How Much Do Tax Distortions Restrict Employment and Output Growth? *

Transcription

1 Ho Much Do Tax Distortions Restrict Emloyment and Outut Groth? * Tax and Structural Reforms in the Euro Area Nikola Bokan University of St Andres Andre Hughes Hallett Vanderbilt University and CEPR Abstract: It is idely acceted that in order to imrove the economic osition of the EU, certain structural or market reforms need to be undertaken. Hoever fe reforms have been undertaken. The reason is that those reforms turn out to be costly in terms of erformance and the cost of financing them at least in the short term. Blanchard and Giavazzi (2003) develo a model based on imerfect cometition in both roduct and labour markets in order to sho the imact of labour market deregulation. Hoever they do not consider ho to finance such reforms or ho to overcome the short run costs. We extend their model to incororate those features, and to include distortionary taxation not only the most likely candidate for reform, but also the most robable instrument for financing the restructuring rocess itself. We establish formally that reforms imly significant short run costs as ell as long run gains; and that the financing of such reforms ill be the main stumbling block. We also come to a number of conclusions hich reverse the Blanchard-Giavazzi results; the comosition of the reform ackage and the distribution of the tax burden both matter. This model therefore sulies ne results on the design and sequencing of reforms, and on hich are the most effective reform instruments in given circumstances. We aly these results to the 24 OECD countries to sho the extent of the distortions, and the elfare gains (and costs) from removing them, in ractice. Keyords: Structural Reform, Wage Bargains, Short vs. Long Run Substitutability JEL Classifications: J58, H23, E24 First Draft, Aril 2006 * We thank Svend Hougaard Jensen, Charles Nolan, Alan Sutherland, and articiants at the Royal Economic Society and a joint CEBR-CEPR conference for helful comments. Financial suort from CEBR at the Coenhagen Business School is gratefully acknoledged. The usual disclaimer alies. University of St Andres and CEBR, Centre for Dynamic Macroeconomic Analysis, School of Economics and Finance, Castlecliffe, St Andres, Fife KY16 9AL, UK. Tel: E- mail: nb85@st-andres.ac.uk Vanderbilt University and CEPR, Deartment of Economics, Box 1819 Station B, Nashville, TN 37235, USA. Tel: a.hugheshallett@vanderbilt.edu

2 1. Introduction In this aer e consider a small theoretical model of age bargaining ith imerfect cometition in the roduct markets and distortionary taxation, in order to understand the likely incentives for, and eventual outcomes of, structural reform. We use the results to exlain olicy makers behaviour and dra certain conclusions about the aroriate design and sequencing of reform ackages. Much of the reform debate has focussed on the need for reforms in the labour market. That, it seems, is based on the analytic and emirical evidence of a negative link beteen economic erformance and age (or unit labour cost) rigidities in many countries (Bruno, 1986). Such a link is observable in the labour and roduct markets of Euroe (Koedijk and Kremers, 1996), here erformance has been measured in rates of groth and emloyment; and market deregulation in cometition olicy, merger codes, and the liberalisation of emloyment ractices. The result has been high costs, taxes and unemloyment; large budget deficits; and sluggish groth. But hile there is little disagreement that economic erformance and market structures are related, the nature of that relationshi and ho the conflicts ithin it can best be resolved is not ell understood. Added to that are the evident costs and inefficiencies caused by high taxes, but visible concerns that tax reductions ill just lead to unfair cometition (fiscal duming) and a race to the bottom. Similarly, the failure to embrace the otential of the ne service industries or a free market in cororate control may be rooted in the conflicts inherent in market liberalisation measures. We examine three roositions in articular. First, e examine the roosition that structural reforms may be hindered by the fact that they involve large costs u front, and tyically only bring benefits in the longer term. Politically sensitive olicy makers may orry that the short term costs outeigh the long term benefits esecially if the latter are uncertain. We also ask if there is a link beteen the need for fiscal disciline and the difficulty of undertaking structural reforms, hich could lead to those reforms not being undertaken. Second, e examine hether the comosition of the reform ackage matters. Secifically hether the distribution of the tax burden matters; hether taxes on earnings matter more than taxes on ayrolls or cororate rofits; and hether the degree of age bargaining oer or roduct market liberalization matters more than either. Third, e examine ho large these tax and market distortion effects are in ractice, hich highlights the contrast beteen the core Euroean economies and the smaller or the more market orientated ones outside. Finally e sho the oer of tax reform and deregulation, in terms of long term elfare gains, but also the high short run costs in the more rigid economies. The urose of this aer is therefore to advance beyond the analysis of hy reforms have not taken lace, the concern of most of the existing literature, to the roblem of defining hich kind of reform ould be the most effective in given circumstances. 2. The Model In earlier ork, Bokan and Hughes Hallett (2006), e modelled the imact of tax distortions on rices and age bargaining behaviour by extending the Blanchard and Giavazzi (2003) frameork to include distortionary taxation. This alloed us to identify the need for, and likely consequences of, structural reforms in any economy.

