Resource co-specialization and supplier concentration in ... · This paper contributes to our understanding of plural sourcing by moving beyond the recognition of the phenomenon to

Paper to be presented at the

35th DRUID Celebration Conference 2013, Barcelona, Spain, June 17-19

Resource co-specialization and supplier concentration in concurrent

sourcingMari Sako

University of OxfordSaid Business School

[email protected]

George ChondrakisUniversity of Oxford

Said Business [email protected]

Paul M. Vaaler

University of MinnesotaCarlson School of Management

[email protected]

AbstractAlthough the preconditions for the simultaneous use of multiple sourcing modes have been explored, the question ofwhat determines how much to make and how much to buy in plural sourcing has been understudied. This study fills thisgap. Using longitudinal data of in-house lawyers at, and law firms used by, Fortune 500 firms, we find empirical evidencethat plural sourcing strategies are tilted in favor of ?make? whenever firms have opportunities for resourceco-specialization (notably between legal resources and R&D, marketing, and top management teams) and a broadportfolio of suppliers that raises the cost of contracting. Our study suggests that plural sourcing research benefits fromexamining a group of transactions within a corporate function, to complement the predominant focus on transaction-leveloutcomes.

Jelcodes:M10,-

1

Resource Co-Specialization and Supplier Concentration in Plural Sourcing: Evidence

from Legal Services Sourcing at Fortune 500 Companies

INTRODUCTION

How do firms organize economic transactions? Since the seminal work on firm

boundaries by Coase (1937, 1988), scholars have addressed this question by identifying

multiple modes (i.e. make, buy, ally, etc.). Prior work in transaction cost economics (TCE)

analyzed the transactional characteristics that lead to choosing one of these modes (Poppo et

al., 1998; Walker et al., 1984; Williamson, 1985) while the resource-based view of the firm

(RBV) suggests that firm boundaries reflect differences in capabilities and knowledge

(Argyres, 1996; Jacobides et al., 2005; Kogut et al., 1996).

Both TCE and RBV employ a comparative analysis of sourcing modes, leading to the

choice of the most efficient institutional arrangement (i.e. market, hierarchy or hybrid) for

each transaction. However, prior research has documented the simultaneous use of multiple

sourcing modes by firms, calling this concurrent sourcing (Parmigiani, 2007; Parmigiani et

al., 2009), plural sourcing (Bradach, 1997; Gulati et al., 2006; Heide, 2003; Jacobides et al.,

2006), or taper integration (Harrigan, 1986; Rothaermel et al., 2006). In contrast to a TCE or

RBV logic, firms are often found to both make and buy the same input.

There is robust theory and evidence on the conditions under which plural sourcing is

likely to occur. For example, Parmigiani (2007) demonstrates that greater overlap between

the firm’s and the supplier’s expertise increases the likelihood of plural sourcing while

Parmigiani et al. (2009) explain that firms may make and buy sets of complementary

components. Performance uncertainty and information asymmetry between buyers and

suppliers (Dutta et al., 1995; Heide, 2003), technological volatility (Krzeminska et al.,

2

forthcoming), and complementarities in incentives or knowledge (Puranam et al.,

forthcoming) can also lead to plural sourcing.

While important in identifying the prevalence of plural sourcing, this research has

provided limited guidance on a logical follow-on question, that is, how to explain the mix of

internal and external sourcing (Parmigiani, 2007:306; Puranam et al., forthcoming).

Assuming that a firm does indeed both make and buy, why would it choose 80% make vs.

20% buy as opposed to 40% make vs. 60% buy? Thus, given the presence of preconditions

for plural sourcing, how do firms decide how much to make and how much to buy?

In addition, the literature does not explore the selection of suppliers and their impact

on the design of plural sourcing strategies. For example, if a firm decides to buy 40% for a

given product or service, this 40% could be outsourced to a single firm or multiple firms.

Existing theory assumes the decisions on the make-buy balance and supplier selection to be

independent of each other. This is despite the fact that behind the ‘façade of the market’ lies

another firm, with its own capabilities and contractual relationship with the focal firm

(Jacobides et al., 2005). To the extent that supplier selection influences the cost of

contracting, it would have an impact on the make-and-buy decision.

In this paper we build on recent progress in the area of plural sourcing and address

these two limitations. In doing so we identify two mechanisms that help better understand the

design and variability in plural sourcing strategies. First, resource co-specialization enhances

the benefits of internalization. By resource co-specialization we refer to the synergistic gains

that arise from the interaction of different resources within firm boundaries (Lippman et al.,

2003; Teece, 1986). When firms outsource part of production they essentially position certain

resources, e.g. personnel, equipment etc. outside of their boundaries. By doing so, they

forego opportunities to harvest synergies resulting from resource combinations that increase

their marginal productivity through on-going interaction, shared expertise, and knowledge

3

spillovers (Dierickx et al., 1989; Kogut et al., 1992; Milgrom et al., 1995). Consequently, in

cases of plural sourcing the extent of integration, or make, will increase when there are

opportunities for resource co-specialization with other firm resources.

Second, we highlight the impact of supplier concentration (Moeen et al., 2013) on the

relative importance of internal versus external suppliers in plural sourcing. Here we suggest

that the number of suppliers chosen to outsource to is not independent of the proportion that

is bought. A large number of suppliers leads, ceteris paribus, to high mundane transaction

costs and reduces the benefits from relational contracting (Baker et al., 2002; Langlois,

2006), resulting in the firm relying more heavily on make in plural sourcing. By contrast,

when firms use a small number of suppliers, they will benefit from relational contracting and

lower transaction costs and consequently buy more from their suppliers (Dyer, 1997; Moeen

et al., 2013).

We develop our arguments in the context of corporate legal services, a setting where

the use of both internal and external suppliers for the same input, known as multi-sourcing, is

common (Sako, 2011; Susskind, 2008). We demonstrate that resource co-specialization

drives the design of plural sourcing strategies as the balance tilts in favor of make when firms

rely more on R&D resources, advertising resources, international presence, and legal

expertise in their top management team. We also find that the extent of insourcing increases

when the firm sources from a large number of law firms. Although this result might sound

counterintuitive, it is explained by the endogenous impact of supplier concentration on the

costs of contracting.

This paper contributes to our understanding of plural sourcing by moving beyond the

recognition of the phenomenon to the identification of specific strategies that firms employ

and their antecedents. Moreover, this study is important in extending the inquiry on firm

boundaries to the domain of corporate functions. The vast majority of an admittedly

4

voluminous literature on vertical integration (or disintegration) has focused on primary

activities (Gibbons, 2005; Williamson, 1985), to the neglect of corporate functions providing

important support activities in the value chain (Porter, 1985). This paper highlights the

importance of investigating firm boundary-setting and the mix of make and buy activities

with a focus on value-adding professional and business services in corporate functions (Sako,

2013).

In the next section we review the relevant literature and explain why corporate legal

services provide an ideal setting to test our predictions. Then, we present in sequence our

hypotheses, data and methodology, and results. The final part offers a concluding discussion.

PLURAL SOURCING: THEORY & EVIDENCE

TCE and RBV are the two dominant theories employed to understand firm boundary

decisions. On the one hand, TCE explains the make-or-buy decision as a response to close

comparison of costs associated with each alternative (Williamson, 1985). Market-based buy

strategies tend to have lower costs and thus dominate comparable make strategies

emphasizing firm ownership of and hierarchical control over suppliers. But cost advantages

of buy over make strategies reverse themselves in exceptional circumstances where hazards

with market-based contracting are substantial. In contrast, RBV contends that the reason an

activity is conducted within the firm is not market failure but rather firm success: the firm as

an institution enjoys an ‘organizational advantage’ in organizing economic activity (Madhok,

2002:536). Thus, RBV explains firm boundary-setting based on competitive interest in

exploiting value from difficult to imitate resource bundles (Barney, 1986).

