The Role of Technological Complexity and Absorptive Capacity in Internalization Decision

Transcription

1 CDE May 2007 The Roe of Technoogica Compexity and Absorptive Capacity in Internaiation Decision Arti Grover Emai: Dehi choo of Economics University of Dehi Working Paper o. 53 Centre for Deveopment Economics Department of Economics Dehi choo of Economics

2 The Roe of Technoogica Compexity and Absorptive Capacity in Internaiation Decision Arti Grover Dehi choo of Economics Abstract Technoogy transfer costs have a profound infuence on the firm s entry mode into a production sharing reationship. To expore this nexus we associate technoogica compexity of the offshored input with the organiationa mode of internationa production sharing by extending the Antràs (2005 mode. We modify the Antràs mode by proposing that the ow-tech input as quaified within the mode cannot be produced in the ow wage south without costy technoogy transfer. The cost of technoogy transfer in turn depends on three factors which are the technoogica compexity of this input the absorptive capacity of the host country and the wages of the host country. Our mode refines the resuts obtained in Antràs (2005. We find that. For high-tech goods intra-firm transfer is preferred vis-à-vis outsourcing ony for intermediate range of technoogica compexity of the off-shored input 2. On the other hand for ow-tech goods where the ikeihood of outsourcing is higher in Antràs intra-firm offshore contract is sti possibe for ow range of technoogica compexity. Our mode has poicy suggestions for host countries which aspire to maximie their benefits from the expoding goba production phenomenon. As the wage gap between the source and the host country fas cost considerations for offshoring disappear. ew sources of comparative advantage shoud therefore be created in the host country by subsidiing technoogy investment and higher education to buid higher absorptive capacity. Keywords: Outsourcing Foreign Direct Investment Technoogy Transfer Absorptive Capacity JEL Cassification: D23 F2 F23 L22 L33

3 ection : Introduction In the eary stage of production fragmentation every firm faces the make or buy choice. The importance of this decision is refected in the recent proiferation of internaiation iterature reating to foreign sourcing. Current research on the theory of internationa sourcing has reveaed that the organiationa structure of the mutinationa firm is infuenced by the degree of standardiation of the good factor intensity of the offshored input intensity of the offshored input in fina good productivity of sourcing firms ega framework and market thickness in the host country. One common denominator across a these factors is that they cruciay impact the cost borne by the sourcing firm. urprisingy the cost of technoogy transfer transmission and assimiation that has been centra to the theory of mutinationa corporations (MC since the ast three and haf decade has been overooked with regard to vertica production transfer. This paper fis the gap by incorporating the cost emanating from technoogy transfer in Antràs (2005 mode and thus reating the internaiation decision of a firm to the technoogica compexity of the offshored input and the wages and absorptive capacity of the host country. Technoogy transfer costs are as crucia in a vertica reationship as in horionta FDI or icensing. Particuary in an outsourcing transaction with an unaffiiated suppier assimiation costs are significant. A survey of Indian Business Process Outsourcing (BPO vendors (The Hindu Business Line 2005 reveas that 25.2% of tota wage cost is spent on training of its empoyees to produce inputs of the quaity standards set by its buyer. Arora et a (2000 in their extensive fiedwork on Indian software BPO industry find that a significant amount of speciaied training for a empoyees incuding the skied empoyees is undertaken after recruitment which asts on an average for 2-3 months. In 2004 Caiber Point Business outions Limited a third party BPO service provider to Hexaware Technoogies made substantia investments in technoogy infrastructure ike fiber optic technoogy for the backbone of Loca area etwork De Inte Xeon ervers and etwork ecurity using tonegate Firewa and ID and Tata Honeywe CCTV. We introduce product/firm specific technoogy transfer cost for the offshored fragment in the Antràs (2005 mode thereby introducing heterogeneity across firms manufacturing a singe product or products of a singe firm. The cost of technoogy transfer varies with the technoogica sophistication of the offshored input. If a firm has n products whose offshored input differ in their technoogica compexity then internaiation decision wi vary for each of these products. Aternativey if a firms in the market have ony one product that differ in the

4 technoogy it empoys for producing its offshored input then again firms diverge in their decision to internaie. We differentiate between vertica foreign direct investment (VFDI and internationa outsourcing (IO in the usua Grossman-Hart-Moore way of contractua bargaining power. Using evidence from existing studies on horionta reationships and current offshoring surveys we distinguish between intra-firm and arm s ength production contract with respect to the technoogy transfer cost borne by the sourcing firm in the two aternative modes. The technoogy transfer cost incurred by a sourcing firm in an interna production transfer is substantia and is a part of its reationship specific investment (RI whie the subsidiary manager has itte incentive to invest in technoogy assimiation. On the same note if the offshore production is contracted to an outside suppier then the suppier has to incur a significant proportion of the technoogy transfer costs whie the sourcing firm has itte motivation to bear the costs of technoogy transmission. Technoogy transfer costs are a function of the compexity of technoogy used to produce the off-shored input the absorptive capacity of the host country and its wage rate. ince the technoogy transfer costs borne by the sourcing firm varies with its organiationa mode so naturay internaiation decision becomes a function of technoogica compexity of the input host country absorptive capacity and the wages. evera studies have found a correspondence between the compexity of a product (whoe or the fragment being offshored and the firm s organiationa structure. Based on fied research conducted in the U ingapore UK and India Aron and ingh (2002 find that ower end processes ike data transformation or customer service which embody ess compex skis are outsourced to a third party. On the other hand compex inputs in the goba vaue chain are offshored to an affiiated suppier. Gereffi et a (2003 describe five organiationa forms of fragmented production based on case studies in the bicyces appare horticuture and eectronics industries. They aso concude that high compexity of a good is compatibe with intra-firm transaction uness the suppier capabiity is high. Davidson and McFetridge (985 studied transactions invoving high-tech products of 32 U based mutinationa companies (MC between 945 and 975. Their ogit estimates indicated that the probabiity of internaiation is higher for transactions invoving products with newer technoogy. Most of these studies focus on transaction cost economies to expain the inear reationship between the tendency for vertica integration and higher technoogica compexity of the input. These modes however do not expain the rising demand for medica eectronics designs of digita devices bought from Asian manufacturers by De Hewett-Packard Motoroa and Phiips but sod under their own brand names. 2

