Services FDI and Manufacturing Productivity Growth: There is a Link

Transcription

1 Services FDI and Manufacturing Productivy Growth: There is a Link Ana M. Fernandes a The World Bank Caroline Paunov b Queen Mary, Universy of London Preliminary and Incomplete This version: April 15 th 2008 Abstract This paper examines the effects of FDI in services on productivy growth of manufacturing plants in Chile between 1992 and We estimate an extended production function where plant output growth depends on input growth and a measure of services FDI. The novelty of our approach is the use of plant-level services usage measures as weights for services FDI. We find a posive effect of services FDI on productivy growth. Our evidence is robust to different productivy measures and tests and not driven by any specific industry. We find no evidence of different effects across small and large plants. Keywords: Productivy Growth, Services liberalization, Foreign Direct Investment, Chile. JEL Classification codes: D24, L8, L9, F2 a Ana Margarida Fernandes. The World Bank. Development Research Group H Street NW, Washington DC, afernandes@worldbank.org. b Caroline Paunov, Queen Mary, Universy of London, Economics Dept, London E1 4NS, c.paunov@qmul.ac.uk. The findings expressed in this paper are those of the authors and do not necessarily represent the views of the World Bank.

2 1. Introduction Foreign direct investment (FDI) inflows into the services sector experienced a boom during the 1990s. 1 By 2002, services accounted for 60% of the world stock of FDI, a four-fold increase relative to the same share in 1990 (UNCTAD, 2004). The main recipients of FDI have been prof-seeking producer services which range from networkintensive services such as electricy, telecommunications and transport to finance and business services. These services are characterized by the facilating and intermediating role which they play for downstream user firms (Francois, 1990). The qualy of producer services is a crucial component of a country s business environment affecting manufacturing firms operations and growth. Therefore, improvements in the performance of producer services are likely to benef all economic sectors. A potentially powerful means to achieve such improvements is FDI in services sectors which can increase the qualy and variety of services available and lower their cost. Moreover, FDI in services may also benef manufacturing firms through spillovers of management, organizational, marketing or technological knowledge from the interaction wh foreign services suppliers. These benefs seem particularly relevant for manufacturing firms in emerging and developing economies as the services inputs and knowledge would otherwise not be available domestically. Markusen (1989) and Rivera- Batiz and Rivera-Batiz (1992) formalize theoretically the benefs from the liberalization of trade in the business services sector. These benefs result from s complementary 1 This boom in mode 3 of trade in services according to the GATS nomenclature (i.e., the case where the services provider moves to the location of the consumer) occurred despe the strong restrictions to FDI in services that remain in place across a large number of countries, particularly in East Asia (Gootiz and Mattoo, 2007). 1

3 wh the liberalization of trade in goods or from specialization and increases in the productivy of sectors and firms that use those services. Despe the relevance of this topic, the effects of backward linkages resulting from the openness of producer services to FDI on manufacturing firms have not been widely documented (Hoekman, 2006). 2 This paper attempts to fill this gap by addressing the following question: did the penetration of FDI into producer services sectors in Chile affect the productivy growth of manufacturing plants between 1992 and 2004? Chile is an interesting economy to study as s services sector has been open to FDI since the late 1980s and received large FDI inflows during the 1990s. Furthermore, the long duration of our manufacturing plant-level panel dataset makes a particularly appropriate set-up to detect the presence of effects from services FDI on plant productivy growth. Our empirical framework estimates an extended production function where plant output growth depends on input growth (including the growth in services inputs) and on a measure of services FDI penetration weighted by plant intensy of services usage. Our analysis relies on the following central and intuive argument: if FDI into services sectors has productivy growth effects, then one should expect plants that use services more intensively to benef more. In order to identify a causal effect of services FDI on manufacturing plant total factor productivy (TFP) growth, we use two-period lagged measures of services FDI penetration weighted by plant-level services usage. Moreover, we control for unobserved fixed differences in productivy growth across plants, 2 There is, however, a lerature that measures the effects of backward linkages from the use of producer services on aggregate productivy and on economic growth, but does not consider the liberalization of services (e.g., Roller and Waverman (2001) on telecommunications, Fernald (1999) on roads, Rajan and Zingales (1998) on financial services). 2