TCE and RBV differ on many dimensions but share assumptions about the choice of

one ‘optimal’ exclusive governance mode that runs counter to long-term use of multiple

sourcing modes. Nevertheless, the simultaneous use of multiple sourcing channels is

something we observe empirically, and numerous studies testify to its systematic use, at

5

different points in production of goods and services in different industry contexts (Bradach,

1997; Heide, 2003; Jacobides et al., 2006; Parmigiani, 2007; Porter, 1980). Plural sourcing,

along with other terms such as tapered integration or concurrent sourcing, has been used to

describe such phenomena.

The benefits and rationale for undertaking plural sourcing are well understood. Some

of the earlier arguments focused on the role of demand uncertainty, whereby firms can avoid

maintaining idle capacity through the use of external suppliers (Adelman, 1949), and the

threat of backwards integration to external suppliers (Harrigan, 1986). Besides these, firms

also have a better understanding of the production process when making and buying and are

thus better at monitoring suppliers (Dutta et al., 1995; Heide, 2003). Finally, ‘knowledge

complementarities’ between the two sourcing modes have been noted (Puranam et al.,

forthcoming). For example, internal and external suppliers can benefit from mutual learning

and from transferring best practices across different organizational arrangements (Bradach,

1997; Cassiman et al., 2006; Parmigiani, 2007).

Of course, these findings have not gone unchallenged by TCE orthodoxy. Williamson

(1985) suggests that plural sourcing is actually an artifact of ill-identified transactional

heterogeneity. Consistent with this argument, He et al. (2006) find that although trucking

firms appear to engage in plural sourcing (through the use of both internal and external

drivers), they are in fact choosing to outsource hauls that are qualitatively different from the

ones undertaken internally. Similarly, Azoulay (2004) demonstrates that variation in project

characteristics guides pharmaceutical companies when choosing to outsource or assign their

own employees on different projects. More generally, the difficulty in identifying when firms

both make and buy ‘exactly the same input’ presents a substantial challenge to plural

sourcing research (Krzeminska et al., forthcoming).

6

Two further limitations of prior theorizing are notable. First, there is little guidance on

the relative balance of make and buy in cases of plural sourcing (Parmigiani, 2007; Puranam

et al., forthcoming). When the preconditions for plural sourcing exist, we still know little

about how firms choose the mix of internal and external procurement.

Second, existing theories do not analyze supplier selection and its impact on the

design of plural sourcing strategy. Although the role of suppliers’ capability and expertise has

been explored (Parmigiani, 2007; Parmigiani et al., 2009), the overall cost of contracting is

assumed to be independent of concentration in the portfolio of suppliers. However, this

premise is unlikely to hold given that the costs of contracting are affected by the nature of the

contractual relationship between any two parties. In particular, one-off arms-length

transactions between parties are characterized by increased costs of contracting because there

are higher ‘mundane’ transaction costs and lower opportunities to benefit from relational

contracting (Dyer, 1997; Langlois, 2006; Moeen et al., 2013). Hence, the degree of supplier

concentration, and its subsequent impact on the cost of contracting, should influence the

make-buy balance in the plural sourcing of products or services.

In order to address these limitations and contribute to our understanding of plural

sourcing we study corporate legal services. This setting is ideal as it is an unambiguous

example of plural sourcing (see below). We are thus able to dispense with the discussion

about whether inputs are actually both internally and externally procured and focus instead on

observed variability in the design of plural sourcing strategies.

In addition, by moving from the level of the individual transaction to that of a

portfolio of transactions (i.e. legal services) we are able to account for patterns of

interdependence between different transactions and firm resources (Moeen et al., 2013;

Puranam et al., forthcoming). Without such focus, it is impossible to observe ‘spillovers’

from different transactions (Mayer, 2006) and the effect of complementarity between inputs

7

(Parmigiani et al., 2009). Firms regularly employ suppliers for a variety of inputs – with

varying degrees of similarity (Krzeminska et al., forthcoming) – and supplier selection is

naturally affected by existing contractual relationships. For example, a corporation using a

law firm for patent filing could choose the same law firm in case some of its patents are

targeted in a patent suit, given the law firm’s expertise and experience in this area. A

portfolio-level analysis ensures that this information is not lost.

The plural sourcing of legal services

The procurement of legal services by corporations presents an ideal context for the

study of plural sourcing (Sako, 2011; Susskind, 2008). Firms typically have an in-house legal

department headed by a general counsel (GC) leading a staff of in-house lawyers, while at the

same time regularly engaging external law firms. Make-and-buy strategies are driven to some

extent by transactional characteristics: routine legal tasks, such as regulatory form filing, may

be undertaken by in-house lawyers while external lawyers may take the lead on more

specialized tasks such as complex acquisitions, litigation, and criminal matters. However, in-

house and external lawyers often work simultaneously, as a team or in parallel, on the same

type of legal matter. For example, combined teams of in-house and external lawyers support

and consult senior managers during acquisitions. Similarly, in-house legal departments

usually undertake patent filing and prosecution with the help of specialist law firms.

The plural sourcing of legal services is not surprising given complementarity between

the in-house lawyer’s firm-specific and the outside counsel’s practice-specific knowledge and

experience (Puranam et al., forthcoming). Important synergies and learning opportunities

arise from such interactions. There are also bargaining and oversight dimensions to consider

in the plural sourcing of legal services (Dutta et al., 1995; Harrigan, 1986). If firms have the

capacity to re-direct legal matters from outside counsel to qualified in-house lawyers, then

these firms also have more power to bargain over the costs of retaining outside counsel. They

8

can also exercise oversight of outside counsel actions to assure high-quality and cost-

effective representation in some legal matter. Conversely, the use of external law firms helps

create competitive pressure to the in-house legal department, thus ensuring a minimum level

of efficiency in its operations (Jacobides et al., 2006).

HYPOTHESES

Co-specialization of legal resources

A central tenet of RBV is that firms create and capture value by deploying different or

unique combinations of resources. This suggests that the value of a resource is not

exogenously determined but rather depends on other surrounding resources. Lippman et al.

(2003) go as far as to state that no resource is firm-specific, and suggest that firm value can

be traced to the presence of co-specialized resources that exist within the legal shell of the

firm. Resource co-specialization then is associated with excess value, or synergy, that results

from the interaction of different resources within firm boundaries.

A number of explanations have been put forward in the literature to explain the

emergence of gains from resource co-specialization.1 Nelson et al. (1982) describe how

organizational resources are engaged in functioning routines that enable firms to economize

on coordination costs. Kogut et al. (1992) explain that the emergence of high-order principles

through organizational membership enhances knowledge creation and sharing.

The generation of synergies resulting from resource co-specialization will affect the

design of plural sourcing strategies, as firms will be less reluctant to outsource production.

Market contracting entails the externalization of resources, which reduces the opportunities

of resources to become co-specialized. Hence, in the case of plural sourcing, there will be

more emphasis on make when resource co-specialization is likely to generate synergies. It is

1 We prefer to use the concept of resource co-specialization as opposed to resource complementarity (Milgrom

et al. 1995) because resource co-specialization emphasizes the dynamic nature of the process of generating

synergy. This synergy is not a characteristic of the resource but develops through ongoing interactions within

firm boundaries. Complementarity results in a more instant generation of surplus value due to the attributes of assets or resources that may lie within or across the firm boundary.

9

important to clarify here that our notion of resource co-specialization is distinct from

complementarity between procurement modes as described by Puranam et al. (forthcoming).

