Union wage-setting and international trade with footloose capital

    Union Wage-Setting and International Trade with Footloose Capital Hartmut Egger, Daniel Etzel PII: DOI: Reference:

S0166-0462(14)00043-X doi: 10.1016/j.regsciurbeco.2014.04.008 REGEC 3053

To appear in:

Regional Science and Urban Economics

Received date: Revised date: Accepted date:

14 December 2012 12 March 2014 27 April 2014

Please cite this article as: Egger, Hartmut, Etzel, Daniel, Union Wage-Setting and International Trade with Footloose Capital, Regional Science and Urban Economics (2014), doi: 10.1016/j.regsciurbeco.2014.04.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

T

ACCEPTED MANUSCRIPT

RI P

Union Wage-Setting and International Trade with Footloose Capital∗ Hartmut Egger† University of Bayreuth CESifo, GEP, and IfW

NU

SC

Daniel Etzel‡ University of Bayreuth

MA

March 12, 2014

Abstract

CE

PT

ED

This paper sets up a general oligopolistic equilibrium model with two countries that differ in the centralization of union wage-setting. Being interested in the consequences of openness, we show that, in the short run, trade increases welfare and employment in both locations, and it raises income of capital owners as well as workers. In the long run, capital outflows from the country with the more centralized wage-setting generate winners and losers and make the two countries more dissimilar in terms of unemployment or welfare. Decentralization of wage-setting can successfully prevent capital outflow and the export of jobs.

AC

JEL codes: F12, F16, J51, L13 Keywords: General oligopolistic equilibrium, Union wage-setting, Asymmetric labor market institutions, Trade liberalization, Footloose capital, Decentralization in union wagesetting

∗ We would like to thank two reviewers for their valuable comments and suggestions. We are also grateful to Frode Meland, Doug Nelson and participants at the 12th G¨ ottingen Workshop on International Economics, the European Trade Study Group Annual Meeting in Copenhagen as well as seminar participants at the University of Bayreuth, the University of Bergen and the LMU Munich for helpful discussion. Part of the project was conducted while Daniel Etzel visited the University of Nottingham and the University of Bergen. The hospitality of both institutions and stimulating discussions are gratefully acknowledged. † University of Bayreuth, Department of Law and Economics, Universit¨ atsstr. 30, 95447 Bayreuth, Germany. E-mail: [email protected]. Phone: +49.921.55-6080. Fax: +49.921.55-6082. ‡ atsstr. 30, 95447 Bayreuth, Germany. University of Bayreuth, Department of Law and Economics, Universit¨ E-mail: [email protected]. Phone: +49.921.55-6083. Fax: +49.921.55-6082.

1

ACCEPTED MANUSCRIPT

1

Introduction

T

It is widely documented that wage-setting institutions differ significantly even between otherwise rather similar countries,1 and it is well-known from the macro-labor literature that such

RI P

differences have severe consequences for the labor market outcome and the well-being of people in the respective countries. Despite the attention, wage-setting institutions have received in the literature, relatively little is still known about the consequences of different wage-setting

SC

institutions in open economies. In particular, the role of labor market institutions at different stages of globalization have not played a prominent role of academic research so far. Filling the gap and providing a more comprehensive picture about how openness affects the macroeconomic

NU

consequences of differing wage-setting institutions is the purpose of this paper. To conduct our analysis, we set up a general oligopolistic equilibrium (GOLE) model along the lines of Neary (2009), with a unit mass of sectors and a small (endogenous) number of

MA

firms in each industry. Enriching this framework with a simple textbook model of monopoly unions gives a general equilibrium version of a unionized oligopoly model with pure economic

ED

rents and involuntary unemployment in equilibrium (see, for instance, Egger and Etzel, 2012). The crucial innovation of our paper is the introduction of capital as a second factor of production into the workhorse GOLE model. To keep the analysis tractable, we assume that capital is employed as fixed input, while labor serves as a variable input in the production process

PT

(cf. Martin and Rogers, 1995). We embed this extended GOLE framework into a two-country model, in which the two economies are symmetric in all respects, except for their wage-setting

CE

institutions. We associate differences in labor market institutions with different degrees of centralization in the wage-setting process and assume that one country is populated by firm-level unions, while the other one is populated by sector-level unions.2

AC

We first study the two economies under autarky and then contrast it to the open economy equilibrium. Thereby, we distinguish two possible scenarios of openness in order to distinguish different stages of globalization. In the first one, we assume that product markets are fully integrated, while capital remains invested at home. This captures the idea that capital owners do not immediately adjust their investment decisions after a globalization shock, and hence we refer to this scenario as the short run. In the short run, trade raises competition in the product market and lowers the ability of unions to set excessive wages for ‘insiders’. This generates an employment and welfare stimulus in both economies and thus raises the magnitude of economic 1 For instance, there are pronounced differences between OECD member countries in the percentage of workers covered by the collective bargaining of unions – ranging from a low level of 15% or less in Korea, Japan and the US to a high level of 90% or higher in Austria, Belgium and France – and the degree of centralization in the wage-setting process – ranging from the country-level in Finland and Norway to the industry-level in Austria, Belgium and Germany and the firm-level in Canada, Japan and the US (see OECD, 2004, Tables 3.3 and 3.5). 2 Focussing on the degree of centralization in union wage-setting as a key aspect of labor market institutions has become common in the literature, since Calmfors and Driffill (1988) have published their hypothesis of a humpshaped relationship between the degree of centralization in collective wage-setting and aggregate unemployment. We do not discuss economy-wide agreements as a third alternative, because agreements at the highest level of centralization have become an exception in the 21st century (see OECD, 2004).

2

ACCEPTED MANUSCRIPT

rents that can be distributed between capital owners and workers, with both groups benefiting from the product market integration. Furthermore, product market integration lowers the wage

T

gap between the two economies, arising from differences in the prevailing wage-setting institu-

RI P

tions, and this reduces the differential in unemployment and welfare ceteris paribus. However, this effect is counteracted by fiercer competition in the global market, which magnifies the employment and welfare differences associated with a given wage gap, because production shifts towards the country that offers the lower production costs in the open economy. As long as

SC

the unemployed receive a compensation which is less than 2/3 of the going wage rate, it is the first effect that dominates, and in this case the result from our analysis is in line previous

NU

findings that differences in labor market institutions are less important in open economies (see Danthine and Hunt, 1994).

In the long run, capital is footloose and searches for the most profitable investment op-

MA

portunities in the global economy.3 This generates capital flows from the country that hosts sector-level unions to the country that hosts firm-level unions, and these flows continue until the return to capital investment, i.e. the profit of the firm, is equalized between the two locations.4 The outflow of capital lowers employment and welfare in the country that hosts the sector-level

ED

union and raises them in the country of capital inflow. Regarding group-specific effects, we find that the additional investment opportunities increase the real income of capital owners in the country that hosts sector-level unions, while workers lose in this economy due to an export of

PT

jobs. Things are different in the country that hosts firm-level unions. Due to capital inflow, firms headquartered in this country lose their competitive advantage vis-´ a-vis foreign producers,

CE

and hence the capital owners running these firms are worse off than in the short run. Workers in this economy are better off than in the short run, because the establishment of new local firms implies additional domestic jobs.

AC

In a final step of our analysis, we investigate to what extent the trend among OECD countries to move towards less centralized forms of union wage-setting (see OECD, 2004) can be successful in securing domestic jobs and thus guarantee gains from trade for domestic workers also in the long run. The message from our analysis is clear: Decentralization can be successful if it occurs early, because in this case it can prevent the capital outflow. If decentralization is a response to the export of jobs, its implications are less promising, because it may be ineffective in stopping an already existing capital outflow and reversing the foreign investment decision of firms. Furthermore, in an open economy decentralization not only affects the domestic labor market but also generates spillovers on foreign workers due to labor market linkages arising 3 There is a difference between capital mobility and the actual movement of capital owners. In line with the footloose capital model, we abstract from mobility of capital owners and assume that the return to foreign investment is repatriated. 4 There is ample evidence that weaker labor market institutions promote the inflow of foreign investment (see, for instance Javorcik and Spatareanu, 2005; Gross and Ryan, 2008). Furthermore, using information on the expansion strategies of French firms in OECD countries, Delbecque, M´ejean, and Patureau (2008, 2014) report that weaker unions in general and those with a lower degree of centralization in the wage-setting in particular lead to higher inflow of foreign investment.

3

ACCEPTED MANUSCRIPT

from integrated product and capital markets. In our model, a movement from sector-level to firm-level wage-setting eliminates the competitive advantage of foreign producers with negative

T

consequences for employment in the short run. Furthermore, if decentralization is successful in

RI P

preventing capital outflow and job exports, it additionally generates long-run losses in terms of foreign employment.

Our paper is of course not the first one that studies the role of unions in open economies. The consequences of product market integration for union wage-setting have been prominently

SC

discussed by Huizinga (1993), Sørensen (1993) and Naylor (1999), whereas Danthine and Hunt (1994) have pointed to the role of centralization in the wage-setting of unions.5 These contribu-

NU

tions are related to the short-run perspective in our model, which deals with the consequences of goods trade. A different strand of the literature has shed light on the interaction between union wage-setting and multinational activity (see, for instance, Mezzetti and Dinopoulos, 1991;

MA

Lommerud, Meland, and Sørgard, 2003; Eckel and Egger, 2009). Leahy and Montagna (2000) emphasize the role of centralization in the wage-setting of unions for the investment decision of multinationals. This literature is closely related to the long-run perspective in our model. Finally, the idea of footloose capital relates our analysis to a relatively small liter-

ED

ature studying union wage-setting in the context of economic geography (see Munch, 2003; Picard and Toulemonde, 2006; Persyn, 2013). All of these studies shed light on important aspects regarding union wage-setting in open economies, but none of them looks systematically

PT

on the role of centralization in the wage-setting process at different stages of globalization.6 The remainder of the paper is organized as follows. Section 2 introduces the theoretical

CE

framework and studies the differences between firm-level and sector-level wage-setting in a closed economy. In Section 3, we consider two open economies that are fully symmetric in all respects, except for the prevailing degree of centralization in union wage-setting, and study how the

AC

movement from the closed to an open economy affects aggregate employment and welfare as well as the real income of capital owners and workers in the short and the long run. In Section 4, we investigate whether decentralization in the country that hosts sector-level unions can be successful in preventing capital outflow and the export of domestic jobs. The last section concludes with a brief summary of the most important results.

