Learning by Supplying

Juan Alcacer

Harvard Business School

Joanne Oxley

Rotman School of Management

University of Toronto

Abstract

Outsourcing in many industries has advanced beyond simple component supply to encompass

manufacturing of entire products, often by suppliers in emerging economies. Understanding the

evolving role and capabilities of suppliers in global supply chains is thus a pressing strategic

issue for suppliers and customers alike. We analyze a novel panel dataset of supply relationships

in the mobile telecommunications industry to answer the following questions: What factors

contribute to a supplier‘s ability to build technological and market capabilities? Does it matter to

whom the firm supplies? Is involvement in product design important, or is manufacturing the key

locus of learning? Do the same types of relationships that support technological innovation also

facilitate successful introduction of own-brand products, or does this require a different ‗locus‘

of learning?

1

Introduction

The globalization of markets and the rise of offshore outsourcing has elevated supply chain

management from an operational issue to a key strategic concern. In the early 2000s, McKinsey

reported that the average consumer electronics firm aimed to outsource almost 75 percent of its

manufacturing; two thirds of US auto companies‘ value added already resided with suppliers;

and contract drug development and manufacturing in pharmaceuticals was growing at close to 20

percent per annum (Doig et al., 2001). As this trend has continued, outsourcing in many

industries has advanced beyond simple component supply to encompass the manufacture of

entire products, often by suppliers in emerging economies. In this new environment,

understanding the evolving role and capabilities of suppliers is a pressing strategic issue for

suppliers and customers alike.

Interest in the strategic implications of evolving supply chains is intensified by speculation

that some firms in emerging economies may successfully parlay their experience as ‗Original

Equipment Manufacturers‘ (OEMs) supplying to major branded producers into positions as

viable world-class players in their industry. In particular there is a concern that this process –

which we dub ‗learning by supplying‘– may come at the expense of previous market leaders

(Pisano and Shih, 2009). In an influential Harvard Business Review article, for example, Khanna

and Palepu (2006: 66-67) note that,‖

Taiwan-based Inventec…is among the world‘s largest manufacturers of

notebook computers, PCs, and servers, many of which it makes in China

and sells to Hewlett-Packard and Toshiba… Inventec has mastered the

challenges associated with sourcing components from around the world,

assembling them into quality products at a low cost, and shipping them to

multinational companies in a reliable fashion. Recently Inventec started

selling computers in Taiwan and China under its own brand name. The

computers have a Chinese operating system and software, so Inventec

doesn‘t compete directly with its customers – yet.

This description suggests that Inventec‘s growing technical and marketing capabilities may

be directly attributable to its experience as an OEM supplier. Other anecdotal evidence reinforces

2

this connection and suggests that, from the perspective of suppliers, building the technological

and marketing capabilities necessary for independent sales success represents an effective –

perhaps even critical – path to increased and sustainable profitability. The popular ―Smiling

Curve,‖ developed by Acer CEO Stan Shih captures this received wisdom:

Recognizing that Acer‘s focus on assembling PCs was keeping the

company in the least profitable segment of the market, CEO Stan Shih

decided to move up the value curve by developing capabilities in

components and distribution. Succeeding in components required

[developing] strong technology…Succeeding in distribution required a

strong brand, established channels and effective logistics. Acer has built

both (Bartlett and Ghoshal, 2000: 134).

A closer look at Acer‘s experience nonetheless suggests a more nuanced view of the learning

by supplying process: Acer‘s increased technical prowess is clearly visible in its international

patenting activity, with cumulative patent holdings going from a mere handful in 1990 to over

600 by 2010. Developing strong marketing capabilities appears to have been a bigger challenge,

however, as Acer has faltered in its efforts to sustain sales of its own-brand products and still

gets most of its earnings from OEM relationships, supplying computers and other electronics

products to industry leaders. Indeed, despite the apparent benefits of moving up the value curve,

to date there are few prominent examples of firms ‗breaking out‘ of their role as OEM suppliers

to become viable world-class competitors.

These observations raise intriguing questions for strategy researchers as well as for OEM

suppliers as they attempt to climb up the value curve in search of greater profits: To what extent

are firms able to leverage their experience as OEM suppliers to build valuable technological

capabilities, and ultimately establish independent (own-brand) sales success? What factors

facilitate these ‗learning by supplying‘ processes? Does it matter to whom the firm supplies? Is

involvement in product design important, or is manufacturing the key locus of learning? And do

the same types of relationships that support technological innovation also facilitate successful

introduction of own-brand products, or does this require a different ‗locus‘ of learning?

3

To date, researchers and managers contemplating these important questions have by

necessity relied on the type of anecdotal evidence cited above, due to a scarcity of relevant

empirical research.1 In this paper we remedy this situation by documenting the extent and nature

of learning by supplying in the context of one important industry, the mobile telecommunications

handset industry. This is a dynamic industry within the electronics manufacturing sector often

featured in popular debates about offshore outsourcing and the migration of capabilities

(Engardio and Einhorn, 2005). We also apply theoretical insights from research in related fields

to generate hypotheses linking the development of technological capabilities (patenting activity)

and marketing capabilities (introduction and sales of own-brand products) to the duration and

extent of an OEM supplier‘s experience in the industry, as well as to characteristics of the

customers served, the nature of the supply relationship, and the initial resources and capabilities

that a supplier brings to the relationship.

Our empirical study exploits a novel dataset detailing all significant supply relationships for

the design and manufacture of complete handsets for major branded producers and operators

over the entire history of the mobile telecom industry. We marry this data with information on

the patenting activities of OEM suppliers and their customers, and on mobile handsets introduced

by these firms over the period 1995-2010. We thus generate an unusually complete picture of

outsourcing in the industry, allowing us to assess the extent to which OEM suppliers learn by

supplying and move up the value curve beyond basic manufacturing to achieve independent

technological innovation and sales of own-brand products.

Our empirical analysis yields robust evidence of learning by supplying in terms of the

accumulation of technological resources. For suppliers in our sample, telecom-related patenting

1 What empirical evidence exists is primarily based on a small number of cases describing the rise of

multinationals from emerging economies (e.g., Khanna and Palepu, 2006; Duysters et al., 2009; Pisano and Shih, 2009). There is also a related body of literature examining the effect of international trade and foreign direct investment on technological and economic convergence or ‗catch-up‘ at the country level. See Athreye and Cantwell (2007) for a useful review and discussion.

4

increases as they accumulate experience in handset supply relationships, particularly when these

relationships involve handset design in addition to manufacturing. We also find evidence of

convergence in supplier and customer technological resources over the course of a supply

relationship, indicating that it indeed matters to whom one supplies in this industry.

Supplementary analysis of supplier selection processes indicates that there is significant inertia in

customer-supplier matches, such that incumbent firms‘ initial supplier choices have a strong

influence on suppliers‘ long-term capability development.

The relationship between supply experience and marketing success with own-brand products

is more complex. We find that a majority of OEM suppliers have attempted to introduce their

own branded products, but that few have done so successfully. This is a striking observation,

given the higher profitability of own-brand sales and the fact that independent sales success is

the professed ultimate goal for many OEM suppliers.2 Moreover, there appears to be little

relationship between a supplier‘s accumulation of technological capabilities and subsequent

success with own-brand products. Our findings thus point to distinct pathways to technological

and market learning, and suggest that suppliers face significantly greater obstacles to learning in

the marketing domain.

Our findings also indicate that customers differ significantly in the extent to which they are

able and willing to share technological and marketing knowledge with suppliers. For example,

even though wireless operators are more likely to delegate design activities to their handset

suppliers, relative to branded producers, these customers do not appear to be a robust source of

technological learning for suppliers. Conversely, when it comes to the accumulation of the

marketing capabilities necessary for independent sales, supplying to operators represents a

particularly important ‗pathway‘ to own-brand introduction and sales, while supplying to market

2 This and other qualitative assessments of the experiences and strategic choices made by industry

participants are derived from field research carried out through interviews with managers at 3 top brand producers, 4 OEM suppliers, and 5 operators, over the period 2007-2012.

