Outsourcing and structural change: shifting firm and sectoral boundaries

Montresor, Sandro; Vittucci Marzetti, Giuseppe Working Paper Outsourcing and structural change: shifting firm and sectoral boundaries Quaderni - Working Paper DSE, No. 566 Provided in Cooperation with: University of Bologna, Department of Economics Suggested Citation: Montresor, Sandro; Vittucci Marzetti, Giuseppe (2006) : Outsourcing and structural change: shifting firm and sectoral boundaries, Quaderni - Working Paper DSE, No. 566, Alma Mater Studiorum - Università di Bologna, Dipartimento di Scienze Economiche (DSE), Bologna, https://doi.org/10.6092/unibo/amsacta/4721 This Version is available at: https://hdl.handle.net/10419/159407 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/ Outsourcing and structural change: shifting firm and sectoral boundaries Sandro Montresor∗Giuseppe Vittucci Marzetti Abstract The paper aims at investigating the structural change implications of outsourcing. In trying to bridge the organizational/industrial and the sectoral/structural analysis of outsourcing, it discusses the rational and the methodological pros and cons of a “battery” of outsourcing measurements for structural change analysis. Their functioning is then illustrated through a concise application of them to the OECD area over the ’80s and the early ’90s. A combined used of them emerges as recommendable in checking for the role of outsourcing with respect to that of other structural change determinants. ∗Corresponding author. Address: Department of Economics, University of Bologna, Strada Maggiore 45, 40125 Bologna, Italy, Tel.: +39 051 2092651, Fax: +39 051 237002, E-mail: [email protected] Keywords: Outsourcing; Input-output analysis; Vertical integration; Manufacturing JEL Classifications: D230, D570, L160, L220, L240, L600, O140 1 1 Introduction Outsourcing has recently become a “hot-topic” of several economic disciplines dealing with the firm, such as industrial organization, labour microeconomics, industrial relations and operation management, just to mention a few.1Accordingly, cases for and against outsourcing are numerous and heterogeneous, depending on the specific approach which is adopted. This upsurging interest for outsourcing processes at the firm level has also had an important cross-disciplinary fertilization. For example, a “new” strand of trade theories has developed on the basis of its “fragmentation” effects (e.g. Kohler, 2004; Jones and Kierzkowski, 2001). Quite surprisingly, instead, the new wave of outsourcing studies has not been accompanied by an as enthusiastic revival of one of the economic fields which for first recognized its relevance: that is, structural change analysis. More than 20 years ago already, Momigliano and Siniscalco (1982b), among others at that time (e.g. Stanback, 1979; Ginzberg and Vojta, 1981; Gershuny and Miles, 1983), recognized that the externalization of production activities from manufacturing firms to specialized producers of business services represented, also and above all, a change in the relative weight of the sectors of an economic system, in terms of both production output and employment. But, despite these seminal contributions, the structural change implications of service outsourcing have remained since then relatively neglected. On the one hand, the increasing availability of firm micro-data and the extraordinary development of firm surveying techniques have made the analysis of its impact on the firm boundaries dominant with respect to that on the boundaries of manufacturing sectors. On the other hand, the reference to the “establishment” as the unit of analysis for building up input-output tables has marginalized its interpretative role of a phenomenon which is claimed to occur at the firm level. This is unfortunate, as the role of service outsourcing in explaining the structural change of economic systems is quite apparent. Indeed, the externalization of business services undertaken by manufacturing firms, while (or rather than) decreasing the industrialization degree of one economy, determines a reshaping of the sectoral boundaries between manufacturing and (business) services. However, quite recently, the increasing pervasiveness of outsourcing has spurred some researchers to reconsider the role that, along with technological change and changes in demand, the kind of organizational change entailed by outsourcing has had on the economic restructuring of developed countries (Dietrich, 1999; McCarthy and Anagnostou, 2004). More precisely, these studies have tried to “decompose” the changes occurred over time in input-output tables data, with the aim of disentangling the relative weight of demand-side and supply-side factors in driving economic restructuring. In fact, these studies generally conclude that the “deindustrialization” arguments that have been used, for example, in accounting for the economic restructuring of Europe from the ’70s to the ’90s, have largely overlooked the extent of outsourcing processes. In so doing, they add, conventional economic views would have underestimated the actual importance and contribution of manufacturing to GDP. In trying to get rid of such a bias, these studies attempt to bridge the 1For a critical survey of the different industrial approaches to national and international outsourcing see Spencer (2005). For a more general account of the issue in economics see instead Montresor et al. (2006). 2 industrial analysis of outsourcing with the intersectoral one of structural change. More precisely, they propose to refer to some sectoral “proxy” of outsourcing to be used along with sectoral proxies of other demand and supply factors of structural change, and to deal with them simultaneously. This paper tries to analyze this conceptual bridge from a methodological point of view. At the outset, it discusses its inspiring rational more in depth, along with its theoretical background (Section 2). The methodological problems which emerge in the bridging are then spelled out (Section 3). At first, a battery of indicators is surveyed which could be used to detect traces of outsourcing in what could be otherwise considered “simple” tertiarization (Section 4). Their functioning and interpretative power is then showed through an illustrative and concise empirical application to a set of OECD countries along the ’80s and the ’90s (Section 5). Section 6 concludes. 