welfare policies, labour taxation and international integration

International Tax and Public Finance, 10, 43–62, 2003©C 2003 Kluwer Academic Publishers. Printed in The Netherlands.

Welfare Policies, Labour Taxationand International Integration

TORBEN M. ANDERSEN [email protected], EPRU and IZA, Department of Economics, University of Aarhus, 8000 Aarhus, Denmark

Abstract

How will international integration affect welfare policies? This paper considers the possibilities of financing publicsector activities (public consumption and social security expenses) by general (wage) taxation in an economywhich becomes more integrated in international product markets. Even if labour is internationally immobile, theincreased mobility of products and hence jobs implies a change in the distortions arising from taxes and socialsecurity contributions levied on labour income. Since financing of social security via general taxation involves acommon resource problem the effects of international integration depend critically on the institutional structureof the labour market. This paper shows that increased international integration inducing more product marketcompetition implies that it becomes more costly to maintain welfare systems financed by general taxation.

Keywords: product market integration, tax distortions, social security, public consumption

JEL Code: F10, J30, H20, H30

1. Introduction

The process of international integration is attracting widespread attention and it is widelyexpected to change economic structures fundamentally. One of the concerns—especiallyamong Northern European countries—is that it may erode the possibilities for maintainingan extended welfare state. A particular concern is whether it will become more difficult tomaintain a large public sector and an extended social security system, which is financed bygeneral taxation or social security contributions. Moreover, some observers are pointing tothe fact that international integration may lead to more volatility and needs for restructuring,which may increase the demand for the parts of public sector activities, which provide(explicit or implicit) social insurance (see e.g., Rodrik, 1997, 1998). The need for socialinsurance arrangements may thus increase at the same time as it becomes more difficultto finance the system. This paper addresses the second part of this problem, namely theproblem of maintaining welfare activities and social security systems financed by generaltaxation in the face of tighter international integration.1

The fact that international integration enhances the mobility of certain tax bases has beenwidely stressed as a major threat to the welfare state (see e.g., Tanzi, 2000). Internationalintegration makes it easier to move the tax base to minimize tax payments, which implies thattax revenues are eroding and it will accordingly be more difficult to finance the welfare stateas international integration proceeds further. However, while mobility of some tax bases

44 ANDERSEN

including in particular financial capital has increased significantly it remains the case thatlabour is fairly immobile—at least in Europe—see OECD (1999). Does this imply that thetax pressure on the welfare state is relieved? After all, capital income taxation contributesa fairly small proportion of total tax revenue in most European countries,2 and the bulkof social expenditures are financed by various forms of taxes levied on wage income. Thisseems to suggest that the mobility problem is not a major threat to the possibility of financingwelfare arrangements and social security by taxes or contributions levied on labour.3

However, this argument overlooks that increased mobility of goods and firms4 also impliesthat jobs become increasingly mobile, hence, even if workers are not very mobile, jobs are.This can in various ways affect labour market performance (see e.g., Andersen, Haldrup andSørensen, 2000) and since European countries have been suffering from structural problemsin the labour market, it is natural to address the question of how product market integrationmay affect wage formation and therefore in turn the consequences for welfare state activities.This paper considers this issue and focuses on the scope for maintaining strong universalor collective elements in welfare state activities. The universal principle prevails if allcitizens have a right to the welfare arrangements (eventually dependent on labour marketstatus), and they all contribute to the financing via general taxation (see Esping-Andersen(1990) on various welfare models). The present analysis considers public provision ofgoods and services as well as social security in the form of unemployment benefits tounemployed members of the labour force and transfers unrelated to labour market status(e.g., pensioners or other groups outside the labour force). Since most European countrieshave a strong universal element in the welfare arrangements, the issue of how it is affected byinternational integration is of direct policy relevance.5 The paper focusses on the interplaybetween labour market institutions and international product market integration. Since bothaspects are complex, it is only possible to study the interaction between the two by restrictingattention to basic mechanisms. However, they are of such a nature that the basic conclusionsof the analysis will apply in a more general setting.

To analyse this issue two fundamental aspects have to be considered. First, we need toclarify how welfare programmes (unemployment benefits) and taxation affect wage forma-tion and employment. Second, we have to address how these effects depend on the extentto which the economy is integrated in international product markets.

For the first part the paper builds on an extensive literature on how unemployment benefitsand taxes affect wage formation. Since the paper takes a macro or general equilibriumperspective on the problem, the institutional structure of the labour market becomes crucialin respect to the perceived connection between receipt of unemployment benefits or otherwelfare arrangements and payment in the form of taxes or social security contributions (seee.g., Atkinson, 1999). The problem is that the public budget becomes a common resource,which in turn may create a negative externality. An example of this is that the costs ofunemployment in the form of larger expenses on unemployment benefits with a universalsocial security model are borne by all participating groups. This crucial link was pointedout by Summers, Gruber and Vergara (1993) and has since been addressed in work on theimportance of the institutional structure in labour markets.6

The second part of the paper builds on a growing literature addressing how tighter productmarket integration and increased mobility of production may affect labour markets even iflabour is immobile (see e.g., Andersen, Haldrup and Sørensen (2000) for an introduction

WELFARE POLICIES, LABOUR TAXATION AND INTERNATIONAL INTEGRATION 45

and references). While the effects of international integration can work through many dif-ferent channels this paper presents a model formulation, which makes it possible to capturethe main qualitative effects in a manageable way which can be combined with the inter-relationship between taxation, social security systems and the institutional structure of thelabour market.