5

leaders strongly inhibits sales of own-brand products. These findings are consistent with

observations in the industry that operators are more likely to involve their handset suppliers in

‗customer-facing‘ activities, such as product definition and marketing. Leading branded

producers, on the other hand, although technologically more advanced, tend to write more

restrictive outsourcing agreements and maintain a greater distance between customers and

suppliers. As a consequence, suppliers working with these firms may find themselves effectively

locked into a subordinate role, thwarting ambitions to move up the value curve and develop as

independent participants in the industry.

By providing the first systematic analysis of the extent and nature of learning by supplying in

one important industry, our study injects much-needed new evidence into ongoing debates about

the impact of offshore outsourcing on firm capabilities and competitiveness. While ‗doomsday

commentators‘ may be correct in their suspicions regarding the ambitions of OEM suppliers to

develop into viable international competitors, their belief that this represents an impending threat

to current market leaders, or to US competitiveness, is almost certainly overblown. In the

concluding section of the paper we discuss additional implications of our study for supply chain

management and for future strategy research.

Theoretical background and hypotheses

We define learning by supplying as the accumulation of technological and marketing capabilities

resulting from a firm‘s experience as an OEM supplier. Consistent with Shih‘s ‗Smiling Curve‘

concept (Bartlett and Ghoshal, 2000), we start with the assumption that suppliers have an

incentive to move up the value curve to become independent innovators and producers of their

own branded products, and that this requires the accumulation of technological and marketing

resources. As suggested in the introduction, learning by supplying so defined has received little

direct empirical study to date. There is nonetheless a rich body of prior research focusing on

related processes through which firms accumulate capabilities and strengthen their competitive

position. These processes include learning-by-doing, learning from international exchange, and

6

learning from alliance partners. In the following, we apply insights from this prior research to

generate hypotheses linking supplier learning (i.e., the accumulation of technological and

marketing capabilities) to the duration and extent of an OEM supplier‘s experience in the

industry, as well as to characteristics of the customers served and of the supplier itself. In

keeping with the empirical focus of our paper, the goal here is not to generate new theory per se,

but rather to apply existing theory to better understand this new and important phenomenon.

The importance of learning from direct production experience, or learning-by-doing, has

been well documented in the economics literature, dating back to early theoretical work by

Arrow (1962). The first empirical studies focused on the shape of individual firms‘ ‗learning

curves,‘ and generated robust evidence that costs tend to decline as a firm‘s cumulative

production volumes increase (Rapping, 1965). Later work found evidence of industry-level

learning curves and learning-by-doing spillovers (Lieberman, 1984; Irwin and Klenow, 1994),

but reaffirmed that a firm‘s own direct experience has the greatest impact on learning.3

More recent studies in this tradition have examined the impact of learning-by-doing on

different aspects of firm performance – notably survival (Baum and Ingram, 1998), productivity

(Darr, Argote and Epple, 1995), and innovation (Penner-Hahn and Shaver, 2005). Researchers

have also begun to explore other types of experience-based learning. For example, a recent

literature rooted in models of trade and endogenous growth (Romer, 1990; Grossman and

Helpman, 1993) examines the link between international trade and innovation. Based on the

premise that trade exposes firms to sources of knowledge that would otherwise be unavailable to

them, scholars have looked for – and found – evidence of ―learning by exporting‖ (e.g., Blalock

and Gertler, 2004; Salomon and Shaver, 2005; Van Biesebroeck, 2005). In an empirical model

that allows for positive feedback between innovation and exporting, for example, Salomon and

Shaver (2005) find that exporting leads to significant increases in both technological innovation

3 Irwin and Klenow (1994), for example, show that firms learn three times more from an additional unit of

their own cumulative production than from an additional unit of another firm's cumulative production.

7

(as indicated by an increase in patent applications) and product innovation (i.e., new product

introductions) at the firm level. Applying these arguments to the accumulation of capabilities by

OEM suppliers leads naturally to the following baseline hypothesis:

H1: An OEM supplier’s accumulation of capabilities will be positively related to the duration

and extent of its accumulated supply activity in the industry.

The basic argument in the learning by exporting literature is that buyers encountered in

export markets are more advanced and/or have different requirements from domestic buyers, and

that exposure to these more sophisticated buyers enables suppliers to accumulate additional

capabilities (Clerides, et al, 1998). A similar theme runs through the writing on international

technology diffusion and foreign direct investment (FDI) in a body of applied literature that

sprang up in the 1980s as East Asian economies emerged as manufacturing powerhouses. Rhee,

Ross-Larson and Purcell (1984: 61), reporting on interviews with Korean firms, for example,

noted that ―Almost half of the firms said they had directly benefited from the technical

information foreign buyers provided: through visits to their plants by engineers or other technical

staff of the foreign buyers, through visits by their engineering staff to the foreign buyers…and

through feedback on the design, quality and technical performance of their products.‖ 4

This premise, that local firms can effectively learn from multinational companies (MNCs) to

whom they supply, has been picked up in more recent work exploring vertical FDI linkages. For

example, in a theoretical model, Pack and Saggi (2001) show that knowledge transfer from

MNCs to their suppliers can benefit both parties involved by increasing efficiency and

innovation, such that MNCs are motivated to actively share knowledge with their suppliers. A

similar conclusion emerges from the supplier involvement literature that has documented the

benefit of knowledge sharing between customers and suppliers, particularly in the auto industry

and other complex manufacturing industries (see, e.g., MacDuffie and Helper, 1997; Takeishi,

4 Quoted in Pack and Saggi, 2001: 390; See also Hobday, 1995, and references therein.

8

2001). A natural implication of these prior findings for our understanding of learning by

supplying is that the accumulation of capabilities by an OEM supplier will be enhanced, all else

equal, when the supplier‘s customers themselves possess superior capabilities, such that they

have a lot to ‗teach‘ the supplier. Thus we have:

H2: An OEM supplier’s accumulation of capabilities will be positively related to the level of

capabilities possessed by the customer(s) served.

Of course it may be that not all suppliers stand to benefit equally from a given supply

relationship. Blalock and Simon (2009), in their study of FDI in Indonesian manufacturing, for

example, find that the positive association between increased FDI and improvements in the

productivity of local firms is moderated by the local firm‘s ‗absorptive capacity‘ (Cohen and

Levinthal, 1990), as proxied by R&D investments and employee human capital. The validity of

this inference is reinforced by research findings on learning in inter-organizational alliances,

wherein learning from alliance partners is conditioned, inter alia, on a focal firm‘s initial stock

of related knowledge (e.g., Sampson, 2004). This suggests the following hypothesis:

H3: An OEM supplier’s accumulation of capabilities will be positively related to the level of

its own pre-existing capabilities.

In addition to highlighting the importance of absorptive capacity for inter-partner learning,

prior alliance research also indicates that the extent of knowledge-sharing and learning is related

to alliance scope – i.e., the range of activities that are jointly undertaken by the partners. In

particular, it has been argued that when an alliance includes design or R&D activities alongside

manufacturing this necessitates an increase in the extent of knowledge sharing among the

participants, relative to ‗pure‘ R&D or manufacturing alliances (Mowery, Oxley and Silverman,

1996; Oxley and Sampson, 2004). This point is also echoed in the supplier involvement literature

where, for example, Takeishi (2001:405) reports that: ―Many auto industry studies showed that

effective supply chain management…involved close, trusting relationships with long-standing

suppliers who were intimately involved with development as well as production of components.‖

9

In the context of learning by supplying these findings imply that supply relationships that

encompass design in addition to manufacturing (i.e. Original Design and Manufacturing or ODM

relationships) should offer greater opportunities for supplier learning than manufacturing-only

(‗pure‘ OEM) relationships:

H4: An OEM supplier’s accumulation of capabilities will be greater in ODM relationships

than in ‘pure’ OEM relationships.

