Interindustry Goods Market Networks and Industry Lobbying for Trade Policy

Transcription

1 Foreign Policy Analysis (2017) 13, Interindustry Goods Market Networks and Industry Lobbying for Trade Policy HAK-SEON LEE James Madison University This study investigates how interindustry goods market networks influence industry lobbying as foreign firms direct investment and local production increase in the United States. The goods market networks consist of each sector s procurement of inputs from other sectors and the sectoral destinations of its outputs. I found that domestic upstream sectors modify their lobbying in ways dissimilar to those of downstream sectors when foreign firms local production and sales increase. Upstream sectors lobby more when US affiliates of foreign firms that procure inputs from their own supply chains gain market share at the expense of domestic firms in investment-receiving sectors. In contrast, downstream sectors lobby less when they save input procurement costs as domestic and foreign firms compete to produce quality goods at a lower price. The empirical results imply that goods market networks provide another theoretical framework, in addition to that provided by factor or sector models, in the analysis of demand side of trade policy. Introduction Foreign direct investment (FDI) by multinational corporations (MNCs) into the United States has significantly increased since the mid-1970s. 1 While previous research in political economy focused largely on how existing or threat of building trade barriers incur tariff-jumping investment (Hamada 1974; Brecher and Diaz- Alejandro 1977) 2 or quid pro quo investment that preempts protectionist measures (Bhagwati 1987; Dinopoulos 1989; Wong 1989; Bhagwati, Dinopoulos, and Wong 1992; Grossman and Helpman 1994), inward investment s impact on the demand side of trade politics has received attention only recently. Existing research shows mixed results on firms reactions to inward FDI. Regarding domestic firms political reactions to FDI inflows, researchers find that local firms lobbying for trade protection decreases because the benefits of pre-existing protectionist rents diminish once foreign firms enter US market and produce locally (Bhagwati and Irwin 1991; Hillman and Ursprung 1993; Blonigen and Feenstra 1996; Thomas 1997; Blonigen and Ohno 1998; Ellingsen and Warneryd 1999). This logic is closely related to the Mundell equivalence: Crossborder factor movement should lead to factor price equalization, just as international trade does (Mundell 1957; Svensson 1984). In contrast to these previous 1 The stock of inward FDI in the United States rose from $51.5 billion in the mid-1970s (2.6% of the total United States nonfinancial corporate net worth) to $425.6 billion in the early 1990s (11.6% in the same measure). See Graham and Krugman (1995). 2 The tariff-jumping FDI argument posits that MNCs have an incentive to engage in direct investment, so as to avoid trade barriers by local production and gain access to a local market. Lee, Hak-Seon. (2017) Interindustry Goods Market Networks and Industry Lobbying for Trade Policy. Foreign Policy Analysis, doi: /fpa VC 2014 International Studies Association

2 HAK-SEON LEE 51 works, more recently Zeng and Sherman (2009) argue that inward FDI may bring in a higher level of competition between foreign and domestic firms, and, as a result, domestic firms are more likely to increase demand for trade protection. These contradictory findings suggest that future research should revisit the linkage between inward FDI and firms lobbying for trade protection in host countries. Regarding the trade policy preference of US affiliates of foreign firms, Goodman, Spar, and Yoffie (1996) find that these firms advocate free trade when they engage in intrafirm trade with their parent firms, but begin to demand trade protection when they replace imports with local production in the United States. This finding complements the conventional wisdom that MNCs advocate free trade in home countries because home-country trade barriers may trigger retaliation by foreign governments, which in turn hurt multinationals intrafirm trade (Helleiner 1977; Milner 1988). Unlike these studies, which focus on how domestic or foreign firms react to inward FDI in their own sectors, Hiscox (2004) investigates how other sectors react to increasing cross-border investment. He finds that once FDI-receiving sectors expand production and pull labor from other sectors, the latter experience decreased returns and then lobby more for compensation. Focusing on interindustry labor flows between FDI-receiving sectors and other sectors, Hiscox (2004) expands the scope of investigation on FDI s impacts on industry lobbying. Moreover, his analysis that borrows a specific factor model by Jones and Neary (1984) permits an empirical test that closely corresponds to the theoretical model. 3 However, it still offers an incomplete explanation of the political impacts from inward FDI on the real economy, which is composed of multiple industries connected not only by labor flows but also by intermediate goods sales and purchases. This study adds another perspective to the approach by Hiscox (2004): It incorporates interindustry goods sales networks in the analysis of FDI and industry lobbying. The networks are represented by an input output (I-O) table that shows each sector s procurement of input goods from other sectors and the sectoral destinations of its outputs (Leontief 1983). Industries that sell their products to final goods producers are upstream of these producers. And industries that purchase intermediate goods from upstream producers are downstream of the sellers. The I-O linkage has been employed for the analysis of interest groups lobbying in trade politics. For example, Gawande and Bandyopadhyay (2000) argue that upstream and downstream sectors have conflicting interests related to trade policies in upstream sectors; upstream sectors demand trade protection while downstream sectors advocate free trade. In a similar vein, Cadot, de Melo, and Olarreaga (2004) and Gawande, Krishna, and Olarreaga (2012) find that upstream sectors pursuit of trade protection is affected by counter-lobbying by downstream sectors, which would experience higher input procurement costs if upstream sectors lobbying for protection succeeded. To analyze how inward FDI and interindustry goods market networks affect industry lobbying, this study sheds light on two elements of the network structure: (i) a given industry s level of goods sales (or purchases) dependencies upon other sectors; and (ii) the industry s position (i.e., upstream or downstream) in the network structure. First, the I-O linkage implies the existence of interindustry interdependencies of goods sales and purchases. Some sectors are more closely 3 The model assumes a domestic economy with three factors (labor and two kinds of capital) in two sectors that produce two goods. In the model, both kinds of capital are specific to sectors in which they are employed, while labor is mobile across sectors in the domestic economy. However, one kind of capital is assumed to be mobile across borders, while the other is not. Labor is not mobile internationally, either. See Hiscox (2004).