The paper is organized as follows: Section 2 develops a model for an open economy whichto different degrees can be integrated in international product markets. Section 3 outlinessome basic mechanisms concerning the interaction between taxation and social securityand labour market performance. Section 4 considers how increased international integrationchanges the effects of taxes and social security systems on labour market performance, andSection 5 addresses the welfare effects. Section 6 provides some concluding remarks, andall technical details have been placed in appendices.

2. A Two-Sector Model

Consider an economy with a continuum of sectors j ∈ [0, 1], which in turn is madeup of subsectors i ∈ [0, 1]. Each subsector produces a differentiated commodity of thesector-specific type of goods and there is one representative firm in each subsector. Somefirms i ∈ T ≡ [0, η] produce traded goods with prices determined in international productmarkets, i.e., p ji = p∗

j i ∀i ∈ T (where p∗j i is exogenous), and the remaining firms i ∈ N ≡

]η, 1] produce non-tradeables with prices p ji determined by the conditions in the domesticproduct market (cf below). An increase in η is taken as a proxy for increased internationalintegration since it captures that the competitive pressure in product markets becomes moreintensified.7

2.1. Households

Households supply labour from which they receive wage income if employed and unemploy-ment benefits if unemployed. Each household supplies a specific type of labour matchingthe labour requirements of one particular sector. Moreover, households own firms and areentitled to profits. Ex-ante there is uncertainty with respect to the labour market status of thehousehold (employed or unemployed). The utility of a representative household h ∈ [0, 1]is given as

Uh = ch − dlh + V (g)

This formulation captures the utility from consumption of the private consumption bundlech (see next page), the disutility of work lh and the utility achieved by access to the goodsand services (g) provided by the public sector (V ′ > 0, V ′′ < 0). Note that d capturingthe disutility of work has the interpretation of a reservation wage, i.e., for any consumptionreal wage above d the household supplies inelastically its working time (normalized tounity), which is a reasonable first approximation to the fact that labour supply elasticities

46 ANDERSEN

are usually found to be small. The budget constraint of the household reads

pch = Ih + πh + TRh

where TRh are lump-sum transfers from the government, πh profits and Ih labour incomegiven as

Ih = (1 − τ)w if employed= b if unemployed

where τ is the tax rate, and b the unemployment benefit (net of taxes).The consumption bundle ch is defined over commodities from the j sectors as

ch =[∫ 1

0c

θ−1θ

hj dj

] θθ−1

and the associated price index is given as

p ≡[∫ 1

0p1−θ

j dj

] 11−θ

The demand by household h of commodities from sector j is given as

chj =[

p j

p

]−θ

ch

The consumption bundle of goods from sector j is similarly defined over the sectorspecific goods as

chj =[∫ 1

0c

θ−1θ

hji di

] θθ−1

with the associated price index

p j ≡[∫ 1

0p1−θ

ji di

] 11−θ

and the demand by household h of commodity i in sector j is given as

chji =[

pji

p j

]−θ

chj

2.2. Government

The government provides public services/goods and social security in the form of transfersrelated to employment (unemployment benefits) and transfers unrelated to the activity level(lumpsum transfers to e.g., pensioners or others outside the labour force).


The public services g are produced by use of commodities from all sectors according to

g =[∫ 1

0g

θ−1θ

j dj

] θθ−1

and the associated price index is defined by p defined above, implying that public demandfor product variety j can be written

g j =[

p j

p

]−θ

g

Similar to the structure of private households, the composite good from sector j is madeup of the specific goods as

g j =[∫ 1

0g

θ−1θ

ji di

] θθ−1

and the associated price index is defined by p j given above, implying that public demandfor product variety i from sector j can be written

gji =[

pji

p j

]−θ

g j

Note that this way of specifying public consumption rules out relative demand shiftsbetween public and private consumption as a source of relative price changes.8 Since weare considering the incentive effects of welfare arrangements this assumption helps isolatingthe direct effects disregarding any relative price effects that may arise if the distribution ofincome affects aggregate demand.

The public sector budget constraint reads

pg + pTR + pB = τ I

where TR is total expenses to transfers, i.e.,

TR =∫

hTRh dh

B is the aggregate expenses on unemployment benefits, i.e.,

B = b(1 −∫ 1

0

∫ 1

0lji di dj)

and I is the aggregate labour income, i.e.,

I =∫ 1

0

∫ 1

0wjilji di dj

The tax is levied only on wage income partially to simplify the model and partially tocapture the empirical fact that general wage income taxation accounts for most of publicsector revenue, cf introduction. Moreover, this formulation allows us to interpret taxes associal security contributions.