2 Theoretical background In organization economics outsourcing is usually dealt with under the “firmboundaries” agenda. Indeed, it is considered as a process symmetric to the integration one,2through which the “vertical scope” of the firm shrinks rather than enlarging. Either because some transactions are “moved” from the firm to the market governance mechanism – following transaction cost economics – or because monitoring costs are such to make external principal-agent relationships more effective – following incomplete contract theories – or because some activities are better carried out through external rather than internal resources and competences – following the resource-based view (Montresor et al., 2006). Associating outsourcing to just a shrink of the firm boundaries could however be contrasted by looking at the firm as an “open system”, whose boundaries integration and outsourcing would rather make, respectively, less and more “permeable” to other organizations and, more in general, to market mechanisms (Jacobides and Billinger, 2005). Indeed, some have argued that outsourcing, through the partnerships, collaborations and agreements which it often entails, would have stimulated the firms to become “extended enterprises” (McCarthy and Anagnostou, 2004) or, similarly, “network firms” (Antonelli, 1988). Sticking to one or the other interpretation has important implications on the nature of the shifts that outsourcing determines in the boundaries of those sectors in which the relevant firms operate. Using a more accurate jargon, outsourcing turns out to have in the two cases different structural change implications. The fact that, for example, a textile firm outsources its machinery maintenance to a specialised service firm, according to the first interpretation just alters, namely diminishes, its economic contribution to manufacturing and, in turn, the economic contribution of manufacturing to the economic system: putting it simply, service outsourcing would induce nothing but a “tertiarization” effect. Following the second interpretation, however, the same outsourcing 2For this reason, it is often referred to simply as “disintegration” or “externalization”. To be sure, in some economic disciplines, namely in business economics, outsourcing is distinguished from other externalization processes of the firm by looking at, for example, the nature and the characteristics of the underlying obligations, the strategic value of the assets involved, etc. In the present paper however, at the risk of being somehow inaccurate, we will treat externalization, disintegration and outsourcing as synonymous. 3 operation would rather entail a restructuring of the textile sector, and of manufacturing in general, following which its contribution to the economic system is actually changed in nature rather than simply diminished. As Momigliano and Siniscalco (1982b) put it, much of what is called tertiarization should be better called integration of services in manufacturing or, possibly, of some manufacturing activities in other manufacturing activities. In the light of the increasing resort that firms make to outsourcing strategies, capturing this particular kind of structural change, often amounting to an extension of manufacturing sectors into non-manufacturing sectors, has become extremely urgent. And as urgent has become the need of disentangling the role of outsourcing in explaining the deep structural changes that most of the developed economies have undergone over the last twenty years. The next sections of the paper aim to move in this direction, by debating some methodological issues (Section 3) and tools of analysis (Section 4), and by presenting some illustrative empirical evidences (Section 5). 3 Outsourcing and sectoral input-output relations In industrial organization, outsourcing is usually retained a process through which a certain firm “switches” from making a certain activity of its production process in-house to buying its outcome from an external contractor, typically another firm (Grossman and Helpman, 2002). Empirical evidences shows that increasingly more this switch occurs for those activities which are not part of the “core competences” of the firm, but rather ancillary ones, externalized to specialized suppliers of other industries: janitorial services and ICT are the most notable example (Abraham and Taylor, 1996). In these and other similar cases, on principle, one would expect that outsourcing, by involving an intersectoral relation, gets somehow reflected into a correspondent change in the relevant input-output table, which of such a kind of relations represents an increasingly more accurate map. In the following, we will refer to it as “intersectoral outsourcing”. Conversely, when firms outsource parts of their production process itself - for example by contracting out the transformation of a certain intermediate input - the evidence of a change at the input-output level will presumably be less visible, as it amounts to an “intrasectoral outsourcing” relation, rather than an intersectoral one. Still, a certain correspondence between the two levels of analysis - that is, the firm and the sector level - could be looked for. However, such a correspondence is just a spurious one, mainly for two methodological problems. The first one has to do with the unit of analysis for the construction of input-output tables.3Indeed, input-output tables are built up by measuring and adding, sector by sector, the deliveries of goods and services which occur between different “establishments”, rather than between different firms, or enterprises, as such. In other words, the so-called “interestablishment deliveries” – that is, deliveries of goods and services between 3In general, intersectoral studies of outsourcing are quite “cavalier” on this point. In Tucker and Wilder (1977), Domberger (1998), Dietrich (1999) and McCarthy and Anagnostou (2004) for example, such a crucial issue is not even mentioned. 4 establishments belonging to the same enterprise – are also accounted as total output of the production unit and thus recorded as either intermediate consumption or gross fixed capital formation by the receiving unit. Apparently, this would represent a serious obstacle in detecting outsourcing starting from input-output tables. Indeed, “data aggregation methods which assign various plants of a single company to different industries in effect ignore multi plant ownership and therefore result in data which are insensitive to major forms of vertical integration” (Woodrow Eckard, 1979, p.105): the comment, raised in an old but effective note on the empirical measurement of vertical integration, actually also applies to vertical disintegration. The insensitivity of input-output tables to outsourcing is however not total and rather depends on the specific case. At the outset, the sensitivity is quite high in front of those outsourcing processes which occur when a certain establishment substitutes services and/or intermediate inputs provided by an establishment of a different firm, for those previously produced within the establishment itself (“pure outsourcing”). This happens, for instance, when a firm decides to outsource the janitorial services previously performed within each establishment to an external specialized service provider. The sensitivity is definitively lower when services and intermediate inputs were previously provided to a certain establishment by another establishment of the same firm: indeed, the substitution, for the latter, of an establishment belonging to a different firm, could virtually leave the correspondent input-output deliveries unchanged but, it should be stressed, just in quantitative terms. Indeed, in terms of value, the substitution would certainly find an input-output manifestation because of the substitution of market prices for “internal prices” in evaluating