development

Analyze and review systems requirements ………implementing the standardized, strategic architecture ………obtaining an end state that results in a best-in-class solution…..

Nascent Technology Coded as 1 when technological standards are nascent or emergent

Vendor will provide Client with newly improved or enhanced commercially available information technology, that could reasonably be expected to have a positive impact in terms of increased efficiency, increased quality, or reduced costs for the Client

31

Service Breadth Number of different IT tasks to be performed within a contract.

Contractual detail Number of pages of the contract

Contract Value Contract document validated with press releases. Log transformed monetary value of the contract.

Rent seeking potential

Proprietary Platform Coded as 1 when the vendor employs proprietary platforms Proprietary technology Coded as 1 when the vendor employs proprietary technologies Standards Licensing Coded as 1 when the vendor employs standards that are owned by the vendor and need to be licensed to the client for use in the contracted task Technological Maturity Coded as 1 when the technology and

the platforms are relatively stable and known to the client

Coded from the service description, the deliverables provided by the vendor and the responsibilities of the client.

Other Task Characteristics

Audit Rights Coded as 1 when clauses denote Audit rights whereby clients have the right to inspect and validate service delivery by the vendor.

Milestones Coded as 1 when the contract contains clauses relating to performance milestones tied to specific outcomes. Example: Customer will demarcate particular milestones in a statement of work (SOW) as dependent upon completion of tasks and/or performance by the Vendor.

Service Level Agreements

Coded as 1 when the contract contains clauses detailing acceptable service levels by the vendor. Example: Exhibit B establishes Service Levels for certain specified Services and groupings of Services to be provided by Vendor from the applicable Effective Date throughout the remainder of the Term.

Communication and Joint Management

Communication Coded as 1 if clauses specified the frequency of interactions between the buyer and supplier, such as frequent status and review meetings and channels of communication such as designating key personnel to oversee responsibilities.

Joint Management Buyer will specify and designate authorized personnel on or before the date of the implementation for reporting problems and the vendor shall make reasonable efforts to resolve the problem.

Party Characteristics

Firm Size, Vendor Size

Hoovers and One Source, Fortune Magazine

Client’s number of employees (log transformed). Vendor’s number of employees (log transformed). Client Listed in Fortune 1000 list. Vendor Listed in Fortune 1000 list. Client Market Power (measures whether the client accounts for more than 10% of revenue for a vendor in the year the contract was signed). Vendor is a publicly traded company (denotes that vendor has access to capital markets).

Vendor Industry Accreditation

Trade and business press, annual financial statements, press releases from vendor

Measures whether vendor has industry accreditation such as ISO, CMM type of standards or ranked as a capable vendor by trade and industry press (e.g., Gartner’s Magic Quadrant, Information Week rankings, DataMonitor Group’s Black Book of Outsourcing etc.) .

32

Table 5. Descriptive statistics (N=149) Variable Mean Std. Dev. Min Max

Prior Relationship 0.49 0.50 0 1 Fair bargaining 0.21 0.40 0 1

Vendor cost performance 0.20 0.32 0 1 Contemporaneous Relationship 0.197 0.39 0 1

Prior Contract Cancellation 0.095 0.29 0 1 Contract Type (Coded as 1 if fixed price) 0.61 0.49 0 1

Contract Value (in millions of dollars) 49.12 18.50 0.5 180 Contract Duration (in months) 46.64 31.60 3 120

Service Breadth 0.38 0.48 0 1 Communication 0.58 0.49 0 1

Complexity 0.95 1.06 0 3 Expectation of future interaction 0.53 0.47 0 1

Vendor Industry Reputation 0.33 0.45 0 1 Client Size 13444.36 42490.45 120 475000

Vendor Size 12680.29 43587.28 32 332548 Milestones 0.267 0.451 0 1

Technological Maturity 0.29 0.43 0 1 Contract Detail (Number of pages) 32 34.85 4 194

Monitoring provisions in Prior Contract 0.267 0.441 0 1 Industry Accreditation 0.33 0.45 0 1

Table 6. Cross-tabulations of Contract types as a Function of Task Characteristics, Future Interaction Potential, and Bilateral Reputation

