Social Dumping: Theoretical and Empirical Aspects

(1)

Social Dumping: Theoretical and Empirical Aspects

Dissertation

zur Erlangung des akademischen Grades des Doktors der Wirtschaftswissenschaften

an der Universit¨at Konstanz

Fakult¨at f¨ur Wirtschaftswissenschaften und Statistik

vorgelegt von

Teodora Dimitrova

29.01.2007

Tag der m¨undlichen Pr¨ufung: 24.6.2006 Referent: Prof. Heinrich Ursprung, Ph.D.

Referent: Prof. G¨unther Schulze, Ph.D.

Konstanzer Online-Publikations-System (KOPS) URL: http://www.ub.uni-konstanz.de/kops/volltexte/2007/2287/

URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-22873

2006

(2)

List of Tables

3.3.1 Distribution of the FACB Indices. . . 62 3.3.2 Child Labour and Industrialization Indicators. . . 64 3.3.3 Regressions of the Index of Freedom of Association and Collective

Bargaining. . . 69 3.3.4 Regressions of the Index of Forced Labour. . . 70 3.3.5 Regressions of the Index of Discrimination at the Workplace. . . . 71 3.3.6 Regressions of Child Labour. . . 72 3.4.1 Regressions of the Index of Freedom of Association and Collective

Bargaining: Developed and Developing Countries. . . 75 3.4.2 Regressions of the Index of Forced Labour: Developed and Devel-

oping Countries. . . 76 3.4.3 Regressions of the Index of Discrimination: Developed and Devel-

oping Countries. . . 77 3.4.4 Regressions of Child Labour: Developed and Developing Countries. 78 3.4.5 Regressions of the Index of Freedom of Association and Collective

Bargaining: Industrialized Countries. . . 80 3.6.1 Countries’ Ranking Score by the OECD Index of FACB. . . 84 3.6.2 Countries’ Ranking Score by the Forced Labour Index. . . 85 3.6.3 Countries with CL above 20%: their level of industrialization and

scores on the indices of FACB. . . 86 3.6.4 Correlations: ILO Index of FACB . . . 87 3.6.5 Correlations: Forced Labour Index . . . 88

(4)

3.6.6 Correlations: Index of Discrimination . . . 88

3.6.7 Correlations: Child Labour . . . 89

3.6.8 Variables Definitions and Data Sources. . . 90

4.3.1 Effects of ILO Index of FACB On US Outward FDI . . . 102

4.3.2 Effects of ILO Index of FACB on Aggregate Exports . . . 104

4.3.3 Effects of ILO Index of FACB on Manufacturing Exports . . . 106

4.3.4 Effects of ILO Index of FACB on US Apparel Imports . . . 109

4.5.1 Employment, Capital Expenditures, and Sales by Nonbank U.S. Multinational Companies, 1988-2002 . . . 113

4.5.2 Variables Definitions and Data Sources. . . 114

List of Figures

1.0.1 The Nexus Between Globalization, Labour Market Institutions, and Labour Market Outcomes. . . 10

2.3.1 The new flow system in equilibrium. . . 25

2.3.2 A graphical illustration of equilibrium. . . 28

2.3.3 The utility functions of the politically active groups. . . 29

2.4.1 The types of equilibria in the political parameters space. . . 34

2.5.1 The welfare effects of capital mobility at different values of the minimum wage . . . 39

4.2.1 US Manufacturing Imports and Exports, 1989 to 2001 . . . 97

(5)

1 Introduction

The present work consists of three logical parts, each of them analyzing the nexus labour standards - labour market outcomes - globalization.

The first part, i.e Chapter 2, is a theoretical analysis of the effects of globalization on the equilibrium labour market policy, employment and wages. The policy instrument considered is a minimum wage set by the government and influenced by two lobby groups. Globalization is modelled as an increase in capital mobility.

The theoretical model developed in this Chapter allows analyzing how globalization affects the political feasibility of the labour standard, i.e. whether its level is reduced or increased as the economy becomes more open.

Chapters 3 and 4 are empirical. Chapter 3 looks at the determinants of four different kinds of labour standards. A major set of explanatory variables consists of different measures of openness. In addition, two variables portraying the political process and the politicians’ preferences are employed. The first measure is the Political Rights Index of Freedom House measuring the level of democracy in the country on a 1-7 scale. The second one is a dummy variable indexing whether the country has established special export processing zones. This variable is interpreted as indicating particular commitment of the governing elite to promoting exports. Chapter 4 looks at an important implicit assumption of the Social Dumping argument. Namely, whether multinational enterprizes are really attracted to countries where labour standards are relatively low, or put differently, whether low labour standards really provide competitive advantages in attracting foreign direct investment and stimulating exports.

This work has been inspired by the recent political debate on social dumping.

Labour movements in many developed countries, NGOs and international organizations such as the ILO and WTO accuse the governments of less developed

(6)

countries of practicing social dumping in the sense that they tolerate low labour standards to create a competitive cost advantage vis-a-vis the developed countries in attracting FDI and increasing exports (Sinn 2001: 3). This strategy undercuts the cost-competitiveness of industries in the developed countries, and the trade unions there, under the threat of plant closures/relocation, see themselves forced to lower their demands (Flecker 1998: 73, Stumpf-Fekete 2000: 72, Rodrik 1997:

24). Thus, competition for investment and exports induces a downward spiral on labour standards, the so called ”race to the bottom” or ”social dumping” (among others s. Hyman 1997:527, Flecker 1998:74, Doerre 1996, Brown, Deardorff and Stern 1996, Davis 1998, Chau and Kanbur 2000, Sinn 2001: 3-6). To stop this

”race to the bottom” developed countries attempt to implement an international harmonization of social conditions and advocate retaliatory trade restrictions to enforce implementation through, for example, the introduction of a Social Clause in the WTO (Sinn 2001: 3).

While this description is in no way rigorous, it first reflects the wide-spread public opinion on the problem. Second, it illustrates the complexity of the problem: the social-dumping argument is a chain argument that rests upon several assump- tions; moreover it contains distinct stages of interaction among different agents.

At one stage, for example, we have the (market) interaction between the partic- ipants on the labour market in a given country. At another stage we have the (strategic) interaction between workers (unions), employer organizations and the government, implementing labour market legislation in the developed countries.

