Political instability and labour market institutions

Political instability and labour market institutions

European Journal of Political Economy 39 (2015) 201–221 Contents lists available at ScienceDirect European Journal of Political Economy journal home...

777KB Sizes 0 Downloads 35 Views

European Journal of Political Economy 39 (2015) 201–221

Contents lists available at ScienceDirect

European Journal of Political Economy journal homepage: www.elsevier.com/locate/ejpe

Political instability and labour market institutions Claudio Lucifora a,b, Simone Moriconi a,c,⁎ a b c

Università Cattolica, Largo Gemelli 1, 20123 Milano, Italy IZA, Schaumburg-Lippe-Str. 5-9/9, D-53113 Bonn, Germany CREA, Université du Luxembourg, 162a avenue de la Faiencerie, L-1511, Luxembourg

a r t i c l e

i n f o

Article history: Received 14 November 2014 Received in revised form 20 May 2015 Accepted 21 May 2015 Available online 29 May 2015 JEL classification: J64 J88 H11

a b s t r a c t This paper investigates the relationship between political instability and labour market institutions. We develop a theoretical model in which political instability creates incentives for a government to introduce labour market regulation in the economy. The distortionary effect of regulation on unemployment effectively puts a constraint on the design of fiscal and public policies. We empirically investigate these predictions using panel data for 21 OECD countries for the period 1985–2006. Our results are consistent with the view that political instability is associated with more regulated labour markets, lower labour taxation, and lower unemployment benefit replacement rates. © 2015 Elsevier B.V. All rights reserved.

Keywords: Political instability Public good provision Labour market institutions

1. Introduction The recent financial crisis has brought employment and growth back to the top of the policy agenda of most industrialised countries. In the public debate, slower growth and poor employment performance have often been associated with lack of governments' commitment to reform extensive labour market regulation. Moreover, policy makers often report a growing concern that frictions in the political process may hinder policies that are oriented to the long term (OECD, 2012; EC, 2013).1 In this context, it is worth asking why labour market regulation is so pervasive in developed countries, and why governments find it so difficult to reform the labour market. This paper explains how labour market regulation may arise as a result of political instability. We show that labour market regulation may be the optimal choice of a government that maximises the welfare of its constituency rather than social welfare. Since workers constitute the majority of voters, the government cares about workers regardless of the political preferences of its constituency. This creates incentives to introduce regulation that creates wage rents, which in turn increases unemployment in the economy. Moreover, the distortionary effect of labour market regulation induces the government to moderate labour taxation to alleviate unemployment and increase social expenditure on unemployment benefits. In this way, the government effectively discourages any fiscal and public policy that is not valued by its constituency. We also show that, when individuals differ in their

⁎ Corresponding author at: Università Cattolica, Largo Gemelli 1, 20123 Milano, Italy. Tel.: +39 0272343393. E-mail address: [email protected] (S. Moriconi). 1 Recognition of the role of political instability in determining market regulation and institutions lies behind many attempts on the part of international organisations, such as the OECD Job Strategy, the EU Lisbon agenda, the World Bank's report Understanding Regulation, and the more recent EU2020 Strategy, urging governments to reform labour markets.

http://dx.doi.org/10.1016/j.ejpoleco.2015.05.003 0176-2680/© 2015 Elsevier B.V. All rights reserved.

