Environment and Planning A 1999, volume 31, pages 575-596

Transaction cost analysis and the methodology of foreign-market entry-mode studies

R B McNaughton Department of Marketing, University of Otago, PO Box 56, Dunedin, New Zealand; e-mail: rmcnaughton@commerce.otago.ac.nz Received 20 May 1997; in revised form 25 October 1997

Abstract. The author continues the discussion of foreign-market entry-mode research begun by O'Farrell, Moffat, and Wood. These authors reviewed and compared the contributions of economic geographers and marketers to our understanding of the channel choices of business service firms. Their central argument was that extant research methods, especially those employed by marketers, suffer significant limitations which impede the advancement of knowledge about channel decision­making. They also implied that the dominant theoretical paradigm, transaction cost analysis (TCA), provides a poor explanation of observed behaviour. They made a significant contribution by bringing together the economic geography and marketing literature on this topic. However, they did not fully review the contributions of the TCA approach to entry-mode studies in marketing, and this may have biased their conclusions about the value of this theory. An alternative account of this literature is provided, along with a response to each of O'Farrell et al's criticisms.

Introduction O'Farrell, Moffat, and Wood (1995) reviewed the contributions made by marketers (1 )

and economic geographers to our understanding of the internationalisation of busi­ness-service firms. The importance of this topic is underscored by the rapid growth in international services trade (Daniels, 1993; Dunning, 1989; ECC, 1991; Enderwick, 1989), and the economic impacts of this activity (Davis and Hutton, 1991; Michalak and Fairbairn, 1993). The choice of entry mode is an important characteristic of inter­nationalisation both from firm (micro) and from policy (macro) perspectives. For the firm, entry-mode choice is a strategic issue as selection of the right channel is a key to successful internationalisation (Keegan, 1984; Root, 1987). From a regulatory perspec­tive, entry mode is associated with the level of involvement of foreign firms in host economies, the level of control which foreign firms have over local operations, and their level of impact on the local economy. Marketers have focused on firm-level considera­tions, asking what modes are used, what factors are associated with the use of particular modes, and how mode decisions are made. Economic geographers have focused on larger issues, providing descriptions of the organisational and spatial patterns of inter­national expansion, and considering economic impacts and regulatory implications.

There is considerable value in bringing together the approaches of marketers and economic geographers. In the preface of a recently published volume on business and geographic approaches to foreign direct investment, Green and McNaughton (1995, page xv) lament that "... the level of research interaction between economic geographers and those in economics and management disciplines is unfortunately low. The pattern (1) O'Farrell et al refer to the contributions of 'economists' and 'management scientists' rather than marketers. However, the studies they review all fall in the domain of marketing, with some overlap in the multidisciplinary field of international business. Transaction cost analysis was developed primarily by institutional economists, although contributions to this paradigm have come from organisational studies, law, and marketing. In the management-science approach mathematical models are applied to problems of channel design (see, for example, Moorthy, 1988; Rangan, 1987). This is not the body of literature to which O'Farrell et al refer, so I clarify by referring more generally to the discipline of marketing.

576 R B McNaughton

of research citations between disciplines reveals a 'one-way flap' where theoretical developments in economics and management disciplines are adopted and influence geographic research, but the reverse is seldom the case". By reviewing both marketing and geographic approaches, O'Farrell et al (1995) contributed to a sharing of ideas that has significant potential to aid our understanding of the internationalisation of business services. New research directions may come from the combination of the geographic approach (characterised by longitudinal and often global empirical descriptions of the internationalisation of particular sectors), with the marketing approach (which is theory driven and tests specific hypotheses usually with cross-sectional data). In a more general context, Cornish (1995) also called for greater recognition by economic geographers of the role marketing plays in the spatial expansion of firms (and implies the need for a greater intermingling of the interests of the two disciplines).

The promise of sharing approaches to understanding entry-mode choice is unlikely to be realised without further stimulus as O'Farrell et al (1995) did not position the studies that they reviewed in the broader marketing literature on channel choice. A more comprehensive review may increase the attention which economic geographers give to this rich body of literature. In particular, transaction cost analysis (TCA) has had a profound impact on thinking in many areas of social science research. It is argu­ably the dominant paradigm in marketing for the investigation of channel choices. Other scholars have recently concluded that new institutional approaches such as TCA have a significant role to play in the theoretical development of economic geography (for example, Conti, 1995, pages 66-72).

My goal in this paper is to present an in-depth review of the TCA-based entry-mode literature in marketing, and to respond to the methodological criticisms raised by O'Farrell et al (1995). The remainder of the paper is structured into three sections. In the first I review the development of the TCA-based marketing literature on entry-mode choice in an attempt to establish a broader perspective for understanding the particular studies reviewed by O'Farrell et al. In the next section I address each of the criticisms raised by O'Farrell et al (1995). In the final section I summarise the argu­ments, and present a vision of how both bodies of literature can benefit from increased interaction.

Channel-choice research The geographic literature O'Farrell et al (1995, pages 686-687) began by reviewing the geographic literature on the internationalisation of business services. It is not within the scope of this paper to extend this review. However, a few comments are necessary. O'Farrell et al's review of the geographic literature is limited to a review of the work of three UK-based researchers or research groups: Leyshon et al (1986; 1987a; 1987b), Daniels (1991), and Dicken (1992). The first two of these focus on the internationalisation of UK producer-services firms (especially accountancy firms); the last describes the global expansion of the banking industry. Thus, the review is (1) limited compared with the volume of geo­graphic research that has been produced on this topic, and (2) biased toward the research approach of UK geographers. This second point is interesting as the marketing literature that is reviewed is almost entirely produced by North American researchers. In marketing there are recognisable differences both in schools of thought and in methodological approach between researchers trained in Europe and North America. Thus, O'Farrell et al (1995) describe the contributions of UK economic geographers and North American marketers, which may confound attempts at comparison.

The primary conclusion of the review is that"... geographical research is distinguished by its careful analyses of the interregional and international expansion of complete

Transaction cost analysis and foreign-market entry 577

industries, such as accountancy and banking. This aggregate approach has permitted a greater understanding of the overall spatial pattern of internationalisation and the general tendencies in entry mode choice" (O'Farrell et al, 1995, page 687). A more extensive review would show that not all geographical research^on the expansion of producer services has been longitudinal, or at the level of complete industries. In fact, a good number of studies have been cross-sectional comparisons of the propensity to export between sectors, and/or are based on surveys (including a study by O'Farrell et al, 1996). See Cornish (1996, pages 1663-1668), and Bagchi-Sen and Sen (1997) for recent reviews of this literature.

Another important observation is that in the geographic literature empirical descrip­tion is often emphasised over theory development. For example, Dahm and Green (1995, page 197), after reviewing the literature on the internationalisation of banking from several disciplines, concluded of geographers' contributions that "Although the descriptive studies of dedicated electronic banking networks are helpful, the recitation of empirical evidence is not matched by a corresponding level of theory." This is an important observation because marketing research on internationalisation is theory driven, and thus has much to offer economic geographers.

O'Farrell et al (1995) discuss the marketing literature on entry mode only in the context of a critique of its methodology. This does little to position individual studies in the broader context of the literature, and may create a biased view. Thus, the remainder of this section is devoted to a description of the marketing literature on channel research, so that the methodological choices of these studies can be better understood.

The marketing literature Approaches to channel choice Physical distribution is one of the four 'Ps' identified with the marketing management paradigm (the other three being product, pricing, and promotion). Although always recognised as important, distribution has not attracted the same level either of research or of managerial interest as have the other 'Ps'. Cox and Schutte (1969) found distribu­tion to be one of the least managed areas of marketing, and Higby (1977) characterised the scholarly study of channels as 'limited'. However, firms are finding that traditional marketing strategies based on pricing, product differentiation, or advertising do not yield long-term sustainable advantages in global markets. Thus, channels have become a focus for export marketing strategy. The interest that firms are showing in their channels is mirrored by an increase in related research. However, Klein (1984, page 186) characterised the literature on export channels as "fragmentary and inade­quate", Munro and Beamish (1987, pages 316-317) commented on the lack of research on foreign distribution arrangements, and Chan (1991, page 56) asserted that "The most striking feature of the [marketing] literature on exporting is the relative dearth of research on distribution". The point is that the literature on mode choice, particularly in foreign markets and for service firms, is less developed than that on other topics of interest in the discipline, but is currently in a period of rapid development.

There are a number of approaches to the study of channel choice, some of which are prescriptive, and others that are normative. A review of the major approaches is found in Rosenbloom (1991), and in other texts on channel management. These include atheoretic 'rules of thumb', financial, management science, microeconomic, and behav­ioural approaches. Because of O'Farrell et al's (1995) focus, the concern here is the significant increase in channel research beginning in the early 1980s. This research stream is associated with the adoption of TCA by several researchers as a paradigm for evaluating channel choices.