3 In addition to tax distortions, e imose to further asects of imerfect cometition in order to generate the need for roduct and labour market reforms. The first arises from imerfectly cometitive roduct markets. In this case, e assume the resence of certain number of the monoolistically cometitive firms each of them roducing a differentiated good. Then, in the labour market, e introduce structural rigidities by assuming a formal age bargaining rocess beteen firms and their orkers. The resence of monoolistically cometitive firms leads to the creation of rents in the economy, the size of hich is determined by the degree of imerfect cometition. At the same time, the existence of a age bargaining rocess leads to a certain distribution of those rents beteen firms and orkers. But distortionary taxation is necessary to comlete the story since, hatever else taxes may do, any reform rogramme needs to be financed. We do not model the dynamics exlicitly in this frameork. But, in order to allo for differences in the effects of reform over time, e follo Blanchard and Giavazzi by imosing a clear cut distinction beteen the short term and the long term. This is achieved by fixing the number of roducers in the market exogenously in the short run, hereas in the long run e allo that number to be determined endogenously by a market entry condition. One can think of this as a er unit entry cost, c, reresenting (or aroximating) certain regulatory or administrative entry barriers resent in the roduct markets. Although there ould be no difference in the outcomes if this cost ere treated as a shado cost, it is erhas better to think of it as real cost hich is roortional to outut. If it is a real cost, firms resent in the market can only earn ure rofits in the short run; in the long run those rofits ould be dissiated in the entry cost. 2.1 The Consumers Problem To model consumtion, e assume that the economy contains a fixed number of orkers-consumers L, indexed by j, ho can choose to either ork, or not to ork. If the orker decides to ork he must suly one unit of labour. If he does not ork he is unemloyed. Labour is therefore indivisible. The utility function for orker j is given by folloing exression j m 1 δ i= 1 δ 1 δ i, j δ δ 1 U = [ m C ] (1) here C i, j reresents individual j s consumtion of the i -th roduct; m reresents number of firms or roducts resent in the market; and δ stands for elasticity of substitution beteen roducts (firms) hich is defined as δ = δ f ( m),.we assume this elasticity to be an increasing function of number of roducts ith f '( m) > o and that δ may be fixed by olicy. This is crucial for disentangling the difference beteen the short and the long run since, by imosing an exogenous number of firms resent in the market, e assume that the elasticity of substitution is constant and exogenous in the short term. But in the long run, it ill be endogenous and determined by the number of roducts or firms that emerge in the long run equilibrium. This secification has to imortant features. First, assuming all the orkers are

4 identical, the utility of the orkers ill not deend directly on the number of roducts, but on the level of aggregate consumtion instead. Second, an increase in the number of roducts increases the elasticity of consumtion beteen them and thereby reduces monooly oer of the individual roducer. This may have indirect consequences for the utility of the individual orker. When making consumtion or labour market decisions, each orker maximises (1) subject to the folloing budget constraint: m PC i i, j = (1 t) jn j + Pr( u)[1 N j] (2) i= 1 here N j takes the value of one if orker j chooses to ork or zero if he or she is unemloyed. r ( u ) can then be interreted as the unemloyment benefit received from government in the case of unemloyment, or equivalently the orker s reservation age. We assume that this social suort, unemloyment benefit or reservation age is decreasing function of the unemloyment rate in the economy as a ' hole: r ( u) < 0. In other ords, the higher the unemloyment rate in the economy, the higher the ressure on the government to finance those benefits; or equivalently the higher the unemloyment, the more illing are the orkers to accet loer age and hence a loer reservation age. Finally P stands for the rice aggregator obtained after solving consumer s otimization roblem. It is given by: m δ 1 δ P= [ Pi ] (3) m i= 1 This exression is slightly different from the standard Dixit-Stiglitz aggregator as a consequence of the assumed form of the utility function at (1). Solving the consumer s otimization roblem, and using the fact that the roblem is symmetric across all consumers, e can obtain an exression for the consumtion that ould maximise utility for the individual consumer. It is given by j [(1 t) r( u)] N j + r( u) (4) P This exression is also roortional to the eriod maximum utility level, and may be used to make elfare comarisons. 2.2 The Firms Problem We assume that each firm roduces a differentiated roduct indexed by i using the same roduction technology hich is linear in labour. Outut is therefore given by 1 Y i = N (5) i Since both individual and aggregate demands are determined by the consumer s 1 Alternatively one can think of fixed coefficient roduction technology in hich caital is assumed to be fixed and normalized to one. Another version of our results starting from a Cobb-Douglas function is available and is set out in Bokan and Hughes Hallett (2006). But the final results don t change.

5 otimization roblem, the firms roblem consists of determining the rices taking demand as given. This allos us to obtain the artial equilibrium demand function for each roduct market. It is given by: Y i Y Pi δ = ( ) (6) m P 2.3 Wage Bargaining and Governments Before describing age bargaining roblem, e need to introduce the tax system. We assume first that both orkers and roducers are obliged to ay certain taxes. Workers need to ay a tax on the ages they earn. In our model, it is assumed that a common average tax rate ill be imosed on every orking orker s age. We also assume that unemloyment benefits are not taxed. Next, roducers need to ay ayroll taxes 2 defined as a certain fixed ercentage of the orkers gross age. Both of these taxes are assumed to be flat taxes. Extensions to a rogressive tax system are ossible, but lead to extremely comlicated exressions hich effectively deny any insight into the scoe for reform. Our flat tax secification meanhile imlies the folloing government budget constraint B = ( t + t ) P ( u)[1 N ] (7) j r j We treat B as being constrained by a ceiling on government debt. That means any increases in exenditures, or reductions in tax rates, must be matched by increases in tax revenues elsehere in the system. This is just an artificial device hich allos us to focus on the cost of financing any reforms. Hoever, deficits do have to be financed by interest ayments or tax revenues. So B ill alays be limited in ractice. Meanhile each firm bargains ith L/m orkers over ages and emloyment in that industry, in both the short and the long run. Intuitively, a fraction L/m of the orkers form a union. That union then bargains ith the firm over ages and the level of emloyment. Indivisibility of labour imlies that orkers can either be emloyed in the firm or be unemloyed. In hat follos, e consider a orld of Nash efficient bargaining solutions. There are to reasons for this. First, the efficient bargaining concet allos ages to be bargained off the labour demand curve, hich imlies that an increase in ages could be achieved ithout an immediate decrease in emloyment. And secondly, some emirical studies (Dobbelaere, 2004) have rejected The Right to Manage Model in favour of an efficient bargaining model as the aroriate exlanation of age bargaining in many Euroean countries. Since the case for structural reform is articularly strong in Euroe, it is imortant to have a model that catures that feature. Assuming risk neutrality for the unions, the age bargaining roblem can be ritten: max{ β log[(1 t ) P ( u)] N + (1 β )log[ P (1 + t ) ] N } (8) i, Ni i r i i i i 2 Or hiring and firing costs; or any rofit or cororate taxes that vary in line ith roduction costs.