48 ANDERSEN

In the following analysis of wage formation it is assumed that the real level of unem-ployment benefits (b), transfers (TR) and public consumption (g) are exogenous—unlessotherwise stated. Consequently, the wage tax rate (τ ) is endogenous adjusting so as tobalance the budget. Note, that the unemployment benefit is not taxable income, see below.

2.3. Firms

All firms are price and wage takers,9 and they produce subject to a technology linking outputto the employed amount of labour given by

yji = 1

α(lji)

α, where 0 < α < 1

which under the assumption of wage taking behaviour implies that labour demand reads

lji =(

wji

pji

)γ

, where γ ≡ (α − 1)−1 < 0

and output supply can be written

yji = 1

α

(wji

pji

)αγ

A key parameter in the following is the elasticity of labour demand wrt to the wage rate.There is a qualitative difference between the tradeable and non-tradeable sectors, since thelatter have the possibility of passing wage increases partially into prices while this is notpossible in the former case. Hence, the labour demand elasticity differs between tradeableand non tradeable firms and we have that

βi = γ for i = T

βi = γ (1 − εpw) = γ

( −θ

αγ − θ

)> γ for i = N

That is, labour demand is less elastic in non-tradeable firms as compared to tradeablefirms since in the former sectors wage increases are partly passed into prices while this isnot possible in the latter sector. In deriving these elasticities it is used that total demandfaced by firm i in sector j can be written

dji =[

pji

p j

]−θ [p j

p

]−θ

(c + g)

2.4. Wage Formation

The labour market is assumed to be imperfectly competitive. While there is no consensuson how to model imperfectly competitive labour markets it is commonly accepted that themonopoly union model in a simple way captures the qualitative implications of different


labour market models at least in respect to generate unemployment, and in the wage re-sponse to general income taxation and the degree of centralization (see e.g., Blanchard andFischer, 1989). Hence, this model is used here in the right-to-manage version to capturesome essential mechanisms in relation to wage formation, public policy and internationalintegration of product markets.

All workers are organized in unions, and workers in a given sector belong to the sameunion. A union sets wages for a fraction φ (0 < φ ≤ 1) of the sectors implying thatthere are 1/φ unions. Labour is mobile across firms in a given subsector, while not amongdifferent sectors and nations. Since sectors are symmetric there is no incentive for workersto move across sectors in equilibrium, and disregarding international mobility matches thecurrent European situation quite well (cf the introduction). The union is able to discriminateits wage policy between firms producing tradeables and firms producing non-tradeables.10

Since nominal issues are not crucial to the problems considered here, it is assumed thatwage setters can determine real wages,11 and the aggregate price level is normalized to one(p ≡ 1).

The union is assumed to be utilitarian and maximizes the expected utility of its mem-bers. Note that there is individual uncertainty wrt to labour market status—employed orunemployed—but there is no aggregate uncertainty. The former implies that an unemploy-ment insurance system has potential welfare benefits, and the latter implies that it is possibleto diversify the unemployment risk. The interesting question is how this affects incentiveswhen the unemployment insurance system is universal, that is, collectively financed.

Since a union represents workers in both tradeable and non-tradeable firms, it has to takethe interaction between the two sectors into account, and it is shown in Appendix A thatthe wages set for tradeable and non-tradeable firms, respectively can be written12

wi = βi (b + d)

(βi + 1)(1 − τ) − ετwi

for i = T, N

where the labour demand elasticity βi (< −1) is given above, and ετw is the effect whichunions perceive that a change in wage policy has on taxes. Note that the wage formula givenabove refers to a symmetric equilibrium, and wages are therefore identical across similarsectors implying that the sector index j can be dropped.

2.5. Equilibrium Conditions

The price of all non-tradeables are determined by the market clearing condition

yji = c ji + gji for all j, i ∈ N

and for tradeables the net export x is given as

xji = yji − (cji + gji) for all j, i ∈ T

Since the model is static, trade has to balance, i.e.,∫j

∫i∈T

pjixji dj di = 0

50 ANDERSEN

Given the wages set by unions, employment is determined by labour demand. We solvefor the symmetric13 equilibrium where pji = pN for i ∈ N , and pji = pT for i ∈ T , andw ji = w N for i ∈ N , and wji = wT for i ∈ T . For later reference define aggregate wages as

w =∫ η

0wi di +

∫ 1

η

wi di

= ηwT + (1 − η)w N

In the following we first lay out some basic mechanisms concerning the interaction be-tween wage formation, social security and taxation, and next we turn to an explicit analysisof the effects of tighter international integration.

3. Taxation and Wage Formation

A crucial aspect of collectively financed welfare arrangements is that they are universal(eventually conditioned on active participation in the labour market). While this may havethe politically attractive feature of offering equal opportunities or the economic advantageof allowing better diversification of risks, it also implies that there is a potential externalityin wage setting. For the tax payer there is no direct link between the taxes paid and thebenefits received, or in the present setting of non-atomistic wage setters between total taxespaid by employed members and what is received in the form of e.g., unemployment benefitsfor unemployed members. To put this differently, by its wage decision a given union affectsthe employment situation for its members which in turn affects the expenses of the socialsecurity system and therefore the tax rate levied on all wage earners to finance the system.This reflects an externality present in any collectively financed system (the problem ofcommon resources). Consider first how this affects wage formation.