The rosy view of mutually-beneficial knowledge sharing in FDI linkages and supplier

relationships seems, on its face, to be at odds with current concerns about the potential threat to

market incumbents posed by ‗upstart‘ OEM suppliers (as described in Khanna and Palepu,

2006). A deeper examination suggests, however, that the apparent contradiction reflects changes

in global supply chains that lead to a violation of a key assumption embedded in models of FDI

linkages and supplier involvement, i.e., that outsourcing is essentially limited to component

supply, and that local suppliers do not have the marketing knowledge or resources to compete

effectively with their MNC customers. As long as these conditions hold, then knowledge transfer

is appropriately viewed as win-win, since the dominant effect is to increase efficiency and

innovation through technical knowledge sharing with component suppliers, and the competitive

threat is minimal. However, now that outsourcing routinely encompasses the manufacture (and

sometimes design) of complete products, knowledge sharing may still generate significant

mutual benefit through increased efficiency, but there is also a real possibility that with active

knowledge sharing the supplier will accumulate the technological and marketing capabilities

necessary to compete effectively against its erstwhile customer(s). Given a sufficiently large

threat of entry, the incentives for the MNC to share knowledge with its suppliers are significantly

reduced (Pack and Saggi, 2001), thus limiting the opportunities for supplier learning.

The above arguments suggest that we should pay close attention to both the means and the

motives for knowledge sharing and cooperation between customers and their suppliers. This idea

also resonates strongly with prior research on the scope of knowledge sharing in R&D alliances.

10

Oxley and Sampson (2004) for example, find that the scope of alliance activities is significantly

influenced by the competitive relationship between partner firms. In particular, these authors

argue that in alliances between direct competitors participants are less willing to increase alliance

scope, even at the cost of reduced efficiency, because the risk associated with knowledge leakage

to competitors is too high. Based on their empirical findings, Oxley and Sampson conclude that

competitive considerations are particularly salient when it comes to joint marketing, and the

sharing of marketing-related knowledge, so that:

…firms decide whether or not to engage in joint manufacturing based on

the needs of the project and the capabilities of the partners, and then

mitigate the hazards posed by joint manufacturing through choice of an

appropriate governance structure. [Conversely], potential hazards raised by

joint marketing are primarily mitigated by a reduction of alliance scope—

i.e., partners simply avoid joint marketing when they foresee problems

(Oxley and Sampson, 2004: 742-743).

These findings thus reinforce the usefulness of examining suppliers‘ accumulation of

technological and marketing resources separately, since it is reasonable to expect that these two

processes represent different ‗loci‘ of supplier learning. Moreover, applying Oxley and

Sampson‘s competitive logic to our context of supplier learning suggests that customers may be

more amenable to involving suppliers in technological activities than in marketing or customer-

related activities, and that knowledge-sharing – and supplier learning – will be more extensive in

the technological domain than in the marketing domain, particularly if customers perceive

suppliers to be potential market rivals.

In sum, application of insights from prior theoretical and empirical research in related areas

to the phenomenon of learning by supplying establishes some base-line predictions, and points to

the importance of examining the accumulation of both technical and marketing capabilities by

supplier firms. In the following section we introduce the empirical setting for our study – the

mobile telecommunications handset industry – and elaborate on the evolving competitive

dynamics in this industry.

11

Evolution of the mobile telecommunications handset industry

The mobile telecommunications handset industry is a relatively new industry. The first

commercial mobile handsets emerged circa 1985, but demand did not take off until the early

1990s; since then, production has increased exponentially. From the beginning of the industry,

the market has been dominated by a handful of powerful branded producers - Nokia, Motorola,

Sony, and Ericsson5- later joined by Samsung and LG. Today, the industry continues to be

dominated by this core group of branded handset producers, but there is a vibrant and growing

set of peripheral providers – some of whom are suppliers that have successfully introduced their

own branded handsets. Nonetheless, industry concentration remains high: in 2010 the five

leading firms still accounted for over 80% of global handset sales, despite the dramatic growth

experienced in the market. Demand growth was particularly strong during the ‗telecom boom‘ of

the late 1990s, when demand outstripped available supply. In contrast to other industries in the

electronics sector, however, outsourcing of manufacturing among the leading branded producers

was quite rare during this period, and firms invested heavily in their own manufacturing plants in

response to the supply shortfall.6

It was only in the post-boom crash of 2000-2001, with excess global production capacity

suddenly present in the industry, that major branded producers turned to outsourcing as a way to

rationalize operations. Many firms sold manufacturing plants to existing electronics

manufacturing firms, most notably Flextronics, Foxconn, and Solectron, opening the door to

significant outsourcing in the industry. This door has since been flung wide open as many more

suppliers came on line, with production at first centered in Europe and North America, but

rapidly shifting to Asia. The scope of outsourced activities has also increased: initially

outsourcing was typically limited to manufacturing-only (OEM) agreements, but in later periods

5 Sony and Ericsson merged their handset businesses in 2001, forming Sony-Ericsson.

6 This decision reflects the rapid pace of technological change in the handset industry during this period,

as well as the dearth of capable suppliers available at the time.

12

branded producers began to outsource handset design responsibilities as well, entering into ODM

agreements with their suppliers. In these agreements the branded producer specifies performance

requirements and selects key components such as the display and core chips, but the supplier

does much of the mechanical and electrical design (Engardio and Einhorn, 2005).

Branded producers sell handsets direct to consumers through retail outlets, and also through

mobile telecom operators around the globe: Cingular, Sprint and Verizon in the US, Virgin

Mobile and Vodafone in the UK, and NTT Docomo in Japan, to name a few prominent

examples. During the telecom boom of the 1990s these operators sought greater control of

handset supply, in part to counter the threat of shortages, but also to more directly influence

handset design and increase hardware and service integration. Because operators had little or no

prior experience in handset production this was typically accomplished through ODM

agreements with established suppliers. The resulting handsets were then distributed exclusively

by the operator under the operator‘s own brand name (Yoffie, Alcacer and Kim, 2012).

The incentives for suppliers to climb up the value curve and compete independently in the

industry are clear, as profit margins for branded producers still significantly exceed those of

OEM suppliers. For example, when Taiwan‘s HTC Corp. successfully introduced its own

branded handsets in 2006, it saw its profit margin increase from 18% in 2004 (in line with peer

OEM suppliers) to 33% in 2008. This latter figure compares quite favorably with leading brand

producers such as Nokia, Samsung, Motorola, etc., whose gross margins were in the range of 25-

30% during this period.7 And indeed, a significant fraction of suppliers have introduced own-

brand handsets in recent years, in pursuit of these higher margins, though few have matched

HTC‘s success in global markets: most own-brand introductions have been limited to suppliers‘

home markets (examples include Ningbo Bird in China, Sewon and Telson in South Korea), and

OEM relationships continue to account for the largest share of these companies‘ revenues.

7 Author estimates based on IQ Capital data, various years.

13

Outsourcing suppliers have also made significant leaps forward in terms of technological

innovation since the beginning of the outsourcing era: while as a group these firms held almost

no telecommunications-related patents in the early 1990s, many are now active innovators and

regularly patent their innovations in the US and elsewhere. Thus it appears that learning by

supplying is a motivating force in this industry, as firms seek to leverage their supply experience

as OEM suppliers to accumulate technological and marketing capabilities and climb up the value

curve.

The co-existence of two distinct categories of mobile handset providers (branded producers

and mobile telecom operators), both of whom source phones from OEM suppliers, introduces a

source of heterogeneity that is particularly useful for our empirical analysis. Industry

commentaries and our own observations suggest that branded producers tend to have

significantly stronger technological capabilities than operators and, as such, suppliers to branded

producers potentially gain access to – or at least exposure to – more advanced technologies.

Observers also note, however, that branded manufacturers have traditionally been quite

protective of their core technologies in relationship to suppliers, limiting outsourcing to less

technologically-advanced and low cost handsets (known as feature phones) or to handsets based

on standards in which they had not previously invested significantly.8 Engardio and Einhorn,

(2005) further note that branded producers have tended to limit suppliers‘ involvement in

customer-facing aspects of the handset design process such as product definition, jealously

guarding their customer relationships. This predictably limits suppliers‘ accumulation of

marketing capabilities and successful introduction of own brand phones as ―intimacy with the

customer‖ is reportedly crucial to success in mobile telecommunications as in other consumer

product industries (Engardio and Einhorn, 2005).