5 In Antràs (2005 mode the decision to internaie depends on the intensity of ow-tech input (the offshored input in fina product. A ow intensity of the offshored input impies esser contribution by the suppier in tota surpus and property rights theory dictates that it is optima for sourcing firm to get the residua rights of contro and hence intra-firm reationship emerges. On the other hand a high intensity of offshored input impies greater contribution by the suppier in tota surpus and optimiation resuts in an outsourcing contract. We extend Antràs (2005 mode by intertwining the roe of intensity of the offshored input and technoogica compexity of this input. Given the absorptive capacity of the host country our mode reveas that at high reative home country wages a good with ow intensity of offshored input is more ikey to get offshored through intra-firm transactions ony for intermediate range of technoogica compexity. At ow and high eves of technoogica compexity the sourcing firm is better off engaging an unaffiiated suppier to produce the offshored input. The intuition for this resut is as foows. When the intensity of offshored good is ow then by Antràs (2005 mode there is greater probabiity for production transfer to occur via the VFDI mode. This makes the profitabiity of the sourcing firm in the VFDI mode more sensitive to technoogica compexity thereby choosing an affiiated suppier ony for intermediate range of technoogica compexity. By making an intrafirm transfer the sourcing firm faces ower host country wages and ower contractua costs but higher costs of technoogy transfer vis-à-vis outsourcing. At ow technoogica compexity the distortion in technoogy transfer investment by the suppier is ow whie at high technoogica compexity the savings from technoogy transfer cost forces the sourcing firm to choose an unaffiiated suppier. If the intensity of the offshored input is high then the intra-firm production transfer is preferred to outsourcing at ow technoogica compexity of the offshored input. In both the case however outsourcing is preferred to VFDI at high technoogica compexity of the offshored input. Our resut is empiricay testified in Borga and Zeie (2004 where the voume of intra-firm trade fas with increasing R&D intensity of the affiiate. Our mode highights the possibiity of different trends that can emerge in the organiationa structure of fragmented production. We have observed offshoring reationships of the form Buid-Operate-Transfer whereby a sourcing firm initiay estabishes an outsourcing reationship with an unaffiiated suppier and then at a ater stage takes over the offshoring unit to make it a captive one. Our mode predicts that reationship of this form wi most ikey transpire for the high-tech goods as in case Aviva Pc. Our mode on the other hand aso The evidence from Borga and Zeie (2004 is reevant to this paper ony to the extent that the intra-firm trade they are taking about is the import of inputs from the parent firm to the affiiate for further processing. 3

6 predicts that a transformation from captive to third party as exempified by Genera Eectric India operations but by and arge for ow-tech goods. The mode deivers impications for poicy on technoogy and absorptive capacity in the host country. In rea word we do not observe outsourcing of inputs embodying compex technoogy on a wide scae. If the host country government subsidies technoogy investment by the domestic vendors then there is higher probabiity of an outsourcing contract for technoogicay compex input given the eve of absorptive capacity. Moreover if a host country enarges its absorptive capacity by heavy investment in education to buid its comparative advantage in inputs embodying compex technoogy then they stand to gain by getting more vaue-added work through both VFDI and outsourcing. The paper beyond this point is organied as foows. ection 2 discusses the iterature associated with this research area. In section 3 we deveop the mode and discuss the consequences of introducing technoogy transfer costs in the Antràs (2005 mode. ection 4 discusses the resuts of the mode and section 5 makes a concusion. ection 2: Reated Literature In this section we intend to assimiate internaiation iterature with the iterature on technoogy transfer costs and contract theory. evera studies have examined the horionta mode of entry by a mutinationa in the presence technoogy transfer costs. For exampe Mattoo Oarreaga and aggi (2004 buid a theoretica mode where estabishing a subsidiary is preferred to acquisition of domestic firm if the cost of technoogy transfer is high. This is because a high cost of transferring technoogy to the acquired firm is associated with a smaer cost advantage over domestic firms and a high acquisition price. Based on the information obtained for twenty-six projects of U.. firms in chemicas and petroeum refining and machinery Teece (977 finds that cost due to technoogy transmission can range from about 20%-80% of a project costs. In our paper we propose that the burden of this cost can be shifted by transferring ownership share and therefore impact the internaiation decision of a firm. A major stumbing bock in reating technoogy transfer iterature to internaiation decision in fragmentation is that there is itte research on vertica transfer of technoogy and the reated costs. Therefore we have to rey on studies reating to horionta technoogy transfer. Horionta technoogy transfer modes do not offer much evidence on the cost sharing pattern between the transferor and the transferee or the resource cost of technoogy absorption by recipient firms. Recenty severa instances from BPOs and sourcing firms as cited in the 4