the outsourced transaction. Such a transaction actually becomes more “permeable” to market mechanisms. One could think of a firm in the automobile industry that closes down a plant producing shock absorbers and starts buying them from specialized suppliers or sub-contractors. This is also the case of a firm which spins off one of its divisions. A last case could instead generate a sort of “over-sensitivity”, rather than insensitivity, of input-output tables to outsourcing. Such as when a certain firm creates, ex-novo, a new establishment for the provision of certain services and/or intermediate inputs to another existing establishment of its own. This could be the case of a firm establishing a plant for the internal provision of IT services, and such services were previously provided within each establishment of the firm, without involving inter-establishment deliveries. Although it is certainly true that this difference in the unit of analysis poses some problems, it has to be noted that, if one deals with service outsourcing, as we actually do in the present paper, such a difference is not so crucial. Indeed, because of the particular nature of the output of such activities, that is neither storable nor transportable, the most frequent situation is probably the one in which, prior to outsourcing, services are provided within each establishment. One might think of janitorial or cleaning services. Firms do not usually set up separate establishments for the “in-house” provision of such services. In fact, implementing what is usually called an Activity-Based Costing and Management (ABCM), and thus keeping service costs under control, is usually one of the reasons for firms to outsource service provision.4 4As pointed out by Domberger, “many private sector organizations, and most public sector 5 However, it is also true that one should not exclude that, instead of resorting to outsourcing, firms could set up a separate establishment for the provision of such services. This is the case of over-sensitivity of input-output data already stressed: although, strictly speaking, this does not represent a case of outsourcing as such, changes in input-output relationships could assimilate it to other more proper cases of it. Notwithstanding, also in the light of its limited extent, accounting it as a special (and indeed “odd”) case of inter-firm outsourcing does not appear totally misleading with respect to the inner rational of the process. Indeed, inter-establishment transactions, even if within the same firm, are more transparent than intra-establishment ones. The second methodological limitation of an intersectoral analysis of outsourcing by means of input-output tables has to do with what is referred to as international “fragmentation” (e.g. Jones and Kierzkowski, 2001) and the distinction between its two main channels: international delocalization and international outsourcing. The former is intended as the set-up of a plant in a foreign country by a domestic firm, whereas the latter usually refers to a situation in which firms contract out parts of their production process to foreign firms.5Indeed, although phenomena of international fragmentation can be detected by referring to input-output tables of imported flows, it is not possible to distinguish at which extent it is due to delocalization rather than international outsourcing.6One again, although relevant, this problem is not so crucial for empirical applications, such as the present one, which intend to capture, rather than the international division of labour, the influence of outsourcing on the sectoral boundaries of an economy. With these caveats in mind, in the following we discuss a “battery” of inputoutput indicators of outsourcing, at different levels of analysis (e.g. sectoral and subsystem) and with different interpretative power (e.g. direct and indirect). 4 Outsourcing measurements for structural change analysis Once outsourcing has been meant as a structural change determinant, possibly substitutive or complementary with respect to other determinants at the sectoral level (such as industrialization and deindustrialization), identifying a consistent sectoral measurement for it becomes necessary, although problematic. ones, cannot account for their internal costs of service provision on a disaggregated basis, that is, in terms of individual service” (Domberger, 1998, p.47). 5To be sure, there is no clear definition of international outsourcing. Indeed, while someone uses the term as a synonym of international fragmentation, measuring it as the ratio of imported intermediate inputs to the total value of domestic production (e.g. Feenstra and Hanson, 1999; Glass, 2004; Jones et al., 2005), someone else instead refers to “international partnerships” (e.g. Van Long, 2005), thus assuming a minimum level of relationship durability among the transactors. 6Input-output tables have been widely used in studies on international outsourcing in manufacturing sectors. In these works international outsourcing is measured as the ratio of imported to domestic intermediate inputs of each sector (Campa and Goldberg, 1997; Feenstra, 1998; Feenstra and Hanson, 1999). Another recent study by Hummels et al. (1998) deals with the issue of the actual complementarity between international production and international trade due to the increasingly important role played by vertical specialization. They use inputoutput data to calculate for each sector the amount of imported inputs embodied in exported goods. 6 Indeed, because of the problems related to the unit of analysis (discussed in Section 3), such measurements cannot be taken for more than outsourcing proxies. Their interpretative power is however different, as we will show in the following. 4.1 The sectoral indicators of service outsourcing The most straightforward way to capture the amount of service outsourcing made by the firms belonging to a certain manufacturing sector is, of course, looking at the correspondent intersectoral flows from the relevant input-output table. For instance, we could analyze the changes occurred over time in the intermediate business service transactions at constant prices (SERV ) made by manufacturing firms per unit of production (Q), that is, for sector i: ∆(SERVi/Qi), where a positive variation could be a signal of outsourcing. Although the most direct indicator, in using it we have to implicitly assume two hypotheses. First of all, returns to scale should be held constant, referring to a common but quite problematic assumption in input-output analysis of structural change. Second, we have to assume that technological progress does not significantly affect service technical coefficients in manufacturing sectors. In particular, if we are interested in cross-sectional structural comparisons, we should retain that this effect is the same across the compared countries. Moreover, given that the present indicator is based on nothing but an inputoutput coefficient, we should also discount the fact that a change of it might be even provoked by cases of “intrasectoral outsourcing”. Indeed, as we will clarify in the next section, this kind of