8 Nokia, for example, has outsourced most CDMA handsets while producing GSM handsets in-house.

More generally, according to data from THT Research, 70% of outsourcing by leading producers was for relatively low-cost feature phones.

14

In contrast, most operators have willingly outsourced manufacturing of even high-end cell

phones (e.g., smart phones), and frequently involving suppliers in all aspects of handset design

and manufacture. For example, in describing HTC‘s relationship with its operator clients (prior

to introduction of its own branded phones) Yoffie et al. (2012:3) note:

Carriers embraced HTC as they gained a greater sense of control over their

product portfolio…Richard Brennan, a former Orange executive [recalled],

―Because HTC were bending over backwards to deliver, you wanted to

make your relationship with HTC work and help the underdog become

successful.‖

Note that operators‘ greater willingness to share knowledge (especially market-related

knowledge), relative to branded producers, is also consistent with Pack and Saggi‘s (2001)

theoretical model, discussed above: mobile operators compete primarily in telecom service

provision and knowledge-sharing with suppliers is thus unlikely to undermine operator profits.

Indeed, if OEM suppliers are able to gain greater insights into the needs of a particular operator‘s

customers, and produce tailored products to suit, this will likely have a positive effect on

operator revenues. In contrast, for branded producers the efficiency benefits of knowledge-

sharing must be balanced against the threat imposed by potential entry of the OEM supplier, as

well as the threat of knowledge leakage to current competitors.

Data

Data for our empirical analysis comes from a wide variety of sources. Our goal for this project

was to assemble a comprehensive dataset covering handset design and manufacturing supply

relationships for all of the major branded producers and operators active in the mobile telecom

industry from the beginning of the outsourcing era to the present day. Extensive search revealed

that, as we suspected, no single source existed that could accomplish this goal. We therefore

drew on a variety of proprietary databases and web-based resources. For the web-based data

search we crawled and compiled relevant information from current and archived pages of

electronic product comparison websites, as well as industry association, news media, and

government websites, to gather information on mobile handset production and outsourcing

15

relationships.9 To increase our confidence in the validity of the outsourcing dyads (customer-

supplier relationships) in our dataset, we include only those dyads that appeared in at least two of

these data sources.

Identification of valid outsourcing dyads is further complicated by the frequent occurrence of

mergers and acquisitions in the mobile telecommunications and electronic manufacturing service

industries during the period of study. To ensure that supply relationships identified in our data

are in fact arrangements between independent firms, we documented the ownership history for

each firm using the Directory of Corporate Affiliations, Mergent Online, ISI Emerging Markets,

Orbis, and archived editions of the IT news website, Digitimes; outsourcing relationships where

the customer and supplier were joined by common ownership were omitted from the dataset.

Consistent with our focus on OEM suppliers‘ accumulation of capabilities, the unit of

observation in most of our empirical specifications is supplier-year, and we aggregate

outsourcing dyads to the firm level for each supplier in each year. For our indicators of supplier

technological and marketing capabilities, as well as firm and relationship characteristics that may

condition the extent of learning by supplying, we drew on data from several additional sources:

data on telecom-related patents come from Thomson Innovation‘s Derwent World Patents Index

(DWPI) database;10

financial data come from Compustat‘s Worldscope Global, ISI Emerging

Markets, Orbis, and Capital IQ; global sales data for branded handsets come from International

9 The following sources were included in the data search: Detectright, FCC Equipment Authorization

System, PDAdb.net, GSM Arena, GSM Choice, Phone Scoop, THT Business Research, World Cellular Handset Tracker, and World Cellular Information Service; see online appendix for detailed information on the scope of data coverage for each source. 10

To compile the patent data we downloaded all patents from DWPI with telecom-related EPI Manual Codes W01 (Telephone and Data Transmission Systems) and W02 (Broadcasting, Radio and Line Transmission Systems) for the period 1990-2011. We assigned patents to firms, first matching sample firms with all related subsidiaries listed in the Directory of Corporate Affiliations, and then matching these to patent assignees in the DWPI database. Finally, we collected assigned patents into ‗patent families,‘ (i.e., patents based on the same invention disclosed by a common inventor and patented in more than one country), to eliminate potential bias arising from differences in the scope of claims across technology classes (Alcacer and Gittelman, 2006). Application dates used in our analysis are based on the year of the first patent application within the family, i.e., the earliest priority year.

16

Data Corporation. Information collected from each of these sources was matched by hand to the

firms (both customers and suppliers) in our dataset.

The dataset emerging from this compilation process comprises observations on 113 unique

firms that supplied complete handsets to 154 branded producers or operators between 1995 and

2010; 13 of these supplier firms were subsequently dropped because they are also branded

producers and derived a significant portion of revenues from sales of their own branded handsets

throughout the period of study; as such, it would be unreasonable to attribute changes in

patenting or own-brand introduction and sales to learning by supplying for these firms.11

Table 1

shows the distribution of the remaining 100 supplier firms across countries, as well as the time

period in which they began supplying handsets. This table illustrates the shifting geography of

the supply base in this industry: operators and producers mainly outsourced handset production

to suppliers in Europe, North America and Japan in the early period, but when outsourcing took

off in the early 2000s, the center of gravity of the handset supply industry shifted towards

emerging markets in Asia. Thus China, South Korea and Taiwan together account for 76% of

new suppliers in the period since 2005.

While we cannot claim that our data are exhaustive and capture every significant handset

supply contract, we are confident that we have assembled the most comprehensive database of

outsourcing relationships to date in this industry, and that there are few major omissions. This

view is bolstered by our conversations with industry experts, who were unable to identify any

significant handset supply relationships that we had missed.

11

The dropped firms are Alcatel, Apple, Ericsson, Hewlett Packard, LG, Motorola, Nokia, Panasonic, Philips, RIM, Samsung, Sony and Sony Ericsson. Note that all of these firms are also important customers to OEM suppliers and, as such, are still accounted for in our final dataset in that capacity.

17

Methods

The goal of our empirical analysis is to examine the relationship between an OEM supplier‘s

accumulation of technological and marketing capabilities and the duration and extent of the

firm‘s cumulative supply experience (H1), the capabilities or sophistication of the customers

served (H2), the supplier‘s own pre-existing capabilities (H3), and the nature of the supply

relationships (H4). We use three proxies for supplier capabilities as dependent variables in our

main analysis – one related to technological capabilities and two related to marketing capabilities

– and all models share the following basic structure:

Supplier capabilities in year t are modeled as a function of experience, etc., up to year t-1.

Thus is the duration and extent of supply experience for supplier i up to year t-1,

captures the capabilities and other relevant characteristics of supplier i‘s customers up to year t-

1, captures the pre-existing capabilities and other characteristics of supplier i at year t-1,

and captures the types of supply relationships that supplier i has been involved in to year t-

1; are year and supplier firm fixed effects and is an error term.12

Because the nature of

the dependent variable varies across models, so does the particular estimation method used, as

detailed below.

Dependent Variables: Our primary measure of suppliers‘ technological capabilities is based on

patent data. PATENTSit, is defined as the number of ‗patent families‘ in successful telecom-

related US patent applications filed by supplieri, in year t.13

Since PATENTSit is a count variable,

we use fixed effect negative binomial estimation in models with this dependent variable.

12

The one exception to the inclusion of firm fixed effects is in the model for introduction of own-brand handsets. As noted below, this regression is estimated using a random effects logit model. 13

We obtain essentially identical results if we replace this annual patent count with a three-year moving average beginning in year t and if we use counts of individual patent applications in place of patent families. This is unsurprising, as average patents-per-family in telecom-related patent classes is quite low.

18

For our analysis of marketing capabilities we construct two firm-level proxies. The first,

OWN BRANDit is an indicator variable marking the first introduction of one or more own-brand

mobile handsets by supplier i. This variable takes a value of 0 in every year preceding the year in

which supplier i introduces its first own brand handset and changes to 1 in the year of

introduction.14

These models are estimated using random effects logit with year dummies. Our

second proxy for market capabilities is BRANDED SALESit, the total number of units of own-

brand handsets sold by supplier i in year t. These sales data, which come from International Data

Corporation (IDC, 2011), are available only for a subset of suppliers (34 firms in all) for the

years 2004-2010, and these are fixed effect linear regressions.

