The emergence of competence creating subsidiaries in UK pharmaceuticals in the 20th century
ABSTRACT
The evolutionary model of the MNE emphasizes that subsidiary specialization is the consequence of the transition from market- to efficiency-seeking FDI. But over 40% of subsidiaries in UK pharmaceuticals had acquired competence creating mandates by 1963, when market- seeking strategies were universal. The paper finds that none of the determinants of the rise of competence creativity in UK pharmaceuticals in recent years can explain what was at least as significant a rise between 1945 and 1963. The paper concludes that the evolution of competence creating subsidiaries is probably more dependent on subsidiary manager agency and path dependency that commonly allowed.
Word count (not including abstract): 13,916.
Keywords: History of FDI and the MNE; Historical Adaptation; Evolution of subsidiaries and headquarters; R&D; Historic FDI in UK pharmaceuticals.
1 What drives creativity? The emergence of competence creating subsidiaries in UK pharmaceuticals in the 20th century.
Introduction: From market-seeking to efficiency-seeking FDI and the emergence of competence creating subsidiaries.
One of the most significant developments in the last twenty years or more of scholarship in international business has been the recognition of how the overarching purpose of foreign direct investment (FDI) by multinational enterprises (MNEs) has switched from being market-seeking to efficiency-seeking, which has transformed the understanding of the role of subsidiaries and the organizational structures of MNEs (Dunning and Lundan, 2008, e.g. p. 190).1 In the earlier market-seeking stage of MNE evolution, MNEs typically had multi-domestic structures, where the ‘parent firm undertakes the first stage of production, for example, the R&D and design work for each of their products in the home country’, and then manufactures overseas (Dunning and Lundan 2008, p. 209 and fig. 7.1). The most explicit statement regarding the relative inferiority of subsidiary status in new product development came through Vernon’s product cycle theory, where subsidiary R&D was deemed unimportant and qualitatively restricted to adapting products for local markets (Vernon 1966). The Uppsala stage theory also assumes that ‘knowledge creation in foreign markets occurs at a late stage’ (Cantwell and Mudambi 2004, p. 40, Johansen and Vahnle 1977).
The switch to efficiency-seeking strategies was associated first, with much greater levels of FDI around the world, and then second, far greater specialization among subsidiaries as they utilized the endowments of their host economies. The subsidiaries best placed to exploit the most valuable knowledge-intensive local endowments were research and development (R&D) facilities, and so, in accordance with this emerging body of theory, the shift to efficiency-seeking investment has led both to a growing geographic dispersal of R&D functions generally, and the growing likelihood that some overseas R&D subsidiaries acquire sufficient capabilities that they are able to acquire greater levels of responsibility for product innovation from the parent company. From the late 1970s foreign affiliate expenditure on R&D increased rapidly, first in the US and then in Europe (Dunning and Narula 1995, Holm and Pederson 2000). The idea of the world product mandate, or competence creation mandate2, began to take shape in the early 1980s among a few leading MNEs (Dunning and Lundan 2008 p. 738, Dunning and Narula 1995 p. 69, Pearce and Papanastassiou 2009, Cantwell and Mudambi 2005), where those few subsidiaries that had displayed sufficient competence were given a charter to contribute to parent company strategic innovation.
2 By the end of the 1980s the globalization of R&D had become an unmistakeable aspect of this evolving strategic behavior (Feinberg and Gupta 2004). It is difficult to overemphasize its significance for the prevailing understanding of the role of MNEs in the world economy. A market-seeking environment understands MNEs essentially as institutions responsible for the transfer of certain kinds of technology, technology sufficiently complex or tacit that other forms of internationalization (such as licensing) failed (Buckley and Casson 1976). The current model, by contrast, no longer sees an MNE as a fairly passive child of market failure, but rather as a far more dynamic organization, increasingly able to combine parent company innovations with those emerging in different subsidiaries to produce wholly new and better products and services that can be sold in ever more locations around the world (Cantwell, 1995; Cantwell and Piscitello, 1999; Cantwell and Janne, 1999; Pearce, 1999).