In symmetric equilibrium (see appendix) the wages in tradeable and non-tradeable firmscan be written as

wT = mTd + (1 − φ)b

1 − τ(1 − φ)where mT ≡ βT

βT + 1

and

w N = m Nd + (1 − φ)b

1 − τ(1 − φ)where m N ≡ βN

βN + 1

Observe that these expressions are not closed form solutions since the tax rate τ is en-dogenous. However, it can be concluded that wages in non-tradeable firms are higher thanin tradeables firms,

w N > wT

reflecting that the competitive pressure is higher in tradeable firms and therefore the wage islower. It is well-known that the effect of unemployment benefits for wage formation dependscritically on whether unemployment benefits are taxed by the same rate as labour incomeor not (Pissarides, 1998). In the present setting unemployment benefits are not taxableincome, since this is a simple way of by-passing the question of whether unemployment


benefits will be changed if say public consumption or other variables are changed. Theassumption is that what is of interest to policy makers is the actual consumption possibilitiesof unemployed.

Considering how the wages respond to changes in welfare activities we find that forunemployment benefits there is both a direct effect in terms of raising the reservation wageof workers and an indirect effect in terms of raising taxes (see appendix). Both effects tendto increase wage demands and we have

∂wi

∂b≥ 0 for i = T, N

For public consumption and transfers unrelated to production/employment, wage formationis only affected via the implied increase in the tax rate, that is,

∂wi

∂z> 0 for i = T, N ; z = TR, g

These results capture the standard result that an increase in public sector activities maylead to a wage increase (see e.g., Alesina and Perotti, 1997).

Turning to the importance of the institutional structure we find that

∂wi

∂φ< 0 for i = T, N

The more comprehensive the union, the more the tax externality is internalized and thelower the wage demands. The reason is simple—the higher φ—the more the union takesinto account that higher wage demands will lead to higher unemployment and thereforehigher public expenses to transfer payments which in turn will have to be financed by taxeswhich lower the disposable income of employed workers.

Welfare policies like unemployment benefits will thus affect wage formation less themore centralized the labour market, since unions realize that the unemployment benefitsare going to its members, i.e.,

∂

∂φ

∂wi

∂b< 0

The same holds for the effects of transfers and public consumption. With fully centralizedwage setting (φ = 1) the negative externality in the social security system is fully internal-ized and the wage rate becomes independent of both unemployment benefits and the taxrate, i.e.,

∂wi

∂b= 0 for i = T, N , and φ = 1

The reason is that the system is actuarially fair at the aggregate level and thereforeessentially works as a way to redistribute income from employed to unemployed, sinceunions are assumed utilitarian, this will have no effects on the incentives underlying wageformation when wage setting is fully centralized. This suggests why countries with a strongelement of universal welfare policies also tend to be economies with highly centralizedlabour markets (see Summers, Gruber and Vergara, 1993; and Moene and Wallerstein,

52 ANDERSEN

1995). The general point here is that the effects of social security systems and taxationcannot be judged independently from the institutional structure of the labour market.

It is interesting to observe that the “tax” externality does not induce a hump-shaped wagecurve a la Calmfors and Driffill (1988). The reason is straightforward, the externality ismonotonously increasing in the degree of decentralization in the labour market.14

It is important to stress that when varying the degree of centralization it is assumed thatthere is no change in the political power relation in the economy. This is consistent with theassumption of the model that policies are decided by a benevolent utilitarian government,but it implies that the analysis disregards possible other effects of centralization in wageformation, some of which may have harmful effects to society.

Using that aggregate employment can be written

l = ηlT + (1 − η)lN

It follows straightforwardly that aggregate employment is decreasing in the level of un-employment benefits, transfers and public consumption, and increasing in the degree ofcentralization of labour market. These effects arise directly from the wage responses out-lined above.

4. International Product Market Integration

Consider now a process of international integration which via a reduction in various barriersto trade as well as implicit and explicit trade costs implies that a larger share of the sectorswill be exposed to international competition, that is, η increases. The immediate effect is areduction of aggregate wages, i.e.,

∂w

∂η< 0

This reflects that a larger fraction of economic activity takes place in firms/sectors inwhich the market power of unions is smaller because the goods produced can be tradedinternationally. As a consequence of this reduction in market power it also follows thatincreases in unemployment benefits and taxes to a lesser extent spillover into wage increases,i.e.,

∂(

∂w∂b

)∂η

< 0

and

∂(

∂w∂z

)∂η

< 0 z = TR, g

The reason is straightforward. With tighter international integration the unions face amore elastic labour demand relation. Accordingly, they are less aggressive in their wagedemands and increases in taxes and unemployment benefits are to a lesser extent shiftedinto wages. This points out that international integration may imply an “implicit” structuralreform of labour markets through the effects it has on union market power. However,


even though international integration in this way may reduce the distortionary effects ofunemployment benefits and taxation on wage formation it does not necessarily followthat the distortionary effects on employment are reduced. Despite the wage moderationthe employment consequences may become larger because tighter international integrationincreases the sensitivity of aggregate employment to wage costs.