202

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

unemployment probability (i.e. there are insiders à la Blanchard and Summers (1986)), the government reduces the unemployment benefit replacement rate when the political power of insiders increases. These results hinge on the existence of an asymmetry between labour market regulation on the one hand, and fiscal and public policy, on the other. We argue that the regulation of labour markets, being embedded in the institutional setting, is hard and costly to change. In contrast, fiscal and public policies are easier to change within the government budget. In this context, with political instability, labour market regulation becomes the state variable that gives the incumbent government an instrument to control future fiscal and public policy. We empirically investigate the main predictions of our theoretical framework using panel data for 21 OECD countries over the period 1985–2006. We specify and estimate an empirical model to explain how political instability is related to labour market regulation, labour taxation, and the unemployment benefit replacement rate. Our results are consistent with the view that political instability induces government to regulate the labour market, to reduce labour taxation, and to lower the unemployment benefit replacement rate. The above results are shown to be robust to a number of extensions, such as the inclusion of alternative political variables, systems with different legal origins, and predetermined indicators of political instability. This paper contributes to the debate on the political determinants of labour market institutions. First, it sheds light on the effects of political instability on the choice of labour market institutions taking as given the political process. Existing papers (see e.g. Wright (1986); Saint-Paul (1996); Pagano and Volpin (2006); Boeri et al. (2012)) have mainly focussed on political economy issues, showing that the level and mix of institutions depend on the characteristics of the median voter. Second, it contributes to the large literature that looks at the economic effects of political instability. In our framework, labour market regulation plays a similar role as public debt in Alesina and Tabellini (1990) and Persson and Svensson (1989): it is a strategic variable that affects the action of future policy makers. In existing studies, a similar role is played by property rights protection (Svensson, 1998) and long-run capital taxation (Devereux and Wen, 1998; Azzimonti, 2011). Other contributions have focussed instead on how political instability, risk, and short-term policy making may negatively affect the design of institutions that foster economic growth (e.g. private and public investments; see Alesina and Perotti (1996); Aidt and Dutta (2007); Busse and Hefeker (2007)). Third, this paper is also related to the literature on labour market institutions and unemployment. Traditionally, studies that analyse the impact of labour market institutions on equilibrium unemployment consider them as exogenous (see Bassanini and Duval (2009); Arpaia and Mourre (2010) for reviews). More recent contributions also investigate how institutions may arise endogenously e.g. due to demand for protection during economic crises (Galasso, 2014), international trade (Raess, 2014), and globalisation (Hessami and Baskaran, 2014). This paper is organised as follows. Section 2 presents the theoretical model. Section 3 describes the data and the empirical strategy. Section 4 presents and discusses the main results, alternative specifications, and a set of robustness checks. Section 5 provides some concluding remarks. 2. A model of political instability and labour market regulation We introduce a static partial equilibrium model to describe a small open economy composed of a private and a public sector. In the private sector there is an imperfectly competitive labour market, where a mass of homogeneous workers inelastically supplies one unit of labour to satisfy aggregate labour demand. In the public sector there is a government, whose policy action occurs in one period (say a legislature). The government has three policy instruments: labour market regulation, fiscal policy, and public policy. Labour market regulation affects the ability of workers to extract wage rents, and creates unemployment in the economy. Fiscal policy consists of the choice of labour taxes to finance public expenditure, while public policy allocates expenditure between alternative public goods. The central assumption of the model is the existence of a natural asymmetry between labour market regulation on the one side, and fiscal and public policy on the other. Labour market regulation is by its own nature broad, pervasive, and is embodied in institutions. These features make it hard and costly to undo regulation set in the past. In contrast, fiscal and public policies are determined within yearly budgets, and any new government sets its own budget.2 To model this asymmetry, we assume the following timing: at stage 0, the government sets labour market regulation; at stage 1, the government sets and implements fiscal and public policy, and implements the labour market regulation previously set at stage 0; at stage 2, after all policies are implemented, wage and employment equilibrium levels are realised. We assume that there are two parties in the political arena, denoted by a and b, which get political support from their respective constituencies A and B. Constituencies A and B are composed of individuals who derive utility from labour income, and have different preferences for public good provision.3 Without loss of generality, we posit that party a is the incumbent government at stage 0. However, between stage 0 and stage 1, a political shock may occur, which replaces the incumbent with its competitor, party b. In this context, we have political instability à la Alesina and Tabellini (1990) and Persson and Svensson (1989): labour market regulation becomes the state variable that gives the incumbent government an instrument to control the fiscal and public policy of its 2 The existence of this asymmetry is also supported by empirical evidence. In countries characterized by pervasive labour market regulation, e.g. Germany, France and Italy, reforms introducing more flexibility in employment contracts have often been delayed by strong opposition and lengthy consultation with the social partners (Deroose et al., 2006). 3 This assumption can be interpreted in terms of an electoral competition model, where all voters derive utility from labour income, regardless of their preferences for public good provision. This view recalls political power theories of labour market regulation, which argue that workers represent the majority of voters such that any party in the political arena aligns its policy platform to their preferences (see Botero et al. (2004); Saint-Paul (1996)). In this section we consider workers that face the same unemployment probability. Later on, we consider an alternative setting where individuals differ in their probabilities of becoming unemployed. This extension finds its micro foundations in a voting model where the median voter can be unemployed, and chooses the level of the unemployment benefit replacement rate.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

203

successor. The incentives of the incumbent government to regulate the labour market depend on its expectations to be replaced in power. By regulating the labour market, the current government discourages its competitor from choosing a fiscal and public policy that is not valued by its own constituency. We solve the model by backward induction and start from the description of the wage and employment outcomes in the private sector at stage 2. We then move to the public policy and the fiscal policy at stage 1. We finally analyse the choice of labour market regulation at stage 0. 2.1. The private sector At stage 2, employment and wages are determined. We assume that the labour market is imperfectly competitive due to the presence of unemployment benefits (v) and labour market regulation (μ). The formation of wages is described as follows4: w ¼ W ðμ; vÞ

s:t:

W ð0; vÞ ¼ v;

v ≥ v;

0

W j N 0;

where

j ¼ μ; v:

ð1Þ

where v̄ is workers' reservation wage. In a perfectly competitive labour market (i.e. when μ = 0 and v = v̄), all workers earn the reservation wage, v̄. In an imperfectly competitive labour market, unemployment benefits and labour market regulation increase wages, W. In Eq. (1), we assume that unemployment benefits never fall below the workers' reservation wage, v ≥ v̄. Each worker faces an aggregate labour demand function5: lðcÞ ∈ ½0; 1;

lðvÞ ¼ 1;

0

lc b 0;



lcc b 0;

and c ¼ w þ τ:

ð2Þ

The aggregate labour demand specified in Eq. (2) is consistent with profit maximising behaviour by firms operating in a perfectly competitive product market. Firms' labour costs (c) include the workers' product wages (w) and the labour tax levied on each unit of labour employed (τ). From Eq. (2), when workers receive their reservation wage and labour is not taxed, there is full employment, while any increase in wages or labour taxes induces unemployment in the economy.6 The workers' expected income is E½I ¼ wl þ vð1−lÞ ≡ v þ ðw−vÞl:

ð3Þ

Eq. (3) describes the effects of μ, v and τ on workers' expected income. Unemployment benefits increase the workers' outside option.7 Labour market regulation and unemployment benefits increase wage rents. Finally, policies increasing labour costs have a negative impact on workers' expected income, as they increase the workers' probability of being unemployed. 2.2. Public and fiscal policy At stage 1, the incumbent government sets the public policy and the fiscal policy. We assume that the objective function of party i = a,b is additively separable in workers' income and the utility from public good provision to its constituency I = A,B h i I I E V i ¼ E½I  þ δGi ;

where i ¼ a; b;

I ¼ A; B:

ð4Þ

In government objective Eq. (4), workers' expected income does not depend on the party in office, as both parties are equally representative of workers' interests. However, party i = a,b maximises the utility from public provision of its constituency I = A,B. We assume that the marginal utility from public goods is higher than the marginal utility from income, i.e. δ N 1. We specify the utility from public good provision of constituency A when party i = a,b is in government office as follows (Svensson, 1998) A

Gi ¼



 1 A g γ ð1−γÞ i

A

where g i ¼ min½γhi ; ð1−γ Þf i 

and

f i þ hi ¼ z:

ð5Þ

4 This can be interpreted as a reduced form Nash bargaining outcome in an equilibrium gross job flow model (Mortensen and Pissarides, 2001. See Saint-Paul (1996); Boeri et al. (2012) for similar approaches). 5 Since workers are homogeneous, they all face the same aggregate demand, and l(c) can also be interpreted as the probability of the representative worker's being employed for a given level of labour costs. We relax the assumption of homogeneous workers in Section 2.4. 6 The assumption of concavity l″cc b 0 ensures that the elasticity of labour demand is increasing with labour costs, i.e. ϵc' N 0, which is a sufficient condition for the concavity of the government's welfare function. This assumption is common in oligopoly models and in the tax incidence literature (e.g. Stern, 1987). It encompasses the idea that policies that increase the cost of labour have complementary effects on unemployment (Coe and Snower, 1997), e.g. via adjustments at the intensive (i.e. via firms' optimal hiring policies) and the extensive (i.e. via the number of firms operating in the market) margins. 7 Later, we discuss the government's choice of the level of unemployment insurance and specify the level of unemployment benefits as a function of the government's choice of the replacement rate ρ.

204

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

Here, g Ai denotes constituency A's utility from public goods h and f, when the allocation of public spending z between the two is chosen by party i = a,b. From Eq. (5), constituency A derives utilities γhi and (1 − γ)fi when party i = a,b sets hi, fi respectively. The utility of constituency B from public good provision, when party i = a,b holds government office is B

Gi ¼



 1 B g γ ð1−γÞ i

B

where g i ¼ min½ð1−γ Þhi ; γ f i 

and hi þ f i ¼ z;

ð6Þ

where g Bi is the utility of constituency B from the allocation of public spending implemented by party i = a,b. Notice that g Bi in Eq. (6) is defined as g Ai in Eq. (5), but with γ replaced by 1 − γ. We assume γ ∈ (1/2, 1]. This assumption describes the heterogeneity of preferences by constituencies A and B: the former constituency maximises its utility by allocating more spending to f, while the opposite holds true for the latter.8 Assume without loss of generality that party a is in office at stage 1. Party a implements the public policy preferred by its constituency A

g a ¼ min½γha ; ð1−γÞf a  with

f a þ ha ¼ z:

ð7Þ

From Eq. (7), and since γ ∈ (1/2, 1], party a sets a composition of public good, which makes the utility obtained by constituency A from consumption of f at least as high as the utility it gets from consumption of h. From Eq. (7) we have γha ¼ ð1−γ Þf a :

ð8Þ

From Eq. (8), as ha = z − fa, party a's optimal allocation of spending on public goods is 

f a ¼ γz;



ha ¼ ð1−γÞz:

ð9Þ

Since γ ∈ (1/2, 1], Eq. (9) implies that party a allocates a higher share of z to public good f. If we substitute Eq. (9) into Eq. (5), we obtain the utility of constituency A from ( fa⁎, ha⁎) A

Ga ¼ z:

ð10Þ

We now turn to the choice of fiscal policy. When setting the optimal level of labour taxes, party a's objective function internalises the preferred combination of public goods of its constituency. Insert Eq. (10) into Eq. (4) h i A E V a ¼ E½I þ δz;

ð11Þ

where the workers' expected income is given by Eq. (3). Total public expenditure includes spending on public goods and unemployment benefits, and is financed by labour taxes. The government's budget constraint is z þ vð1−lÞ ¼ τl:

ð12Þ

Party a chooses the level of τ that maximises Eq. (11), subject to the budget constraint (12) and aggregate labour demand (2). The first order condition is 0

δl ¼ −½w−v þ δðτ þ vÞlc :

ð13Þ

Eq. (13) defines the optimal fiscal policy. The incumbent government at stage 1 sets the labour tax at the level where the marginal benefit of the tax (in terms of higher utility from public goods, on the left hand side of the equation) equals its marginal cost (in terms of foregone employment). The stage 1 equilibrium labour tax is an implicit function of labour market regulation and unemployment benefits, that is, T(μ , v). We have (the proof is in Appendix A.2): Proposition 1. An equilibrium level of the labour tax exists τ⁎ = T(μ , v) such that (i) T ð0; vÞ ¼ τ max ; (ii) T(μ , v) b τ max, for any μ N 0 and v N v̄; (iii) if T(μ , v) b τ max, then Tj' b 0 with j = μ, v.

8

This readily follows from the application of the maximin criterion for welfare maximisation. See also Appendix A.1.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

205

The optimal tax rule in Eq. (13) implies that when μ = 0 and v = v̄, – i.e. in perfectly competitive labour market – the government sets the highest possible labour tax τ max, as in (i) above.9 However, when μ N 0 and v N v̄, the employment penalty of the labour tax increases, and induces the government to set τ⁎ b τ max, as in (ii) above. Finally, any increase in μ and v induces the government to cut the labour tax, as in (iii) above.10 In other words, Proposition 1 suggests that labour market regulation and unemployment benefits reduce the government's willingness to levy taxes. This ‘tax moderation’ effect alleviates unemployment and increases the share of workers that benefit from wage rents. 2.3. Political instability and labour market regulation At stage 0, the incumbent government sets labour market regulation. The incumbent party has an exogenous probability p of being replaced by its competitor at stage 1 (Alesina and Tabellini, 1990). Assume without loss of generality that party a is in office at stage 0. If party a remains in office at stage 1, it will set the public policy (9) preferred by its constituency, who will enjoy utility G Aa given by Eq. (10). However, if party a is replaced in office, party b at stage 1 will maximise the utility from public good provision of its constituency, described by Eq. (6), and set the public policy ( fb⁎, hb⁎). In this case, constituency A will only get utility G Ab b G Aa (see Appendix A.1, for details). The expected utility from the provision of public goods for constituency A at stage 0 is then given by weighting GAa and GAb by their probabilities (1 − p) and p, respectively: h i A A A E G ¼ ð1−pÞG a þ pG b ¼ ϕðp; γÞz

with ϕðp; γ Þ ¼ 1−2p þ

p ; γ

ð14Þ

where 0 b ϕ(p, γ) b 1 denotes the degree of uncertainty associated with party a's public policy due to political instability.11 It follows that ðiÞlim ϕðp; γÞ ¼ lim ϕðp; γÞ ¼ 1; p→0 0