outsourcing might decrease SERVi/Qivia an increase of the gross production of sector iitself. In trying to overcome these problems, one might want to use an alternative indicator of service outsourcing, suggested by McFetridge and Smith (1988), that is: the change in the ratio between the intermediate business service transactions of a certain sector i(SERVi) and the value of wages and salaries of the same sector (LABRi): ∆(SERVi/LABRi). Indeed, as we will also argue in what follows, outsourcing usually implies a substitution of primary inputs, mainly labour, for intermediate inputs, so that a positive variation of the previous ratio could be signaling service outsourcing. However, also to use this measure as an indicator of outsourcing we have to make an important implicit assumption: the price of business services in wageunits have to be stable over time. In making cross-section comparisons, for example, we have to assume a constant relative cost of labour across countries, an assumption which hardly holds true for economic systems with different levels of development. Moreover, the same indicator tends to vary, and thus become less reliable, whenever a change in the labour productivity of a certain sector is not properly reflected in the correspondent monetary wages. Thus, also the present indicator, as the previous one, is just an imperfect indicator of service outsourcing at sectoral level. Both of them are affected by different phenomena, not all related to outsourcing. However, the “noise” by which they are affected can be deemed as less problematic when, and if, they both signal traces of outsourcing, while contrasting signals would recommend caution. 7 4.2 The input-output technical coefficients The effects that outsourcing brings about at sectoral level are not just limited to the direct ones associated to an increase of SERVi/Qi. More in general, through any kind of outsourcing, intra-firm transactions, or better to say, “intra-establishment deliveries”, which cannot be caught by national accounts and input-output data, actually shift outside the firm (the establishment) and thus become measurable by them. In other words, by shifting the boundaries of a certain establishment, outsourcing brings about an increase in the intermediate consumptions of the sector to which it belongs, which comes from organizational changes, and not from technological ones. More precisely, through this mechanism outsourcing affects the intermediate consumption, the total production and the value added of the sectors in which it occurs. Of course, these effects are different depending on the sectoral classification of the establishments themselves. In what we called “intrasectoral outsourcing”, the “outsourcee” and the “outsourcer” belong to the same sector: ceteris paribus, we can view outsourcing as increasing the intermediate consumption and the gross output of the same sector, while leaving its value added hardly affected. In what we called “intersectoral outsourcing”, on the contrary, outsourcing involves units of production belonging to different sectors, such as in the case of producer services analyzed in the previous section: ceteris paribus, in the “outsourcer” sector, because of outsourcing, gross production can be viewed as unchanged, while its value added diminishes. At the outset, these effects determine a change in the relevant input-output coefficients, that is in the correspondent elements aij of matrix Adefined as: A=Wˆ q−1(1) where ˆ qand Wstand for, respectively, the diagonalized vector of sectoral gross production and the matrix of intersectoral production flows. And these changes can be read accordingly. When in a certain sector j there are evidences of “intrasectoral outsourcing” processes, its gross output (Qj) increases, because the correspondent intrasectoral inflows (wjj) increase for accounting reasons. On the other hand, the inflows out of the main diagonal (wij with i6=j) do not change. For this reason, outsourcing determines, first of all, a reduction in the extra-diagonal technical coefficients for sector j, that is, aij (with i6=j). In addition, as the sectoral value added, which is not affected by duplication, remains unchanged, the increase of wjj turns out to be greater than that of Qj, because Qjis the sum of all the intermediate inflows plus the value added of the sector. Accordingly, outsourcing also makes the technical autocoefficients ajj increase. Summing up, “intrasectoral outsourcing” in a certain sector jmight be expected to determine a reduction in the technical coefficients aij (with i6=j) and an increase in the technical coefficient ajj. Let us stress that this is due just to accounting reasons, not related to the production side (for a formal treatment see Appendix A). These effects do not occur, instead, in the case of “intersectoral outsourcing”, because the gross output of the “outsourcer” sector remains relatively stable. Therefore, disentangling organizational changes from technological ones becomes in this case nearly impossible. Additional information could however be obtained by looking at an important related sectoral ratio to which we will now turn. 8 Finland, France, Germany, Greece, Hungary, Italy, Japan, Korea, Netherlands, Norway, Poland, UK and US, with respect to the middle ’90s (see Appendices B and D). Data availability also allows us to retain a sectoral disaggregation for manufacturing (including construction) of 14 sectors for the OECD6 over the ’80s, and of 17 sectors for the OECD18 in the middle ’90s (see Appendix E).18 At the outset, we stress that such an application is just intended to be illustrative of the methodological pros and cons of the various indicators, rather than explorative of the economic structure and structural change of the OECD area.19 We also stress that, rather than following the presentation order of the outsourcing measurements, the application starts by looking for the most general traces of outsourcing, at the subsystem level, for then moving to more specific insights at the sectoral level. 5.1 Searching for “system” traces of service outsourcing To start with, although a rough measurement, cross-country average values of the sectoral vertical integration degree and of the business services integration provide some interesting insights about those sectors which are “structurally” more disintegrated than others, and for which outsourcing could thus be more relevant. Across the 18 OECD countries considered in the middle ’90s, 5 are the sectors in which, on average, extra-sectoral labour contributions (direct and indirect) are particularly relevant (Table 2).20 However, when the role of natural resources intensity is discounted for, the number of the most disintegrated sectors narrows to 3: basic metals, chemical and transport equipment subsystems, for which, unlike the other two (i.e. food, beverages and tobacco, and coke, petroleum, and nuclear fuel) no manufacturing or agricultural sector out of the main diagonal plays a pivotal role. Quite interestingly, these are also among the sectors in which production services have the greatest average labour weight in the middle ’90s. More in general, the two rankings are quite similar, supporting our tentative interpretation of a relationship between vertical disintegration and integration in business services.21 Our structural analysis seems thus aligned with what other contributions find through case-studies at the firm level, in particular for transport equipments and chemicals (Domberger, 1998). As for basic metals, instead, the result is somehow counterfactual, having in mind the relative “closeness” to raw materials of the sector compared to the other manufacturing ones. However, the data show a prominent role of horizontal linkages over the vertical ones for manufacturing, 18In the definition of business sector services we have followed OECD conventions (50–74 ISIC Rev.3) (see Appendix C). In order to avoid, as much as possible, distortions coming from sectoral aggregation, calculations have been carried out at the maximum level of disaggregation, and then the results have been reaggregated as required. 