4 observable plant characteristics, industry-level time-varying heterogeney, region-level time-varying heterogeney, and year fixed effects. The microeconomic evidence provided by our study contributes to the lerature on the impact of services liberalization for economic growth and for the performance of services users. At the macro level, Mattoo et al. (2006) and Eschenbach and Hoekman (2006) show that countries wh liberalized services sectors grow faster once all standard correlates of growth are controlled for. 3 Based on computable general equilibrium models, Konan and Maskus (2006) and Jensen et al. (2007) suggest that services liberalization could lead to large gains in GDP in Tunisia and Russia, respectively. 4 The main mechanism underlying these gains is the increase in the number of business services available for manufacturing users as a result of the rise in services FDI. This increase of the number of services inputs increases the TFP of manufacturing firms through a Dix- Stiglz-Ethier framework. 5 While these two papers make an important contribution by showing the potential gains from services liberalization, their theoretical framework assumes rather than shows the ensuing improvements in the efficiency of services users. At the industry level, Francois and Woerz (2007) show that the increased openness of business services through exports and FDI had strong posive effects on exports, value added, and employment of 3-dig manufacturing industries in OECD countries between 1994 and Fernandes (2007) finds posive and significant effects of liberalization of finance and infrastructure on labour productivy of downstream manufacturing industries across several eastern European countries from 1997 to See the comprehensive survey on the topic of trade in services by Hoekman (2006) for further references. 4 Markusen et al. (2005) also show important GDP gains from services liberalization based on general equilibrium simulations for a hypothetical country. In their model, the presence of foreign-owned services firms allows final goods producers to rely on more specialized expertise. 5 See Dix and Stiglz (1977) and Ethier (1982). 3

5 At the firm-level, Arnold et al. (2007a) show significant posive effects of services liberalization - capturing FDI, privatization and deregulation - in the Czech Republic on manufacturing firms TFP between 1998 and Similarly, Arnold et al. (2007b) find significant posive effects of banking, telecommunications, and transport reforms on Indian manufacturing firms TFP between 1990 and Finally, Javorcik and Li (2007) provide evidence of a posive effect of the presence of FDI in the retail sector in Romania on the TFP of manufacturing suppliers to the retail sector. We should note that by exploing the heterogeney in services usage at the plant level, our study differs from these firm-level studies since they capture the dependence of manufacturing firms on services using industry-level coefficients from input-output tables. Finally, by considering the potential role of knowledge spillovers from services providers to manufacturing users, our study relates also to the lerature on vertical knowledge spillovers from FDI in manufacturing, which are shown to be more important than horizontal spillovers by Javorcik (2004), Kugler (2006), and Blalock and Gertler (2008). 6 A specific rationale provided in this lerature for forward linkages (i.e., from a foreign supplier to local buyer) is that foreign suppliers provide assistance to firms on the most effective way to use the inputs they are supplying (Gorg and Greenaway, 2004). Our main results are as follows. First, we find evidence of a posive and significant effect of FDI in services on productivy growth of Chilean manufacturing plants that use those services more intensively. The estimates from our preferred specification suggest that an increase in the services FDI linkage variable of the same magnude of that which 6 The main rationale for strong spillovers through backward linkages (i.e., from foreign buyers to a local supplier) found by Javorcik (2004) for Lhuania and by Blalock and Gertler (2008) for Indonesia is that multinationals are likely to benef from sharing knowledge wh their local manufacturing suppliers, whereas they are likely to minimize the leakage of information to their domestic competors. 4

6 occurred between 1992 and 2004 would add 1.1 percentage points to annual plant productivy growth in Chile, all else constant. The economic impact of this estimate is that backward linkages from FDI in services to downstream manufacturing users account for about 4.7% of the observed increase in manufacturing productivy growth in Chile (24%) during the sample period. This economic impact is que meaningful in light of the finding by Haskel et al. (2007) that spillovers from manufacturing FDI explain a roughly similar share (5%) of manufacturing TFP growth in the U.K. over the period. Given that a large fraction of FDI inflows into the services sector in Chile consisted in the acquision of incumbent firms - many of which privatized since the late 1980s - our estimated impact is likely to be an under-estimate relative to the potential impacts in countries where services FDI inflows are directed at the privatization of services providers or at the creation of new firms. We obtain qualatively similar findings using a measure of plant TFP growth based on production function estimations following Levinsohn and Petrin (2003). Second, our results are robust to variations in the definion of services FDI penetration, to the use of three-period lagged measures of services FDI penetration, to the elimination of potential outliers, and to the control for measures of domestic competion. Third, our findings are robust to the use of alternative measures of TFP growth based on production function estimation following Olley and Pakes (1996) or Ackerberg et al. (2007), and on simpler growth accounting plant TFP growth indices. Fourth, our results are not driven by any specific industry. Finally, we find suggestive evidence of a stronger effect of services FDI for plants in differentiated product industries but no evidence of different effects across small and large plants. These significant benefs of services FDI for productivy 5

7 growth of manufacturing plants may capture to some extent an unmeasured decline in qualy-adjusted services prices as well as the spillover of managerial and organizational knowledge or technical skills from foreign services providers to manufacturing plants. The remainder of the paper proceeds as follows. Section 2 describes recent trends in FDI in the services sector in Chile. Section 3 discusses the expected effects of FDI in services sectors and the available evidence. Section 4 describes the data. Section 5 describes our empirical specification. Section 6 discusses our main results and the results from robustness checks. Section 7 discusses extensions to our main results. Section 8 concludes. 2. Trends in Services FDI in Services in Chile We begin by presenting evidence on the substantial increase in FDI in services in Chile. Over the last three decades, liberalization, privatization, and deregulation reforms in Chile opened s economy to trade and investment more than any other country in Latin America (Moreira and Blyde, 2006). The entry of foreign capal into Chile is governed by Decree Law 600 which is in place since 1974 and grants equal treatment to foreign and domestic investments in mining, manufacturing, and most services sectors including telecommunications (Warren, 2000). 7 In the 1980s, the majory of FDI inflows were related to Chile s comparative advantage in the extraction and processing of natural resources. However, from the mid-1990s onwards, FDI inflows into the services sector took on a leading role, wh a peak in 1999, as shown in Figure 1. Electricy and water, transport and telecommunications, and business services represented about 60% of net 7 Specifically, Decree Law 600 regulates the condions of market entry, capalization, and remtances of foreign capal (ECLAC, 2000). Services sectors where FDI is somewhat restricted and equal treatment to domestic and foreign investments is not granted are professional services such as engineering, legal, and other technical services (Moreira and Blyde, 2006). 6