ðiiÞϕp b 0;

0

γ→12

ϕγ b 0;

ð15Þ

ðiiiÞ lim ϕðp; γÞ ¼ 0: γ→1

p→1 Condition (i) states that when p = 0 (independently of γ) or γ = 1/2 (independently of p), there is no political instability, and constituency A is certain that party a's public policy is going to be implemented (as ϕ(0, γ) = ϕ(p, 1/2) = 1). Condition (ii) suggests that an increase in p and γ makes constituency A uncertain about the implementation of its preferred public policy. Moreover, from Eq. (14), we see that the negative impact of p on ϕ is larger when γ is high, and vice versa. In the limit, as in (iii) above, p = 1 and γ = 1 implies that constituency A is certain that party a's public policy is not going to be implemented (i.e. (1,1) = 0). We derive the impact of political instability on the optimal level of labour market regulation by substituting Eqs. (3) and (14) into Eq. (4) and obtaining the objective function of party a at stage 0: h i A E V ¼ v þ ðw−vÞl þ ωðδ; p; γÞz

with ωðδ; p; γ Þ ¼ δϕðp; γ Þ;

ð16Þ

where ω(δ, p, γ) denotes the expected marginal utility from the provision of public goods with political instability. This depends on both δ and ϕ. If v = v̄ (we relax this assumption in the next section), party a chooses μ to maximise Eq. (16) subject to the agent optimality constraint (13). The first order condition is (see Appendix A.3 for details) 0

Vμ ¼

  0 ðw−vÞðδ−1Þl0c ðw−vÞ 0 0 þ ½1−ϕðp; γ Þ T μ lc þ ½ωðϕ; p; γ Þ−1 τ þ v lc ; : 0 δ w |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} μ |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl} social expenditure ðN0 if ωb1Þ wage rents ðb0Þ

ð17Þ

tax moderation ðN0 if ϕb1Þ

This shows that labour market regulation produces three different effects on welfare. The first is an effect on wage rents, due to a reduction in employment. The second is a tax moderation effect, as regulation induces the government to cut the labour tax in order to alleviate unemployment. The third is a social expenditure effect, as regulation increases the number of unemployment benefits' recipients, thus the share of spending on unemployment benefits in the government budget in Eq. (12). While the welfare effect of wage 9 The generality of our approach does not exclude a priori the possibility that the government's optimal choice is to impose a labour subsidy, i.e. a negative labour tax. However, when δ N 0, the government's optimal choice of τmax is positive. Since the labour tax creates unemployment, a part of the tax revenues is used to finance unemployment benefits. 10 In fact, under the assumption lcc″ b 0, any increase in μ and v increases the elasticity of the demand for labour with respect to labour costs, and thus the size of the employment penalty of the labour tax. 11 Persson and Svensson (1989) interpret ϕ(p,γ) as a measure of the time inconsistency of public policy, meaning that the incumbent government at stage 0 (i.e. party a) cannot commit to its announced policy due to an exogenous probability of being replaced by a government that will implement a different policy.

206

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

rents is always negative, the welfare effects of tax moderation and the social expenditure effect are a priori ambiguous, and depend on political instability. When ϕ = 1 (i.e. no political instability), the public policy of the incumbent is implemented for sure, thus public good provision is beneficial to its constituency. For this reason, there is no tax moderation effect, as the incumbent government does not have any reason to constrain the fiscal policy. The social expenditure effect on welfare is negative, since spending on unemployment benefits reduces the budget for public policy. In this case, the government optimally chooses the corner solution μ⁎ = 0, – i.e. no rentcreating labour market regulation. However, when ϕ b 1, political instability makes the public policy of the incumbent uncertain, thus reducing the expected benefits of public provision for its constituency. In this case, the government finds it optimal to restrain the fiscal policy of its successor and the tax moderation effect is positive. Moreover, when ω b 1, the expected marginal utility from public goods of the incumbent is lower than the marginal utility of income and the social expenditure effects is also positive. When the positive tax moderation and social expenditure effects balance the negative wage rents effect, (i.e. Vμ' = 0), the government optimally chooses an interior solution, μ⁎. This can be expressed as an implicit function of political instability. We have the following (the proof is in Appendix A.3)12: Proposition 2. An equilibrium level of labour market regulation exists μ∗ = M(p, γ) such that (i) M(0, γ) = M(p, 1/2) = 0; (ii) there exists a pair p b 1 and γ b 1s.t. M(p, γ) N 0 for any pNp and γ Nγ; (iii) if M(p, γ) N 0, then Mk′ N 0 with k = p, γ. With ϕ = 1, the incumbent finds it optimal not to regulate the labour market (i.e. μ∗ = 0), as in (i) above. In this way, it does not constrain the fiscal policy (in fact τ = τ max from Proposition 1), and maximises the utility from public goods of its constituency. However, when ϕ b 1/δ, the government finds it optimal to regulate the labour market (i.e. μ ∗ = M(.) N 0), as in (ii) above. By regulating, the incumbent forces its successor to keep labour taxes low (since τ ∗ b τ max). Any increase in political instability induces the incumbent government to regulate the labour market more, as in (iii) above, and forces its successor to reduce taxes on labour (since Tk′ = Tμ′ Mk′ b 0). Proposition 2 suggests that political instability makes it optimal for governments in a legislature to set labour market regulation that creates unemployment in the economy. This happens for two main reasons. First, the unemployment effect of labour market regulation can be optimally contained by the decision to moderate labour taxes, as political instability reduces the benefits of raising taxes to finance public goods. Second, politically unstable governments prefer to allocate a higher share of public spending to unemployment benefits, as the marginal utility of labour income is higher than their constituencies' expected utility from public goods.13