578 R B McNaughton

The transaction cost approach The TCA approach frames channel design as a problem of interorganisational gover­nance, emphasising the fact that decisions about how to distribute a product or service are also implicitly decisions about how to organise the firm. The primary progenitors of the resulting research stream are Anderson and Weitz (1986), Dwyer and Oh (1988), and Heide and John (1988; 1992). Others have addressed the specific issue of inter­national market entry: for example, Anderson and Coughlan (1987), Anderson and Gatignon (1986), Gatignon and Anderson (1988), and Klein (1989). Recently, researchers have investigated the use of channels in foreign markets (see, for example, Bello and Lohtia, 1995; Klein et al, 1990), and (particularly relevant here) have considered the unique circumstances of service firms entering international markets (for example, Erramilli, 1991; Erramilli and D'Souza, 1995; Erramilli and Rao, 1993).

The modern development of TCA is largely attributable to Williamson (1975; 1979; 1981; 1983; 1985; 1991), although the genesis of this theory can be traced to the much earlier work of Coase (1937), Commons (1934), Knight (1921), and Simon (1955). Two main branches of transaction cost reasoning have emerged: (1) the measurement-costs approach, and (2) the rent-appropriation or hold-up approach (Hallwood, 1990). In the first the actual costs incurred in a transaction, for example, writing and monitoring a contract, are emphasised, whereas in the second the potential costs of an unforeseen contingency, usually negative behaviour on the part of a partner, are emphasised. These two approaches are related as the potential for negative behaviour increases measure­ment costs. The rent-appropriation school has had a greater influence on transaction cost reasoning in marketing, although most authors do not make a distinction, and refer both to measurement costs and to rent appropriation as sources of transaction costs.

Hallwood (1990) provides an excellent account of the development of transaction cost economics, and its various adaptations in several disciplines. Heide (1994) provides a recent overview of TCA from the marketing perspective, along with a comparison with resource-dependency theory (Pfeffer and Salancik, 1978), and relational contract­ing theory (MacNeil, 1978; 1980). Shelanski and Klein (1995) provide a review and extensive bibliography of empirical applications of transaction cost economics. In a seminal text on complex organisations, Perrow (1986) also provides an overview of TCA, placing it in the context of other economic theories of the organisation (for example, agency theory). Perrow also provides a critique that draws on sociological approaches to understanding organisational governance.

In its basic form, TCA sees the governance of economic exchange as a choice between the market and a hierarchical control structure. The choice between these two modes is governed by an assessment of their relative efficiency. The process of exchange gives rise to transaction costs that may cause market failure in the sense that the market is an efficient means of mediating exchange. The three aspects of exchange processes responsible for creating transaction costs are: (1) investment in transaction-specific assets, (2) external uncertainty, and (3) internal (or behavioural) uncertainty (Williamson, 1985).

Transaction-specific assets are those dedicated to a particular relationship, and are not easily redeployed. As these assets are unique, they must be safeguarded to mini­mise the risk of opportunistic exploitation (Williamson, 1985). The 'external' and 'internal' components of uncertainty are judged relative to a contractual relationship. External uncertainty is a property of the decisionmaking environment, and refers to the extent that relevant contingencies are too numerous or unpredictable to be anticipated in a contract. Thus, mechanisms must exist to allow adaptations to occur as an exchange relationship develops. Internal uncertainty relates to the problem of ascertaining con­tractual compliance and monitoring performance.

Transaction cost analysis and foreign-market entry 579

Williamson (1975) originally proposed that it is the interaction between asset specificity and external uncertainty which explains a firm's governance decision. How­ever, these factors have generally been treated as separate explanatory variables in subsequent empirical work. Klein et al (1990, page 199), for example, argue that in the context of foreign markets either transaction-specific assets or uncertainty are sufficient in themselves to result in market failure. Their rationale is that external uncertainty is already high in foreign markets, so any increase in asset specificity will increase transaction costs. Further, because external uncertainty is high, it influences transaction costs independent of the level of asset specificity. Klein et al found no statistical support for the inclusion of an interaction effect in their model.

The implications of transaction costs for forward integration into distribution stem from their tendency to be low in highly competitive markets, thereby providing little incentive to substitute internal organisation for market exchange. In contrast, when the market is unable to impose pricing or behavioural constraints, firms are expected to internalise transactions to reduce the cost of exchange. A limit on integration is imposed by the fact that organisations themselves are not perfect, and transaction costs also exist within corporate structures. These are called 'organisation' or 'bureaucratic' costs, and include investments in legal, administrative, and operating infrastructures (Davidson and McFetridge, 1985). A general way of looking at channel choice is that firms make decisions about the trade-offs between the benefits of integration, and the costs of the resource commitments to gain that control (Anderson and Gatignon, 1986).

Whereas Williamson (1975) originally proposed that hierarchical governance was the only means to control transaction costs, recent theoretical extensions include the idea that the desirable features of internal organisation can also be achieved within the context of interfirm relationships. Coughlan (1985), Jeuland and Shugan (1983), and McGuire and Staelin (1983), for example, argue that interfirm channel agreements can be structured so that they are virtually indistinguishable from integrated channels. Stinchcombe (1985) suggests that unilateral provisions can be included in contracts to give the functional equivalent of hierarchical authority, and even Williamson (1985; 1991) admits that bilateral trading relationships can be designed to minimise potential governance problems and costs.

A further consideration is that, although TCA necessarily emphasises transaction rather than production costs (the cost of physical distribution), the implication is that the objective is to minimise the sum of transaction and production costs in making integration decisions (John and Weitz, 1988; Williamson, 1985, pages 92-94). Thus, channel volume is also an important variable as the costs of control can be spread over a larger number of sales units when channel volumes are high. The inclusion of production costs in most transaction cost models creates a link with earlier micro-economic studies. In fact, the acceptance of TCA may in part be due to its logical extension of microeconomic and (to a lesser extent) behavioural analyses of channel selection. This point is particularly important because O'Farrell et al (1995) argue that the inability to consider production costs prevents TCA from being used to consider exporting as a foreign-market entry mode.

Empirical studies of entry mode TCA studies of entry mode are focused on three primary characteristics of market exchange: (1) asset specificity, (2) uncertainty, and (3) channel volume (John and Weitz, 1988, pages 340-343; Shelanski and Klein, 1995, page 339). Most researchers have found a positive relationship between channel integration and asset specificity (Anderson, 1985; Anderson and Coughlan, 1987; John and Weitz, 1988; Klein et al, 1990). John and Weitz (1988) found a positive relationship between external uncertainty and integration.

580 R B McNaughton

However, Klein et al (1990) report mixed results, with environmental volatility being positively related to the use of foreign sales subsidiaries, but not to the hierarchical option of serving the foreign market from a domestic location. Conversely, environ­mental diversity was found to be related to exporting directly from a domestic location, but not to the establishment of a foreign subsidiary. Anderson (1985) found no relation­ship between channel integration and market unpredictability, and Dwyer and Welsh (1985) found no relationship with market heterogeneity. Lilien (1979), and Klein et al (1990) found that channel volume is positively related to integration, whereas John and Weitz (1988) found only one of their two production-cost constructs to be significant, and Anderson(1985) found no significant relationship with production cost.

Many of the researchers who have used TCA to explain channel choice argue that it is necessary to adapt or extend the model to some extent to account for particular circumstances. Heide and John (1988, page 21), for example, comment that "middle range theoretical extensions ... are needed to enable TCA to address specific classes of situations not adequately addressed in the global specification". John and Weitz (1988) added motivational variables to the TCA framework to improve their explanation of sales-force compensation. Heide and John (1988) drew on resource dependence theory to clarify how small firms with limited resources safeguard transaction-specific invest­ments, and Heide and John (1992) extended TCA with relational norms to explain buyer control over suppliers.

Anderson and Coughlan (1987) developed a model of channel integration in foreign markets in which TCA variables were combined with nine additional explanatory variables: maturity of product, product-service level, product differentiation, legal restrictions, presence of an established integrated channel, relationship of the product to the company's core business, the importance of trade secrets, number of competitors with integrated channels, and the cultural similarity of the foreign country. The presence of transaction-specific assets was the most important determinant of integration; product differentiation, maturity, and cultural similarity were also positively associated. Kim and Daniels (1991) extended this mix of TCA and non-TCA variables, and again found transaction-specific assets to be the most important factor explaining the use of integrated channels. Sales size, the time the product has been on the market, and the experience of the firm were also found to be significantly related to channel integration.

Huang (1996) recently argued that models of foreign entry mode are not 'eclectic' enough, and tested a model that included ten variables grouped into transaction-specific, environmental, firm-specific, and organisational factors. Like previous studies, he found evidence of a strong relationship between entry mode and transaction-related costs. Some of the non-TCA variables were not significant, and only one (locational advantages) had a larger coefficient than transaction costs.