8 FDI inflows into Chile during the period. 8 While FDI inflows into mining and manufacturing originated mostly in the U.S., Canada, and U.K., FDI inflows into services originate mostly in Western European countries (OECD, 2007). The substantial FDI inflows resulted in a growing stock of FDI in the main services sectors in Chile: electricy and water, transport and telecommunications, and business services, as shown in Figure 2. In order to evaluate the importance of FDI in specific services sectors, we plot the ratio of sectoral FDI stocks to sectoral GDP in Figure 3. 9 The evidence shows a significant increase in the ratio of FDI to output in most services sectors between the first half of the 1990s and the period thereafter. The large inflow of FDI into services in Chile during the 1990s reflects first and foremost the worldwide increase in services FDI (UNCTAD, 2004). A prime reason for this development was the interest of multinationals (MNCs) to become global services providers by gaining access to domestic and regional markets - particularly in the developing world. This included mainly crucial services sectors, notably electricy, financial services, and telecommunications. Global MNCs such as the Spanish firm Endesa in the electricy sector identified Chile s largely privately-owned services sectors as good investment opportunies to consolidate their posions in Latin America (ECLAC, 2000). In sectors such as electricy, firms were first privatized before 1990 and then acquired by foreign players. The fact that firms were already privately-owned increased their attractiveness to global investors. 3. The Effects of Services FDI 8 Net FDI inflows are defined in the Appendix. 9 FDI stocks and GDP in services sectors are defined in the Appendix. 7

9 FDI in services sectors can have various types of benefs for the host country: price changes, qualy improvements, increased variety of services available, and knowledge spillovers Effects on the Services Sector First, the entry of foreign MNCs into services sectors is likely to increase competion in local markets and result in a reduction in the price of services. The reason is that incumbent firms - particularly in electricy and telecom sectors - can no longer retain the rents they obtained from being previously monopoly providers. Indeed, there is evidence of price decreases for Chile. Stehmann (1995) argues that in the telecom sector, FDI led to more competion, particularly in the markets where foreign MNCs entered early such as the long-distance market. Also, Pollt (2004) shows that prices in the Chilean electricy sector fell over the period. For a large group of 80 developed and developing countries Claessens et al. (2001) find that increased shares of foreign equy in the banking sector led to stronger local competion and reduced margins over the period. Second, FDI in services sectors may result in better qualy services. This improvement in qualy may result both from increased competion and from the superior technological, organizational, and managerial know-how of foreign services providers. 11 In the specific case of the electricy and, to some extent, the telecommunications sectors, qualy relates to the reliabily of service provision. FDI can provide the necessary 10 FDI in services also entails potential costs: (1) foreign ownership in sectors that are inherently monopolistic such as electricy or telecom may result in higher prices unless the regulatory system is very well defined and managed by the government, and (2) foreign ownership may crowd out domestic firms for example in the banking sector (UNCTAD,2004). 11 See Shelp et al. (1984) for a description of the process of technology transfer from multinational parents to their affiliates in insurance and in engineering and consulting services across developing countries. 8

10 finance for the major investments required for the modernization and expansion of existing networks. This argument is especially relevant for Chile which experienced a period of strong economic growth over the 1990s that resulted in increased demand - and thus increased pressure on the existing networks - for infrastructure services such as electricy while s domestic savings rates remained stable (OEDC, 2003). UNCTAD (2004) provides evidence of an impact of FDI on the reliabily of services in Latin America over the period. 12 World Bank (2004) shows improved service qualy in the electricy sector across Latin American countries as a result of privatization often to foreign MNCs and of deregulation. 13 Third, the presence of FDI in services sectors may result in a greater variety of services being provided. 14 For example in the case of the banking sector, this may include newer and more technologically advanced services, such as telebanking. Indeed, Denizer (1999) shows that as a result of the entry of foreign MNCs, Turkish banks introduced a number of modern planning, budgeting, and management information systems as well as electronic banking techniques. Evidence of an increase in the number of innovative financial products available as a result of foreign presence in the banking sector is provided by Akbar and McBride (2004) for Hungary and by Cardenas et al. (2003) for Mexico. Fink et al. (2002) show that higher shares of foreign equy in the telecommunications sector resulted in a tripling of the fixed telephone line penetration in 12 We should note, though, that despe a strong foreign presence in s electricy sector, Chile experienced an energy crisis in While the crisis was mainly caused by a drought, brought to light some technical failures and problems of coordination and transparency in the activies of power-generating firms (Gabriele, 2004). OECD (2003) argues that further improvements in energy infrastructure will be necessary for the continued growth of the Chilean economy. 13 Evidence of improvements in product qualy in other services sectors such as transport or professional services is not available as some equy and operational restrictions to FDI in those sectors remain (worldwide and in Chile in particular) and the existing inflows of FDI are fairly recent. 14 Increased variety can also be defined to cover the variety of markets being served: e.g., the provision of services to regions or types of clients to which they were not available before. 9