Whereas they focus on the excess value resulting from the simultaneous use of both market

contracting and hierarchical governance, we emphasize the excess value resulting from the

interaction of internalized resources with other firm resources.

Legal resources can be combined with other firm resources, giving rise to

opportunities for exploiting resource co-specialization. Thus, instead of regarding lawyering

as necessary fixed overhead costs for the corporation, in-house legal resources may be seen

as a source of competitive advantage (Bagley, 2008; Orozco, 2010). But exactly what firm

resources give rise to co-specialization opportunities, which in turn increase reliance on the

internal legal department?

International firms

Internationalization as a process compounds the complexity of all managerial tasks

(Carpenter, 2002; Prahalad, 1990; Sanders et al., 1998). As multinationals enter foreign

markets which are more distant in geographic, cultural, and administrative dimensions

(Ghemawat, 2001), they suffer from greater ‘liability of foreignness’ (Zaheer, 1995). In

relation to legal tasks, complexity multiplies with the number of foreign jurisdictions in

which the company has presence, as it has to deal with diverse institutional environments.

This additional complexity creates opportunities for generating value through

regulatory arbitrage. This term is used to describe cases when parties take advantage of a gap

between the economics of a deal and its regulatory treatment, restructuring the deal to reduce

or avoid regulatory costs (Fleischer, 2010:227). Examples include transfer pricing, effective

cross-jurisdictional tax planning, investment decisions based on tax, or other regulatory

incentives. Yet, a strong legal expertise within the firm is required in order to identify such

opportunities and fully capture their benefits (Bagley, 2008). As Marchant et al. (1999)

10

explain, legal expertise largely relies on tacit knowledge of the context and past experience

with the specifics of the situation. Hence, frequent interaction between the corporate legal

department and, say, the corporate accounting department (for international tax planning) or

the strategic planning department (for foreign direct investment or M&A) is likely to create

important benefits for international firms through the identification and realization of

unexplored opportunities. The potential to exploit opportunities from the co-specialization of

legal with other resources, arising from multi-jurisdictional international presence, is likely to

favor more ‘make’ in the plural sourcing of legal services. Thus:

H1: The more internationalized a firm is, the greater its reliance on the internal

sourcing of legal services.

R&D intensive firms

Besides international firms, we also expect R&D intensive firms to rely more on their

internal legal departments. Firms investing in new technology must choose an appropriate

strategy to appropriate gains by protecting their intellectual property, for example by

choosing between secrecy and patenting. Exactly what to patent and how to patent is a know-

how that arises from the co-specialization of legal and technical knowledge. Internal patent

lawyers tend to be highly knowledgeable about the company’s unique technologies (Somaya

et al., 2007). They can interact with the firm’s R&D department to discuss patentable ideas

from an early stage and thus help to increase the share of the value appropriated from

innovation. For example, Reitzig et al. (2009) have found that intermediate levels of cross-

functional involvement between the legal and R&D department increase the speed of patent

grants.

In case studies of innovating firms there is also increased evidence on the role and

importance of internal legal experts in the management of intellectual property – for example

through various committee memberships (Fox, 1998; Grindley et al., 1997). Hence, the value

11

of in-house lawyers will be higher for R&D intensive firms due to the co-specialization of

legal and technical resources. We therefore hypothesize that:

H2: The higher the R&D intensity of a firm is, the greater its reliance on the internal

sourcing of legal services.

Advertising intensive firms

Similarly to R&D intensive firms, firms that rely on advertising to compete will

benefit from the co-specialization of legal resources. Advertising relies on the use of

intangible assets, such as brand names and trademarks, which require protection from

competition. Increased interaction between the legal and marketing departments will

therefore help firms devise a trademark strategy that is informed by and exploits the legal

opportunities and limitations in the use of trademarks (Cohen, 1986, 1991). This will enable

firms to reduce brand dilution and create stronger brand names. Increased communication

between lawyers and marketers is important in this process as lawyers can identify threats to

the intellectual capital of the firm that are not immediately obvious to marketing stuff

(Peterson et al., 1999). In view of these gains, firms with high advertising intensity will rely

more on their internal legal department. Thus:

H3: The higher the advertising intensity of a firm is, the greater its reliance on the

internal sourcing of legal services.

Legal expertise in TMTs

Legal resources may be used directly by top management teams (TMT) to hone their

capability for making strategic choice. Defined as ‘the ability of a TMT to communicate

effectively with the counsel and to work together to solve complex problems’, ‘legal

astuteness’ is a valuable dynamic capability (Bagley, 2008:378). By incorporating legal

know-how in business decisions, ‘legally astute’ TMTs use legal tools to create options (e.g.

12

in litigation), and convert regulatory constraints into opportunities (e.g. by proactively going

beyond the letter of the law in environmental compliance to improve financial performance).

Legal astuteness impacts on the orientation of both TMTs and company lawyers. As

for TMTs, the upper echelons perspective points out that organizational outcomes reflect the

values and cognitive bases of powerful actors in the organization (Hambrick et al., 1984).

Consequently, examining prior experiences and background of TMT members help us

identify the ways they shape and direct the resources at their disposal. In the legal context,

TMTs with legal expertise may create more structures and processes to improve the

effectiveness of the in-house legal function. Thus, the presence of a functional TMT member

influences organization design, and equally, the TMT composition is affected by changes in

organization structure (Menz, 2012).

Legal astuteness also affects the orientation of in-house corporate lawyers. They

would be expected to be proactive in identifying complementarities between business

opportunities and legal know-how. They are, in effect, counsels (combining legal and

business advice) or entrepreneurs (giving priority to business objectives rather than legal

analysis), rather than cops (limiting their advice to legal mandates) (Nelson et al., 2000). In

short, these ‘decision consultants’ (Rosen, 1984) give legal advice in the context of the

business. Thus, legally astute TMTs rely much on the GC’s capacity to create value from the

co-specialization of legal resources and business-context knowledge within the firm. Thus,

the more legally astute the TMT, the more likely the GC is a member of the TMT and the

more likely in-house lawyers rely on co-specialization of legal and firm-specific business

knowledge.

H4: The general counsel’s membership in the top management team will result in

greater reliance on the internal sourcing of legal services.

13

Supplier concentration

Besides resource co-specialization, the degree of concentration of suppliers affects the

design of plural sourcing strategy. As explained before, heterogeneity in supplier selection

has not been explored in the context of plural sourcing. In essence, the analytical calculus that

prompts a firm to choose plural sourcing is assumed to be independent of the composition of

supplier portfolio. Here, however, we suggest that different degrees of supplier concentration

impose different costs of contracting to the outsourcing firm. These costs depend on the

nature of the contractual relationship. Low scale, arms-length transactions are characterized

by high costs of contracting as the parties do not have much incentive to invest in firm-

specific transacting platforms and systems (Baldwin et al., 2003; Langlois, 1992, 2006). In

addition, infrequent interactions with suppliers do not generate inter-organizational trust and

the associated benefits of relational contracting and increased commitment (Baker et al.,

2002; Dyer, 1997; Sako et al., 1998). Hence, in the presence of high costs of contracting, we

expect lower reliance on external providers.

In-house legal departments may choose to procure legal services from a large number

of law firms in pursuit of increased expertise in some area (e.g. regulatory compliance in a

specific industry) or lower cost. This model is closer to arms-length-style market governance,

as firms scan the environment for the best possible provider for their specific legal needs.

Alternatively, legal departments could outsource to a concentrated panel of law firms

involved in relational governance. This mode involves less flexibility in terms of supplier

selection but the firm benefits from reduced costs of contracting. For example, having an on-

going and committed relationship with a law firm facilitates communication (as there is no

need to identify points of contact or firm background), and ensures that the client firm will

get the top partners for their case.