As the transition from market-seeking to efficiency-seeking FDI has transformed the role of the subsidiary, so correspondingly it has also been the catalyst for a change in the structure of the MNE. Outside a few pioneers, multi-domestic or ethnocentric organisations were universal and, moreover, in a market-seeking environment subsidiaries were able to operate under some considerable autonomy with ‘minimal parental interference’ (Dunning and Lundan 2008, p. 187). After all their primary function was simply to place a given product into the host market. By contrast, in an efficiency-seeking environment, parents have to reconfigure the entire multinational organisation away from a series of dyadic HQ- subsidiary relationships into an increasingly differentiated network. Here the flows of knowledge are no longer necessarily from the centre to subsidiaries, but increasingly from subsidiaries to all other nodes on the network (Kuemmerle, 1999; Nohria and Ghoshal, 1994; Birkinshaw and Hood, 1998). By the early 2000s the large MNE was ‘evolving into… the role of an orchestrator of production… the decision-making nexus of the MNE has come to resemble the central nervous system of a much larger group of interdependent… activities’ (Dunning and Lundan, 2008, p. 739). The imperative for better internal co-ordination of resources and information has prompted many MNEs to become polycentric or geocentric organisations (Hedlund 1986).
Underpinning this explanation of the specialization and orchestration of subsidiary R&D activities around the world was a renewed emphasis on the importance of location advantages. Efficiency-seeking MNEs became increasingly aware of the potential advantages for new technology creation available to them in different locations around the world (Dunning 1998, Cantwell 2009), so encouraging subsidiaries to adapt, ‘tapping into local sources of competence’ (Ensign, Birkinshaw and Frost 2000, p. 149-50).
The central tenets of the evolutionary model of the MNE can thus be summarized as follows: a shift from market- to efficiency-seeking investment strategies precipitated a move to a much greater level of FDI
3 overall, with vastly increased specialization among subsidiaries, the most important of which was the emergence of competence creating subsidiaries. The increased specialization of subsidiaries and, for some, their growing contribution to parent new product development, in turn co- evolved with the emergence of new organizational forms, notably the move from ethnocentric to geocentric structures. The shift from market- to efficiency-seeking strategies is thus central to understanding the emergence of the modern MNE.
This shift can in turn be explained by the change in the wider global economic environment, where a series of institutional changes introduced the liberal world trade order increasingly from the late 1960s onwards which had the effect of increasing competition. This was made ‘possible by the reduction of barriers to trade between countries’ (Dunning and Lundan, 2008, p. 192). Pearce and Papanastassiou (2009) are even more explicit about this historical process. The multidomestic hierarchy with its centralization of R&D emerged because of tariffs and other restraints on trade from the 1930s onwards, reaching its ‚apotheosis… in perhaps the mid-1960s‛ (p. 5), and it was only after that subsidiary specialization began.
The key historical event here was the Kennedy Round of General Agreement on Tariffs and Trade (GATT), which began in 1964 and was the first multilateral trade agreement to remove any trade barriers after World War Two. The 1967 Kennedy Agreement was supported by other, near contemporaneous but less significant events (the European Economic Community formed a partial free-trade area among six nations in 1958, for instance), and was ultimately superceded by the more far-reaching removal of trade barriers in GATT’s Tokyo Round from 1980. But the opening of the Kennedy Round discussions meant that 1964 was when the era of de- globalisation began to draw to a close (Jones, 2005). So with the increasing removal of trade barriers initially after 1967, and then more clearly from 1980, global competition increased.
There has emerged therefore a consensus that the evolution of the MNE in the last thirty to forty years ,with its associated increase in the scale and specialization of subsidiary activities, and all co-ordinated by a geocentric organizational structure, has been driven by increasing levels of international competition. And while no single indicator of levels of competition exists that would cover such a global transformation of international business, the conventional proxy of looking at trends in international trade (as an indicator of an economy’s relative openness to international competition [Cantwell and Piscitello 2010]) gives empirical support to the association (WTO 2010, UN 1960). Moreover, substantial supporting evidence is found in the literature on subsidiary evolution, where survey respondents, for example, support the association of the emergence of subsidiary mandates with increased competition, the need to maintain competitive advantage, and so the importance of tapping into local sources of knowledge and benefiting from spillovers.