3 52 Interindustry Networks and Trade Policy Lobbying connected to each other because of a relatively large volume of goods sales between them. It is these nearby sectors for example, the ones with the strongest goods market connections that disturbances such as inward FDI into a given sector should most affect. In contrast, distant sectors that have few sales to (or purchases from) a FDI-receiving sector would not receive a substantial impact from the investment. As a result, nearby sectors will respond to inward FDI more strongly than distant sectors do. Second, a given industry s position is also important, because whether the industry in question is upstream or downstream of inward FDI may determine positive or negative impacts from the investment to the industry. Inward FDI into various US industries can illustrate how positive or negative impacts can be generated to upstream or downstream sectors. When foreign automobile firms undertake a greenfield investment and produce cars and trucks in the United States, the investment may affect the business environments of domestic suppliers of automobile parts (i.e., upstream of the US automobile industry) in different ways. Some domestic suppliers businesses are highly dependent upon their shipments to the automobile industry. Although in some cases the new arrivals become customers for existing US suppliers, foreign automobile companies producing in the United States usually procure the majority of their inputs from their parent firms via intrafirm trade or from home-country suppliers. Once US automobile producers lose market share to foreign entrants, US suppliers to domestic producers will receive a negative demand shock, and suppliers will then devote more resources to lobbying for compensation, ceteris paribus. 4 In contrast, downstream sectors reaction to inward FDI will differ from that of upstream sectors. For example, if the steel industry receives greenfield investment and increases output, the price of steel products in the domestic market will decrease, ceteris paribus. In this case, domestic automobile producers that are highly dependent upon the steel industry for their procurement of steel products should experience an increased rate of effective protection due to lowered procurement costs. In addition, more severe competition between new foreign entrants and domestic firms in the steel industry may benefit the automobile industry by providing quality products at a lower price. Then the automobile industry will be less likely to increase, or will even decrease, lobbying for trade protection, ceteris paribus. The empirical findings of this study confirm that a given sector s lobbying in trade politics is decisively influenced by its position in interindustry sales networks: Upstream sectors lobby more, and downstream sectors lobby less. Given two facts that inward FDI into the United States has been increasing since the 70s, and that the United States is currently the largest recipient of cross-border direct investment in the world the empirical findings of this study suggest that foreign firms investment into upstream sectors are more beneficial to US economy than investment into downstream sectors and thus should be encouraged by US foreign economic policies. This study proceeds as follows. The following section explains how inward FDI generates demand and supply shocks to upstream and downstream sectors via interindustry networks. Next, a set of hypotheses regarding industry demand for trade protection are empirically tested and the results are presented. The concluding section discusses the implications of this study on the trade politics literature and suggests directions for future research. 4 I assume that when foreign investment takes place, the affected sectors have a purely political reaction and do not respond by changing some aspects of their business or, at the extreme, exiting from the sector.