14

We therefore expect a company’s willingness to invest in relational governance with a

selected few law firms to result in a high degree of outsourcing and a smaller in-house legal

department (Moeen et al., 2013). By contrast, companies collaborating with a large number

of law firms will increase their reliance on internal legal resources. Thus, we predict the

following:

H5: The larger the number of external law firms employed by a firm, the greater its

reliance on the internal sourcing of legal services.

EMPIRICAL METHODS

Data and sampling

To evaluate our theoretical framework and test our hypotheses, we first collect

information on the size and composition of internal legal departments and their relationships

with external law firms. This information is not publicly available, so we use proprietary

survey and secondary data collected by ALM Legal Intelligence, a research unit within the

American Lawyer Media Group. In particular, we used ALM’s annual survey of in-house

legal departments as well as ALM’s reports on corporate activity, including litigation, M&A

transactions, and corporate bankruptcies. These data are available from 2004 until 2011 for

Fortune 500 companies2.

To be included in our sample, we require firms to have ALM annual survey data as

well as data on firm operations in Compustat corporate and industry segment files. Our final

dataset is an unbalanced panel consisting of 1230 observations from 285 firms (i) observed

over up to eight years (t). Our reported number of observations drops to 945 as several

statistical analyses below require the inclusion of one-year lagged variables.

2 For 2004 and 2005 the data are only available for Fortune 250 companies.

15

Dependent variable

Consistent with other empirical work on plural sourcing (Parmigiani, 2007;

Parmigiani et al., 2009), we seek comparable information on internal and external suppliers

of legal services. Ideally we would have information on hours worked on specific matters by

in-house lawyers at corporations as well as hours worked on the same matters by lawyers at

outside law firms on behalf of these client corporations. Such data are not publicly available

as lawyers at both types of organizations treat such information as sensitive and strictly

confidential.

We do, however, have ALM survey data on the number of in-house lawyers and the

number of external law firms providing outside legal work for these firms. Thus, we

construct our dependent variable based on the count of in-house lawyers (as reported in the

ALM annual survey of general counsel (GC)) and include the count of external law firms as a

right-hand side term. Our dependent variable is the natural log of the annual count of in-

house lawyers per $100 billion in firm sales. Changes in the dependent variable then reflect

changes in the in-house lawyer count attributed to factors unrelated to organic growth.

An increase (decrease) in the dependent variable can be attributed to two possibilities:

a) increased (decreased) reliance on internally made legal services by in-house lawyers rather

than externally bought legal services by outside law firms; or b) an increase (decrease) in the

overall amount of legal work that needs to be undertaken on behalf of the firm by both

internal and external providers. In order to account for the second possibility and isolate

changes in the make-and-buy mix, we include controls for the overall amount of legal work

undertaken and employ multiple regression analyses that incorporate dynamic processes and

account for possible reverse causality.

16

Independent variables

Internationalization. We measure the degree of internationalization as a count of the

number of countries where each firm operates via subsidiaries in a given year. Although

different measures of internationalization have been suggested, we use the number of

countries as it more accurately captures the complexity associated with operating in multiple

jurisdictions. Even modest sales or presence in a country requires basic knowledge of and

compliance with substantive laws and legal processes.

R&D intensity. We calculate R&D intensity as R&D expenses divided by sales.

Advertising intensity. We calculate advertising intensity as advertising expenses

divided by sales.

General counsel in TMT. In order to examine the GC’s participation in the firm’s

TMT, we construct a dummy variable taking the value of one when the GC has the title

‘Senior Vice President’ or ‘Executive Vice President’ and otherwise taking the value of zero.

Data on the GC’s role were obtained from S&P’s Capital IQ database which includes

biographical and other information on key professionals, executives and directors.

Law firms. In order to measure the degree of concentration in the provision of external

legal services we use a count of external law firms working for a given client corporation

surveyed. This variable is included in ALM’s annual survey and asks GCs to identify the law

firms that undertook legal work on their behalf in a given year. All other equal, the higher the

number of law firms, the lower the degree of concentration in the provision of external legal

services across the group.

Control variables

We include a number of time-varying firm controls including firm debt exposure,

calculated as total debt divided by total assets, firm profitability, calculated as EBITDA

divided by sales, and the natural log of the count of firm employees. We also control for the

17

effect of product diversification using the entropy measure of diversification, where industry

segments are defined by 4-digit SIC codes. In addition, we include a count of new

acquisitions started in a given firm as well as a count of new litigation cases started in a given

year where the firm was involved either as defendant or plaintiff.3 Both acquisitions and

lawsuits are important controls as they may prompt a significant increase in the amount of

legal work to be done in-house or externally. To account for the possibility that the additional

work may come with some delay, we also include one-year lagged value of these two counts.

To better control for the overall amount of legal work undertaken by the firm in each

year, we also use annual selling, general and administrative expenses (SG&A) as a proxy for

legal expenses. This variable is reported by Compustat and includes all commercial expenses

of operation not directly related to production, including GC office expenses and legal fees

paid to law firms. We are thus able to observe changes in the legal service make and buy

propensity as opposed to changes in the overall amount of legal work undertaken. Finally, we

include a dummy variable which is equal to one when the firm is in bankruptcy proceedings

in a given year. Table 1 reports summary statistics and correlations for the variables.

- Insert Table 1 about here -

Econometric specifications

Standard fixed- or random-effects specifications allow us to control for unobserved

firm characteristics that influence the legal services make-and-buy decision. However, our

data have two additional characteristics that need to be accounted for. First, the in-house

lawyer count is likely to be influenced by past observations. Including a lagged value of the

dependent variable as a regressor allows us to account for autocorrelation but the presence of

both a lagged dependent variable and fixed-effects can render estimates inconsistent (Nickell,

3 ALM only collects data for ‘important’ acquisitions and lawsuits as reported in trade publications. Although

this probably results in loosing some information, this is unlikely to bias our results as reporting criteria are

similar for all firms in our sample. In addition, the biggest acquisitions and lawsuits are likely to have the most important effect on legal service sourcing dynamics.

18

1981). Second, several of our regressors may be endogenously determined or predetermined,

that is, correlated with current or past disturbances. To address these concerns we employ a

dynamic panel data estimator with GMM-type instruments (Arellano et al., 1991). This

estimator is ideal for ‘small T, large N’ panels like ours. In addition, GMM estimators are

robust to heteroskedasticity in the cross-section and unknown patterns of serial correlation

(Arellano, 1987).

More specifically, we estimate a statistical model of the form:

!!" ! !!!! !!! ! !!!!" ! !! ! !! ! !!"

where !!" is the natural log of the count of lawyers divided by sales in firm i of year t, !!" is a

vector of explanatory variables, !! is a year effect, !! is a time-invariant firm-specific effect

and !!" is the error term.

The GMM estimator originally proposed by Arellano et al. (1991) uses first

differencing to remove unobservable firm-specific effects and then instruments the

endogenous variables using lagged levels of the series. However, this approach is

problematic in our setting as the first-differenced GMM estimator is found to have large

finite sample bias and poor precision when time series are short and persistent (Blundell et

al., 1998). Instead, we employ the system GMM estimator suggested by Arellano et al.

(1995) and Blundell et al. (1998). This approach uses lagged differences as instruments for

equations in levels, in addition to lagged levels as instruments for equations in first

differences. The system estimator introduces an additional assumption that changes in

instrumenting variables are uncorrelated with the fixed effects. All models were calculated

using the xtabond2 Stata module (Roodman, 2006).