4
Increasing global competition led to the emergence of integrated MNEs, which in turn enabled some subsidiaries to exploit their location’s comparative advantage in knowledge creation and emerge with competence creating mandates for their R&D activities (Cantwell and Mudambi, 2005, Table 2; Dunning and Lundan, 2008; Wilkins, 1970 and 1974). The evolutionary model has not yet reached the point where the relative contributions of changes in competiton, emerging organizational forms, and developing location advantages can be separately analyzed. But as this introduction has indicated, levels of competition do emerge as the principal determinant, and these in turn are largely determined by governance structures and tariff and non-tariff restrictions to international trade. Given that the world economy has moved from positions of being more or less liberal over the course of the twentieth century, it is reasonable to expect that the move towards efficiency seeking FDI and corresponding subsidiary specialization seen in recent years might be mirrored earlier in the century, before retrenchment prompted MNEs to move to market-seeking FDI and so away from subsidiary specialization from the 1930s to the early 1960s.
Of course it is important to be clear at the outset that competence creating- type innovativeness among subsidiaries does have a long history. There have been several notable episodes of subsidiaries developing crucially important innovations for the subsequent life cycle of the parent company. Singer’s dominance of the global sewing machine industry came from the creation of a novel sales system developed in its UK sales subsidiary. Shell was notable for relying on R&D that emerged from its US subsidiary. And it was the German subsidiaries of GM and Ford that first developed effective ‘compact’ models in the 1970s, for instance, to take only three of the most obvious examples (Godley, 2006; Casson and Godley, 2007; Köhler, 2010). What is not yet known, however, is whether these were isolated episodes or not, exceptions that proved the rule of competence-exploiting HQ- subsidiary relationships in a de-globalised world.
The empirical data to support the globalization-induced emergence of subsidiaries with competence creating mandates is weighted to surveys of subsidiaries in the North America, UK and Scandinavia in the 1990s (Mudambi, 1999; Pearce and Papanastassiou 2009; Ensign, Birkinshaw and Frost 2000; Davis 2000; Iwasa and Odagiri 2004). These surveys clearly show the anticipated specialization among subsidiaries and the acquisition among a minority of a competence creating or product mandate. Pearce and Papanastassiou have largely focused on conceptualising a range of product mandates, and their results show that large minorities of subsidiaries emerged with relatively weak mandates (where the subsidiary’s contribution to parent company product innovation was significant but relatively minor), and correspondingly smaller minorities with strong mandates. Further investment and development may, for an even smaller minority, lead to a more elevated status of Centre of
5 Excellence (Ensign, Birkinshaw and Frost 2000, Holm and Pedersen 2000). But such development is incremental and path dependent, building on the base of a subsidiary having earlier acquired some minimal level of competence creativity. The critical transition is therefore that from competence exploiting to competence creating status.
In the recent studies respondents universally indicated that the process of R&D specialization was increasing, thus giving some support to their contention that this was a historical process. In fact, further support emerges from surveys of historical inward direct investment in British manufacturing, notably from Bostock and Jones’ (1994) work (which itself built on Dunning, 1958 – see Appendix). This comprehensive survey of overseas entrants in British manufacturing showed that possessing any kind of R&D function among subsidiaries in British mechanical and electronic engineering was a rarity. In 1935, for example, only 11 out of 97 engineering subsidiaries had any kind of R&D activity in the UK, and none were engaged in anything other than adapting parent company technologies for the British market. Bostock and Jones’ data show a big increase in inward FDI into British engineering by 1963. But despite this increase in volume, there is no evidence of any increases either in the share of subsidiaries engaged in R&D, or any change in the nature of that subsidiary R&D. Among subsidiaries of foreign MNEs in UK engineering in 1963 R&D remained universally adaptive, in contrast to the 1980s and 1990s (Bostock and Jones, 1994; Appendix).