4 HAK-SEON LEE 53 Inward FDI and Interindustry Goods Market Networks I assume that the domestic economy is in equilibrium, and that the only relevant disturbance to the equilibrium is inward FDI. 5 I then address how the investment can trigger demand and supply shocks to upstream and downstream sectors, and how these sectors modify lobbying as a response to the investment. A given sector s reaction to events in FDI-receiving sectors is an increasing function of its input or output dependency on the sectors in question. I treat interindustry dependency as the proportion of a given sector s total purchases (or sales) that involve transactions with its upstream (or downstream) sector. A detailed explanation on the operationalization of interindustry dependencies is provided in the following empirical analysis section. Upstream Sectors Reaction to Inward FDI In specific circumstances, inward FDI may generate a positive or negative demand shock to domestic upstream sectors. In other circumstances, though, there can be no demand shock. Below, I explain a number of situations in which upstream sectors receive different demand shocks and how they modify lobbying accordingly. Once inward FDI occurs, two factors interact and may generate a demand shock to domestic upstream sectors. First, foreign firms procurement patterns of intermediate inputs need attention. On the one hand, foreign firms procurement may have supply rigidities for example, no change in quantity of goods supplied in the short run in the face of changes in the price of the good, which is a typical result of either intrafirm trade between parent firms and foreign affiliates or long-term supply contracts between multinationals and home-country suppliers. On the other hand, foreign firms may procure inputs from the most competitive suppliers without supply rigidities. The second factor that should influence the formation of a specific demand shock is changes in domestic market share between US and foreign firms in FDI-receiving sectors. If there is no significant change in domestic market share, no demand shock is generated to upstream sectors, ceteris paribus. A gain or loss in market shares by domestic firms vis-à-vis newly entered foreign firms may transmit positive or negative demand shocks to upstream sectors. Once specific demand shocks are generated, upstream sectors may modify their lobbying as a response to the shocks. In this process, upstream sectors level of competitiveness (i.e., import-competing or exporting) may also affect the modification of industry lobbying. Table 1 presents a summary of hypothesized changes in political efforts by upstream sectors responding to downstream FDI inflows. The table shows how (i) the presence or absence of supply rigidities by US affiliates of foreign firms; and (ii) domestic firms market share changes in FDI-receiving sectors should determine types of demand shocks transmitted to import-competing or exporting upstream sectors. Subsequently, the upstream sectors receiving different kind of demand shocks modify their lobbying accordingly. Cases of each demand shock are explained below. 5 Since this study deals with the effects of inward FDI, outward FDI by US multinationals is assumed to be in equilibrium as well. I consider only FDI rather than portfolio investment, which, by definition, does not alter the control of corporate assets. Nor do I investigate strategic alliances or licensing agreements between foreign competitors and domestic firms, because they do not involve flows of capital across borders. The Commerce Department defines a foreign investment as direct when a single investor has acquired a stake of 10% or more in a US firm. See Graham and Krugman (1995).

5 54 Interindustry Networks and Trade Policy Lobbying Table 1. Hypothesized Changes in Political Effort by Upstream Producers as a Consequence of Downstream Foreign Direct Investment Downstream Foreign Firms Procurement Patterns Supply Rigidities No Supply Rigidities Market Share Change in Downstream Sector Type of Demand Shock to Upstream Producers Upstream Producers Competitiveness Upstream Producers Response Loss of Market Share by Downstream Domestic Producers Negative Shock Increase lobbying Exporting Increase lobbying or Take no action Take no action Take no action N/A Importcompeting No Change or Gain in Market Share by Downstream Domestic Producers No Shock or Positive Shock* Importcompeting Exporting No Shock or Positive Shock Importcompeting Take no action or Increase lobbying** Exporting Take no action (Notes. *In a case of a gain in market share. In a case of an increase in total industry shipments after inward FDI. This assumes that upstream suppliers do not initially possess a long-term supply contract with arriving foreign firms downstream. If not substantially dependent upon downstream sectors. In a case of no shock. **In a case of positive shock.) No Demand Shock Before discussing cases of a positive or negative demand shock, it is necessary to point out that in some situations inward FDI transmits no shock to upstream sectors. First, in a case in which there are no supply rigidities, if there is no increase in total industry shipments in FDI-receiving sectors, domestic upstream sectors may not receive any demand shock regardless of their level of competitiveness. In this case, market share changes between domestic and foreign firms in FDI-receiving sectors are irrelevant because any firm in the receiving sectors will simply procure the most competitive inputs regardless of place of origin. 6 If domestic upstream sectors are competitive (and thus exporting), newly entered foreign firms are more likely to purchase inputs from the competitive producers. Since we assume there is no increase in total industry shipments in FDI-receiving sectors, the competitive suppliers simply have new customers (i.e., US affiliates of foreign firms) that replace previous customers (i.e., US firms). Nor will importcompeting upstream sectors receive any new demand shock from changes in ownership structure and market shares between domestic and foreign firms in FDI-receiving sectors. If the import-competing upstream producers have been lobbying for trade protection, they simply continue to do so, ceteris paribus. In a case in which supply rigidities do exist, if domestic upstream sectors still sell the same amount of inputs to domestic firms in FDI-receiving sectors, then no shock will be generated, ceteris paribus. This result may occur when US affiliates simply substitute previous exports to the United States with local production. In 6 An assumption is that international trade is relatively open so that importation of inputs is not restricted.