To address this issue first note that international integration implies an increase in aggre-gate employment due to the reduced market power of unions, i.e.,

∂l

∂η> 0

Turning to the employment consequences of welfare policies we find that they are nu-merically larger the more integrated the economy is, since

∂

∂η

(∂l

∂z

)< 0 for z = b, g, TR

That is, even though the international integration works to lower the shifting of taxes andunemployment benefits into wages, the net effect is that the employment consequences are(numerically) larger, i.e., a given increase in tax rates or unemployment benefits have alarger distortionary effect on aggregate employment, the more integrated the economy is. Itis interesting to note that the distortionary employment effects are larger, the less centralizedthe wage formation process (φ low), that is, the more decentralized wage formation is, themore international integration increases the distortions on employment of welfare policies.

5. Welfare Implications

The welfare implications of taxation, public consumption and social security in the formof unemployment benefits and lump sum transfers can now easily be considered. Althoughthe market power of unions implies that employment is inefficiently low, i.e.,

∂�

∂li> 0 for i = T, N

the welfare consequences cannot directly be measured by the employment consequencessince aggregate welfare under a utilitarian welfare criterion is given as (see Appendix D)

� = ηpT yT + (1 − n) pN yN − d(ηlT + (1 − η)lN ) + V (g)

Since tighter international integration affects not only employment, it follows that thegeneral welfare effects cannot be inferred simply from observing that employment increases.It is, however, possible to show that the standard argument for aggregate welfare gains frominternational integration applies, since international integration unambiguously improveswelfare, i.e.,

∂�

∂η> 0

54 ANDERSEN

The reason is straightforward: international integration reduces the market power of unionsand therefore the distortion due to imperfectly competitive labour markets.

The interesting question, however, is how international integration affects welfare policies.To this end let us consider the three main categories in turn.

Consider first public consumption. Since the benefits of public consumption are not di-rectly affected by international integration it follows that the change in the distortions arisingfrom financing public consumption is the channel through which international integrationaffects the optimal level of public consumption, and we have

∂gopt

∂η< 0

The optimal level of public consumption is reduced when product markets are integrated,because the costs of financing public consumption go up (the tax distortions increase).

An increase in public lump sum transfers financed by income taxation reduces welfareand the consequences are larger the more open the economy, i.e.,

∂�

∂T< 0 and

∂

∂η

(∂�

∂T

)< 0

Similarly, unemployment benefits have a welfare cost which are larger the more integratedthe economy,

∂�

∂b< 0 and

∂

∂η

(∂�

∂b

)< 0

Note that these results only capture the cost side of unemployment benefits since theformulation adopted here does not address the welfare gains (reduction of inequality, betterinsurance) of unemployment benefits and other transfers. The reason being that the distribu-tion of income does not have any aggregate consequences under the specification adopted,and hence under the utilitarian welfare criteria it comes to play no role for aggregate welfare.This is obviously a special case, but the finding of increasing costs of providing transferseither to unemployed or unrelated to production/employment will have to be taken intoaccount also when explicitly considering the welfare benefits of such transfers.

6. Concluding Remarks

This paper has highlighted some important channels through which international productmarket integration affects a universal welfare model. Specifically, we have considered pro-vision of public consumption, transfers unrelated to labour market status and unemploymentbenefits. In all cases it was found that tighter international integration puts the public sectorunder pressure, not because the tax base as such is very mobile, it is not since labour isassumed immobile across nations, but because the distortionary consequences of taxationincrease when product markets become more integrated.

An important point of the present study was that the institutional structure underlyingwage formation is critical for the effects of taxes and unemployment benefits on labourmarket performance. This is due to the negative externality implied by welfare arrangements


financed by general taxation. Hence, the more centralized the labour market the more theexternality is internalized. Or to put it differently, it is less costly to maintain universalwelfare arrangements if labour markets are highly centralized. Empirical support for theimportance of this mechanism has recently been provided by Summers, Gruber and Vergara(1993), Alesina and Perotti (1997) and Daveri and Tabellini (2000).

The importance of the institutional structure may be of interest in its own right, sinceinternational integration may be expected to lead to more decentralized labour marketstructures (see e.g., Danthine and Hunt, 1994). Hence, there may be a “double pressure”on the public sector from international integration—directly by increasing the distortionaryeffects of public sector activities financed by wage taxation and indirectly by changinginstitutional structures in the direction of more decentralized labour market structures,which will tend to reinforce the direct effect. Even if it is questioned whether internationalintegration in itself leads to more decentralization in wage formation, it remains a factthat in many countries there has been a trend towards more decentralized wage settingarrangements in recent years (see Flanagan, 1999; Booth et al., 2001). The present analysissuggests that this may contribute to increase the costs of maintaining universal welfarepolicies.

An important premise for this study is that labour markets are characterized by structuralproblems due to rent-seeking behaviour. The possibility of exerting rents is larger in aless internationally integrated economy, and therefore integration works as an implicitlabour market reform reducing wages and increasing employment. Despite these effectsit was found that other consequences of structural labour market problems, namely, thedistortions arising from taxation are increased. It follows that international integration isnot a perfect substitute for structural labour market reforms, and therefore a need for labourmarket reforms remains relevant if rent-seeking behaviour is dominating labour marketperformance.