11 86 developing countries between 1985 and 1999, while Guislain and Qiang (2003) point to the explosion in the number of mobile subscribers over the same period. For Chile, ECLAC (2000) shows that FDI in the telecommunications sector led to the provision of a wider range of products and services (in addion to an increase in the number of telephone lines). Fourth, the entry of FDI into services sectors may result in leaking of managerial, marketing, and organizational know-how from foreign services providers to domestic services providers. Miroudout (2006) documents these knowledge spillovers for the banking, telecom, transport sectors in various developing countries. These knowledge spillovers often result in improvements to product qualy and increased variety of services provided by the services firms that benef from them. The four types of potential effects from FDI in services sectors just described - reduction of services prices, improvements in services qualy, increased variety of services available, and knowledge spillovers - are likely related to improvements in productivy growth of in the services sector, for both foreign as well as domestic providers. We should note that in Chile FDI inflows into the services sector occurred mostly through a wave of mergers and acquisions by foreign MNCs of services firms, some of which had been privatized since the late 1980s. The fact that MNCs often acquired the best performing services firms as opposed to opening entirely new subsidiaries could reduce the potential for posive effects of FDI on the performance of Chilean services firms. Nevertheless, there is evidence that those posive effects 10

12 materialized: e.g., Pollt (2004) shows a significant improvement in labour productivy in the Chilean electricy sector due to foreign entry during the period Effects on the Manufacturing Sector The crucial hypothesis that we test in this paper is whether the FDI-induced improvements in services sectors just discussed have dynamic benefs for downstream manufacturing users in terms of their TFP growth. If present, these dynamic benefs could be classified as pecuniary (or rent) spillovers which are a by-product of market interactions (Griliches, 1992). Manufacturing plants benef from pecuniary spillovers if they use services and if the improvements in the qualy of services are (i) not fully appropriated by services providers and/or (ii) not incorporated in services price deflators. 16 Regarding (i), for services sectors which tend to be characterized by imperfect competion, the most likely reason why providers do not appropriate the entire surplus from better and more diversified services is their inabily to perfectly price discriminate. 17 If the presence of FDI increases substantially the degree of competion in some services sectors, then intensive competive pressure would be the reason preventing services providers from appropriating that surplus. 18 Regarding (ii), price 15 Fischer et al. (2003) provide addional evidence of substantial labour productivy gains in Chilean electricy firms (in both generation and distribution) between 1987 and Vergara and Rivero (2005) provide evidence of sustained TFP growth in the Chilean electricy, gas, and water sectors, the transport and communications sectors, and the financial and business services sectors between 1986 and Though the authors do not account explicly for, FDI is likely to have played a role in TFP growth of the services sectors, given that the study s sample period covers the period of expansion of FDI in services. 16 New services could be thought of as qualy improvements to services: new services replace less suable alternatives that manufacturing plants had to use previously. 17 First-degree price discrimination is rarely possible by services providers since would require perfect knowledge of consumers willingness to pay for each un of the service purchased. Regulatory restrictions on prices further diminish services providers abily to perfectly discriminate. 18 In both cases, services providers charge a price that is lower than the qualy of the services provided. 11

13 deflators for services sectors in Chile do not capture perfectly the improved qualy and variety of services (e.g., through hedonic methods). Thus, if one estimates a significant effect of services FDI on measured TFP growth of manufacturing plants that may to some extent capture the pecuniary spillovers from an unmeasured decline in qualy-adjusted services prices. In Section 5, we give more analytical content to this possibily, drawing on Griliches and Lichtenberg (1984). Note, however, that even if price deflators captured perfectly qualy improvements there could be under-pricing of services due the market structure prevailing in the services sectors and to the inabily of services providers to perfectly price discriminate. FDI in services could also benef manufacturing plants through spillovers of soft technology linked to managerial, organizational, marketing know-how and technical skills. Learning by manufacturing plants could result from demonstration effects, personal contacts, and manager or worker turnover. 19 Griliches (1992) distinguishes knowledge spillovers from pecuniary spillovers since, in principle, only the former can lead manufacturing plants to potentially use that knowledge to advance their own innovation capabilies. While this provides a conceptually clear categorization, in practice pecuniary spillovers may become knowledge spillovers if downstream users of better services apply some of the embodied knowledge in services to improve their own productivy growth (Branstetter, 2001). First, for knowledge-intensive business services - such as marketing, some financial services, legal services, accounting, technical and other consultancy services (including those related to information technology and to research and development) - the actual service provided is a knowledge-intensive input 19 See Malerba (1992) for a discussion of alternative learning processes for plants. In particular, the author examines the role of learning by interacting wh suppliers as a spur for incremental technical change by plants. 12