6 here β reresents an exogenously determined index of union bargaining oer; and here t and t reresent the average tax rates aid by emloyees and emloyers resectively (0 t, t <1). This formulation imlies that unions ant to maximize the net age surlus from emloyment, the first term ithin the brackets, hile firms ant to maximize their net rofit reresented by the second term. Several imortant consequences of market regulation no follo. On the roduct market side e have c and δ =δ f(m). Reductions in the entry cost, c, can be thought as the removal of administrative restrictions; or the relacement of some state oned monoolies by market firms. The degree of roduct substitutability in the markets is broken into to arts. First, a olicy comonent (δ ) hose increase could reresent some market liberalisation measure, or a reduction in some domestic/external trade barrier hich has the effect of increasing roduct substitutability. These are matters hich lie under government control. The second element, f(m), is an index of market com-etition hich increases ith the number of firms. If e change δ by olicy, e changeδ. But m may then change. So, in ractice e seak of a net change to δ. Finally, on the labour market side e have β reresenting bargaining oer hose increase can be interreted as the increase in the degree of the orkers oer over age and emloyment decisions ranging from rights to strike, emloyment rotection legislation, severance conditions, firing costs, or other collective matters. In addition both tyes of taxes reresent regulatory instruments under direct government control. 3. Solving for equilibrium outcomes In order to roceed, e solve the model in three stes. First e solve for short run artial equilibrium values for relative rices and real ages. These ill be used to obtain the short run general equilibrium rices and ages. After obtaining those values, e can solve for the corresonding long run equilibrium values. 3.1 Short run artial equilibrium relationshis Equilibrium demand for each roduct, and hence emloyment, ill be determined by (6). Since orkers and firms bargain over both ages and emloyment, and since emloyment is already determined as a function of outut, our bargaining roblem can be resolved by substituting (6) into (8) and then alloing orkers and firms to bargain over ages and rices. First order conditions for relative rices and real ages are given by Pi P δ + β 1 = r ( u) δ (1 + t ) and (9)

7 i β P i 1 β = + r ( u) P 1 t + P 1 t (10) Using the exressions above, e can solve for short run artial equilibrium real ages and relative rices as functions of the regulatory arameters in the model. In fact: Pi P = [1 + µ ] ( u) and (11) r 1 + βµ (1 t) β( t + t) + t i = r ( u) P (1 + t)(1 t) (12) here µ reresents a relative rice mark-u in a broad sense, defined as δ ( t + t) 1 µ = + ( δ 1)(1 t) δ 1. (13) It is easy to sho that this mark-u is increasing function of both taxes on ages aid by emloyees, and the ayroll tax aid by emloyers, hen δ >1. That is, µ δ = > 0 t ( δ 1)(1 t ) and µ δ (1 + t ) = t ( δ 1)(1 t ) 2 >0 (14) hen δ >1. This result is to be exected since, in the case of increases in ayroll taxes, it is otimal for roducers to bargain for higher rices; hereas in the case of an increase in the taxes aid by emloyees, the latter ill demand higher ages. Hoever, that ould lead roducers to require an even higher mark-u in order to revent rofits from changing too much their ability to do so being limited only by the degree of inter-roduct substitutability. These results also sho that µ reresents a mark-u in relative rices, reflecting the combined rents to the firm and the derived rents to the ork force. µ may therefore be negative if relative rices are belo average for good i. In fact, µ > ( t + t) /(1 t) for δ 1; and it is a decreasing function of δ hich reaches its minimum value at θ = ( t + t ) /(1 t ) hen δ (a minimum hich increases ith t andt ). Hence e can think of µ ( t + t ) /(1 t ) as the degree of market distortion due to imerfect cometition; and ( t + t ) /(1 t ) as the degree of tax distortion due to the tax regime. Thus there ill therefore alays be some distortion, even under erfect cometition, so long as there are taxes. Hence, µ<0 is ossible if δ<1. Hoever, that imlies a mark-don on irreducible inut costs hich is clearly unsustainable in the long term. Firms ill either go out of business; or restructure to enter ne markets for goods that are more easily substitutable for those roduced elsehere. And as that haens, the average level of substitutability, δ, ill necessarily rise again. Corollary 1, section 4, shos that this rocess ill go on until no more firms exit and δ 1 (more firms enter than leave).