6 HAK-SEON LEE 55 this case, upstream sectors level of competitiveness makes no difference, because even competitive upstream suppliers cannot sell their products to newly entered foreign firms that rely on their own supply chains, such as parent firms or homecountry suppliers. When inward FDI generates no demand shock to domestic upstream sectors, they will have no incentive to modify their lobbying and thus take no action, because the sectors in question experience no change in business environments, ceteris paribus. Positive Demand Shock There are a few cases in which inward FDI may trigger a positive demand shock to upstream suppliers. First, in a case in which there are no supply rigidities, there will be a positive demand shock to domestic upstream sectors when there is an expansion of goods production in sectors receiving FDI, ceteris paribus. In this case, a positive demand shock to upstream suppliers will trigger contrasting responses from import-competing and exporting suppliers. Import-competing suppliers will have an incentive to lobby more because the positive demand shock increases the value of protection; once imports are excluded, there are more customers for the domestic suppliers, and hence more to gain by achieving protection (Ellingsen and Warneryd 1999). In contrast, exporting suppliers producing competitive inputs are more likely to increase sales to expanding sectors receiving FDI, and they are less likely to modify lobbying, ceteris paribus. Second, in a case in which supply rigidities do exist, domestic upstream sectors will receive a positive demand shock when US firms gain market share at the expense of foreign firms that rely on their own supply chains. In this case, domestic suppliers (regardless of their level of competitiveness) will be able to increase sales to FDI-receiving sectors and thus have no incentive to modify lobbying, ceteris paribus. The positive demand shock of this case can be generated either with or without an expansion in total industry shipment in FDI-receiving sectors, provided that supply rigidities exist. Negative Demand Shock Inward FDI will generate a negative demand shock to upstream sectors when two conditions are met: (i) if newly entered foreign firms have supply rigidities in procurement patterns; and (ii) if the foreign firms gain market share at the expense of domestic producers in FDI-receiving sectors. Procurement patterns by US affiliates are mandated by foreign ownership of the firms, which usually rely on intrafirm trade with parent firms, long-term supply contracts with home-country suppliers, and the suppliers accompanying investment into the United States. All of these procurement patterns provide supply rigidities. Foreign firms in the United States import more inputs than domestic firms do. The Bureau of Economic Analysis (BEA) reports that US affiliates rely on imported inputs to a much greater degree than do US-owned firms. In 1989, the import share of inputs by US affiliates was 17%, while that of US-owned firms was 11%. 7 The significant difference in import propensity between foreign and USowned firms arises because MNCs usually undertake intrafirm trade when they engage in FDI, particularly at the early stage of the investment, to procure input goods unavailable from host country suppliers (Goodman et al.1996; Crystal 1998). Furthermore, the portion of intrafirm trade in total US imports has been substantial. In 1982, intrafirm trade accounted for 32% of total US imports of goods and services, increasing to 37% in 1993 (Whichard and Lowe 1995). 8 7 In 1994, the portion of imported inputs increased to 19% for US affiliates, while that of US-owned firms remained at 11%. See Zeile (1998). 8 This measurement includes imports from foreign affiliates to US parent firms and those from foreign parent firms to US affiliates.

7 56 Interindustry Networks and Trade Policy Lobbying Moreover, during the 1980s, two-thirds of total imports by the US affiliates came from their parent firms or other foreign affiliates (Zeile 1997). The US affiliates thus have had higher import propensities, and the majority of their imports have been supplied through intrafirm trade. Long-term supply contracts between multinationals and home-country suppliers are another source of supply rigidity. If transaction costs for switching suppliers are substantial, firms usually sign long-term contracts with their suppliers (Caves 1996). In such a situation, US affiliates would rely on previously contracted suppliers instead of US suppliers. When foreign firms invest in the United States, their home-country suppliers may undertake their own FDI following their customers. 9 Foreign suppliers have two incentives to do so. First, when foreign firms produce in host countries, they may require specifically designed parts that local suppliers cannot provide. Thus, the dependence of the original investing firm on differentiated products along with requirements for close communication and easy goods movement between supplier and purchaser, as well as substantial investment from both sides in a long-term relationship creates incentives for suppliers to choose a location close to their customers. The second incentive to spur home-country suppliers to follow their customer firms abroad is provided by domestic content regulations that require firms to use a certain amount of domestically produced inputs in producing outputs. 10 The US upstream suppliers may have several options in their response to a negative demand shock. First, in an extreme case, they might close their business. This exit option would be the last resort, though, and less likely to be a short-term response to inward FDI. A more plausible short-term solution would be to lobby the US government to set up policies that discourage inward FDI by foreign multinationals. However, the United States has adopted relatively liberal and open FDI policies from post-world War II till now, with the notable exception of regulations on inward FDI in US national security-related industries (Graham and Marchick 2006). Thus, lobbying the US government to restrict inward FDI would not be a realistic option for upstream suppliers. Firms or industries more frequently choose to lobby for trade protection in order to get relief from an unfavorable business environment. When import-competing upstream suppliers receive a negative demand shock triggered by inward FDI and foreign firms supply rigidities, the upstream sectors are more likely to increase lobbying for trade protection, ceteris paribus. If an import-competing upstream sector is highly dependent upon FDI-receiving sectors, the sector will accordingly devote substantial resources to lobbying. In contrast, if the sector is not highly dependent upon investment-receiving sectors, it would not modify lobbying in a significant way. An exporting supplier s reactions to a negative demand shock will hinge on the level of sales dependency upon FDI-receiving sectors. If the supplier experiences a significant negative demand shock because its sales to the investment-receiving sectors are critical to its business, even the exporting supplier may increase lobbying for compensation. In contrast, if the exporting supplier targets primarily foreign markets and is not highly dependent upon domestic customers, a negative shock after inward FDI will be negligible. In this case, the exporting supplier will be less likely to modify its lobbying, ceteris paribus. 9 The arrival of foreign input providers might also involve increased competition for the domestic firms supplying inputs in markets other than the sector receiving the original FDI. 10 Domestic content requirements to MNCs have usually been imposed by developing countries to encourage the host countries industrialization. Yet Japanese automobile companies practice of using Japanese parts when they produce cars overseas brought the issue to Europe and the US. See Krugman and Obstfeld (2003).