14 upon which manufacturing plants may rely to improve their innovation capabilies and thus their productivy growth (Kox and Rubalcaba, 2007). 20 The capabily of routine problem-solving as part of everyday project work and the instructions and know-how for installing and using new equipment and systems exemplify intangible knowledge flows between a knowledge-intensive business services provider and s (manufacturing) client (den Hertog, 2002). Second, the usage of newer services (e.g., internet banking) by manufacturing plants may embody technological knowledge that allows them to improve their production and operations (e.g., by increasing the effectiveness of their investments in information and communications technology). This type of forward knowledge spillovers is more likely the more pervasive and lasting is the foreign presence in services sectors. Furthermore, the increased reliabily of service provision resulting from FDI that we discussed above may (1) allow manufacturing plants to optimize their machinery usage (e.g., production processes are less disrupted due to electricy outages) and (2) provide the incentives for plants to use technologically more advanced production processes that depend on telecom or internet/data connection. Both of these possibilies capture dimensions of technological change and thus would imply a posive effect of services FDI on plant productivy growth. These examples epomize the overlap between pecuniary and knowledge spillovers which will characterize our main results throughout the paper. 4. Manufacturing Plant-Level Data 20 The providers of knowledge-intensive business services can act as facilators of innovation for manufacturing plants as they share wh them experience and ideas regarding best practice solutions for technological and business problems based on their observation of localized tac knowledge across their clients, which can be particularly relevant for small plants (Muller and Zenker, 2001). 13

15 The main dataset used in our analysis is the ENIA (Encuesta Nacional Industrial Annual), which is the annual manufacturing survey of Chilean plants wh more than 10 employees. The dataset is an unbalanced panel capturing plant entry and ex. The original dataset covers the period and includes an average of 4913 plants per year (varying from 5379 in 1996 to 4395 in 1999) classified into 4-dig ISIC revision 2 industries. 21 After addressing various data problems our final estimating sample has 57,025 observations. The Appendix provides details on how the final sample was obtained and shows the sample composion across years and 3-dig industries as well as summary statistics for the variables that enter our econometric analysis. The ENIA survey collects plant-level information on sales, employment, raw materials, investments (buildings, machinery and equipment, transportation, and land) which we will use to construct output and inputs for the empirical productivy growth specifications described in Section 5. All nominal variables are expressed in real terms using appropriate deflators and capal is constructed applying the perpetual inventory method formula as described in the Appendix. A particularly interesting feature of the ENIA survey is that collects information on plant-level expendures on a variety of services: advertising, banking commissions and interest payments, communications, insurance, legal, technical, and accounting services, licenses and foreign technical assistance, rental payments, transport, other services and water. Regarding electricy, the ENIA survey provides information on both the quanty consumed and the corresponding expendures. On the one hand, this 21 The ENIA dataset has been widely used for the analysis of a variety of micro-level issues by Pavcnik (2002), Alvarez and Lopez (2004), and Bergoeing and Repetto (2006), among others. Note that the dataset provides information by plant and not by firm. However, according to Pavcnik (2002) more than 90% percent of firms during the period were single-plant firms. Therefore plant information corresponds to a large extent to firm information. 14

16 information allows us to include the quanty of electricy and an aggregate of other services (appropriately deflated) as inputs in our production function specified in Section 5.1. On the other hand, this information allows us to construct plant-specific weights of services usage - to be used in Section for four categories of services: electricy and water, transport and communications, financial, insurance, and business services, and real estate. Figure 4 shows that the average share of services (in total sales) used by Chilean manufacturing plants increased substantially over the period, wh business services being the most important component. Figure 5 also shows substantial differences in the average usage of services across manufacturing industries. However, almost all of the variation in plant-level services usage is due to variation whin industries rather than across industries. 5. Empirical Specification 5.1 Basic Framework In this section we present the reduced form framework used to estimate the impact of FDI in services sectors on the productivy growth of Chilean manufacturing plants. We consider a production function for plant i in industry j at time t: 22 Y = A F X, S ) (1) ( where Y is output, X is a vector of five inputs (skilled and unskilled labour, materials, electricy, and capal), S is a services input, and A is a plant-specific index of Hicksneutral TFP measuring the plant s technology factor and s efficiency in transforming inputs into output. The separation of the services input from the rest of inputs will be 22 For simplicy we om the industry subscript j from Equations (1)-(5). 15