Independent Variables: Our first set of independent variables measures the duration and extent

of a focal firm‘s supply experience in the mobile telecom handset industry. SUPPLY TIMEit-1 is

our measure of supply duration, and indicates how long supplieri has been supplying mobile

telecom handsets (up to year t-1), i.e., the number of years from the first observation of a supply

relationship for supplieri in our data to year t-1. To measure the extent of the accumulated supply

experience we use CUM. CUSTOMERSit-1. This measure is based on the number of customers

supplied in each year by supplieri, cumulated to year t-1.

The second set of independent variables captures relevant characteristics of the customers

that a supplier serves. To capture the extent to which a supplier‘s customers can provide a

window on the technological frontier we include a variable CUSTOMER PATENTSit-1 equal to

the maximum number of patent families in the telecom-related patent applications of any one of

supplieri‘s customers in year t-1. We also create additional customer counts that distinguish

different types of customers: to capture the extent to which a supplier‘s customers provide a

window on the marketing frontier, we include the count variable CUM. LEADERSit-1 which is

14

In the results reported in Table 4a, the supplier exits the sample in years following the introduction of its first brand since this is the focal event for our analysis: in this way our analysis mimics the structure of a hazard rate model but does not require any assumptions about the baseline hazard rate.

19

the summation to year t-1 of supplier i‘s customers that are among the top five branded

producers in terms of market share in mobile telecom handsets in a given year;15

and to account

for potential differences in the knowledge-sharing incentives of branded producers and operators

we include CUM. OPERATORSit-1, defined as the summation to year t-1 of supplier i‘s

customers that are mobile telecom operators.

Our third set of independent variables addresses other relevant characteristics of supplier

firms: HAS PRIOR PATENTSit-1 is a dummy variable equal to 1 if the supplier has successfully

applied for one or more telecom-related patents prior to year t.16

To account for the effect of

changes over time in firm size, revenues and R&D investments, we include LOG ASSETSit-1,

LOG SALESit-1, LOG R&Dit-1, and LOG FIRM AGE it-1 in all model specifications.17

Finally, in some specifications we partition the cumulative customer count according to

whether the supplier‘s agreement with a given customer in a given year is an OEM agreement,

i.e., manufacturing only, or an ODM agreement (design and manufacturing).18

Similar to our

basic customer count, these variables, CUM. OEMit-1 and CUM.ODMit-1 are counts of the

number of agreements of each type that a supplier has in each year, cumulated to year t-1.

Summary statistics for all of these variables are shown in Table 2. Panel A provides statistics

for the full sample of 100 suppliers; Panel B gives comparable statistics for the subset of 83

suppliers for which we have complete financial data and are thus included in the regression

analysis reported below.19

As these summary statistics indicate, suppliers with missing financial

15

Market share data comes from Gartner group‘s ―Mobile device – markets share‖ reports for 2000-2011. 16

We get essentially identical results in models where this variable is replaced by a lagged dependent variable (the number of patent families in supplieri‘s patent applications in year t-1). As noted by Nickell (1981), however, inclusion of a lagged dependent variable can result in inconsistent estimates in fixed effects panel regressions; our use of a dummy variable effectively avoids this problem. 17

Financial variables are dollar-denominated logged values. 18

There are a very small number of design-only agreements in our data, which we combine with ODM agreements in the results reported here. Breaking out these agreements as a separate category produces essentially identical results. 19

Missing data on own-brand introductions and own-brand sales further restrict the samples used in the regressions reported in Tables 4 and 5, as explained below. The inclusion of firm fixed effects causes additional observations to drop out of each regression for those suppliers where there is no variation in the

20

data tend to be slightly younger, less experienced, and hold fewer patents than the industry

average.

To get a first look at the extent of learning by supplying in the mobile handset industry we

examine the evolution of capabilities for the suppliers in our sample as a group. Of the 100

suppliers in the full sample, about half (51 out of 100) held at least one telecom-related patent at

the time that they first entered a handset supply agreement; the median number of telecom-

related patents held by these 51 firms was seven. Six firms also introduced an own-brand handset

from the outset. As one would expect, these were all diversifying consumer electronics firms

(Japan‘s JRC, Kokusai, Mitsubishi, NEC and Oki, plus the German company Siemens). Thus,

from the beginning, some suppliers entered the industry with significant technological and

marketing resources. By the end of the period, supplier capabilities had nonetheless grown

significantly: 75% of the firms in the sample had successfully applied for at least one telecom-

related patent by 2010, and the median number of patents held among patenting firms had

increased to 20. In addition, over 60% of the firms had introduced one or more own-brand

handsets onto the market. Average sales of these own-brand handsets nonetheless remained

relatively low, and very few OEM suppliers could be said to pose a serious competitive threat to

leading producers: among the 34 suppliers for which we have sales data, average cumulative

sales of own-brand handsets to 2010 amounted to 23 million units, compared with an average of

451 million units by the leading branded producers. Given that the suppliers with sales data tend

to be among the largest and most successful firms in our dataset, this suggests that in general

suppliers have found it quite difficult to build the marketing capabilities necessary to compete

independently in the telecom handset industry. There is nonetheless significant variation among

relevant dependent variable (e.g., in Table 3 for suppliers that do not patent at any time during the sample period). Note, however, that re-estimation of the models using the entire sample and without financial data and firm fixed effects produces materially identical results for all of the variables of interest. See note 23 for more on this.

21

the firms in our sample in terms of both patenting and own-brand success, as the summary

statistics in Table 2 indicate.

Results

The estimation results in Table 3 explore the relationship between supplying and the

accumulation of technological capabilities as evidenced by changes in supplier patenting. The

dependent variable in these negative binomial regressions is PATENTS.20

All models include

firm and year fixed effects which reduces the number of supplier firms in these estimations to 67,

because suppliers that never patent are dropped from the analysis.

Models 1 and 2 explore the basic relationship between the duration and extent of supplying

activity and the accumulation of technological capabilities (H1). These results indicate that it is

not merely the duration, but rather the extent of supply activity that seems to matter for supplier

learning: the coefficient on SUPPLY TIME is insignificant while CUM. CUSTOMERS is

positive and significant in all specifications. Among the variables capturing relevant supplier

characteristics we observe positive and significant coefficients on LOG SALES and HAS PRIOR

PATENTS, consistent with absorptive capacity arguments (H3); these effects are consistent

across specifications. Coefficients estimates for logged values of assets, R&D spending, and firm

age are insignificant in Models 1 and 2, and are at best only marginally significant in most other

specifications.21

When it comes to the type of customers served by the supplier (Models 3-5) we see some

particularly interesting results. Model 3 introduces our measure of the technological

sophistication of customers served by the supplier (CUSTOMER PATENTS) and subsequent

models add cumulative counts of market leaders (CUM. LEADERS, Model 4) and operators

(CUM. OPERATORS, Model 5). Here we see a positive association between the technological

20

For convenience, firm and time subscripts are omitted from variable names hereafter. 21

Re-running these estimations with an R&D stock variable (cumulated over 3 years) in place of R&D produced essentially identical results, as did a specification that replaced log R&D and log Sales with log R&D/Sales.

22

sophistication of a supplier‘s recent customers and the firms‘ own subsequent patenting activity,

suggesting that suppliers indeed benefit from exposure to the frontier technologies of their

customers (H2).22

And while it does not appear to matter whether the customers are market

leaders, it does matter if they are operators: CUM. OPERATORS carries a negative and

significant coefficient in all specifications, suggesting that technological learning is lower for

firms supplying to operators, all else equal.

In Model 6 we replace the cumulative customer count with the partitioned count, separately

cumulating OEM and ODM customers. Consistent with prior research (and our H4) these results

suggest that suppliers learn from their customers and build capabilities more rapidly when they

collaborate in design in addition to manufacturing: CUM. ODM carries a positive and significant

coefficient, while CUM. OEM is insignificant. Thus, it appears that ODM relationships are

indeed an important pathway to the accumulation of technological capabilities.