A further test of this apparent absence among subsidiaries in British engineering of any with sufficient creativity to be thought of as analogous to possessing a competence creating mandate comes from examining historic US patent data, compiled by NBER economists (Hall, Jaffe, and Trajtenberg, 2001). This source lists all patents granted in the US by company name from 1963 onwards. Not one single UK engineering subsidiary had a US patent assigned to it between 1963 and 1969, confirming that R&D activities among subsidiaries in UK engineering remained universally competence exploiting. By contrast by the early 1990s, Cantwell and Mudambi (2005) showed that 24% of subsidiaries in UK engineering had acquired US patents, so providing their empirical proof of the increase in the share of competence creating subsidiaries in an era of efficiency-seeking investments and changing organizational forms.
The evolutionary model therefore seems broadly to be supported by both global data and a simple examination of longitudinal trends in patenting among one relevant population of subsidiaries. But before this can be accepted as the standard explanation of the evolution of subsidiary R&D creativity, further tests are surely necessary. One obvious yet simple test would be to perform the same historical survey of UK subsidiary R&D activity but in a different sector. If the central propositions – that increasing global competition, the diffusion of geocentric structures, and the emergence of relevant location advantages were the determinants of
6 the emergence of competence creating mandates in the most advantageous locations – are valid, then they must either be valid in all cases or otherwise qualified. This paper therefore concentrates on patterns of R&D activities among subsidiaries in the UK over the course of the entire twentieth century, but in pharmaceuticals rather than in engineering. The pharmaceuticals sector has been selected deliberately. Pearce (1999) and Davis (2000) have recently shown that pharmaceuticals subsidiaries were among the most significant sector for subsidiary R&D specialization in the UK and Denmark. It seems reasonable to assume that in pharmaceuticals, the most knowledge-intensive sector of all, any such specialization will have occurred relatively early and so have left the greatest amount of historic evidence from which to draw conclusions.
Competence Creating Subsidiaries in British Pharmaceuticals in the twentieth century – empirical data.
The pharmaceuticals sector internationalised both relatively early and extensively. Remarkable breakthroughs in synthetic chemistry and biologicals in the 1890s allowed German firms to become the global leaders, with Swiss and French firms as followers (Bernschneider-Reif et al, 2002; Burhop, 2009; Kobrak, 2002; Foreman Peck 1995, Quirke 2009). Despite important advances among US and UK firms, the Germans, Swiss and French retained their technological lead through the 1930s and up to World War Two (Cantwell 1991; Corley, 2003; Corley and Godley, 2011; Slinn 2008; Quirke and Slinn 2010). That war acted as an enormous catalyst for technological development in the sector, with the leading US firms developing significant advantages in pharmaceutical manufacturing, the profits of which were ploughed back into R&D. Combined with the decimation of the German and French industries, this provided the platform for the internationalisation of the US industry in the 1950s, 1960s and 1970s (Athreye and Godley, 2009). By the 1970s foreign (largely US) MNEs dominated UK pharmaceutical output (Corley, 2003). During the 1980s and 1990s the increasing costs of R&D caused substantial restructuring, which in turn prompted dramatic growth in the amount of R&D conducted in UK pharmaceuticals overall, rising from £22 million in 1970 to £2,000 million in 1995, a more than tenfold real increase in expenditure. By the end of the 20th century the UK was clearly at the forefront of global pharmaceuticals R&D (ABPI 1992 and 2000).3 Despite the presence of some prominent indigenous firms, this increasing R&D focus in the UK pharmaceuticals was largely the outcome of efforts by the subsidiaries of foreign MNEs. There are sadly, however, no official or industry-level data that distinguish between indigenous and subsidiary unit R&D activities for the twentieth century (Corley 2003: 27).