Characteristic Likelihood of Cost-plus Contracts Test of Difference If Yes If No t-statistic (p value)Complexity greater than median 0.63 0.19 3.09 (0.00)Contract length greater than 3 years 0.44 0.23 1.57 (0.11)Contract value greater than median 0.57 0.25 3.21 (0.01)Prior Relationship 0.45 0.36 1.41 (0.20) Fair bargaining in prior relationship 0.30 0.58 -2.41 (0.01)Vendor Fair Cost performance 0.92 0.32 5.31 (0.00)Prior Contract Cancellation 0.43 0.36 0.37(0.60) Expectation of Future Interaction 0.51 0.16 4.13 (0.00) Contemp. Relationship 0.78 0.31 4.78 (0.00) Buyer’s Future Interactions with Other sellers 0.43 0.38 0.523(0.68) Buyer’s past history with other sellers 0.34 0.54 2.028 (0.04) Sellers future interactions in buyer industry 0.28 0.46 -2.22(0.02) Sellers past interactions in buyer industry 0.29 0.44 -1.59(0.12)

Notes. The 2nd column shows the likelihood of cost-plus contracts being used if the characteristics in column 1 is true. The 3rd column shows the likelihood of cost-plus contracts being used if the characteristics in column 1 is not true.

33

Table 7. Static Analysis and Contract Choice Estimates Without Correcting Endogeneity (i) Static Case (ii) Including Future Interaction Fixed Price Fixed Price Future Interaction -0.419(0.106)****

Task Characteristics

Complexity -0.174(0.085)** -0.243(0.090)***

Vendors rent seeking 0.063 (0.02) ** -0.035(0.01)**

Ln(Contract Value) -0.166(0.140)* -0.279(0.141)**

Ln(Contract Length) -0.462(0.166)*** -0.496(0.169)***

Communication and Joint Management

Communication 0.115(0.24) -0.056(0.255)

Joint Management 0.196(0.140)* 0.200 (0.173)*

Monitoring Terms

Audit Rights -0.852 (0.307)*** -1.00 (0.307)***

Milestones -0.049(0.137) -0.00 (0.192)

Service Level Agreements 0.086 (0.120) 0.067 (0.132)

Vendor Size and Client Size Controls, Market power, Vendor Accreditation

X X

Constant 1.10 (0.70)* -0.160 (0.317) Notes. N=149. Coefficients shown are based on two-tailed t-tests (*** 1% level of significance, ** 5% level and * 10%); Standard Errors are displayed in parentheses.

Table 8. Probability of Future Interaction Clauses Future Interaction Instrument variables: Dispute Resolution Terms Arbitration Clauses -0.006 (0.004)*

Product Restrictions 0.455 (0.173) ***

Exclusivity Provisions 0.240 (0.165)*

Insurance Terms 0.207 (0.160)*

Re-pricing provisions 0.301 (0.160) **

Control variables:

Exit Clauses -0.366 (0.163) **

Parties’ future responsibilities Delivery Time 0.069 (0.33)

Quality Norms 0.085 (0.290)

Size and Bargaining Power Client Market Power -0.068 (0.292)

Vendor is a Fortune 1000 firm -0.505 (0.189) ***

Client is a Fortune 1000 firm 0.128 (0.328) Vendor is a Publicly Traded Firm -0.207 (0.292)

Vendor Industry Accreditation 0.215 (0.147)*

Ln(Client Size) -0.011 (0.052)

Ln(Vendor Size) 0.097 (0.056)*

Constant -0.037 (0.439) Coefficients shown in the above tables are based on two-tailed t-tests (*** 1% level of significance, ** 5% level and *

10%); Standard Errors displayed in parentheses.