2

The closed economy

We start our formal analysis with a detailed model description and a characterization of the autarky equilibrium. 5 Bastos and Kreickemeier (2009) and Kreickemeier and Meland (2013) have embedded the model of unionized oligopoly into a general equilibrium environment. 6 A more extensive literature review can be found in the working paper version of this manuscript.

4

ACCEPTED MANUSCRIPT

2.1

Model description

We consider an economy that is populated by L workers, each of them supplying one unit of

T

labor, and K capital owners, each of them supplying one unit of capital. Capital is required as

RI P

a fixed input for starting up and operating firms, while labor is used as a variable input in the production process. Product markets are modeled along the lines of Neary (2009), who provides a workhorse for studying oligopolistic competition in a general equilibrium environment. Capital

SC

owners are entrepreneurs and thus receive operating profits as a return on their capital input. There is no imperfection in the capital market and free entry of firms. On the contrary, there is

NU

imperfection in the labor market due to union wage-setting. Preferences and consumer demand

We assume that preferences are described by an additively separable utility function over a

MA

continuum of different goods z that are defined on the unit interval, with the sub-utility for each of these goods being quadratic. The utility function of consumer c is given by

ED

Uc [{xc (z)}] =

Z

1

0

1 axc (z) − bxc (z)2 dz, 2

(1)

PT

R1 and his/her budget constraint equals 0 p(z)xc (z)dz ≤ Ic , where p(z) denotes the price of good z, and Ic is income of consumer c. Provided that the budget constraint is binding, the solution to the consumer’s utility maximization problem, gives his/her inverse demand function for good z:

CE

p(z) = [a − bxc (z)]/λc , where λc is the consumer’s marginal utility of income, which is a function R1 R1 of the first and second (uncentered) moments of prices, µ ≡ 0 p(z)dz and σ ≡ 0 p(z)2 dz, respectively, as well as income, Ic : λc ≡ (aµ − bIc )/σ. This gives

AC

To determine economy-wide consumer demand, X(z), we aggregate xc over all consumers.7

where A ≡ (K + L)a, λ ≡

p(z) = P

c λc

1 [A − bX(z)], λ

= (Aµ − bI) /σ, and I ≡

(2) P

c Ic .

Eq. (2) can be interpreted

as the indirect demand of a representative consumer, whose preferences can be used as welfare criterion. Technology, production, and competition In each sector, an endogenous number of firms, n(z), produces a homogeneous sector-specific output. Firm number, n(z), is finite and firms therefore take into account their impact on price p(z), when setting quantities in Cournot competition. However, in view of a continuum of industries, firms rationally ignore their impact on economy-wide variables, such as λ or I. 7 Throughout our analysis, we focus on parameter configurations that ensure participation and non-satiation of all consumers.

5

ACCEPTED MANUSCRIPT

Regarding production, we assume that firms in all industries employ the same technology. They invest one unit of capital as a fixed input and must hire one unit of labor for each unit of output

RI P

T

they want to produce. Denoting output of firm j in industry z by yj (z), considering product Pn(z) market clearing, i.e. k=1 yk (z) = X(z), and accounting for demand function (2), we can write

firm-level profits as follows:



n(z)

Πj (z) ≡ λπj (z) = A − b

k=1



yk (z) − λwj (z) yj (z).

SC

X

(3)

As explained in Neary (2009), λπj (z) can be interpreted as real profits at the margin, and

NU

changes in this variable do not exert direct welfare implications. However, such changes are still instructive as they indicate adjustments of the competitive environment in the product market.

MA

Throughout our analysis we focus on the case of a positive supply of all firms and, therefore, restrict our attention to parameter configurations that lead to A > λwj (z) for all j and z. Labor market institutions

ED

We assume that wages are unilaterally set by unions before firms set their employment level, produce and sell their products to consumers. Unions maximize an objective function V = (w − w)ℓ, ¯ where ℓ is the number of employed union members, which, in the case of a closed shop,

PT

equals the employment level of all firms in which the respective union is active, w is the union wage, and w ¯ ≡ βw ˜ is unemployment compensation, which is a constant fraction β ∈ (0, 1) of

CE

a country’s economy-wide average wage, w. ˜ In the background, there is a proportional tax on both sources of labor income, wages and unemployment benefits, which provides the revenues for financing unemployment compensation. This income tax has the attractive feature of being

AC

a lump-sum instrument, which allows for redistributing resources towards those who do not have a job, without affecting the maximization problems of capital owners, firms, and unions in our model. Unions can be organized at the sector of the firm level.8 In the former case (index s), unions Pn(z) represent the workforce of the whole industry, i.e. ℓ = k=1 lk (z), and set a uniform sector-wide wage rate. In the latter case (index f ), unions represent the workforce of a single firm and set

a specific wage rate for this group of workers. Taking stock, we can express the objectives of sector-level and firm-level unions in the following way: n(z)

 X V s (z) = ws (z) − w ¯ lk (z), k=1

  Vjf (z) = wj (z) − w ¯ lj (z).

(4)

8 Since unions are small in the aggregate, they take economy-wide variables as given and hence the outcome of their maximization problem would be unaffected if nominal wages and unemployment benefits were divided by a common deflator, such as the consumer price index or λ−1 .

6

ACCEPTED MANUSCRIPT

2.2

The autarky equilibrium

The equilibrium outcome is characterized by the solution to a three-stage game in which capital

T

owners decide on firm entry at stage one, unions enter and set wages at stage two, while firms

RI P

choose employment and compete in quantities at stage three. We solve this three-stage game through backward induction.

SC

Output competition at Stage 3: Under Cournot competition, firms set their output to maximize profits (3) subject to yj (z) ≥ 0. The (interior) solution to this maximization problem is given by the first-order condition, which

P

k6=j

yk (z) − λwj (z) 2b

.

(5)

MA

yj (z) =

A−b

NU

can be reformulated to

Sector-level unions set a uniform industry-wide wage and since the profit-maximization problem is the same for all firms in this industry, we have wk (z) = w(z) and thus yk (z) = y(z) for all k = 1, ..., n(z). Firm-level unions set a wage that is only binding for workers of a specific firm.

ED

However, since firms have perfect foresight, producer j rationally anticipates symmetry of all competitors, implying wk (z) = w−j (z) and thus yk (z) = y−j (z) for all k 6= j.9 We can thus

A − λw(z) , b(n(z) + 1)

yjf (z) =

A + (n(z) − 1)λw−j (z) − n(z)λwj (z) , b(n(z) + 1)

(6)

CE

y s (z) =

PT

reformulate Eq. (5) for the case of sector-level and firm-level unions, respectively:

where λw(z) and λwj (z) can be interpreted as real wages at the margin.

AC

Wage-setting at Stage 2:

Substituting lj (z) = yj (z) in union objectives (4), accounting for (6), and maximizing the resulting expressions for w(z) and wj (z), respectively, gives the first-order conditions dV s (z) A − 2λw(z) + λw ¯ = = 0, dw(z) b(n(z) + 1)

dVjf (z) dwj (z)

=

A + (n(z) − 1)λw−j (z) − 2n(z)λwj + n(z)λw¯ = 0. b(n(z) + 1)

Due to symmetry of all firms and unions in industry z, we can now set wj (z) = w−j (z) = w(z). Solving for wages, therefore gives λws (z) =

A + λw ¯ , 2

λwf (z) =

A + n(z)λw ¯ n(z) + 1

in the case of sector-level and firm-level unions, respectively. 9

We use subscript −j for referring to all firms differing from j.

7

(7)

ACCEPTED MANUSCRIPT

Capital allocation and firm entry at Stage 1: Capital owners make the investment decision to maximize their profit income. Substituting wage

T

rates (7) into output functions (6) and noting further that Πj = byj2 , we can calculate firm-level

1 Π (z) = b



A − λw ¯ 2b (n(z) + 1)

2

1 Π (z) = b f

,



n(z) (A − λw) ¯ 2 (n(z) + 1)

2

.

(8)

SC

s

RI P

profits

Firm indices have been neglected because all firms in an industry are symmetric. Differentiating (8) with respect to n(z), we see that real profit income at the margin shrinks in the number of

NU

competitors. Hence, income maximization of capital owners requires an equal number of firms in all industries and thus an allocation of K according to the no arbitrage condition Π(z) = Π for all z. With a unit mass of industries, we therefore get n = K and, since in equilibrium

MA

industries are symmetric in all respects, we can omit sector indices from now on. Furthermore, in view of the ex-post symmetry of sectors, we can set w ¯ = βw. Equipped with this insight, we can now solve for wages in the symmetric autarky equilibrium: A , 2−β

ED

W s ≡ λws =

W f ≡ λwf =

A . 1 + n(1 − β)

(9)

PT

It is easily confirmed that n > 1 implies W s > W f , so that our model reproduces the textbook result that sector-level unions set higher wages than firm-level unions (see Calmfors and Driffill, 1988).