8 We therefore need to make a clear distinction beteen the short run here δ<1 might be ossible; and the longer run here, as a result of restructuring and market entry, it is not. For the uroses of this aer, e imose δ 1 hereafter. 3.2 Short run general equilibrium Since in a symmetric equilibrium all roducers need to charge the same rice, and since not all of them can have relative rices larger than one in a general equilibrium, all relative rices must be equal to one in the general equilibrium setting. Substituting that into (11) rovides us ith the folloing condition for the reservation age: 1 r ( u) = (15) 1 + µ Taking tax rates as temorarily fixed, this exression imlicitly determines the short run unemloyment rate hich is a consequence of the assumed fixed short run coefficient of the elasticity of substitution. Substituting (15) into (12) e obtain a solution for the short run general equilibrium real age: 1 + βµ (1 t ) β( t + t ) + t i = P (1 + t )(1 t )(1 + µ ) (16) 3.3 Comarative Statics in the Short Run Proosition 1: Short run real ages are an increasing function of labour s bargaining oer if and only if the mark-u, broadly defined, is greater than the share of the total tax burden on the er unit net age received by emloyees: or, equivalently, if the folloing condition (market distortions exist) is satisfied hen δ>1: t + t µ > (17) 1 t Proof: The first derivative of short run equilibrium real age is ositive if (17) holds, since i µ (1 t) t P t = >0 (18) β (1 t )(1 + t )(1 + µ ) hen δ>1 so that µ>0. But the result also holds if δ<1, since µ<-1 hen 0<δ 1. In contrast to Blanchard and Giavazzi (2003) ho claim that short run equilibrium real ages are an increasing function of bargaining oer, e have shon that their result actually deends on the relationshi beteen the existing tax rates in the economy and the value of the mark-u (i.e. on the imact of market oer vs. the strength of the tax distortions). If the tax system is such that the mark-u is larger than the share of total tax receits in the net age received by emloyees, then the Blanchard and Giavazzi result does hold. But if not, their result is reversed.

9 On the other hand, taking the first derivative of the short run equilibrium real age ith resect to that mark-u, shos that short run real age is alays a decreasing function of the mark-u and hence an increasing function of the degree of substitution in the roduct markets. This result as obtained by Blanchard and Giavazzi as ell. It is consequence of to oosite effects. The first is a artial equilibrium effect, in hich an increase in the mark-u leads to higher rents to the firms. That in turn leads to an increase in the orker s nominal age since they receive a share (β) of those rents. The second is a general equilibrium effect channelled through rices. An increase in the mark-u leads to an increase in rices hich reresents a loss to the orkers since they can buy less of a more exensive roduct. Since the magnitude of the latter effect is greater than the former, the real age received by labour falls. Next e consider the consequences of a change in the to tyes of taxes on short run real ages and on the reservation age. Proosition 2: The short run equilibrium real age is alays a strictly decreasing function of ayroll taxes, hereas it is unaffected by changes in age taxes. Proof: Substitute the broad mark-u, (13), into the solution for short run equilibrium real ages, and take first order derivatives ith resect to t and t. The intuition behind this conclusion comes from the effect of tax changes on the mark-u. Evidently the mark-u is less resonsive to changes in the ayroll tax than it is to changes in taxes aid on ages (see (14); t <1).Thus, in the case of an increase in ayroll taxes, real ages must fall because firms can alays increase their mark-u by enough to more than comensate for the increase in the ayroll tax: see again (14). The burden is therefore artly transferred to the orkers. But if there is an increase in age taxes, orkers ill demand higher ages. Nevertheless, firms are able to comensate for this increase by raising their mark-u by more than they could have done in the case of a ayroll tax. That results in an increase in the general rice level such that real ages remain unaffected. Proosition 3: The short run equilibrium reservation age is alays a strictly decreasing function of both tyes of taxes. Proof: (14) and (15) together imly the result. This result is also intuitive since the equilibrium reservation age is inversely related to the mark-u, and the mark-u is increasing in both tyes of taxes. Proosition 3 therefore imlies that the equilibrium unemloyment rate ill increase ith increases in both tyes of taxes. But it is imortant to note that the size of the effect on the reservation age, and hence on the unemloyment rate, differs deending on hich tax rate has been changed. A change in the tax on ages has larger effect on the equilibrium unemloyment rate than a change in the ayroll taxes: (14) imlies µ / t > µ / t in (15).

10 4. Long run equilibrium relationshis 4.1 Entry and Exit In the long run, firms can restructure or enter ne markets. We assume that firms need to ay a fixed entry cost hich is a roortion of (the rice of) their outut. This means that firms ill enter the market so long as rents cover those entry costs. Since firms get a share (1 β ) of the total rents from hich taxes need to be aid, e can define the share of net rents available to cover entry cost as follos: ( r 1 β )[1 (1 + t ) (u)] (19) Using (15) e can no exress the maximum accetable entry cost as a function of the mark-u, bargaining oer and taxes. It is given by µ t c = (1 β ) 1+ µ (20) Hoever the mark-u itself is no longer exogenous since the elasticity of substitution coefficient ill change because the number of firms, and hence the number and varieties of goods, changes hen firms enter and exit the market. In fact, the number of firms and the degree of substitution beteen goods ill adjust through entry and exit until the rents, (19), are fully consumed by the entry costs (20). In other ords the number of firms, and thereby the degree of cometition, must be such as to totally dissiate any excess rofits/rents over entry cost. Recall that this imlies δ 1, since otherise there are no rofits to be dissiated and no firms ill survive. Hence: Proosition 4: The number of firms, goods and emloyment ill eventually rise if tax rates of either tye are increased; or if market regulation measures loer the degree of substitutability (the degree of cometition) beteen goods and roducers. Proof: The first derivative of the maximum accetable entry cost is ositive, c (1 + t ) = (1 β ) µ (1 + µ ) 2 (21). Combining (21) ith (14), or ith µ / δ <0 from (13), gives the result. It is imortant to see hat is going on here in case roosition 4 aears counterintuitive. Increasing tax rates of either tye increases the mark-u that firms can imose, and hence the costs (and rents) they are reared to ay in order to enter the market. Moreover, that mark-u ill have increased by more than the original increase in tax rates: that follos from (14). Hence, the number of firms and degree of cometition has to fall in the short term, although rofits and rents ill rise. That s hat (21) imlies. But if rents have risen, ne firms ill enter the market and, in the longer term, the number of firms, goods and emloyment ill rise again. Hence roosition 4 and (21) together imly:

11 Corollary 1: More firms (goods, emloyment) enter the market in the long term than leave in the short term. Proof: The changes in the short term mark-us, µ j, are given by (14); and the subsequent (long term) adjustments by the artial derivatives from (21), (23) belo, and (13), once the ne degree of substitutability has been established. Putting these together, the total change is µ δ c dµ = µ jdj δ c µ j = t,t, δ (22) here the second term on the right reresents long term changes. But, using (21), (23) and (13), the square bracket is negative if 2t t < 0. That alays holds, irresective of δ, so long as both taxes are resent. Given (14), that result confirms roosition 4. Corollary 2: A olicy of reducing age taxes ill be more effective than reducing ayroll taxes for increasing the number of firms, goods or emloyment. But a olicy of market liberalisation (deregulation) that raises the level of cometition beteen roducers ill be more effective than either at lo levels of cometition (defined by δ(δ-1)<1- t, but less effective if cometition or taxation are already high. Proof: Cometition, and the number of goods and firms all increase if the alloable level of entry costs increases. By (21), that requires the mark-u µ to rise. The result no follos by comaring the artial derivatives in (14) ith each other, and ith µ / δ <0 from (13). Note that (21) imlies that the number of firms increases ith the entry costs they are reared to ay in order to enter a ne market, and ith the ease ith hich their goods can be substituted for others (δ). Notice also that emloyment increases because (u) / c <0 follos from (24) belo. r Finally, by substituting (13) into (20) and rearranging, e can solve for the long run elasticity of substitution as a function of the regulatory arameters. That solution is: δ (1 β )(1 t ) = (23) c (1 β ) t Hence, using (23) and (13) in (15) and (16), e can also solve for the long run reservation age and the long run real age. Their equilibrium values are given by: and 1 c β r ( u) = (1 β )(1 + t ) i P 1 c βt = (1 + t )(1 t ) (24) (25) The introduction of taxation in this model has therefore increased the comlexity of the solution, but it is straightforard to see the effects of the regulatory arameters on the equilibrium reservation age, real ages and emloyment.

12 4.2 Comarative Statics in the Long Run Proosition 5: Long run equilibrium reservation age (unemloyment rate) is alays a decreasing (increasing) function of labour s bargaining oer. Proof: The first derivative of r (u) ith resect to labour s bargaining oer is alays negative: c < 0. (26) 1 β )(1 + t ) ( Proostion 6: It is also easy to see that the long run equilibrium real age is alays a decreasing function of bargaining oer. Proof: Taking first order derivatives in (25), e obtain t /[(1 + t )(1 t )] hich is also negative. This result also differs from the standard result in the literature; as ell as from the Blanchard and Giavazzi (2003) version in hich long run equilibrium real ages are not affected by changes in bargaining oer. The question is hy does the introduction of taxation into this economy roduce such different redictions? To exlain roositions 5 and 6, consider a ermanent increase in labour s bargaining oer. In the short run, this leads to a rise in real ages since the share of the rofits going to the orkers ill have increased. But that means the rofits available to firms ill be reduced and it ill be harder to satisfy the requirement imosed by the entry condition the more so, the greater is β. Therefore the number of the firms resent in the market ill decrease. A decrease in the number of firms imlies a decrease in the elasticity of substitution faced by the remaining firms. That means that firms ill be able to charge higher rices. Workers ill demand higher ages to comensate. But, because firms have market oer (and because taxation increases the mark-u that this imlies; and because the tax edge increases the nominal age claim orkers have to make in order to reserve their take home ay), these age increases ill be assed on in rice increases. That leads to reduction in the real age finally received by the orkers. If taxation ere to go to zero, this effect ould vanish as (25) ould be indeendent of β. Similarly, it ould also vanish even if markets ere to become fully cometitive since δ imlies c (1-β) t in (23), hich makes i / P indeendent of β in (25). Hence either distortionary taxation or imerfect cometition, or both, is resonsible for this result. Finally e consider the effects in the change of taxes on reservation and real ages. Proosition 7: The long run reservation age is not affected by changes in the taxes aid by emloyees, but is a decreasing function of the taxes aid by emloyers. By contrast, the long run equilibrium age is an increasing function of the taxes aid by emloyers and a decreasing function of the taxes aid by emloyers. Proof: The first derivative of r (u) ith resect to t is zero, and ith resect to t is

13 1 c β (1 β )(1 + ) t 2 (27) hich is negative so long as c+β <1. Similarly the first derivative of the long run real age ith resect to t is (1 + t 1 c β )(1 t hereas the first derivative ith resect to t is given by 1 c βt 2 (1 + t ) (1 t ) 2 ) (28), (29). Of these to exressions, the first is alays ositive and the second alays negative so long as c+β <1. Hoever, it is easy to check that c+β <1 alays holds if δ 1 (imlying µ 0) since t Policy Imlications: 5.1 Short run costs, long run gains. Based on the model develoed above, e can dra some useful olicy conclusions. Consider first a simle scenario in hich the government ants to decrease the taxes aid by emloyers. Let us also assume that the government either ants to kee the budget balanced, or needs to kee it ithin some strict uer bound such as the 3% of GDP limit imosed by the Eurozone s Stability and Groth Pact. Wage taxes ill have to rise therefore. What are the short and long run effects of this olicy? Are reform measures of this kind really more effective for groth and emloyment than a reduction in age taxes? Or ould deregulation in either the roduct or labour markets roduce a better result? To anser questions of this kind, e resent an examle and then to sets of olicy imlications hich use the results in sections 3 and 4. They highlight one imortant area of economic reform: loering the burden of fiscal olicy and tax distortions. According to Proosition 2, the short run increase in the taxes aid by emloyees needed to kee the budget in balance ill not affect real ages, hereas the lanned reduction in ayroll taxes ould lead to an increase in the real age through its ositive effect on the mark-u. But that increase in taxes aid by emloyees ould have the oosite effect on the mark-u, comared to a reduction in the ayroll tax. Moreover, in the short run, it ould lead to a decrease in emloyment since the negative age tax effect ill be larger than the ositive ayroll tax effect. The overall imact of this tye of olicy ould therefore be to increase unemloyment in the short term. It might have been better to have just reduced age taxes; or simly to have removed the requirement to kee the budget balanced. But in either case, there are significant disincentives hich might block a structural reform rogramme of this kind. It entails a short run loss in economic erformance, olitical loss of face, and counter-roductive outcomes if budget balance is enforced just at a oint hen abandoning fiscal restraint might have risked destabilising the budget.