7 introduction confirm our beief that technoogy absorption is aso a substantia proportion of costs and hence may impact internaiation decision of the sourcing firm. Thus we need to rey on a combination of horionta technoogy transfer research information based on case studies of BPOs and captive offshored production units (VFDI reports or surveys pubished in popuar media. The other strand of iterature which we incorporate in our mode is contract theory which has been known to infuence the sourcing firm s decision to internaie since Grossman and Hart (986. Incompeteness of contracts is an inevitabe feature that sets in when transaction happens between two independent entities as in case of Antràs and Hepman (2004 and Antràs ( With incompete contracts these modes show that the bargaining power of the MC is higher with VFDI mode vis-à-vis outsourcing. Besides this difference between VFDI and outsourcing Antràs and Hepman (2004 aso emphasie that the organiationa fixed costs is higher for VFDI. In such an economy more productive firms venture into VFDI. Based on evidence provided by Dunning (993 Antràs (2003 assumes that if a good requires RI in capita and abor then the sourcing firm aways contributes to capita investment. This impies that for a capita-intensive good the sourcing firm contributes more to aggregate surpus and optimaity requires integration with the suppier. Antràs (2005 is aso based on the same principe as Antràs (2003 where the two inputs are abeed as high-tech and ow-tech instead of capita and abor. The sourcing firm makes RI in high-tech input and therefore we expect intrafirm production transfers for high-tech good. However with time as the intensity of high-tech input fas it can be outsourced. This highights that the degree of standardiation is aso higher for a product that is outsourced reative to a product that is produced by a MC subsidiary. To understand these two strands of research together in one framework we spit technoogy transfer costs into transmission and assimiation costs. Transmission cost is the cost to shift codified knowedge ike bueprint formuas management techniques customer ist or tacit knowedge ike know-how information gained from experience which is usuay borne by the transferor. Assimiation cost is the expenditure on R&D by the suppier cost of training workers to adapt to new technoogy or acquiring new technoogy from technoogy market. These costs are typicay borne by the recipient firms uness the host government makes it mandatory for the investing foreign firm to make investment on technoogy absorption or acquisition. For the VFDI mode we woud expect high technoogy transmission costs in proportion to technoogy assimiation cost. This is justified by the high bargaining power of the parent firm in an intra-firm transaction which induces it to invest in costy technoogy transmission but at the 5

8 same time reduces the incentive of the subsidiary manager to invest in assimiation of the technoogy. Our insight is speed out in the survey by Chuang and Chang (993 on foreign affiiates and domestic icensee firms (and joint ventures in Taiwanese pharmaceutica industry. They find that foreign subsidiaries in the host country do not give much importance to the cost of technoogy transfer in their profitabiity anaysis. At the same time icensee firms are very carefu about the cost of technoogy. Their resuts are expained by emphasiing that domestic firms rey on externa market to channe the ingestion of technoogy and thus have to bear pecuniary expenditure and adaptation costs. On the other hand a subsidiary obtains technoogy from its parent firm which precudes any transaction in the technoogy market and therefore has itte incentive to assimiate technoogy or its costs. Teece (977 survey found that technoogies coser to the frontier are transferred to a subsidiary vis-à-vis an arm s ength agent. ince the cost of transferring technoogy is positivey reated to its age a parent firm spends more resources for transmitting technoogy to a subsidiary vis-à-vis an arm s ength unit. UCTC (987 aso finds empirica evidence in the cases of U and German firms where intra-firm technoogy transfer is far more significant than that taking pace between independent parties. This is a refection of the MC preference towards fuy controing the assets transferred to overseas estabishments. ince fu contro of technoogy transferred is not granted in case of an arm s ength contract the MCs may not prefer to bear the costs of technoogy transferred to an arm s ength agent. In the current context technoogy transfer by a sourcing firm to an outside suppier may aso be imited by the fact that a third party vendor (TPV usuay has more than one cient. Therefore if a cient transfers its technoogy to the suppier it undertakes a risk that its technoogy maybe used by the suppier to serve other cients as we. Therefore any rationa sourcing firm wi not transfer its technoogy to its TPV to the extent possibe. Hence the TPV has to invest on its own training and technoogy acquisition contrary to a captive (subsidiary unit which can depend on its parent firm for technoogy. Exampes can be found in the Indian third party BPO companies ike Visuaoft Technoogies Ltd Zensar Technoogies igate Goba outions etc. who have to spend a considerabe proportion of their revenues on technoogy acquisition and absorption. ince a sourcing firm has itte incentive to invest in technoogy transmission in an arm s ength reationship it is therefore the technoogy assimiation and acquisition costs which gain more importance. This is aso supported by the Grossman and Hart theory reating to the bargaining power of the unaffiiated suppier vis-à-vis the affiiated one. Chudnovsky (99 report on north-south technoogy transfer finds that technica assistance to oca suppiers is 6

9 crucia for meeting their performance metric however this is precisey the area where assistance from technoogy providers are missing. Egan and Mody (992 find that in a subcontracting reationship the buyer is wiing to transmit ony the minimum information required to get the product out of the production cyce. If the product must adhere to stringent quaity specifications before being accepted then it is entirey eft to the suppier s discretion to take up the contract get invoved in the manufacturing process and produce the good of requisite quaity at owest possibe cost. Thus the suppier incurs most of the technoogy transfer or adaptation expenditure. Our assumption is impicit in a theoretica mode by Barte et a (2005. An increase in the rate of technoogica change in their mode increases outsourcing because it aows the sourcing firms to use the services of the suppier based on eading edge technoogies without incurring the sunk costs of adopting these new technoogies. The assumption impicit in their anaysis is that it is aways the suppier of the input who subsumes the cost of technoogy in an outsourcing reationship. Assumption : In case of VFDI the parent firm incurs a significant share technoogy transfer costs whie in case of outsourcing it is the unaffiiated input suppier who bears a arge proportion of this cost. ection 3: The Mode: Consider a word with two countries the deveoped north and the ow wage south and a good y produced with abor ony. We borrow demand function and production function from Antràs (2005 mode given by ( and (2 respectivey. Consumer preferences are such that a unique producer i 2 of good y faces the foowing isoeastic demand function: = λ p( y ( i i ( Where p(i is the price of good y(i and λ is a given parameter known to the producer. The fina good y is produced using two inputs high-tech x and ow-tech x with intensity (- and respectivey. y = x h x By assumption the outh acks the capabiity to produce the high-tech input ike R&D. Thus it is ony the ow-tech input that can be offshored. Unike Antràs 3 (2005 the production of the ow- h (2 2 Firms in Antràs (2005 mode are homogeneous and hence the subscript i is not present in his mode. 3 In his mode there is a one to one reationship between abor and output of ow tech input. 7

