Book Search:  


Google full text of our books:


Privileging Industry:
The Comparative Politics of Trade and Industrial Policy
Fiona McGillivray

Book Description | Endorsements | Table of Contents

COPYRIGHT NOTICE: Published by Princeton University Press and copyrighted, © 2004, by Princeton University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. Follow links for Class Use and other Permissions. For more information, send e-mail to

This file is also available in Adobe Acrobat PDF format

Chapter 1

Redistributive Politics

DEMOCRATIC PROCESSES are intended to ensure equal representation. Yet the "one man, one vote" ideal does not mean that everyone benefits equally from political processes. Redistribution is the essence of politics, and electorally motivated politicians have incentives to redistribute resources among voters, privileging some at the expense of others. Who wins and who loses from such redistribution is at the heart of democratic politics and is the subject of this book.

Desiring to keep their jobs, politicians are ever willing to give to those who can keep them in power: winners are not chosen randomly. I focus on the winners and losers from trade and industrial policy. Some might argue that trade policy is a fundamentally international issue. Yet despite its consequences for the international flows of goods and services, trade policy is, like industrial policy, highly redistributive. Those people associated with the production of a protected good are privileged, while the population as a whole suffers from higher prices or higher taxes. There is remarkable variance in which industries win and how they are privileged; for instance, imported leather handbags face high tariff barriers in the United States, the Germans heavily subsidize the marine propeller industry, carpet firms are favored with government loans in Belgium, while the Spanish strongly support EU (European Union) quotas for the toy industry.1

Democratic institutions are ostensibly designed to serve the majority. Why then should democratic politicians want to enrich some chosen few and forsake others? The answer is political survival. Leaders assist those who can help them keep their jobs; yet who can most effectively help them keep their jobs depends upon the institutional context in which the politicians serve. As such, who the winners and losers are is shaped by domestic political arrangements. Institutional rules affect which industries legislators wish to protect as well as whether they can achieve protection for these groups.

The field of comparative politics seeks to identify the dimensions in which political systems differ and the impact of these differences. Although the myriad of combinations across all of these different dimensions produces a virtually limitless number of possible institutional arrangements, I focus on the impact of two prominent features of democratic institutions: the electoral rule and the strength of political parties. These two features have deservedly received much critical attention as they shape many aspects of political behavior; for instance, the number of parties, parties' policy positions, the level of political violence, the duration of government, and public policy outcomes.2 My goal, however, is to explain which industrial groups receive preferential treatment through the redistributive effect of trade and industrial policy.

The institutional features of the electoral rule and the strength of political parties play a key role in sorting the winners from the losers through two processes. First, the combination of the electoral rule and industry geography affects which industries legislators want to protect. That is to say, they induce preferences over which groups to protect. Second, the electoral rule and the strength of parties affect which industries legislators are able to protect. The legislative incentives created by these features determine how legislators' induced preferences are aggregated into actual policy. Rogowski (1987, 1998; Rogowski and Kayser 2002) is one of many scholars who consider the mapping from electoral rule to trade policy as a single step.3 As I shall argue, considering only one of these two processes is misguided. While Rogowski's and Kayser's work is extremely insightful, particularly in terms of explaining aggregate levels of protection, I believe it fails to account for the pattern of winners and losers under different electoral systems. Desperately wanting to help an industrial group is not the same as being able to help an industrial group. Similarly, simply because politicians are in a position to privilege an industrial group does not mean they will. It has to be in their political interest to do so. The provision of assistance requires both a will and a means.

The theory in this book is about redistribution. The logic of the theory tells us which groups of voters politicians will favor with redistributive policies. Trade and industrial policy are one means for targeting benefits to voters (Godeck 1985). Other policies, like public spending programs or welfare transfers, also have a redistributive component. So why would politicians use trade and industrial policy rather than other more "efficient" means to redistribute? Rodrik (1995) has been vocal in his criticisms of the trade literature for precisely this reason. While alternative tax-and-spend redistributive policies might have greater economic efficiencies, the question is, are they politically as effective? Throughout the book I argue that trade and industrial policy can be a politically efficient way to target key voters. Whether or not an industry is a good vehicle through which to redistribute income depends on the industry's geography and how it maps into electoral jurisdictions. Politicians do not want to protect any industry per se. They want to assist groups with precisely the right size, spread and location, to target benefits to politically important groups of voters.

While those wedded to the efficiency of alternative redistribution might remain unconvinced of the superior political efficacy of protection, trade policy has another great advantage over direct redistribution: it is opaque (Magee, Brock, and Young 1989). It is hard to know how much has been redistributed and to whom. If the government raises taxes to hand out additional benefits, such as increased welfare transfers, these acts are readily transparent. Tracing the beneficiaries of the U.S. sugar quota is not. Of couse the beneficiaries know who they are, but it is difficult for most voters to figure this out, far less to determine how much extra they themselves are paying for a pound of sugar. It is extremely difficult for voters to judge whether or not the sugar quota is based on sound economic reasoning (to help the beleaguered industry compete with subsidized EU sugar) or political vote-buying (pork for Florida's voters). Trade and industrial policies are also an indirect method of redistributing between geographic regions. For example, the U.S. sugar quota redistributes resources to Florida, Louisiana, and Hawai from the other forty-seven states. Many countries institute rules preventing direct transfers to regional groups; for example, in the United Kingdom, government spending to different regions is strictly controlled by a set of rules that assess need, based on criteria such as demographics and unemployment. However, the allocation of research-and-development funds cannot so easily be monitored or legislated against.

More broadly, there are many different ways that governments can give and take between industries: subsidies, tax exemptions, low-interest loans, debt reduction, tariffs, and quotas. Unfortunately, what makes trade and industrial policy attractive to politicians--its opaqueness--makes it difficult to study analytically. Conceptually straightforward, as a practical matter it is extremely hard to directly assess the level of assistance any particular industry receives. I believe this is why so many of the empirical results in the literature disagree.4 Looking at different measures of protection can lead to very different conclusions about its causes. As a result, comparative trade research has stalled. Rather than contribute another voice arguing for one measure over another, this research focuses on deriving additional consequences of trade and industrial policy. By deriving alternative dependent variables and testing hypotheses that relate to aspects of trade that are not dependent on direct measures of trade assistance, I hope to forward the study of comparative trade and industrial policy.

The arguments in this book also contribute to the literature on U.S. government redistribution. Scholars of U.S. politics disagree about which groups of voters are "purchased" with government transfers. For instance, Dixit and Londregan (1986, 1998; see also Lindbeck and Weibull 1987) argue that parties target policy transfers to swing voters. In contrast, Cox and McCubbins (1986) argue that parties are more likely to use transfers to reward loyal voters. Both of these arguments focus on only one step of the two-step mapping I propose. Using a comparative framework to consider both steps, I argue that the Dixit and Londregan (1986, 1998) hypothesis applies to one subset of countries--for example, the United Kingdom, Austalia, and Canada--while the Cox and McCubbins predictions apply to a different subset of countries--for example, Sweden, Germany, and Belgium. The predictions in this book are not unique; however, they are sensitive to comparative political institutions and are testable against other theories of redistribution.

Next, and throughout this chapter, I focus on how governments redistribute policy between industries. I use the case of the cutlery industry in the United Kingdom, Germany, and the United States to illustrate how the geographic and structural features of industry and the institutional characteristics of political systems affect how governments distribute trade and industrial assistance.


The focus of this book is to explain cross-country differences in which industries governments choose to assist. Take, for example, the cutlery industry: highly protected in some countries, it had been left to the vagaries of market forces in others. In Germany, Britain, and the United States, the cutlery industry had long been in decline,5 but the recessions of the 1970s threw the industry into crisis.6 Although the cutlery industry lobbied loudly for protection in all three countries, there was considerable variation in each government's willingness to assist. Successive governments in Britain heavily protected the cutlery industry; its effective tariff rate was as high as 30 percent, far greater than the average level of protection in Britain. By comparison, the German government did much less to support its cutlery industry, giving less than a half of the assistance that the British did, and in the United States, the cutlery industry received only token tariff protection from its government.7 Based on national security legislation, the U.S. Department of Defense was restricted from purchasing foreign stainless-steel flatware. Apparently using German forks undermined national preparedness.8

Why did the British lack the political will to kick out cutlery's crutch of protection? Industry pressure alone was not responsible; if that had been the case in the United States, Britain, and Germany, we would have seen the highest level of protection in the United States, where the industry was largest, or in Germany, where the industry was well organized. In fact, what we observe is protection directly inverse to the size of the industry and uncorrelated with the level of organization. What is missing is a grasp of the political incentives to supply protection. Later in this chapter, I will argue that British cutlery's political leverage was, in large part, due to its regional concentration in the type of electoral districts that make or break governments. In Britain, the cutlery industry was small in size, but its firms were regionally clustered in politically important districts. Cutlery's geography made it an ideal vehicle for the government to redistribute income toward key groups of voters; hence cutlery's high levels of protection. Under the German and the U.S. electoral systems, however, cutlery's geography made protecting the industry much less politically profitable. In the latter case, regional concentration became the cutlery industry's Achilles' heel, the reason it failed to win protection. Throughout the book, I argue that it is the joint effects of an industry's geography and the electoral system that determine the political opportunities for industries, such as cutlery, to gain industrial assistance. This approach differs from extant explanations in which the industry's structure and ability to lobby matters most. Rather than focus on the incentives facing industry to organize and demand protection, I focus on the incentives facing governments to reward particular groups of voters. There are many different ways that government can redistribute resources among voters. Trade and industrial policy can be a politically efficient way to target key voters. Whether or not an industry is a good vehicle through which to redistribute income depends on the industry's geography. The purpose of this book is to show when governments use trade and industrial policy for political goals and to show why aiding an industry can be a politically efficient way for government to redistribute from one group to another.

Who the winners are and who the losers are from trade and industrial policy depends on the interaction of industry geography and the electoral system, a two-stage process. In the first stage of the argument, legislators have preferences induced over which industries they want to assist as a function of each industry's geographic distribution and the electoral rule. An industry's geography relative to political jurisdictions induces these preferences, determining whether or not legislators want to privilege an industry. In the second stage, the electoral rule and the strength of parties determine how legislators' induced preferences are aggregated into actual policy. It is one thing for a legislator to want to protect an industry and another for her to be able to do so.