CE

Substituting (9) into (6) and accounting for the symmetry of industries, we can calculate firmlevel output and employment under the two labor market regimes: y s = ls = W s (1−β)/[b(n+1)], y f = lf = W f n(1 − β)/[b(n + 1)]. Higher wage claims of sector-level unions lead to higher

AC

production costs and lower firm-level output and employment than in the case of firm-level unions. With firms and industries being symmetric in equilibrium, economy-wide employment R1 equals nl = 0 n(z)l(z)dz. Denoting the unemployment rate by u and focussing on parameter

configurations for which not all workers find a job in equilibrium, total employment under the two labor market regimes is given by (1 − us )L =

nA(1 − β) , b(n + 1)(2 − β)

(1 − uf )L =

n2 A(1 − β) . b(n + 1)[1 + n(1 − β)]

(10)

In an interior equilibrium with involuntary unemployment, i.e. u > 0, labor supply is a nonbinding constraint and thus aggregate employment independent of labor endowment L (see Brecher, 1974, for a similar result). Furthermore, sector-level unions generate a stronger labor market imperfection leading to higher unemployment than in the case of firm-level unions i.e. us > uf . With prices being the same in all industries, indirect utility of the representative consumer is a monotonic transformation of the total real income of the economy (see Neary,

8

ACCEPTED MANUSCRIPT

2009). Denoting by λI total real income at the margin and by P = λp the consumer price index, total real income is given by λI/P . Noting further that a binding budget constraint requires

T

that aggregate revenues, P ny = P (1 − u)L, equal aggregate income, λI, we can safely conclude

RI P

that λI/P = (1 − u)L. Hence, us > uf implies that real economy-wide income is lower with sector-level than with firm-level unions: (λI/P )s < (λI/P )f .

In a final step, we analyze how different degrees of centralization in union wage-setting affect Φ ≡ (1 − u)LW , by the consumer price index gives Φ P

s

nA(1 − β) = , b [n + 2 − β] (2 − β)



Φ P

f

=

n2 A(1 − β) . b [(n + 1) + n(1 − β)] [1 + n(1 − β)]

(11)

NU



SC

the two income groups, capital owners and workers. Dividing economy-wide labor income,

Similarly, we can make use of Π = by 2 to calculate economy-wide profit income Ψ = nΠ:

Ψ P

s

1−β = n+1

 s Φ , P

 f   Ψ n(1 − β) Φ f = . P n+1 P

(12)

ED



MA

Ψs = Φs (1 − β)/(n + 1), Ψf = Φf n(1 − β)/(n + 1). Dividing by P , we obtain

Both workers and capital owners are better off with wage-setting at the firm than at the sector

PT

level.10

3

The open economy

CE

Let us now consider trade between two countries, i = 1, 2, that are identical in all respects, except for their labor market institutions, and whose economies are of the type analyzed in Section 2. Differences in labor market institutions are captured by differences in the degree of

AC

centralization in union wage-setting, and we assume that country 1 hosts sector-level unions, whereas country 2 hosts firm-level unions. We abstract from international shipment costs and consider fully integrated product markets. Labor is internationally immobile and we distinguish two scenarios with respect to capital mobility. In the first one, we assume that the capital investment decision is given and thus firm allocation the same as in the closed economy. We refer to this scenario as the short run because it captures the idea that de-investment of capital takes time. In the long run, capital is footloose and invested where it generates the highest return, which may be at home or abroad.11 If capital is invested abroad, profit income is 10

If unemployment compensation is not too generous, those who have a job (as well as those who are unemployed) in both scenarios are better off with sector-level than with firm-level wage-setting, whereas the total negative effect of sector-level wage-setting on labor income is a result of higher unemployment. 11 Using the endogeneity of capital investment as criterion to distinguish the long run from the short run is common in the literature (see, for instance, Blanchard and Giavazzi, 2003). In line with this literature, we interpret the opening up of a country as a single policy intervention, which has short-run effects for a given capital investment and long-run effects when investment is endogenous (rather than treating product and capital market integration as two distinct phenomena). Of course, one could further enrich the picture of possible scenarios by accounting for a ‘medium run’, in which investment can be adjusted within the country but capital

9

ACCEPTED MANUSCRIPT

repatriated to the home country. To characterize the open economy equilibrium, we can follow the analysis in Section 2 step demand function

RI P

 1 xt (z) , pt (z) = ¯ 2A − b¯ λ

T

by step. We first sum up consumer demand in the two economies. This gives the global indirect (13)

¯ ≡ λ1 + λ2 denotes where superscript t is introduced for referring to trade variables and, λ clearing condition, firm j’s profits are given by 

nt (z)

k=1



¯ j (z) yj (z), yk (z) − λw

NU

¯ t (z) = 2A − b Πtj (z) ≡ λπ j

X

SC

the world representative consumer’s marginal utility of income. Applying the product market

(14)

MA

where nt (z) ≡ n(z) + n∗ (z) is the total number of domestic and foreign firms (with an asterisk referring to foreign country variables). Solving the firm’s profit maximization problem, we can calculate j’s optimal output

¯ ∗ (z) + (n(z) − 1) λw(z) ¯ ¯ j (z) 2A + n∗ (z)λw − (n(z) + n∗ (z)) λw b (n(z) + n∗ (z) + 1)

ED

yj (z) =

(15)

PT

as a function of the own as well as the domestic and foreign competitors’ wage rates, wj (z), w(z), and w∗ (z), respectively.

To solve for the unions’ wage-setting problem, we substitute (15) into the union objectives

CE

in (4) and maximize the resulting expressions for the respective union wage rates. As formally shown in the Appendix, this gives a system of two equations that characterize the optimal wage

AC

choices for a given capital allocation: ¯ ¯2 ¯ 1 (z) = 2A [2 (n1 (z) + n2 (z)) + 1] + n2 (z) (n1 (z) + n2 (z)) λw λw 2 3n1 (z)n2 (z) + 2n2 (z) + 4n1 (z) + 4n2 (z) + 2 ¯w (n2 (z) + 1) [2n1 (z) + n2 (z) + 1] λ ¯1 + , 3n1 (z)n2 (z) + 2n2 (z)2 + 4n1 (z) + 4n2 (z) + 2 ¯ ¯1 ¯ 2 (z) = 2A [n1 (z) + 2n2 (z) + 2] + n1 (z) (n2 (z) + 1) λw λw 2 3n1 (z)n2 (z) + 2n2 (z) + 4n1 (z) + 4n2 (z) + 2 ¯w 2 (n2 (z) + 1) (n1 (z) + n2 (z)) λ ¯2 + . 2 3n1 (z)n2 (z) + 2n2 (z) + 4n1 (z) + 4n2 (z) + 2

(16)

(17)

Regarding the capital allocation, we distinguish between the short and the long run. cannot flow internationally. While we can show that the initial capital allocation also describes a medium-run equilibrium, we are not able to prove uniqueness of this equilibrium analytically. Since the medium run is not in the center of this paper’s interest, we do not further discuss it here.

10

ACCEPTED MANUSCRIPT

A short-run trade equilibrium with immobile capital Since investment decisions are given in the short run, capital allocation is the same as under

T

autarky, i.e. n1 (z) = n2 (z) = n(z). Accounting for w ¯i = βwi , we can therefore simplify wage

RI P

rates (16) and (17) in the following way:

2A [(2n + 1) + 2n(1 − β)] , (n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1) 2A [(2n + 1) + (n + 1)(1 − β)] = , (n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)

!
¯ 2 sr W2sr ≡ λw

SC

!
¯ 1 sr = W1sr ≡ λw

(18) (19)

where superscript ‘sr ’ refers to the short run. Substituting (18) and (19) into Eq. (15) yields

NU

short-run equilibrium output levels y1sr = W1sr (n + 1)(1 − β)/[b(2n + 1)], y2sr = W2sr 2n(1 − β)/[b(2n + 1)]. Due to symmetry of all producers in country i, we we can compute total em-

MA

ployment according to (1 − ui )L = nli . Accounting for li = yi , this gives 2nA(n + 1)(1 − β) [(2n + 1) + 2n(1 − β)] , b(2n + 1) {(n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)} 4n2 A(1 − β) [(2n + 1) + (n + 1)(1 − β)] . (1 − usr 2 )L = b(2n + 1) {(n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)}

ED

(1 − usr 1 )L =

(20) (21)

As formally shown in the Appendix, product market integration provides an employment stim-

PT

ulus in both countries and this stimulus is essential for gains from trade in our model. Similar to the closed economy, aggregate employment is equal to total real income, and hence we can infer from a positive employment effect that product market integration leads for both countries

CE

to higher welfare in the open economy than under autarky. In addition, we can determine the relative importance of wage-setting institutions for the

AC

macroeconomic performance in the open as compared to the closed economy by looking at the sr s f sign of ∆usr − ∆u, with ∆usr ≡ usr 1 − u2 and ∆u ≡ u − u . In the Appendix, we show that

the sign of ∆usr − ∆u is equivalent to the sign of −1 + (1 − β)(n2 − 1), which can be positive or negative. However, noting that even for countries with generous unemployment compensation schemes the assumption of β < 2/3 seems realistic,12 we can conclude that n ≥ 2 is sufficient for ∆us < ∆u when focusing on empirically relevant parameter domains. In this case, product market integration reduces the impact that differences in the degree of centralization exert on key macroeconomic variables, such as unemployment and welfare, which is in line with findings from previous research (see Danthine and Hunt, 1994). The following proposition summarizes the main insights from the analysis above. Proposition 1 Product market integration increases total employment and aggregate welfare in both countries, irrespective of the degree of centralization in the wage-setting process. Furthermore, provided that β < 2/3 and n ≥ 2, differences in the the degree of centralization in union 12

Gross replacement rates are smaller than http://www.oecd.org/dataoecd/60/8/49971171.xlsx)

11

2/3

in

all

OECD

countries.

(Source:

ACCEPTED MANUSCRIPT

wage-setting are less important for unemployment and welfare in the short-run open economy

T

than under autarky. Proof. Analysis in the text and derivation details in the Appendix.