10 tech input depends not ony on the empoyment of abor but aso on the abor productivity T i where A(i is the firm specific or product specific technoogica compexity of the ow tech input. x = x L x T i T 2 i > 0 T i 2 A( i A( i < 0 ( 0 = 0 T ( 0 < A ( i < T = A high eve of A(i impies a more advanced technoogy within the cass of technoogies avaiabe in the technoogy market 4. In Acemogu Antràs and Hepman (2006 more advanced technoogy is impicity more productive. Aron and ingh (2002 expain the concept of revenue distance where a higher revenue distance of a production stage impies ower contribution to revenue and vaue-addition and hence ower productivity. In their mode inputs using ess compex skis have arge revenue distances and hence ower productivity. To ease interpretation of A(i and how it is different from we can consider an exampe from a consuting firm. A consuting project can be treated as a fina good y produced using two inputs - x and x. A consutant s strategic anaysis of the cient s probem is a high-tech input h an input avaiabe in the north ony whie data anaysis of the cient is a ow tech input which can be offshored. If the consuting project requires reativey ess amount of data anaysis vis-à-vis a consutant s strategic anaysis then the project is intensive in high-tech input and the parameter is ow for such a project. Data anaysis for the project can be done using two techniques varying in their technoogica compexity - A or exce. Technoogica compexity A(i for A is higher than A(i for Exce and accordingy efficiency for data anaysis is higher in A. In the above exampe i refer to the different kinds of consuting projects handed by a firm one that requires sophisticated data anaysis and the other that doesn t. Our mode shows that variation in technoogica compexity of the ow-tech input across projects induces difference in their organiationa modes. Assumption 2: The production of ow -tech input is inear in productivity and abor. x = T i L x ( L 0 = 0 x = x ( L = x ( L L x = x = x x x x L x (3 ( 3 4 The technoogy that we refer to in case of offshoring is typicay a standardied one avaiabe in the technoogy market. 8

11 A(i = 0 means that the technoogica compexity of the offshored input is too ow to justify production in the outh and A(i = impies that high technoogica compexity makes productivity of southern abor high enough to match the productivity of the northern abor. If technoogica sophistication adds to productivity it cannot come without cost. A higher eve of technoogica compexity has to be matched by a corresponding rise in efforts to transmit and assimiate the technoogy. In the exampe of the consuting project discussed above there are costs of running data anaysis in A icense costs and training costs. Exce which has a ower technoogica compexity has a ower technoogica transfer cost vis-à-vis A. This is straightforward and foows directy from Teece (977 observation that more recent technoogy embodying more compex mechanisms require more resources to be transferred whether they are transmitted by the sourcing firm or have to be acquired technoogy transfer cost is a positive function of technoogica compexity. Teece (977 study supports our view that technoogies coser to the frontier embody more compex mechanisms and hence require more resources to be transferred. In addition to this cost Chuan and Chang (993 mode suggests that technoogy transfer cost may aso depend on many factors and most interestingy on the mode of technoogy transfer absorptive capacity of host country and the eve of technoogica deveopment of the host country. To endogenie the technoogy transfer cost with respect to the mode of organiing production fragmentation we use assumption. The buyer (suppier understands that there is itte incentive for the suppier (buyer to invest in technoogy assimiation (transmission in an intra-firm (externa production contract and hence she decides to take a sma fixed payment TT ( TT from the suppier (buyer in ieu of its unverifiabe and insignificant investment in technoogy transfer. In case of an intra-firm (externa production transfer the technoogy transfer cost incurred by the buyer (suppier is given by C defined beow in equation (4 whie in an outsourcing contract (VFDI the buyer (suppier incurs a sma fixed cost TT ( TT that these fixed payments are insignificant and cose to 0. Assumption 3: TT 0 TT 0.. To simpify agebra and without oss of generaity we assume Absorptive capacity of the host country has been cited as crucia to technoogy transfer costs by Baranson (970 Mattoo et a (2005 Teece (977 Pack and aggi (997. Eicher and Kaaitidakis (997 mode the host country absorptive capacity and emphasie the importance of oca human capita necessary to absorb FDI technoogy. Long (2005 deas with 9