19In this vein, the present application represents a selection of a more extended empirical work carried out on the same area (Montresor and Vittucci Marzetti, 2006). 20In Table 2 sectors have been ranked by referring to total production input-output tables. However, the ranking remains substantially unchanged when the reference is to domestic production flows (the Spearman correlation index is as high as 0.941). 21The Spearman correlation index is 0.684 and becomes 0.798 when the two “resource intensive” sectors previously analyzed are not considered. 15 Table 2: Vertical disintegration and business services integration per industrial subsystems - middle ’90s – cross-country average values Industrial subsystems Avg % values Vertical disintegration Rank Business services integration Rank Low vertical integration Coke, refined petroleum products and nuclear fuel 9.9 1 35.8 1 Food products, beverages and tobacco 26.3 2 20.6 14 Motor Vehicles, Trailers and Semitrailers 35.1 3 28.4 4 Chemicals 35.4 4 35.3 2 Basic Metals 37.4 5 30.2 3 Middle vertical integration Rubber and Plastics Products 44.2 6 25.7 7 Electrical machinery & Apparatus, nec 45.4 7 24.9 8 Office and computing machinery - Communication equipment - Medical, precision and optical instruments 46.9 8 27.6 5 Other non-metallic mineral products 48.5 9 26.3 6 Machinery and equipment, nec 49.4 10 23.4 10 Other transport equipment 49.6 11 21.2 13 Wood, products of wood and cork 50.2 12 18.2 16 High vertical integration Manufacturing, nec; Recycling 52.2 13 19.4 15 Construction 52.4 14 21.4 11 Pulp, paper, paper products, printing and Publishing 53.2 15 24.9 9 Fabricated metal products 54.2 16 21.3 12 Textiles, textile products, leather and footwear 64.6 17 17.8 17 Source: our calculation on OECD I-O Database and 60-Industries GGDC Database along with a high integration of services in the basic metal subsystem, revealing a process of great restructuring. In searching for traces of “system” outsourcing, interesting results emerge also in terms of cross-country structural comparisons and of structural change. As for the former, Table 3 shows some interesting country peculiarities.22 Business services have a relatively lower integration in manufacturing in all the transition economies considered, namely Czech Republic, Poland and Hungary. Hungary, in particular, although the country in which in the 1990s there was the largest increase in the service share, both in terms of value added and labour, and where manufacturing labour productivity grew faster (see, for instance, Landesmann, 2000), reveals the least integrated business services in manufacturing: thus suggesting that the increase in the service sector was mainly due to final services, whereas producer services still lag behind. At the opposite extreme, the highest integration of business services in manufacturing is shown 22Also in this case, the analysis has been accomplished both with respect to total and domestic flows input-output tables. The relative country ranking is quite similar (the Spearman correlation index is 0.83). 16 Table 3: Business services integration in the manufacturing subsystem - middle ’90s – Weight of business services on the manufacturing subsystem in terms of hours worked Country Business services integration Total flows Rank Domestic flows Rank United Kingdom 29.1 1 26.6 2 Netherlands 29.0 2 25.9 4 France 28.9 3 26.4 3 Australia 26.9 4 24.2 6 United States 25.7 5 27.2 1 Canada 25.4 6 24.9 5 Norway 25.1 7 21.4 8 Spain 24.9 8 21.0 11 Korea 24.2 9 21.1 10 Finland 23.5 10 19.0 14 Italy 23.3 11 18.6 15 Japan 22.9 12 21.4 9 Germany 22.5 13 19.7 12 Czech Republic 21.8 14 19.4 13 Poland 21.4 15 17.4 16 Denmark 20.3 16 22.1 7 Greece 19.1 17 14.9 18 Hungary 15.2 18 16.1 17 Source: our calculation on OECD I-O Database and 60-Industries GGDC Database by the UK, which underwent in the 1980s a deep process of economic restructuring (see, for instance, Matthews and Gardiner, 2000). While the high positions of Australia, the United States and Canada are also confirmed by other studies (e.g. Domberger, 1998).23 Coming to the structural change implications, a first set of indications comes from the analysis of the levels and patterns of change in the average degree of both vertical integration and business services integration in manufacturing subsystems. At the outset, vertical disintegration at the subsystem level appears a quite recent result in OECD manufacturing, of the middle ’80s, and indeed a switch with respect to the early ’80s. Apart from transport equipment, all the 14 retained manufacturing sectors increased, rather than decreased, their average vertical integration at the beginning of the ’80s (Figure 2). At that stage only motor vehicles and other transport sectors in OECD6 seemed to have started undergoing a process of vertical reorganization of labour, being it due to technological change or other causes, such as outsourcing. Moreover, unlike vertical disintegration, the increase of the labour weight of services on manufacturing subsystems was already occurring in the early ’80s (Figure 3).24 Combining the 23In passing, it is worthwhile observing that the country ranking we got in terms of business services subsystem integration is substantially different from that obtained working with value added and employment shares of business services in total economy. This suggests that “simple” sectoral indicators of tertiarization could be misleading in interpreting more complex cases of structural change (Montresor and Vittucci Marzetti, 2006). 