At a third stage we possibly have the interaction (again strategic) between firms’

lobbies and the government elite in the host country, assuming the multinational firms have enough bargaining power: the (strong) multinational investors may buy off the political elite in developing countries and make use of the threat to exit in order to obtain beneficial tax and labour policies. Finally, the WTO may set the stage for some agreement at an international level on the level of some

(7)

labour standards. I will come back to these stages again when I discuss the structure of the theoretical part. This discussion of the different stages of interaction is also useful with respect to the inference we can make from empirical evidence:

when we interpret given evidence in favour of or against the social dumping hy- pothesis, it should be with some caution because it may, and it often does, relate only to a given stage.

This general definition of social dumping provides us with some starting point.

Interestingly, no clear-cut and commonly accepted definition of the term exists so far, although the political discussions on the issue have at times, become very heated. For the purposes of the present work, the following points deserve more attention:

1. What is the object of the debate? Which labour standards are meant?

2. Which groups dominate the debate? Does the political cleavage really run between developed and developing countries?

3. What are the transmission channels through which social dumping can take place?

4. The social dumping argument rests upon the implicit assumption that foreign investors, predominantly coming from industrialized countries, pursue cost-effective strategies and seek to relocate their operations in countries with lower labour standards in order to escape from tough home-country institutions. Some authors specify that this point relates particularly to low-skilled labour intensive industries. The validity of this assumption is predominantly an empirical issue, and it will be scrutinized in a separate chapter of this thesis.

5. Which labour standards are on the agenda for international harmonization,

(8)

which countries advocate harmonization, and is an international agreement on labour standards feasible?

Let me begin with the identification of the parties to the social dumping debate.

Although some scholars identify a political cleavage between ”developed” and

”developing” countries or, alternatively, between ”industrialized” and ”newly industrialized” countries, what actually divides the countries is the level of adopted welfare state policies. In other words, it is more correct to say that the debate is between countries with more developed welfare states (which provide for higher labour standards) and countries with less developed welfare states. Thus, on the one side of the debate are the Northern European countries like Germany, Belgium and Denmark, which have highly developed welfare states in practically all dimensions (s. for example, Nickell 1997, Traxler and Woitech 2000). As far as the other developed countries are concerned, there exists some room for vari- ation. The US and UK, for example, have no child or forced labour, have good health and safety conditions at work, but have low employment protection such as layoff restrictions, union coverage, minimum wages, annual and sickness leave, unemployment benefits, etc. Thus, which countries stand on which side of the debating line depends on the particular considered labour standard.

The general political discussion and the attempts to arrive at an international harmonization of labour standards mainly focus on the following four labour standards: prohibition of child labour (the ILO 182 and 138 conventions on the Worst Forms of Child Labour and on Minimum Age, respectively), prohibition of forced labour (ILO 29 Forced Labour Convention and the 105 Convention on the Abolition of Forced Labour), non-discrimination (ILO 111 Discrimination (Em- ployment and Occupation) Convention and the 100 Equal Remuneration Conven- tion), and the right to organize and collectively bargain (ILO 87 Convention On Freedom of Association and Protection of the Right to Organize and the 98 Con-

(9)

vention on the Right to Organize and Collective Bargaining). These four labour standards are designated as core labour standards. The main contestants are the US, the UK and the other Continental European countries on the one side, and the so-called Newly Industrialized counties such as China, Thailand, Hong Kong, Singapore, The Republic of Korea, etc. on the other. The latter countries have reached high levels of industrialization (and manage to competitively produce and export to the developed countries) but have more lax effective legislation on the above listed labour standards than their trading partners. It is also on these core labour standards that I will focus in the subsequent empirical chapters.

To understand the concept, it is useful to look at the transmission channels through which social dumping might take place. This also sheds some light on the theoretical model that will be presented in Chapter 2. As Mosley (1990: 160) points out, social dumping may occur ”in at least three inter-related ways:

• through the displacement of high-cost producers by low-cost producers from countries, in which wages, social benefits and the direct and indirect costs entailed by protective labour legislation are markedly lower;

• firms in high labour cost countries would be increasingly free to relocate their operations, thereby strengthening their bargaining power, vis-`a-vis their current workforce (or national authorities), to exert downward pressure on current wages and working conditions;

• individual states might be tempted to pursue a low-wage and perhaps even anti-union secondary labour market strategy as part of their efforts to catch up economically.”

The first point emphasizes international trade (or the integration of goods markets) as a transmission mechanism: the firms are assumed to remain in their home country but their products compete now with imports from countries with

(10)

more lax labour regulation. The second point stresses the increased mobility of capital (or the integration of financial markets) as the transmission mechanism:

firms directly relocate some of their operations in countries with more lax labour regulation. The third point focuses on politicians’ preferences and the role of the government. Thus, the social dumping problem has arisen because world product and financial markets have become much more closely integrated in the past 20 years, while labour has remained predominantly immobile and its protection (through a set of labour market institutions) is still a national-domain policy.

This gives the motivation for the theoretical analysis.

To capture these relationships, the theoretical part of the thesis depicts two stylized economies, which differ only in their labour market institutions (one has a binding minimum wage and the other has a flexible wage, i.e. it is deregulated) and studies the effects of increased capital mobility on the labour market outcomes (LMO) such as wages and employment, and on the political feasibility of the labour market institution (LMI) in these two economies. The model portrays the second and third channel according to Mosley (1990) through which social dumping might take place. The interdependencies are illustrated in the following Figure 1.0.1.

(11)

Figure 1.0.1: The Nexus Between Globalization, Labour Market Institutions, and Labour Market Outcomes.

The first two sub-sections, 2.2 and 2.3, of the theoretical part, ”The Base Model”

and ”The Effects of Regulation”, view labour market outcomes as a result of the interaction of market forces and institutional factors: these sections analyze how wages and employment are determined in each of the economies, with and without (exogenous) regulation. This interaction is illustrated in the bottom part of the figure (s. the bottom oval in Figure 1.0.1). The focus is on the interaction between the agents operating on the labour market, i.e. workers and firms. As such, in terms of the above discussion on the complexity of the social dumping argument, it represents the first stage of market interaction.

The following sub-section 2.4, ”The Political Equilibrium”, views labour market institutions as determined by the interaction of market forces and the political

(12)

process: it analyzes the determination of the policy instrument as an outcome of a political process in which the government takes into account both the lobby groups’ interests and the economic welfare effects of the standard. The second- stage interactions are illustrated by the middle oval in the figure 1.0.1.