8 HAK-SEON LEE 57 To summarize various cases discussed above, when MNCs produce in the United States, domestic upstream sectors will either take no action or increase lobbying for protection. Also, as shown in Table 1, import-competing upstream sectors have more numerous circumstances in which they increase lobbying for trade protection. My theory of interindustry goods market networks and supply rigidities along with FDI-receiving sectors market share changes and upstream producers competitiveness yields two hypotheses: Hypothesis 1: The more an upstream sector depends upon sales to FDI-receiving sectors, the more effort it will devote to lobbying for trade protection, ceteris paribus. Hypothesis 2: An import-competing upstream sector will lobby more for trade protection than does an exporting sector when foreign firms production and sales increase in the United States, ceteris paribus. Downstream Sectors Reactions to Inward FDI When inward FDI occurs, domestic downstream sectors that procure inputs from FDI-receiving sectors may realize economic gains. First, in a case of greenfield FDI that involves newly established production facilities, there will be an increase in industry shipments in FDI-receiving sectors, ceteris paribus. 11 This increase will lower the price of the goods FDI-receiving sectors produce, and downstream sectors have economic gains because they can save procurement costs. Another source of gain to downstream sectors is intensified competition between domestic and foreign firms in FDI-receiving sectors. Previous studies suggest that more severe competition between foreign and domestic firms in host countries should encourage both to produce high-quality products at a lower price. As a result, downstream sectors gain as more foreign firms undertake local production (Pack and Saggi 1997; Blomstrom and Kokko 2001; Barrios, Goerg, and Strobl 2005; Kugler 2006). 12 The level of purchase dependency of a downstream sector upon FDI-receiving sectors will determine the relative size of gains. Sectors purchasing a relatively large amount of inputs from FDI-receiving sectors will gain more, while other sectors purchasing a smaller amount gain less. Any gain from lower procurement cost may decrease downstream sectors vulnerability to international competition, ceteris paribus. If the marginal savings from decreased lobbying exceeded the marginal losses in effective protection caused by lower trade barriers, downstream sectors might actually decrease lobbying for protection, ceteris paribus Some or all of that increase can be exported. However, as an empirical matter this assumption is unrealistic, because foreign investors usually undertake market-oriented FDI to sell output in the United States, which is the largest national market. 12 A caveat is that there may be potential changes in industry concentration ratio after inward FDI occurs. If all of inward FDI takes the form of mergers and acquisitions, then not only would there be no initial increase in output, but the resulting investment might also increase market concentration in upstream sectors, which would then subject downstream sectors to stronger oligopoly pricing. Under these circumstances, downstream sectors might be worse off. This prospect would give them a motive to seek higher levels of protection on their own outputs. These sectors might also seek redress by raising antitrust objections to M&A or seeking to impose or invoke barriers to the investment. In contrast, greenfield foreign investment would result in no increase in market concentration, as it involves no M&A, and industry output would increase. If the investing firm is a new arrival, it will actually reduce market concentration, ceteris paribus. 13 If we assume a minimal fixed cost for lobbying, the gains from lowering procurement cost and the magnitude of decreasing lobbying efforts will be a step function rather than a linear function. However, I assume that if a sector is spending a substantial amount of resources for lobbying, there should be a net decrease of lobbying, ceteris paribus.