14 But in the long term, the sequence of events is reversed. By roosition 7, the net long run effect of an increase in the age taxes needed to comensate for our reduction in ayroll taxes, ould no lead to an increase in long run real ages; and to a decrease in the unemloyment rate (since the reservation age, hich also increases, is negatively related to unemloyment). The last statement follos because the rise in age taxes ill not affect the reservation age (roosition 7); but any comensating fall in ayroll taxes ill increase the reservation age, reflecting a fall in unemloyment, even if t has had no effect. This outcome reflects our results in section 4 because the increase in age taxes has a larger effect in increasing the marku, and hence real ages and the reservation age, than the decrease in ayroll taxes has in decreasing it. In other ords, there is a demand side effect desite the neutral budget changes; and the distribution of the burden of taxation matters a great deal. This examle therefore confirms a quite idely acceted argument that structural reforms (a reduction in the burden of taxes in this case) ould be beneficial from the long run ersective; but ould certainly induce short run costs, both in terms of economic erformance (indicated here by the increase in the short run unemloyment rate) and in its olitical imlications. And this short run-long run conflict is then made very much orse by the resence of the budget restraint. That, if nothing else, might be enough to derail the reform efforts. But the long term effects are entirely ositive. The question therefore is hether the (discounted) long run benefits ould be judged to outeigh the short run costs in ractice. To anser that fully requires an exlicitly dynamic analysis, and another aer. This aer gives the comarative statics vie and a matching assessment of the long term financial imlications in terms of real ages, elfare (consumtion) and the government s budget balance. 5.2 Which reform rogrammes are most effective? Folloing the results of section 4, it is natural and imortant to ask hich reform strategy ill be most effective in terms of increasing the number of firms, roducts and hence emloyment in an economy. As noted in corollary 2, to be effective here means being effective in getting the mark-u or accetable cost of entry to fall as taxes or labour or roduct market regulation falls. But if a measure is effective in that sense, then it ill raise real ages and the reservation age at the same time (by (24) and (25)). That then imlies an increase in elfare and a decrease in unemloyment. Hence, the meaning of being effective is unambiguous: it can be assessed simly by determining hich reform instrument has the largest imact on real ages and hence elfare. From corollary 2, e already kno that a reform of age taxes, t, is the more effective of the to tax instruments. We also kno that deregulation of the roduct markets ill be better that tax reform ifδ ( δ 1) < 1 t, from hich e can calculate the maximum value of δ, δ max, such that market liberalisation ould be the referred otion for a given tax rate on ages. Folloing the same aroach, e can comare the size of the artial derivatives of i / P ith resect to β, δ and t. After some algebra, this yields the folloing:

15 Corollary 3: a) Product market liberalisation is more effective (elfare enhancing) as a reform rogramme than deregulating the labour market if δ < + (1 β )(1 t ) t 2, (30); 1 or if δ > 1+1/θ here θ is the tax distortion defined in section 3. b) Labour market deregulation is more effective than tax reform if t < (1 β )( δ 1) / δ (31) Proof: Comare ( i / P) / δ, ( i / P) / β and ( i / P) / t in absolute size. Corollaries 2 and 3 rovide a set of easily comuted sufficient conditions to assess the relative efficiency of each tye of reform rogramme, each condition being exressed as the maximum δ value that can hold if the given instrument is to be more effective. Thus, for the existing OECD and EU members, e find (Table 3): i) Product market liberalisation is more effective than tax reform if δ 1.5. ii) Product market liberalisation is more effective than labour market deregulation if δ 2 (if β 0), or if δ 1.5 (if β 0.25). iii) Tax reform is better than labour market deregulation if δ 1.3 (β 0) or if δ 1.5 (hen β 0.25), but for δ values u to 4 or 5 if β = 0.5. Tax reform is therefore almost alays the more effective reform instrument for the range of arameter values that are likely to hold in the Euroean or OECD economies. The only excetion ould be in an economy ith severe labour market distortions (β > 0.5). In that case, labour market deregulation is likely to be more effective. 5.3Structural reform requires joint, not single deregulation measures One might exect tax reductions to lead to less short run cometition in the roduct markets even if higher disosable incomes and increased rofits imly higher activity levels and extra emloyment. Indeed, from Proositions 2 and 3 e already kno that real ages, emloyment, and hence elfare, ill rise ith tax cuts made individually. But it easy to see from (13) that the rice mark-u, µ, ill also rise as the total tax burden falls, imlying that the degree of cometition in the roduct market falls as rents to firms rise. And in the long run, e kno that real ages and elfare ill fall ith cuts in age taxes, hile they rise ith cuts in the taxes aid by emloyers (Proosition 7). Hence e must face the ossibility that a tax reform that reduces the overall tax distortions in the economy may roduce short run gains in elfare and emloyment; but losses in the longer run because the extra cometition in the roduct markets (hich ill start to increase after firms rents rise) ill rise by less than tax distortions fall. That is, because tax reductions ill loer rices and ages (but rices by less than ages, in vie of Proosition 7), rice distortions ill rise relative to the tax distortions in the long term and ill remain higher. We can demonstrate that formally: Proosition 8: Structural and tax reform measures should be conducted jointly in order to avoid increasing relative distortions/rigidities in other arts of the market, and hence a net loss in deregulation or liberalisation overall.