The last sub-section, 2.5, of the theoretical chapter, ”The Effects of Openness”, first analyzes how opening to international capital mobility between a laisser-faire and a regulated economy affects the endogenously determined level of the labour standard. Second, it investigates the effects of non-perfect capital mobility from the perspective of a small open economy. In these scenarios the government maximizes political support by taking into account the globalization-induced changes in the stake-holder interests (s. the upper oval in the figure for illustration).

The theoretical analysis in this thesis thus contributes in two important respects to the existing literature on social dumping. It, first, explicitly takes into account the economic cost as well as the benefit of the introduced labour market regulation. The bulk of the existing literature relies on perfectly functioning labour markets, which automatically implies that the introduction of any regulation comes at a net cost for the economy, which can only be justified for redistrib- utional reasons. Of course, some studies do focus on imperfect labour markets (for example, Agell 1999); these studies are, however, interested in other issues and consider mostly closed economies, which by definition does not allow any conclusions about social dumping effects. The theoretical part of this thesis is to my knowledge the first that considers this point in the context of the social dumping debate.

Second, as already mentioned, the analysis encompasses the endogenous determination of the level of the labour market regulation. Most existing studies in the field treat the labour standard as exogenous and study how international

(13)

integration affects wages and employment with and without regulation. The influential paper of Davis (1998), for example, shows that increased trade between a country with a binding minimum wage and a country with flexible wages leads to increased unemployment in the regulated country. One could therefore specu- late that the minimum wage is a burdensome instrument of regulation and that it will, as a consequence, be reduced or even abolished in the course of globalization. However, such a conclusion cannot be drawn because of the exogeneity of the regulation. Endogenizing of regulation allows us to draw some conclusions with respect as to how labour market institutions will be adapted as globalization progresses.

An important result of the theoretical analysis is that globalization acts as a political equalizer: the policy adjustment goes against the interests of the stronger lobby. The intuition for this result is that in an integrated economy the minimum wage is less effective as a tool for redistribution and the government is thus less willing to deviate from the social optimum in favour of the stronger lobby. The analysis provides thus no support for the argument that globalization will necessarily erode labour market regulation: the regulated country may or may not be driven to move towards the flexible-wage model. Moreover, complete deregulation will never prevail in both countries, if laissez-faire is economically inefficient.

In Chapter 3 I identify the determinants of labour market regulation empirically.

Specifically, I investigate whether the level of exports (aggregated at different levels) and FDI have an effect on the level of the four core labour standards. In doing so I also account for the political-interest effect regarding globalization, which, as shown in the theoretical part may influence labour market regulation. This is admittedly an effect that is difficult to capture empirically. I try to capture it by including a dummy for the existence of export-processing zones as a regres- sor. My reasoning is as follows. These zones are targeted at foreign investors

(14)

and important exporters. The labour market, tax and environment conditions in these zones are in almost all cases exempted from the legislation covering the rest of the economy. As the literature on the export-processing zones shows (s.

for example, Kusago and Tzannatos, 1998 and ILO, 1998) the regulation in these zones is in many countries very lax. In most of the cases regulation explicitly forbids any form of workers’ association and joint action, there is no regulation of working hours, etc. Thus, it may be argued that this variable captures government aspirations with respect to exports and investment and its readiness to compromise working conditions for this objective. The empirical results point to the fact that the direct effect of globalization (i.e. the effect of FDI and trade) is positive on the level of labour standards; political aspirations for more exports and investment, however, have an adverse effect, particularly, for the right of freedom of association and collective bargaining.

In the second empirical part, Chapter 4, I investigate one of the underlying as- sumptions of the social dumping argument: are multinational investors really attracted to countries with low labour standards? In particular, I focus on the right of freedom of association and collective bargaining as measured by the ILO index. The empirical results show that low respect for freedom of association does not appear to play a critical role in the location decisions of the average in- vestor. Low labour standards do however provide certain industries a competitive advantage, which they make use of. An example is the textile industry, which is intensive in low-skilled labour; in this industry low cost of labour seems to be a major motive for investment. Thus, although the basic assumption underlying the social dumping argument is not generally confirmed by the empirical evidence, it may play a role for some industries that make intensive use of low-skilled labour.

The empirical analysis presented in Chapters 3 and 4 contributes to the social dumping debate first because I use new and more robust measures, which focus

(15)

on the de facto respect for labour standards. Measuring labour standards has up to now been the bottleneck of the empirical research in the field. The new measures that I use allow me to include a larger number of countries in the sam- ple. In particular, my dataset of 170 countries not only includes the developed economies but also a considerable number of developing economies. Finally, the highly disaggregated trade data allow drawing more accurate conclusions about the relevance of the social dumping argument than has hitherto been possible.

(16)

2 Globalization and Labour Markets Deregula- tion

1

2.1 Introduction

One of the most important aspects of globalization is that firms become increasingly mobile: advances in IT, easier access to information about foreign environments, reduced administrative barriers and transport costs allow them nowadays to relocate production activities across borders more easily than ever.

This increased mobility of firms is partly reflected in the growth of foreign direct investment. World direct investment flows increased from approximately 3,5 bio.

US dollars in the first half of the 70s to 233,4 bio. US dollars two decades later (Lipsey, 1999, p.327). The share of FDI in total capital flows increased from 5,8 to 26,2 percent during the same period, op. cit. p. 316.

While the growth of FDI indisputably induces important economic benefits such as higher returns to capital, technology and management spillovers, etc., some observers see in it a potential danger. In particular, concerns have been raised that footloose firms will target investment towards locations where labour standards are lowest and where labour practices are least management restrictive.

In a response to intensified competition individual states might even be tempted to pursue a policy of lowering labour standards, e.g. lowering minimum wages, relaxing employment protection legislation (such as restrictions on layoffs), re- stricting union penetration and centralization, etc. This strategy is labelled ”social dumping”. Extreme proponents to this view go even further and argue that social regulations will spiral downward in a so-called ”race to the bottom” to the

1This Chapter is based upon the joint paper Dimitrova and Tchipev [2004]. Large parts are reproduced without further citation. For this reason I will stick in this Chapter to thewe-form.

(17)

end that labour markets will become completely deregulated worldwide in the long run. While this argument is put forward primarily in the political arena and is taken with skepticism by most academic economists (Freeman, 1998, p.