2.4. Unemployment insurance The analysis so far has assumed that workers are homogeneous and unemployment benefits are exogenously set at the workers' reservation wage. In this section, we present an extension of the baseline model where the incumbent party a cares about insiders (e.g. because insiders are a fraction of its constituency A), as in Blanchard and Summers (blanchardsummers), and chooses the unemployment benefit replacement rate, ρ = v/w at stage 0. We substitute v = ρw into the wage formation function (1), and rewrite it as follows:14

w ¼ Ψðμ; ρÞ s:t:

ρw ≥ v;

0

Ψ j N 0;

where j ¼ μ; ρ:

ð18Þ

Party a's employment objective is a weighted average of the insiders' employment probability, π0, and the aggregate employment probability (i.e. the labour demand function), l. The weight of insiders in the objective function is α(p, γ), and this depends on the uncertainty of public policy: EðlÞ ¼ α ðp; γÞπ0 þ ½1−α ðp; γ Þl;

where 0 b α ðp; γ Þ b 1:

ð19Þ

12 We assume also that the second order conditions are satisfied, which requires some assumptions about the third order derivatives of l(c), as is generally the case in optimal taxation problems. 13 Overall, this result can be interpreted in terms of an electoral competition model as political instability shifting the preferences of the median voter towards labour market rents and lower public good provision. 14 Eq. (18) is the exact counterpart of w − W(μ , ρw) = 0, which is obtained by inserting v = ρw into Eq. (1). Note that since unemployment benefits are now fully indexed to wages, we need to take into account the indirect impact of ρ and μ on v. Thus, in moving from Eqs. (1) to (18), we impose W′v b 1/ρ, which guarantees Ψ′μ N 0 and Ψ′ρ N 0.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

207

The relation between preferences for insiders and policy uncertainty is described as follows: ðiÞ lim α ðp; γ Þ ¼ lim α ðp; γÞ ¼ 0; p→0 0

ðiiÞ α p N 0;

0

γ→12

α γ N 0;

ð20Þ

ðiiiÞ lim α ðp; γ Þ ¼ 1: γ→1

p→1 Eqs. (19) and (20) convey the idea that party a's concerns towards insiders increase with the uncertainty of its public policy. When p = 0 or γ = 1/2, party a is certain than its public policy is going to be implemented (as ϕ(0, γ) = ϕ(p, 1/2) = 1), and does not value insiders, as in (i) above. In this case, from Eq. (19), the government employment objective is E(l) = l, as in our baseline model. Conversely, when p = 1 and γ = 1, party a is certain that its public policy is not going to be implemented, and only cares about insiders, as in (iii) above. In this case, as shown in Eq. (19), the government objective totally neglects the effects of labour market regulation on aggregate employment, i.e. E(l) = π0. Condition (ii) above shows that any increase of p and γ increases the uncertainty of party a's public policy, and raises its concern for insiders. Whenever party a's public policy is not fully enforceable, party a values both insiders and aggregate employment as in Eq. (19). Substitute E(l) and v = ρw in Eq. (16) to obtain the objective function of party a, which is the incumbent government at stage 0: h i A E V ¼ ρw þ ðw−ρwÞEðlÞ þ ωðδ; p; γÞz:

ð21Þ

The equilibrium at stage 0 is that combination of μ and ρ which maximises Eq. (21) subject to the agent optimality constraint given by the optimal tax rule τ⁎, implicitly defined in Eq. (13). Note that since the incumbent government at stage 1 is not concerned about labour market regulation, it does not consider insiders in its objective function. When there is policy uncertainty, i.e. ϕ b 1, μ influences the workers' expected income only through wages, and ρ affects the expected income both through wages and the degree of social insurance. Hence, the government uses μ to target wages and sets ρ based solely on insurance considerations. At the equilibrium level of labour market regulation μ⁎, the impact of ρ on welfare is described by (see Appendix A.4 for details) 0   V ρ μ ¼



α ðπ o −l Þ − 1−ρ |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

insiders0 rents ðb 0 if πo Nl Þ



ð1−l Þ 0 0 þ ½1−ϕðp; γÞ−α ðp; γ ÞT^ ρ lc − ðωðϕ; p; γÞ−1Þ : : 1−ρ |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl {zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl } tax moderation ðN0 if 1−ϕNα Þ

ð22Þ

social expenditure ðN0 if ω b 1Þ

Eq. (22) shows that the unemployment benefit replacement rate has three effects on welfare. The first is a direct impact on insiders' rents, which is negative when insiders at equilibrium, have a higher employment probability than the average worker, i.e. πo N l*.15 The second is a tax moderation effect, which is positive when the actual expectations to see party a's public policy implemented exceed concern for insiders, i.e. 1 − ϕ N α. The third is the effect on social expenditures, which is positive when ω b 1. When 0 b ϕ b 1 − α and ω b 1, the first order condition V 0ρ μ  ¼ 0 admits an interior solution. In this case, the government optimally chooses ρ⁎ as an implicit function of political instability and preferences for insiders.16 We have the following proposition (the proof is presented in Appendix A.4): Proposition 3. In the presence of political instability, an equilibrium level of the unemployment benefit replacement rate exists ρ∗ = ϒ(p, γ) such that ϒk′ ⋚ 0, where k = p,γ. In particular, ϒk′ ≤ 0 if αk′ ≥ Θϕk′ with Θ b 0. From Eq. (22), a lower unemployment benefit replacement rate has a positive welfare effect, due to insiders' rents, and a negative welfare effect, in terms of tax moderation and social expenditure. Proposition 3 states that political instability affects the equilibrium level of the unemployment benefit replacement rate, by modifying the net balance between these effects on welfare. When αk′ ≥ Θϕk′, political instability makes the government relatively more concerned for insiders' than for unemployment and public good provision. In this case, political instability produces net welfare gains from insiders' rents, and the government optimally chooses a lower unemployment benefit replacement rate. However, when α k′ b Θϕk′, political instability makes the government relatively more concerned for unemployment and social expenditure. In this case, political instability produces net welfare losses from insiders' rents. Accordingly, the government optimally chooses a higher unemployment benefit replacement rate. In other words, Proposition 3 states that political instability induces the government to reduce (increase) the unemployment benefit replacement rate, when it raises (lowers) concern for insiders relative to the provision of public goods.17 15 This is in the spirit of the one period model by Blanchard and Summers (1986), where initial membership in the group is exogenous and the insiders' employment probability depends on a comparison between the size of the group and aggregate employment. 16 Also in this case we assume that the second order conditions of government's maximisation problem are satisfied. 17 This result can be interpreted in terms of an electoral competition model where political instability shifts median voter's preferences towards labour market insiders i.e. a higher wage and a lower unemployment benefit replacement rate.