16 Proof: Proositions 2 and 3 imly real ages, emloyment and elfare rise ith cuts in either t ort, and hence reductions in the overall level of tax distortion θ, since (1 + δθ ) (1 + µ ) θ / t and θ / t are both ositive. But µ = andδ = by (13). Hence ( δ 1) ( µ θ ) dµ µ µ δ δ = +. = [1 µ ] < 1 (32) dθ θ δ θ δ 1 since µ > 1/ δ is guaranteed if δ>1. Meanhile δ (1 + µ ) = θ ( µ θ ) 2 > 0 µ δ and = 0 θ δ 1 > (33) under the same conditions. Consequently 0> dµ > dθ; and rice distortions increase, d(µ-θ) = dµ-dθ>0, hen tax distortions are reduced. Hence a reduction in θ must be matched by an increase inδ to offset the relative rise in µ and absolute fall in δ. 6. Emirical Evidence To evaluate the ractical significance of our results, e use the OECD s Tax Data Base and unemloyment figures from the OECD s Main Economic Indicators. The former sulies t andt defined as the all in average tax rate on manufacturing ages and cororate incomes, inclusive of social security contributions; the latter, unemloyment rates on a standard definition. 3 For the remaining data, e set β (age bargaining arameter) at 0.25, being the mid-range estimate from the Layard, Nickell and Jackman (1991) study, and then consider β = 0 and β = 0.5 (decentralised and centralised bargaining resectively) as alternatives. Finally e set δ at 3.5 for the short run substitutability beteen roducts, and δ = 10 for the long run. These figures are based on the fe ithin-eriod roduct substitutability studies in the literature and may be comared to δ = for erfectly cometitive markets. 4 All data are for The Baseline Case As a benchmark, Table 1 records the tax and rice distortions, as they stood in 2005, for the 24 OECD economies and the EU as a hole. There is considerable variation, but three features stand out. First, all of Euroe (Ireland exceted) suffers greater tax and rice distortions than the US. But outside Euroe, only Canada does. Likeise, core Euroe (Belgium, France, 3 The OECD figures agree ith Eurostat s ESA95 data, excet that the latter does not searate emloyer from emloyee social security contributions. As a result, e don t have consistent data for the smaller states of the EU (Estonia, Cyrus, Latvia, Lithuania, Malta, and Slovenia) ho are not yet members of the OECD. Slitting those contributions beteen emloyers and emloyees gives us rough estimates of the figures in tables 1-4 for those countries. Those figures are available on request. 4 Ogaki and Reinhart (1998a,b) suggest for the US, hile develoing countries have loer figures hich again suggests ould be about right for the OECD economies. Ravn et al (2004) refer 2.0; Paadaki et al (2004) ; and Gali et al (2003) calculate mark-us hich imly δ = 3.3 for the EU. Long run figures come from midoint US estimates in Duca and VanHoose (2000).

17 Germany and Seden in this instance) are noticeably more distorted than the EU as a hole; and the Netherlands, Czech Reublic, Hungary, Poland and Finland come close. Most of the others are similar to the US and Canada. In most cases tax distortions and rice distortions are equally serious; but in the case of the Netherlands, Poland, Finland and Denmark rice distortions are the more serious (as imlied by the high values of c, reflecting above average mark-us), hile in France and Italy tax distortions are more serious. There is therefore a small vs. large distinction in terms of cometitive markets. Second, countries can be groued by the strength of their market distortions: (i) Core Euroe: Belgium, Germany, France, Italy Seden and the EU-25 (µ >1.5). (ii)the Hasburgs: Czech Reublic, Hungary, Poland, Slovakia, Finland, Netherlands (1.5>µ>1.4) 5. (iii) Perihery Euroe: Austria, Denmark, Greece, Sain, Noray (1.4>µ>1.1) (iv) Small and Anglo-Saxon economies: the US, UK, Portugal, Sitzerland, Canada, Australia ith1.1> µ>0.95; and (v) Recovery economies: Jaan, Ireland (µ<0.95; the small transition economies, not shon, also fit into this grou). This grouing, hile arbitrary, remains unchanged for different values of δ (and β). Third, tax distortions are larger than rice distortions in the EU-25: Belgium, Czech Reublic, Denmark, Germany, France, Italy, Netherlands Seden, Austria, Finland, Poland, Hungary and Slovakia. But rice distortions are more imortant in Sain, Greece, Ireland, Portugal, the UK and the non-eu economies, reflecting erhas the size of the domestic markets but more likely a loer incidence of taxation. 6.2 Welfare Comarisons Table 2 considers the elfare changes that could be achieved by reducing the distortionary effects of taxation, in line ith the examles given in section 5, or by direct market liberalisation measures. Recall from section 2.1, utility can be calculated from the otimal consumtion values (4). Table 2 records those utility values, and the initial real and reservation ages in the first three columns. Those figures ut the less distorted economies somehat ahead, in elfare and emloyment, of those ith higher distortions rincially tax distortions, given the results of the last section. Columns (4) and (5) therefore comute the elfare changes if all countries adot US tax rates in lace of their on 6. All the EU countries (bar Ireland) ould benefit in the short run, but by varying amounts ranging from 40%-50% for the hard core (Belgium, Germany, Italy) to 20%-30% for France, the Scandinavians and the Hasburgs; but only 14%, 5% and 2% for Sain, Portugal and the UK. But Ireland ould lose 7%, reflecting her lo tax rates and strong groth; and Jaan, Australia and Sitzerland likeise. These are all short term changes derived using results from section 3. The alternative strategy is market liberalisation. If all countries removed all barriers to cometition, as must haen (ceteris aribus) in the long term in this model, then e 5 With surrisingly little violence to history: the Netherlands as under Hasburg rule for a limited time, Poland only artly, but Finland never as. And Austria is aroximately a Hasburg here. 6 This is the standard comarison in this literature: see Bayoumi et al (2004), Coenen et al (2006).