7), the question whether increased economic integration among countries with different labour market regimes will put pressure for adjustment of national policies, remains interesting and relevant. Will countries retain their labour market institutions, or will there be some adjustment as they become more integrated through trade and/or FDI? Will there be a tendency for institutional conver- gence? Which labour market regime, if any, will prove to be the most successful in the global economy?

We address these issues in the context of the Europe-America dichotomy. The reason for this focus is that Europe and America are main trading and investment partners² with significantly different labour market regimes³. The existing theoretical literature on the Europe-America dichotomy is substantial (notable contributions are Wood 1994, Krugman 1994 and 1995, Bertola and Ichino 1995, Davis 1998, Acemoglu and Newman, 2002). This literature seeks to explain the stylized fact that ”rigid” Europe has witnessed higher unemployment and a more compressed wage structure than ”flexible” America by taking into account that the two economies operate in the same international environment and are thus

2The EU share of the US inward direct investment stock was 60 percent in 1989 and 65 percent in 2000. The EU share of the US outward investment stock was 43 percent in 1989 and 46 percent in 2000 (OECD, 2001, p.409). Moreover, EU (resp.EEC) has been a net exporter of direct investment to the US both cumulative over the period 1982-2000 and in each single year of this period except in the period 1991-1995 (OECD 2001, p.408-415 and OECD 1993, p.

242-248).

3Nickell 1997: 60-63, Tables 4 and 5 and Traxler and Woitech 2000: 144, Table 1, for example, summarize the institutional properties of the OECD labour market regimes like union density and union coverage, bargaining centralization, union participation in public policy, ex- tension agreements, employee participation rights on company boards, payroll tax rate, etc.

While the European countries cannot be easily lumped together, the estimates show that continental European countries taken as a unit dispose of much greater welfare states than the US and UK do. Thus, in the literature in the field Europe as a unit has often been referred to as

”rigid”, and the US as ”flexible” with regard to the operation of the labour market.

(18)

exposed to the same external shocks (technical progress, trade, etc.). In the above studies, which are perhaps the most influential in the field for the moment, the divergent labour market experiences have been attributed to the different labour market institutions, which are assumed to be exogenous. We follow the tradition of this literature to model Europe and America as two identical economies, which are exposed to a common shock: in this case, an increase in capital mobility. We do, however, endogenize the pertaining labour market regulations. This modelling strategy allows studying not only how the effects of a given policy change as the economy opens, but also how openness affects the policy itself through the political process. To portray the political process we rely on the Grossman-Helpman approach (Grossman and Helpman, 1994) that builds upon the common agency model by Bernheim and Whinston (1986). As it is well known this modelling approach implies a government that sets economic policy as if it maximized a weighted sum of social welfare and contributions from organized interest groups.

We focus on the conflict between workers and capital owners⁴. In this framework these two interests lobby the government on minimum wage legislation.

Furthermore, the analysis considers an imperfect labour market. This framework thus accommodates the argument by Agell (1999) that potential efficiency benefits of labour market regulation may be important in the context of globalization.⁵

4This focus is motivated by the open economy perspective that is adopted here. In the context of globalization union leaders often voice concerns about increased bargaining power of firms due to their increased mobility and call for stronger cooperation among workers (s.

comment of Rodrik 1997: 24). In this chapter we consider homogeneous labour and do not analyze any conflict among workers. For an analysis of the conflict between workers with different employment opportunities in a closed economy see Saint-Paul (2000).

5In an imperfectly competitive framework labour market institutions that promote wage compression may be beneficial in various ways: e.g., through facilitating the expansion of high- productivity sectors, creating incentives for human capital accumulation and providing social insurance when private one is undersupplied (s. Agell 1999 for a review of these arguments).

Further, wage compression may reduce wasteful use of resources in monitoring (Acemoglu and Newman 2002) and may stimulate technological development that increases the productivity in low-paid occupations (Acemoglu 2003).

(19)

Adopting a perfectly competitive labour market would imply that, other things being equal, more regulated countries should have a lower per capita labour income, which would be at odds with the fact that living standards within the developed world have been converging while the institutional variety among these countries remains substantial (Freeman 1998, Acemoglu and Newman 2002). To model the labour market, we build on the dual labour market model of Bulow and Summers (1985) by introducing capital in the secondary sector. In this model identical workers are accidentally assigned to either high-paying (primary) or to low-paying (secondary) jobs in equilibrium. The inefficiency of the labour market thus stems from the inability of primary firms to observe the effort of their workers, which results in a larger than optimal secondary sector.

The dual model itself does not say anything about the nature of the goods produced in the two sectors. The usual interpretation is that primary jobs are offered by large manufacturing establishments such as IBM while fast-food outlets, such as McDonalds, are typical suppliers of secondary jobs (Bulow and Summers 1986, p. 380, Saint-Paul 1996, p. 3). For the purposes of the present issue, however, the nature of the goods produced by the two types of jobs is not important. As recent theoretical work has shown good and bad jobs may coexist within the same industry (s. evidence presented in Bhaskar, Manning and To, 2002) or even within the same firm (Saint-Paul, 1996). The key insight in Saint-Paul’s model is that the firm has two basic methods at its disposal to motivate its workers:

offer wages higher than the reservation wage or invest in more intensive monitoring. Under some circumstances the firm may find it optimal to combine the two methods and split its workforce into two. Namely, part of the workers are prevented from shirking by means of higher wages (these are the primary workers who are envisaged as those enjoying greater employment security), and part by means of more intensive monitoring (these are the secondary workers, which are used by the firm as a buffer against fluctuations in product demand). Note that

(20)

from this perspective the skill content of primary and secondary jobs need not be different.

Saint-Paul’s model of internal dualism offers an important insight into the forces at work behind the wage inequality among similar workers. Nevertheless, we model dualism as an intersectoral phenomenon. This amounts to assuming exogenous supply of primary and secondary jobs and is done mainly for simplicity.

A model of internal dualism where firms optimally decide how to split their workforce into primary and secondary jobs could generate similar interdependencies.

In any case, one would expect labour market institutions to reduce the number of secondary jobs and increase that of primary, perhaps at the cost of some unemployment. Moreover, as firms become increasingly mobile one would expect to see some shift of secondary jobs occurring from regulated to deregulated countries.

These are basically the interdependencies that arise in the framework.