To address concerns about simultaneity bias we treat the number of law firms and

general counsel in TMT variables as endogenous while the remaining explanatory variables

as predetermined. From the control variables, selling, general and administrative expenses are

19

endogenous as they proxy overall legal expenses. Litigation cases are also treated as

endogenous given that larger legal departments are more likely to take cases to court. The

remaining control variables are treated as exogenous. We use the forward orthogonal

deviation transformation instead of first differencing given that our panel is unbalanced. This

minimizes data loss while preserving the orthogonality among the errors (Arellano et al.,

1995; Roodman, 2006). Finally, we use standard errors that are robust to heteroskedasticity

and arbitrary patterns of autocorrelation within firms.

RESULTS

Descriptive statistics and preliminary analysis

Descriptive statistics and pair-wise correlations are reported in Table 1. The sample

mean for In-house lawyers is 5.37, suggesting that firms employ on average 3.62 lawyers per

billion dollar of sales. The raw number of in-house lawyers at Fortune 500 firms sampled

exhibited substantial variation ranging from a handful to more than 1200 in certain firms such

as General Electric. The sample mean of Outside law firms is 1.83, meaning that firms

reported on average nine outside law firms as undertaking legal work on their behalf in a

given year. Those numbers ranged from one to more than 50 outside law firms. Pair-wise

correlations are largely intuitive. Variables corresponding to the five hypotheses for testing

all exhibit the predicted positive sign when correlated with the dependent variable.

Figures 1A-E present results from Lowess analyses4 of In-house lawyers and make

and buy determinants related to Hypotheses 1-3 and 5 (Internationalization, R&D intensity,

Advertising intensity, and Outside law firms), and from comparative bar-chart analyses of In-

house lawyers related to Hypothesis 4 (General counsel on TMT). The pattern of results in

all figures indicates preliminary support for Hypotheses 1-5.

4 Lowess analyses compute linear regressions around each observation of given determinant (e.g.,

Internationalization) with neighborhood observations chosen within some sampling bandwidth and weighted by

a tri-cubic function. Based on the estimated regression parameters, In-house lawyers values are computed. Combinations of determinants and In-house lawyers are then connected, yielding a Lowess curve.

20

- Insert Figures 1A-E about here -

Regression analysis

Table 2 presents the results of our regression analysis. Models 1-2 present fixed-

effects specifications while Models 3-5 present results from the system GMM estimator.

Simpler fixed-effects models suggest that the lagged dependent variable, In-house lawyersit-1,

explains roughly 30% of variation in the dependent variable. We noted previously, however,

that the lagged dependent variable coefficient estimates in fixed-effects regressions are

typically biased downwards (Nickell, 1981). When the number of time-periods, t, are few as

here, the downward bias can be substantial as corrected system GMM regression results

demonstrate. Coefficients on In-house lawyersit-1 jump from 0.310 and 0.305 in Models 1-2

to 0.973, 0.882 and 0.883 in Models 3-5. All estimates are significant at the 1% level.

Model 3 reports one-step system GMM results with the lagged dependent variable

and controls only. Models 4-5 are fully-specified and report one- and two-step system GMM

results (the two-step estimator is corrected for downward bias in the computed standard

errors (Windmeijer, 2005)). Diagnostics at the bottom of each column suggest that key

estimation assumptions hold. The Arellano-Bond z test for second- and higher-order auto-

correlation is not statistically significant. This is not weakened by the number of instruments

as instrument count is lower than the number of firms (Roodman, 2006). Hansen’s J test fails

to reject the null hypothesis that the instruments generated are exogenous.

- Insert Table 2 about here -

We look primarily to results in Models 4-5 for evidence to evaluate support for our

hypotheses. Results support our predictions and the importance of resource co-specialization

in designing plural sourcing strategies. Consistent with Hypothesis 1, we observe positive

coefficients for Internationalization significant at the 1% level. Firms operating in more

countries increase their dependence on in-house lawyers. Consistent with Hypotheses 2 and

21

3, we see positive coefficients on R&D intensity significant at the 10% level and positive

coefficients on Advertising intensity significant at the 5% level. Consistent with Hypothesis

4, we observe positive coefficients on General counsel on TMT significant at the 10% level,

suggesting that elevating senior legal personnel to top management ranks also increases in-

house lawyer counts.

Recall that Hypothesis 5 explains that the design of the portfolio of external legal

service providers is not independent of the proportion of legal work that is externally

procured. Our analysis confirms the hypothesized positive relation between the number of

law firms supplying to a company and the propensity to internalize legal work. The

coefficient for the number of external law firms is positive and significant at the 10% level,

suggesting that the proportion of legal work undertaken within the firm boundaries increases

with the number of external service providers. Although this result might sound counter-

intuitive, it is consistent with a relational contracting logic. Firms employing a large number

of suppliers do not develop the mechanisms that reduce contractual uncertainty and, as a

result, face higher contracting costs.

From the control variables we find that firm profitability is negatively correlated with

the extent of internal procurement of legal services. The coefficient for SG&A expenses, a

proxy for the overall amount of legal work undertaken by the firm, is also negative and

significant. This might reflect the ‘natural’ tendency to outsource legal work due to low asset

specificity and increased expertise in the marketplace. Finally, the coefficient for the number

of acquisitions undertaken is negative and significant. Given that corporate acquisitions

require specialist legal knowledge during negotiations, due diligence, contract design etc, it is

reasonable to expect that more outsourcing of legal work will take place.

22

Robustness checks

These results prove robust to several changes in model specification and sampling.

For example, we excluded firms from our sample that were involved either in antitrust suits

or in bankruptcy proceedings – see Model 6. We also obtain consistent results when re-

measuring the dependent variable, In-house lawyers, as a simple count and then analyzed in

Poisson panel count models – see Models 7-8. Results remain essentially unchanged when

employing different variable definitions (e.g. we varied Internationalization to account for

the number of foreign jurisdictions with differing (from US) legal system and rule-of-law

traditions) and when excluding possible outliers (e.g. R&D intensity values above 30%).

Another issue we explored further is the measurement of supplier portfolio

concentration. On average, we expect that a high number of law firms undertaking work for

the focal firm is associated with low concentration in the supply of legal services. However,

this relationship is less clear when there are changes in the type of representation. For

example, assume firm A employs 4 different law firms in four different practice areas

(M&As, patent litigation etc) while firm B similarly employs 4 law firms but in one practice

area. As it stands, our measure suggests that both corporations have the same level of supplier

concentration. However, firm B clearly has a less concentrated portfolio of suppliers given

that it has 4 different suppliers for the same practice area. To account for this, we counted the

number of law firms undertaking work for the focal firm in each practice area (this

information is available from ALM) and then averaged these. A high number of this measure

suggests that the focal firm has multiple suppliers in the same practice area, which is

indicative of low concentration in the provision of legal services. Consistent with our results,

the coefficient for this variable remains positive and significant at the 6% level.

23

DISCUSSION AND CONCLUSION

A substantial body of research has examined the conditions under which plural

sourcing is likely to occur. Moving beyond the question of when it occurs, the issue of how

firms choose the balance between making and buying in plural sourcing has remained under-

explored. This study fills that gap. In doing so we identify two mechanisms that dictate the

make-buy balance in plural sourcing, namely resource co-specialization and supplier

concentration. Our framework enables us to predict the make-and-buy mix in the context of

legal services provision.

The empirical findings support the view that when making and buying, firms exploit

resource co-specialization and supplier selection to tip the balance in favor of in-house

production and delivery. We noted statistically significant increases in the reliance on in-

house lawyers with increases in the number of countries where the firm had operations, with

increasing R&D and advertising expenses as a percentage of sales, when the CG is part of the

firm’s TMT, and as the number of law firms providing work for the firm increases. Overall,

these findings provide broad-based support for our make-and-buy framework that explains

how firms decide to balance internal and external activities.