RI P

We are also interested in the group-specific effects of product market integration. Substituting Wi from (18) and (19) as well as (1 − ui )L from (20) and (21) into Φi = (1 − ui )LWi , and dividing the resulting expression by price index P , gives

 Φ1 sr 2nA(n + 1)(1 − β) [(2n + 1) + 2n(1 − β)] = (22) P b [(2n + 1) + (n + 1)(1 − β)] {(n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)}  sr 4n2 A(1 − β) [(2n + 1) + (n + 1)(1 − β)] Φ2 = . (23) P b [(2n + 1) + 2n(1 − β)] {(n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)}

NU

SC



Similarly, we can substitute yi into Ψi = nΠi = bnyi2 to compute economy-wide profit income

Ψ1 P

sr

=

(n + 1)(1 − β) 2n + 1



Φ1 P

ED



MA

sr sr sr at the margin in the two economies: Ψsr 1 = Φ1 (n + 1)(1 − β)/(2n + 1) and Ψ2 = Φ2 2n(1 − β)/(2n + 1). Dividing by price index P gives

sr

,



Ψ2 P

sr

=

2n(1 − β) 2n + 1



Φ2 P

sr

.

(24)

In the Appendix, we show that both labor and profit income are unambiguously higher in the

PT

(short-run) open than in the closed economy. However, this does not mean that all agents benefit from product market integration. In particular, workers with a job in the closed and the open goods trade.

CE

economy may experience an income loss due to wage moderation of unions and thus may oppose The following proposition summarizes the effects of product market integration on group-

AC

specific income.

Proposition 2 Product market integration increases real income of capital owners and workers, irrespective of the degree of centralization in the wage-setting process. Proof. Analysis in the text and derivation details in the Appendix. A long-run trade equilibrium with footloose capital In the long run, capital owners adjust their investment decisions (at no cost) to increase their profit income. From the short-run analysis, we know that the initial capital allocation leads to higher wages and thus lower profits in country 1 and this triggers capital flows towards country 2 which continue until profit income is equalized between the two economies. Capital allocation in the long run can thus be characterized by two no arbitrage conditions: Π1 (z) = Π2 (z) ≡ Πt (z) and Πt (z) = Πt for all z. With linear demand, profits can only be equalized if output levels are the same in both countries and all industries, and this yields wi (z) ≡ w for all z and i = 1, 2, according to (15). Thereby, factor price equalization establishes a unique equilibrium with a 12

ACCEPTED MANUSCRIPT

symmetric firm allocation across industries, i.e. ni (z) = ni for all z, and a firm allocation across countries that renders sector-level wage-setting equivalent to firm-level wage-setting in country

T

1, i.e. n1 = 1, n2 = 2n − 1.13

RI P

With the equilibrium firm allocation at hand, we can now calculate employment, welfare and ¯ i = λw, ¯ n1 = 1, group-specific income in the long-run open economy equilibrium. Setting λw and n2 = 2n − 1, we can rewrite (16) and (17) in the following way: 2A , 1 + 2n(1 − β)

SC

¯ lr = W lr ≡ (λw)

(25)

where superscript ‘lr’ refers to the long run. Substituting the latter into (15) gives firm-level

NU

employment and output in the long-run open economy equilibrium: y lr = W lr 2n(1 − β)/[b(2n + 1)], with y1sr < y lr < y2sr . Adding up firm-level employment (output) over all local producers

  1 − ulr 1 L=

MA

gives economy-wide employment 4nA(1 − β) , b(2n + 1) [1 + 2n(1 − β)]

  1 − ulr 2 L=

4n(2n − 1)A(1 − β) . b(2n + 1) [1 + 2n(1 − β)]

(26)

ED

Comparing (26) with (20) and (21), we see that capital outflow lowers employment in country 1, whereas capital inflow raises employment in country 2. From inspection of Eq. (10), we can further note that if unemployment compensation is not too generous and the number of

PT

competitors not too small, the negative employment effect triggered by capital outflow in country 1 may be strong enough to reverse the positive short-run effect of product market integration.

CE

The higher is n, the more capital flows from country 1 to country 2, and the stronger is the negative employment effect in country 1. The higher is β, the smaller is firm-level employment and the smaller is ceteris paribus the number of domestic jobs replaced by foreign ones in the

AC

case of capital outflow.

¯ i /P , which equals: To determine country-specific welfare, we must look at total real income λI  ¯ lr 4nA(1 − β) (2n + 1) + 2n2 (1 − β) λI1 = , P b(2n + 1) [1 + 2n(1 − β)] (2n + 1) + 2n(1 − β)  ¯ lr 4nA(1 − β) (2n − 1)(2n + 1) + 2n2 (1 − β) λI2 = . P b(2n + 1) [1 + 2n(1 − β)] (2n + 1) + 2n(1 − β)

(27) (28)

according to (25) and (26).14 As formally shown in the Appendix, the inflow of capital unam13 In a supplement, which is available upon request, we provide a formal proof for the existence and uniqueness of the long-run open economy equilibrium. There, we also consider outcomes with one of the two countries specializing only on a subset of industries and show that this is inconsistent with an equilibrium (when ignoring the integer constraint). !
  14 lr lr 2 ¯ 1 lr = Φlr Total income of country 1 is given by λI 1 + Ψ1 = Φ1 1 + 2n (1 − β)/(2n + 1) , while total income  !
 2 2 lr lr ¯ 2 lr = Φlr of country 2 is given by λI 2 + Ψ2 = Φ2 1 + 2n (1 − β)/(4n − 1) . Furthermore, the value of total lr lr domestic output equals ni (P y) = Φi [1 + 2n(1 − β)/(2n + 1)]. Putting together and substituting for y lr and n1 = 1, n2 = 2n − 1, we obtain Eqs. (27) and (28).

13

ACCEPTED MANUSCRIPT

biguously raises welfare in country 2, whereas capital outflow has negative welfare consequences in country 1. However, this does not mean that country 1 is worse off in the long-run open econ-

T

omy equilibrium than under autarky. On the contrary, provided that capital owners repatriate

RI P

their profits from their foreign production activity, welfare losses associated with capital outflow are unambiguously lower than the welfare gains from product market integration in the short run.

Regarding the impact of the degree in union wage-setting on the relative macroeconomic

SC

performance in the two economies, we show in the Appendix that capital outflow raises both the employment as well as the welfare differential between the two economies, and this effect is

NU

strong enough to render the respective differentials larger than under autarky. Hence, the finding in Danthine and Hunt (1994) that in an open economy differences in wage-setting institutions are less important for the economic performance of countries does no longer hold – at least in

MA

our setting – if one accounts for international capital mobility as an important feature of open economies. The following proposition summarizes the impact of capital mobility on aggregate employ-

ED

ment and welfare as well as their differentials between the two economies. Proposition 3 In the long run, capital inflows increase aggregate employment and welfare in country 2, while capital outflows reduce employment and welfare in country 1. Furthermore,

PT

welfare is definitely higher in the long-run open economy equilibrium than under autarky. The same is true for employment in country 2, while it is not clear in general whether in the long run openness increases or reduces employment in country 1 relative to autarky. Finally, in the

CE

long run open economy, the employment and welfare differentials between the two economies are even more pronounced than under autarky.

AC

Proof. Analysis in the text and derivation details in the Appendix. With regard to the group-specific effects of capital mobility, we first look at labor income and calculate15



Φ1 P

lr

=

4nA(1 − β) , b[1 + 2n(1 − β)][(2n + 1) + 2n(1 − β)]

(29)



Φ2 P

lr

=

4n(2n − 1)A(1 − β) . b[1 + 2n(1 − β)][(2n + 1) + 2n(1 − β)]

(30)

Intuitively, capital outflow harms workers in country 1. However, this does not mean that workers also lose relative to autarky, because losses from capital outflow do not necessarily destroy all the benefits from product market integration. To be more specific, we find that workers are the more likely better off in the long-run open economy equilibrium than under autarky, the more generous is unemployment compensation and the weaker is product market From Footnote 14, we know that (Φi /P )lr = ni y lr [1 + 2n(1 − β)/(2n + 1)]−1 . Substituting y lr and n1 = 1, n2 = 2n − 1, then gives (29) and (30). 15

14

ACCEPTED MANUSCRIPT

competition. Due to capital inflow and the creation of new local jobs, workers in country 2 are unambiguously better off in the long-run open economy than in the short run or under autarky.

T

In a final step, we look at total real capital income. Abstracting from costs of capital

RI P

reallocation, in the long run profit income must be the same in the two economies, and it is given by16 

Ψ P

lr

=

8n3 A(1 − β)2 . b(2n + 1)[(2n + 1) + 2n(1 − β)][1 + 2n(1 − β)]

(31)

SC

Capital mobility improves investment opportunities of capital owners in country 1, who are therefore unambiguously better off in the long-run open economy equilibrium than in the short

NU

run or under autarky. Things are different for capital owners in country 2. The inflow of capital reduces the competitive advantage of country 2 firms, with negative consequences for the market position of these producers. As a consequence, capital owners in country 2 lose relative to the

MA

short-run open economy equilibrium, while they are still better off than under autarky. Proposition 4 In the long run, better investment opportunities reinforce the short-run stimulus of trade on total real capital income in country 1. Capital outflow lowers real labor income in

ED

country 1 relative to the short-run open economy, but these losses need not be high enough to destroy all benefits from product market integration. Capital inflow lowers income of capital owners in country 2, but does not entirely destroy this group’s benefits from product market

PT

integration. Finally, capital inflow augments the short-run gains of workers in country 2.

CE

Proof. Analysis in the text and derivation details in the Appendix.