18 have the long run gains of columns (6)-(10) of Table 2. Here e exloit the results of section 4. In columns (6)-(8), all rice distortions have vanished (δ ) but the original tax distortions remain. That yields fairly uniform gains of 28%, similar to the short run gains from reducing tax distort-ions in the EU. But if e combine liberalisation ith reduced taxes, e get much larger gains in articular for the hard core EU economies (column (10)). The latter gain by 65%-90%; the Hasburgs and Scandinavians by 50%; and the rest by 20%-30%. Reducing tax distortions may be the biggest art of this, but liberalisation in the roduct markets is an essential comlement (if not equal artner) in EU economies. That underlines the imortance of making joint reforms (roosition 8) Short run costs, long run gains. These results confirm that tax reforms are tyically more effective than other reform instruments hen alied individually, as e sa in Table 3. The same is erhas less obvious hen reforms are undertaken jointly. The other clear result from our theoretical model is that any reform rogramme may be blocked because the short run costs. Section 5.1 illustrated ho that might haen, and Table 4 rovides some numerical examles for the tax reform case. Here e calculate the revenue losses if tax rates are reduced to the US level, using the changes in short term and then the long term real ages from sections 3 and 4 to do so. To estimate the changes in budget balances, e need to add in the falls in exenditure due to falling unemloyment (other exenditures are assumed fixed). We suose unemloyment to be roortional to the reservation age so that, in the long term, changes in unemloyment benefit ayments equal changes in the unemloyment rate. Table 4 then shos that core Euroe ould suffer significant losses in tax revenues, of 20% to 30% in the short term, and the rest of the EU-25 ould suffer smaller losses. But those outside (and Ireland, Portugal and the UK) ould gain. The icture is better in the long term: revenue losses for core Euroe shrink to 5%-10%, hile the rest start to gain ith imroved activity levels. The long term budget ositions therefore vary beteen losses of 3% to 8% for Germany, France, Italy and Belgium; and gains of 10%-30% elsehere in the EU, or 40% in the UK and outside. The conflict beteen short run losses and long run gains is therefore very clear. In fact, a number of countries ould find that they breach the Stability Pact s 3% of GDP deficit limit, in both the short term and long term, if they folloed this reform rogramme. That includes the current sinners (France, Germany, Italy, Greece), but also Hungary, the Czech Reublic, Poland and Slovakia. This might be enough to dissuade them from undertaking any reforms at all. 7. Conclusions We have taken a standard model of the labour market in an economy ith imerfect cometition in the roduct and labour markets, and extended it to allo for distortionary taxation. From the equilibrium outcomes of this model, e conclude: a) There is a difference beteen the short run and long run outcomes of reform. The short run involves significant costs or losses in erformance, but the long run effects

19 are almost uniformly favourable. Structural reform rogrammes are therefore very likely to be avoided, or abandoned if undertaken, because of their short run costs. b) Fiscal restraints, such as imosed by Euroe s Stability Pact, exaggerate this effect and make reform less likely. c) The choice of reform instrument matters. Tax reform usually has a larger effect than market deregulation or deregulation. d) The advantage of using age taxes vanishes if markets become fully cometitive (δ ) or union bargaining oer becomes too large (β 1). e) Pro-cometitive liberalisation more generally is likely to be more effective than tax reform if the level of cometition is very lo or taxes are high, but less useful if the degree of market cometition is already strong. f) Excessive union oer (β 1) ill mean no reforms are ossible since the long term gains vanish (c 0) and e are left ith the case here short run costs dominate. From the numerical results, e see that both the tax distortions and the degree of imerfect cometition (rice mark-us) is likely to be quite large in some OECD economies, rincially those at the centre of the Eurozone. The distortions in those outside the Eurozone, or in a transition/recovery hase, are significantly loer. As a result, there are imortant gains to be had (in elfare and income terms) by reducing those distortions: tyically imrovements of 25%-30% in the long run (but less in the short run) if the tax distortions are reduced to the US level, and maybe double that if the roduct markets are liberalised at the same time. These gains are of the same order as, but somehat larger, than those reorted by Bayoumi et al (2004) and Coenen et al (2006). We get larger figures because those earlier studies rely on numerical simulations of a calibrated model and are therefore unable to cature long run in full, and have not considered the effects of simultaneous reforms or liberalisation in to sectors (taxation, roduct markets) at once. The size of the gains e reort therefore emhasise our conclusion: joint reforms are desirable, if not necessary. Finally, e have shon that tax reform is likely to be more imortant than market reforms in Euroe unless the roduct and labour market constraints are severe. But the short run costs of tax reform could easily be quite large and revent any reforms from taking lace. References: Bayoumi T, D Laxton, and P Pesenti (2004) Benefits and Sillovers of Greater Cometition in Euroe: A Macroeconomic Assessment Staff Reort 182, Federal Reserve Bank of Ne York. Blanchard O and F Giavazzi (2003), Macroeconomic Effects of Regulation and Deregulation in Goods and Labour Markets, Quarterly Journal of Economics, 118,