Subcontractors for tractors: Theory and evidence on flexible specialization, supplier selection, and contracting☆
Introduction
Many industries, particularly in developing countries, are characterized neither as vertically integrated firms nor as a set of independent buyers and suppliers but as networks. Suppliers provide specialized inputs to several buyers selling related but different products and buyers have more than one supplier for the same input. The resulting investment pattern on the part of suppliers has been one of flexible specialization and is considered to be an optimal response to demand uncertainty and costly capacity-building or inventory-holding.1 In such environments, there is considerable variation in the terms of the contracts faced by a set of sellers who differ in terms of how specific their assets are with respect to the main buyer.2
While there is a large literature on the determinants of the boundaries of the firm that highlights the importance of relationship-specific investments, we know very little about how relationship-specific investments affect contracts when the boundaries of the firm are given.3 Moreover, the existing literature treats specificity as being driven purely by technology. In a network or cluster setting, given that investment is characterized by flexible specialization, the degree of asset specificity with respect to any particular buyer is also partly a matter of choice.
In this paper we address these questions both theoretically and empirically. We use primary data to examine a particular buyer–seller network in Pakistan characterized by high uncertainty, weak contracting, and costly capacity building. This description fits the industrial sector of most developing countries quite well, although the relevance of the framework is not limited to these countries. Our initial analysis of the network reveals several interesting and somewhat puzzling facts. We find that there is substantial variation in how suppliers are treated – prices differ by as much as 25% and quantities by a factor of three across different suppliers supplying the same product in the same year. Upon further examination we find that surprisingly, it is the “tied” suppliers (those that choose higher levels of specific investments) that are treated as second preference suppliers, not only in terms of receiving more unstable and lower orders but lower prices as well.4
We then develop a theoretical model to understand these findings. We take the existence of buyer–seller networks as given and address two main questions: do suppliers of the same product who differ in how specific their assets are with respect to the same buyer, receive different prices and distribution of orders? What governs the variation in how specific a supplier's assets are in relation to one buyer?5
Our theoretical model has three key ingredients. First, relationship-specific investments (as opposed to general investments) increase the surplus within the relationship but lower the flexibility of a seller to cater to the outside market. This decrease in flexibility is costly when demand is uncertain. Second, suppliers are of different “types” i.e., ex ante qualities. Holding the level of investment constant, higher types generate a higher level of surplus both within the relationship and in the outside market. Third, higher types are more likely to find a buyer in the outside market. Because of the first and the third features, higher types face a greater marginal cost of undertaking relation-specific investments. Thus for the same level of orders, higher types invest less than lower types. Therefore the model predicts that even if the assembler prefers high types in general, some low type vendors might be kept as marginal suppliers because of their greater willingness to invest in assets specific to the buyer, especially when demand is very uncertain.
The model generates further implications that are examined using the primary data set we collected on a sample of annual vendor-product specific contracts between Millat Tractors Ltd. (MTL) and its sellers (locally referred to as vendors) for a period of 10 years. The MTL data is attractive for several reasons. First, the focus on a single large buyer ensures that the comparison between contracts is meaningful. Second, we have detailed contractual outcomes including prices paid to a supplier for a given product (tractor part) and quantities scheduled every quarter for each product (henceforth “part”) from the vendor for over a decade. Finally, given the assembler has multiple vendors supplying the same part, we are able to make cleaner comparisons by contrasting contracts between two vendors with different degrees of specificity but which supply the same part. Unlike a majority of the empirical literature on relationship-specific investments, our comparisons are therefore not confounded with other effects that may be specific to a product yet not related to relationship-specificity. Our measure of specificity is the vendor's response to what fraction of its machinery will go to waste if MTL stops buying from it and this measure is confirmed through various means such as relating it to a vendor's production processes.
In addition to the differential treatment results, we find that tied vendors indeed have lower unit production costs, though this makes it even more surprising that they are treated as second preference vendors. However, cost is not the only consideration of the assembler. It cares a great deal about timely and defect-free delivery. This suggests that, as in the model, ex ante quality (type) differences between vendors can explain why MTL does not treat the cheaper tied vendor as its first preference vendor. Indeed, further empirical results show that vendors with greater asset-specificity perform worse both in terms of timely and defect-free delivery. In terms of our theoretical model, low type vendors act as capacity buffers because they are more willing to both undertake higher levels of specific investment and face greater uncertainty.
Our work is related to the theoretical literature on property-rights. The key distinction is that in our setup only the vendors undertake investments, and so optimal ownership is not the key question. Also, unlike our model, in this literature, specific investment is purely technology driven, and firm heterogeneity and selection issues are not emphasized.
Our work is closely related to the empirical literature on asset specificity and how it affects the nature of contracting.6 In this literature, the effect of asset specificity is typically shown either on contract duration (e.g., Joskow, 1987) or on certain contract provisions (e.g., Lyons, 1994 on the use of formal contracts, Gonzales et al., 2000 on extent of subcontracting, Woodruff, 2002 on the likelihood of vertical integration, and Baker and Hubbard (2004) on the pattern of asset ownership). While we too study the effect of asset specificity on contracts, our focus is on prices and quantities of orders, and their variability over time and across subcontractors. More generally, our work is related to the recent empirical literature on contracting where controlling for unobserved heterogeneity is an important theme (Chiappori and Salanie, 2003).