The preponderance of evidence is that TCA variables are able to explain a signifi­cant amount of the observed variance in channel choice. After reviewing the substantial transaction-cost literature, Shelanski and Klein (1995, page 335) concluded: "We believe the empirical literature, on the whole, is remarkably consistent with the predictions of TCE [transaction cost economics]—more so than is typically the case in economics". There is only one study in which transaction cost economics were found to be sub­stantially wanting in an empirical comparison of frameworks for explaining mode of foreign-market entry. Madhok (1993) compared the transaction cost (internalisation), competitive strategy, and organisational capability perspectives in their ability to explain the choice of subsidiary, or joint venture (nonequity agreements were excluded because of their few occurrences in the sample). The results of this study suggest that com­petitive strategy and organisational capability are critical to the entry decision, whereas transaction cost considerations are not. Madhok attributes this surprising finding to a

Transaction cost analysis and foreign-market entry 581

number of factors. Perhaps most important is his assertion (following Forsgren, 1989) that transaction costs may be most important for newly internationalising firms, but less so for experienced multinationals (which formed his sample). The argument is that the international nature of multinational enterprises (MNEs) mitigates firm-specific disadvalitages.^FuTthermore, for the large MNE, ̂ overall optimisation is^more impor­tant than is local efficiency in any particular set of transactions.

Transaction costs and service-firm entry mode With the above as background, the position in the literature of Erramilli and Rao's (1993) study of service-firm entry mode [which receives considerable attention from O'Farrell etal (1995)] can be better understood. Their article is the first attempt in the marketing literature to account for the particular circumstances of service firms enter­ing international markets. Previous entry-mode research focused almost exclusively on manufacturing firms (see Agarwal and Ramaswami, 1992, for a review of this litera­ture). This despite a clear recognition in the marketing literature of the differences between manufacturing and service firms, based on characteristics such as low capital intensity and inseparability (see, for example, Bowen et al, 1989; Larsson and Bowen, 1989). Erramilli and Rao argued that the assumption that market modes are the default choice for situations characterised by low asset specificity only holds true if the benefits of integration are related solely to a reduction in transaction costs, and the costs of integration are always high. As neither is necessarily the case for service firms, they relax the assumption that low-specificity transactions will automatically be handled by shared control channels. Instead, they posit that the choice is contingent on other factors which influence the relative costs and benefits of integration.

To test this proposition, Erramilli and Rao include several non-TCA variables (capital intensity, inseparability, country risk, cultural distance, and firm size) as mod­erators of the relationship between asset specificity and the propensity to use shared control modes. This model was tested on a sample of US service firms. Consistent with TCA, they found a negative relationship between asset specificity and use of a shared control mode. However, they also found evidence that this relationship is strengthened by lower levels of capital intensity, but weakened by higher levels. The preference for shared control modes when asset specificity is low, and full control modes when it is high is intensified by inseparability, country risk, and smaller firm size.

This study is significant as a test of the robustness of TCA arguments to the partic­ular circumstances of service firms. It contributes to Williamson's (1985, page 105) call for more specific detail in the application of transaction-cost theory. Erramilli and Rao (1993, page 33) conclude that TCA is "useful and universally applicable", but that it requires extensions to be effective as a general theory of entry mode. This is consistent with Boddewyn et al's (1986) view that existing paradigms should, with appropriate modifications, be able to account for the circumstances of service firms. Erramilli and Rao (1993, page 33) contend that "The conventional TCA approach actually represents a special case of this [their] paradigm, dealing with situations wherein internal organi­sation costs are high and non-TCA incentives to integrate are low."

Response to critique of methods Definition of entry-mode choice and levels of aggregation O'Farrell et al's (1995) specific criticisms of the methods used in entry-mode research are addressed in the following sections. The first problem concerns the definition of the dependent variable—channel choice. Specifically, which modes to include in an analysis, and the level of aggregation of the choices. O'Farrell et al argue that (1) some researchers have excluded certain modes from their analysis for the sake of expediency,

582 R B McNaughton

and (2) high levels of aggregation are used simply to serve the requirements of logistic regression and because TCA cannot explain complex mode choices (1995, pages 688-689). They assert that "A meaningful investigation of entry-mode choices demands the maximum degree of disaggregation in the specification of the dependent variable, ideally at the level of the modes actually used, without aggregation" (page 689).

The question of which modes to include is a matter of the specific research question being addressed, and the range of channels commonly used by firms in the sector(s) being observed. O'Farrell et al (1995) focus on Erramilli and Rao's (1993) decision to omit firms which serve foreign markets by exporting. Erramilli and Rao rationalise their decision by referring to Hennart's (1989) opinion that TCA is not appropriate for comparing exporting with foreign direct investment (FDD. The argu­ment is that the choice between exporting and FDI is based on production costs rather than on transaction costs, as it relates to production in different locations. The decision not to include exporting is appropriate given that Erramilli and Rao's model does not include a production-cost variable, and the role of moderators is emphasised in the relationship between asset specificity and the propensity to share control. The influence of moderators would likely be more difficult to detect with additional modes, and consequently a more disaggregate dependent variable.

A distinction between exporting and FDI (in terms of the properties of a distribu­tion channel) implies that a distinction between trade and investment can be made. O'Farrell et al argue (1995, page 684) that this is not the case for services (and is increasingly untrue in manufacturing sectors as well). The dimension that underlies channel choice is the level of control afforded by a channel (Gatignon and Anderson, 1988). Exporting can be placed on a conceptual continuum of control, falling somewhere between a foreign sales subsidiary, and an intermediate mode such as a joint venture (McNaughton, 1996). As discussed in the review above, the implication in most TCA models is that channels are chosen to minimise the sum of transaction and production costs. Klein et al (1990), for example, included exporting along with FDI modes in their analysis, and found strong support for a production-cost effect on channel choice. This effect existed not only between exporting and FDI modes, but also between hierarchical, intermediate, and market modes of FDI. Further, transaction costs were also found to discriminate between exporting and FDI modes. I reported a similar result in my investigation of the foreign market channels used by software firms (McNaughton, 1996).

Both transaction and production costs are important in explaining the choice of channels, including exporting. It would be useful, in fact, for future research to be focused more on the interaction of these two sets of costs. One result might be an improved understanding of the selection of intermediate modes such as joint ventures and various forms of alliances. A working hypothesis is that, if transaction costs are moderate to high and production costs are also high, a firm with limited resources (relative to the magnitude of production costs) may be prohibited from integrating, but will be motivated to quasi-integrate to the extent possible through partnership arrange­ments (Bello and Lohtia, 1995, page 84).

Researchers other than Erramilli and Rao have made different decisions about the inclusion and exclusion of particular modes. Empirical tests of internalisation theory, for example, are concentrated largely on the choice between market entry through wholly owned subsidiary or licensing (for example, Rugman, 1980). The reason for this is the focus of internalisation theory on explaining FDI, and hence the existence of multinational firms. Thus, interest is largely constrained to the choice between direct investment and licensing. Further, internalisation emphasises the transfer of know-how, rather than the physical distribution of products or services. In contrast, the marketing

Transaction cost analysis and foreign-market entry 583

literature is concerned with the transfer of products or services from producers to intermediaries or final consumers.

Statistical techniques Q'Earrell et al (1995Xnote_ that in most TCA-based^studiesoL entry mode, cross-sectional surveys have been used, and choice models calibrated with the aid of logistic regression. Although survey methods and analysis by means of logistic regression are popular, there is no one methodology prescribed by transaction cost economics. Indeed, Williamson (1985, page 104) describes eight types of microanalytic approach that have been used to 'test' TCA predictions empirically. In many cases, logistic regression (or multinomial logistic regression—MNL) is the appropriate technique because the depen­dent variable is noncontinuous. The channel-selection process involves a managerial choice between competing designs, and these are generally recognised by practitioners and researchers alike as falling into a limited number of descriptive categories.

Conceptually, channel categories can be ordered based on degree of ownership control (Chu and Anderson, 1992; Gatignon and Anderson, 1988). Thus, they can also be treated as ordinal data. An ordinal dependent variable can be analysed by means of ordered-response logistic regression (ORL). The primary advantage of ORL over MNL is parsimony, as only one set of parameters is estimated, reflecting the relationship between each independent variable and the ordering of the dependent variable (Ghu and Anderson, 1992). The MNL model, in contrast, requires that one category of the dependent variable be selected as the 'base mode'. A set of coefficients is estimated for each additional level of the dependent variable. These coefficients show the utilities associated with each level relative to the base mode.

Specification of multiple-entry decisions O'Farrell et al (1995) note that a major decision in entry-mode research is that of which entry decision(s) to sample (1995, pages 689-690). They discuss the examples of Erramilli (1991) who leaves this decision to the respondent, Erramilli and Rao (1993) who collected information on up to six entries selected by the respondent, and Kim and Hwang (1992) who collected information on a recent decision. These approaches reflect an attempt to overcome recall problems, but may introduce a bias toward more recent and successful entries.

Another approach is to collect information for the entry mode relating to the firm's most important export product in its largest foreign market. This approach was used by Klein et al (1990), and McNaughton (1996). O'Farrell et al (1996) collected data on the first foreign market entered, and the current largest foreign market. This emphasis on the largest market may be particularly relevant for smaller and medium-sized firms that have a limited number of product-market combinations.