Decentralization in union wage-setting

AC

4

In this section, we analyze how a shift from sector-level to firm-level wage-setting in country 1 affects the capital allocation in our model. Decentralization in union wage-setting not only refers to a common trend within OECD countries over the last few decades, but also captures a possible form of policy intervention that aims to destroy those factors which render a country unattractive for the investment of capital (see Delbecque, M´ejean, and Patureau, 2008, 2014). That this is a relevant policy option can be inferred from the observation that in the aftermath of the Eurozone crisis, the European Council has suggested to “review the wage-setting arrangements, and, where necessary, the degree of centralization in the bargaining process, [. . . ], while maintaining the autonomy of the social partners in the collective bargaining process” (European Council, 2011) as one promising instrument to stabilize the system. But can decentralization be a successful reform and allure footloose capital? To answer this question, it is worth noting that with firm-level wage-setting everywhere, the real wage at the margin is the same in both locations and given by W lr , while firm-level output is y lr . This 16

Substituting y lr into Ψ = nby 2 and P = 2(A − bny), it is straightforward to compute (Ψ/P )lr in (31).

15

ACCEPTED MANUSCRIPT

outcome does not depend on the international capital allocation, and hence the no arbitrage condition for investment does not depend on this allocation. Therefore, the equilibrium capital

T

allocation is undetermined if both countries host firm-level unions. To solve this problem, we

RI P

impose the additional assumption that capital owners will only adjust their investment decisions if de-investment increases their profit income. This implies that decentralization is ineffective and leaves all long-run equilibrium variables unchanged if union wage-setting occurs after a long-run equilibrium with firm allocation n1 = 1 and n2 = 2n − 1 has been established (ex-

SC

post decentralization, in short). On the contrary, if decentralization occurs prior to the capital outflow (ex-ante decentralization, in short) it is fully effective and bans the long-run incentives

NU

for de-investment in country 1.

Since ex-post decentralization does not alter the long-run equilibrium outlined in Section 3, we focus on the impact of ex-ante decentralization in the subsequent analysis. Noting that

MA

wage claims and output after the reform are given by W lr and y lr , respectively, we can calculate aggregate employment materializing under firm-level union wage-setting in both economies (and symmetric firm allocation n1 = n2 = n). This gives: 4n2 A(1 − β) , b(2n + 1)[1 + 2n(1 − β)]

ED

(1 − ur )L =

(32)

where superscript r indicates post-reform (or post-decentralization) variables. With output per

PT

firm increasing and the number of active firms remaining constant, total employment in country 1 is higher than in the short-run open economy equilibrium. Hence, the decentralization in

CE

union wage-setting is not only successful in abolishing the incentives for capital outflow, but it also provides an additional short-run stimulus for domestic employment, because it lowers union wage claims and thus the competitive disadvantage of domestic firms in the international

AC

market. Of course, this generates negative spillovers on country 2, where firms lose their competitive advantage and hence end up with lower output and lower employment than in the (pre-decentralization) short-run open economy equilibrium. Since there is no capital inflow in the long run, aggregate employment in country 2 unambiguously falls in response to decentralization in the wage-setting of country 1 unions. With welfare being directly linked to aggregate employment, it is immediate that the positive employment effects in country 1 are associated with welfare gains, while the employment reduction in country 2 is accompanied by welfare losses. With respect to the group-specific effects of ex-ante decentralization, we can first look at labor income. Dividing Φr = W r (1 − ur )L by the common price deflator yields 

Φ P

r

=

4n2 A(1 − β) . b[1 + 2n(1 − β)][(2n + 1) + 2n(1 − β)]

(33)

As formally shown in the Appendix, the short-run effects of ex-ante decentralization on workers

16

ACCEPTED MANUSCRIPT

in country 1 are not clearcut in general and depend on the competitive environment in the product market as well as the generosity of unemployment compensation. If unemployment

T

benefits are small and competition sufficiently strong, workers in country 1 are worse off after

RI P

the decentralization in the wage-setting of local unions. However, this does not mean that workers should oppose the reform. Decentralization in union wage-setting, while generating short-run losses, may still be to the benefit of workers, because it helps avoiding the capital outflow and thus the even more disastrous long-run outcome in Eq. (29). In country 2, workers

SC

face double losses from ex-ante decentralization in the wage-setting of country 1. Firms in country 2 hire less workers in the short run and there is no capital inflow and hence no creation

NU

of new jobs in the long run. Real profit income equals the respective expression for the long-run open economy equilibrium in (31), and we can thus infer the impact of ex-ante decentralization on capital owners from the respective discussion in Section 3. Capital owners in country 1 are

MA

better off after the change in the local wage-setting institutions than in the short run open economy equilibrium or under autarky. Capital owners in country 2 lose relative to the (predecentralization) short run open economy equilibrium but are still better off than in the closed economy.

ED

The following proposition summarizes the main insights from the analysis above. Proposition 5 For the effectiveness of decentralization in union wage-setting, the timing is im-

PT

portant. If decentralization occurs after the capital outflow, it is not successful in restoring the initial capital allocation. However, if decentralization occurs early, it can prevent the capital outflow with positive consequences for domestic employment, welfare, and possibly the real income

CE

of workers. This success comes at the cost of negative spillovers on country 2, where employment, welfare, and real labor income shrink in response to decentralization in the wage-setting of

AC

country 1 unions. At least in the long run, capital owners are not affected by the decentralization in country 1, because they can always enforce the outcome of firm-level wage-setting by relocating their investment accordingly. Proof. Analysis in the text and derivation details in the Appendix. The general recommendation from our analysis for policy makers who aim at securing domestic jobs in an open economy is clear. Act early to prevent capital outflow, because it may be difficult (if not impossible) to reverse the investment decisions of domestic capital owners once they have set up their production facilities abroad. The costs of responding late to new challenges in an open economy may be even more significant if agglomeration forces are at work. In this case, a government that aims at persuading domestic capital owners to invest at home instead of abroad may have to pay the full agglomeration rent – in addition to the direct costs of de-investment in the foreign country – when these capital owners have already closed their domestic plants because of the strong local wage-setting institutions.

17

ACCEPTED MANUSCRIPT

5

Concluding remarks

T

This paper presents a general oligopolistic equilibrium model with unionized labor markets and two countries that differ in the degree of centralization in union wage-setting. In this framework,

RI P

we investigate how openness alters the way in which the degree of centralization in union wagesetting affects key macroeconomic variables, such as welfare and unemployment. Thereby we distinguish two forms of openness: a short-run scenario, in which product markets are fully

SC

integrated, while capital remains invested at home (and firm allocation is as in autarky); and a long-run scenario, in which capital is mobile. In the short run, product market integration lowers the scope of unions to set excessive wages. This has positive effects on welfare and economy-

NU

wide employment in the two countries, with both income groups – capital owners and workers – benefiting from the overall increase in production.

In the long run capital is internationally mobile and searches for the best investment oppor-

MA

tunities worldwide. As a consequence, it moves to the country with less centralized wage-setting and lower labor costs. The capital outflow reduces welfare and employment in the country with

ED

the higher degree of centralization in union wage-setting and alters the distribution of income in this economy significantly. While workers are worse off due to an export of jobs, capital owners benefit from having access to better investment opportunities. Things are exactly the opposite in the country with the more decentralized level of wage-setting. Due to an inflow of capital this

PT

country experiences a welfare gain and an employment expansion. Furthermore, while workers benefit from the creation of new local jobs, capital owners are worse off, because their firms lose

CE

the competitive advantage in the product market. To round off the discussion in this paper, we have looked at the consequences of decentralization of wage-setting in the country with the more severe labor market imperfection. The results

AC

from this analysis make clear that such a reform can be successful in preventing capital outflow when it occurs before the relocation of capital starts. Such early attempts to decentralize union wage-setting can indeed be essential for securing benefits of product market integration in the long run and for rendering globalization a success story for all income groups. On the contrary, if decentralization starts after capital owners have adjusted their investment decisions in the long run, the reform is less promising and may fail to restore the initial capital allocation. In this case, long-run losses of some income groups may be unavoidable, rendering strong and persistent opposition by the respective income groups a real threat to globalization. While we hope that this paper broadens the understanding of how different wage-setting institutions shape the outcome in open economies, it is clear that the analysis builds on many simplifying assumptions which limit the ability of our model to inform policy makers on how to solve real world problems. One restrictive feature of our analysis is the assumption of an integrated global goods market and zero trade costs. To make sure that our results are not driven by this assumption, we have analyzed a modified model variant with segmented goods markets and positive trade costs in a supplement to this paper. There, we show that this modification 18

ACCEPTED MANUSCRIPT

reduces the tractability of our model, but it does not change the main message from our analysis regarding the role of union wage-setting for capital flows and the timing of labor market reforms

T

that can successfully prevent capital outflow. Another restrictive assumption in our model is the

RI P

immobility of workers. Whereas a detailed discussion on how migration alters the insights from our analysis is beyond the scope of this paper, it is worth noting that the higher probability of getting a job abroad may be a key rationale for emigration in our setting. If migration were possible, workers would follow capital in the long run, and this points to differences in the

SC

degree of centralization in union wage-setting as an important source of agglomeration, with industry production concentrating in those locations that offer the least restrictive labor market

NU

institutions.

MA

References

Bastos, P., and U. Kreickemeier (2009): “Unions, Competition and International Trade in General Equilibrium,” Journal of International Economics, 79(2), 238–247.

ED

Blanchard, O., and F. Giavazzi (2003): “Macroeconomic Effects of Regulation and Deregulation in Goods and Labor Markets,” Quarterly Journal of Economics, 118(3), 879–907.

PT

Brecher, R. A. (1974): “Minimum Wage Rates and the Pure Theory of International Trade,” Quarterly Journal of Economics, 88(1), 98–116.