208

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

3. Data and empirical strategy The model in the previous section shows that political instability creates incentives for the government to increase labour market regulation that gives rise to unemployment in the economy. Since it is hard to undo labour market regulation in the time span of a legislature, it becomes optimal for the government to alleviate unemployment by cutting the labour tax. The model also shows that when individuals differ in their unemployment probability (i.e. there are labour market insiders), the government reduces the unemployment benefit replacement rate when the political power of insiders increases. The above structure generates three main theoretical predictions. First, political instability is associated with higher labour market regulation. Second, political instability also induces lower labour taxation. Third, political instability may determine a lower unemployment benefit replacement rate. In the empirical analysis, we look for evidence consistent with these predictions using a unique dataset that combines information on political systems and labour market institutions for 21 OECD countries over the period 1985–2006.18 One advantage of focusing on OECD countries is that they constitute a rather homogeneous group in terms of economic development and labour market standards. While there is considerable heterogeneity across countries in terms of institutional arrangements (e.g. union coverage and bargaining coordination see Aidt and Tzannatos (2008) for a review), in all countries workers exert their political power through collective bargaining, thus influencing the design of labour market regulation. We adopt the following specifications: 0

ð23Þ

0

ð24Þ

0

ð25Þ

μ it ¼ a1 Hit þ Xit l þ α i þ λt þ ϵit ; τit ¼ a2 H it þ Xit m þ α i þ λt þ χit ; ρit ¼ a3 Hit þ Xit n þ α i þ λt þ ν it ;

Here, the degree of labour market regulation μ, the labour tax τ, and the unemployment benefit replacement rate ρ depend on the degree of political instability H in country i at time t. We also include a vector of control variables X, country fixed effects ai, and time fixed effects λt, to take into account time invariant country specific features and common time shocks. We measure μ with a composite indicator that combines union density, an inverse measure of the corporatism of collective bargaining, and an index of employment protection legislation.19 We define τ as the average effective tax rate on labour (Carey and Rabesona, 2002). Finally, we compute ρ as the average gross replacement rates across various earnings levels, family situations, and durations of unemployment. From our theoretical predictions, we expect an estimated a1 N 0 in Eq. (23), a2 b 0 in Eq. (24), while the sign of a3 in Eq. (25) depends on whether political instability increases or reduces the government's concern for insiders, thus it is ambiguous a priori. Consistent with our theoretical model, we define political instability as H = 1 − ϕ(p, γ). A problem when it comes to measurement of H, is that this is not an easily observable dimension of government's action. Existing papers adopt an indirect approach, and proxy H by the size of general government expenditure, or the relative incidence of unproductive public spending, which are correlated with the number and size of political constituencies (see Holsey and Borcherding (1997) for a review). However, there are many other reasons than political instability, which may be consistent with the observed pattern of public expenditure, e.g. public sector inefficiency (Tanzi and Schuknecht, 2000) or myopic policy making (Aidt and Dutta, 2007). We adopt an alternative approach, and measure H in terms of political inputs to the policy making process. A country does not present only one relevant dimension, but multiple political and constitutional features that may influence political instability. The political economy literature (Persson and Tabellini, 2004; Persson and Tabellini, 2006) has classified these political features, distinguishing those that are structural to the political system (e.g. related to a country's electoral system), and the features of the incumbent legislature (e.g. related to the composition of majority and opposition in parliament). The effect of these political features on political instability has to be measured in a way that accounts of their multiple interdependences. Accordingly, to construct our empirical indicators of political instability, we rely on principal component analysis, and extract the relevant variation from both structural and political features of p and γ. The resulting synthetic indicators are time varying, and reflective of both changes in the political system and the legislature.20 In the empirical analysis, we specify H as a linear combination of p and γ and estimate separate coefficients in Eqs. (23)–(25). This also derives from our theoretical analysis. In fact, from Eq. (15), we have that ϕ ¼ 1−2p þ γp , which implies H = 2p − p/γ. Turning to the components of H, political turnover p is constructed using a principal component analysis on the following variables. First, a dummy for the lack of any plurality rule in a country's electoral system (Nowtal = 1), a dummy for a mixed proportional– plurality rule (PrPlu = 1), and a dummy for a pure proportional electoral rule (Pr = 1). These three rules favour more frequent changes of government, and a higher political turnover than with a pure plurality system (Persson and Tabellini, 2004). Second, a dummy for the adoption of closed party lists (CL = 1). Systems with closed lists produce higher political turnover than systems without lists or 18 The countries are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, The Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, The United Kingdom, and the United States. Note that the time series stop in 2006, to avoid the social and political turmoil caused by the economic crisis (see Appendix B for details on the data sources). 19 The composite indicator is obtained by standardising the individual measures to have zero mean and unit standard deviation and summing them (Duval, 2008). 20 Variation in the characteristics of the legislature comes from yearly changes in the government and opposition coalitions in all countries. Variation in the characteristics of the political system is associated with the comprehensive electoral reforms carried out in Italy and New Zealand during the 1990s. Our results are robust to the exclusion of these countries.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

209

Fig. 1. Patterns of political turnover and heterogeneity of political preferences. Notes: Country averages 1985–2006. Data from World Bank DPI.