In the model the minimum wage reduces profits, benefits workers in the aggregate (provided it is set sufficiently high), reduces wage inequality and increases unemployment. The model, moreover, gives rise to a double-peaked social welfare function: one local social-welfare maximum occurring at no intervention, and the other at some binding level of the minimum wage. To the extent that social welfare matters in the political process (in the Grossman-Helpman set-up it is assumed to enter the government’s objective function) this translates into institutional variety: some countries may end up at the no-regulation peak, others at the high-regulation peak. This argument provides some intuition for the first result: similar countries may have substantially different labour market regulations, and these differences may arise because of small differences in the political process.

The main results however relate to the influence of economic integration via cap-

(21)

ital mobility. First, we analyze the effects of international capital mobility on a laissez-faire and on a regulated economy, which are otherwise identical, keeping the policy in each country fixed. This is the kind of comparative-statics analysis performed in the related literature. We find that the protective effects of regulation (in terms of higher wages for the low-wage workers) spread over to the deregulated country, while unemployment remains restricted to the regulated one and increases after the opening. In this respect the results parallel those of Davis (1998)⁶.

Second, when we allow for an endogenous policy response, we find that the increase in capital mobility may lead to either more or less strict regulation in the regulated country. The direction of the policy response turns out to be related to the relative political influence of the economic interests involved. In particular, we find that deregulation occurs whenever the labour interest is more influential in the political process. Furthermore, if the regulated country is assumed to be a small open economy, the policy response always goes against the interests of the stronger lobby, i.e. the minimum wage is reduced, or even abolished, whenever the workers’ political influence is stronger, and it is increased whenever the capital owners dominate in the political process. In other words, economic integration turns out to work as a political equalizer in the model. The intuition for this result is that the minimum wage is less effective as a tool for redistribution in an integrated economy and the government is then less willing to deviate from social welfare maximization in order to accommodate the stronger political interest.

In sum, the results from the theoretical analysis do not lend support to the argument that globalization will necessarily erode the established labour market regulations: the regulated country may or may not adopt the flexible-wage par-

6It should be noted that we focus on inequality among similar workers, while Davis (1998) addresses inequality between skilled and unskilled workers.

(22)

adigm; but there has been identified no scenario where the deregulated regime will be adopted, should this be less efficient (in terms of the utilitarian welfare function). In other words, worldwide deregulation does not appear to represent an equilibrium outcome, if regulation can be justified on efficiency grounds.

The rest of this chapter is structured as follows. Section 2.2 presents the basic model. Section 2.3 considers the effects of the exogenous policy. Section 2.4 looks at the endogenous determination of the policy, and Section 2.5 derives the effects of the opening. Section 2.6 concludes this chapter.

2.2 The Baseline Model

Consider an economy with two types of infinitely lived individuals, workers and capital owners. The only source of income for the workers is their wages, and for the capital owners it is firms’ profits. There are two types of competitive firms, primary and secondary firms. Each secondary firm uses one unit of specific capital and labour. Its profits are equal to the marginal value product of capital by the homogeneity of degree one of the production function Y(L, K). Primary firms produce exclusively with labour and make no profit. Throughout the chapter we abstract from product market interactions and keep the relative price of the goods produced by primary and secondary firms fixed. A composite numeraire good is defined. The output of both primary and secondary firms is measured in terms of the numeraire. Labour input is measured in efficiency units. A worker contributes one unit of effective labour, if he exerts efforte, e >0. There are only two levels of effort, e and 0. If a worker shirks⁷, i.e. exerts zero effort, he contributes nothing to production. A major difference between secondary and primary firms is that the former can perfectly monitor the effort of their workers while the latter

7The model described in this Section draws heavily on Shapiro and Stiglitz (1984). It is set in continuous time and analyzes only steady state equilibria. Similar two-sector extensions of the Shapiro-Stiglitz model are developed in Bulow and Summers (1986) and Jones (1987).

(23)

cannot. Primary firms are conscious of the effort elicitation problem and take workers’ preferences into account in their employment decisions⁸. There are N identical and risk neutral workers with instantaneous utility separable in income and effort and normalized to

U = w−e (2.2.1)

wherewis the wage andeis the effort level. Workers maximize expected lifetime utility

V = E

∞

Z

0

Utexp(−rt)dt (2.2.2)

The effort decision of a worker employed in a primary firm is based upon a comparison between his lifetime utility if shirking,V^{P S}, and that if not shirking, V^{P N}. In the steady state we have

rV^{P N} = wp−e+sn(Vâlt−V^{P N}) (2.2.3) rV^{P S} = wp+ss(Vâlt−V^{P S}) (2.2.4) where the instantaneous interest rate r(r > 0), the separation rate for non- shirkers sn(sn > 0) and for shirkers ss(ss > sn) are exogenous parameters; Vâlt is the lifetime utility of a worker fired from the primary sector, and wp is the primary wage. Workers choose not to shirk wheneverV^{P N} ≥V^{P S}. Substituting forV^{P N} and V^{P S} from Eq. 2.2.3 and 2.2.4 into this no-shirking condition yields

wp ≥ (r+ss)e ss−sn

+rV^alt (2.2.5)

Noting that there will be no shirking in equilibrium, the corresponding asset equation forV^alt can be written as

rV^alt = ws−e+a(V^{P N}−V^alt) (2.2.6)

8Competitive firms producing with labour only do pay efficiency wages in Bulow and Sum- mers (1986) as well.

(24)

wherews is the secondary sector wage, and a is the instantaneous probability of a worker fired from the primary sector to find a job in that sector again. The first term in Eq. 2.2.6, ws −e, is the instantaneous utility of a worker outside the primary sector. Since in the absence of intervention secondary jobs are freely accessible, the latter is equal to the instantaneous utility of having a secondary job. If ws = e, the instantaneous utility of having a secondary job equals the instantaneous utility of being unemployed⁹, thus, without intervention, the model is consistent with voluntary unemployment. The flow probabilityacan be found from the steady state condition

a(N −Lp) = snLp (2.2.7)

where Lp denotes employment in the primary firms. Normalizing the marginal product of labour in the primary sector to one and using the zero-profit condition for the primary firms, the equilibrium in the case of no intervention can be described by the following three equations inws, Lp and Ls (employment in the secondary sector):

1 = ws+c0+c1

· N N −Lp

¸

(2.2.8)

ws = Y_L⁰(Ls,K)¯ (2.2.9)

N = Lp+Ls (2.2.10)

Let us denote the solution of the above system (w⁰_s, L⁰_s, L⁰_p). Eq. 2.2.8 is obtained by first solving the system consisting of Eq. 2.2.3, 2.2.6 and 2.2.7 fora,V^{P N} and Vâlt, and then substituting the solution for Vâlt into the no-shirking condition 2.2.5, whereby c0 and c1 denote the constants _sêr

s−sn and _s^esⁿ

s−sn, respectively. Eq.