17 useful in the discussion below. Assuming a Cobb-Douglas functional form for the production function, taking logarhms, and first-differencing Equation (1) we obtain: d lny 5 = d ln A + m= 1 η * d ln X +η * d ln S m m s, (2) where η m is the contribution of growth in input contribution of growth in the services input to output growth. m X to output growth and η s is the For Chilean plants, the services input is measured as growth in nominal services expendures deflated by a services price deflator (i.e, growth in real services), as detailed in the Appendix. 23 Improvements in the qualy and variety of services available for manufacturing plants possibly linked to the presence of FDI in the services sector are not fully captured by the services price deflators in Chile, as mentioned in Section 3. Following Griliches and Lichtenberg (1984), we can think of a discrepancy d between the change in the actual services price deflator, d ln ps, and the change in the true (qualy-adjusted) services price deflator * * d ln ps : d ln ps = d ln ps + d. If the change in nominal services expendures is correctly measured d ln NS = d ln NS, then growth * in real services (the term entering Equation (2)) is calculated as d ln S = d ln NS d ln ps. However, growth in the true quanty of services is given by d ln S * * * * = d ln NS d ln ps. Thus, d ln S d ln S = d ln ps ( d ln ps ) = d. 23 Equation (2) also includes the contribution of electricy growth to output growth. However, in that case a quanty (rather than a deflated nominal expendure) is used. This implies that the issues related to pecuniary spillovers for other services described below do not apply to the electricy input. We should note, however, that electricy quanty may not capture improvements in the reliabily of provision, which as we mentioned in Section 3 could be one of the important effects of FDI in the electricy sector. For simplicy, we lim the discussion below to growth in other services. 16

18 Making the simplifying assumptions that output is correctly measured and services is the only mismeasured input, Equation (2) can be re-wrten in true growth rates as: 24 5 * m d lny = d ln A + η m * d ln X + η s ( d ln S + d ). (3) m= 1 where growth d ln A is actual productivy growth. This implies that measured productivy * d ln A based on Equation (2) deviates from actual productivy growth by the discrepancy in the change of the services prices deflator: d ln A = d ln A * + η d s. As Griliches and Lichtenberg (1984) state, this equation is a definional relationship between measured productivy growth and actual productivy growth. The crucial hypothesis that we test in this paper is whether FDI in services sectors affects plant productivy growth in the manufacturing sector. This effect could result from pecuniary spillovers showing up in measured productivy through mismeasured improvements in qualy and variety of services in services prices deflators. However and equally important is the possibily that FDI in services sectors may generate knowledge spillovers for manufacturing plants, and pecuniary spillovers themselves can result in knowledge spillovers, as discussed in Section 3. Therefore, we allow measured plant productivy growth to depend on a measure of the presence of FDI in services FDIsl 1 (detailed in Section 5.2) as: d ln A FDIsl Z λ, (4) = β fdi _ s 1 + γ Z 1 + t 24 This is a strong assumption given the problems associated wh the measurement of output and material inputs by deflated revenues and expendures in the absence of data on the physical quanties of output and of material inputs, respectively. Plant productivy measures obtained based on deflated revenues and deflated expendures combine true plant efficiency and price-cost markups. If the mark-ups increase wh efficiency as in the models developed by Bernard et al. (2003) and Katayama et al. (2003), then productivy measures are correlated wh true plant efficiency. In this paper, we abstract from the issues of measurement of output and materials to focus on the effect of services FDI on manufacturing TFP growth. 17

19 where Z 1 is a vector of addional control variables potentially affecting plant productivy growth, and λ t are time specific effects. Combining Equation (4) wh Equation (2) and adding a stochastic residual we obtain a preliminary version of our empirical specification: d lny 5 m fdi _ s FDIsl 1 + η m * d ln X + η s * d ln S + γ z Z + λt ε. (5) m= 1 = β + A posive β fdi _ s in Equation (5) indicates a beneficial impact from the presence of FDI in services sectors on productivy growth, i.e., on output growth controlling for input growth of manufacturing plants. Before discussing various econometric issues associated wh the estimation of Equation (5), we present our measure of services FDI. 5.2 Services FDI Linkage Measure To estimate the effects of services FDI on manufacturing productivy growth in Chile, we make the working hypothesis that plants that are relatively heavier users of services should (ceteris paribus) experience disproportionately more benefs from an increased presence of FDI in services sectors than plants that are less heavy users of services. 25 To measure the intensy of plant services usage, we use information on plant expendures on various types of services described in Section 4, as a ratio to plant sales. One could argue that using information on a plant s usage of services from foreign providers as separate from domestic providers would be better. However, to the extent 25 This assumption is close in spir to that made by Rajan and Zingales (1998) in the estimation of the benefs of access to finance on growth across industries. 18