CE

Calmfors, L., and J. Driffill (1988): “Bargaining Structure, Corporatism and Macroeconomic Performance,” Economic Policy, 3(6), 14–61. Danthine, J.-P., and J. Hunt (1994): “Wage Bargaining Structure, Employment and Eco-

AC

nomic Integration,” Economic Journal, 104(424), 528–541. Delbecque, V., I. M´ ejean, and L. Patureau (2008): “Social Competition and Firms’ Location Choices.,” CEPII Working Paper 2008-12. (2014): “Labor Market Institutions and Firms’ Location Choices,” Review of World Economics, 150(1), 115–148. Eckel, C., and H. Egger (2009): “Wage Bargaining and Multinational Firms,” Journal of International Economics, 77(2), 206–214. Egger, H., and D. Etzel (2012): “The Impact of Trade on Employment, Welfare, and Income Distribution in Unionized General Oligopolistic Equilibrium.,” European Economic Review, 56(6), 1119–1135. European icy

Council

Coordination

(2011): for

“The

Euro

Competitiveness

Plus

and 19

Pact.

Stronger

Convergence.

Economic

Conclusions

of

Polthe

ACCEPTED MANUSCRIPT

European

Council

24/25

March

2011,”

Brussels,

Belgium,

April

20,

2011:

http://www.consilium.europa.eu/uedocs/cms data/docs/pressdata/en/ec/120296.pdf.

T

Gross, D. M., and M. Ryan (2008): “FDI Location and Size: Does Employment Protection

RI P

Legislation Matter?.,” Regional Science and Urban Economics, 38(6), 590–605. Huizinga, H. (1993): “International Market Integration and Union Wage Bargaining,” Scan-

SC

dinavian Journal of Economics, 95(2), 249–255.

Javorcik, B. S., and M. Spatareanu (2005): “Do Foreign Investors Care about Labor Market

NU

Regulations?.,” Review of World Economics, 141(3), 375–403. Kreickemeier, U., and F. Meland (2013): “Non-traded Goods, Globalization and Union Influence.,” Economica, 80(320), 774–792.

MA

Leahy, D., and C. Montagna (2000): “Unionisation and Foreign Direct Investment: Challenging Conventional Wisdom,” Economic Journal, 110(462), C80–C92. Lommerud, K. E., F. Meland, and L. Sørgard (2003): “Unionised Oligopoly, Trade Lib-

ED

eralisation and Location Choice,” Economic Journal, 113(490), 782–800. Martin, P., and C. A. Rogers (1995): “Industrial Location and Public Infrastructure,”

PT

Journal of International Economics, 39(3-4), 335–351. Mezzetti, C., and E. Dinopoulos (1991): “Domestic Unionization and Import Competi-

CE

tion,” Journal of International Economics, 31(1-2), 79–100. Munch, J. R. (2003): “The Location of Firms in Unionized Countries,” Scandinavian Journal

AC

of Economics, 105(1), 49–71. Naylor, R. (1999): “Union Wage Strategies and International Trade,” Economic Journal, 109(452), 102–125.

Neary, J. P. (2009): “International Trade in General Oligopolistic Equilibrium,” unpublished manuscript, University of Oxford. OECD (2004): OECD Employment Outlook. OECD, Paris. Persyn, D. (2013): “Union Wage Demands with Footloose Firms and Agglomeration Forces.,” Regional Science and Urban Economics, 43(1), 142–150. Picard, P. M., and E. Toulemonde (2006): “Firms Agglomeration and Unions,” European Economic Review, 50(3), 669–694. Sørensen, J. R. (1993): “Integration of Product Markets When Labour Markets Are Unionized,” Recherches Economiques de Louvain, 59(4), 485–502. 20

ACCEPTED MANUSCRIPT

Appendix

T

Derivation of Eqs. (11) and (12)

RI P

Since, by definition, economy-wide labor and profit income must add up to total income, i.e. Φ + Ψ = λI, we can write Φ = φλI and Ψ = ψλI, where φ and ψ denote the income shares attributed to workers and capital owners, respectively. For the two wage-setting institutions, we

n+1 , n+2−β n+1 φf = , (n + 1) + n(1 − β)

SC

can thus calculate

1−β , n+2−β n(1 − β) ψf = . (n + 1) + n(1 − β) ψs =

NU

φs =

(34) (35)

Noting further that a binding budget constraint implies that total income must equal total

MA

revenues, λI = P ny, we can calculate total real labor and capital income, Φ/P = φ(1 − u)L and Ψ/P = ψ(1 − u)L, respectively. Using (10), (34) and (35), we obtain Eqs. (11) and (12). QED

ED

Derivation of Eq. (15)

Maximizing profits (14) for yj (z), gives the first-order condition dΠtj (z)/dyj = 0. Solving the P ¯ j (z). We can now latter for yj , gives the best-reply function 2byj (z) = 2A − b k6=j yk (z) − λw

PT

note two things: first, a structurally identical best-response function can be calculated for any other producer k 6= j; second, due to perfect foresight, firm j rationally anticipates that all

CE

competitors of country i = 1, 2 choose the same output in equilibrium. Introducing an asterisk for indicating foreign variables, we can thus rewrite the best response function of firm j in the

AC

following way:

yj (z) =

¯ j (z) 2A − b (n(z) − 1) y(z) − bn∗ (z)y ∗ (z) − λw , 2b

(36)

where y(z), y ∗ (z) refers to the common output of domestic and foreign competitors, respectively. Accounting for the symmetry assumption of domestic and foreign competitors in the first orderconditions of the respective producers, we can furthermore calculate ¯ 2A − bn∗ (z)y ∗ (z) − byj (z) − λw(z) n(z)b ¯ ∗ (z) 2A − b (n(z) − 1) y(z) − byj (z) − λw , y ∗ (z) = (n∗ (z) + 1) b y(z) =

21

(37) (38)

ACCEPTED MANUSCRIPT

¯ ¯ ∗ (z) refer to the common wage rates of domestic and foreign competitors where λw(z) and λw of firm j, respectively. We can now solve system (37) and (38) for y(z) and y ∗ (z). This gives

T

¯ ∗ (z) − (n∗ (z) + 1) λw(z) ¯ 2A − byj (z) + n∗ (z)λw , ∗ (n(z) + n (z)) b ¯ ¯ ∗ (z) 2A − byj (z) + (n(z) − 1) λw(z) − n(z)λw . y ∗ (z) = ∗ (n(z) + n (z)) b

RI P

y(z) =

(39) (40)

SC

Substituting (39) and (40) into (36), finally gives (15). QED

NU

Derivation of Eqs. (16) and (17)

Since sector-level unions choose a uniform wage rate for all employees in the respective sec-

MA

tor, we can set w1j (z) = w1 (z) in (15) to determine industry-wide employment in country 1: Pn(z) j=1 l1j (z) = ny1 (z). Substituting the latter into the union objective of sector-level union in (4), gives

ED

  ¯ 2 (z) − (n2 (z) + 1) λw ¯ 1 (z) [w1 (z) − w ¯1 ] n1 (z) 2A + n2 (z)λw V1 = . b (n1 (z) + n2 (z) + 1) Maximizing V1 for w1 (z) gives the first-order condition dV1 /w1 (z) = 0, which can be reformulated

PT

to

(41)

CE

¯ ¯ ¯1 ¯ 1 (z) = 2A + n2 (z)λw2 (z) + (n2 (z) + 1) λw . λw 2 (n2 (z) + 1)

In a similar vein, we can substitute (15) together with l2j (z) = y2j (z) into the objective function

AC

of firm-level unions in (4), which gives   ¯ 1 (z) + (n2 (z) − 1) λw ¯ 2 (z) − (n2 (z) + n1 (z)) λw ¯ 2j (z) [w2j (z) − w ¯2 ] 2A + n1 (z)λw V2j (z) = . b (n1 (z) + n2 (z) + 1) Maximizing this objective for w2 j(z) gives the first-order condition dV2j /dw2j = 0. Rearranging terms and noting that w2j (z) = w2 (z) must hold due to ex-post symmetry we can calculate ¯ ¯ ¯2 ¯ 2 (z) = 2A + n1 (z)λw1 (z) + (n1 (z) + n2 (z)) λw λw . 2n1 (z) + n2 (z) + 1

(42)

Eqs. (41) and (42) constitute a system of two equations, which jointly determine wage rates ¯ 1 (z) and λw ¯ 2 (z) in (16) and (17). QED λw

The impact of product market integration on economy-wide employment s Using (10) and (20), we can show that the sign of ∆u1 ≡ (1−usr 1 )L−(1−u )L is equivalent to the

sign of γ¯1u ≡ (2n+1)(2n2 +2n+1)+(1−β)(n+1)(2n2 +5n+1)+(1−β)2 2n(n+1) and thus positive. 22

ACCEPTED MANUSCRIPT

f In a similar way, we can infer from (10) and (21) that the sign of ∆u2 ≡ (1 − usr 2 )L − (1 − u )L is

equivalent to the sign of γ¯2u ≡ (2n+1)(2n+3)+(1−β)(n+1)(2n2 +3n+3)+(1−β)2 2n(n+1), and

T

hence is also positive. Putting together, we can thus conclude that product market integration

RI P

stimulates employment in both locations. QED

The impact of product market integration on rent sharing

n2 (1 − β) > 0, [(2n + 1) + (n + 1)(1 − β)][(n + 1) + (1 − β)]

NU

s sr ∆ψ 1 ≡ ψ1 − ψ =

SC

Looking first at country 1, we can infer from a comparison of (34) and (48) that

while for country 2, we get

n(1 − β) > 0. [(2n + 1) + 2n(1 − β)][(n + 1) + n(1 − β)]

MA

sr f ∆ψ 2 ≡ ψ2 − ψ =

(43)

(44)

Hence, we see that capital owners in both countries are able to extract a larger share of economic

ED

rents in the short-run open economy equilibrium than under autarky.