Privatization of social security is often proposed as a solution to the problems faced bypublicly financed welfare systems. While decentralization of social security surely reducesthe tax externality and therefore the distortionary effects, it is also the case that in termsof risk diversification it can offer less than a universal system which pools the risks ofall agents in the economy. It is an interesting topic for future research to clarify howinternational integration affects the trade-off between risk sharing and the distortionaryeffects of taxation.

Appendix A: Wage Setting

Consider the union which represents sectors j ∈ J where the set J constitutes a fraction φ

of all sectors. The expected utility of members which are employed in both tradeable andnon-tradeable firms can be written∫

J

[∫ η

0(lji((1 − τ)wT − d) + (1 − lji)b) di +

∫ 1

η

(lji((1 − τ)w N − d) + (1 − lji)b) di

]dj

Note that labour supply (exogenous time constraint) is one in all sectors. The union sets thewage for tradeable and non-tradeable firms taking into account the labour demand relation

56 ANDERSEN

and the effects its decisions have on the public budget. Since the problem is symmetric forall sectors the sector index is dropped on the wage. The first order condition for the wagedecision problem can be written

βi ((1 − τ)wi − d − b) + (1 − τ)wi − wi

liIJ

∂τ

∂wi= 0 for i = T, N

Denoting

εiτ ≡ wi

liIJ

∂τ

∂wi

yields the wage setting rule given in Section 2.4. In this expression IJ denotes total labourincome generated by the union, i.e.,

IJ =∫

J

[∫ η

0(ljiwT di +

∫ 1

η

ljiw N di

]d j

From the public sector budget constraint we find by using that φ I = IJ

IJ∂τ

∂wi= φli

(−bβi w−1i − τ(1 + βi )

)

Inserting we find that the wage relation can be written

wi = βi

βi + 1

d + (1 − φ)b

(1 − τ(1 − φ))for i = T, N

or

wi = mi w∗ for i = T, N

where

m j = β j

β j + 1for j = T, N

and

w∗ = d + (1 − φ)b

(1 − τ(1 − φ))

The interpretation being that the mark-up which differs across tradeable and non-tradablefirms determines how much the wage is increased beyond the base level w∗.

Appendix B: Wage Effects

It follows straightforwardly that

∂wi

∂b= βi

βi + 1

[(1 − φ)

(1 − τ(1 − φ))+ (d + (1 − φ)b)(1 − φ)

(1 − τ(1 − φ))2

∂τ

∂b

]for i = T, N


and

∂wi

∂z= wi

(1 − φ)

(1 − τ(1 − φ))

∂τ

∂zfor i = T, N and z = g, TR

It follows that

∂τ

∂z> 0 for z = b, g, TR

Since

∂w

∂z= η

∂wT

∂z+ (1 − η)

∂w N

∂zfor z = b, g, TR

we have

∂

∂η

∂w

∂z= ∂wT

∂z− ∂w N

∂z

The results in the text now follow directly by observing

∂wT

∂z<

∂w N

∂zfor z = b, g, TR

Finally note that

∂w

∂φ< 0

and

∂

∂φ

∂w

∂z< 0 for z = b, g, TR

Appendix C: Aggregate Prices

First note that by aggregating the private and public sector budget constraints we get

p(c + g) = ηpT yT + (1 − η)pN yN

The demand for any non-tradeable can be written

dN =(

pN

p

)−θ

p−1(ηpT yT + (1 − η)pN yN )

using that supply of non-tradeables equals demands, and that p = 1 we get(1 − (1 − η) p1−θ

N

)pN yN = ηp1−θ

N pT yT

Using that the supply of non-tradeables can be written

yN = 1

α

(w N

pN

)αγ

58 ANDERSEN

and that the supply of tradeables can be written

yT = 1

α

(wT

pT

)αγ

from which we get(1 − (1 − η) p1−θ

N

)ηp1−θ

N

m N

mT=

(mT

m N

pN

pT

)αγ−1

Since PT is exogenous, it follows that PN is independent of b and τ, and therefore g andTR. Note that the LHS of this expression is increasing in pN while the RHS is decreasingin pN . For pN = pT = 1 (the latter equality follows from p = 1) it follows that the LHSis less than the RHS from which it follows that pN > pT . It is easily verified that

∂pN

∂η> 0

Next to prove that

m N

pN>

mT

pT

implying that the product real wage is higher in non-tradeable firms than in tradeable firms,i.e.,

w N

pN>

wT

pT

This can be proved by contradiction. Assume m N /pN ≤ mT /pT in which case it followsthat the RHS is less than or equal to one, but the LHS is larger than one. Hence, a contradictionas it follows that m N /pN > mT /pT .