24It should be noted that, keeping technical coefficients and production organization con17 Figure 2: Vertical integration degree of OECD6 manufacturing – cross-country average values: 1980-1990 – Weight of sectors on the relative subsystems (hours worked) – total flows at constant prices Figure 3: Business services integration in manufacturing subsystems in the OECD6 – cross-country average values: 1980-1990 – Weight of business services in manufacturing subsystems (hours worked) – total flows at constant prices 18 (a) Vertical integration degree (b) Weight of business services Figure 4: Total manufacturing: 1980-1990 two results, it seems that, although increasingly more important for manufacturing, in the early ’80s business services did not enter in it as substitute yet. This has possibly occurred instead in the middle ’80s, as the vertical integration of business services in manufacturing further increased on average and was accompanied, as we saw, by the vertical disintegration of the latter. Interesting results also emerge by analyzing the changes intervened in the individual countries. By referring to the whole manufacturing (Figure 4), for example, one can contrast UK with the US. Indeed, while the former reveals a remarkable increase of the business services integration in manufacturing, the latter is the only country in which the weight of services in manufacturing during the ’80s, rather than increasing, decreased to an appreciable extent (-5.3%), thus hinting at a possible different strategy of US manufacturing firms, which got more vertically integrated. As interesting are the peculiar patterns of change at the level of individual subsystem. The analysis of transport equipment (Figure 5), for example, reveals a notable disintegration in Japan over the whole period and far beyond the others,25 thus confirming the outcomes of other studies at the firm level on the restructuring of the Japanese sector (see, for instance, Womack et al., 1990). stant, the weight of business services on manufacturing subsystems tends to growth over time when, as it is generally assumed, the rate of growth of labour productivity in manufacturing is greater than in services. Accordingly, service outsourcing only accelerates the pace of this “natural” tendency. 25In 1990 the vertical integration degree of the Japanese transport equipment sector was just 0.34, while in the United States it was nearly twice as much (0.65). Moreover, from the early to the late ’80s the rate of change of the vertical integration degree of this sector was -29.4%, while in the US it was positive and equal to 5.5%. The main findings get confirmed when the vertical integration degree is calculated with respect to domestic, rather than total, production flows. 19 (a) Vertical integration degree (b) Weight of business services Figure 5: Transport equipment: 1980-1990 5.2 Sectoral value added-gross production ratios: are they reliable? As we have argued in Section 4.3, although built up by using simple sectoral data, the value added-gross production ratio should be regarded as an outsourcing measurement which refers to the subsystem level. Accordingly, we should expect a certain correlation between such a measurement and that applied in the previous section. In order to verify this expectation, we calculated the average sectoral value added-gross production ratios over the ’90s across the same sectors and countries (OECD18) of the previous sections, worked out the relative cross-country averages and ranked the sectors accordingly (Table 4) (see Appendix B for the dataset description). Such an operation has been done after having checked for an eventual bias in the V A/Q ratio due to the sectoral proximity to raw materials. As the bias did not turn out evident,26 we have then analysed the relation between the average sectoral V A/Q ratios and the percentage labour share of the sectors in the relative subsystems (Figure 6). Quite interestingly, their correlation turns out to be as high as 0.862 and the two rankings are quite similar, except for some sectors, in particular: textiles, leather and footwear; other non-metallic mineral products; and construction (Figure 6). These and other outliers deserve a special attention, as they actually reveal the different “disintegration” rational the two measures are able to capture. First of all, as we said in Section 4.5, while the vertical integration degree does not change because of phenomena of intrasectoral disintegration, the same does not hold true with respect to the sectoral V A/Q ratio, which tends to decrease when establishments belonging to a certain sector outsource to establishments 26E.g., basic metals has got a ratio equal to 0.261, while the ratios of fabricated metals product and electrical & optical instruments sectors are equal to, respectively, 0.401 and 0.353. The coke, refined petroleum products and nuclear fuel sector has the lowest V A/Q ratio (0.202), though it is certainly one of the most structurally “closer” to raw materials among the manufacturing sectors. 20 Table 4: Average sectoral V A/Q ratios of OECD18 manufacturing sectors over the ’90s – cross-country average values Industrial sectors Avg V A/QaVertical integrationb Coke, refined petroleum products and nuclear fuel 0.202 9.9 Basic Metals 0.261 37.4 Food products, beverages and tobacco 0.264 26.3 Motor Vehicles, Trailers and Semitrailers 0.268 35.1 Chemicals 0.332 35.4 Wood and products of wood and cork 0.349 50.2 Electrical and Optical instruments 0.353 46.4 Textiles, textile products, leather and footwear 0.361 64.6 Other transport equipment 0.363 49.6 Rubber and Plastics Products 0.368 44.2 Machinery and equipment, nec 0.376 49.4 Pulp, paper, paper products, printing and Publishing 0.380 53.2 Manufacturing, nec; Recycling 0.400 52.2 Fabricated metal products 0.401 54.2 Other non-metallic mineral products 0.405 48.5 Construction 0.421 52.4 Source: our calculation on OECD I-O Database and 60-Industries GGDC Database aAverage sectoral value added-gross production ratios 1990-2000 bWeight of sectors in the relative subsystems in terms of hours worked – Total flows classified in the same sector. Apparently, this is what happened during the 1990s in the textile sector, where phenomena of “intrasectoral outsourcing” were highly frequent. Although to a lesser extent, the same holds true also for wood, wood products and cork; basic metals; and paper products, printing and publishing. Quite interestingly, what emerge from the data can be related to the recent technical changes occurred in these sectors, especially in the last two, involving a reduction in their minimum efficient scale. We refer in particular to the emergence of mini-mills in the steel production (e.g. Audretsch and Feldman, 1996) and to the massive computerization occurred in printing and publishing in the last years (e.g. Domberger, 1998). A different argument holds with respect to constructions, other non metallic mineral products and rubber and plastics products, which appear less disintegrated in terms of V A/Q than what would be in terms of the relative sectorsubsystem ratio. As we pointed out in Section 4.5, this could be due to the peculiar market structure of these sectors, of the first in particular, in which monopolistic rents are quite diffuse and hinder the outsourcing revealing power of the V A/Q ratio. Although both the kinds of interpretations need to be confirmed with other data, suffice here to notice how the two indicators of disintegration/integration should be used in a complementary way to have a better understanding of the investigated phenomenon. 