20 that domestic services providers improve their qualy and variety and lower their prices due to the presence of FDI in their sector - through increased competion or knowledge spillovers - a plant s total services usage will capture adequately the benefs which a manufacturing plant in Chile may derive from services usage for s productivy growth (beyond the effects that growth in services inputs per se generate for output growth). To measure the presence of FDI in each services sector, we compute a measure of net FDI inflows based on data from the Chilean Foreign Investment Commtee for each year between 1992 and Net FDI inflows are obtained by subtracting from the annual FDI inflows the corresponding annual FDI outflows which refer to foreign investors repatriation of capal, profs, and dividends. But net FDI inflows do not adequately capture the importance of FDI in a services sector in a given year because they neher account for all past investments nor for the size of the sector. Therefore, we cumulate the net FDI inflows using the perpetual inventory method formula to construct an FDI stock for each services sector, as described in the Appendix. 26 Then, we compute the ratio of the FDI stock in a services sector to the corresponding sector s output (GDP) obtained from the Chilean Central Bank. This ratio gives a measure of FDI penetration in a services sector. 27 To construct our final measure that captures both the presence of FDI in services and plant usage of those services, we weigh the FDI penetration ratio in a services sector by the intensy of plant usage of those services in year t. The final services FDI linkage measure FDIsl (using the notation in Equation (5)), is obtained as: 26 Our approach is similar to that followed to construct R&D stocks (e.g., Coe and Helpman, 1995) and public capal stocks (e.g., Munnell, 1990). 27 Note that this measure differs from the measures of the presence of FDI in a sector used tradionally in the lerature on spillovers from FDI, i.e., the share of a given sector s sales or employment accounted for by foreign firms. In the absence of a survey of services firms in Chile, such measures cannot be computed. 19

21 K k FDIsl = α * FDI kt, where FDI kt is the FDI penetration ratio in services sector k k = 1 k and α is the intensy of usage of services from sector k by plant i in year t. The sum is computed over four groups of services sectors: (1) electricy and water, (2) transport and communications, (3) financial, insurance, and business services, and (4) real estate. 28 Addional details on the construction of services FDI linkage measure are provided in the Appendix. Our services FDI linkage measure draws upon the measure used by Arnold et al. (2007a), but differs from by relying on a plant-specific time-varying measure of services usage instead of measures of services usage based on input-output table coefficients. 29 A first shortcoming of measures based on input-output tables is that they provide information on average industry usage only. However, an industry average is unable to identify the heavy users of services whin the industry. Indeed, as mentioned in Section 4, evidence from our data suggests a large degree of heterogeney in services usage across plants whin 3-dig industries in Chile. A second shortcoming of measures based on input-output tables is that they provide information for a single year. This is a particularly restrictive assumption for services usage during our sample period as the 1990s saw an increase in the importance of the linkages between services and manufacturing resulting from processes of outsourcing or splintering (Pilat, 2000; Francois and Woerz, 2007). Figure 4 indeed shows a substantial increase in the usage of services by manufacturing plants in Chile over the sample period. 28 The groups of services sectors that we use are dictated by the availabily of sectoral output (GDP) from the Chilean Central Bank. Business services encompass advertising, legal, technical, and accounting services, licenses and foreign technical assistance, and other services. 29 Both our measure as well as that in Arnold et al. (2007a) are inspired by the seminal work of Javorcik (2004) and Blalock and Gertler (2008) for vertical spillovers whin manufacturing. 20

22 5.3 Econometric Issues and Final Specification First, we note that during most of our sample period Chile experienced strong economic growth, which may have resulted simultaneously in manufacturing productivy growth but also in attracting FDI inflows into services. Thus, an estimated posive effect of the services FDI linkage variable could be due to s spurious correlation wh the dependent variable through the strong economic growth channel. Hence, Equation (5) includes year fixed effects which also account for economy-wide technological progress. Second, plant unobservables, such as managerial abily, may affect both plant productivy growth and the services FDI linkage, especially through the intensy of services usage. To address this possibily, we allow the residual ε in Equation (5) to include a plant-specific component f i such that ε = f + u i, where u is an independent and identically distributed disturbance. Consequently, the estimation of our final specification uses plant fixed-effects regressions. Third, in addion to fixed plant unobservables, time-varying factors could also be correlated wh the services FDI linkage variable and wh plant productivy growth. The omission of those factors could bias the estimated coefficient on the services FDI linkage variable in Equation (5). Main candidate factors are contemporaneous FDI in manufacturing and mining. The services FDI linkage variable in Equation (5) might be proxying for the effects of FDI in these other sectors. To address this possibily, we include in the vector of addional controls Z FDI linkage variables for manufacturing and mining whose construction is described in the Appendix. The coefficient on the services FDI linkage variable could also pick up differences in services usage by foreign or exporting plants, which in turn may exhib different productivy growth rates relative 21

23 to other plants. Similarly, productivy growth and services usage may be intrinsically different for small versus large plants. Hence, we add to Z dummies to control for foreign ownership, export status, and plant size. 30 Further, is possible that certain manufacturing industries experience faster technological progress or changes in their market structure relative to other industries, wh potential consequences for plant productivy growth. Chilean regions may also exhib differential growth rates over time due to the evolving nature and importance of agglomeration economies. To account for these two possibilies we include industry-year interaction effects and region-year interaction effects in Z. Finally, the main challenge for the estimation of our empirical specification is the potential endogeney of the services FDI linkage variable wh respect to manufacturing productivy growth. One possible mechanism for such endogeney would be that manufacturing industries whose plants experience faster productivy growth on average lobby the government for services liberalization. In that case, the estimated effect of the services FDI linkage could be biased upwards. However, several reasons, particularly related to the timing of events, suggest that this potential reverse causaly is unlikely to hold in the Chilean case. First, manufacturing industries may have lobbied for services liberalization through privatization and the attraction of FDI inflows. However, Chile had a fairly liberal FDI regime since well before our sample period, as mentioned in Section 2. Second, the privatization of services firms started in the late 1980s partly to solve a public defic problem and partly based on the potential efficiency gains from private 30 We define three size categories based on plant-level total employment as in Stumpo and Alarcon (2001): small plants have less than 50 employees, medium plants have 50 to 200 employees and large plants have more than 200 employees. We also experiment wh other size categories and found the results to be robust. 22