2.2.9 says that the wage in the secondary sector equals the marginal product of labour. ¯K denotes the number of secondary firms (which equals also the aggregate capital endowment since each firm is endowed with one unit of capital).

9The latter is zero: unemployment benefits are not considered.

(25)

Eq. 2.2.10 says that the sum of primary and secondary employment equals total labour force¹⁰. We assume that w_s⁰ > e, i.e. in the case of no intervention there is full employment and equilibrium is given by equations 2.2.8 - 2.2.10.

2.3 The Effects of Regulation

In this Section we describe the effects of the government’s policy. We assume that the government can directly set the wage in the secondary sector. The choice set W is given by the closed interval [w_s⁰,1], where w_s⁰ is the secondary-sector wage obtained in the absence of intervention and 1 is the wage in the primary sector.

If the government chooses w = w⁰_s, i.e. the policy of no intervention, equilibrium is given by Eqs. 2.2.8 - 2.2.10. Alternatively, the government may choose w ∈ (w_s⁰,1], i.e. a binding minimum wage. The imposition of a binding minimum wage entails that workers fired from the primary sector cannot immediately obtain a job in the secondary sector. In equilibrium this leads to involuntary unemployment: primary jobs are rationed due to the no-shirking condition, secondary jobs are rationed due to the binding minimum wage. This implies that in each instant a worker is in one of three possible states, each associated with different lifetime utility: employed in a primary firm, employed in a secondary firm, and unemployed. Therefore, a new flow system characterizes the equilibrium (see Figure 2.3.1 below).

The arrows in Figure 2.3.1 indicate the flows between the three states, in steady state, the letters above indicate the instantaneous probability of a worker in a

10If the secondary wage that solves the above system is below e, equilibrium is found by solving a system consisting of 2.2.8, 2.2.9 and the following two equations

ws=e Lp+M+Ls=N whereM is unemployment (here voluntary).

(26)

PSfrag replacements

Lp M Ls

sn

ap

as

bap

Figure 2.3.1: The new flow system in equilibrium.

given state to get in the respective flow. Note that workers fired from the primary sector enter directly the unemployment pool, there is no arrow from Lp to Ls. There is no arrow from Ls to M either, which indicates that secondary sector workers cannot be falsely accused of shirking and fired. The unemployed acquire primary jobs at rate ap and secondary jobs at rate as. The parameter b reflects how successful an unemployed worker is on average in finding a primary job relative to a secondary worker. Thus, the equilibrium conditions are as follows:

rVâlt = ap(V^{P N} −Vâlt) +as(V^sec−Vâlt) (2.3.1) rV^sec = ws−e+bap(V^{P N} −V^sec) (2.3.2)

snLp = apM +bapLs (2.3.3)

snLp = (ap+as)M (2.3.4)

N = Lp+M +Ls (2.3.5)

Eqs. 2.3.1 and 2.3.2 define the lifetime utility of the unemployed (Vâlt) and that of the employed in the secondary sector (V^sec). The next two equations say that the flows into and out of the primary sector (Eq. 2.3.3) as well as the secondary sector (Eq. 2.3.4) must be equal. Eq. 2.3.5 says that the sum of employment in both sectors and unemployment (M) equals total labour force. Solving Eq. 2.2.3 and Eqs. 2.3.1 - 2.3.5 simultaneously for M, ap, as, V^{P N}, V^sec, Vâlt, and substituting the solution forVâlt into 2.2.5 yields a no-shirking condition in wp, ws, Ls and b.

Sincewp has been normalized to 1, ws =w, as set by the government, and Ls is given by the requirement that marginal value product of labour equals the wage

(27)

(Eq. 2.2.9), it remains only to pin down the parameterbto close the model. One possibility would be to assume that b= 1, i.e. the typical secondary worker and unemployed have equal chance of finding a primary job at any point of time. It may be argued however that the involuntarily unemployed have an incentive to search harder since the utility gain from obtaining a primary job for them is larger.

This would implyb <1. Moreover, the government may provide assistance in the job search to the unemployed to compensate them for the adverse effects of the minimum wage. In any case, we will treatb here as exogenous and will consider the extreme case of b= 0. This significantly simplifies the equilibrium dynamics as the flows into and out of the secondary sector drop out. Setting b = 0 yields the following no-shirking condition

wp ≥ e+c0+c1

µ N −Ls

N −Ls−Lp

¶

(2.3.6) The model in the binding minimum wage case reduces then to the following two equations in Lp and Ls.

Lp = α(N −Ls) (2.3.7)

w = Y_L⁰(Ls,K)¯ (2.3.8)

Eq. 2.3.7 is obtained by solving 2.3.6 for Lp and setting wp = 1. It implies that employment in the primary sector is proportional to the labour force outside the secondary sector with a constant of proportionalitya≡ ¹⁻₁^e₋⁻_e^c₋⁰_c⁻^c¹

0 . It follows that unemployment¹¹ is also proportional with M = (1−α)(N −Ls). Employment in the secondary sector is directly obtained from the condition that marginal product equals the minimum wage, Eq. 2.3.8.

We are now in a position to derive the welfare effects of the policy instrumentwon the politically active groups. To focus on the social conflict between capital and labour we consider just two lobby groups that comprise the whole population: the

11The condition that unemployment is involuntary isV^alt <

∞

R

0

(w−e) exp(−rt)dt and it is assumed that it is satisfied in equilibrium.

(28)

one lobby group is that of the capital owners and the other that of the workers¹². We assume that the two lobbies have overcome the collective action problem¹³ and each maximizes the aggregate utility of its members. Let us denote the aggregate lifetime utility of the capital owners byVK and that of workers by VU. VK andVU are obtained by double integration, once over time for each individual, and once over individuals. Given that the number of workers in each of the three possible states does not change over time in equilibrium we can change the order of integration and write Vi =Ui

∞

R

0

exp(−rt)dt, i=K, U where Ui, i=K, U, stands for the aggregate instantaneous utility found by summing over individuals. Since the aggregate lifetime utilities are proportional to UK and UU we can use these latter measures in the subsequent analysis without loss of generality. They are defined as follows:

UK ≡ Y(Ls,K)¯ −wLs (2.3.9)

UU ≡ (1−e)Lp+ (w−e)Ls (2.3.10) The following graph should be useful in the derivation of the functions UK(w) and UU(w) (see Figure 2.3.2 below).