In the case of cutlery, the initial link between industry geography and induced preferences is similar in the United States and Britain (the electoral rule is the same); however, in the second stage, differences in how legislative preferences are aggregated into policy strongly disadvantage a small, regionally concentrated industry such as cutlery in one case but strongly advantage it in the other. Thus the second stage is as important as the first. As I hope to demonstrate, most extant explanations of how governments' choose to redistribute fail because they examine only one of these two stages.

Obviously, the type of electoral system and the strength of parties are not the only political factors that come to bear on the setting of tariffs or on legislative policy making; for example, the role of international negotiations is a common topic of trade research (Grossman and Helpman 1995; Levy 1997; Marvel and Ray 1983; Pahre 2001, 2002). Yet focusing on these features of the political system as a starting point provides leverage in understanding how industry geography and political structure influence other aspects of the political process of trade. For instance, while it is true that international agreements might prohibit policies that governments would otherwise pursue, governments are unlikely to negotiate treaty exemptions for industries that they do not want to protect anyway. On the other hand, they are likely to fight long and hard at the negotiating table for those industries that they deem essential for their political future.

In this chapter, I start by assessing our current understanding of how industry structure and geography influence industrial and trade policies. This issue is far from resolved on empirical grounds. Findings conflict because there is no general, unambiguous relationship between industry geography and political influence. Rather, the relationship between industry protection and industry geography is contingent upon institutional features of the political system. Following this review, I outline my approach in this book and consider some of the data and measurement issues that make testing so difficult in this field. A large part of the problem is that trade and industrial policy lack transparency. In Germany, cutlery gets three times the subsidies that steel gets; however, steel is heavily protected by tariffs. The question is, what size of subsidies to cutlery is equivalent to the dollar amount of tariffs won by steel? Compounding this problem, different governments tend to prefer different policy instruments to favor industries: the Swedes often opt for debt guarantees; the British generally favor regional grants and development subsidies; the Germans make liberal use of tax exemptions; and "Buy America" legislation and tariffs tend to be the United States's policy instruments of choice.

To overcome these problems I use a combination of case studies and statistical tests. Most empirical tests focus on assessing the value of protection, or assistance, an industry receives and relating this to economic and political factors.9 Unfortunately, even in the case of direct tariffs (which are becoming increasingly rare), industrial assistance and protection rarely leave a smoking gun. As such, many direct tests are inconclusive. Although I do use industry tariff data to test the argument, I supplement this tariff-based test by using my theoretical explanation to derive auxiliary hypotheses. These hypotheses suggest new dependent variables, such as the dispersion of stock prices within capital markets and government influence in the targeting of plant closures within declining industries.

As fans of courtroom dramas know, a case without a smoking gun often needs to be built on circumstantial evidence, and the more extensive and the more varied the evidence, the better the prospects for conviction. Testing new dependent variables does more than simply provide confirmatory evidence of direct tests. Many theoretical perspectives may produce parallel predictions on a particular dimension. When theories make such parallel predictions, it is impossible to distinguish between them without observing how they fare against the evidence on a different dimension upon which they do not agree. Generating additional hypotheses and testing new dependent variables not only supplement the direct evidence supporting one set of theories; it also allows us to differentiate between the arguments within this set.


So why did the cutlery industry receive higher levels of trade and industrial assistance in Britain than in Germany or the United States and in fact higher levels than most other British industries? Protection and industrial assistance are both highly redistributive. By using tariffs, subsidies, voluntary export restrictions, or other nontrade barriers (hereafter NTBs) to protect one industry from international competition, governments either increase the price at which the industry can sell its wares or lower the industry's cost for producing its wares. This assistance provides economic rents for the industry.10 Those associated with the industry, its employees and owners, benefit from protection. In the short term it is costly for investors to shift capital between industries and costly for workers to find new jobs.11 Both labor and management have incentives to demand protection for their industry. That said the ease with which labor and capital can shift between industries varies across countries and over time; this will affect the extent that political divisions are based on industry rather than class (Hiscox 2001). For example, in Germany better unemployment benefits make it less costly for labor to shift between industries. Nonetheless, in an industry like cutlery, where it is difficult for workers to carry over traditional skills to other high-paying jobs, the political division in Germany remains firmly along industry lines.

If the industry wins trade assistance, there are often additional kick-on effects for the immediate community associated with the industry. For example, if without trade protection a firm would close, then protection helps the entire local workforce, since the rise in unemployment associated with plant closure would hurt all workers' ability to demand higher wages. The effects of plant closure are often very severe, particularly in small "company towns." When the main employer in a town fails, so do many of the subsidiary and service businesses. Additionally, the local tax base and property values often collapse, hurting the town as a whole.

Protection and industrial assistance target benefits at those people and areas associated with the particular industry. Yet these benefits are not without costs. While protection allows an industry to extract rents through charging higher prices, this harms the welfare of consumers who must pay these, higher prices and, in the case of some NTBs, pay higher taxes. These costs are dispersed, however; everyone suffers slightly through higher prices. The benefits are concentrated among producers.12 This concentration of benefits and diffusion of costs means producers are more motivated to organize and lobby than consumers.13

There can be cases where the costs of protection are concentrated. For instance, if the product of a protected industry is an input for another domestic industry, then that second industry suffers. Cutlery firms, as well as companies that make products such as oil equipment and plastic moldings, are hurt when tariffs are imposed on imported steel.14 Indeed, President Bush's imposition of steel tariffs in 2002 had severe adverse affects on U.S. auto makers.15 More generally, rival foreign industries are harmed by protection, since it reduces demand for their products. Foreign governments might impose retaliatory tariffs, which will hurt domestic export industries. However, the likelihood of any particular exporter being hurt by these retaliatory tariffs is small; hence, it is unlikely that export industries will coordinate to oppose protectionist legislation.16 Of course, as the size of the protectionist demands becomes large, so does the probability that any single exporter will be hit with retaliatory tariffs; this increases the probability that an antiprotectionist coalition will form (Destler and Odell 1987).17

While it is always important to bear these exceptions in mind, in most cases, protection and industrial assistance provide concentrated benefit to those associated with the industry and to those in the immediate geographic vicinity. In contrast, the costs of protection and industrial assistance are evenly spread over the entire population. From a purely microeconomic viewpoint, the costs of trade assistance are widely believed to outweigh the benefits.18

Redistribution is the heart of my theory. Protection or industrial assistance privilege some groups at the expense of others. Who the winners are and who the losers are is the stuff of politics, not economics. Therefore, with respect to the underlying economic model, I work with the simple stylized fact that protection and industrial assistance provide concentrated benefits to areas associated with the protected industry, with the costs being dispersed over everyone. In the short term, it is costly to relocate capital or retrain workers for a new job. I assume labor and capital within the industry and individuals in the regional community associated with the industry have similar trade preferences. The political division is both between industries and between geographic regions.

Maintaining protection for the cutlery industry meant that at the height of its protection, British consumers paid about 30 percent (this is the approximate effective tariff) more for cutlery than they otherwise would. This protection also harms the interests of foreign cutlery workers. Why would the British government choose to privilege British workers in the cutlery industry at the expense of the general public and workers overseas? The latter is perhaps easy to explain: foreigners do not vote on the government's future. But why help cutlery workers at the expense of the general voting public? Most of the literature focuses on an industry's motivation to organize and its ability to lobby as the reason for its political clout.19 Such theories would argue that cross-national differences in the level of protection result from cross-national differences in the structure of an industry.20 These arguments implicitly assume that across all countries, the biggest and loudest demanders for protection are the ones who receive the most assistance. However, politicians do not automatically respond to the loudest demands. They listen to the industries that affect their chances of reelection. Hence, which industries receive assistance depends upon how industry geography and political institutions interact to create an incentive for the government to help an industry. Sure, an ability to shout the loudest might help, but politicians have to be willing to listen.

As a vehicle to motivate my arguments, I concentrate on the case of British cutlery. Using this as an illustrative example, I show that many extant arguments are deficient. In the process of evaluating these arguments, I introduce the components necessary for my theoretical approach.


The town of Sheffield, in South Yorkshire, is home to the United Kingdom's table cutlery, handknives, and handtools industry. The manufacture of Sheffield knives can be traced from the fourteenth century. The English poet Geoffrey Chaucer (ca. 1340-1400), said in Canterbury Tales of the miller of Trumpington, "A Sheffeld thwitel baar he in his hose."21 For such a small and industrially fragmented industry, it has received a remarkable level of assistance (Cable 1983). Successive governments in Britain have heavily protected the cutlery industry. As mentioned earlier, in the 1970s, the effective tariff rate for cutlery was around 30 percent (Cable 1983).22 In the late 1970s and 1980s, the cutlery industry received additional assistance via government grants and quantitative trade restrictions. The British cutlery industry also obtained voluntary export restrictions on a variety of products from Japan and South Korea (Tweedale 1995). Cutlery received higher protection than the average industry in the United Kingdom and higher protection than the cutlery industry in Germany and the United States. In the 1970s and 1980s the effective tariff rate on cutlery in Germany was around 6 percent. The German government supplemented this with a variety of domestic subsidies, mainly in the form of regional grants, which were worth about 1.3 percent of the value added in 1972 (Glismann and Weiss 1980; Weiss 1983). The United States has largely ignoredthe plight of its cutlery industry. Although it received an effective tariff rate of about 5 percent in the 1980s (Lavergne 1981), it received few domestic subsidies. Since the 1950s there have been a series of on-and-off quotas; however, these have been largely ineffective (USITC 1978, 1985). In no case did state aid prevent cutlery's decline, but in the United Kingdom and Germany, it significantly slowed the process. The decline of the cutlery industry in the United States has been drastic, with cutlery almost completely eliminated from Connecticut, a state that together with New York once formed the basis of the U.S. cutlery industry.