Finally, note that this implies that

lT > lN

Appendix D: Employment Effects

From the expression for aggregate employment we find that

∂l

∂z= η

∂lT

∂z+ (1 − η)

∂lN

∂z< 0 for z = b, g, TR

∂l

∂φ= η

∂lT

∂φ+ (1 − η)

∂lN

∂φ> 0

∂l

∂η= (lT − lN ) + (1 − η)

∂lN

∂pN

∂pN

∂η> 0

∂

∂η

(∂l

∂z

)= ∂lT

∂z− ∂lN

∂z+ (1 − η)

∂

∂η

(∂lN

∂z

)< 0 for z = b, g, TR


Appendix E : Welfare Implications

We have

ψ = c − dl + v(g)

where

c = (1 − τ)I + π + TR + b(1 − l)

τ I = g + b(1 − l) + TR

Hence,

c = I + π − g = ηpT yt + (1 − η)pn yn − g

Using that αT yT l−1T = wT

pTand αN yN l−1

N = w NpN

, we have

ψ = η (wT , pT ) + (1 − η) (w N , pN ) + v(g)

where

(wi , pi ) =(

1

αwi − d

)l

(wi

pi

)

and

w =l(

wipi

)wi

(1

αwi (1 + ε) − dε

)< 0

ww = 21

αl ′

(wi

pi

)+

(1

αwi − d

)l ′′w

(wi

pi

)< 0

p =(

1

αwi − d

)l ′w

(wi

pi

)(−wi

p2i

)> 0

It follows straightforwardly that

∂ψ

∂η= (wT , pT ) − (w N , pN ) + (1 − η) p

∂pN

∂η> 0

and

∂

∂η

(∂ψ

∂z

)= w (wT, pT )

∂wT

∂z− w (w N , pN )

∂w N

∂z− wpN (w N , pN )

∂pN

∂z

Since,

w (wT , pT )∂wT

∂z− w (w N , pN )

∂w N

∂z= ∂w∗

∂z

1

w∗ [wT w (wT ) − w N w (w N )] < 0

and

∂w w (w, p)

∂w= 1

α(1 + ε)2l − l ′w dε > 0

60 ANDERSEN

It follows that

∂

∂η

(∂w

∂z

)< 0 for z = b, g, TR

Optimal public consumption (tax-financed) is determined by the modified Samuelson ruleaccording to

MC(g) ≡ 1 +(

η (wT , pT )∂wT

∂τ+ (1 − η) (w N , pN )

∂w N

∂τ

)∂τ

∂g= v′(g)

The LHS gives the costs of public consumption as the sum of the direct resource cost plusthe distortion.

We have that

∂MC(g)

∂η=

( (wT , pT )

∂wT

∂τ− (w N , pN )

∂w N

∂τ

)∂τ

∂g+ (1 − η) p

∂pN

∂η

∂τ

∂g

+(

η (wT , pT )∂wT

∂τ+ (1 − η) (w N , pN )

∂w N

∂τ

)∂

∂η

(∂τ

∂g

)

From the public budget constraint we have

∂τ

∂g= τ

g[1 − τ(1 + ε)(ηlT + (1 − η)lN )]

Hence,

∂

∂η

(∂τ

∂g

)= −τ 2

g(1 + ε)(lT − lN ) − τ 2

g(1 + c)(1 − η)

∂lN

∂pN

∂pN

∂τ> 0

Accordingly,

∂MC(g)

∂η> 0

from which it follows that the optimal g is decreasing in η.

Acknowledgments

Comments and suggestions by anonymous referees as well as seminar participants at theUniversity of Copenhagen and at RECEP (Moscow) are gratefully acknowledged.

Notes

1. The issue of integration and risk is addressed in Andersen (2000a).2. In European Commission countries the revenues raised from taxation of corporations is less than 3% of GDP

in 1997, and taxation of capital income of various forms yields even less revenue, see EU (2000).3. Obviously, taxes levied on some forms of consumption may be threathened. However, if labour is immobile

this can be counteracted by changing from consumption to income taxation.4. Important examples include the pressure on indirect taxes due to border trading and e-commerce.


5. A distinction is often made between Scandinavian countries which are taken to be more universal in theirwelfare arrangements as opposed to the continental model which is taken to rely more on private insurancearrangements. However, the difference between the two systems is more nominal than real since the insurancemodel relies on mandatory insurance arrangements with a collective element in financing, i.e., contributions areusually a flat rate levied on some income measure. This does not preclude that there are substantial differencesacross countries in terms of e.g., replacements ratio and the distributional profile of the arrangements.

6. See e.g., Holden and Raaum (1991), Holmlund (1993) Jackman (1990). For surveys on the role of centralizationsee Calmfors (1983) and Moene, Wallerstein and Hoel (1993).

7. Allowing for explicit trade frictions makes it possible to endogeneously determine which goods becometradeables and which become non-tradeables, and also to address explicitly the consequences of reduced tradefrictions as the cause of tighter international integration, see e.g., Andersen (2000b). However, the qualitativeimplications are the same as those captured by the more simple way of modelling international integrationused here. See Hau (1998) for a similar approach to the modelling of integration of product markets.

8. See Andersen (2000a) for an analysis of the effects of relative demand shifts arising when public and privateconsumption are differently distributed over tradeables and non-tradeables.

9. Imperfect competition in the form of monopolistic competition could easily be introduced without changinganything qualitatively.