12 the issue of training cost in fragmented production. However our mode differs from his in two aspects. We endogenie the technoogy transfer cost and then reate it to the issue of internaiation. Per contra Long (2005 focuses on expaining incompete offshoring to a ow wage nation in the presence of exogenous training costs. In our mode we propose that to produce ow tech input in the south there is an additiona cost of equipping each southern abor empoyed to produce the ow-tech input with the firm or product specific technoogy. The technoogy transfer cost whether transmission or assimiation depends on the host country wages its absorptive capacity and the technoogica compexity of producing the ow-tech input. Assumption 4: The functiona form for technoogy transfer cost is the same irrespective of the mode of organiation of fragmented production. C = C q C < 0 q [ ( w ξ A( i ] 2 C > 0 C A i A i ( ( 2 > 0 Where q(. the efficiency of technoogy transfer is a function of wages in the south w and the absorptive capacity of the host country ξ. Higher wages impy higher productivity of abor 5 and hence efficiency in absorbing the transferred technoogy. At the same time the inherent capabiity of the popuation measured by say the educationa standard is aso a crucia factor in determining the efficiency of transfer. This hypothesis has been supported by Teece (977 study which found a negative reationship between cost of transferring technoogy and host country s absorptive capacity. Usuay a countries maintain some statistics on the human capita figures ike R&D iteracy rate skied abor to unskied abor ratio or investment on human capita formation. Thus a sourcing firm can form a perception of the absorptive capacity of the country hosting its production. By assuming a inear production function for C ( ( ˆ = Ω A i C[ q( w ξ ] x we get a inear cost function. (4 As in Antràs (2005 we consider three possibe organiationa forms: ( Vertica integration in the orth/ Domestic outsourcing (DO (2 Unaffiiated uppier in the outh: Outsourcing and (3 Affiiated uppier in the outh: VFDI. 5 This can aso be rationaied by using the efficiency wage theory; however we choose not to use this terminoogy as wages in our mode are exogenous. 0

13 Vertica Integration or outsourcing in the orth Antràs (2005 assumes that vertica integration and domestic outsourcing in the north are equivaent because of compete contract enforcement. To maintain this supposition in our mode we need to additionay assume that a firms in the north have identica absorptive capacity and hence require no technoogy transfer to produce the ow-tech input. Demand and production function is given by ( and (2 respectivey. Assuming that one unit of abor produces one unit of the input the profit of the northern firm is given by: Π = λ ζ x ( h x w x h w x ( Where ( ( ζ = This case is exacty the same as in Antràs (2005. Profit maximiing price and equiibrium profit is given by: w p ( = (5a Π w = λ ( (5b Internationa Outsourcing- Unaffiiated suppier in south Assumption and 3 together impy that the technoogy transfer cost in an outsourcing reationship is borne by the suppier. The RI for the sourcing firm comprises of its commitment to producing the high tech input ony. Assumption 5: As in Antràs (2005 competition among southern suppiers of ow-tech input drives their profit to ero. A transfer payment T from the suppier to the sourcing firm has to be aowed for such that the profit of the outsourcing partner is driven to ero. The profit function for the sourcing firm outsourcing to a TPV in the south is: Π o = φ R w = φ x + T ( ( λ ζ x x w x + T h h h Where R denotes the tota revenue from the reationship and φ is the share of the sourcing firm in the tota vaue of the reationship. It is aso a measure of the bargaining power of the sourcing firm. In Antràs (2003 mode RI by the suppier (sourcing firm is in abor (capita investment whie in Antràs (2005 it is the resources committed to produce the ow tech (high-

14 tech input. Besides the RI in ow-tech or high-tech input production our mode has an additiona component of RI which is incurred by the suppier or the sourcing firm contingent on the organiationa mode. The suppier (sourcing firm has to make RI in technoogy transfer costs in case of outsourcing (VFDI. The unaffiiated suppier maximies: Π Π o o = = = ( φ R w x C L T x Using ( 3 and (4 we get: L T ( ( φ( λ ζ x x w L Ω i Cˆ [ q( w ξ ] h x w + Ω i Cˆ ( ( [ q( w ξ ] ( φ λ ζ x x The term ( w + Ω A i Cˆ q w T A( i h ( [ ( ξ ] T ( i x x T is the Average Efficiency cost (AEC adjusted for productivity for producing x the ow tech input. Profit maximiation of the two agents and setting T so as to make the suppier break even eads to the foowing expression for the sourcing firm s ex ante profits and profit maximiing price in IO equiibrium: Π o = λ Po = φ [( ( ] ( w ( AEC + ϕ 2 φ ( φ ( w ( AEC (6a ( φ The profit maximiing price in Antràs (2005 when outsourcing is chosen is given by: p ( w ( w = φ ( φ Our price equation (6b is anaogous to the above equation except that the southern production cost is augmented to incude the technoogy transfer costs adjusted for productivity enhancement due the sophistication of technoogy. Let = Θ Π Π O Internationa outsourcing is preferred to domestic outsourcing in north if Θ < that is (6b (6c 2

15 w AEC φ > φ > L ( φ ( + φ ( 2 φ = φ ( + φ ( 2 Assumption 6: To further simpify agebra we assume that ˆ [ q( w ξ ] w and ξ. That is Cˆ [ q( w ξ ] = w. C( ξ ~ Using the above assumption w w Θ < if φ φ C is ineary separabe in AEC = ω > L ( φ. (7 w Where ~ AEC + Ω i C ( ξ ( = w T A( i ω and ( φ L are given for given A(i so we examine the behavior of. ( AEC w = 0 if: A( i i Ω A( i η = η Ω A( i Ω T i + Ω i T = A( i ~ i C( ξ A( i T + ~ i Ω i C( ξ AEC w That is the easticity of the cost of technoogy transfer with respect to technoogica sophistication A(i is equa to the weighted easticity of productivity of southern abor with respect to the technoogica compexity weighted by * technoogica compexity at which (8 hods to be A io. Proposition : The function reaches minima at (8 ( ( + ~. Lets define the Ω A( i C ξ * ( AEC A io when w = 0. A( i Mathematicay the cost function is convex with respect to the technoogica compexity and the productivity function is concave then second order conditions confirm our assertion that ( AEC is a convex function. Intuitivey the proposition impies that at ower eves of technoogica compexity the increment to productivity is higher than the increment to technoogy transfer costs. At higher eve of w 3