24 ownership (Bran and Saez, 1994). Third, the response of FDI inflows depends to a significant extent on external factors that are outside the control of the Chilean government and s manufacturing sector. As mentioned in Section 2, services FDI inflows were part of a global trend that led MNCs to invest in Chile in order to establish their posion in services whin Latin America. While foreign investors may have been attracted to the relatively sound performance of some of the recently privatized services firms in Chile, as hypothesized by ECLAC (2004), there is no clear evidence that they were attracted by the performance of the manufacturing users of those services. The idea of early benefs from the 1980s privatization in terms of improved firm performance is actually questioned by Hachette et al. (1992). This suggests that foreign participation in the services sector late in the 1990s was really the driver for the efficiency improvements mentioned in Section 3. This argument is in favor of our study finding significant effects of FDI in services on manufacturing productivy growth over our sample period. Notwhstanding, strong productivy growth of manufacturing plants in Chile may have provided an addional incentive for foreign investors in services sectors expecting subsequent strong demand for services. Our choice of a services FDI linkage variable that is based on services FDI stocks helps in migating the potential reverse causaly problems. Services FDI inflows would be much more vulnerable to reverse causaly issues as productivy growth of manufacturing plants might driving those. Moreover, our final empirical specification includes the services FDI linkage variable lagged two periods, instead of being lagged one period as in Equation (5). The question that arises is whether the two-period lagged services FDI linkage measure is exogenous relative to current plant productivy growth. This will not be the case if plant 23

25 productivy growth is serially correlated over time. While plant productivy levels tend to exhib serial correlation in micro datasets, there is no reason to expect this for plant productivy growth. Indeed, our tests for first-order autocorrelation in plant productivy growth rates based on the Baltagi-Wu locally best invariant (LBI) test provide no evidence of serial correlation. We therefore argue that the use of a services FDI linkage measure based on FDI stocks and lagged two periods is a reasonable way to migate potential reverse causaly problems. The considerations above lead to our final empirical specification: d ln Y + γ fdi _ r = β fdi _ s FDIres FDIsl jt γ Z z + J 1 6 j = 1 m = 1 + γ η * d ln X ind j m m ind * year + γ + η * d ln S s reg reg * year + γ fdi _ i + λ + t FDIndl f i + u jt 2 (6) where FDIsl 2 is the two-period lag of the services FDI linkage variable, FDIl 2 and FDIres jt 2 are the two-period lag of the manufacturing and the mining FDI linkage variables, respectively. Note that in the actual estimation the coefficients on input growth are allowed to differ across 2-dig industries (ISIC classification). Finally, recall that the vector Z 1 includes dummies for foreign ownership, export status, and size, that ind * year and reg * year are industry-year interaction fixed effects and region-year interaction fixed effects, and u is an i.i.d. disturbance. 6. Results 6.1 Main Results 24

26 Our main results are shown in Table 1. Column (1) presents the results from a plant fixed-effects specification that follows previous studies in measuring the usage of services by manufacturing plants using industry-level coefficients from the Chilean 1996 input-output table. 31 We find insignificant effects of services FDI penetration weighted by the input-output coefficients on plant productivy growth. In Column (2), we estimate a modified version of Equation (6) including only the one-period lag of our services FDI linkage variable (based on plant-level weights for services usage). 32 We find a very strong and posive effect of services FDI on manufacturing plants productivy growth. However, the specification in Column (2) does not take into account the potential endogeney of the services FDI linkage variable and therefore the effect of services FDI on productivy growth is likely to be over-estimated. Indeed, that seems to be the case as the use of the two-period lag of the services FDI linkage variable in Column (3) reduces substantially the magnude of the effect. However, the estimated coefficient on the services FDI linkage is still posive and significant at the 1% confidence level. In Column (4) we add the manufacturing and mining FDI linkage variables and find the coefficient on the services FDI linkage variable to be unaffected. Finally, Column (5) estimates the complete empirical specification in Equation (6) i.e., controls for plantlevel observables and industry-year and region-year interaction effects. This is our preferred specification. The standard errors in Table 1 and all subsequent tables are robust and clustered at the plant-level to allow for possible correlation across 31 The details on the computation of the measure are provided in the Appendix. 32 Due to space constraints, we om from all tables the coefficients on input growth since that would require the addion of 54 coefficients = 6 (inputs) * 9 (2-dig industries). These coefficients are available from the authors upon request. 25