10. While this may seem restrictive, bargaining usually leaves some room for wage differences across firms.The assumption serves the technical purpose here of avoiding that sectoral differences disappear in equilib-rium. With equal wages, output prices would be the same, and thus employment independent of the sectoralcomposition of the economy.

11. In the case of nominal contracting this would amount to assuming that the nominal wage is fully indexed. Thisassumption rules out the price-externality effect which has been stressed as one mechanism through which thedegree of centralization in labour market may make a difference, see e.g., Calmfors and Driffill (1988) andAlesina and Perotti (1997).

12. Note that in the case where the effects of changed wages on the public budget are accomodated either bychanges in public consumption or changes in transfers we have ετw = 0 and we get the standard monopolywage formula appearing in the literature, i.e.,

w j = βi

1 + βi

(b + d)

(1 − τ)

in which case the wage is increasing in both b and τ .13. As is well-known due to the symmetry across sectors there will be no trade in equilibrium i.e., nx ji = 0 for

all i ∈ T . However, clearly incentives are affected by the possibility for trade.14. Alesina and Perotti (1997) conjecture a hump-shaped relation, but this is not supported by the present analysis.

See also Daveri and Tabellini (2000).

References

Alesina, A. and P. Perotti. (1997). “The Welfare State and Competitiveness.” American Economic Review 87,921–939.

Andersen, T. M. (2000a). “International Integration, Risk and the Welfare State.” Scandinavian Journal of Eco-nomics (to appear).

Andersen, T. M. (2000b). “Product Market Integration, Wage Dispersion and Unemployment.” Working Paper.Andersen, T. M., N. Haldrup and J. R. Sørensen. (2000). “Labour Market Implications of EU Product Market

Integration.” Economic Policy 30.Atkinson, A. B. (1999). “The Economic Consequences of Roling back the Welfare State.” Munich Lectures in

Economics. The MIT Press.Blanchard, O. and S. Fischer. (1989). Lectures on Macroeconomics. Cambridge: MIT Press.Booth, A., M. Burda, L. Calmfors, D. Checchi, R. Naylor and J. Visser. (2001). “The Future of Collective Bargaining

in Europe.” In T. Boeri, A. Brugiavini, and L. Calmfors (eds.), The Role of Unions in the Twenty-First Century.Oxford University Press.

Calmfors, L. (1993). “Centralisation of Wage Bargaining and Macroeconomic Performance—A Survey.” OECDEconomic Studies 21, 161–191.

62 ANDERSEN

Calmfors, L. and J. Driffill. (1988). “Bargaining Structure, Corporatism and Macroeconomic Performance.”Economic Policy.

Danthine, J.-P. and J. Hunt. (1994). “Wage Bargaining Structure, Employment and Economic Integration.”Economic Journal 104, 528–541.

Daveri, F. and G. Tabellini. (2000). “Unemployment, Growth and Taxation in Industrial Countries.” EconomicPolicy 30.

Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism. Princeton: Princeton University Press.European Commission. (1999). “The EU Economic Review.” European Economy 69.European Commission. (2000). “Structures of the Taxation Systems in the European Union: 1970–1997.” Eurostat.Flanagan, R. J. (1999). “Macroeconomic Performance and Collective Bargaining: An International Perspective.”

Journal of Economic Literature 37, 1150–1175.Hau, H. (1998). “Openness and Real Exchange Volatility: Theory and Evidence.” Mimeo.Holden, S. and O. Raaum. (1991). “Wage Moderation and Union Structure.” Oxford Economic Papers 43, 409–423.Holmlund, B. (1993). “Wage Setting in Private and Public Sectors in a Model with Endogenous Government

Behavior.” European Journal of Political Economy 9, 149–162.Holmlund, B. (1998). “Unemployment Insurance in Theory and Practice.” Scandinavian Journal of Economics

100, 113–114.Jackman, R. (1990). “Centralized and Decentralized Pay Determination in Nordic Countries.” In L. Calmfors (ed.),

Wage Formation and Macroeconomic Policy in the Nordic Countries. Oxford University Press.Moene, K. O., M. Wallerstein and M. Hoel. (1993). “Bargaining Structure and Economics Performance.” In R.J.,

Flanagan, et al. (eds.), Trade Union Behaviour, Pay Bargaining and Economic Performance. Clarendon PressOxford.

Moene, K. O and M. Wallerstein. (1995). “How Social Democracy Worked: Labor Market Institutions.” Politicsand Society 23, 185–211.

OECD. (1999). Policy Briefs—Open Markets Matter: The Benefits of Trade and Investment Liberalization, Oc-tober, OECD (Paris).

Pissarides, C. A. (1998). “The Impact of Employment Tax Cuts on Unemployment and Wages; The Role ofUnemloyment Benefits and Tax Structure.” European Economic Review 42, 155–184.

Summers, L., J. Gruber and R. Vergara. (1993). “Taxation and the Structure of Labor Markets: The Case ofCorporatism.” Quarterly Journal of Economics, 385–411.

Tanzi, V. (2000). “Globalization and the Future of Social Protection.” IMF Working Paper WO/00/12.

welfare policies, labour taxation and international integration

Documents