Finally, we view our work as a contribution to the emerging literature on contracting and organizational choice in the industrial sector in developing countries. The presence of significant uncertainty and transactions costs in these economies provide a fertile ground for testing many predictions of the theory of contracts and organizations. While a rich and growing empirical literature on contracting and organizational choice exists in the context of agriculture in developing countries, there is relatively little work in the context of industry (exceptions include Banerjee and Duflo, 2000, McMillan and Woodruff, 1999, Banerjee and Munshi, 2004).7
The plan of the paper is as follows. In Section 2 we discuss the key features of the environment drawing on several case studies on subcontracting in buyer–seller networks from different parts of the world. In Section 3 we present the theoretical model. In Section 4 we examine the case of MTL and is supplier network and interpret our empirical findings in terms of the theoretical model. Section 5 concludes.
Section snippets
Buyer–supplier networks: key features and puzzles
In this section we discuss key features of a particular buyer–seller network in Pakistan. The network presents some potential puzzles that motivated the model developed in the next section. We also document that these features hold more generally in other buyer–seller networks.
The environment
The model focuses on the relationship between a single assembler, and its vendors. Everyone is assumed to be risk-neutral. The assembler is unable to make some parts in-house and needs to outsource. Vendors produce parts that are then converted into output using some technology by the assembler. For simplicity, we assume that there is only one part that the assembler needs from vendors. The production technology involves the assembler using one unit of this part to produce one unit of the final
Empirical evidence
The motivating empirical results presented in Section 2 showed that not only is there supplier multiplicity and differential treatment of MTL suppliers but that somewhat surprisingly, suppliers that choose to invest specifically for MTL receive less favorable treatment in terms of price and quantity levels and the variability of orders. The model developed above presented a theoretical framework which shows that this puzzle can be resolved once we allow for ex ante supplier heterogeneity. It
Conclusion
The relationship between a tractor assembling firm in Pakistan and its subcontractors offers important insights about contracting and asset specificity within a buyer–supplier network. The presence of demand uncertainty makes undertaking relationship-specific investments costly on the part of suppliers. This cost is likely to be more, the more able and versatile the supplier. Therefore, there is a chance for low quality suppliers to survive because of their greater willingness to undertake
References (28)
Manufacturer–supplier relationships in Japan and the concept of relationship-specific skill
Journal of the Japanese and International Economies
(1989)Factors affecting supplier quality performance
Journal of Operations Management
(1997)- et al.
Quality management practices and their relationship to buyer's supplier ratings: a study in the Korean automotive industry
Journal of Operations Management
(2001) Non-contractible investments and vertical integration in the Mexican footwear industry
International Journal of Industrial Organization
(2002)- Amir, I., Isert, K., Khan, S., 1995. Note on the local suppliers to the tractor industry in Pakistan, Mimeo. Lahore...
- et al.
Subcontractors for Tractors: Theory and Evidence on Flexible Specialization, Supplier Selection, and Contracting, Working Paper
(2004) - et al.
Contractibility and asset ownership: on-board computers and governance in U.S. Trucking
Quarterly Journal of Economics
(2004) - et al.
Reputation effects and the limits of contracting: a study of the Indian software industry
Quarterly Journal of Economics
(2000) - et al.
How efficiently is capital allocated: evidence from the knitted garment industry in Tirupur
Review of Economic Studies
(2004) Comparisons of the competitive position of Canadian, Mexican, and U.S. suppliers
The Journal of Supply Chain Management: A Global Review of Purchasing and Supply
(2000)
Testing contract theory: a survey of some recent work
The expanding buyer–supplier partnership for product and process technology development in the Korean automobile industry
International Journal of Automotive Technology and Management
Cited by (24)
The dynamics of outsourcing relationships in global value chains: Perspectives from MNEs and their suppliers
2019, Journal of Business ResearchCitation Excerpt :But the market transaction costs for coordinating activities are high, there would be an incentive to internalize the exchange relationships through vertical integration. This might be the case when one party has made investments in assets that are specific to the relationship, and which increase the level of dependence of that party on the another (Andrabi, Ghatak, & Khwaja, 2006; Gietzmann, 1996; Scherrer-Rathje, Deflorin, & Anand, 2014). The discussion above suggests that potential power asymmetries depend upon a variety of contextual factors, and that they are likely to change over time as those factors themselves change with consequent effects for the power asymmetries between the parties.
Multi-sourcing as an entry deterrence strategy
2013, International Review of Economics and FinanceA study of using RST to create the supplier selection model and decision-making rules
2010, Expert Systems with ApplicationsInput specificity and global sourcing
2010, Journal of the Japanese and International EconomiesRent a womb: Surrogate selection, investment incentives and contracting
2009, Journal of Economic Behavior and OrganizationCitation Excerpt :Also, our work is closely related to the literature on asset specificity and its implications for the nature of contracting (for a review of the issues, see Tirole, 1999).2 Our model is based on a paper by Andrabi et al. (2006, henceforth AGK) that studies how tractor parts are subcontracted in Pakistan. AGK analyzes how much relationship-specific investment each supplier makes for the buyer, who gets higher and more stable orders and at what price.
Supplier Networks at the Integrated Peripheries of the Automobile Industry: The Case of Keiretsu Suppliers in Thailand
2023, Tijdschrift voor Economische en Sociale Geografie
- ☆
This paper is dedicated with great admiration to Pranab Bardhan, whose pioneering work in development economics combining economic theory and econometric analysis with rich institutional detail has been an inspiration to us all. We thank Abhijit Banerjee, George Baker, Robert Gibbons, Oliver Hart, Alexander Karaivanov, Michael Kremer, Rocco Macchiavello, W. Bentley MacLeod, Kaivan Munshi, Canice Prendergast, Tomas Sjöström, Jeffrey Williamson, Chris Udry, two anonymous referees, the editor, Mark Rosenzweig and several seminar audiences for helpful feedback. We are grateful for all the support and information provided by the Lahore University of Management Sciences and Millat Tractors Ltd. All errors are our own.