The structural features of the cutlery industry are similar in all three countries. Cutlery is a small industry in decline. It consists of a heterogeneous set of firms, meaning in this case that many of the firms within the industry have different preferences over trade policy. It is industrially decentralized, meaning that production is spread over a number of firms. Yet cutlery is highly geographically concentrated, since most of its firms are located in close proximity to each other. In the next section, U.K. cutlery serves as a platform from which to assess how industry competitiveness, size, industrial concentration, and geographic spread affect an industry's political leverage.

Declining Industries

The U.K. cutlery industry has been in decline since the U.S. Tariff Act of 1890.23 However, the British recession of the late 1970s and early 1980s created a major crisis in the industry.24 In 1980, the industry had a gross output per employee, wage per quarter, value-added per employee, and fixed capital per employee below the manufacturing average (see table 1.1). In the industrialized world, labor-intensive, low-wage, low-value-added industries tend to receive higher than average levels of protection.25 These industries have a comparative disadvantage in a world market where they face steep competition from foreign imports.26 This was true for the U.K. cutlery industry. In the 1980s, foreign imports had a 38 percent share of the U.K. cutlery market (see table 1.1). The long-term prospects for the U.K. cutlery industry are bleak.27 Part of the reason declining industries demand, and win, protection is that it is costly for workers and owners to shift out of the industry. In labor-intensive industries, large numbers of voters must find new jobs. Politicians often argue that state aid is not intended to prop up declining industries such as cutlery; rather, it is to ease the adjustment process for workers and owners (Baldwin 1989).28

There is little doubt that the plight of the cutlery industry--company bankruptcy, heavy job loses, and surging import competition--spurred the government to assist the industry in the 1970s with exemptions from tariff reductions, grants, and quotas (Tweedale 1995). Across countries, there is strong, robust evidence that industries facing a surge in import competition receive more favorable levels of trade assistance (Trefler 1993; Riezman and Wilson 1992; Rodrik 1995). However, not all of the United Kingdom's struggling industries received as favorable levels of assistance in this period as the cutlery industry did; for example the shipbuilding industry was left to sink or swim with minimal political support.29 This suggests that cutlery's declining competitiveness was only part of the explanation for its privileged levels of government assistance.

Industry Size

The cutlery trade is a small industry and this typically works against an industry's ability to demand protection (see table 1.1). In 1980, cutlery had about nine thousand workers; today that number is about two thousand (HMSO 1983; ONS 2001). That is not a lot of votes. Larger industries have more workers--hence they have more voters, plus they have greater access to funds for lobbying (Becker 1983; Busch and Reinhardt 1999; Caves 1976; Pincus 1975). The plight of large industries is also more likely to be covered in the media and to arouse public sympathy. On the other hand, being ignored by the media can have benefits; smaller-sized-industries are less likely to arouse antiprotectionist opposition than are larger industries (Mayer 1984). Smaller industries are easier to organize and can also be "cheaper" to protect (Olson 1965). Crossnationally, however, studies have found that industries with larger numbers of employees (large size) tend to receive more favorable levels of assistance (Anderson and Baldwin 1987). Cutlery's small size probably put it at a political disadvantage vis-à-vis other industries seeking protection.

Industrial Concentration

Industrial concentration is defined as the degree to which a small number of firms in an industry dominate the market.30 In 1980, 55 percent of cutlery workers were employed in firms with fewer than two hundred employees--a far higher proportion than the average for all manufacturing industries (see table 1.1). Small firms, such as Schofields or C. W. Fletcher, characterized the industry. There were a few large firms, such as Viners; in fact, the output of the five largest firms accounted for 43 percent of the industry's output (Cable 1983). Nonetheless, compared to other industries, cutlery was relatively industrially decentralized.

Industrially concentrated industries, where the benefits of protection are shared among a few large firms rather than many smaller firms, have stronger incentives to organize and lobby.31 For an industrially decentralized industry, the costs of lobbying probably exceed the benefits of protection. Individual firms have incentives to free ride on the efforts of others (Cornes and Sandler 1996; Olson 1965). If these firms want to organize and be effective, they need to coordinate lobbying activities. This can be difficult to achieve. On average, industrially concentrated industries win higher levels of trade protection than industrially decentralized industries (Anderson and Baldwin 1987; Riezman and Wilson 1992). Cutlery's industrial decentralization should have undermined its ability to organize as an industry and demand assistance. Yet as I describe in the next section, cutlery had a prolific lobbying history.

Firm Heterogeneity

The structure of firms is important in determining trade preferences (Milner 1988). In a heterogeneous industry, many of the firms within the industry make different products, play different roles in the production process, or rely in different ways on export and domestic markets. In the case of cutlery, firms had different trade preferences because some firms--typically the larger ones--used semifinished cutlery imports in the production process. Viners, for example, cared less than smaller firms about the artificially high cost of U.K. steel because a large part of its business was importing semifinished products, finishing them, and then stamping "made in Sheffield" on the final product. Using the Sheffield trademark on semifinished imports created much fury among the smaller firms.32 If firms have different preferences about how the government should help, this can undermine the industry's lobbying efforts (Milner 1988). Cutlery certainly has a rich and complex history of lobbying in the United Kingdom. However, it does not have an effective lobbying history, in part because the industry was divided over goals (Cable 1983).

Cutlery is a highly fractured industry on both management and labor sides, with numerous business and union associations. Cutlery has four manufacturing associations: the Cutlery and Silverware Association (CSA), the Federation of Cutlery Manufacturers (FBCM), the Sheffield Spoon and Fork Makers Association, and the Manufacturing Silversmiths Association (Cable 1983; Tweedale 1995). Among the things they disagree about is the type of trade assistance the industry should receive: tighter protection or more restrictive quotas on semifinished products. Unionized cutlery workers are split between three rival unions: the National Union of Goldsmiths, Silversmiths, and Allied Trades (NUGSAT), the Transport and General Workers (TGWU), and the General and Municipal Workers (GMWU). The unions have typically lobbied for the industry (Cable 1983) but, like the manufacturing organization, are at odds over goals. The industry is split between four manufacturing associations and three unions. These divisions partly reflect the industrial heterogeneity of the industry and partly reflect the structure of unions in the United Kingdom. Cutlery was organized, but industry heterogeneity undermined cutlery's ability to coordinate effectively to demand protection. Thus it remains to be explained why cutlery won such high levels of government assistance.

Geographic Concentration

Numerous scholars have contended that geographic distribution affects an industry's ability to lobby and secure assistance.33 Geographical concentration relates to the spatial concentration of firms within an industry. The cutlery industry is about as geographically concentrated as it is possible to get--over 50 percent of the industry is located in the town of Sheffield in South Yorkshire.34 Most scholars argue that geographic concentration has a positive effect on an industry's level of protection. Nevertheless, there is disagreement whether it is spatial proximity or asset specificity that positively affects a geographically concentrated industry's ability to organize.

Industries that are spatially proximate are thought to benefit from lower transactions costs in organizing industry-wide lobbying activities and mobilizing industry workers to vote (Busch and Reinhardt 1999; Chwe 2001; Pincus 1975; Schonhardt-Bailey 1991). Spatial proximity makes it easier for firms to coordinate lobbying activities (Chwe 2001). The spatial proximity of workers and firms in Sheffield lowers communication costs within the industry. These workers share social networks (down at the pub--the British version of the water-cooler effect). If communication costs are low, firms are able to sanction free riders; this makes it less costly for firms and workers to organize as an industrial lobby.

Others argue that geographically concentrated industries are more politically active because their assets are highly specific (Alt et al. 1996). As it is difficult for labor and capital to shift out of such industries, they have incentives to organize politically to defend their jobs and investments. Cutlery workers are highly skilled, but only in the cutlery trade; these skills do not transfer easily to other industries. In Sheffield, it is difficult for unskilled laborers to find new employment (Hayter 1985). Nowadays, the local council is the largest single employer in the city. Labor has incentives to organize, raise funds, and get the vote out in an effort to save immobile jobs.

In a different line of argument, economic geographers make the point that a geographic concentration of firms in an industry--geographic agglomeration--improves the competitiveness of individual firms (Krugman 1991; Sabel 1989; Scott 1988).35 We might therefore expect geographically concentrated industries to be less in need of protection. In the case of U.K. cutlery, however, innovation in the industry stalled in the 1800s despite the benefits of geographic agglomeration (White 1997).36

The empirical findings on the effect of geographic concentration on the level of protection are ambiguous. In some studies, geographic concentration is positively related to levels of protection. In others the effect is negative, and in still others, no significant effect is found (Anderson and Baldwin 1987; Hansen 1990; Lavergne 1983; Nelson 1988; Pincus 1975; Rogowski 1997, 1998; Schonhardt-Bailey 1991). Part of the problem is that in many studies, industrial concentration is used as a proxy for geographical concentration (for example, Cable and Rebelo 1980). However, an industry that is industrially concentrated is not necessarily spatially concentrated. For example, the car industry in Australia is industrially concentrated in a handful of firms, but car plants are spread throughout the country. In an important step forward, Busch and Reinhardt (1999, 2003) construct a measure of geographic concentration that captures the spatial relationship between geographically proximate firms. They categorize geographic concentration as a decreasing function of the physical distance from plants of workers to the national midpoint of an industry. Using NTBs in the United States and Europe, they find strong evidence that geographically concentrated industries receive more favorable trade assistance. Although Busch and Reinhardt's measure of geographic concentration captures spatial concentration, this does not necessarily measure an industry's regional concentration. For example, the footwear industry in the United States is heavily located in the states of Maine and Missouri. According to Busch and Reinhardt's measure of spatial dispersion, the footwear industry is geographically dispersed; however, footwear is regionally concentrated in only two of fifty states. Later in the book I argue that when the theoretical focus is the government's decision to supply protecion, rather than the ability of industry to demand protection, regional concentration is a more relevant concept than spatial concentration.