The two lines starting at points I and A represent the marginal product of labour in the primary and the secondary sector, respectively. The curve passing through point D is the right-hand side of 2.2.8, the no-shirking condition in the case of no intervention. The position of the horizontal line JC measures the effort levele. In the case of no intervention employment in the primary sector equals the distance ID and employment in the secondary sector - the distance EB. The aggregate

12A standard result in the Grossman-Helpman model is that the equilibrium policy departs from the social-welfare maximizing one due to the existence of an unrepresented part of the population. Even though in the present framework all individuals are organized, the same result is obtained because of the unequal power of the two lobby groups.

13A source of additional tension within the workers’ group may be their ex-post heterogeneity.

Nevertheless, the fact that they are identical ex-ante and the uncertainty about who will end up unemployed make it reasonable to assume that workers act as a homogeneous group.

(29)

PSfrag replacements

I

J

−→^Lp

G

H D

E

F

A

B

C 1

e w_s⁰

←−^Ls

Figure 2.3.2: A graphical illustration of equilibrium.

welfare of the capital owners is found as the area of the triangle ABE and that of the workers as the area of the rectangles IDFJ and EBCF, which represent the aggregate utility of the workers employed in the primary and the secondary sector, respectively. The introduction of a just binding minimum wage only marginally affects point E (the intersection of the marginal product in the secondary sector and the policy level), while shifting the no-shirking condition to point G¹⁴. After the introduction of a binding minimum wage, therefore, aggregate welfare of the workers employed in the primary sector is given by the area IGHJ. Further increasing the minimum wage implies that the point E moves along the marginal product line towards point A whereby the employment in the secondary sector is progressively diminishing and the no-shirking condition is progressively relaxed (shifted to the right). We can now proceed with the algebraic derivation of the functionsUK(w) and UU(w). The results are illustrated in Figure 2.3.3.

Capital owners’ utilityUK(w) is continuous and decreasing at a diminishing rate

14This can be seen from (16), where the right-hand side goes to infinity when Lp goes to N−Ls. Since Ls equals the distance EB it follows that with a binding minimum wage the no-shirking condition is asymptotic to the vertical line going through point E.

(30)

PSfrag replacements

wint

UK

1 UU

Ω

limUu(w)

w→w0 s w>w0 s

UU(w⁰_s) lim Ω(w)

w→w0 s w>w0 s

ΩU(w⁰_s)

w⁰_s w

Figure 2.3.3: The utility functions of the politically active groups.

over the whole range [ws,1]. Differentiating 2.3.9 and using 2.3.8 and 2.2.9 yields

U_K⁰ (w) =−Ls <0 (2.3.11)

where, for any wage,Ls is obtained from Eq. 2.3.8. Workers’ aggregate utility is discontinuous at w = w⁰_s. This is so because the introduction of a just binding minimum wage tightens the no-shirking condition relative to its initial position as illustrated in Figure 2.3.2. The utility loss from a just binding minimum wage for the workers as a whole is

UU(w⁰_s)− lim

w→w0 s w>w0 s

UU(w) = (1−e)(1−α)(N −L⁰_s)>0 (2.3.12)

where (1−e) is the loss in instantaneous utility of a primary worker who becomes unemployed, and (1 − α)(N − L⁰_s) is the size of the affected workforce. For w∈(w_s⁰,1] the functionUU(w) is continuous and differentiable. Using 2.3.7 and 2.3.10 we find

UU(w) = (1−e)αN + (w−wint)Ls, for w > w_s⁰ (2.3.13)

(31)

wherewint denotes the constantα+ (1−α)e <1. Differentiating this expression yields

U_U⁰ (w) = (w−wint)dL

dw +Ls, (2.3.14)

where, for any wage, Ls and ^dL_dw = _Y⁰⁰ ¹

LL(Ls,K)¯ are found from Eq. 2.3.8. We assume that w_s⁰ < wint, which guarantees that the function UU has a well defined maximum over the interval (w⁰_s,1]¹⁵. Furthermore, it is assumed that this maximum is higher than UU(w⁰_s), i.e. in the aggregate, workers can gain from the imposition of a minimum wage. This implies a conflict of interest between capital owners and workers with respect to the policy. The workers’ preferred policy is some binding minimum wage, while capital owners’ preferred policy is no intervention. It is also of interest to look at the utilitarian social welfare function Ω(w)≡UK(w) +UU(w). This definition together with 2.3.11 and 2.3.14 imply:

Ω⁰ =U_K⁰ +U_U⁰ = (w−wint)dLs

dw ≷0 as w≶wint (2.3.15) It is clear that labour market regulation in the model entails both efficiency costs and benefits, in terms of Ω, leaving the net effect ambiguous. When we start at the laissez-faire equilibrium, ”marginal” costs dominate ”marginal” benefits.

This is captured in an extreme way here by the jump in Ω, which is equal to the jump in UU, when a just binding minimum wage is introduced. Further, as it is clear from 2.3.15, Ω is increasing over the range (w⁰_s, wint), i.e. over this range it is the marginal benefits that become dominant, and to the right of wint Ω is decreasing again, i.e. it is the marginal costs that dominate. A similar two-peak relationship between efficiency and regulation may be obtained in more general settings as well. Here the costs of regulation are associated with a surge in involuntary unemployment. The benefits stem from shifting workers

15Since _dw^dL = _Y00 ¹

LL(Ls,K)¯ <0 we haveU_U⁰ (w)>0 for all w∈(w_s⁰;wint), i.e. UU(w) reaches its maximum somewhere afterwint.

(32)

from low-productivity to high-productivity sectors¹⁶. The interpretation would be different, if one views the presence of primary and secondary jobs as intra- rather than as an inter-sectoral phenomenon. On the benefit side, the shift from secondary to primary jobs induced by the minimum wage would potentially relocate resources from monitoring to productive activities. On the cost side, it would reduce the flexibility of the firms and entail that production opportunities in periods of economic boom be not fully realized. The presence of both costs and benefits to labour market institutions provides a natural explanation of the institutional variety among the leading industrialized counties. Accordingly, in the model two different institutional settings, the laissez-faire and the binding minimum wage one, may generate the same aggregate welfare¹⁷. This follows from the fact that either of the two peaks Ω⁰ ≡Ω(w_s⁰) or Ω^int≡Ω(wint) may be higher in the framework depending on the constellation of parameters.