Tables 4a and 4b results explore the relationship between supplying and the accumulation of

marketing capabilities. The dependent variables in these regressions are OWN BRAND (Table

4a) and BRANDED SALES (Table 4b). These models mirror the specifications in the previous

table, and are particularly interesting when compared with the results on technological

capabilities. First, it does not appear that accumulation of technological capabilities is a

necessary or sufficient condition for successful introduction of own-brand products - the

coefficient on HAS PRIOR PATENTS is insignificant across specifications. Indeed, the only

consistently significant predictor of own-brand introductions is the supplier‘s cumulative

experience producing handsets for operators (CUM OPERATORS). Recall that this is opposite

to the observed effect of operator experience on technological capabilities, where supplying to

22

If we replace CUSTOMER PATENTS with a measure based on the average number of patent families held by supplier i‘s customers instead of the maximum, the estimation returns an insignificant coefficient on this variable, while all other coefficients remain essentially unchanged. This suggests that, as we suspected, the maximum customer patent holding better captures the ‗technological frontier‘ to which a supplier is exposed. Logging the CUSTOMER PATENTS variable does not materially change any of the reported results.

23

operators was associated with lower capability building. This result is nonetheless consistent

with the idea that operators by necessity involve their suppliers more deeply in product definition

and other customer-facing activities.

The direct effect of involvement in design activities is captured in Model 6, where we see a

positive and marginally significant coefficient on CUM ODM, the cumulative count of OEM

agreements, but the coefficient on CUM OPERATOR remains positive and significant,

suggesting that operators are more open in their supplier relationships even relative to branded

producers in ODM agreements with their suppliers.

The results for BRANDED SALES in Table 4b add further nuance to this inference of

differential access. Despite the small sample size in these regressions we see a strong positive

coefficient on CUM CUSTOMERS. There is no statistical difference in the effects for operators

or ‗peripheral‘ branded producers, but supplying to market leaders has a significant dampening

effect on sales of own-brand phones (MARKET LEADERS carries a strong negative and

consistently significant effect in Models 4-6). Taken together these results support the contention

that suppliers are better able to capture valuable market-related knowledge when they supply to

operators than when they supply to other branded producers, particularly if these branded

producers are market leaders. In other words, the extent of learning by supplying - and the ability

to capitalize on that learning - is shaped to a significant extent by who a firm supplies to, not just

on the extent of supply activity.

Robustness and Alternative Explanations

The empirical observations reported above are quite robust to a variety of alternative measures

and methods, as detailed in footnotes accompanying discussion of the results. However, one

major concern when drawing inferences from this or any other empirical study of the co-

evolution of resources and organization is endogenous matching and unobserved heterogeneity

(Hamilton and Nickerson, 2003). For example, if it is the case that branded manufacturers

systematically choose the ‗most capable‘ available supplier, then one might worry that these

24

suppliers tend to patent more, or sell more of their own-brand handsets, not because of their

supply activity per se, but as a natural consequence of the firm‘s particular (perhaps

unobservable) capabilities. By including supplier and year fixed effects in our regressions we are

able to rule out the simplest form of selection bias associated with potential endogenous

matching, i.e., that observed differences in the extent of learning by supplying are merely a

reflection of different starting points in supplier capabilities, since we are estimating within-firm

effects rather than cross-sectional differences.23

Inclusion of fixed effects is only a partial remedy for potential selection bias, however, since

it does not exclude the possibility that the learning curves of some firms (i.e., more ‗capable‘

firms) have steeper slopes. However, since we separately examine two types of learning –

technological and market learning – as well as different types of customers – market leaders,

other branded producers, and operators– our analysis speaks not only to the overall learning rate

of a supplier, but also to how different types of learning vary depending on the type of customer.

In particular, the contrasting patterns observed in Tables 3 and 4 point to the existence of quite

distinct development pathways for technological and marketing capabilities for suppliers in the

mobile handset industry: since supplying to operators appears to dampen technological learning

while enhancing the accumulation of marketing capabilities it cannot simply be the case that

operators are selecting firms that are somehow more able learners than those selected by branded

producers.

23

One drawback of including firm fixed effects in the regressions is that suppliers drop out of any regression where there is no variation in the dependent variable for that firm. As a consequence, the samples included in the different models reported above are not completely overlapping, raising the possibility that differences in estimated effects are in part driven by the inclusion or exclusion of particular firms from the analysis. To ensure that this is not the case we re-ran all of the models in Tables 3, 4a and 4b without firm fixed effects, and also without financial data, so recouping the dropped observations. This exercise (results not shown; available from the authors on request) produced essentially identical coefficient estimates for all our main variables of interest, suggesting that our results are not dependent on particular sample restrictions.

25

The idea that different types of customers may foster different types of learning highlights a

final source of potential bias which we cannot mitigate entirely, however: potential assortative

matching between different types of buyers and suppliers. Here the concern is that learning

outcomes are influenced by the particular buyer-supplier matches observed in our data. Although

we control for as many observable differences in buyer and supplier characteristics as possible in

our regressions, some aspects of these matches may be unobservable, introducing the potential

for selection bias. In the estimation results reported in Table 3, for example, we saw that supplier

patenting was positively associated with suppliers‘ own prior patenting, and with customer

patenting. This begs the question of whether ‗high tech‘ buyers (those with a strong patent

portfolio) are matching with ‗high tech‘ suppliers (i.e. those that already hold telecom-related

patents), and whether this may in part be driving the observed results. To address this possibility

we perform additional analysis of individual customer-supplier dyads, to evaluate whether (and

to what extent) a supplier‘s technological portfolio becomes more similar to the portfolios of the

specific customer(s) it serves, relative to those of otherwise similar potential customers. This

allows us to evaluate whether the specific identity of the customer matters for supplier learning

by looking at the direction as well as the rate of change in supplier capabilities.

For this analysis we constructed a dyadic measure of the technological overlap (TECH

OVERLAPijt) between every potential customer-supplier pair and supplier-supplier pair in each

year of the sample period.24

The precise construction of this variable is as follows:

√( ) (

)

Here, Fit and Fit are patent class distribution vectors (within telecom-related patent classes) for

firms i and j respectively in year t (See Jaffe, 1986, for further details.) We then constructed a

24

Note that because construction of the TECH OVERLAP requires that both customer and supplier have patents our analysis is restricted to dyads where the supplier has at least one patent at the beginning of the supply agreement.

26

matched sample by associating each ―active‖ (observed) customer-supplier dyad with one or

more counterfactual (non-active) customer-supplier dyads using a coarsened exact matching

(CEM) method (Blackwell et al, 2009). To do this, for the supplier in each active dyad, we

match with one or more alternative suppliers based on the following criteria (measured at the

onset of the active supply relationship, t0); (i) the counterfactual suppliers are in our sample but

have never had nor do they ever have a supply relationship with the specific customer in the

active dyad, (ii) they are technologically close to the active supplier (i.e. high TECH OVERLAP

between the two suppliers); (iii) they are similar in terms of patenting levels (i.e. similar values

for PATENTS). We then pair these counterfactual suppliers with the customers in the relevant

active dyad over the duration of the active dyad (i.e. from t0 to tend), to create a sample of

equivalent non-active dyads.25

To evaluate the extent of technological convergence between customers and suppliers over

the course of a supply relationship, we first compare the average change in technological overlap

from t0 to tend for active customer-supplier dyads, with the change in overlap for customers and

(alternative) suppliers in non-active dyads over the same period. This gives us an estimated

average treatment effect, where ‗treatment‘ occurs over the course of a supply relationship. This

test reveals a small but statistically significant positive difference in means of 0.040 (std. error

0.016; unadjusted t-stat 2.64). Thus we see a greater increase in technological overlap over the

duration of observed supply relationships than for otherwise similar counterfactual customer-

supplier pairs over the same period.

To probe this result further we perform OLS regression on annual observations of TECH

OVERLAP over the period t0 to tend for each dyad in the pooled sample of active and non-active

dyads. This regression includes dyad and year fixed effects, allowing us to control for additional

25

The number of non-active dyads matched with each active dyad depends on the number of counterfactual suppliers that are in the same ―strata‖ as the active supplier, based on the automated cem data coarsening algorithm (Blackwell et al, 2009: 527).