21 Figure 6: Sectoral V A/Q ratios vs vertical integration degrees 5.3 Service outsourcing: what input-output technical coefficients have to say? The final point to address is how consistent the previous indicators of “system” integration of business services into the production of manufactured goods are with the more “direct” sector-based indicators of outsourcing we have presented in Section 4, that is SERV/Q and SERV/LABR. At the outset, let us observe that, once calculated for the OECD18 in the middle ’90s, the rank correlation index between SERV/Q and SERV/LABR is very low (0.31). And also the relative ranking shows how SERV/LABR does not turn out very reliable in cross-country comparisons, as possibly affected quite a lot by the relative cost of labour (Table 5). Second, the ranking made according to SERV/Q is not consistent with the results we have previously obtained at the subsystem level (see Table 3),27 thus suggesting how these sectoral indicators should be retained measures of changes in service outsourcing practices at the sectoral level over time, with differences and some degree of complementarity with respect to the ones analysed at the subsystem level. The replication of the analysis carried out in Section 5.1 for manufacturing as a whole and for transport equipment can be of some help in illustrating this point (Figures 7 and 8). Quite interestingly, as far as the European countries are concerned, the two data series are quite consistent between them and decisively supportive of the outsourcing hypothesis. Indeed, these countries show significant increases in both the variables over the whole period. In particular, the data provide further evidence on the economic restructuring of the UK manufacturing during the ’80s, and especially in the last five years.28 However, it has to be noted that the 27The correlation indexes, both linear and rank, between SERV/Q in manufacturing and the business services integration in manufacturing subsystem are definitively quite low. The linear correlation is 0.42, while the rank one is even lower and equal to 0.325. As for SERV/LABR, its rank correlation with the indicator at the subsystem level is as low as 0.063, while the correspondent linear correlation is nearly null (-0.013). 28For manufacturing as a whole, for example, SERV/Q increased more than 66%, while SERV/LABR of 60.6%. 22 Table 5: Business services expenditure per production unit (SERV/Q) and on labour compensation (SERV/LABR) in manufacturing in middle 90s Country SERV/Q Rank SERV/LABR Rank United Kingdom 0.186 1 0.752 9 France 0.183 2 0.855 4 Denmark 0.183 3 0.721 15 Norway 0.178 4 0.829 6 Germany 0.177 5 0.624 17 Japan 0.177 6 0.775 8 Australia 0.176 7 0.986 2 Poland 0.171 8 1.128 1 United States 0.168 9 0.752 10 Italy 0.158 10 0.840 5 Netherlands 0.152 11 0.751 11 Spain 0.152 12 0.750 12 Czech Republic 0.139 13 0.986 3 Finland 0.128 14 0.667 16 Greece 0.127 15 0.788 7 Canada 0.121 16 0.593 18 Hungary 0.115 17 0.737 13 Korea 0.111 18 0.728 14 Source: our calculation on OECD I-O Database and 60-Industries GGDC Database integration of business services in UK manufacturing is largely underestimated by the two indicators of the present section, both in terms of levels and of rates of changes, as it clearly emerges comparing Figures 4-5 with Figures 7-8. This suggests how considering both direct and indirect intersectoral relationships matters in dealing with outsourcing as much as with other processes of economic restructuring. Apart from Canada and US, where data do not show significant tendencies towards service outsourcing29 (the only exception is represented by the US transport equipment sector in the late ’80s) a special attention should be paid to the case of Japan, and of the Japanese transport equipment sector in particular, for which the data seem to be somehow inconsistent. Indeed, as we saw in Section 5.1, during the ’80s the growth rate of business services integration in the transport equipment subsystem was of 30.8%, and accompanied by a related process of sectoral disintegration (-29.4%), thus clearly signalling an intense process of economic restructuring within the subsystem. This fact gets somehow confirmed by the data on SERV/LABR: for transport equipments, the overall growth rate of the indicator for the ’80s was nearly 25%, although this increase mainly occurred in the last five years.30 29In the United States during the ’80s the overall rate of change of SERV/Q for the whole manufacturing is negative and equal to -5.2%, whereas the rate of change of SERV/LABR, though positive, is relatively small (6.9%). 30At the same time, the indicator increased a lot also for machinery & equipment (+35.4%), the main supplier sector of transport equipment. As for basic metals and fabricated metals products, the other two manufacturing sectors with relevant labour share in the transport equipment subsystem, the rates of change of SERV/LABR were, respectively, 24.1% and -2.9%. Furthermore, looking at the average levels of the indicators in the two sectors, Japan 23 (a) Total Manufacturing (b) Transport Equipment Figure 7: Intermediate business services expenditure per production unit (constant prices): 1980-1990 (a) Total Manufacturing (b) Transport Equipment Figure 8: Intermediate business services expenditure on labour compensation (current prices): 1980-1990 However, the same pattern is not revealed by the data on SERV/Q. Indeed, unlike for the other countries, in the Japanese sector there was no significant increase in business services expenditure per production unit during the 1980s (see Figure 7(b)), and the same does hold true also for its traditional supplier sectors, that is, machinery & equipment, basic metals, rubber and plastics products and other fabricated metal products.31 In synthesis, it seems that the shows the highest values. 31Although sectoral input flows at constant prices from business services increased more than 89.8%, the overall increase in sectoral gross production was 95.6%, thus determining a reduction in the coefficients of -2.94%. Looking at the Japanese data on SERV/Q for machinery & equipment; basic metals; rubber and plastics products; and other fabricated metal products, the overall rates of change for the 1980s were, respectively, −11.5%, 9.9%, −16.6%, −2.6%. (With respect to the same period, in the UK the rates of change in the corresponding sectors were 40.9%, -23.1%, 70.9% and 43.6%.) In addition, examining the average level of SERV/Q over the ’80s, rather than its change rates, it turns out that it is 24 Appendix E Sector classification Sector ISIC Rev.3 Codes Food products, beverages and tobacco 15-16 Textiles, textile products, leather and footwear 17-19 Wood and products of wood and cork 20 Pulp, paper, paper products, printing and publishing 21-22 Coke, refined petroleum products and nuclear fuel 23 Chemicals (including pharmaceuticals) 24 Rubber and plastics Products 25 Other non-metallic mineral products 26 Basic metals 27 Fabricated metal products (except machinery and equipment) 28 Machinery & equipment ...Machinery and equipment, nec 29 ...Office and computing machinery - communication equipment - medical, precision and optical instruments 30,32-33 ...Electrical machinery and apparatus, nec 31 Transport equipment ...Motor vehicles, trailers