PT

The impact of product market integration on the employment and welfare differential between the two economies In the closed economy the employment differential between the two countries is given by ∆u ≡

CE

(1 − uf )L − (1 − us )L = (us − uf )L: n(n − 1)A(1 − β) , b(n + 1)(2 − β) [1 + n(1 − β)]

(45)

AC

∆u =

according to (10). Furthermore, accounting for (20) and (21), we can compute the respective differential in the short-run open economy: ∆usr =

2n(n − 1)A(1 − β) . b {(n + 1)(1 − β) [(3n + 1) + 2n(1 − β)] + (2n + 1)}

(46)

Combining (45) and (46), it is straightforward to show that the sign of ∆u − ∆usr is equivalent to δusr ≡ −1 + (1 − β)(n2 − 1). Hence, ∆u − ∆usr is positive if β < 2/3 and n ≥ 2. Finally, noting that aggregate employment equals total real income in the two economies and and that total real income is a suitable welfare measure in our model, we can conclude that product market integration lowers the employment and welfare differential between the two economies. QED

23

ACCEPTED MANUSCRIPT

Derivation of Eqs. (22)-(24) ¯ i and P (1 − ui )L = λI ¯ i , we Noting from the derivation of Eqs. (11) and (12) that Φi + Ψi = λI

2n + 1 , (2n + 1) + (n + 1)(1 − β) 2n + 1 , = (2n + 1) + 2n(1 − β)

RI P

T

can compute

(n + 1)(1 − β) , (2n + 1) + (n + 1)(1 − β) 2n(1 − β) = . (2n + 1) + 2n(1 − β)

ψ1sr =

φsr 2

ψ2sr

(47) (48)

SC

φsr 1 =

Substituting (20) and (21) together with (47) and (48) into Φi /P = φi (1 − ui )L and Ψi /P =

NU

ψi (1 − ui )L, allows us to compute Eqs. (22)-(24). QED

The impact of product market integration on real labor income

MA

Looking first at country 1, we can note that total real labor income (and thus the welfare of workers) in the short-run open economy is higher than, equal to, or smaller than in the sr closed economy if ∆Φ − (Φ/P )s >, =, < 0. In view of (11) and (22), we can 1 ≡ (Φ1 /P )

ED

3 2 furthermore show that the sign of ∆Φ 1 is equivalent to the sign of γ1 (n, β) ≡ 4n + 6n + 4n + ! 2
!
1 + (1 − β) 2n + 7n + 2 (n + 1) − (1 − β)2 3n2 − 6n − 1 (n + 1) − 2(1 − β)3 n (n − 1) (n + 1).

Noting that γ1 (n, β) > 0 holds for any possible combination of n ≥ 1 and β ∈ (0, 1), we

PT

can conclude that trade increases real income and welfare of workers in country 1. Looking at country 2, we can note that total labor income in the open economy is higher than, equal sr s to, or lower than under autarky if ∆Φ 2 ≡ (Φ/P ) − (Φ/P ) >, =, < 0. In view of (11) and

CE

(23), we can furthermore show that the sign of ∆Φ 2 is equivalent to the sign of γ2 (n, β) ≡ (2n + 1)(2n + 3) + (1 − β)[(n + 1)2 (2n + 1) + (2n + 1)2 + 1] + 2(1 − β)2 n(n2 + n + 2). Noting

AC

that γ2 (n, β) > 0 holds for any n ≥ 1 and β ∈ (0, 1), we can thus safely conclude that product market integration increases real income and welfare of workers in country 2. QED

The impact of trade and capital mobility on aggregate employment Let us first look at country 1. The employment effects of capital outflow are determined by ¯ u ≡ usr − ulr , which in view of (20) and (26) is equal to the sign of γ¯u (n, β) ≡ the sign of ∆ 1 1 1 1   −(n − 1) (2n + 1) + 2(n + 1)(2n + 1)(1 − β) + 4n(n + 1)(1 − β)2 . Since γ¯1 (n, β) ≤ 0 holds for all

n ≥ 1 and β ∈ (0, 1), aggregate employment in country 1 must be lower in the long run than in

the short-run open economy. Furthermore, to see whether capital mobility reverses the positive ˜ u ≡ us − ulr . employment stimulus from product market integration, we have to look at ∆ 1

1

1

˜ u is equivalent to the sign of γ˜ u (n, β) ≡ From (10) and (26), it follows that the sign of ∆ 1 1 ! 2
(2n + 3) − 2(1 − β) 2n − n − 2 . It is obvious that γ˜1u (n, 1) = 2n + 3 > 0, while γ˜1u (n, 0) = −4n2 + 4n + 7, where γ˜1u (0, 0) = 7 and limn→∞ γ˜1u (n, 0) < 0 imply that the sign of γ˜1u (n, 0)

is ambiguous. In addition, we can show that γ˜1u (1, β) = 7 − 2β > 0. Accounting for the properties of γ˜1u (n, β), we can therefore conclude that openness may have negative long-run 24

ACCEPTED MANUSCRIPT

employment effects in country 1 if n is sufficiently large, while β is sufficiently small. Turning ¯ u ≡ usr − ulr ≥ 0. to country 2, we can note that capital provides an employment stimulus if ∆ 2

2

2

RI P

T

Noting that, in view of (21) and (26), the sign of ∆u2 is equivalent to the sign of γ¯2u (n, β) ≡   (n − 1) (2n + 1) + (1 − β)(2n2 + 4n + 1) + 2n(n + 1)(1 − β)2 and thus positive for any n > 1 and β ∈ (0, 1), we can safely conclude that capital inflow reinforces the employment stimulus from product market integration. QED

SC

The impact of trade and capital mobility on aggregate welfare

NU

Let us first look at country 1. The welfare implications of capital outflow can be inferred from !
!
¯ 1 /P sr . Noting from our previous analysis that (λI1 /P )sr = ¯ 1 /P lr − λI ¯ U ≡ λI the sign of ∆ 1 ¯U (1 − usr 1 ) L, it follows from (20) and (27) that the sign of ∆1 is equivalent to the sign of   γ¯1U (n, β) ≡ −(n − 1) (2n + 1)2 + 2(1 − β)(2n + 1)(2n2 + 2n + 1) + 4(1 − β)2 n(n + 1)2 . Since

MA

γ¯1 (n, β) < 0 holds for any n > 1 and β ∈ (0, 1), it is clear that capital outflow lowers wel-

fare relative to the short-run open economy. To see, whether this detrimental effect can be strong enough to reverse the positive welfare implications of product market integration, we !
!
s !
s ¯ 1 /P lr − λI/P ¯ ¯ ˜ U ≡ λI have to determine the sign of ∆ . Noting that λI/P = (1 − us )L, 1

ED

˜ U must be positive because γ˜ U (n, β) ≡ (2n + 1)(2n + we can infer from (10) and (27) that ∆ 1 1

3) + 4(1 − β)(n + 1)2 + 4(1 − β)2 n2 > 0 holds for any n > 1 and β ∈ (0, 1). Therefore, we

PT

can safely conclude that welfare in country 1 is higher in the long run open economy equilibrium than under autarky. To determine the welfare effects of capital inflow in country 2, !
!
¯ 2 /P sr . Noting that (λI2 /P )sr = (1 − usr ) L and ¯ 2 /P lr − λI ¯ U ≡ λI we can evaluate ∆ 2

2

CE

¯ U is equivalent to the sign of accounting for (21) and (28), we can show that the sign of ∆ 2   γ¯2U (n, β) ≡ (n − 1) (2n + 1)2 + (1 − β)(4n3 + 10n2 + 6n + 1) + 2(1 − β)2 n(n2 + 3n + 1) . Since

AC

γ¯2U (n, β) is positive for any n > 1, β ∈ (0, 1), we can thus safely conclude that capital inflow amplifies the positive short-run welfare gains from product market integration. QED

The impact of trade and capital mobility on the employment and welfare differential In the closed economy, the employment differential between the two countries is given by (45). In lr the long-run open economy, the respective differential is given by ∆ulr ≡ (1 − ulr 2 )L − (1 − u1 )L:

∆ulr ≡

8n(n − 1)A(1 − β) , b(2n + 1) [1 + 2n(1 − β)]

(49)

according to (26). Combining (45) and (49), we can show that the sign of ∆u−∆ulr is equivalent to the sign of δulr ≡ −(6n + 3) − 2(1 − β)(4n2 + 7n + 2) − (1 − β)2 8n(n + 1), which is negative. In a similar vein, we can compare (46) and (49) to see that the sign of ∆usr − ∆ulr is equivalent to the sign of δ˜lr ≡ −(7n + 3) − 2(1 − β)(n + 1)(5n + 2) − (1 − β)2 8n(n + 1) and thus negative. u

25

ACCEPTED MANUSCRIPT

Let us now turn to the welfare differential. Accounting for (27) and (28), we can show that

T

the real income differential between the two countries in the long-run open economy equilibrium ¯ 2 /P )lr − (λI ¯ 1 /P )lr : ˜ lr ≡ (λI is given by ∆U 8n(n − 1)A(1 − β) . b[1 + 2n(1 − β)][(2n + 1) + 2n(1 − β)]

RI P

˜ lr ≡ ∆U

(50)

SC

¯ ¯ ˜ ≡ (λI/P Noting further that ∆U )f − (λI/P )s = ∆u holds in the closed economy, we can ˜ − ∆U ˜ lr is equivalent to the sign of infer from comparing (45) with (50) that the sign of ∆U