In the absence of any reliable industry-wide sources, data on subsidiary level R&D activities here have been drawn from two sources. The most recent observations are drawn from a survey of subsidiaries’ R&D activities
7 (including UK pharmaceuticals) conducted in 1994 by Robert Pearce and Marina Papanastassiou (Pearce 1999; Pearce and Papanastassiou, 2009). Following the current convention among international business scholars (e.g. Rangan and Sengul, 2009), 1994 is considered to be sufficiently close to the end of the 20th century as to accurately capture the trends in creativity among subsidiaries in UK pharmaceuticals in the most recent period of reglobalisation to the present. For the earlier observations, data is drawn from the a series of projects by John Dunning, Geoffrey Jones and Andrew Godley on the historic population of subsidiaries of foreign multinationals in the UK, including pharmaceuticals (Bostock and Jones, 1994; Dunning, 1958; Godley, 1999; Godley, 2000; Godley, 2003; Jones and Bostock, 1996).
These successive research projects were mostly funded by successive grants from the UK Economic and Social Research Council, and so the resulting dataset has been through successive peer review scrutiny and post-award evaluation processes. Furthermore a succession of articles have been published in highly ranked journals (e.g. see Bostock and Jones 1994; Godley 2003), and the outcomes have been widely cited (e.g. Wilkins 2009, Dunning and Lundan 2008, Godley and Casson 2010). While none of the dataset authors argue that the historic populations of subsidiaries are complete, they do claim that any omissions are essentially insignificant. The population of historic subsidiaries is therefore sufficiently complete that any results from its analysis ought to be considered reliable. Indeed, when compared with more recent survey data, arising from response rates varying around the 30 percent level, there is a strong basis for presuming that the results from the historic population are likely to be inherently more reliable than the contemporary data.4 This does not mean that the historic dataset is perfect. While for many of the most important variables the historic data are comprehensive and robust, for some desirable variables – unit-level annual expenditure on R&D, for example - no data exist. Nevertheless the historic dataset offers easily sufficient evidence to subject the long term relationship between the emergence of competence creating subsidiaries and changes in the levels of competition, in the MNE organizational form, and in location advantages in host economies, to far greater scrutiny than has hitherto been the case. The analysis below follows a series of pair-wise comparisons of each of the proposed drivers of the change in subsidiary behavior and the changes in the share of competence creating subsidiaries in UK pharmaceuticals over the period.
Proposition One. Comparing the Emergence of Competence Creating Subsidiaries and levels of Competition in UK Pharmaceuticals in the Twentieth Century
The shift from market- to efficiency-seeking behavior of MNEs is in its simplest form largely explained by increasing levels of competition. The evolutionary model of the MNE is therefore essentially Darwinian, with
8 the forces of competition driving change. Empirical data apparently strongly support the association of increased competition leading to increasing levels of competence creativity among UK pharmaceuticals subsidiaries in recent years. By the 1990s Pearce reports that 80 percent of pharmaceuticals R&D subsidiaries were in both large and small ways contributing to parent company product development (Pearce 1999, pp 164-5, n. 26).
For the historic population of pharmaceuticals subsidiaries it is known that from 1907 onwards all pharmaceutical production subsidiaries in the UK needed invest in laboratory capacity to enable them to comply with basic quality control requirements; they all therefore conducted adaptive R&D from that point on. Assessments as to whether any subsidiary R&D contributed to new product development have been done initially through a close reading of the relevant business history and history of science literatures (see Appendix for sources). This represents an enormous body of relevant documentary material. There is no a priori reason why self- reporting survey data might lead to more or less exaggerated claims for subsidiary R&D significance than in the historic sources. In the end, as Cantwell and Mudambi make clear, assessing whether any specific subsidiary has acquired a competence creating mandate is essentially a judgment based on empirical information. But judgment needs to be verified. In the absence of a fully comprehensive historical record of all subsidiaries, the NBER database of historic patents has been consulted again (as with the comparison of historic subsidiaries in engineering above) in order to obtain some base measure of R&D creativity among the subsidiaries in British pharmaceuticals in 1963. Evidence of registering a US patent by a UK subsidiary provides the verification of at the very least the acquisition of a minimal threshold of R&D creativity to suggest that the unit was not simply a competence exploiting subsidiary.5
Method The initial proposition being investigated here is whether there was the expected positive correlation between changes in the share of the population of subsidiaries that had acquired competence creating type levels of R&D creativity and changes in the level of competition in the UK pharmaceuticals sector. At a global level the literature strongly associates these two phenomena, and further associates changes in the levels of competition with changes in the governance of international trade; increased liberalization leads to increased competition. Of course there were other determinants of levels of international trade, changes in the costs of transport being only the most obvious, but governance regimes are widely seen as the most important. The direction of change in liberalization has varied, breaking the 20th century into sub-periods according to the prevailing governance structure of international trade (Jones 2005).