with open lists, because they disconnect the re-election prospects of politicians from their efforts in office (Persson and Tabellini, 2000). Third, we use the inverse margin of majority of the executive over the opposition in parliament (Invmaj). The formation of a minimally-winning (or, even worse, minority) government indicates greater political turnover than for a large majority (Persson et al., 1997, 2000). Finally, the number of veto players who drop from the government in a given year is denoted by Stabs. In both presidential and parliamentary systems, a drop in the number of veto players signals a higher propensity for the government to terminate before the end of the legislature.21 The variable measuring the heterogeneity of political preferences, γ, is constructed by carrying out a principal component analysis on the following variables. First, a dummy measuring the maximum ideological distance between the executive and the four main parties in the legislature (Maxpolar = 1), which measures the polarisation of the preferences of the government coalition in parliament (Azzimonti, 2011). Second, two indicators of government and opposition fractionalisation (Govfrac and Oppfrac, respectively) measure the effective number of the parties in a coalition. Third, two indicators of the dispersion Govdisp of the government and of the opposition Oppdisp measure the share of seats among parties within the same coalition. Party fractionalisation and dispersion in parliament are complementary measures of parliamentary fragmentation (Beck et al., 2001) and give a voice to minority ideological and possibly extremist positions, increasing the heterogeneity of political preferences (Tsebelis, 1995; Tsebelis and Chang, 2004).22 The vector Xit includes additional control variables: two dummies for membership in the European Union EU and the European Monetary Union Euro, as well as an indicator Crisis for the occurrence of an economic crisis. (See Appendix B for more details on the definition of the variables used in the empirical analysis.)

3.1. Descriptive statistics Fig. 1 plots the mean values of political turnover and heterogeneity of political preferences by country, highlighting cross-country differences along the two dimensions of political instability. The sample means (dashed horizontal and vertical lines) partition the graph into four quadrants, which indicate alternative policy settings. Countries are mainly distributed in the bottom-left (low political instability) and upper-right (high political instability) quadrants, suggesting that the two indicators of political instability are positively correlated. In the bottom-left quadrant we find mainly Anglo-Saxon countries and Japan. These are characterised by a low probability of political turnover, due to plurality electoral systems and a low incidence of veto players; and low heterogeneity of political preferences, due to the presence of few parties in parliament. In the upper-right quadrant, we find Continental European and Nordic countries. These combine high levels of political turnover, due to pure or mixed proportional electoral systems and a high incidence of veto players; and a high heterogeneity of political preferences, due to the prevalence of multi-partyism.23 Mediterranean European countries, in the bottom-right quadrant, show low heterogeneity of political preferences and high political turnover. Compared to other European countries, France and Switzerland present a lower political turnover, due to a presidential constitutional system in France, and a majority electoral rule in Switzerland. 21 For presidential systems, the veto players are the President, the largest party in the legislature, and the largest party in the Senate. For parliamentary systems, veto players are defined as the Prime Minister and the three biggest coalition members. See Tsebelis (2002) for a comprehensive review of the role of veto players in modern political systems. 22 It may be argued that party fractionalisation and dispersion also reduce cabinet duration and could therefore equally well be considered as proxies of political turnover. However, this effect is second order, as it is always mediated by an increase in heterogeneity of political preference in parliament. See Carmignani (2003) for a review of the literature on the impact of political fragmentation on public policy. 23 Note that we check the robustness of our results to the exclusion of the Nordic countries, where political turnover may not indicate myopia but accountability, due to the high levels of civicness and trust in the political institutions.

210

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

Fig. 2. Political instability and labour market institutions. Notes: Differences between the averages for 2004–2006 and for 1985–1987.

Fig. 2 shows the relation between labour market institutions and a synthetic indicator of political instability.24 For each variable, we report the difference between its 2004–2006 average and its 1985–1987 average to allow for heterogeneity in (time invariant) unobserved factors. Changes in political instability appear to be positively correlated with changes in labour market regulation, and negatively correlated with both changes in labour taxation and changes in the unemployment benefit replacement rate. This descriptive evidence is consistent with the predictions of the theoretical model. The experiences of New Zealand and Portugal can prove useful to illustrate our arguments. While the two countries have roughly similar per capita incomes, they differ considerably 24 The synthetic indicator of political instability is obtained by summing up the two standardised measures of p and γ (see Appendix B for details about the PCA and standardisation).