9 58 Interindustry Networks and Trade Policy Lobbying Consideration of potential economic gains generated by foreign firms local production and sales transmitted to a downstream sector produces the third hypothesis. Hypothesis 3: The more a downstream sector depends upon FDI-receiving sectors for procurement of inputs, the less effort it will devote to lobbying for trade protection, ceteris paribus. 14 Testing Hypotheses To test my hypotheses, I adopt a quantitative approach for a comprehensive analysis of multiple interindustry goods sales networks and industry lobbying. The unit of analysis is an individual US manufacturing industry in a given time period. A total of 18 manufacturing industries are covered: two-digit Standard Industrial Classification (SIC) codes from 20 to 38, with the exception of These industries are tradable sectors and data are readily available. 16 The time period is from 1981 to 1990, which spans five US Congresses (from the 97th to the 101st). This study focuses on the experiences of the 1980s for a practical reason: Data for the dependent variable industry level campaign contributions to candidates for US House of Representatives is readily available only for this decade. 17 In addition to the data-availability issue, the 1980s provides a suitable timeframe because inward FDI into the United States increased at a noticeably larger pace in the 1980s than in previous decades. While FDI inflows were less than half a percent of gross national production in the 1970s, they reached 1% at the beginning of the 1980s, and almost one and a half percent by the end of the decade (Graham and Krugman 1995). Table 2 provides a comparison chart on intra- and interindustry transactions for each manufacturing sector at the SIC two-digit level. As shown, interindustry transactions constitute the majority of these types of transactions, even though the ratio varies across sectors. The cross-industry variation implies each sector s level of dependency on other sectors, and my hypotheses predict that the level of dependency affects each sector s lobbying propensities. As explained below, a sector s sales and purchase dependencies upon other sectors are measured to represent cross-industry variation. To analyze industry campaign contributions, I employ the ordinary least squares (OLS) estimation with time period dummies that control for unexplained period and trend effects. In addition, OLS with panel-corrected standard errors (PCSEs) estimation is adopted because all of the error processes cannot be independent of one another in a panel data set with time variances (Beck and Katz 1995). Each model is estimated with one-period lagged independent variables to allow extra time for independent variables to influence the dependent variable. Most campaign contributions are made before members of Congress are elected and before they vote in the next congressional period. Thus, to test my theory empirically, it is essential to investigate how independent variables in the previous period affect campaign contributions in the current period. This paper focuses on impacts from different levels of foreign firms market share in the United States across industries, rather than on impacts from changing levels of inward 14 Supply rigidities in the upstream sector arising from intrafirm trade or long-term supply contracts are irrelevant to the downstream sector it is concerned only with price and availability of inputs. 15 SIC code 21 is the tobacco industry, and FDI into the industry is included in the SIC For the primary sectors (agriculture, forestry, and fishing), I-O Benchmark data cannot be converted to SIC two-digit codes. The service and construction sectors are not included because they were not tradable in the 1980s. 17 Data for other periods are available, but the cost of mapping it into SIC categories and coding it is very high. See Fordham and McKeown (2003).

10 HAK-SEON LEE 59 Table 2. Comparison of Intra- and Interindustry Transactions Standard Industrial Classification (SIC) Ratio of Intra-industry Transactions Ratio of Interindustry Transactions* 20 Food Textile Apparel Lumber and Wood Furniture Paper Printing and Publishing Chemicals Petroleum and Coal Rubber and Plastics Leather Stone, Clay and Glass Primary Metals Fabricated Metals Industry Machinery Electrical Machinery Transportation Equipment Instruments (Notes. *Interindustry transactions include each manufacturing sector s transactions with other manufacturing sectors and also those with nonmanufacturing sectors such as service sectors.) investment in a given industry across different periods. Still, industry fixed effects analyses are added for a reverse causality test: Generalized methods of moments (GMM) and error correction model (ECM). I measure industry lobbying for trade protection by calculating the total dollar amount of campaign contributions made by each manufacturing industry to electoral candidates for the House of Representatives who voted for protectionist bills in each of the five Congresses. The existing literature suggests that interest groups lobbying may target different groups of legislators. Interest groups lobby not just friend but also foe legislators in order either to counteract the influence of opposition interest groups (Austen-Smith and Wright 1994), 18 or target fence-sitters to persuade them to vote for favored policies by contributors (Rothenberg 1992). Interest groups campaign contributions have been employed for the analysis of lobbying in trade politics (Grossman and Helpman 1994; Grossman and Helpman 2002; Gawande 1998; Fredriksson 1999; Baldwin and Magee 2000; Gawande and Hoekman 2006; Leibman and Reynolds 2006; Gawande, Krishna, and Olarreaga 2009; Gilbert and Oladi 2012). Political action committees (PACs) are private interest groups organized to support political candidates to promote legislation that represents the PACs special interests. For example, Magee (2002) found that PACs reward candidates for adopting favorable policy positions in various economic and social issues, such as trade policy and gun control. In addition to contributions via PACs, interest groups lobby legislators through soft-money spending (Drope and Hansen 2004). However, in the case of soft money, it is not easy to identify direct links between interest groups and candidates, because the money is given to political parties first and then sent to specific candidates (Ansolabehere and Snyder 2000). A majority of PACs in the 1980s were individual corporations such as Kraft Foods or General Electric. Still, there exist business associations representing each manufacturing industry at the two-digit SIC level. Examples include the 18 Baumgartner and Leech (1996) argue that findings from Austen-Smith and Wright (1994) are erroneous because they use cross-sectional variance models where a longitudinal one is needed.