27

time-varying supplier characteristics and, most importantly, to examine the relationship between

technological overlap and supply time: for active dyads, DYAD SUPPLY TIME is set to the

number of years from the onset of the supply relationship (t0) to year t; for non-active dyads,

DYAD SUPPLY TIME = 0 in all years. The results of this analysis are presented in Table 5. In

Model 1, we retain all of the matched non-active dyads in the analysis (i.e. 1-to-k matching) and

include weights in the regression to compensate for differential strata sizes. In Model 2 we

restrict the non-active sample to one per active dyad (i.e. 1-to-1). These results show quite

clearly that the technological overlap between a supplier and its customer increases with

relationship-specific experience – the coefficient on DYAD SUPPLY TIME is positive and

significant in both models. Together with the difference in means test this result indicates that a

supplier‘s technological portfolio indeed converges with that of its customer over the course of a

supply relationship, providing powerful additional evidence of learning by supplying.

While the analysis of dyadic technological overlap is useful in ruling out assortative

matching on technological capabilities, we are still left with the possibility that differential

supplier learning reflects assortative matching along other dimensions. As a final step we

therefore examine this issue directly by estimating supplier selection models, first for all buyers,

and then separately for different subsets of buyers (results shown in Table 6). These are

conditional logit models, estimated on a dyadic dataset comprising all potential buyer-supplier

dyads over the study period.26

The dependent variable in each case is ACTIVE DYAD, and

independent variables include the supplier characteristics used in the main regressions, plus a

dummy variables capturing whether the buyer and supplier are located in the same region, and

how long they have been working together. If there is indeed active assortative matching among

26

For the models shown a supplier is at risk of selection as soon as it comes into existence. Qualitatively similar results were obtained for a variety of different sample definitions, for example, when suppliers are considered at risk of selection only in a 3-5 year window around the observation of an active dyad.

28

buyers and suppliers then we would expect to see significant differences in the observed

selection criteria for different categories of customers.

These selection models generate some interesting results, but none that are indicative of

problematic assortative matching among different classes of buyers and suppliers. In Model 1,

using the full sample of dyads, the most significant positive predictors of an active dyad are

DYAD SUPPLY TIME and SAME REGION. Thus there is strong evidence of the perceived

benefits of proximity, and of continuity or path dependence in buyer-supplier relationships – the

probability that a supplier will serve a particular buyer in a given year is strongly related to the

length of time that the buyer and supplier have worked together up to that point. Buyers also

show a preference for younger, smaller and more R&D-intensive suppliers with high revenues,

all else equal. Surprisingly, supplier patenting is not significantly associated with selection.

Models 2-5 show similar estimation results for subsamples of buyers – operators only, branded

producers, market leaders, and high-tech buyers (i.e. buyers that are in the top 5% of patenters

among the firms in our data). While there are some small differences in these results – market

leaders seem to be less concerned with choosing a supplier located in the same region for

example – the general logic guiding supplier selection appears to be quite similar for all of the

different subsets of firms that we observe.

Model 6 replicates the selection analysis using just the first few years that we observe in our

data, leading up to the take-off in outsourcing around 2000: It is particularly interesting to see

that in these early years, dyad-specific experience and location appear to be even more salient

decision criteria – the coefficients on both variables is larger than in the estimations using other

subsamples, and most other variables carry insignificant coefficients. This seems to suggest that

buyers took a quite opportunistic and relatively haphazard approach to supplier selection in the

early days of outsourcing in the industry, usually confining their search to their immediate

geographic vicinity. When combined with the strong inertia apparent in these buyer-supplier

relationships, this again reinforces the inference that the patterns of capability development

29

observed in our data are not merely an artifact of endogenous selection or assortative matching

between buyers and suppliers in the industry, and indeed reflect a process of learning by

supplying.

Discussion and Conclusion

The findings reported above provide important new evidence on the extent of learning by

supplying in an industry that has witnessed significant offshore outsourcing in the last decade.

This is a first-order issue given continuing debates on the implications of outsourcing for the

migration of technological and market leadership. By highlighting contrasting patterns of

technological and market capability development for suppliers serving different types of

customers in the industry our results reveal an interesting and nuanced pattern of learning by

supplying. The evidence of technological learning is strong and unequivocal: for the suppliers in

our sample, patenting increases as a firm accumulates experience in handset supply relationships,

particularly when these involve both manufacturing and design of handsets. Our dyadic analysis

reinforces this inference, as the technological overlap between a buyer and a supplier increases

over time, as dyad-specific experience accumulates. Evidence on the accumulation of marketing

capabilities is more mixed: although most of the suppliers attempt to introduce their own branded

handsets at some point, few such introductions have generated large sales volumes. Thus it

appears that, contrary to the more alarmist commentary in the popular press, the progression

from trusted supplier to threatening competitor among electronics manufacturing firms is far

from inevitable.

Our findings also indicate that it matters a lot to whom you supply, although sometimes in

counterintuitive ways. For example, even though operators are more likely to delegate design

activities to their suppliers, these relationships generate relatively modest technological learning

for suppliers, relative to serving branded producers. When it comes to building marketing

capabilities, however, selling to operators is associated with a greater likelihood of own-brand

introductions, while supplying to market leaders strongly inhibits sales of own-brand products.

30

Thus, our results point to the existence of quite distinct development pathways for technological

and marketing capabilities for OEM suppliers in the mobile handset industry. Our findings also

reinforce anecdotal evidence that leading producers have responded to the threat of potential

competition from OEM suppliers by writing tight outsourcing agreements that can restrict the

ability of a supplier to sell own-brand phones, particularly in markets currently served by the

customer. This inference is also consistent with anecdotal evidence that leading producers take

these restrictive agreements seriously and react strongly to violations: When BenQ began selling

phones in China under its own brand name in 2004, for example, Motorola promptly pulled the

contract under which BenQ had previously designed and manufactured millions of handsets for

the company (Engardio and Einhorn, 2005).

Our post-hoc supplier selection analysis indicates that there are high perceived switching

costs and strong inertia in customer-supplier matches. As such, early supplier selection decisions

– even if they are based on relatively crude decision criteria - may have a strong and lasting

influence on suppliers‘ long-term capability development and market opportunities. One

important implication of these findings for OEM suppliers is that decisions regarding what types

of customers to pursue (operators versus branded producers; market leaders versus regional

players) should give consideration to the priority placed on the introduction of own-brand

products. If the ultimate goal of an OEM supplier is to develop as a branded producer in its own

right, then pursuing handset supply agreements with mobile operators may be the smarter choice;

suppliers working with leading branded producers may find themselves effectively locked into a

subordinate role, thwarting ambitions to move up the value curve to become viable independent

participants in the industry. More generally, firms should think carefully about both the means

and the motives for potential customers to share knowledge and support capability development

among their suppliers when investing resources in customer search or negotiating supply

agreements.

31

Our primary goal in this paper was to empirically document an important but under-explored

phenomenon. To this end we applied insights from existing theory in related areas, rather than

developing entirely new theory. We nonetheless believe that our study raises issues that are

worthy of consideration in future theoretical work aimed at understanding supplier relationships

in complex global supply chains. In particular, the divergent patterns of learning that we observe

with respect to technological and marketing capabilities speak to the different incentives for

knowledge sharing among supply chain participants in these two domains. Incorporating this

insight into future theoretical work on vertical spillovers in international supply chains is likely

to uncover important boundary conditions for prior results, and may generate additional

implications for the benefits of knowledge transfer from MNCs to local suppliers under different

conditions. This in turn has implications for the extent and nature of spillovers to the local

economy, so that further theoretical developments along these lines can usefully inform FDI

policy.

Our findings help put to rest the most apocalyptic view of offshore outsourcing and

capability migration, but they also temper the rosy characterization of buyer-supplier

relationships common in the supplier involvement literature and in some models of international

technology diffusion. We have argued that current outsourcing relationships diverge in

significant ways from the component supply agreements that were the subject of prior research in

these areas and, as a consequence, it is less clear whether customers are able to effectively

capture benefits generated by knowledge sharing with suppliers. Thus a natural complement to

our study of learning by supplying would be an exploration of how supplier learning impacts

customers‘ own capability development and/or market success. This is clearly beyond the scope

of the current study, but represents an exciting opportunity for future work.