16 technoogica compexity the increment to technoogy transfer costs is much arger than its contribution to increasing southern productivity. One can justify this because had the technoogy transfer cost not be prohibitivey high one woud have observed the offshoring of advanced stages of production as we. ~ Hence + Ω i C( ξ fas for ow eves of technoogica compexity A(i < T A( i A > * ( i A i o ( it rises. Equation (7 can be depicted by figure 6. * A i o. For Reative wage L ( φ. + Ω T ~ ( A ( i C ( ξ ( A( i ω Domestic Internationa Outsourcing Outsourcing Domestic Outsourcing 0 A ( i o * A io A ( i o A(i Technoogica compexity Figure : Tradeoff between northern production and Internationa Outsourcing The sourcing firm stands to gain from IO vis-à-vis DO due to ower host country wages whie it stands to ose due to costs from contractua distortions and suboptima RI in technoogy transfer. Any technoogica compexity beow A (i impies a greater distortion due to incompete o contracts than it saves costs due to cheap southern abor. imiary any A(i above A (i o increases technoogy adaptation cost more than it adds to productivity and aso increases distortions due to incompete contracts. Proposition 2: Ony at intermediate eves of technoogica compexity is internationa outsourcing preferred to domestic outsourcing. At ow and high eves of technoogica sophistication domestic outsourcing dominates internationa outsourcing. Proposition 3: To host outsourcing contracts with the fu range of technoogica compexity the host country shoud possess a minimum eve of absorptive capacity. Anaogousy the range of 6 The curvature and sope of the curve depends on parameters ike φ and ξ 4

17 technoogica compexity avaiabe for outsourcing can be increased if the absorptive capacity is raised. If we ook at equation (7 we can derive that A( i * > 0 for A ( i > A i o and A( i < 0 for ξ ξ A < * ( i A i o. Thus outsourcing expands at both ends with a rise in absorptive capacity. Vertica Integrated suppier in the outh or VFDI We retain Hart-Moore (990 premise that the sourcing firm has a right to higher share in surpus (bargaining power in an intra-firm transaction vis-à-vis a market transaction. This assumption is given mathematicay in Antràs (2005 as: φ ( δ = δ + φ > φ (9 Where δ is the proportion of output expropriated by the sourcing firm if the manager of the subsidiary is fired. In Antràs (2005 the expression for the profit maximiing price in case of VFDI is anaogous to equation (6c with φ repaced by φ. In our mode the mutinationa firm assumes the technoogy transfer costs to train the abor in a VFDI contract. As in outsourcing T is set such that competition among suppiers drive their profit down to ero. The profit function of a mutinationa is given by: Π f = φ R w = φ x [ ] i Cˆ q( w ξ + T ( ( λ ζ x x w x Ω i Cˆ [ q( w ξ ] h h Ω h L x T x i RI on the part of the integrated suppier comprises of its resources committed to produce the ow-tech input ony with insignificant expenditure on technoogy absorption. The subsidiary manager maximies: Π ( φ R w x T f = + T First order conditions for maximiation of the MC profits under VFDI yieds price: P f = ( w φ T w ( φ i In case of VFDI the presence of technoogy transfer cost does not distort prices since they are incurred by the mutinationas whie the amount of southern abor empoyed is determined by the 5

18 subsidiary. Thus technoogy transfer cost is ike a fixed cost to the mutinationa. Equiibrium profit of the MC is given by: Π f = λ ( + φ( 2 [ ] ( ( ˆ A i C q( w ξ ( w φ ( φ i Ω ( φ Assumption 7: Let φ = ½ as in Antràs (2005 w T w (0 Using assumption 6 and 7 we get that for VFDI to yied a positive profit is 7 : ( + δ + 2 ~ ( δ C ( ξ 2 δ A ( i < Ω = b ( Per contra Π o >0 for a range of A(i. This resut comes by because in case of outsourcing it is the unaffiiated suppier that makes RI in technoogy transfer whie the resuting gain in productivity is aso enjoyed by the sourcing firm. In case of VFDI the MC makes RI in technoogy transfer whie both parties enjoy the productivity gain. Proposition 4: The sourcing firm stands to ose from VFDI (in absoute terms if the technoogica compexity of the ow-tech input is higher than a critica eve b defined in (. To evauate the reative prevaence of VFDI vis-à-vis DO we compare (0 with (5b the respective profit functions of the sourcing firm in the two aternative modes of organiation. Let = Θ 2 Π Π f Then VFDI is preferred to production in north if Θ < 2 that is ( φ A( i ξ ω > L2. (2 T ( φ A( i ξ φ = φ i ( + φ( 2 Ω i Cˆ q( w ξ φ L 2 ( w [ ] φ 7 For >½ or <½the numerator is aways positive. 6

19 Let Ψ = L ( φ A( i ξ. Ψ i 2 T i 2. i T i L = i L i 2 i [ T ] 2 i i T Let the eve of technoogica compexity at which + Ψ i * = 0 be A i f. The first term in the above tota derivative denotes the effect of increase in technoogica compexity on technoogy transfer costs which is positive and the second term gives its effect on abor productivity due to a sma increase in A(i. As in assumption 6 we suppose that at ower eves of technoogica compexity the second effect dominates that is for A(i < * A i f an increase in technoogica compexity increases the profitabiity of the sourcing firm by increasing productivity of the ow tech input produced in the south than can be offset by an increase in technoogy transfer cost. * A i f Proposition 5: For reasons corresponding to proposition for A(i < Ψ i < 0. It reaches minimum at say * A i f and then it rises. With this proposition equation (2 has been depicted graphicay in figure 2 8. Reative wage L 2 ( φ A ( i ξ. T ( A ( i ω Domestic Outsourcing Vertica FDI Domestic Outsourcing * 0 A ( i A i f A ( i f b A(i f Technoogica compexity Figure 2: Tradeoff between northern production and Vertica FDI 8 The curvature and sope of the curve depends on parameters ike φ ξ and δ 7