In the case of my sample of Canadian software firms, the most important product-market combination accounted for an average of 56% of the firm's revenues, and in 76% of the cases it was the first product the firm had ever exported. This is obviously still a biased sampling of all possible product-market combinations, but does account for the bulk of many firms' foreign activity. Assuming limited resources, and that firms do not divide their resources evenly between markets or products, the largest market likely consumes the bulk of foreign-marketing resources. Thus, a sample of largest markets may produce a bias toward finding integrated entry modes as more resources are available to increase control. Bell (1995) attempted to overcome this problem by collecting data on the first foreign market entered, the largest current market, and entry into top target markets for his sample of Finnish, Irish, and Norwegian software firms. He also tracked the order of market entry, and the entry mode used, allowing him to build a more dynamic picture of channel choice.

584 R B McNaughton

The question of which entry-mode decision(s) to sample has no easy answer. Researchers must make decisions based on the specific research question being addressed. O'Farrell et al (1995) suggest that researchers limit themselves to a partic­ular decision (for example, the first foreign market entered) so that all firms provide information on a comparable decision. They note that this might confine studies to firms in the early stages of internationalisation to ensure the availability of informants who were involved in the decision. This suggestion is consistent with the trend in the literature of examining the internationalisation of smaller firms, and firms in sectors (such as software) that have internationalised comparatively recently. Another approach would be to move beyond the survey methodology that is pervasive in this field, and explore channel decisions in experimental settings. For example, decisionmaking elements might be elicited from experienced managers, and then an experiment could be designed in which those elements were manipulated to design a channel for a hypo­thetical product - market combination.

Mode choice and cooperative organisational networks O'Farrell et al (1995) argue that the extant literature has focused on characteristics of transactions and the mode of entry (in their terminology, the "destination end of the decision"), while neglecting the organisational environment at the "origin end". In particular, they suggest looking at the link between cooperative organisational networks and the internationalisation process. This suggestion is consistent with theoretical developments both in economic geography and in marketing in which greater emphasis is placed on the embeddedness of firms in their surrounding environment. Yeung (1994) reviewed the development and influence of the organisational-network concept in economic geography, and parallel developments in marketing are reviewed here.

The strongest advocates in marketing for a network approach are the European-based Industrial Marketing and Purchasing (IMP) group. This group's work has four distinguishable variants. Easton (1992) describes these lines of inquiry in terms of metaphors: networks as relationships; networks as structure; networks as processes; and networks as positions. 'Networks as relationships' are associated with the inter­action approach which emerged from the first pan-European IMP study (Hakansson, 1982). In the interaction approach the focus is on the dyadic relationships between firms, and the implication is that networks result from the aggregation of these relationships. 'Networks as structures' is a more holistic view, implying that networks exist to match heterogeneous resources and demands. From this point of view, net­works evolve in an ecological process which allows firms to transform resources to meet their needs. In the 'networks as position' approach the focus is on the location of firms relative to others in the network. It is closely associated with the work of Mattsson (1984; 1987). There is a concern for the role of the firm, its importance to other firms, and the strength of relationships with other firms. It provides a powerful language for describing and analysing the strategic position of a firm. In the fourth approach, 'networks as process', the focus is on the dynamic aspects of relationships in a network. This approach is concerned with questions such as how firms are coordinated in a network, how power is distributed between firms in a network, and how networks change over time. Together these approaches have created an influential paradigm for the analysis of a range of marketing and purchasing functions.

A number of IMP researchers have commented on the potential of network approaches to yield insight into internationalisation processes, particularly those of smaller firms that are reliant on relationships with others for their expansion (for example, Axelsson and Johanson, 1992; Bridgewater, 1992; Forsgren, 1989). A recent study by Coviello (1994) of the internationalisation of New Zealand software developers

Transaction cost analysis and foreign-market entry 585

is a good example of the application of network theory to the internationalisation of service firms. Coviello sought to characterise empirically the influence of linkages and network evolution on the international marketing activities of these small knowledge-intensive firms. She found that linkages with partners were often formed to 'outsource' marketnfelated activities, allowing^soffware firms to achieve^internatiorial^growth with­out a well-developed in-house marketing function. Although channel choice was not addressed explicitly, the implication is that resource constraints which prevent adequate investment in distribution functions may be overcome by developing relationships with partners. In a similar vein, Bell (1995) is in favour of a network interpretation of his finding that European software firms initiate exporting because of domestic linkages (especially with clients, that is, 'followership'), sectoral targeting, and industry trends.

In addition to industrial networks, there are two other bodies of literature in which the importance of the domestic organisational environment as a source of competitive advantage in foreign markets is promoted. The first is Porter's (1990) work in which the postulate is that competitive advantage is created and sustained through a highly localised process in which differences in national economic structures, values, cultures, institutions, and histories contribute to competitive success. In particular, Porter promotes the idea of 'clustering', in which related groups of successful firms and industries emerge to gain leading positions in the international market. These clusters function through a complex mix of competition, cooperation, and emulation. As Crocombe etal(1991, page 30) describe it, "The complex web of interactions within these clusters can provide a major source of competitive advantage throughout the entire economic system. Often such clusters are geographically concentrated, making the interactions closer and more dynamic."

The second body of literature concerns the concept of 'milieu', which is "... a particular form of network in which the spatial, cultural and psychological proximity between actors determine a local and specific mode of regulation" (Cova et al, 1995, page 300). Cova et al see the concept of milieu as offering a solution to the problem which IMP researchers often encounter, of delimiting network boundaries by focusing on the embeddedness of a network of relations in a well defined spatial framework or territory. This concept is related to that of 'industrial districts' which are special, largely extrametropolitan, areas in which certain types of firm cluster to gain advan­tage through cooperation, and economies of scale and scope (see Brusco, 1992; Cooke and Morgan, 1990; Grotz and Braun, 1993). This body of literature is well known to economic geographers, and was recently reviewed by Camagni (1995).

The literatures on industrial networks, Porterian clusters, milieu, and industrial districts have all served to increase the emphasis placed on the organisational environ­ment at the 'origin end'. Well-known and widely tested theories of internalisation, such as Dunning's eclectic theory (1995), and Johanson and Vahlne's (1990) stages model, have been reinterpreted to take account of the influence of networks. There have also been several recent calls for further incorporation of network concepts into the mainstream of North American marketing research (for example, Thorelli, 1986; Webster, 1992). Further, 'network thinking' has had an impact on the industrial policies of several countries, including Denmark, Canada, Australia, and New Zealand, which have implemented programmes to foster the development of networks as 'launch pads' for export activity (Harper, 1993).

Most important to the discussion here is whether network theory can appreciably enhance our understanding of entry-mode choice (particularly in comparison with TCA). From a network perspective, channel design cannot be separated from the overall process by which firms enter foreign markets and develop their foreign operations over long periods of time. Network research does not focus on channel design per se.

586 R B McNaughton

Instead, it is concerned with the number of business relationships developed with other firms in the foreign market, the actors responsible for developing those relationships, and the way in which existing relationships in domestic markets are utilised in the entry process (Axelsson and Johanson, 1992). Thus, the primary observation suggested by this approach is that the way in which a firm enters a foreign market will be conditioned by its existing relationships which offer contacts and help in developing new partners (Johanson and Mattson, 1988).

Thorelli (1986, page 37) sees networks as falling somewhere between markets and hierarchies, although he notes that Williamson (1975) would likely treat most networks as markets. The (superficial) similarity between TCA and the industrial networks paradigm (INP) prompted Johanson and Mattsson (1987) to compare the two perspec­tives. They identified a number of important differences in theoretical framework, problem orientation, basic assumptions with regard to human behaviour (opportunism versus trust) and the nature of relationships, and unit of analysis. They illustrated these differences with regard to the study of internationalisation, and showed how the two approaches can lead to different conclusions. In particular, TCA offers only hierarchi­cal integration as a means of safeguarding specific assets, whereas the INP suggests that multinational firms can use their network positions to externalise some of their activities without losing control of crucial assets. Overall, they conclude that the focus of TCA is in explaining the existence of a specific economic institution, the multi­national enterprise, whereas the INP is more concerned with the internationalisation process. Further, the deterministic tendencies of TCA make it less suitable for strategy analysis.

The essential question here is whether TCA can be extended to account for network organisational forms, and their influence on entry-mode choices. Williamson's concept of bilateral governance is similar to the notion of relationships in networks (Johanson and Mattsson, 1987, page 46), however, Williamson originally argued that bilateral governance is not a stable institutional form. More recently, Williamson (1991) described stable hybrid governance forms that have their own 'disciplined rationale', and which arise from coordinated adaptation to the economic environment. Others, such as Zajac and Olsen (1993), are working on broadening the TCA framework to account for joint value maximisation between partners, and the process by which partners create and claim value in a relationship. These developments may well lead to a transaction-cost theory of internationalisation in which the organisational net­work of firms relationships, both domestically and abroad, is considered.