Implications for theory and research

Our study has important implications for strategy theory. First, our study extends the

plural sourcing research by going beyond important but still basic questions of whether and

when firms engage in this practice to theorize about firm-level mechanisms, including

resource co-specialization and supplier selection, that determine the mix of sourcing modes.

Second, these co-specialization opportunities exist for resources lying in different

corporate functions. Thus, whilst our empirical context is the legal department as a focal

corporate function, other corporate functions also engage in plural sourcing by balancing

internal and external resources within the firm. In particular, in-house accountants work

24

alongside outside accounting and audit firms, in-house engineers with outside engineering

consultant firms, in-house marketing departments with marketing and PR agencies, and

internal strategy consultants with outside consulting firms. In these and other contexts, we

think our theoretical framework can provide researchers with relevant guides for making

inferences about make and buy strategies.

A third research implication relates to methods. We suspect that the dearth of research

on sourcing decisions in legal services is due to the absence of good data and the difficulty in

making causal inference. Despite their still relatively short time-series (i.e. seven years) and

limited cross-sectional coverage (i.e. Fortune 500 firms), our panel data provide us with

advantages relative to single firm case narratives (e.g. Smith, 2001) or one-time cross-

sectional surveys (e.g. Schwarcz, 2008).

Implications for practice

The findings of this study draw practitioners’ attention to a few, yet important, aspects

of the sourcing decisions in legal services. Some of them may appear paradoxical. First, in

legal circles, the last two decades have seen vigorous debate on the proper size and scope of

work for in-house lawyers. One view championed by the US giant, General Electric (Smith,

2001), argues for substantially increasing the number of in-house lawyers. In-house lawyers

are expected to increasingly play a dual role of being a lawyer and a business partner (Green,

2012). The chief legal officer (CLO) is said to be ‘one of the mightiest figures in the C-

suite’5. This is because the 2002 Sarbanes-Oxley Act, the 2010 Dodd-Frank Act and the 2008

financial crisis have heightened the need for compliance and risk management, making

companies turn to lawyers to prevent corporate bosses from going to jail and to fend against

endless threats of lawsuits. Our study indicates that whilst the power and status of GC may be

5 ‘A guardian and a guide’, Schumpeter column, The Economist April 7, 2012

25

on the rise, the resulting trend towards insourcing leads to business benefits only if resource

co-specialization benefits are exploited.

Limitations and further research

A number of limitations of this study are worth noting. The first limitation in the

analytical framework is self-imposed to simplify the analysis. Plural sourcing in legal

services in full manifestation involves multiple sourcing modes (make, buy, ally) from

multiple types of providers other than law firms, including LPO providers, legal technology

providers, and contract lawyers. By limiting our analysis to the co-existence of ‘make’ and

‘buy’ only from law firms, we did not take into account the impact of the growth of providers

other than law firms on the number of law firms and the consequent make-buy balance.

The second limitation lies in available data. Our study links the number of in-house

lawyers to the number of outside law firms, not the number of external counsel or hours spent

by these lawyers. In an ideal world, we would have better data on internal to external legal

personnel or hours worked. We deal with this limitation by including a number of control

variables and employing a dynamic panel data estimator.

A third limitation relates to sampling and generalizability. We think our framework

and evidence can be generalized to other large, established firms in the US. But this leaves

many other firm types where we are reluctant to generalize and advise additional study on

make and buy strategies. The discount consumer services purchasing giant, Groupon, was

founded in 2008, but did not have a full-time in-house counsel until 2011 when it already

operated in 48 countries generating $1.6 billion in annual revenue6. Groupon’s history

suggests that explanation of plural sourcing strategies for legal services in entrepreneurial

firms may require quite different frameworks and assumptions.

6 ABA Journal. 2011. The new normal: Make or buy in the age of the free-agent lawyer. ABA Journal.com.

October 26. Available electronically at: http://www.abajournal.com/legalrebels/article/in-house_lawyers_make_or_buy/!

26

Conclusion

We began this study by asking how firms develop plural sourcing strategies important

to their survival and success. We end it with a call for strategy researchers to continue

developing plural sourcing frameworks and evidence relevant to different organizational

forms and different product and service producing activities. We chose legal services and

demonstrated how such frameworks and evidence could help us understand the relative size

of in-house legal departments. Hitherto, strategy scholars have undervalued the analysis of

what Porter (1980) and others refer to as “support activities” to the firm. Our study suggests

that they can be essential to the firm’s strategy for gaining and maintaining competitive

advantage. In this context, we should count all the inside and outside professionals as part of

the search for an optimal make and buy mix in corporate functions. Research on firm

boundaries will be enriched by placing plural sourcing decisions in corporate functions on a

par with vertical (dis)integration decisions for the primary activities of the firm.

27

REFERENCES

Adelman MA. 1949. The Large Firm and Its Suppliers. The Review of Economics and

Statistics 31(2): 113-118. Arellano M. 1987. Computing robust standard errors for within-groups estimators. Oxford

Bulletin of Economics and Statistics 49(4): 431-434. Arellano M, Bond S. 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The Review of Economic Studies 58(2): 277-297. Arellano M, Bover O. 1995. Another look at the instrumental variable estimation of error-components models. Journal of Econometrics 68(1): 29-51. Argyres N. 1996. Evidence on the role of firm capabilities in vertical integration decisions. Strategic Management Journal 17: 129-150. Azoulay P. 2004. Capturing Knowledge within and across Firm Boundaries: Evidence from Clinical Development. American Economic Review 94(5): 1591-1612. Bagley CE. 2008. Winning legally: the value of legal astuteness. academy of Management

Review 33(2): 378-390. Baker G, Gibbons R, Murphy KJ. 2002. Relational contracts and the theory of the firm. Quarterly Journal of Economics 117(1): 39-84. Baldwin CY, Clark KB. 2003. Where do transactions come from? A perspective from engineering design. Paper presented at the Saint-Gobain Centre for Economic Research 5th Conference, Paris, France. Barney JB. 1986. Strategic factor markets: Expectations, luck, and business strategy. Management Science 32(10): 1231-1241. Blundell R, Bond S. 1998. Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics 87(1): 115-143. Bradach JL. 1997. Using the plural governance form in the management of restraurant chains. Administrative Science Quarterly 42(2): 276. Carpenter MA. 2002. The implications of strategy and social context for the relationship between top management team heterogeneity and firm performance. Strategic Management

Journal 23(3): 275-284. Cassiman B, Veugelers R. 2006. In search of complementarity in innovation strategy: Internal R&D and external knowledge acquisition. Management Science 52(1): 68-82. Coase RH. 1937. The nature of the firm. Econometrica 4(16): 386-405. Coase RH. 1988. The nature of the firm: Meaning. Journal of Law, Economics, and

Organization 4(1): 19-32. Cohen D. 1986. Trademark Strategy. Journal of Marketing 50(1): 61-74. Cohen D. 1991. Trademark Strategy Revisited. Journal of Marketing 55(3): 46-59. Dierickx I, Cool K. 1989. Asset stock accumulation and sustainability of competitive advantage. Management Science 35(12): 1504-1511. Dutta S, Bergen M, Heide JB, John G. 1995. Understanding Dual Distribution: The Case of Reps and House Accounts. Journal of Law, Economics, & Organization 11(1): 189-204. Dyer JH. 1997. Effective interim collaboration: how firms minimize transaction costs and maximise transaction value. Strategic Management Journal 18(7): 535-556. Fleischer V. 2010. Regulatory arbitrage. Texas Law Review 89: 227-290. Fox SP. 1998. Intelelctual property managementL From theory to practice. In Profiting from

Intellectual Capital. Sullivan P (ed.), Wiley: New York. Ghemawat P. 2001. Distance still matters. Harvard Business Review 79(8): 137-147. Gibbons R. 2005. Four formal(izable) theories of the firm. Journal of Economic Behavior &

Organization 58: 200-245.