20 The prevaence of DO beow outsourcing. A (i and above f A (i can be expained as in case of internationa f Proposition 6: Ony at intermediate eves of technoogica compexity is vertica FDI preferred to domestic outsourcing. At ow and high eves of technoogica sophistication domestic outsourcing dominates vertica FDI. Proposition 7: To host VFDI contracts with the fu range of technoogica compexity the host country shoud possess a minimum eve of absorptive capacity. Anaogousy the range of technoogica compexity avaiabe for intra-firm contracts can be increased if the absorptive capacity is raised. If we ook at equation (2 we can derive that A ( i * > 0 for A ( i > A i and f ξ A < * ( i A i f. Thus VFDI expands at both ends with a rise in absorptive capacity. A ( i ξ < 0 for Comparing Internationa Outsourcing with VFDI To compare the profit functions of the sourcing firm in the two aternative regimes of organiation of fragmented production we ook at equation (6a and (0. Let Θ Π = Π o f Θ = ( + φ( 2 ~ ( + φ( 2 + ( φ δ (2 ( δ Ω i C( ξ [ ( ( ] δ + φ ~ ( ( C( ξ ( δ δ + φ( δ + Ω A( i φ Using equation (9 and φ = ½ we can simpify the above equation to: Θ = 2 2 ~ ( + δ ( 2 ( δ Ω i C( ξ 2 ( ( ( ( + Ω( ( ( ~ A i C ξ + δ δ To have meaningfu comparison between VFDI and internationa outsourcing we need to hod ( + δ + 2 ~ ( δ C ( ξ 2 δ A ( i < Ω = b and for A ( i > b either IO or DO exists but not VFDI. 8

21 Let us now ook at the partia derivative of Θ with respect to A ( i. Θ A i ( Θ A i ( Ω i ~ ( ( ( δ = Θ C ξ ( ~ A i + Ω( ( A( i C ξ ~ + 2 ( + δ ( 2 ( δ Ω i C( ξ < 0 that is as technoogica compexity increases the profitabiity from VFDI increases if: 2 ( δ 2δ ( ~ (( + ( δ C( A i < Ω ( = a (3 ξ Ceary a<b ese production wi never be offshored via intra-firm contract. Θ A i ( > 0 that is as technoogica compexity increases the profitabiity from outsourcing increases if if ( δ 2δ ( Ω ~ (( + ( δ C ( ξ ( + δ + 2 ~ ( δ C ( ξ 2 2 δ < A( i < Ω (4 Hence a < A( i < b then Θ A i ( > 0 Proposition 8: If in equiibrium internationa outsourcing occurs for technoogica compexity beow a then a sma increase in technoogica compexity sti ess than a wi induce a regime switch from internationa outsourcing to VFDI. Proposition 9: If in equiibrium VFDI occurs in the range a < A( i < b then an increase in technoogica compexity in this range wi switch the organiationa form to internationa outsourcing. Thus at higher eve of technoogica compexity the organiationa form of fragmented production is ikey to be an externa one. We can consider two reevant and interesting cases that come up with this mode. For < ½ ( δ ( + δ < and technoogica compexity + A ( i ~ 2 > Ω the function 2 C ( ξ Ψ representing a MC s profitabiity with VFDI is more sensitive to A(i than the simiar function 9

22 [ L ( φ. AEC ] for outsourcing 9. It is intuitive because at ow 0 there is higher probabiity for the sourcing firm to consider a VFDI contract vis-à-vis outsourcing (Antràs Then if we introduce technoogy transfer costs the sourcing firm has to take into account these costs to offshore the input. Transfer costs are a convex function of technoogica compexity and hence raise the sensitivity of the MC in an intra-firm transfer vis-à-vis externa contract at ow. In figure 3 we consider the case of ow and therefore we have drawn the VFDI curve steeper than the IO curve. At high reative northern wagesω if internationa outsourcing equiibrium occurs beow technoogica compexity a then there is a tendency to switch to VFDI (equation 3 proposition 8. The figure shows that ow eves of technoogica compexity beow A (i goes with DO ony. For the range of technoogica compexity between o A (i and P o we may have IO and between P and a we have VFDI. Again between a and because Θ A i ( > 0 for A ( i > a. A (i we have IO o It is observed that the rate of growth of wages in countries which host offshoring contracts is very high at about 20% per annum. What do we expect of the reative prevaence of the two organiationa forms of internationa production sharing? In figure 3 we show the impact of fa in northern reative wages from ω to ω. The bod ines define the new range of technoogica compexity for internationa production sharing. We observe a fa in off-shoring at the two ends of technoogica compexity. If the fa in wages is not arge the region for VFDI may not be impacted at a whie on the other hand if the fa in reative wages is very arge internationa outsourcing may be competey wiped out. Therefore with a fa in north-south wage differentia one moves from mutipe regime switch situation where the regime switches twice as the technoogica compexity increases to a unique regime switch situation where the regime switches from VFDI to IO and finay if the wage differentia is ow enough we have a situation of a pervasive VFDI. The resut is intuitive because a fa in reative northern wages represents a oss in comparative advantage of the ow wage south. As we have observed rising wages in offshoring destinations ike India our mode suggests to buid a new source of comparative advantage or ese witness its growth and empoyment emanating from the offshoring industry fa. One such source of comparative advantage has been discussed in Acemogu Antràs and Hepman (2006 where better contracting institutions can infuence the eve of production sharing by impacting reative productivity of the fina good sector. There can 9 Assuming that this critica A(i is ower than technoogica compexity. 0 We assume that for >½ the good to be high-tech and for <½ the good is ow-tech. A ( i we draw the VFDI function steeper than the IO function at a eves of f 20