11 60 Interindustry Networks and Trade Policy Lobbying Association of Food Industries (20), the American Textile Manufacturers Institute (22), the American Apparel Manufacturers Association (23), the American Chemical Council (28), the American Petroleum Institute (29), and the Rubber Manufacturers Association (30). Other trade associations represent the SIC three-digit level, though, such as the American Iron and Steel Institute (331 & 332), the Alliance of Automobile Manufacturers (371), etc. Recently, Fordham and McKeown (2003) coded all corporate PAC campaign contributions between 1981 and 1990 into the two-digit SIC categories. The original data, though, includes all contributions made to any candidates regardless of those candidates preferences over trade policy. Yet, it is reasonable to assume that some candidates prefer protectionist policies while others advocate free trade. From the original data, I compiled data comprising campaign contributions made only to candidates who voted for protectionist bills, assuming that these contributions are a measurement of each industry s lobbying efforts for protectionist trade policies. In this paper, protectionist candidates are defined and measured in two ways candidates (i) who voted for any protectionist trade bills; and (ii) who voted for all of those bills in each Congress. Campaign contributions to protectionist candidates can be considered political efforts either to realize rents from protectionist trade policies or to compensate for a loss caused by the surge of imports. A caveat is that pressure groups might have made financial contributions to their favored candidates seeking benefits from other policies, not necessarily from trade policy. Still, industry contributions to protectionist candidates can be a valid measurement of corporate lobbying activities in trade politics because we can infer contributors lobbying purpose from their recipients voting records. 19 While some researchers found that campaign contributions do not directly affect legislative members voting patterns (Chappell 1982; Wright 1990; Hall and Wayman 1990; Wawro 2001), other researchers found not only that campaign contributions are a reflection of friendly giving to electoral candidates, but also that monetary contributions have independent impacts on the roll call voting of representatives (Fleisher 1993; Bronars and Lott 1997; Fellowes and Wolf 2004; Gordon 2005; Roscoe and Jenkins 2005). I employ three main explanatory variables: a given sector s own inward FDI as well as the sector s sales and purchase dependencies upon FDI-receiving sectors. The BEA publishes records of the total assets and sales values of both US affiliates of foreign firms and foreign affiliates of US firms at two- and three-digit SIC levels. 20 While the asset measurement of inward FDI is the portion of assets owned by foreign firms in total industry assets, the sales measurement indicates the portion of sales by US affiliates of foreign firms in total industry sales. Of the two measurements of FDI, the sales measurement is more relevant for the purpose of this paper, which is to investigate political impacts from the local production and goods sales by US affiliates. The own-sector inward FDI variable is employed to examine whether increasing foreign firms local production reduces incentives to lobby for protection or increase protectionist demand by domestic firms. This variable is constructed by the portion of shipments of US affiliates in total industry shipments. NIFDI i ¼ IFDI i =K i where IFDI is inward FDI measured by sales of US affiliates; NIFDI is normalized IFDI, and K is total industry shipments. Inward FDI variable in the regression models (Tables 5 and 6) indicates NIFDI. 19 The list of protectionist bills used to create the dependent variable is provided in Appendix The BEA collects data on FDI by means of mandatory surveys of the US affiliates of foreign firms and foreign affiliates of US firms.

12 HAK-SEON LEE 61 To investigate how interindustry goods market networks affect industry lobbying, two explanatory variables are created: a given sector s sales dependency, and its purchase dependency upon other sectors. These variables are compiled from the I-O accounts data by the BEA. The I-O data have their own industry classification codes, and I have converted them into two-digit SIC codes that are used for other main variables industry campaign contributions and inward FDI. 21 Sales dependency of sector i is the portion of shipments from sector i to sector j in sector i s total shipments; purchase dependency is similarly defined as SD ij ¼ s ij =s i and PD ij ¼ p ij =p i where SD ¼ sales dependency; s ¼ sales; PD ¼ purchase dependency, and p ¼ purchases. Own sector and another sector are indexed by i and j, respectively. 22 The impact of FDI in a given sector (sector i) on downstream purchases or upstream sales is taken to be proportional to the normalized investment into an investment-receiving sector (sector j). Inward investment s impact on an upstream or downstream sector will then be proportional to SD ij ðnifdi j Þ and PD ij ðnifdi j Þ respectively, where NIFDI j is a normalized shipment measure of inward investment in sector j. To arrive at the total impact of inward FDI on a given sector i, I then sum up the entire individual sector impacts: NSTOT i ¼ X j SD ijðnifdi j Þ and NPTOT i ¼ X j PD ijðnifdi j Þ where NSTOT i is the normalized total sales impact and NPTOT i is the normalized total purchase impact. A given sector s sales dependency variable (NSTOT) is employed to investigate how the sector s sales dependency upon FDI-receiving sectors will affect the sector s lobbying (Hypothesis 1). To test my conditional hypothesis (Hypothesis 2), which is to investigate import-competing upstream sectors reaction to inward FDI, I create an interaction term between NSTOT and the level of total industry imports. A given sector s purchase dependency (NPTOT) is employed to investigate how the sector s purchase dependency upon FDI-receiving sectors will influence the sector s lobbying (Hypothesis 3). Control variables that are conventionally believed to affect industrial trade policy preferences are also included in the regression models. The industry total shipments variable, which represents the size of an industry, is employed to test whether larger industries receive more attention from the government and thus are more likely to be active in lobbying (Lavergne 1983). A sector s import penetration level is predicted to be positively associated with industrial lobbying for trade protection because a higher level of import penetration is likely to trigger domestic import-competing producers to demand a higher level of trade protection, ceteris paribus. Total industry exports are predicted to be negatively related to industrial lobbying for protection, because export-oriented sectors are more concerned with global markets than with domestic markets; these sectors are less likely to lobby for trade protection due to potential retaliation in foreign markets 21 Background information for the I-O data is provided in Appendix These measures implicitly ignore the elasticity of demand and supply, an assumption generally consistent with input output analysis.