Cutlery is an industry that is both spatially and regionally concentrated. How does this geographic concentration affect cultery's political clout? Cutlery has a strong history of lobbying for import-relief. For such a small, industrially decentralized industry, it has made noisy demands for aid. Unfortunately, it is hard to determine the impact of geographic concentration in this case. Spatial concentration in one town probably explains why cutlery is highly organized; however, firm industrial heterogeneity has prevented the coordination of lobbying activities. One might expect that geographic concentration--most cutlery firms working together in the same city--would generate coordination over time. The spatial proximity argument assumes that workers from different plants and firms rub shoulders together because they work closely together. Yet cutlery firms are described as aggressively independent.37 In Sheffield, spatial proximity seems to breed contempt, not cooperation. Spatial proximity by itself does not ensure coordinated lobbying activity. A local mechanism is needed to get workers from different firms together, even in the extreme case where they all work in the same city. In the United Kingdom, the union and manufacturing associations assume this role, although in this case to the industry's detriment. Social networks in Sheffield might have lowered the costs of communication, and specific assets have increased workers' and managements' incentives to organize, but the fragmented union structure has meant that workers are either socializing within their union (preaching to the converted) or socializing across unions with less effect. While on average, geographic concentration might lower the costs of coordination and lobbying, in the case of cutlery, fragmented unions and industrial heterogeneity forestalled the creation of a united lobbying front. Cutlery's lobbying efforts were largely ineffective (Cable 1983). Thus it still remains to be explained why cutlery gained much higher assistance than many other troubled industries did.

Electoral Concentration

Electoral concentration is an industry's spatial concentration relative to political jurisdictions;38 is an industry geographically concentrated in a few electoral districts or dispersed over many electoral districts? An industry that is concentrated in one electoral district is probably such an important employer in that district that its representative cannot afford to ignore the industry's demands. However, only one representative cares strongly about the fate of the industry. Most scholars argue that electorally dispersed industries have more political clout; as an industry becomes more electorally dispersed, more legislators care about its fate, which makes it easier to create voting blocks and build legislative majorities (Caves 1976; Pincus 1975; Schattschneider 1935). As in the case of geographic concentration, the empirical findings on the effect of electoral concentration are mixed.39 Scholars frequently use measures of electoral concentration as a proxy for both electoral and geographic concentration (see studies in Anderson and Baldwin 1987). Given that the predicted effect of geographic concentration is positive and of electoral concentration is negative, it is not surprising that findings are often insignificant and nonrobust. It is possible for an industry to be geographically concentrated but electorally dispersed if the region contains a large number of electoral districts (Busch and Reinhardt 1999; McGillivray 1995, 1997).

That said, there is evidence that electorally dispersed industries have political clout, particularly when industry size is taken into account (Glismann and Weiss 1980; McGillivray 1997; Pincus 1975; Tharakan 1980). A large, electorally dispersed industry is likely to receive favorable protection.40 Such an industry gains from both sides of theoretical debate. Being large, even though dispersed, the industry is still sufficiently important in each district that legislators care about it, and being in a large number of districts helps the industry create voting blocks. By contrast, a large, electorally concentrated industry has a single legislator who cares passionately but who is unlikely to be able to provide assistance by herself.41 For small industries, electoral dispersion is a trade-off. If an industry is thinly spread although present in a large number of districts, its impact on jobs and the local economy is too insignificant for legislators to care. For a small industry, electoral concentration might be best since then at least one legislator might care.

The cutlery industry in Britain is electorally concentrated in the constituencies of only six Members of Parliament. It is a major employer in those districts (in 1981, approx 15 percent of the Sheffield workforce were in the cutlery and tools trade).42 However, only about 1 percent of legislators care directly about cutlery's fate. Cutlery's electoral concentration should work against its efforts to secure trade assistance. Yet as we shall see shortly--given the electoral rule and the structure of the party system--cutlery was located in the right 1 percent of districts in the British case.


While there is some disagreement over the directionality and relative importance of the factors above, it would appear that a small industry with heterogeneous firms that is industrially dispersed over many firms, geographically dispersed yet electorally concentrated, is the least likely industry to receive assistance. By these standards, the U.K. cutlery industry is damned on all measures except geographical dispersion. And the benefits from geographic concentration are negligible given cutlery's heterogeneous firms and fragmented union structure. Yet, despite all these disadvantages, cutlery secures relatively high levels of government assistance versus other troubled U.K. industries, and certainly more than its rival cutlery industries in the United States and Germany.

What accounts for these differences in success? The cutlery industry is somewhat larger in Germany and the United States. It is more geographically dispersed in the United States, and the German industry is more competitive. But otherwise, the cutlery industries in all three countries are structurally very similar (Hayter 1985). Therefore, the question remains, why was British cutlery more successful in gaining assistance? Next, I argue that to understand crossnational differences in cutlery's level of government assistance, one has to look at the interactive effect of the industry's geography and political institutions.

The Interactive Effects of Political Institutions

The joint effects of an industry's geography, the electoral rule, and the strength of parties determine the political opportunities for industries such as cutlery. An industry's geography is defined as its size, its location, and its spread over electoral districts and regional locals. In the U.K. case, cutlery is a small industry, electorally concentrated over six districts and geographically concentrated in the town of Sheffield in Yorkshire. Cutlery is unlikely to have widespread parliamentary support; only six legislators directly care about the fate of cutlery. However, deriving legislators' induced preferences is only part of the story--equally important is how legislators' induced preferences are aggregated into policy decisions. In the United Kingdom, backbench Members of Parliament are not the important policy makers. Political parties are strong, and the party in government makes the policy decisions (Bailey 1979). The fate of cutlery depends on whether or not the ruling party elite cares about winning Sheffield's electoral districts. The loss of thousands of jobs in the cutlery industry would be devastating for Sheffield and the surrounding countryside in South Yorkshire. Yet why should a national party be concerned about these local effects? The answer lies in the marginality of these districts and their importance in determining which party wins the next election.

The United Kingdom is a plurality system with single-member districts. As Duverger (1954) first showed, such systems devolve to two-party competition, with the party that captures the greatest number of seats forming the government. As a practical matter, party support varies across electoral districts. Some districts, such as those in the home counties, strongly support the Conservative party, while others, such as the North of England and the West Country, strongly support the Labour party. Except under extreme conditions, it is rare for these safe seats to change hands. Other districts are party competitive; the elections are close and the seats regularly change hands. It is these latter marginal districts that are the key to electoral success.

The party in government could adopt trade and industrial policies that reward its key supporters in safe districts. While such policies might ensure that the government wins an even higher proportion of the vote in these districts at the next election, it does not help it capture or hold the marginal districts it needs to maintain a legislative majority. Alternatively, the government might target assistance to opposition strongholds in an attempt to lure voters away from the opposition. While such attempts might work, the government has so much ground to make up in opposition strongholds that it is unlikely to capture these districts. Such a policy might succeed in increasing the government's vote share but is unlikely to help it maintain its seat share. Since governments in strong-party majoritarian systems care primarily about maintaining a legislative majority, assistance is best provided to those districts where electoral support is most likely to win the election.43 Simply put, a dollar is best spent on the district where its impact is most likely to influence who wins the district. Policies that improve the popular vote, per se, politically misallocate resources.

Vote differentials between the top two parties are the easiest way to determine which seats are party competitive. Marginal seats are commonly defined as seats that can be gained with a swing of 5 percent or less (Butler 1975; Butler and Kavanagh 1974, 1980; Butler and King 1966; Butler and Pinto-Duschinsky 1971). Another measure of key seats is the 5 percent of government seats held by the smallest majority. Although parties do not officially define what a marginal seat is, they often have lists of critical seats (Butler 1975; Butler and Kavanagh 1974; Butler and Pinto-Duschinsky 1971). A large third-party vote can obscure the identity of key seats. Contemporary newspaper accounts often provide a clear indication of which seats will be critical at the next election. Usually, these correspond to those seats with close vote totals between the two major parties.

Today's Sheffield is one of the safest, most militant Labour strongholds in the country; every seat in South Yorkshire bar Sheffield Hallam is Labour held.44 West Yorkshire and South Wales are the regions heavily concentrated with marginal districts. Traditionally, the outer regions of Sheffield, for example, Sheffield Heeley and Sheffield Hillsborough, contained party-competitive seats. Sheffield Heeley is immediately south of the city center and was a marginal seat particularly in the early 1970s when it could be gained with a swing of 1 percent or less.45 More generally, this region of the country gathered a large number of Liberal votes, so three-way races were common.46 Where the Liberals drew a high third-party vote, the outcome of the election was harder to gauge. Hillsborough and Hallam--where the Liberals gained almost 20 percent of the vote--were also party-competitive seats. I argue that successive governments, both Conservative and Labour, came to cutlery's aid during the recessions of the 1970s because the industry was located in party-competitive seats. Ignoring cutlery's plight would have meant the loss of politically important marginal seats for the ruling party.

Some might want to dismiss the marginality arguments, since cutlery was distributed over only a few marginal districts and not all of Sheffield's seats were marginal. The government has less incentive to privilege such safe seats. However, such complaints fail to account for the importance of these marginal seats. In the closely run 1970 and 1974 elections, such marginal districts were seen as "make or break" for the incumbent government (Butler and Kavanagh 1974; Butler and Pinto-Duschinsky 1971). Furthermore, cutlery is a small industry, so even though some of the benefits go to the electorally less important safe districts, the cost of protecting cutlery is relatively small. No other industry allows the government to so accurately target benefits to such electorally important districts.

Changes in the political landscape since the 1970s provide further evidence. Sheffield is now a safe Labour region, and the Conservatives have little prospect of capturing seats there. Given this Labour stronghold, Conservative governments have little incentive to continue to privilege cutlery, and neither do Labour governments. Lack of government support is not the only reason the cutlery industry has waned, but the end result is that cutlery is now reduced to a cottage industry in safe Labour seats. Nowadays, it has fewer than two thousand workers.47

Political parties are strong in both Germany and the United Kingdom.48 Differences in the electoral mechanism help explain why the cutlery industry gained more government assistance in the United Kingdom. Unlike the United Kingdom's majoritarian system, proportional representation (PR) systems, like that of Germany, do not have "marginal districts." Districts are multimember and seats are distributed based on the proportion of votes within each district. Germany is a mixed-member proportional representation system. Half of parliament is elected from single-member districts and the other half is selected from a list vote. Despite these single-member districts, overall the system is proportional since the overall allocation of seats in the Bundestag is decided by the nationwide party vote. Each party is awarded list seats in addition to their district seats until the total number of seats is proportional to the number of votes the party received.49 If the ruling party in Germany loses a "marginal" single-member district seat, its votes are not wasted votes, since these votes will be used to help determine the overall distribution of seats in the Bundestag.