23 be yet another source of comparative advantage which is the host country s absorptive capacity and technoogy expertise. A high eve of absorptive capacity and proficiency in technoogy can sustain a higher technoogicay sophisticated good by owering the cost of technoogy transmission and hence widen the range of off-shoring. Proposition 3 and 7 advocates this point. VFDI : L 2 ( φ A ( i ξ. T ( A ( i Reative wage DO IO VFDI IO Domestic Outsourcing (DO IO L : L ( φ. + Ω T ~ ( A ( i C ( ξ ( A( i ω ω' 0 A ( i P Q a R o A ( i o b Technoogica compexity A(i Figure 3: Possibiity of mutipe switches for high-tech good For > ½ aong with other parametric restrictions on δ and the VFDI profitabiity function is ess sensitive to technoogica compexity of the offshored input than the sourcing firm s profit function in IO. The intuition is again derived from Antràs (2005. The probabiity for outsourcing is higher for high and hence the cost of the suppier assumes greater importance and making it more sensitive to technoogica compexity. Again for A(i>a (from equation 4 the profit from IO is expected to be higher vis-à-vis VFDI. This impies that we woud expect a ow-tech input to be contracted internay for ower range of technoogica compexity but externay for higher A(i. In this case if reative northern wages fa to a eve sayω then it is VFDI which is competey wiped out and we observe ony internationa outsourcing regime. 2

24 Reative wage ω DO VFDI Internationa Outsourcing VFDI Domestic Outsourcing IO : L 2 : L ( φ ( φ A ( i ξ. T + Ω. T ( A ( i ~ ( A ( i C ( ξ ( A( i ω' 0 A ( i f a A ( i o b Technoogica compexity A(i Figure 4: Possibe Regimes with ow-tech good ection 4: Discussion A point worth noting is that in both the cases above it is aways outsourcing which is preferred at higher eves of technoogica compexity of the offshored input irrespective of the vaue of. This is because at higher eves of A(i the high cost of technoogy transfer is a strong disincentive for the sourcing firm to undertake an intra-firm production transfer. In case of VFDI the MC makes RI in technoogy transfer whie the ensuing productivity gain is shared by the suppier as we. As the technoogica compexity crosses a threshod the MC is no onger wiing to bear the cost of technoogy transfer and is better off sharing a arger part of the surpus in return for the unaffiiated party s RI in technoogy transfer. Our resut shares a simiarity with Barte et a (2005. An increase in the speed of technoogy in their mode encourages domestic outsourcing vis-à-vis intra-firm production transfer. Our mode proposes that a firm with higher compexity of technoogy wi aways choose to outsource it provided the host country has a threshod eve of absorptive capacity. The forces driving simiar resuts in the two modes are however different. In the cosed economy mode of Barte et a (2005 acceeration in the pace of technoogica change raises the technoogy adoption costs of the fina good firm and hence increases the per-period unit cost of producing in-house. This shifts the demand for outsourcing outwards irrespective of its service price because it aows firms to 22

25 use services based on eading edge technoogies without incurring the arge and recurrent fixed costs of adopting these new technoogies. Perhaps a widey hed notion is that firms do not outsource the production of technoogicay compex inputs. The trend to buy technoogicay compex inputs from unaffiiated suppiers is not competey absent though. For instance De contracts out the design for notebooks Persona Computers digita teevisions. Hewett-Packard seeks externa assistance to deveop servers and printers. Motoroa purchases designs for its cheapest phones from unaffiiated suppiers. These firms acquire compete designs of digita devices from Asian deveopers and modify them to suit their own specifications and finay stamp their own brand name. The trend is fast spreading from eectronics sector to navigation systems pharmaceutica and even consumer goods. For exampe Boeing is working with HCL Technoogies an Indian third party service provider to co-deveop software ranging from navigation systems and anding gear to the cockpit contros. imiary 20% of Procter & Gambe s new product ideas come from externa source. The reason for outsourcing compex technoogica products within the basic stage of production can aso be rationaied by the fact that these products require speciaied skis and knowedge which can be offered by ony a broad network of speciaists. That is perhaps the reason why many pharmaceutica companies have begun to outsource basic research. To ensure that the possibiity of outsourcing at higher technoogica compexity does not remain a theoretica opportunity ony we need a dynamic invovement by the host country in gobaiation. The mode thus deivers impications for the need of an active technoogy poicy in the host country. ince the sourcing firm is not ikey to make an intra-firm production contract at high eves of technoogica compexity the host country government shoud subsidie the domestic vendors technoogy investment so as to enhance its overa participation in the goba production. ome eading companies have simutaneousy adopted a mix of captive and outsourced services wherein some of the more compex and core processes are being handed by the captive unit. Credit card companies for instance have compex technoogies in pace to anaye customer behavior. If a country has ow absorptive capacity its third party outsourcing service providers may get trapped in ow vaue-add work as is depicted in figure 3. Given the possibiity of mutipe switches it is possibe for a TPV to jump to high vaue add and technoogicay compex work if the country enhances its absorptive capacity through investment in human In contrast to Antràs (2005 Acemogu Aghion and Ziibotti (2005 show that firms coser to technoogy frontier (intensive in high tech input are more ikey to outsource to focus on R&D. It is ikey that the inputs of a high-tech good are more technoogicay compex than the inputs of a ow tech good. Thus their mode is aso capabe of generating a resut simiar to ours that more technoogicay compex inputs are outsourced. 23