34

Table 9. Expectation of Future Interaction and the Form of Contracts

(i) Bivariate Probit (ii) Bivariate Probit (iii) Bivariate Probit

Fixed Price Expectation of

future interaction Fixed Price Expectation of

future interaction Fixed Price Expectation of

future interaction Constant 1.844 (0.592) *** 0.114 (0.354) 2.388 (0.852)*** 0.145 (0.372) 2.607(0.771) *** -0.014 (0.43)

Expectations of Future interaction -1.704 (0.618) *** -1.713 (0.401)*** -1.714 (0.188)***

Task Characteristics

Complexity principal component -0.116 (0.052)** -0.138 (0.055)*** -0.211 (0.076)***

Vendor’s Rent Seeking Potential Principal Component

-0.102 (0.051) ** -0.051 (0.030)** 0.102 (0.007)*

Log(Contract Value) -0.071 (0.051)* -0.251 (0.145)* -0.072 (0.104)

Ln(Contract Length) -0.276(0.119)** -0.293(0.115)*** -0.298(0.123)***

Learning and Communication

Communication 0.149 (0.051)* 0.114 (0.127)

Joint Management 0.213 (0.152)* 0.250 (0.174)*

Monitoring Terms

Audit Rights -0.636 (0.221)**

Milestones -0.159 (0.117)* Service Level Agreements 0.187 (0.165)* Instruments for future interaction

X X X

Controls for parties future responsibilities and exit clauses

X X X X X X

Controls for size and Bargaining Power

X X X X X X

Notes. N=149. Coefficients shown are based on two-tailed t-tests (*** 1% level of significance, ** 5% level and * 10%); Standard Errors are displayed in parentheses. The size and bargaining power variables are Vendor Size, Client Size, Market power, Vendor Accreditation, Client Fortune 1000, Vendor Fortune 1000, Vendor is publicly traded

35

Table 10. Relationship History, Market Experience and the Form of Contracts

(i) Bivariate Probit (ii) Bivariate Probit (iii) Bivariate Probit

Fixed Price Expectation

of Future Fixed Price Expectation

of future Fixed Price Expectation

of future Fixed Price Expectation

of future

Constant 1.801 (0.723)** 0.076

(0.370) 1.49 (0.921)* -0.069

(0.424) 2.837 (0.779)*** -0.104

(0.427) 2.62 (0.810)*** -0.112

(0 .391) Expectations of Future interaction -1.72 (0.185)*** -1.408 (0.458)** -1.392 (0.502)*** -1.649(0.205)***

Task Characteristics

Complexity principal component -0.220(0.068)** -0.282(0.092)*** -0.284 (0.092)*** -0.234(0.074)***

Rent seeking principal component -0.087 (0.09) -0.074 (0.056)* -0.064 (0.450) -0.097(0.089) *

Log(Contract Value) -0.083 (0.063)** -0.102 (0.084)* -0.102 (0.140) -0.100 (0.08)*

Ln(Contract Length) -0.324(0.127)*** -0.352 (0.161)** -0.368(0.183)*** -0.274(0.112)**

Relationship History

Contemp. Relationship 0.165(0.249)* 0.345 (0.318)* 0.364 (0.292)* 0.244(0.203)*

Prior Relationship 0.370 (0.194)** 0.367 (0.225)* 0.347 (0.183)* 0.381(0.205)*

Fair Bargaining 0.438 (0.204)* 0.443 (0.174)** 0.410 (0.282)*

Fair Cost Performance -0.862 (0.383)** -0.865 (0.502)* -0.790(0.415)*

Prior Cancellation -0.399 (0.365) * -0.387(0.327)* -0.372 (0.305)*

Market History of Buyer

Similar Contracts of Buyer in Past 0.007 (0.005)* 0.035(0.008)**

Similar Contracts of Buyer in Future 0.225 (0.087)** 0.068 (0.026)*

Vendor’s Market History

Mergers and Acquisitions -0.024 (0.022)

Past Contracts in Buyer Industry 0.119 (0.092)*

Future Contracts in Buyer Industry 0.396 (0.146)**

Controls for communication and learning X X X X

Controls for monitoring terms X X X X

Instruments for future interaction X X X X

Parties’ future responsibilities & exit clauses X X X X X X X X

Controls for size and Bargaining Power X X X X X X X X

The regresson method is Bivariate Probit. N=149. Coefficients shown are based on two-tailed t-tests (*** 1% level of significance, ** 5% level and * 10%); standard errors displayed in parentheses. The size and bargaining power variables are: Vendor Size, Client Size, Client Fortune 1000, Vendor Fortune 1000, Client Market power, Publicly traded Vendor, Vendor Accreditation.