Recent methodological approaches to measuring cultural distance and country risk O'Farrell et al (1995) discuss the concepts of cultural distance and country risk as they influence channel choice, and focus on the problem of measuring these concepts. Cultural distance is an explanatory variable that is associated with the internationalisa-tion-process model (Johanson and Widersheim-Paul, 1975; Johanson and Vahlne, 1977; Luostarinen, 1970; 1979). Following Cyert and March's (1963) conceptualisation of bounded rationality in decisionmaking, this model sees internationalisation as a learning sequence where an initially limited perception of alternatives and a selective search leads to a constrained choice regarding markets and entry modes. Proximity is impor­tant to the decision process as greater information is likely to be available for 'near' as opposed to Tar' markets. As experience is gained, more alternatives will be considered, and operations will gradually expand into more 'distant' countries. The implication for channel choice is that cultural distance encourages shared control modes because of increased information gathering and therefore integration costs (Gatignon and Anderson, 1988; Kogut and Singh, 1988). Similarly, firms are thought to be more likely

Transaction cost analysis and foreign-market entry 587

to enter high-risk countries with shared control modes because of the need for flexibility (and integrated modes are less flexible because of high switching costs) (Anderson and Gatignon, 1986).

Cultural distance and country risk are not primary explanatory variables in the TCA:approacHr"However71hey are^ihcludedby Erlralhilir^nd\Raol(1993) as potential moderators of asset specificity. They hypothesise that the inverse relationship between asset specificity and use of shared control modes grows stronger with increasing cultural distance between the home and host countries, and with increasing country risk. This hypothesis is not supported for cultural distance, and only weakly for country risk. O'Farrell et al (1995) express concern that Erramilli and Rao adopt the cultural-distance measures calculated by Kogut and Singh (1988), based on Hofstede's (1980) indices of power distance, uncertainty avoidance, individuality, and masculinity or femininity. The problem, as they see it, is that "... 'cultural distance' is not based upon the investor's actual perceptions of the cultural distance between the USA and potential host countries. It is assumed implicitly (the assumption is rarely explicit) that such constructs reflect investor's perceptions and can explain the individual behaviour of firms" (O'Farrell et al, 1995, page 691). A similar criticism is levelled at measures of country risk.

In most studies the assumption is not in fact (implicitly or explicitly) that cultural distance or country risk reflect investors' perceptions. Rather, the assumption is that these characteristics are exogenous and constant across firms for a given host country (Agarwal and Ramaswami, 1992, page 3). In the literature on export motivation a distinction is made between the external and internal factors which propel firms to increase their foreign commitments (Wiedersheim-Paul et al, 1978). The economic and cultural proximity of export markets are considered to be external factors (Bilkey, 1978; Czinkota and Johnston, 1983; Jain, 1987), along with levels of competition in domestic and foreign markets (Bilkey, 1978; Cooper and Kleinschmidt, 1985; Czinkota and Johnston, 1983), the receipt of unsolicited foreign orders (Bilkey, 1978; Bilkey andTesar, 1977; McNaughton, 1992; Reid, 1981), and export incentive programmes. Internal factors revolve around the characteristics of a firm's decisionmakers. Examples include managerial aspirations toward growth and profit (Cavusgil and Naor, 1987; Kizilback and Maile, 1977), innovativeness (Lee and Brasch, 1978; Thomas and Araujo, 1985), and international orientation and experience (Cheong and Chong, 1988; Cunningham and Spigel, 1971).

Cultural distance is a characteristic of the external environment to which firms must adapt. Some firms may have internationally experienced personnel, good inter­national networks, and/or very adaptable business procedures, and thus are not very sensitive to cultural distance (that is, they can easily adapt to cultural differences). Other firms may have colloquial personnel, poor overseas contacts, and/or inflexible procedures, and be very sensitive to cultural differences. Thus, managerial perceptions are better seen as moderators of the influence of exogenous considerations like cultural proximity and country risk. For example, internationally experienced managers may perceive less 'distance' than do inexperienced managers. An obvious problem, though, is the necessity to distinguish between the perceptions of individual managers, and the ability of the firm to act on those perceptions. A particular manager's perception of cultural distance may not be shared by others in the same firm, he or she may not have influence within the firm, or the firm's organisational abilities may not be consistent with the manager's perception.

Benito and Gripsrud (1995) recently compared three measures of the 'distance' between countries and its relationship to international experience and company resources. The three measures were Kogut and Singh's (1988) index, economic distance

588 R B McNaughton

(based on difference in gross national product per capita), and the air distance between capital cities. These three measures were found to be interrelated (r = 0.69,0.58, and 0.72, respectively), with the highest correlation between cultural and air distances. They found only weak support for a relationship between international experience and distance, and no support for a relationship between company resources and distance. The strongest relationship between experience and a distance measure is for air distance. Although Benito and Gripsrud do not comment on the significance of this, it would seem to suggest that 'distance' is a characteristic of the external environment which creates organisational difficulties and costs which firms can lessen through a learning process.

Scale development Many of the measures used in entry-mode studies are not directly observable, and are thus 'latent' variables that are measured by multiple-item scales. O'Farrell et al (1995, pages 692-694) provide a thorough description of the scale-development process as advocated by Churchill (1979). However, in their discussion they emphasise the statistical assessment of reliability, while deemphasising the assessment of validity. In addition to measuring reliability via Cronbach's a, exploratory-factor analysis (includ­ing all possible scales within a single analysis), followed by second-order factor analysis of individual scales, is often used to assess the internal consistency of scales. It is also common to assess discriminant validity through the use of confirmatory-factor analysis. The content validity of individual scale items is usually determined by focus groups or in-depth interviews with potential respondents, and by pretesting the scales.

Statistical technology cannot overcome all of the problems identified by O'Farrell et al (1995). Perhaps the most important of these is inadequate sampling of the domain of possible items relevant to a construct. However, this is more likely to be a problem with relatively large and ambiguously defined domains, such as those for the constructs cultural distance and country risk. In contrast, the constructs of interest in TCA are tightly defined in theory, and so have relatively small domains. This is particularly so as researchers are beginning to divide these domains into even smaller segments. For example, knowledge-based asset specificity is now typically measured separately from the specificity of physical assets (see, for example, Bello and Lohtia, 1995; McNaughton, 1996), and diversity is distinguished from volatility as a source of uncertainty (see, for example, Klein et al, 1990; McNaughton, 1996).

O'Farrell et al (1995, page 692) assert that "There are few established scales with proven psychometric properties to measure the constructs such as cultural distance or country risk, let alone others relevant to the analysis at firm level of decisions over internationalisation." Although this may be true with regard to cultural distance and country risk, in many studies the same items have been used to assess asset-specificity and uncertainty constructs. Independent assessments of their validity and reliability have suggested that they are acceptable measures of the constructs (for example, compare the scales used by Anderson, 1985; Bello and Lohtia, 1995; Gatignon and Anderson, 1988; Klein et al, 1990; and McNaughton, 1996). Most importantly, the fact that these constructs generally behave as predicted by TCA theory helps to establish their validity. As O'Farrell et al (1995, page 694) note, "... any process of statistical testing should be designed in relation to a clear theoretical framework ... if the constructs clearly reflect the theory involved, and the data measure the former effec­tively, then we need to take account of the statistical tests."

Having defended the use of existing scales, it would be remiss not to admit that further work on scales that measure TCA constructs is needed. Many of the existing scales exhibit only moderate reliability. [For example, ten of Anderson's (1985) thirteen scales have a coefficients below 0.7 and three of those are below 0.5.]

Transaction cost analysis and foreign-market entry 589

The effect of using scales with low reliability is to attenuate any relationship that exists between the constructs they measure. Thus, statistical tests in which unreliable scales are used are biased toward insignificance, and an effect must be powerful to be detected. Interestingly, uncovering significant transaction cost effects has not generally been a problem. TPFarrell et al (1995, page 694) note thatT"There are~Te^lms^riiufiiair behaviour where there are major difficulties in specifying theoretical constructs that can be tested statistically..." This is true of TCA because, as yet, no one has devised a way of directly measuring transaction costs. Until someone devises a creative way of doing so, this stream of research will continue to rely on multiple-item scales.

Isolating determinants of entry-mode decisions The final criticism raised by O'Farrell et al (1995) is that entry-mode research needs to go beyond scaling of attitudes toward theoretical constructs, and find out what factors are actually considered by managers when making an entry-mode decision. Implicit in this suggestion is that we need to understand better the decisionmaking process itself.

This recommendation is laudable as there is a paucity of research in which the factors that managers consider in their internationalisation decisions, or which charac­terisesthe channel-selection process are considered. Theories of channel choice generally imply a rational process [see chapter 6 in Rosenbloom (1991) for an example of the steps in a rational planning approach to channel choice]. TCA, for example, does not characterise the channel-decisionmaking process itself, but infers a rational approach as" firms' are thought to choose the optimal'form ofgovernance by "attempting to minimise transaction costs.