NU

δUlr ≡ −(6n + 7)− 4(1− β)(n2 + 3n + 2)− (1− β)2 4n(n + 2) and thus negative. Finally, noting that ¯ 2 /P )sr −(λI ¯ 1 /P )sr = ∆usr , it follows from (46) and (50) that the sign of ∆U ˜ sr ≡ (λI ˜ sr −∆U ˜ lr ∆U 2 2 lr is equivalent to the sign of δ˜ ≡ −(6n + 3) − 4(1 − β)(2n + 3n + 2) − 4(1 − β) n(n + 2) and thus U

negative. Putting together, we can therefore conclude that both the employment differential and the short run or under autarky. QED

MA

the welfare differential are more pronounced in the long-run open economy equilibrium than in

ED

The impact of trade and capital mobility on total real labor income

PT

For country 1, we can infer the impact of capital outflow on total real labor income from the sign ¯ Φ ≡ (Φ1 /P )lr − (Φ1 /P )sr . In view of (22) and (29), we can conclude that γ¯Φ (n, β) ≡ −(n − of ∆ 1  1  1) (2n+1)2 +4(1−β)(n+1)(2n2 +3n+1)+2(1−β)2 (n+1)(8n2 +7n+1)+4(1−β)3 (n+1)n < 0 ¯ Φ < 0, so that capital outflow reduces total real labor income in country 1 relative to implies ∆ 1

the short-run open economy equilibrium. To see whether this income loss is strong enough to 1

CE

reverse the positive real income stimulus from product market integration, we have to determine ˜ Φ ≡ (Φ1 /P )lr − (Φ/P )s . In view of (11) and (29) we can conclude that the sign the sign of ∆

AC

˜ Φ is equivalent to the sign of γ˜Φ (n, β) ≡ (2n + 3) − 4(1 − β)(n2 − 2) − 4(1 − β)2 (n2 − 1). of ∆ 1 1 It is easily confirmed that γ˜1Φ (n, 1) = 2n + 3 > 0, while γ˜1Φ (n, 0) = 15 + 2n − 8n2 , where γ˜1Φ (0, 0) = 15 > 0 and limn→∞ γ˜1Φ (n, 0) < 0 imply that the sign of γ˜1Φ (n, 0) is ambiguous. Accounting for the properties of γ˜1Φ (n, β), we can therefore conclude that openness may generate long-run welfare losses of workers if n is sufficiently large, while β is sufficiently small. Turning to country 2, we can infer the impact of capital inflow on total real labor income by determining ¯ Φ ≡ (Φ2 /P )lr − (Φ2 /P )sr . Accounting for (23) and (30), we can conclude that the sign of ∆ 2  !
 ¯ Φ > 0, so γ¯2Φ (n, β) ≡ (n − 1) (2n + 1)2 + (1 − β) 2n2 + 4n + 1 + 2(1 − β)2 n(n + 1) > 0 implies ∆ 2

that capital inflow amplifies the positive welfare implications for workers triggered by product market integration. QED

The impact of trade and capital mobility on total real capital income Let us first look at country 1, where we can infer the impact of capital inflow on total real capital ¯ Ψ ≡ (Ψ1 /P )lr − (Ψ1 /P )sr . Accounting for (24) and (31), income from determining the sign of ∆ 1  ¯ Ψ is equivalent to the sign of γ¯Ψ (n, β) ≡ (n−1) (2n+1)2 (3n+1)+ we can show that the sign of ∆ 1 1 26

ACCEPTED MANUSCRIPT

 2(1− β)n(n + 1)(2n + 1)(4n + 3)+ 4(1− β)2 n2 (n + 1)(3n + 2) , which is positive for any n > 1 and β ∈ (0, 1). Hence, we can safely conclude that capital outflow reinforces the positive short-run

T

impact of product market integration on total real capital income. Turning to country 2, we can

RI P

infer the impact of capital inflow on total real income of domestic capital owners from the sign ¯ Ψ ≡ (Ψ2 /P )lr − (Ψ2 /P )sr , which is equivalent to the sign of γ¯ Ψ ≡ −(1 − β)n(n − 1) and thus of ∆ 2

2

negative. This implies that capital inflow lowers total real income of capital owners in country 2 relative to the short-run open economy equilibrium. To see whether this negative impact is strong

SC

enough to reverse the short-run benefits of this income group from product market integration, ˜ Ψ ≡ (Ψ2 /P )lr −(Ψ2 /P )f . Accounting for (12) and (31), we can show we can look at the sign of ∆ 2

NU

˜ Ψ is equivalent to the sign of γ˜Ψ ≡ 4n2 +12n+7+(1−β)12n(n+1)+(1−β)2 4n2 that the sign of ∆ 2 2 and thus positive. Hence, capital owners in country 2, while losing from capital inflow, are better

MA

off in a long-run open economy equilibrium than under autarky. QED

Derivation of Eq. (33)

Accounting for φ = Φr /(λI)r and (λI)r = Ψr + Φr , we can calculate φr = (2n + 1)/[(2n + 1) +

ED

2n(1 − β)]. Substituting the latter together with Eq. (32) into (Φ/P )r = φr (1 − ur )L, establishes Eq. (33). QED

PT

The impact of ex-ante decentralization in union wage-setting on employment and welfare

CE

Let us first look at country 1, where we can infer the employment effects of ex-ante decenˆ u ≡ usr − ur . Accounting tralization in union wage-setting from determining the sign of ∆ 1 1 1 u ˆ for (20) and (32), we can show that the sign of ∆ is equivalent to the sign of γˆu (n, β) ≡ 1

1

AC

(n − 1)[(2n + 1) + 2n(n + 1)(1 − β)]. Since γˆ1u (n, β) > 0 holds for any n > 1 and β ∈ (0, 1) we can conclude that ex-ante decentralization in the wage-setting of domestic unions stimulates employment and – in the absence of international capital flows – also welfare in country 1 relative to the short-run open economy equilibrium in Section 3. These positive aggregate effects of decentralization also extend to the long run, because we know from the previous analysis that capital outflow, which is prevented by the reform of country 1’s wage-setting institutions, is associated with a decline in employment and welfare in country 1. For country 2, we can infer the employment and welfare effects of ex-ante decentralization in union wage-setting ˆ u ≡ usr − ur , which, in view of (21) and (32), is equivalent to the sign of from the sign of ∆ 2

2

2

γˆ2u (n, β) ≡ −(1 − β)n(n − 1) and thus negative. We can therefore conclude that ex-ante decentralization in the wage-setting of country 1 unions lowers employment and welfare in country 2 relative to the short-run and – since there are no international capital flows after the reform – the long-run open economy equilibrium in Section 3. QED

27

ACCEPTED MANUSCRIPT

The impact of ex-ante decentralization in union wage-setting on total real labor income

RI P

T

For country 1, we can infer the short-run impact of ex-ante decentralization in the wage-setting ˆ Φ ≡ (Φ1 /P )r − (Φ1 /P )sr . In view of local unions on total real labor income from the sign of ∆ 1

ˆ Φ is equivalent to the sign of (n − 1)ˆ γ1Φ , with of (22) and (33), we can show that the sign of ∆ 1 γˆ1Φ ≡ (2n + 1)2 + (1 − β)2n(n + 1)(2n + 1) − (1 − β)2 2n(n + 1)2 − (1 − β)3 4n2 (n + 1). It is

SC

easily confirmed that γˆ1Φ (n, 1) = (2n + 1)2 > 0, while γˆ1Φ (n, 0) = −2n3 + 2n2 + 4n + 1, where γˆ1Φ (1, 0) = 1 > 0, and limn→∞ γˆ1Φ (n, 0) < 0 imply that the sign of γˆ1Φ (n, 0) is not clearcut in general. To be more specific, there exists a unique n ˆ (β) > 1 such that γˆ1Φ (n, 0) > 0 if n < n ˆ (β),

NU

while γˆ1Φ (n, 0) < 0 if n > n ˆ (β). Accounting for the properties of γˆ1Φ (n, β), we can therefore conclude that ex-ante decentralization in the wage-setting of domestic unions can lower total real labor income in country 1 relative to the short-run open economy in Section 3, if n is

MA

sufficiently high, while β is sufficiently small. Otherwise, workers in country 1 benefit form this change in local wage-setting institutions. Regarding the long-run implications of the ex-ante decentralization in domestic union wage-setting, it is straightforward to infer from (29) and (33)

ED

that (Φ1 /P )r − (Φ1 /P )lr > 0. Furthermore, we can determine the short-run consequences of ex-ante decentralization in the wage-setting of country 1 for total real labor income in country ˆ Φ ≡ (Φ2 /P )r − (Φ2 /P )sr , which, in view of (23) and (33), is 2 when looking at the sign of ∆ 2

PT

equivalent to the sign of γˆ2Φ ≡ −(1 − β)n(n − 1) and thus negative. Since we also know from the previous analysis that capital inflow renders workers in country 2 better off, we can conclude that decentralization in the wage-setting of country 1 unions lowers total real labor income in

AC

QED

CE

country 2 relative to the short-run and long-run open economy equilibria analyzed in Section 3.

28

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

SC

RI P

T

!"#$%"#$&'( ! • "#!$%&'()#*!+&(%&!,-.#!'#//(&.!(&!-!.#&#*-0!%0(.%1%0('/($!#2+(0(3*(+4!4%)#05! • "#!'6%,!6%,!$#&/*-0(7-/(%&!(&!+&(%&!,-.#!'#//(&.!-88#$/'!#410%94#&/!-&)!,#08-*#! (&!-&!%1#&!#$%&%495! • "#!6(.60(.6/!/6#!)(88#*#&/(-0!(41-$/!%8!1*%)+$/!4-*:#/!-&)!$-1(/-0!4-*:#/! (&/#.*-/(%&5! • "#!'/+)9!/6#!$%&'#2+#&$#'!%8!)#$#&/*-0(7-/(%&!(&!+&(%&!,-.#!'#//(&.!(&!%1#&! #$%&%4(#'5!