9 Because the institutional structure of international trade, and hence international competition, differed so markedly from one sub-period to another, the most appropriate method is to treat each sub-period as a discrete event. Thus the method adopted here has been to conduct a series of event studies and to see whether the actual behavior of pharmaceutical subsidiaries was consistent with that expected over each successive event. The benefit of conducting repeated event studies according to Morck and Yeung, is that they are then ‘perhaps the most direct test of causality’ (2011, p. 48). A high, expected correlation between the ex ante predicted and the observed outcomes over successive events is strong evidence from which to infer causality.
Event studies have been used frequently by financial economists eager to understand investor expectations of firm value with respect to announcements about mergers and acquisitions, or legislative changes, or other events that might be supposed to have an impact on the value of the firm (Warren-Boulton and Dalkir 2001). The logic relevant to the current application is that if, as recent survey evidence indicates, MNE managers believe that the competitive environment will influence subsidiary specialization, and that changes in the global governance of international trade influence international competitiveness, then it follows that MNE managers would have been more likely to pursue strategies that would have led to an increase in the share of competence creating subsidiaries after an announcement of a global commitment to increased liberalization of world markets, and vice versa.
Financial economists wanting to examine the impact of some announcement on the value of a firm typically calculate the ex ante probability of an event associated price change and then compare this with the observed outcome. Given the current generality of the evolutionary model of the MNE and the responsiveness of MNE structures to competition, it would not be possible credibly to estimate any ex ante probabilities with any precision. But estimating the likely direction of change in subsidiary behavior under different conditions of competitiveness and comparing these with the actual outcomes during the different event windows is entirely reasonable. First, therefore it is important to get some credible index of competitiveness in UK pharmaceuticals over the course of the twentieth century and so identify the start and end points of each event window.
Drawing on the earlier discussion of the close correspondence between levels of openness and competition in an economy, perhaps the best single index of competition in UK pharmaceuticals would be to examine trends in UK pharmaceuticals trade. Tariff and non-tariff barriers have operated in UK and global pharmaceuticals at different times and at different levels of exclusion, so impacting levels of competition in the UK sector. Table 1 suggests that the UK pharmaceuticals sector was highly competitive both before and immediately World War 1, with levels of openness higher then
10 than at any time before or since. War time constraints led to much reduced openness between 1914-1918, and 1939-1945, but openness fell also between 1919-1929, and further between 1929-1939, the classic era of Depression and autarchy. The pharmaceuticals industry data show that outside the Great Depression and war, the forces of competition were at their most attenuated around 1960. This is consistent with the controversies surrounding the issue of ‘monopoly profits’ in the pharmaceuticals industry that erupted in the US and UK at that time (the Kefauver Senate hearings 1959-1963, and the UK Government’s Sainsbury Committee, 1965-1966). Levels of competition slowly increased by 1970, before subsequently accelerating to approach pre-World War 1 levels by 1990 (see Appendix Table A3).
As far as the UK pharmaceutical sector is concerned, it is therefore possible to divide the twentieth century into five separate sub-periods, or windows, each one differentiated both from the preceding and succeeding window in terms of trends in openness, or some major dislocation. Table 1 lists the twentieth century sub-periods in chronological order, highlighting the trends in openness in the pharmaceutical sector during the sub-period, and then the expected change in the share of competence creating subsidiaries in the period. The table then goes on to show the actual outcomes (Column 4).