and semitrailers 34 ...Other transport equipment 35 Manufacturing, nec; recycling 36-37 Construction 45 References Abraham, K. G., Taylor, S. K., 1996. Firms use of outside contractors: Theory and evidence. Journal of Labor Economics 14 (3), 394–424. Adelman, M. A., 1955. Concept and statistical measurement of vertical integration. In: Stigler, G. J. (Ed.), Business Concentration and Price Policy. NBER-Priceton University Press, pp. 281–322. Antonelli, C., 1988. The emergence of the network-firm. In: Antonelli, C. (Ed.), New Information Technology and Industrial Change: the Industrial Case. Kluwer Academic Pubblishers, Dortrecth, pp. 13–22. Audretsch, D. B., Feldman, M. P., 1996. Innovative clusters and the industry life cycle. Review of Industrial Organization 11, 253–273. Baranzini, M., Scazzieri, R., 1990. Economic structure: Analytical perspectives. In: Baranzini, M., Scazzieri, R. (Eds.), The Economic Theory of Structure and Change. Cambridge University Press, Cambridge, Ch. IX, pp. 227–333. Campa, J., Goldberg, G., 1997. The evolving external orientation of manufacturing industries: Evidence from four countries. Federal Reserve Bank of New York Economic Policy Review 4, 79–99. Diestel, R., 2005. Graph Theory, 3th Edition. Springer-Verlag Heidelberg, New York. 31 Dietrich, M., 1999. Explaining economic restructuring: an input-output analysis of organisational change in the European Union. International Review of Applied Economics 13 (2), 219–240. Domberger, S., 1998. The Contracting Organization. A Strategic Guide to Outsourcing. Oxford University Press, Oxford. Feenstra, R. C., 1998. Integration of trade and disintegration of production in the global economy. Journal of Economic Perspective 12 (4), 31–50. Feenstra, R. C., Hanson, G. H., 1999. The impact of outsourcing and hightechnology capital on wages: Estimates for the United States, 1979-1990. Quarterly Journal of Economics August, 907–940. Gershuny, J., Miles, I. D., 1983. The New Service Economy: The Transformation of Employment in Industrial Societies. Praeger, New York. GGDC, February 2005. Data sources and methodology of the 60-industry database of the Groningen Growth and Development Centre. http://www. ggdc.net. Ginzberg, E., Vojta, G., 1981. The service sector of the U.S. economy. Scientific American 3. Glass, A. J., 2004. Outsourcing under imperfect protection of intellectual property. Review of International Economics 12 (5), 867–884. Gossling, W., 1972. Productivity Trends in a Sectoral Macro-Economic Model. Input-Output Publishing, London. Grossman, G., Helpman, E., 2002. Integration versus outsourcing in industry equilibrium. The Quarterly Journal of Economics 117, 85–120. Gupta, S., Steedman, I., 1971. An input-output study of labour productivity in UK. Oxford Bulletin of Economics and Statistics 33, 21–34. Hummels, D., Rapoport, D., Yi, K.-M., 1998. Vertical specialization and the changing nature of world trade. Federal Reserve Bank of New York Economic Policy Review 4, 79–99. Jacobides, M. G., Billinger, S., January 2005. Designing the boundaries of the firm: from “make, buy or ally” to the dynamic benefits of vertical architecture. London Business School - Strategic and International Management Working Paper Series, working Paper. Jarillo, J., 1993. Strategic Networks: Creating the Borderless Organization. Butterworth-Heinemann, Oxford. Jones, R., Kierzkowski, H., Lurong, C., 2005. What does evidence tell us about fragmentation and outsourcing? International Review of Economics and Finance 14, 305–316. Jones, R. W., Kierzkowski, H., 2001. A framework for fragmentation. In: Arndt, S. W., Kierzkowski, H. (Eds.), Fragmentation: New production patterns in the world economy. Oxford University Press, Oxford and New York, pp. 17– 34. 32 Kohler, W., 2004. Aspects of international fragmentation. Review of International Economics 12 (5), 793–816. Laffer, A. B., 1969. Vertical integration by corporations, 1929-1965. Review of Economics and Statistics February, 1969, 91–93. Landesmann, M., 2000. Structural change in the transition economies, 19891999. Economic Survey of Europe 2/3, 95–123. Matthews, H., Gardiner, V. (Eds.), 2000. The Changing Geography of the United Kingdom. Routledge, New York. McCarthy, I., Anagnostou, A., 2004. The impact of outsourcing on the transaction costs and boundaries of manufacturing. International Journal of Production Economics 88, 61–71. McFetridge, D. G., Smith, D. A., 1988. The Economics of Vertical Disintegration. Fraser Institute, Vancouver. Milberg, W. S., 1991. Structural change and international competitiveness in Canada: an alternative approach. International Review of Applied Economics 5, 77–93. Momigliano, F., Siniscalco, D., 1982a. The growth of service employment: a reappraisal. BNL Quarterly Review 142, 269–306. Momigliano, F., Siniscalco, D., 1982b. Note in tema di terziarizzazione e deindustrializzazione. Moneta e Credito 138, 143–181. Momigliano, F., Siniscalco, D., 1984. Technology and international specialization. BNL Quarterly Review 150, 257–284. Montresor, S., Mazzanti, M., Pini, P., 2006. The general profile of the outsourcing firm: Empirical evidence for a local production system of Emilia Romagna. mimeo, University of Bologna. Montresor, S., Vittucci Marzetti, G., 2006. Outsourcing and structural change in the OECD area. mimeo, University of Bologna. OECD, 1995. The OECD input-output database. sources and methods. http: //www.oecd.org/sti/stan. OECD, 2004. STAN database. http://www.oecd.org/sti/stan. OECD, 2005. Input-output tables (2002 edition). http://www.oecd.org/sti/ stan. Pasinetti, L. L., 1973. The notion of vertical integration in economic analysis. Metroeconomica 1. Rampa, G., 1982. Commento a Momigliano e Siniscalco. Moneta e Credito 139, 475–479. Rampa, G., Rampa, L., 1982. Sul mutamento tecnologico nell’economia italiana, 1959-1975: Un’analisi input-output. Ricerche economiche XXXVI (4), 303– 341. 33 Spencer, B. J., 2005. International outsourcing and incomplete contracts. Working Paper 11418, National Bureau of Economic Research. Sraffa, P., 1960. Production of Commodities by Means of Commodities. Cambridge University Press, Cambridge. Stanback, T. M., 1979. Understanding the Service Economy. Johns Hopkins University Press, Baltimora. Stigler, G. J., June 1951. The division of labor is limited by the extent of the market. The Journal of Political Economy 59 (3), 185–193. Tucker, I. B., Wilder, R. P., 1977. Trends in vertical integration in the U.S. manufacturing sector. Journal of Industrial Economics 26 (1), 81–94. Van Long, N., 2005. Outsourcing and technology spillovers. International Review of Economics & Finance 14 (3), 297–304. Williams, K., Haslam, C., Williams, K., Johal, S., 1998. Cars: Analysis, History, Cases. Berghahn Books, USA. Womack, J. P., Jones, D. T., Roos, D., 1990. The Machine That Changed the World : The Story of Lean Production. Harper Collins Publishers, New York. Woodrow Eckard, Jr, E., September 1979. A note on the empirical measurement of vertical integration. Journal of Industrial Economics 28 (1), 105–107. Zaghini, E., 1967. Una nota sui subsistemi di Sraffa. Studi Economici 22, 290– 305. 34