13 62 Interindustry Networks and Trade Policy Lobbying Table 3. Descriptive Statistics Obs. Mean Standard Deviation Minimum Maximum Industrial Campaign Contributions to , ,901 1,967 2,377,794 Protectionist Candidates (who voted for any protectionist bills) Industrial Campaign Contributions to , ,775 1,400 2,377,794 Protectionist Candidates (who voted for all protectionist bills) Inward FDI Sales Dependency Purchase Dependency Imports (1995 $bn) Exports (1995 $bn) Shipments (1995 $bn) Government Sales (1995 $bn) Capacity Utilization Industry Concentration once trade barriers are set up in the domestic market. A sector s total sales to the federal government is employed because an industry that sells heavily to the government is hypothesized to lobby more because of its sensitivity to government procurement patterns (Lichtenberg 1989; Zaleski 1992; Hansen, Mitchell, and Drope 2005). Another control variable is industry capacity utilization. Researchers have found that a lower level of industry capacity utilization is more likely to be associated with a higher level of protectionist demand (Strange 1979; Cassing, McKeown, and Ochs 1986; Hansen 1990). Lastly, the conventional wisdom in analyzing the demand side of trade politics is that a highly concentrated industry is better able to overcome the collective action problem between firms in the industry than are less concentrated industries. Acknowledging that industry preferences and actual lobbying activities cannot be equated because of the free-rider problem, this paper adopts another control variable: industry concentration ratio (measured by the market share of the four largest firms in each industry). Tables 3 and 4 provide basic statistics for each variable and the correlation matrix between variables, respectively. Tables 5 and 6 report the regression results on the analysis of industrial campaign contributions. The dependent variable is measured differently in each table. In Table 5, the variable is total industry contributions to candidates who voted for any protectionist bills in each election cycle. In Table 6, the variable is contributions to candidates who voted for all protectionist bills in each period. A smaller number of representatives did vote for all of protectionist trade bills, and thus, the total amount of contributions made by industries is relatively smaller in the latter. Each table adopts a total of six regression models. Model I is a baseline model including only control variables. The main explanatory variables are employed in Model II, and an interaction term is added in Model III. Model IV adopts the same variables as Model III, but PCSEs estimation is used. Models V and VI adopt industry fixed effects models: GMM and ECM, respectively. A given sector s inward FDI is seen to be negatively associated with contributions for trade protection, as shown in Models II through IV in both tables. This result shows that increased local production and sales by foreign firms in the United States should weaken industry lobbying, confirming the conventional argument that cross-border capital flows reduce industries incentives to lobby for

14 HAK-SEON LEE 63 Table 4. Correlation Matrix Inward FDI Sales Dependency Purchase Dependency Sales Dependency * Imports Imports Exports Shipments Gov t Sales Capacity Utilization Industry Concentration Inward FDI 1.0 Sales Dependency Purchase Dependency Sales Dependency * Imports Imports Exports Shipments Gov t Sales Capacity Utilization Industry Concentration

15 64 Interindustry Networks and Trade Policy Lobbying Table 5. Interindustry Goods Sales Network Structure and Industry Campaign Contributions, I II III IV V VI Lagged Industrial Contributions (0.126) Inward FDI (0.396)*** (0.376)*** (0.295)*** 0.33 (0.745) (0.774) Sales Dependency 10.2 (1.726)*** (2.072)** (1.401)*** (6.189) (6.479) Purchase Dependency (0.523)*** (0.483)*** (0.388)*** (1.157) (1.317) Sales Dependency* Imports (0.223)*** 0.89 (0.086)*** 0.52 (0.256)** (0.319)* Imports, (0.005)*** (0.004)*** (0.004) (0.002)*** (0.005) (0.005) Exports, (0.008)*** (0.006)*** (0.006)*** (0.003)*** (0.011)** (0.011)** Shipments, (0.0009)*** (0.0008)*** (0.0007)*** (0.0003)*** (0.001)* (0.001)* Government Sales, (0.003)*** (0.002)*** 0.02 (0.002)*** 0.02 (0.001)*** (0.04)*** (0.042)*** Capacity Utilization 0.02 (0.006)*** (0.005)*** 0.01 (0.004)** (0.002)*** (0.003) (0.003) Industry Concentration (0.004)*** (0.004)*** (0.003)*** (0.002)*** (0.005) (0.008) Observations Adjusted R-squared N/A 0.50 (Notes. Dependent variable: Industrial Campaign Contributions to Protectionist Candidates (who voted for any protectionist bills). Standard errors in parentheses. All estimations include constant. I, OLS with period dummies with control variables only; II, OLS with period dummies; III, OLS with period dummies with an interaction term; IV, OLS with PCSEs; V, GMM; VI, ECM. Significance at *10%, **5%, ***1%. Coefficients shown in millions US billion dollars.)