Despite the assumption that channel-decisionmaking processes are rational, there is relatively little evidence in the literature to suggest that channel selection is informed by extensive analyses (Maignan and Lukas, 1997). Ford et al (1984) found that firms rarely perform cost-benefit analyses of market-entry options. Indeed market research often follows the choice of channel rather than preceding it. Similarly, Kleinschmidt and Ross (1984, page 13), having interviewed managers in eighty-five high-technology firms, concluded that "... external information sources have less impact on export market decisions than internally collected data", and that a very small amount of time is spent collecting primary data.

One of the first studies in which process elements of channel decisions have been characterised is that of Calof (1993). In studying the mode-change decisions of Canadian manufacturing firms, he found that only 33% of respondents described their channel-decisionmaking processes as rational. Further, only 35% of the processes included a formal study, in half of the decisions no outside experts were consulted, and only 18% considered more than one mode. I found similar results in a study of the channel-decision processes of Canadian computer software firms (McNaughton, 1997). In my study, the proportion of decisions made with a rational approach, by means of formal studies, and consulting externally was even smaller. Calof also found a significant (although proportionately small) relationship between decision-process elements and channel performance, whereas I found no evidence for such a relation­ship. In a study of UK service firms, O'Farrell et al (1996) found that a minority of firms (18%) considered alternative entry modes, and only about half of those carried out any form of systematic comparison between them. Following the methodology outlined by O'Farrell et al (1995, pages 694-695), O'Farrell et al (1996, page 115) also ascertained that the most important factors influencing entry mode where (1) "To take advantage of orders from foreign country", (2) "To exploit UK based managerial expertise", and (3) "To maintain standards of service quality".

590 R B McNaughton

O'Farrell et al (1996, page 114) conclude that their evidence contrasts with "... research which has modelled entry mode choice by service companies on the assump­tion that firms do exercise choices between modes, making rational evaluations and decisions within something like a transaction cost framework". Calof cautions against this interpretation, arguing that managers' 'gut feel' may actually constitute a complex and rational process that draws on considerable experience. Many activities, such as consulting customers or other partners, are simply part of the ongoing process of acquiring industry experience, and not an activity specifically performed in relation to a channel decision. Further, in smaller firms [which constitute my sample and, to a considerable extent, O'Farrell et al's (1996) as well] few people are involved in the decisionmaking process. Aharoni (1966) demonstrated some time ago that the exten-siveness and complexity of a foreign-entry-decisionmaking process is dependent on who is entrusted with the guiding process. Thus, extensive planning processes may have more to do with the need for accountability in large firms than they do with enhancing the quality of decisions. Owner operators are free to make decisions without having to prepare formal plans to back them up.

Conclusions In this paper I have continued the discussion of service-firm foreign-market entry-mode research began by O'Farrell et al (1995). Their contribution was in comparing the contributions of economic geographers and marketers, and raising several criticisms of the methods used by marketing researchers, especially those following the transaction cost paradigm. O'Farrell et al (1995) did not position the few studies of service-firm entry mode in the broader marketing literature on channel choice. To balance their critique, in this paper I have reviewed the marketing literature in more detail, and addressed each of O'Farrell et al's (1995) criticisms.

Empirical tests of TCA have shown remarkably consistent results, often explaining a significant proportion of the variance in channel choices. Results showing an inverse relationship between asset specificity and channel integration are nearly universal, for example, whereas results with regard to production costs are mixed between studies. Marketers have only recently started to explore the robustness of TCA to the special circumstances of service firms, small firms, and firms in high-technology sectors. Initial results are encouraging, and suggest that TCA arguments can be extended to account for specific firm and/or market characteristics.

Entry-mode research is difficult, and O'Farrell et al (1995) raised important method­ological issues. However, when understood in the proper context, the methodological choices that most researchers have made are appropriate. Specifically: 1. The choice of modes to include in a study is based on theory, the specific research question, and the range of modes typically used in a sector. TCA can explain exporting as well as FDI, especially when a production-cost variable is included in the TCA model. 2. Channel choice is often treated as dichotomous because of an inherent interest (arising from theory) in full versus shared control modes. Further, the dichotomous case has not been examined in all studies, and entry mode has also been treated both as a multinomial choice, and as an ordered one. 3. The specification of multiple entry decisions is a matter of sampling from the possible product-market transaction sets of target firms. For service firms, many of which are smaller, the most important product-market combination often captures the majority of a firms' economic activity. 4. There are bodies of research within the marketing literature involved with the influence of the organisational environment on channel choice, most notably the

Transaction cost analysis and foreign-market entry 591

industrial network paradigm. In several important internationalisation models efforts have been made to incorporate the embeddedness of firms in networks of domestic and foreign relationships. 5. Cultural distance and country risk should be treated as exogenous variables. Man­agers' perceptions are best seen as moderators of the influence of such exogenous variables. 6. The use of multiple-item scales is a 'necessary evil', as a method has not been devised to measure transaction costs directly. TCA constructs are well defined in theory and have small sampling domains. 7. There have been some attempts to characterise the channel-selection process, but more research is needed. The fact that the decisionmaking process may be quick, and not based on an extensive evaluation of alternatives, is not in itself contradictory to transaction cost theory.

O'Farrell et al (1995) suggest that future research should be longitudinal, and focused more on individual decisionmakers. Economic geographers have produced longitudinal descriptions of the internationalisation process, although usually at the level of whole sectors. Firm-level TCA-based studies in marketing have largely been cross-sectional. However, Dahlstrom and Nygaard (1997) recently analysed longitudinal changes in transaction costs within franchised distribution channels, and there is considerable potential for applying this approach in other channel contexts. In respond­ing to O'Farrell etal's (1995) criticisms a number of research requirements have been identified. One is improvement in measures of transaction costs, and another is the interaction between transaction and production costs. Theoretical extensions of TCA to account for the organisational environment of firms are also needed. Finally, I suggest a move away from the survey methodology that is pervasive in this field. Channel-decisionmaking process elements might be elicited from managers by means of protocol analysis, and experiments could be designed in which managers manipulate that information to design a channel for a hypothetical product.

References Agarwal S, Ramaswami S N, 1992, "Choice of foreign market entry mode: impact of ownership,

location and internalization factors" Journal of International Business Studies 23 1 -27 Aharoni Y, 1966 The Foreign Investment Decision Process (Graduate School of Business

Administration, Harvard University, Boston, MA) Anderson E, 1985, "The sales person as outside agent or employee: a transaction cost analysis"

Marketing Science 4 234 - 254 Anderson E, Coughlan A, 1987, "International market entry and expansion via independent or

integrated channels of distribution" Journal of Marketing 5171 - 82 Anderson E, Gatignon H, 1986, "Modes of foreign entry: a transaction cost analysis and

propositions" Journal of International Business Studies 17 1 - 26 Anderson E, Weitz B, 1986, "Make or buy decisions: a framework for analysing vertical integration

issues in marketing" Sloan Management Review 27 3 -19 Axelsson B, Johanson J, 1992, "Foreign market entry—the textbook view vs the network view",

in Industrial Networks: A New View of Reality Eds B Axelsson, G Easton (Routledge, London) pp 218-234

Bagchi-Sen S, Sen J, 1997, "The current state of knowledge in international business in producer services" Environment and Planning A 29 1153 - 1174

Bell J, 1995, "The internationalization of small computer software firms: a further challenge to 'stage' theories" European Journal of Marketing 29(8) 60-75

Bello D C, Lohtia R, 1995, "Export channel design: the use of foreign distributors and agents" Journal of the Academy of Marketing Science 23(2) 83-93

Benito G R, Gripsrud G, 1995, "The internationalisation process approach to the location of foreign direct investments: an empirical analysis", in The Location of Foreign Direct Investment: Geographic and Business Approaches Eds M B Green, R B McNaughton (Avebury, Aldershot, Hants) pp 43-58

592 R B McNaughton

Bilkey W J, 1978, "An attempted integration of the literature on the export behaviour of firms" Journal of International Business Studies 9 (Spring/Summer) 33 - 46

Bilkey W J, Tesar G, 1977, "The export behaviour of smaller-sized Wisconsin manufacturing firms" Journal of International Business Studies 8 93 - 98

Boddewyn J J, Halbrich M B, Perry A C, 1986, "Service multinationals: conceptualisation, measurement and theory" Journal of International Business Studies 17 41 - 57

Bowen D E, Siehel C, Schneider B, 1989, "A framework for analyzing customer service orientation in manufacturing" Academy of Management Review 14(1) 75 - 95

Bridgewater S, 1992, "Informal networks as a vehicle of international market entry: future research directions", RP54, Warwick Business School, University of Warwick, Coventry

Brusco S, 1992, "Small firms and the provision of real services", in Industrial Districts and Local Economic Regeneration Eds F Pyke, W Sengenberger (International Institute for Labour Studies, Geneva) pp 177 - 196