Which groups of voters do German parties target with policy benefits? In strong-party PR systems, the parties in government control the policy decision. Parties care about maximizing votes, since these translate almost directly into seats. The electoral process provides incentives for parties to target benefits to their core industrial supporters and to favor industries located in regional voting strongholds (even if the industry votes for another party, its fate affects the prosperity of the local economy). Whether or not cutlery wins protection in Germany depends on whether or not it supports a party in government or, at a minimum, whether the town of Solingen, where cutlery is based, is a partisan stronghold for a party in government.

Like British cutlery, German cutlery has a long historic tradition and has also been in decline for much of the twentieth century (Boche 1997; Harrigel 2000). Despite a brief comeback in the1950s, the industry struggled to compete during recessions in 1970s and 1980s (Hayter 1985). Cutlery is geographically concentrated in the town of Solingen in the Ruhr region of Germany. It is the major industry in Solingen; about a quarter of the Solingen's workforce are employed in the cutlery and tools trade (Sträter 2003). The industry was, and is, larger than that in the United Kingdom (it has also been more competitive than the British cutlery industry).50 In Solingen in the 1970s and 1980s there were around six thousand workers and forty cutlery firms, such as Henckels, Puma-Werk, Merkur, and Trident. Nowadays, the cutlery industry in Solingen is much more reduced in size (Sträter 2003). Cutlery workers are traditional supporters of the Social Democratic party (SPD), and this region of Rhurgebiet, Solingen-Remscheid, is a strong base of support for the SPD.51

At the federal level, the SPD was in power through much of the 1970s. This benefited cutlery, although obviously not to the same extent as cutlery was favored in the United Kingdom. Cutlery is a small industry within the SPD's large group of industrial supporters. Given its relative size, it is perhaps not surprising that it won only limited assistance compared to British cutlery. On the other hand, because cutlery is a small industry, protecting it would have been relatively "cheap." However, the SPD was not alone in determining government policy. In the 1970s, the SPD was in government with the Free Democratic party (FDP). Most (PR) Proportional Representation governments are multiparty, and this affects how parties' preferences are aggregated. In order to form a government, the SPD needed, and still needs, malleable policy goals in order to accommodate its coalition partners' preferences. In general we should expect trade and industrial policy to be a compromise of what each party in the coalition government wants. Indeed, when the SPD was in coalition government with the Greens in 1998, the SPD cooperated on energy reforms that hurt smaller businesses, such as cutlery, to accommodate its coalition partner's desire for tougher environmental regulations.52 This does not imply that PR systems are generally less protectionist than majoritarian systems. How parties' preferences are aggregated in coalition governments strongly affects the pattern of redistribution in the strong-party PR case.

The United States is a majoritarian system with single-member districts, much like the United Kingdom. However, unlike in the United Kingdom, the party system is weak and legislative outcomes are not determined by party elites. Instead, policy depends upon a majority coalition of legislators forming to pass legislation. Unlike the United Kingdom's or other high-party discipline systems, legislative votes are not strictly along party lines, and the coalition that forms for one bill might be completely different from the coalition that forms for another. It is these coalition dynamics that determine whether or not a legislator wins import protection for her industry. In both the United States and the United Kingdom, the majoritarian system induces similar preferences in legislators, but the aggregation of these preferences differs drastically between these systems.

In the postwar period, a small, highly geographically concentrated industry like cutlery, located largely in Connecticut and New York, found it very difficult to build the legislative majority coalition it needed to win protection. In weak-party systems, agenda-setting rules in the legislature are extremely important in determining outcomes. The cutlery industry was larger in the United States than it was in either the United Kingdom or Germany. In 1981 it was composed of about two thousand companies and around one hundred thousand workers--however, many of these firms were in the specialized and highly competitive knife trade, concentrated in Oregon (USITC 1985). Stainless steel flatware production was clustered in parts of upstate New York and in Connecticut (USITC 1985).53 The industry is more industrially concentrated than that of its European rivals; in 1981, the top four companies controlled 85 percent of total production in stainless steel flatware (USITC 1985). Taken together, these factors should have favored the cutlery industry with higher-than-average levels of protection. From 1956 to the early 1970s the government used escape-clause action to place on-again, off-again quotas on Japanese cutlery; however, these had little impact (USITC 1978).54 Recessionary pressures and a flood of cheap foreign imports hit the U.S. cutlery industry in the late 1970s, and it made repeated pleas for quotas to be reinstated. In 1978 and 1983 industry appeals for quotas were denied.55 The industry tariff remained about 5 percent--significantly lower than in the United Kingdom and Germany (Lavergne 1981).

Why did U.S. cutlery fail to get the protection it sought, and what might have helped it? Had cutlery been more politically dispersed, it would have had a presence in more legislators' districts. Unfortunately, cutlery is a small industry. If spread any thinner, its employment impact on any district would have become so small that no legislator would have cared sufficiently about it to have helped. Given its small size, cutlery's best chance would have arisen if it were geographically concentrated in the district of a representative with institutional leverage, such as the chairmanship of legislative budgetary or trade and industry committees (as we shall see in the case of footwear later). But it was not in such a district. Before a representative obtains a strong institutional position, she requires seniority. Unlike the U.K. case, where marginality helped, a representative from a party-competitive district in the United States is unlikely to develop the seniority required to gain institutional leverage. Under the U.S. political system, cutlery's political geography is part of the reason it failed to secure the financial assistance it needed to stay open.

How Does Industry Geography Matter?

The case of cutlery reveals that the relationship between industry size, spread, and location and the level of trade assistance is a contingent one that depends upon the interactive effects of political institutions. It is hardly surprising that the empirical evidence on the effect of geographic concentration or size on trade assistance is ambiguous. Depending on the type of political system, the effect of industry size is conditional on electoral spread and location or on party-industry affiliations. Geographic concentration can affect which industries win protection, but I argue it is regional, rather than spatial, concentration that matters. In the case of the steel industry, which is discussed in chapter 4, steel plants are spatially dispersed but regionally clustered in every country case. I show that, in some cases, regional concentration undermined steel firms' political struggle, while in other cases, regional concentration was the key to steel firms' political success. In PR systems, the incentives to target one region over another arise from economic and demographic differences, which structure party affiliation. Political jurisdictions by themselves are of no importance. In contrast, in majoritarian systems, political jurisdictions are everything. Within a specific region, parties compete only in those political jurisdictions where the underlying economics and demographics make the seat party competitive. I argue there is no general relationship between industry geography and political influence across all countries.

Rival Political Explanations

I argue that variance in the level of government assistance to the cutlery industry in the United Kingdom, the United States, and Germany are largely explained by differences in party strength and the electoral rule. There are, however, many other institutional variables that influence political clout and government support that I do not explore in the context of this single case. For example, Germany and the United States have federal bicameral political systems, while the United Kingdom has basically a unitary structure. The federal structure creates an extra layer of veto players--actors whose agreement is necessary to create government policy (Tsebellis 2002). It is conceivable that these veto players hindered the national government's attempts to protect the cutlery industry. U.S. individual states or German Lander might well veto trade and industrial policy that lead to an unfavorable regional redistribution of resources. This implies that regionally concentrated industries rarely win protection. On the other hand, trade and industrial policy is a multidimensional policy issue, so one could also imagine legislative bargains that buy-off coalitions of U.S. states or German Landers.

A related argument is that Germany and the United States did protect their cutlery industry to the same exent as the United Kingdom, but that I overlooked this by focusing solely on national trade and industrial policies. In federal systems, industrial assistance is often distributed at the subfederal level. Cutlery might have received much higher levels of government aid once the policies of German Lander and U.S. states are accounted for.

An alternative theoretical argument is that smaller district size in the United Kingdom made legislators more vulnerable to special interests than in the United States or Germany (Rogowski 1987). PR systems--like Germany's--tend to have geographically larger districts than majoritarian systems like the United Kingdom. These large PR districts, however, are multimember, not single-member, districts. This is an important difference. Party candidates in multimember races typically target niche groups of voters in order to win one of many seats (Cox 1990). I see no reason why these legislators are less vulnerable to special interests. I will return to this argument later in the book, when I discuss how electoral rules affect the level of protection.

I do not compare the role of unions in aiding the cutlery industry to win protection. I argue that the fragmented structure of U.K. unions undermined the industry's attempts to demand protection. It is possible that the centralized wage bargaining in Germany impaired the industry's ability to effectively demand protection. Cutlery workers constitute a minuscule portion of the Metalworkers Union. It is unlikely the union would fight hard for such a tiny industry. On the other hand, it is possible that the Metalworkers Union did pressurize the government for redundancy packages on behalf of cutlery workers. Governments can use many different types of policy instrument to privilege key groups of voters. Although I do discuss the role of unions elsewhere in this book, the theoretical focus is how governments supply protection, not how industries demand protection.

Yet another explanation for cutlery's high level of trade protection in the United Kingdom is the national pride argument. There is a great deal of sympathy among the British public for the plight of the Sheffield cutlery industry and the nostalgia this conjures for industrial dominance long gone. The main express train from London to Sheffield is still called the Master Cutler. In part, this cultural resonance explains why cutlery receives higher protection than, say, shipbuilding. People are less nostalgic about British ships; they don't inherit sets of them from their grannies.56 Protecting this kind of industry has a big electoral payoff; one that goes beyond the marginal districts in the north of England. Hence, one might argue that successive governments have protected the cutlery industry because of its cultural resonance. However, it is possible to find another industry that is as not favored with government assistance but, nonetheless, viewed wih some wistfulness (i.e., the U.K. clock industry). Nor is the national-pride argument particularly useful in a comparative context. One could argue that Germany and the United States do not offer the same levels of protection to their cutlery industries because the industry does not have the same cultural resonance in these countries. This is hard to test. It is certainly true that there is no train from Frankfurt to Solingen called the Master Cutler, yet Solingen's cutlery industry is regarded as a national symbol of German innovation and craftsmanship. These types of cultural variables undoubtedly play a role in the government's decision to supply protection. Nonetheless, in a comparative context, I argue that the variables doing most of the work are the electoral mechanism and party structure.