28

Green S. 2012. In-house lawyers increasingly in dual legal/business roles. Corporate Counsel 9 March. Grindley PC, Teece DJ. 1997. Managing intellectual capital: Licensing and cross-licensing in semiconductors and electronics. California Management Review 39(2): 8-41. Gulati R, Puranam P. 2006. Complementarity and Constraints: Why Firms Both Make and Buy the Same Thing, Working paper. Hambrick DC, Mason PA. 1984. Upper echelons: the organization as a reflection of Its top managers. academy of Management Review 9(2): 193-206. Harrigan KR. 1986. Matching vertical integration strategies to competitive conditions. Strategic Management Journal 7(6): 535-555. He D, Nickerson JA. 2006. Why do firms make and buy? Efficiency, appropriability and competition in the trucking industry. Strategic Organization 4(1): 43-69. Heide JB. 2003. Plural governance in industrial purchasing. Journal of Marketing 67(4): 18. Jacobides MG, Billinger S. 2006. Designing the boundaries of the firm: From "make, buy, or ally" to the dynamic benefits of vertical architecture. Organization Science 17(2): 249-261. Jacobides MG, Winter SG. 2005. The co-evolution of capabilities and transaction costs: Explaining the institutional structure of production. Strategic Management Journal 26: 395-413. Kogut B, Zander U. 1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science 3(3): 383-397. Kogut B, Zander U. 1996. What firms do? Coordination, identity, and learning. Organization

Science 7(5): 502-518. Krzeminska A, Hoetker G, Mellewigt T. forthcoming. Reconceptualizing plural sourcing. Strategic Management Journal. Langlois RN. 1992. Transaction-cost economics in real time. Industrial and Corporate

Change 1(1): 99-127. Langlois RN. 2006. The secret life of mundane transaction costs. Organization Studies 27(9): 1389-1410. Lippman SA, Rumelt RP. 2003. A bargaining perspective on resource advantage. Strategic

Management Journal 24(11): 1069-1086. Madhok A. 2002. Reassessing the fundamentals and beyond: Ronald Coase, the transaction cost and resource-based theories of the firm and the institutional structure of production. Strategic Management Journal 23: 535-550. Marchant G, Robinson J. 1999. Is knwoing the tax code all it takes to be a tax expert? On the development of legal expertise. In Tacit Knowledge in Professional Practice. Sternberg RJ, Horvath JA (eds.), Taylor & Francis: New Jersey. Mayer KJ. 2006. Spillovers and governance: An analysis of knowledge and reputational spillovers in information technology. Academy of Management Journal 49(1): 69-84. Menz M. 2012. Functional Top Management Team Members : A Review, Synthesis, and Research Agenda. Journal of Management 38(1): 45-80. Milgrom P, Roberts J. 1995. Complementarities and fit: Strategy, structure, and organizational change in manufacturing. Journal of Accounting & Economics 19(2/3): 179-208. Moeen M, Somaya D, Mahoney JT. 2013. Supply portfolio concentration in outsourced knowledge-based services. Organization Science 24(1): 262-279. Nelson RL, Nielsen LB. 2000. Cops, counsel, and entrepreneurs: constructing the role of inside counsel in large corporations. Law & Society Review 34(2): 457-494. Nelson RR, Winter SG. 1982. An evolutionary theory of economic change. Harvard University Press: Cambridge.

29

Nickell S. 1981. Biases in dynamic models with fixed effects. Econometrica 49(6): 1417-1426. Orozco D. 2010. Legal Knowledge as an Intellectual Property Management Resource. American Business Law Journal 47(4): 687-726. Parmigiani A. 2007. Why do firms both make and buy? An investigation of concurrent sourcing. Strategic Management Journal 28(3): 285-311. Parmigiani A, Mitchell W. 2009. Complementarity, capabilities, and the boundaries of the firm: the impact of within-firm and interfirm expertise on concurrent sourcing of complementary components. Strategic Management Journal 30(10): 1065-1091. Peterson RA, Smith KH, Philip CZ. 1999. Trademark Dilution and the Practice of Marketing. Journal of the Academy of Marketing Science 27(2): 255-268. Poppo L, Zenger T. 1998. Testing alternative theories of the firm: Transaction cost, knowledge-based, and measurement explanations for make-or-buy decisions in information services. Strategic Management Journal 19: 853-877. Porter M. 1985. Competitive Advantage: Creating and Sustaining Superior Performance. The Free Press: New York. Porter ME. 1980. Competitive strategy. Free Press: New York. Prahalad C. 1990. Globalization: the intellectual and managerial challenges. Human Resource

Management 29(Spring): 27-38. Puranam P, Gulati R, Bhattacharya S. forthcoming. How Much to Make and How Much to Buy? An Analysis of Optimal Plural Sourcing Strategies. Strategic Management Journal. Reitzig M, Puranam P. 2009. Value appropriation as an organizational capability: The case of IP protection through patents. Strategic Management Journal 30(7): 765-789. Roodman D. 2006. How to do xtabond2: An introduction to difference and system GMM in Stata, Center for Global Development working paper 103. Rosen RE. 1984. Lawyers in Corporate Decision-Making. In Department of Sociology. University of California Berkeley. Rothaermel FT, Hitt MA, Jobe LA. 2006. Balancing vertical integration and strategic outsourcing: effects on product portfolio, product success, and firm performance. Strategic

Management Journal 27(11): 1033-1056. Sako M. 2011. General Counsel with Power? In Novak Druce Working Paper. Novak Druce Centre for Professional Service Firms. Sako M. 2013. Professionals between market and hierarchy: a comparative political economy perspective. Socio-Economic Review 11: 1-28. Sako M, Helper S. 1998. Determinants of trust in supplier relations: Evidence from the automotive industry in Japan and the United States. Journal of Economic Behavior &

Organization 34(3): 387-417. Sanders WG, Carpenter MA. 1998. Internationalization and firm governance: the roles of CEO compensation, top team composition, and board structure. Academy of Management

Journal 41(2): 158-178. Schwarcz SL. 2008. To Make or to Buy: In-House Lawyering and Value Creation. The

Journal of Corporation Law 33(2): 498-570. Smith L. 2001. Inside outside: How businesses buy legal services. American Legal Media: New York, NY. Somaya D, Williamson IO, Xiaomeng Z. 2007. Combining patent law expertise with R&D for patenting performance. Organization Science 18(6): 922-937. Susskind R. 2008. The End of Lawyers? Rethinking the Nature of Legal Services. Oxford University Press: Oxford. Teece DJ. 1986. Profiting from technological innovation: Implications for integration, collaboration, licensing and public policy. Research Policy 15: 285-305.

30

Walker G, Weber D. 1984. A transaction cost approach to make-or-buy decisions. Administrative Science Quarterly 29: 373-391. Williamson OE. 1985. The economic institutions of capitalism. The Free Press: New York. Windmeijer F. 2005. A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of Econometrics 126(1): 25-51. Zaheer S. 1995. Overcoming the liability of foreignness. Academy of Management Journal38 38(2): 341-363.