Data availability and other limitations also constrain the reach of our findings and point to

opportunities for future research. For example, with fine-grained data at the subsidiary level, one

could provide a more complete treatment of learning by supplying, comparing the accumulation

32

of capabilities by independent OEM suppliers to that of in-house suppliers, i.e. subsidiaries of

the major branded producers. Another important opportunity for future research is to examine the

actual structures and processes adopted by outsourcing firms to facilitate – or impede – supplier

learning. Despite the relative richness of our data we are forced to rely on anecdotal evidence

from interviewees and industry commentators for our characterization of the knowledge-sharing

practices of different types of customers in the mobile telecom industry. This places clear

limitations on our ability to derive more nuanced managerial implications from the results, as

well as potentially limiting the generalizability of our findings. Extending our study into other

industry settings would also be useful to explore additional factors that condition the extent to

which customers perceive their OEM suppliers to be a credible competitive threat, and how this

impacts suppliers‘ ability to develop the capabilities necessary to climb up the value curve.

Despite these common limitations, our study generates practical insights for managers of

OEM suppliers as they attempt to move up the value curve in search of higher profits, and

provides a useful window on capability development in global supply chains. We believe that

our research also injects important new evidence into continuing debates surrounding offshore

outsourcing and the migration of capabilities and competitiveness from incumbent producers to

upstart suppliers. There is much work still to do, and we hope that future research building on the

current study will continue to contribute to strategic management and public policy in this

important area.

33

References

Alcacer JA, Gittelman M. 2006. How do I know what you know? Patent examiners and the

generation of patent citations. Review of Economics and Statistics 88(4): 774-779.

Arrow K. 1962. The economic implications of learning by doing. Review of Economic Studies

29: 155-173.

Athreye S, Cantwell J. 2007. Creating competition? Globalisation and the emergence of new

technology producers. Research Policy 36: 209-226.

Bartlett C, Ghoshal S. 2000. Going global: Lessons from late movers. Harvard Business Review

(March-April): 132-142.

Baum JAC, Ingram P. 1998. Survival-enhancing learning in the Manhattan hotel industry, 1898-

1980. Management Science 44: 996-1016.

Blackwell M, Iacus S, King G, Porro G. 2009. cem: Coarsened exact matching in Stata. The

Stata Journal 9 (4): 524-546.

Blalock, G, Gertler PJ. 2004. Learning by exporting revisited in a less developed setting. Journal

of Development Economics 75: 397-416.

Blalock G, Simon DH. 2009. Do all firms benefit equally from downstream FDI? The

moderating effect of local suppliers‘ capabilities on productivity gains. Journal of International

Business Studies 40: 1095-1112.

Clerides SK, Lach S, Tybout JR. 1998. Is learning by exporting important? Micro-dynamic

evidence from Colombia, Mexico and Morocco. Quarterly Journal of Economics 113(3): 903-

948.

Cohen WM, Levinthal DA. 1990. Absorptive capacity: A new perspective on learning and

innovation. Administrative Science Quarterly 35: 128-152.

Darr ED, Argote L, Epple D. 1995. The acquisition, transfer and depreciation of knowledge in

service organizations: productivity in franchises. Management Science 42: 1750-1762.

Duysters G, Jacob J, Lemmens C, Jintian Y. 2009, Internationalization and technological

catching up of emerging multinationals: A comparative case study of China‘s Haier group.

Industrial and Corporate Change 18(2): 325-349.

Engardio P, Einhorn B. 2005. Outsourcing innovation. Bloomberg BusinessWeek. March 21.

http://www.businessweek.com/magazine/content/05_12/b3925601.htm [accessed April 13 2012].

Grossman G, Helpman E.1993. Innovation and Growth in the Global Economy. MIT Press:

Cambridge MA.

Hamilton B, Nickerson J. 2003. Accounting for endogeneity in strategic management research.

Strategic Organization 1: 53-80

Hobday M. 1995. Innovation in East Asia: The Challenge to Japan. Edward Elgar: London.

Irwin D, Klenow PJ. 1994. Learning-by-doing spillovers in the semiconductor industry. Journal

of Political Economy 102: 1200-1227.

http://www.businessweek.com/magazine/content/05_12/b3925601.htm

34

Jaffe AB. 1986. Technological opportunity and spillovers of R&D: Evidence from firms‘

patents, profits, and market value. American Economic Review 76: 984-1001.

Javorcik BS. 2004. Does foreign direct investment increase the productivity of domestic firms?

In search of spillovers through backward linkages. American Economic Review 94: 605-627.

Khanna T, Palepu KG. 2006. Emerging giants: Building world-class companies in developing

countries. Harvard Business Review (October): 60-69.

Macduffie JP, Helper S. 1997. Creating lean suppliers: diffusing lean production through the

supply chain. California Management Review 39 (4): 118-151.

MacGarvie M. 2006. Do firms learn from international trade? Review of Economics and

Statistics 88 (1):46-60.

Mowery DC, Oxley JE, Silverman BS. 1996. Strategic alliances and inter-firm knowledge

transfer. Strategic Management Journal 17 (Winter): 77-91.

Nickell SJ. 1981. Biases in dynamic models with fixed effects. Econometrica 49(6): 1417-26.

Oxley JE, Sampson RC. 2004. The scope and governance of international R&D alliances.

Strategic Management Journal 25: 723-749.

Pack H, Saggi K, 2001. Vertical technology transfer via international outsourcing. Journal of

Development Economics 65(2): 389-415

Penner-Hahn J, Shaver JM. 2005. Does international research and development affect patent

output? An analysis of Japanese pharmaceutical firms. Strategic Management Journal 26(2):

121-140.

Pisano G, Shih W. 2009. Restoring American competitiveness. Harvard Business Review (July):

1-13.

Rapping L. 1965. Learning and World War II production functions. Review of Economics and

Statistics 47: 81-86

Rhee Y, Ross-Larson B, Pursell G. 1984. Korea’s Competitive Edge: Managing Entry into

World Markets. Johns Hopkins University Press: Baltimore.

Romer P. 1990. Endogenous technological change. Journal of Political Economy 98(5): S71-

S102.

Salomon R, Shaver JM. 2005. Learning by exporting: New insights from examining firm

innovation. Journal of Economics and Management Strategy 14 (2): 431-460.

Sampson RC. 2004. The cost of misaligned governance in R&D alliances. Journal of Law,

Economics and Organization 20(2): 484-526.

Takeishi A. 2001. Bridging inter and intra-firm boundaries: management of supplier involvement

in automobile development. Strategic Management Journal 22: 403-433.

Van Biesebroeck J. 2005. Exporting raises productivity in Sub-Saharan African manufacturing

firms. Journal of International Economics 67 (2): 373-391.

Yoffie D, Alcacer J, Kim R. 2012. HTC Corp. in 2012 Harvard Business Publishing. No. 712-

423

35

Table 1: Frequency of new suppliers by country-of-origin & year

1994-1999 2000-2004 2005-2009 Total firms

FINLAND 0 1 0 1

FRANCE 2 0 1 3

GERMANY 3 0 1 4

ITALY 1 0 0 1

SPAIN 0 0 1 1

SWEDEN 1 0 0 1

SWITZERLAND 0 1 0 1

UNITED KINGDOM 1 0 0 1

Europe 9 2 3 13

CANADA 0 3 0 3

USA 4 7 1 12

North America 4 10 1 15

CHINA 1 11 4 16

HONG KONG 0 1 1 2

JAPAN 16 0 0 16

MALAYSIA 0 1 0 1

SINGAPORE 0 1 1 2

SOUTH KOREA 2 10 1 13

TAIWAN 2 13 6 21

Asia 21 37 13 71

ISRAEL 0 1 0 1

Other 0 1 0 1

Total 34 50 17 100

36

Table 2: Summary statistics (supplier