Throughout this chapter, the role of the cutlery case has been purely illustrative. In chapter 2, I set up the theoretical framework and derive predictions as to which groups of voters are favored with redistributive policies under each of three types of political system: strong-party majoritarian systems (like the United Kingdom), weak-party majoritarian systems (like the United States), and strong-party PR systems (like Germany). I explore how industry geography--size, spread, and location--interacts with features of the electoral system to determine which industries win protection. As the case of cutlery shows, it is important to consider both steps of this process: how industry geography affects legislators' induced preferences and how legislative preferences are aggregated into policy. This second step is as important as the first. In the case of cutlery, the initial link between industry geography and induced preferences is similar in the United States and in Britain (the geography of the industry and the electoral rule are the same); however, in the second stage, differences in how legislative preferences are aggregated into policy strongly disadvantage a small, geographically concentrated industry such as cutlery in one case, and strongly advantages it in the other.

I do not examine weak-party PR systems, of which Brazil would be an obvious case. There are several reasons for this, the primary one of which is the relative indeterminacy of predictions for this case. As Barry Ames has documented, in Brazil's weak-party system, legislators use many different tactics to get reelected. In the formal sense there are multiple equilibriums. This means that individual legislators can come to power based upon very different support groups. While in some cases these might be based on industrial groups, in other cases they are not. Although I do speculate about the Brazilian case in chapter 2, given the relative ambiguity of prediction, I do not intend to explore it in the detail I do for the other cases.


Other scholars who use the electoral rule to explain patterns of protection offer quite different arguments from those presented here. Rogowski and Kayser (2002) argue that consumer groups are more powerful in majoritarian than in PR systems. The key feature of their argument is the greater seat-vote elasticities in majoritarian than in PR systems. This leads to greater electoral penalties in majoritarian than in PR systems (for example, a 5 percent loss in votes translates into a bigger loss of seats in majoritarian than in PR systems). Because alienating voters has higher political risk in majoritarian systems, Rogowski and Kayser predict that consumers are more politically powerful in majoritarian than PR systems. However, they roll the relationship between the electoral rule and trade policy into one step: how parties' vote share is translated into seat share. This ignores how parties' induced preferences are aggregated into policy.

In majoritarian systems, votes are not equally valuable everywhere. In the U.K. system, the most valuable votes for national parties are those in marginal seats. Aiding an industry in a marginal district is where the income effect from trade protection has the biggest impact on the government's reelection chances. This hurts consumers, but the government can afford to lose votes in safe seats that it has either little hope of winning or little risk of losing. In the PR case, Rogowski and Kayser's argument implies that since all votes are equally helpful in obtaining seats in PR systems, the governing parties want to enrich all producer groups to maximize their support. However, Rogowski and Kayser ignore how multimembered district and coalition politics encourages parties to target niche groups of voters (Cox 1990; Myerson 1993). Party-sector affiliations and the low cost of new party entry make targeting all producer groups a risky political strategy. It is useful to construct a hypothetical case. Suppose party A typically represents farmers and party B typically represents steelworkers. If party A is in government (and B is not), then it could no doubt increase its support among its rival's traditional supporters by increasing steel protection. Yet, given steelworkers' predisposition toward party B, A's expected yield of additional votes, for each dollar of protection, is small. Further, the cost of enriching steelworkers is to harm farmers, who must pay higher steel prices. While it might well be true that PR parties are more sensitive to producer interests than consumer interests, farmers are nevertheless consumers as well as producers.57 As such their interests are hurt by party A's attempts to lure the traditional supporters of party B. Multimembered districts allow new party entry. As party A seeks to increase its support, a new party that exclusively represents farmers' interests can make gains at A's expense. In attempting to attract additional supporters, party A risks alienating its core supporters.

So while it is true that in PR systems all votes are equally valuable, it is not true that they are all equally easy to obtain (see also Cox 1990; Myerson 1993). The electoral rule and industry geography affect which industries parties want to protect, while, the electoral rule and the strength of parties also affect which industries legislators are able to protect. Which producer groups actually receive government assistance depends on the dynamics of coalition bargaining in multiparty government.

The existing work on government redistribution of transfers in the field of U.S. politics also tends to focus on one or other of these two steps. Lindbeck and Weibull (1987, 1993) and Dixit and Londregan (1986, 1998) argue that governments purchase votes by distributing money to regions teeming with swing voters. Transfers to swing voters have a bigger effect on the government's reelection chances than do transfers to their own or opposition voters. Cox and McCubbins (1986) assume parties are risk averse and argue that targeting swing voters is a risky strategy. Parties will, first and foremost, target policy to benefit the party loyal. Hence, governments purchase votes by investing in regions where they already have high support. However, both arguments assume parties are strong. I assume that U.S. parties are weak. In the U.S. case, the key policy makers are coalitions of individual legislators. I contend that the Dixit and Londregan swing-voter argument is better suited to a strong-party majoritarian system, like the United Kingdom's. In contrast, the predictions of the Cox and McCubbin model, where parties reward loyal voters, are a better fit for the strong-party PR case. However, Cox and McCubbins's argument is quite different from the one I propose. I do not assume that parties are risk averse in one type of electoral system and risk acceptant in the other. Although in the PR case, parties' induced preferences make them act as if they were risk averse. The ease of new party entry makes it least costly to target groups within your own set of party supporters.

The basic theory is about redistribution, not trade. Restated to refer to all types of government transfers, my thesis is that the government weighs the costs and benefits of using policy to buy votes to win reelection. In PR systems with strong parties, one vote is as good as any other vote. The least costly votes to buy are those from individuals who are ideologically predisposed to vote for you in the first place. Hence, parties are predisposed to gaining votes within their traditional constituencies. In majoritarian systems with strong parties, one vote is not as good as any other vote. Votes in marginal districts are more valuable, since winning the marginal districts is the key to winning the election. It costs more to buy off these voters, but it is worth the expense because their votes are more valuable. In both cases it is extremely costly to lure away voters who have strong opposition-party loyalty. In PR systems it is cost effective to target loyal core voters; in majoritarian systems it is cost effective to target voters in marginal seats. While governments in PR systems appear risk averse and governments in majoritarian systems appear risk acceptant, this behavior is shaped by the structure of the political system.


An industry's political clout does not always remain constant over time. As discussed earlier, cutlery's political influence diminished once the electoral landscape changed and Sheffield's electoral districts became safe Labour seats. Here I explore the dynamic aspects of industry protection and ask what political factors change trade and industrial policy.58

As the Sheffield case suggests, in strong-party majoritarian systems a change in the marginality of districts affects an industry's political influence. Empirically, marginality is reasonably stable in the medium term. Yet voter realignment--such as occurred in Canada in the 1980s--can cause a major shift in marginal districts. While such sudden changes are rare, economic development and shifts in the salient political issues produce gradual shifts in the distribution of marginal seats. In both Australia and the United Kingdom textiles have maintained their privileged positions because their semirural locations have remained party competitive throughout the postwar period. Other industries have seen their fate change. Austin-Rover, a United Kingdom automobile producer, was concentrated around Birmingham in the Midlands region. In the 1980s, its largest plant, Longbridge, employed about fourteen thousand workers.59 Not only was it the biggest plant in the Rover group, it was also the biggest employer in Birmingham. For various reasons, the firm operated at a loss for decades, but successive governments supported Austin-Rover with massive bailouts (Marklew 1995). By 1984, British Leyland, Austin-Rover's owner, had absorbed $3 billion in subsidies. Given the scale of the plant, closure would have devastated not only the local economy but economic conditions throughout the Midlands. In addition to the fourteen thousand Longbridge workers, around fifty thousand workers were employed in subsidiary companies that supplied Longbridge.60

Allowing Longbridge to fail for a long time posed too high a political liability, and successive governments sought to save it. Yet in the late 1980s, after almost a decade of heavy public backing, Thatcher's Conservative government drew the line at future bailouts.61 While the case of economic hardship remained unchanged, the political costs had lessened. Longbridge was on the southern edge of Birmingham, and its surrounding electoral districts were a mix of middle-class semirural housing and council estates. In the 1959, 1964, 1970, 1974, and the 1979 elections these districts held the key to victory.62 In an area so rich in marginals, no government could risk harming it; whether the government was Labour or Conservative, it bailed out Longbridge.63 Yet shifts in marginality undermined Longbridge's privileged position. The marginals of the 1980s were the East Midlands, South Wales, Northwest England, and Outer London. By 1986, only eight of seventy-seven Conservative held marginals had significant motor interests (marginals here are generously defined by majorities of less than 10 percent in the 1983 general election).64 Without its dependence on these districts, the government felt safe to break up and sell off Austin-Rover.65 In a strong-party majoritarian case like this, it is a change in marginality that strongly shifts the pattern of protection.

In a weak-party majoritarian case, a change in the agenda-setting power of the industry's representatives affects the level of protection, as does a change in the industry's coalition partners. For example, the U.S. footwear industry's fate depended upon the agenda-setting power of its representatives. Until the 1990s, footwear was relatively protected from imports.66 A small industry, it was largely located in Maine and Missouri; however, it was part of a coalition with textiles and apparel, which extended its geographic influence. Part of the reason footwear was a valuable coalition partner was that its representatives had political clout. The heart of its power lay in the important committee assignments held by the senators from Missouri and Maine (McGillivray and Schiller 1998; Schiller 1999). Senator Danforth (R-MO) was chair of the Finance Subcommittee on Trade from 1981 to 1986. Senator Mitchell (D-ME) was a member of the Finance committee and then Senate majority leader from 1989 to 1994. These legislators were able to block legislation that damaged the footwear industry. With the departure of these institutional leaders, foot-wear lost its political clout, and it was deserted by its coalition partners (textiles and apparel) since it no longer had anything to offer. Political analysts commented, "Having Mitchell as a majority leader was crucial. Mitchell would block the tiniest footwear provision, it would never get through because of him. They really lost their champion when they lost him."67 Today, Missouri's shoe industry is almost completely gone, its production moved overseas.68

In PR systems a change in the party composition of a coalition government shifts the distribution of protection. Parties have incentives to target niche groups of voters,69 and tend to have ties to particular industries. For instance, in Ireland, Fianna Fail is associated with the construction industry, while the German textile industry is associated with the Christian Democratic Union (CDU). When the CDU left office in 1969, political support for textiles bottomed out (see chapter 6). Similarly, the Swedish paper and pulp industries won substantial subsidies in the late 1970s and early 1980s because the party they supported, a small center party, moved into government.70 Of course, if the same parties continually remain in government, such as the Christian Democrats in Italy or Belgium, then the industries that support them should see their assistance remain largely unchanged.