Calof J L, 1993, "The mode choice and change decision process and its impact on international performance" International Business Review 2(1) 97 -120

Camagni R, 1995, "Global network and local milieu: towards a theory of economic space", in The Industrial Enterprise and Its Environment: Spatial Perspectives Eds S Conti, E J Malecki, P Oinas (Avebury, Aldershot, Hants) pp 195-216

Cavusgil S T, Naor J, 1987, "Firm and management characteristics as discriminators of export marketing activity" Journal of Business Research 15 221-235

ChanT S, 1991, "Export expansion process for electronics: a study of channel integration strategy" Journal of Global Marketing 4(4) 55 - 68

Cheong W K, Chong K W, 1988, "Export behaviour of small firms in Singapore" International Small Business Journal 6(2) 34-41

Chu W, Anderson E M, 1992, "Capturing ordinal properties of categorical dependent variables: a review with applications to modes of foreign entry" International Journal of Research in Marketing 9 149 -160

Churchill G A, 1979, "A paradigm for developing better measures of marketing constructs" Journal of Marketing Research 16 64 - 73

Coase R, 1937, "The nature of the firm" Economica 4 386-405 Commons J R, 1934 Institutional Economics (University of Wisconsin Press, Madison, WI) Conti S, 1995, "Four paradigms of the enterprise system", in The Industrial Enterprise and Its

Environment: Spatial Perspectives Eds S Conti, E J Malecki, P Oinas (Avebury, Aldershot, Hants) pp 59-89

Cooke P, Morgan K, 1990, "Learning through networking: regional innovation and the lessons of Baden-Wurttemburg", Regional Industrial Research Report 5, University of Wales College of Cardiff, Cardiff

Cooper R G, Kleinschmidt E J, 1985, "The impact of export strategy on export sales performance" Journal of International Business Studies 16 (Spring/Summer) 37 - 55

Cornish S L, 1995," 'Marketing matters': the function of markets and marketing in the growth of firms and industries" Progress in Human Geography 19 317-337

Cornish S L, 1996, "Marketing software products: the importance of 'being there' and the implications for business service exports" Environment and Planning A 28 1661 -1682

Coughlan A, 1985, "Competition and cooperation in marketing channel choice: theory and application" Marketing Science 4 110-129

Cova B, Mazet F, Salle R, 1995, "Milieu 2 - network 0: the concept of milieu in search of network boundaries", in Interaction, Relationships and Networks IMP Group Competitive Papers Volume 1 Eds P Turnbull, D Yorke, P Nande, 11th International Conference, Industrial Marketing and Purchasing, Manchester Federated School of Business and Management, Manchester, pp 299-324

Coviello N, 1994 Internationalizing the Entrepreneurial High Technology, Knowledge-Intensive Firm unpublished PhD dissertation, Department of Marketing, University of Auckland, Auckland

Cox R, Schutte T F, 1969, "A look at channel management", in Marketing Involvement in Society and the Economy Ed. P McDonald (American Marketing Association, Chicago, IL) pp 98 -105

Crocombe G T, Enright M J, Porter ME, 1991 Upgrading New Zealand's Competitive Advantage: The Porter Project on New Zealand (Oxford University Press, Oxford)

Cunningham M T, Spigel R 1,1971, "A study in successful exporting" British Journal of Marketing 5(1)2-12

Cyert R D, March J G, 1963 A Behavioral Theory of the Firm (Prentice-Hall, Englewood Cliffs, NJ)

Transaction cost analysis and foreign-market entry 593

Czinkota M R, Johnston W J, 1983, "Exporting: does sales volume make a difference?" Journal of International Business Studies 14 (Spring/'Summer) 147 -153

Dahlstrom R, Nygaard A, 1997, "A longitudinal analysis of transaction costs in franchised distribution channels", in Proceedings of the 26th EM AC Conference Volume 1 Warwick Business School, Coventry, May, pp 304-328 :. .". ."-': :'"'.'• •'"

DahmDM, Green M B, 1995,"Transnationalbranchbanking 1976 -1986: an empiricist's approach", in The Location of Foreign Direct Investment: Geographic andBusiness ApproachesEds M B Green, R B McNaughton (Avebury, Aldershot, Hants) pp 43-58

Daniels PW, 1991, "Producer services and the development of the space economy", in The Changing Geography of Advanced Producer Services Eds P W Daniels, F Moulaert (Belhaven Press, London) pp 135 -150

Daniels P W, 1993 Service Industries in the World Economy (Blackwell, Oxford) Davidson W H, McFetridge D G, 1985, "Key characteristics in the choice of international

technology transfer mode" Journal of International Business Studies 16 5 - 21 Davis H C, HuttonT A, 1991, "Producer services exports from the Vancouver metropolitan region"

Canadian Journal of Regional Science 14 371 - 389 Dicken P, 1992 Global'Shift: The Internationalisation of Economic Activity 2nd edition (Paul

Chapman, London) Dunning J H,1989, "Multinational enterprises and the growth of services: some conceptual and

theoretical issues" Services Industries Journal 9(1) 5-39 Dunning J H, 1995, "Reappraising the eclectic paradigm in an age of alliance capitalism" Journal of

International Business Studies 26 461 -491 Dwyer F R, Oh S, 1988, "A transaction cost perspective on vertical contractual structure and

interchannel competitive strategies" Journal of Marketing 52 (April) 21-34 Dwyer F R, Welsh M A , 1985, "Environmental relationships of the internal political economy of

marketing channels" Journal of Marketing Research 22 397-414 Easton G, 1992, "Industrial networks: a review", in Industrial Networks: A New View of Reality

Eds B Axelsson, G Easton (Routledge, London) pp 1-28 ECC, 1991 Employment in the Service Economy Economic Council of Canada (Canada

Communication Group, Ottawa) Enderwick P (Ed.), 1989 Multinational Service Firms (Routledge, London) Erramilli M K, 1991, "The experience factor in foreign market entry behavior of service firms"

Journal of International Business Studies 22 479-501 Erramilli M K, D'Souza D E, 1995, "Uncertainty and foreign direct investment: the role of

moderators" International Marketing Review 12(3) 47 - 60 Erramilli M K, Rao C P, 1993, "Service firms' international entry-mode choice: a modified

transaction-cost analysis approach" Journal of Marketing 57(3) 19 - 38 Ford D, Lawson A, Nicholls J F, 1984, "Developing international marketing through overseas sales

subsidiaries", in International Marketing Management Ed. E Kaynak (Praeger, New York) pp 136-154

Forsgren M, 1989 Managing the Internationalization Process: The Swedish Case (Routledge, London) Gatignon H, Anderson E, 1988, "The multinational corporation's degree of control over foreign

subsidiaries: an empirical test of a transaction cost explanation" Journal of Law, Economics, and Organization 4 305 - 336

Green M B, McNaughton R B, 1995 The Location of Foreign Direct Investment: Geographic and Business Approaches (Avebury, Aldershot, Hants)

Grotz R, Braun B, 1993, "Networks, milieux and individual firm strategies: empirical evidence of an innovative SME environment" Geografiska Annalerl5B 149 -162

Hakansson H, 1982, "An interaction approach", in International Marketing and Purchasing of Industrial Goods Ed. H Hakansson (John Wiley, Chichester, Sussex) pp 10 - 27

Hallwood C P, 1990 Transaction Costs and Trade Between Multinational Corporations: A Study of Offshore Oil Production (Unwin Hyman, Boston, MA)

Harper D A, 1993, "An analysis of interfirm networks", report prepared for the Ministry of Commerce, Contract No 584, New Zealand Institute of Economic Research, Wellington

Heide J B, 1994, "Interorganizational governance in marketing channels" Journal of Marketing 58 71-85

Heide J B, John G, 1988, "The role of dependence balancing in safeguarding transaction specific assets in conventional channels" Journal of Marketing 52 20-35

Heide J B, John G, 1992, "Do norms matter in marketing relationships?" Journal of Marketing 56 32-44

594 R B McNaughton

Hennart J F, 1989, "Can the 'new forms of investment' substitute for the 'old forms'? A transaction costs perspective" Journal of International Business Studies 20 211 - 234

Higby M, 1977, "An evaluation of alternative channels of distribution: an efficiency model", MSU Business Studies, Division of Research, Graduate School of Business Administration, Michigan State University, East Lansing, MI

Hofstede G, 1980 Culture's Consequences: International Differences in Work-related Kz/w&y (Sage, Beverly Hills, CA)

Huang H, 1996, "A re-examination of entry mode strategies for small firms from NICs: are we eclectic enough?", in Competitive Advantage Through Global Networks, Conference Proceedings Eds V Gray, V Llanes, Academy of International Business South-East Asia Regional Conference, 17 - 20 June; Executive Education Programme, University of Otago, Dunedin, pp 351-361