31

Table 1. Descriptive statistics and pairwise correlations

Descriptive statistics Pair-wise correlations

Variable Mean St.Dv. Min Max (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

1) In-house lawyers 5.37 1.10 1.44 8.44 1.00

2) Internationalization 2.24 1.48 0.00 4.55 0.36* 1.00

3) R&D intensity 0.02 0.05 0.00 0.53 0.30* 0.41* 1.00

4) Advertising intensity 0.01 0.03 0.00 0.21 0.17* 0.14* 0.07* 1.00

5) General counsel on TMT 0.79 0.41 0.00 1.00 0.19* 0.02 0.05 0.06 1.00

6) Outside law firms 1.83 0.93 0.00 3.95 0.14* 0.21* 0.12* 0.02 0.09* 1.00

7) Diversification 0.90 0.36 0.00 2.05 0.01 0.17* -0.11* 0.09* 0.00 0.03 1.00

8) Debt 0.04 0.07 0.00 0.46 -0.02 0.04 0.00 0.02 -0.10* 0.06 0.00 1.00

9) Profitability 0.16 0.13 -0.31 0.77 0.40* 0.24* 0.20* 0.10* 0.11* 0.14* 0.00 0.10* 1.00

10) Employees 3.67 1.11 0.00 6.32 -0.08* 0.24* 0.06 0.10* 0.08* 0.29* 0.15* 0.01 -0.03 1.00

11) SG&A 6.64 2.93 0.00 10.79 -0.20* 0.24* 0.28* 0.28* 0.07* 0.10* 0.08* -0.07* -0.12* 0.36* 1.00

12) Litigation 0.13 0.44 0.00 3.00 0.21* 0.18* 0.29* 0.12* 0.02 0.18* -0.04 0.04 0.23* 0.12* 0.06 1.00

13) Acquisitions 0.09 0.33 0.00 3.00 0.08* 0.09* 0.08* -0.01 -0.01 0.05 0.01 0.20* 0.22* 0.12* -0.04 0.15* 1.00

14) Bankruptcy 0.00 0.06 0.00 1.00 0.04 -0.01 -0.03 -0.02 -0.02 0.08* 0.01 0.03 -0.07* 0.04 -0.13* 0.07* 0.10* 1.00

N=945, * statistically significant at the 5% level

32

Table 2. Panel data regression models of corporate legal services make and buy balance

Dep. Variable In-house lawyers No of lawyers

(count)

(1) (2) (3) (4) (5) (6) (7) (8)

Fixed

effects

Fixed

effects

System

GMM

System

GMM

System

GMM

System

GMM

Fixed

effects

Poisson

Random

effects

Poisson

Lagged dep. variable 0.310** 0.305** 0.973** 0.882** 0.883** 0.882**

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

Explanatory variables

No of countries (log) 0.055 0.041** 0.040** 0.039** 0.043 0.270**

(0.818) (0.005) (0.008) (0.008)

(0.661) (0.000)

R&D intensity 2.878* 0.770† 0.764† 0.912† 0.518† 2.407**

(0.028) (0.072) (0.078) (0.051) (0.068) (0.000)

Advertising intensity 0.350 2.442* 2.389* 2.469* 1.636* 1.240*

(0.778) (0.012) (0.012) (0.017) (0.013) (0.049)

General counsel in TMT 0.090* 0.090† 0.090† 0.0938† 0.180** 0.161**

(0.034) (0.094) (0.079) (0.082) (0.000) (0.000)

No of law firms (log) 0.023 0.041† 0.041† 0.0389† 0.019** 0.014*

(0.112) (0.069) (0.072) (0.076) (0.000) (0.017)

Control variables

Product diversification 0.006 0.021 0.022 0.000 0.0301 0.103**

(0.937) (0.653) (0.641) (0.990) (0.236) (0.000)

Debt ratio -0.139 -0.213 -0.179 -0.229 -0.218 -0.234 -0.405* -0.605**

(0.686) (0.562) (0.132) (0.165) (0.200) (0.144) (0.010) (0.000)

Profitability -1.092** -1.026** -0.444* -0.346* -0.343* -0.290† -0.057 -0.227**

(0.000) (0.000) (0.037) (0.029) (0.045) (0.075) (0.486) (0.006)

No of employees (log) -0.344** -0.344** 0.003 -0.009 -0.010 -0.012 -0.000 -0.024

(0.000) (0.000) (0.825) (0.593) (0.553) (0.502) (0.995) (0.356)

Selling, general & admin

expenses (log)

-0.118** -0.117** -0.018 -0.040** -0.040** -0.035* 0.093** -0.049**

(0.005) (0.005) (0.204) (0.008) (0.007) (0.032) (0.000) (0.000)

No of litigation cases -0.028+ -0.040* 0.065 0.046 0.047 0.053 0.000 -0.010

(0.096) (0.013) (0.169) (0.141) (0.132) (0.106) (0.923) (0.175)

No of litigation casest-1 -0.021 -0.009 -0.014 -0.012 -0.012 -0.059† -0.010 -0.000

(0.529) (0.765) (0.746) (0.722) (0.730) (0.084) (0.174) (0.922)

No of acquisitions -0.003 -0.012 -0.030 -0.055† -0.055† -0.079* -0.004 -0.019+

(0.911) (0.722) (0.369) (0.088) (0.089) (0.020) (0.682) (0.057)

No of acquisitions-1 -0.034 -0.035 -0.014 -0.029 -0.028 -0.0301 -0.018* -0.027**

(0.270) (0.266) (0.683) (0.380) (0.391) (0.440) (0.036) (0.002)

Bankruptcy -0.463* -0.467* -0.480 -0.389 -0.382 0.0210 -0.0228

(0.049) (0.045) (0.411) (0.344) (0.356) (0.625) (0.601)

Sales (log) 0.141** 0.331**

(0.000) (0.000)

Constant 5.952** 5.658** 0.325 0.674* 0.676* 0.680* 0.264

(0.000) (0.000) (0.257) (0.012) (0.013) (0.012) (0.339)

Year dummies Yes Yes Yes Yes Yes Yes Yes Yes

N (number of firms) 945(285) 945(285) 945(285) 945(285) 945(285) 921(280) 945(285) 945(285)

Wald x2 (R2) (0.344) (0.361) 3119.9 4885.9 5060.1 5006.39 218.64 347.7

Arellano-Bond test AR(1) -3.80** -4.15** -3.97** -4.10**

Arellano-Bond test AR(2) 0.31 0.23 0.23 -0.42

No of instruments 126 249 249 248

Hansen test 106.2 224.74 224.74 214.95

p-values in parentheses, † p ! 10%, * p ! 5%, ** p !1%

33

Figures 1A-E. Locally-weighted scatter-plot smoothed and comparative bar-chart analyses of in-house

lawyer counts

!

24

68

log

law

ye

rs_

pe

r_sa

les

0 20 40 60 80 100No_countries

bandwidth = .8

Lowess smoother

In-h

ou

se l

awy

ers

No of countries

24

68

log

law

ye

rs_

pe

r_sa

les

0 .1 .2 .3 .4 .5rndintensity

bandwidth = .8

Lowess smoother

In-h

ouse

law

yer

s

R&D intensity

24

68

logla

wye

rs_per_

sa

les

0 10 20 30 40 50no_of_lawfirms

bandwidth = .8

Lowess smoother

In-h

ouse

law

yer

s

No of law firms

24

68

log

law

ye

rs_

pe

r_sa

les

0 .05 .1 .15 .2advert_intensity

bandwidth = .8

Lowess smoother

In-h

ouse

law

yer

s

Advertising intensity

24

68

10

loglawyers_per_sales

0 1

In-h

ou

se l

awy

ers

CG in TMT

No Yes

1A 1B

1C 1D

1E