Different political systems suggest different dynamic patterns in how assistance changes over time. I exploit these differences to derive additional hypotheses and new dependent variables. In particular, I focus on how a change in government affects redistributive policy. These dynamic implications of the theory are explained in greater detail in chapter 2.


Which states are the most protectionist? This is a common question in trade politics. While most efforts have compared autocratic with democratic regimes (see, for example, Mansfield et al. 2000), some attention has been paid to the role of the electoral rule. Although, as I shall argue, some of the emphasis placed on this question is misguided, it is worth exploring the logic behind arguments before considering the empirical evidence.

Magee, Brock, and Young (1989), Rogowski (1989), and Mansfield and Busch (1995) suggest PR systems are less protectionist than majoritarian systems because they are characterized by larger electoral districts.71 Since larger districts are more likely to be industrially heterogeneous, legislators representing such districts must weigh a wider array of interests, which isolates them from parochial interests. The following example captures the gist of the argument. Suppose industry A has 100 workers. In a constituency of 1,000 voters, industry A represents 10 percent of the electorate. In a constituency of 100,000 voters, industry A represents only 0.1 percent of the electorate. Refusing to protect 0.1 percent of her electorate is unlikely to affect a politician's reelection chances. Refusing to protect 10 percent of her electorate is much riskier. In this way, the large districts that characterize PR systems insulate PR governments from industry demands. Along similar lines, Magee, Brock, and Young (1989) argue that in the United States, the president, whose constituency is the whole country, is less protectionist than senators, who in turn are less protectionist than congressmembers. While this argument has much to commend it, it focuses only on the desires of individual legislators to protect industries and says nothing about whether they can influence policy.

Rogowski and Kayser (2001) predict that producer groups are more powerful in PR systems. This implies that PR systems are more protectionist than are majoritarian systems. The model developed here says nothing about whether PR or majoritarian electoral systems are characterized by higher average levels of protection. Because the relationship between the electoral rule and industry geography is a contingent one, the level of protection in strong-party majoritarian systems depends on how many districts are marginal and how industries are dispersed across these districts. In PR systems, parties do not maximize the welfare of a large constituency; rather, they target niche groups of voters. In weak-party majoritarian systems, the level depends on how industries are dispersed across districts and how much institutional clout the representatives from each district have. Industries' geography and the political system interact to determine the distribution of protection. Within this context, the average level of protection is vastly insufficient to describe the range and extent to which different industries receive assistance. We might, for example, find that a few industries get lots of protection in majoritarian countries (a small number of industries in marginal districts with high protection), but that many industries get lower levels of protection in PR countries (from multiple parties bargaining in a coalition government). Within this conception, it is unclear that the average level of protection is a useful statistic with which to compare very different distributions of trade protection. Even if the mean protection level is regarded as the appropriate measure for deciding whether nations are more or less protectionist, such claims cannot be directly attributed to the electoral system since the average level of protection depends, for example, on the number of marginal districts and how industries are distributed across them.

Recent work by Shigeo Hirano (2002) on the impact of electoral reform in Japan (in particular, in moving from multimember to single-member districts) also speaks to this issue. Under the old multimembered district system he finds that legislators targeted and drew their support from a subsection of constituents in the electoral district. Following reform and the introduction of single-member districts, legislators want to target and draw support more broadly from within what are generally smaller electoral districts. The overall effect of the electoral system on protection is ambigouous.

The empirical evidence is mixed. The majoritarian countries in Europe, France, and the United Kingdom were, until recently, regarded as the most protectionist countries in the EU. The smaller countries, such as Italy and the Netherlands, tend to be more free-trade oriented. Table 1.2 contains the averages for the 1980s for a variety of trade measures (Dutt and Devashish 2002). Looking across the table at the different measures of trade assistance, no clear pattern of protection emerges for any one country. True, on average, Canadian tariffs are higher than Belgian tariffs, but quotas are more common and subsidies (as a percentage of the GDP) tend to be higher in Belgium. Overall, Belgium has larger trade flows ({exports + imports}/GDP), but this is not surprising.72 Small countries tend to be heavily dependent on their export trade and have stronger incentives to support free-trade policies. Since many of the PR countries in Europe are small, their free-trade position could have more to do with their size than with their type of electoral system. Nonetheless, this does not stop PR countries from bending the rules and finding other ways to support their domestic industries. While some countries appear free-trade oriented within Europe, they use nontransparent policy instruments to help struggling industries on their home turf. Among the Italian political parties, there is a widespread consensus that free trade is preferential to protectionist policies (Grilli and La Noce 1983). However, as a percentage of the GDP, Italian industrial subsidies are frequently higher than those in Germany, the United Kingdom, the United States, Australia, and New Zealand.73 Federal countries, like Germany or Canada, are accused of only sounding free-trade oriented because they know the sub-federal-level states will step in to help industry. In Germany, Lander (state level) and city governments often step in to aid industry.74

Table 1.2 lists different measures of trade assistance for various countries in the 1980s. New Zealand has the highest average tariffs (Switzerland, the lowest); the highest subsidies (as a percentage of the GDP) are in Sweden (the lowest in the United States);75 quotas are most heavily used in Portugal (least used in Canada); the highest import duty as a percentage of imports are in Australia (lowest in Germany); Belgium makes the most extensive use of NTBs (lowest NTB coverage in Norway); however the highest trade flows are in Belgium (the lowest are in the United States).76 It does appear that the majoritarian countries tend to prefer to use tariffs, while PR countries often prefer to use subsidies.77 Although this is not true across the board; the United Kingdom has some of the lowest tariffs and quotas.78 The fact that there are so many different ways a government can privilege an industry begs the question as to why certain governments tend to favor one form of policy instrument over another. Mansfield and Busch (1993) find that PR countries tend to have more NTBs than majoritarian systems; however, this does not necessarily mean that PR systems are more protectionist. It could mean governments in PR systems prefer to use NTBs rather than other policy instruments, such as tariffs, to assist industries (although tariffs and NTBs tend to be correlated [Rodrik 1995]). At the end of the book, in chapter 6, I discuss the endogenous choice of policy instrument. I argue that the strength of the party system affects whether governments prefer to use geographically blunt policy instruments, like tariffs, to redistribute, or geographically targetable policy instruments, like subsidies. However, leaving aside for the moment the question of why governments choose different policy instruments, table 1.2 reveals that in both PR and majoritarian countries, governments find ways to privilege their favored domestic industries.


As the previous section revealed, lack of policy transparency makes it difficult to measure how governments redistribute public assistance to industries.79 In part, this is because trade and industrial policy is multidimensional. Not only can government provide different levels of assistance to each industry, but government can also use different policies such as tariffs, quotas, subsidies, tax breaks, procurement, or regulation to assist each industry. Comparing the monetary value of subsidies versus tariffs, or procurement policies versus low-interest loans, is extremely tricky. As table 1.2 revealed, looking at only one of these policies can cause serious bias in the results. Today's governments use a complicated mix of different types of assistance. If we look just at subsidies, we risk ignoring other forms of assistance. Although international agreements often restrict or limit the amount of assistance, nations have become adept at bending the rules and finding different ways to privilege industries. For example, governments might resort to bureaucratic harassment. Britain taxes French perfumes at the same high rate as alcohol. The French make Japanese videos travel to the town of Poitiers in central France for customs clearance.80 Research and development support has replaced subsidies as the most popular form of state aid--in large part because it is geographically flexible.81 In Germany, research and development subsidies are a popular way to target benefits to specific regions through the creation of industrial parks.82 Within the EU, it has become particularly hard to measure which countries are pushing for which particular industries because the decision-making process is nontransparent. While we do get slivers of information--for instance Britain pushed for textile protection within the EU, yet it fought the hardest against a sensitive label (excluded from tariffs) for carpets under the Generalized System of Preferences (Cable and Rebelo 1980)--it is extremely difficult to know who is helping whom.

To test the hypotheses derived in chapter 2, what I ideally need is clear, comparable measures of the total assistance granted to each industry. Unfortunately, such measures are hard to come by. Tariffs are probably the easiest protection to measure, but even calculating the effective tariff from the nominal tariff is a complex task.83 The quantitative analysis of NTBs is very difficult. We have data on the presence of NTBs across industries. However, we do not have comparable data on the degree of NTBs across industries.84

Price differentials offer an innovative and promising way to measure industry protection. If we ignore problems of transportation costs and economies of scale, if truly free trade exists, then arbitrage should ensure that prices in every country are the same. However, barriers to trade distort these prices. Rogowski and his colleagues (1999; Rogowski with Kayser 2002) use deviations from average world prices as measures of protection for different products across the world. They find evidence that trade is more open in majoritarian systems. One advantage of this method is that it provides a measure of all the different types of government assistance to industry in comparable units. Of course, its biggest limitation is that it probably reflects other factors that have nothing to do with barriers to trade, such as transportation costs, market size, and demand. Controlling for these other effects is tricky (Rodrigez and Rodrik 2000).85

Our inability to accurately measure levels of assistance severely limits our ability to test our theories. I propose several solutions to overcome these problems. I use two different types of dependent variable to test the theory outlined in chapter 2. The first dependent variable is a direct test of how much income from tariffs goes to industries and electoral districts in Canada and the United States--two majoritarian countries with strong and weak parties, respectively. While I have argued that tariffs make up only one means of assistance, I use tariffs from the 1970s, when they were still the predominant tool for protection. Additionally, I limit the statistical analysis to Canada and the United States, neither of whose tariff policy was subverted to a customs union at the time. By the 1970s, tariff levels within much of Europe were regulated by the European Economic Community or European Free Trade Area.