Jain C S, 1987 International Marketing Management (Kent Publishing, Boston, MA) Jeuland A, Shugan S M, 1983, "Managing channel profits" Marketing Science 2 239 - 272 Johanson J, Mattsson L G, 1987, "Interorganizational relations in industrial systems: a network

approach compared with a transaction cost approach" International Studies of Management Organization 17(1) 34-48

Johanson J, Mattsson L G, 1988, "Internationalization in industrial systems—a network approach", in Strategies in Global Competition Eds N Hood, J Yahlne (Croom Helm, New York) pp 369-379

Johanson J, Vahlne J, 1977, "The internationalisation of the firm- a model of knowledge development and increasing foreign commitments" Journal of International Business Studies 8 (Spring/Summer) 23-32

Johanson J, Yahlne J, 1990, "The mechanism of internationalisation" International Marketing Review 7(4) 11-24

Johanson J, Wiedersheim-Paul F, 1975, "The internationalization of the firm: four Swedish cases" Journal of Management Studies 12 303 - 322

John G, Weitz B, 1988, "Forward integration into distribution: empirical test of transaction cost analysis" Journal of Law, Economics, and Organization 4 337 - 354

Keegan W, 1984 Multinational Marketing Management 3rd edition (Prentice-Hall, Englewood Cliffs, NJ)

Kim J, Daniels J D, 1991, "Marketing channel decisions of foreign manufacturing subsidiaries in the US: the case of the metal and machinery industries" Management International Review 31123-138

Kim W C, Hwang P, 1992, "Global strategy and multinationals' entry mode choice" Journal of International Business Studies 23 29 - 52

Kizilbach A H, Maile C A, 1977, "Export marketing in a changing economic environment" Journal of Small Business Management 15(1) 1 - 6

Klein S, 1984, "Vertical integration of export channels", in Proceedings of the Annual ASAC Conference 5(3) 182-191

Klein S, 1989, "A transaction cost explanation of vertical control in international markets" Journal of the Academy of Marketing Science 17 253-260

Klein S, Frasier G, Roth Y, 1990, "A transaction cost analysis model of channel integration in international markets" Journal of Marketing Research 27 196-208

Kleinschmidt E J, Ross R E, 1984, "Export performance and foreign market information: relationships for small high technology firms" Journal of Small Business Canada 2(2) 8-23

Knight F H, 1921 Risk, Uncertainty and Profit (Houghton Mifflin, Boston, MA) Kogut B, Singh H, 1988, "The effect of national culture on the choice of entry mode" Journal of

International Business Studies 19 411 -432 Larsson R, Bowen D E, 1989, "Organisation and customer: managing design and coordination

of services" Academy of Management Review 14 213-233 Lee W Y, Brasch J J, 1978, "The adoption of export as an innovative strategy" Journal of International

Business Studies 9(1)85 -93 Leyshon A, Daniels P W, Thrift N, 1986, "UK producer services: the international dimension

producer services", WP1, Department of Geography, University of Liverpool, Liverpool, and Department of Geography, St David's University College, Lampeter

Leyshon A, Daniels P W, Thrift N, 1987a, "Large accountancy firms in the UK: operational adaptation and spatial development producer services", WP2, Department of Geography, University of Liverpool, Liverpool, and Department of Geography, St David's University College, Lampeter

Transaction cost analysis and foreign-market entry 595

Leyshon A, Daniels P W, Thrift N, 1987b,"Internationalisation of professional producer services: the case of large accountancy firms development producer services", WP3, Department of Geography, University of Liverpool, Liverpool, and Department of Geography, St David's University College, Lampeter

LilienG, 1979, "Advison2: modeling the marketing mixdecision for:industrial products" Mawagemew/ Science 25 191-204

Luostarinen R, 1970 Foreign Operations (Helsinki School of Economics, Helsinki) Luostarinen R, 1979 Internationalization of the Firm (Helsinki School of Economics, Helsinki) McGuire T W, Staelin R, 1983, "An industry equilibrium analysis of downstream vertical

integration" Marketing Science 2 161 -191 McNaughton R B, 1992, "The export behavior of small Canadian manufacturing firms" Professional

Geographer 44 Y70-180 McNaughton R B, 1996, "Foreign market channel integration decisions of Canadian computer

software firms" International Business Review 5(1) 23 - 52 McNaughton R B, 1997, "The foreign market entry mode and channel change decision processes

of small computer software firms", in Proceedings of the 26th EMAC Conference Volume 2, Warwick Business School, Coventry, May 863 - 875

MacNeil I R, 1978, "Contracts: adjustments of long-term economic relations under classical, neoclassical and relational contract law" Northwestern University Law Review 72 854-905

MacNeil I R, 1980 The New Social Contract (YBIQ University Press, New Haven, CT) Madhok A, 1993 Mode of Foreign Market Entry: An Integrative Study unpublished PhD thesis,

Faculty of Management, McGill University, Montreal Maignan I, Lukas B A, 1997, "Entry mode decisions: the role of managers' mental models" Journal of

Global Marketing 10(4) 1-22 Mattsson LG, 1984, "An application of a network approach to marketing: defending and changing

market positions", in Changing the Course of Marketing: Alternative Paradigms for Widening Marketing Theory Eds N Dholakia, J Arndt (JAI Press, Greenwich, CT) pp 194-225

Mattsson L G, 1987, "Management of strategic change in markets as networks perspective", in The Management of Strategic Change Ed. A Pettigrew (Basil Blackwell, Oxford) pp 234-256

Michalak W Z, Fairbairn K J, 1993, "The producer services in a peripheral economy" Canadian Journal of Regional Science 11 353 -372

Moorthy K S, 1988, "Strategic decentralization in channels" Marketing Science 7 335-355 Munro H J, Beamish P W, 1987, "Distribution methods and export performance", in Managing

Export Entry and Expansion: Concepts and Practice Eds P J Rosson, S Reid (Praeger, New York) pp 316-331

O'Farrell P N, Moffat L,Wood PA, 1995, "Internationalisation by business services: a methodo­logical critique of foreign-market entry-mode choice" Environment and Planning A 27 683-697

O'Farrell P N,Wood PA, Zheng J, 1996, "Internationalization of business services: an interregional analysis" Regional Studies 30 101 - 118

Perrow C, 1986 Complex Organizations: A Critical Essay 3rd edition (McGraw-Hill, New York) Pfeffer J, Salancik G, 1978 The External Control of Organizations (Harper and Row, New York) Porter M E, 1990 The Competitive Advantage of Nations (The Free Press, New York) Rangan V K, 1987, "The channel design decision: a model and an application" Marketing Science

6156-181 Reid S D, 1981, "The decision maker and export entry and expansion" Journal of International

Business Studies 12101-112 Root F, 1987 Entry Strategies for International Markets (Lexington Books, Lexington, MA) Rosenbloom B, 1991 Marketing Channels: A Management View 4th edition (The Dryden Press,

Chicago, IL) Rugman A M, 1980, "Internalization as a general theory of foreign direct investment: a reappraisal

of the literature" Weltwirtschafliches, Archiv 116 2 Shelanski H A, Klein P G, 1995, "Empirical research in transaction cost economics: a review and

assessment" The Journal of Law, Economics and Organisation 11 335 - 361 Simon H A, 1955, "A behavioral model of rational choice" Quarterly Journal of Economics 69

99-118 Stinchcombe A L, 1985, "Contracts as hierarchical documents", in Organization Theory and Project

ManagementEds A L Stinchcombe, C Heimer (Norwegian University Press, Oslo) pp 121-171 Thomas M J, Araujo L, 1985, "Theories of export behaviour: a critical analysis" European Journal of

Marketing 9(1) 42-53

596 R B McNaughton

Thorelli H B, 1986, "Networks: between markets and hierarchies" Strategic Management Journal 7 37-51

Webster F E, 1992, "The changing role of marketing in the corporation" Journal of Marketing 56 1-17

Wiedersheim-Paul F, Olson C, Welch L S, 1978, "Pre-export activity: the first step in international-isation" Journal of International Business Studies 9 (Spring) 47-58

Williamson O E, 1975 Markets and Hierarchies: Analysis and Antitrust Implications (The Free Press, New York)

Williamson O E, 1979, "Transaction cost economics: the governance of contractual relations" Journal of Law and Economics 22 233 - 262

Williamson O E, 1981, "The economics of organization: the transaction cost approach" American Journal of Sociology 87 548 - 577

Williamson O E, 1983, "Credible commitments: using hostages to support exchange" American Economic Review 73 519 - 540

Williamson O E, 1985 The Economic Institutions of Capitalism (The Free Press, New York) Williamson O E, 1991, "Comparative economic organization: the analysis of discrete structural

alternatives" Administrative Sciences Quarterly 30 June, 209-296 Yeung H W, 1994, "Critical review of geographical perspectives on business organisations and

the organisation of production: towards a network approach" Progress in Human Geography 18 460-490

Zajac E J, Olsen C P, 1993, "From transaction cost to transactional value analysis: implications for the study of interorganizational strategies" Journal of Management Studies 30 131 -145

p © 1999 a Pion publication printed in Great Britain