2.4 The Political Equilibrium

In this Section the policy is endogenized. We assume that it is determined in a political process which takes the form of a two stage game (Grossman Helpman 1994). At the first stage the two lobbies offer the government contribution schedules λi(w), w ∈W, i=K, U; at the second stage the government chooses w∈W so as to maximize the weighted sum of contributions and net social welfare:

Γ(w)≡γKλK(w) +γUλU(w) + [Ω(w)−λK(w)−λU(w)] (2.4.1) Net social welfare (the term in brackets) receives a weight of 1, contributions - weights ofγK and γU respectively, where we assume γK ≥1, γU ≥1. The para-

16The relevance of such benefits has been discussed in Bulow and Summers (1986) in the context of industrial policy, and in Agell (1999) in the context of labour market policy.

17The same idea has been expressed by Acemoglu and Newman (2002). However, they do not build a political economy model nor do they analyze the impact of globalization on institutional change.

(33)

metersγK andγU can be regarded as a measure of the strength of each lobby: the higher itsγ, the more effective the lobby is in influencing the policy. For example, in the extreme caseγK =γU = 1 both lobbies are powerless and the government is a social welfare maximizer. Another extreme case would be when one of the lobbies has infinite power (γi → ∞). In this case this lobby could induce the government to set any policy ˆwby simply offering a contribution schedule that pays an arbitrarily small amount ² > 0 at ˆw and zero anywhere. There are different reasons why the two lobbies may not be equally effective in influencing the government. Rama and Tabellini (1998), for instance, argue that the nature of the contributions (labour offers support demonstrations, while capital offers cash) as well as the cost to organize into an active lobby may differ between capital and labour (Rama and Tabellini 1998, p. 1306).

At the first stage of the game each lobby offers a contribution schedule. Following Grossman and Helpman (1994) and Rama and Tabellini (1998) we focus on the Nash-equilibrium supported by the so called ”truthful” contribution schedules.

These take the form

λi(w, Ri)≡max [Ui(w)−Ri,0], i=K, U (2.4.2) i.e. they pay the government, for any w, the excess welfare of lobby i at w relative to a given reservation utility. The problem of choosing an optimal contribution schedule for lobby i reduces then to the problem of choosing an optimal reservation utility Ri. The equilibrium is a combination of two reservation utilities (Rêq_K, R_Uêq) and a corresponding policy wêq = argmax

w∈W

Ω(w) + (γK − 1)λK(w, R^eq_K) + (γU−1)λU(w, R^eq_U) such that no lobby can improve its net welfare by readjusting its reservation utility, given the reservation utility of the other lobby. Net welfare is defined as gross welfare minus contributions

U_i^net(w)≡Ui(w)−λi(w) (2.4.3)

(34)

The two definitions 2.4.2 and 2.4.3 imply:

U_i^net(w) = min [Ui(w), Ri] (2.4.4) This highlights the basic trade-off in choosing an optimal reservation utility. On one hand,Ri shouldn’t be chosen too low because it is an upper bound on what the lobby can get. On the other hand, it shouldn’t be too high because then the contribution is small and the lobby cannot effectively influence the policy. The following Lemma 1 says that it is not optimal for a lobby to ask for more than what it can get.

Lemma 1 In equilibrium each lobby gets its reservation utility:

R_iêq=U_i^net(wêq), or equivalently Rêq_i ≤Ui(wêq).

(Proof in the Appendix)

We now proceed to characterize how the equilibrium policy and reservation utilities depend upon the political parametersγK and γU. The objective is to divide the parameter space (γK, γU) into a set where the equilibrium policy is laissez- faire (let us denote this set Λ⁰ ), and a set where the equilibrium policy is a binding minimum wage (let us denote this set Λ^m).

Lemma 2 The equilibrium policy satisfies

w^eq = argmax

w∈W

γKUK(w) +γUUU(w). (Proof in the Appendix)

An immediate implication of Lemma 2 is that the equilibrium policy is homogeneous of degree zero in γK and γU. This leads to the main

Proposition There exists a critical ratio γ^∗, such that for γK/γU > γ^∗ the equilibrium policy is laissez-faire and forγK/γU < γ^∗ the equilibrium policy is a binding minimum wage. (Proof in the Appendix)

This Proposition shows what the sets Λ⁰ and Λ^m look like. Among all points (γK, γU) that satisfy the general requirement γK ≥ 1, γU ≥ 1, those that lie to

(35)

the left and above the straight lineγK =γ^∗γU belong to Λ⁰, and those that lie to the right and below this line belong to Λ^m. The precise slope of the borderline γK =γ^∗γU depends upon the economic parameters of the model as illustrated in Figure 2.4.1 below.

PSfrag replacementsγK γK γK

γU

γU (1,1) (1,1) (1,1)

Λ⁰ Λ⁰

Λ⁰

Λ^m Λ^m

Λ^m

case 1: Ω⁰ >Ωînt case 2: Ω⁰ = Ωînt case 3: Ω⁰ <Ωînt Figure 2.4.1: The types of equilibria in the political parameters space.

For example, if Ω⁰ > Ω^int, i.e. social welfare is maximized by no intervention, then the point (1,1) must lie to the left and above the lineγK =γ^KγU (remember that for γK = γU = 1, w^eq = argmax

w∈W

Ω(w)). Hence in this case γ^∗ < 1. If, on the other hand, Ω⁰ <Ωînt, i.e. social welfare is maximized by a binding minimum wage (wînt), then the point (1,1) must lie to the right and below the line with slopeγ^∗ and henceγ^∗ >1. Finally, if Ω⁰ = Ωînt, then the point (1,1) lies exactly on the borderline, γ^∗ = 1. It is clear that when Ω⁰ = Ωînt and γK =γU = 1 the equilibrium policy is not unique: it may be w_s⁰ or wint. In fact, it can be shown that, whatever the slope of the borderlineγ^∗, when (γK, γU) lies exactly on that line, the equilibrium policy is not unique: it can be laissez faire or a binding minimum wage. The sets Λ⁰ and Λ^m have therefore an intersection, which is given by the lineγK =γ^∗γU.

Let’s proceed now to the equilibrium policy to the right and below this line. By the Proposition it is a binding minimum wage. By Lemma 2 it must satisfy

Social Dumping: Theoretical and Empirical Aspects