One advantage of this empirical setup is that I am able to test the interactive effects of industry size, location, and electoral concentration. In Canada I find that, as expected, industries highly electorally concentrated in marginal districts receive the most favorable levels of protection. In the United States, industries located in safe districts receive more favorable levels of protection than industries located in marginal districts; large, electorally dispersed industries are able to secure the most favorable levels of protection. This analysis helps unravel the mixed findings in the literature on the effect of geographic and electoral concentration on the level of industry protection. Busch and Reinhardt (1999) argue that this is largely a measurement issue.86 I argue that the effect of electoral or geographic concentration on political clout depends on the interactive effect of institutional variables.

While the results in chapter 3 support the arguments for majoritarian systems, they do not provide a test of the predictions for strong-party PR systems. Neither do they control for government assistance through other policy instruments.

Much of the extant literature has dealt with the problem of multiple policy instruments by either ignoring the problem or attempting to construct indices of protection. While the latter attempts are potentially rewarding, I prefer another tack. The second dependent variable is an indirect measure of the value of government assistance to an industry--industry stock prices. In chapter 5, I test auxiliary hypotheses about the effect of government change on redistributive policy using the relative price of industry stock prices in PR and majoritarian systems.

Stock market data provide a valuable and, I believe, a much underused measure of government policy.87 Actors in stock markets, particularly large institutional investors, have incentives to, on a case-by-case basis, examine the impact of government policy and bureaucratic regulation on an industry's profitability at a level of detail impossible for a political scientist looking at many industries across many countries to duplicate. What is more, stock markets aggregate each investor's assessment and report the information in comparable units across all industries and across all nations--namely, the value of an industry.

The French government's rebate plan of 1994 provides an excellent example of the stock market's ability to capture the value of protection. In 1994, the French government offered a rebate to consumers willing to trade in their old cars for new, less-polluting models. The stated goal of the rebate was to help the environment. However, because it was tied to the purchase of French automobiles, the rebate also helped French automakers. One way of measuring the value of the government subsidy to Peugeot-Citroen is to estimate how much the rebate affected Peugeot-Citroen sales in 1994. It is estimated that about 7 percent of Peugeot-Citroen's increased sales in 1994 were due to the government subsidy.88 Alternatively, we could look at what happened to the price of Peugeot-Citroen's stock on the French stock market--did it rise with the news of the government rebate? Private investors are highly efficient at incorporating this type of information into their expectations about industry share prices. If the future profits of an industry are expected to change because of an alteration in trade or industrial policies, then the value of equities in that industry will adjust accordingly. Further, a comparison of stocks allows us to compare the impact of very disparate policies. For example, were the rebates to Peugeot-Citroen more or less valuable than the research and development subsidies awarded to the French chemical company Rhodia? Comparing these two policies without an intimate knowledge of both markets is like comparing apples and oranges, yet stock prices provide directly comparable, although admittedly noisy, measures. Since monetary units or percentage changes in market capitalization are easily compared, stock-market data makes it possible to compare the value for Chrysler when the Austrian government pays a third of the cost of a new Chrysler factory, to the value of the grants to Volkswagen from the German state of Saxony, to the value of the French government's rebates on trade-ins for Peugeot-Citroen cars.89

Unfortunately, stock-market-based measures are not perfect, since many factors other than the value of government policy affect stock prices. This volatility makes the measure noisy. Despite this noise, there is considerable evidence that financial markets react as expected to shifts in party government and to changes in domestic industrial policy or trade shocks (Quinn and Jacobson 1989). Using U.S. stock prices, Mahdavi and Bhagwati (1994) find that the announcement of Voluntary Export Restrictions does increase the stock price of affected industries. Herron, Lavin, Cram, and Silver (1999) found that the profits of different industry sectors in the United States reacted as expected to anticipated political change in the 1992 presidential election.90

I exploit the efficiency with which stock markets incorporate information to test the dynamic implications of the model. In addition to predicting which industrial groups are likely to receive protection as a function of the electoral system, the theory also predicts how the winners and losers of the competition for industrial assistance change in response to political change. For example, in PR systems, trade and industrial policy represents a bargained compromise between the parties in government. When the government changes, so will the nature of this bargain, with issues that the outgoing parties cared about being dropped and the issues of the incoming parties being incorporated. With such a change, the winners' stock prices should rise while the losers' fall. In majortarian systems, a change in marginality (with strong parties) or a change in the agenda-setting power of legislators (in weak party systems) produces policy changes.

An event analysis of investors' changing valuation of Peugeot Citroen stock is one way to assess the policy impact of government change in France.91 However, such a direct test is problematic over a wide range of industries because it requires a detailed knowledge of which industries are associated with which parties or legislators. Such information can be readily compiled for a particular industry in a particular country--for example, from newspaper accounts and political contributions (where relevant)--but for a cross-national, cross-industry study coordinating, this data is impractical. This data constraint prevents me, in a large N setting, from assessing who the winners and losers are from a policy change. Yet a policy change produces winners and losers. The winners' stocks should rise and the losers' stocks should fall: a divergence in stock price accompanies policy change. I create a measure of stock-market price dispersion by asking how much price changes for individual industries vary from the average market price change. Since this dispersion measure increases with policy changes, it provides a means to assess policy change, and I regress this measure on political change.

It is critical to recognize that in these tests it is not the average market price, per se, that I care about, but rather the extent to which industries deviate from the market average. Following a change in government there might be a 1 percent increase in the average market price. This average change is not of interest here. Rather, is this 1 percent change the result of all industry groups experiencing a 1 percent gain or are the winners from a change in industrial policy gaining 11 percent, while the losers are losing 9 percent? The former implies no redistributive implications from the change in government, while the latter suggests a change in the set of industries privileged from trade and industrial policy. I assess how the dependent variable "price dispersion," responds to political and economic change in both PR and majoritarian systems across fourteen stock markets, from 1972 to 1996.92 While it is not a direct assessment of the quantity of assistance provided to each industry, the comparison of stock data allows me to test an auxiliary hypothesis derived from the theory.

In between the two empirical chapters, 3 and 5, I provide a test of the causal plausibility of the theory using another type of dependent variable. It is traditional in the trade literature to think about competition between industries. In most cases this is appropriate, since most assistance policies affect all firms in an industry equally. Yet this is not always the case. The theory I advance concerns redistribution, and it asks under what conditions government chooses to privilege one group over another. While I couch the theory in terms of industries, the theory is equally applicable at a subindustry level. In chapter 4, I examine the government's role in plant closures within the steel industry in Australia, Belgium, Germany, Sweden, the United Kingdom, and the United States.

By the early 1980s the steel industry in each of these countries was under immense pressure from cheap imports and a falling price for steel. In each county it was clear that the steel industry needed urgent restructuring, which in all cases meant closing some plants. In most cases the government played an active role in choosing which plants remained open and which were closed: a redistributive choice. Steel manufacturing involves massive plants that typically dominate the employment in the areas where they are situated. The choice of which plants to close has enormous redistributional consequences, since plant closures result in high levels of unemployment and economic stagnation for the local economy. As I shall demonstrate, the decision to keep certain plants open and others close was not a decision based on economic efficiencies. The decisions were primarily political.

For example, in the early 1980s, the U.K. Conservative government implemented a radical restructuring plan for the steel industry. The plan called for the closure of steel plants in both Wales and the Northeast of England, while the Ravenscraig plant in Scotland was kept open. Such a plan was clearly not formulated on an efficiency basis: the Welsh plants were far more efficient than Ravenscraig. Indeed, the British Steel Corporation (the nationalized industry that ran steel) and a parliamentary select committee both opposed keeping Ravenscraig open. Politics holds the key as to why the government imposed the economic misery on communities in the Northeastern England rather than in the far greater loss-making plant in Scotland. While the regions of Wales and Northeastern England strongly support the Labour party, Ra-venscraig to the Southeast of Glasgow near Motherwell included the few remaining seats that the Conservatives held in Scotland. As we shall see in chapter 4, the Ravenscraig case demonstrates how governments in majoritarian systems with strong parties redistribute in favor of marginal districts.

In a PR system, the theory predicts that governments will favor the party faithful. The restructuring of Belgium steel illustrates this well. In December 1981, a Catholic-Liberal coalition took over from the socialist-center government. This new coalition was dominated by northern, Flemish interests. Unlike the previous government, it pushed for reform of the steel industry. Specifically, the new government restructured and closed southern (Wallonian) steel plants (Bain 1992; Capron 1986). Although the government provided public funds to the southern plants, the one-shot subsidies were designed to speed their closure and put a stop to their long-term drain on the public purse. Closing the older Wallonian steel plants and redistributing assistance as a reward to Flemish constituencies in the north was both economically and politically rational for the new government.

The steel case studies emphasize the centrality of redistribution to understanding trade and industrial policy. Given the policy instruments at its disposal and the geographic distribution of industrial groups relative to political jurisdictions, government privileges those groups upon whom they depend for political success. Chapter 4 demonstrates that, given its choice of policy instrument, the government privileges groups to maximize political ends. However, this begs the question of why certain governments tend to favor one form of policy instrument over another. I discuss this endogenous choice of policy instrument in the final chapter of the book, chapter 6.


The purpose of this book is to show how governments use trade and industrial policy for political goals. In particular, governments use these policies to privilege certain groups. As such the pattern of trade protection and industrial assistance depends upon how an industry is distributed and what political system is in place. Who wins and who loses from trade and industrial policy depends upon a two-stage mapping. First, the geographic distribution of an industry relative to political jurisdictions determines which industries legislators wish to help. Second, the electoral system and the strength of parties determine how these induced preferences are aggregated into policy. It is not enough to simply have legislators that want to help your industry; you need legislators who can also convert their preferences into policy.

In the next chapter I provide a detailed exposition of the theory. In particular I analyze how the industry geography required to obtain government assistance varies according to the strength of political parties and the electoral rule.

Return to Book Description

File created: 8/7/2007

Questions and comments to:
Princeton University Press

New Book E-mails
New In Print
PUP Blog
Princeton APPS
Sample Chapters
Princeton Legacy Library
Exam/Desk Copy
Recent Awards
Princeton Shorts
Freshman Reading
PUP Europe
About Us
Contact Us
PUP Home

Bookmark and Share