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

211

in terms of both policy setting and labour market institutions (Botero et al., 2004). In the 1980s, New Zealand was a parliamentary democracy based on a pure winner-takes-all voting system, with low employment protection and a highly corporatist system, which was effective at moderating wage externalities associated with collective bargaining. In contrast, Portugal was a young parliamentary democracy, with strictly regulated (by constitutional rights) employment protection and collective relations laws. These two countries experienced different political developments over the next three decades. In New Zealand, the 1993 referendum paved the way to a switch from a plurality to a proportional electoral system, which increased the representation of smaller parties in Parliament. This reform increased political instability, and created uncertainty in public policy. On the side of labour market policy, starting in the early 1990s, successive New Zealand governments passed regulations that increased insiders' bargaining power, favoured wage externalities of collective bargaining,25 and reduced both labour taxes and unemployment benefit replacement rates (Freeman, 2008). In Portugal, a series of constitutional reforms favoured the election of stable centre-left, progressive governments. This process led to the implementation of shared policy platforms, and induced Portuguese governments to continuously erode the constitutional rights guaranteed to insiders and increase social expenditures (Abreu and David, 2010). While New Zealand and Portugal can be considered extreme cases, the experience of the United Kingdom, France, Italy, Germany, the Netherlands, Denmark and Spain, provides comparable evidence of countries that during the sample period experienced decreasing political instability and reduced wage rents, while increasing social expenditures. This was not the case in countries such as Belgium, Norway, Canada and Australia, in which political instability increased. 4. Results Table 1 presents the estimates of Eqs. (23)–(25) by ordinary least square (OLS). In columns [1]–[3] we regress the aggregate indicator of political instability on labour market regulation μ, the labour tax τ, and the unemployment benefit replacement rate ρ, respectively. In column [1], one unit increase in political instability is associated with a 0.33 unit increase in μ, suggesting that political instability leads a government to increase labour market regulation. In column [2], an increase by one unit in political instability leads to a decrease by 0.92 percentage points in τ, which is in accordance with a tax moderation effect. While our theoretical prior regarding the relation between political instability and the unemployment benefit replacement rate are ambiguous, the results in column [3] indicate that an increase by one unit in political instability is associated with a decrease by 3.95 percentage points in ρ. The specification in columns [1]–[3], however, does not allow for separating the effects of the determinants of political instability (e.g. political turnover and heterogeneity of political preferences) as highlighted in the theoretical model (see conditions (15)). This is done in columns [4]–[6], where we enter the two factors contributing to political instability separately.26 The results show that political turnover has a larger effect on labour market institutions than does the heterogeneity of preferences. An increase by one standard deviation in political turnover leads to an increase in μ by +0.87 percentage points, a decrease in τ by 1.48 percentage points, and a decrease in ρ by 7.13 percentage points. An increase by one standard deviation in the heterogeneity of preferences is associated with an increase in μ by +0.12 percentage points, a decrease in τ by 0.71 percentage points, and a decrease in ρ by 2.72 percentage points. In columns [7]–[9], we test whether the impact of political turnover is larger when preference heterogeneity is high and vice versa, as implied by Eq. (14), adding an interaction term to the previous specification. The overall effect of political instability on labour market institutions is magnified by the interaction term, supporting the view that political turnover (preference heterogeneity) has a larger effect when preference heterogeneity (political turnover) is above the OECD average. To get a better idea of the economic magnitudes implied by the above results, we build on the policy developments in New Zealand and Portugal, previously discussed, and compute the overall predicted impact of a change in political turnover and preference heterogeneity on labour market institutions. Our estimates indicate that a reduction in political instability such as the one induced by the Portuguese process of political stabilisation determined a 0.22 decrease in the indicator of labour market regulation (which is about 20% of the total variation in the average OECD country),27 an increase in labour taxes by +0.79 percentage points, and an increase in the unemployment benefit replacement rate by + 2.39 percentage points. As a polar case, a rise in political instability such as the one determined by the electoral reforms implemented in New Zealand, implied a +0.63 increase in the indicator of labour market regulation, a decrease in labour taxation by 1.29 percentage points, and a decrease in the unemployment benefit replacement rate by 6.54 percentage points. 5. Sensitivity analyses We now check the robustness of our main set of results by running a number of additional sensitivity analyses. In Table 2, we allow for potentially confounding features of the political factors that have been shown to be important features of government's choices of labour market institutions. These factors, if omitted, may bias the estimated impact of political instability on labour market institutions. Columns [1]–[3] add a dummy capturing a left-wing government political orientation (Left = 1). Political orientation may affect 25 In particular, there was a move towards a fully decentralised and uncoordinated system of collective bargaining. Labour protection was also increased along with a role for state arbitration courts in setting wages. See Castles and Mitchell, 1993 for details. 26 Note that since political turnover and preference heterogeneity are standardised, the size of their coefficients is directly comparable. This is not possible with the aggregate indicator of political instability, as it is not standardised. 27 Computed at the Portuguese averages (1.02 for political turnover and −0.48 for preference heterogeneity) and standard deviations (0.09 for political turnover and 0.40 for preference heterogeneity), the estimated coefficients in Table 1 (columns [7] to [9]) suggest an effect on labour market regulation of about −0.22 = 1.03 * (−0.09) + 0.14 * (−0.40) + 0.20 * [(1.02 * (−0.40)) + (−0.48 * (−0.09))]. Similar calculations apply to labour taxation and the unemployment benefit replacement rate. See Table B-1 for country averages and standard deviations.

212

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

Table 1 Political instability and labour market institutions: Main specification.

Political instability (H)

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

μ

τ

ρ

μ

τ

ρ

μ

τ

ρ

−0.92*** (0.12)

−3.95*** (0.44) −1.48*** (0.32) −0.71*** (0.17)

−7.13*** (1.39) −2.72*** (0.37)

0.95 456

0.93 456

1.03*** (0.12) 0.14*** (0.04) 0.20*** (0.04) 0.93 456

−2.09*** (0.34) −0.80*** (0.17) −0.78*** (0.16) 0.95 456

−8.22*** (1.41) −2.88*** (0.37) −1.38*** (0.39) 0.93 456

0.33*** (0.05)

Political turnover (p)

0.87*** (0.12) 0.12*** (0.04)

Preference heterogeneity (γ) Turnover* heterogeneity (p*γ) R sq. N

0.92 456

0.95 456

0.93 456

0.93 456

Notes: μ, τ and ρ stand for labour market regulation, the labour tax, and the unemployment benefit replacement rate, respectively. Political instability is obtained as the sum of political turnover and preference heterogeneity as these are standardised measures with mean 0 and standard deviation 1. OLS estimates with robust standard errors in parentheses. All specifications include dummies EU, Euro and Crisis plus country and time fixed effects. Significance levels: *: 10% **: 5% ***: 1%.

a government's stake in terms of the equity versus efficiency trade-off, as well as a government's attitude to labour income and redistribution (see Høj et al. (2006)). Columns [4]–[6] include two variables characterising the phase of the political cycle, i.e. a dummy for less than two years left to the end of the legislature (Yrcurnt ≤ 2 = 1) and a dummy for less than two years of office by the current government (Yroffc ≤ 2 = 1). The phase of the legislature may influence the government's incentives to make policy decisions involving long-term deferred benefits and short-term costs (see Høj et al. (2006); Dal Bó and Rossi (2011)). Columns [7]–[9] control for a measure of party strength, i.e. the average age of the main parties in parliament (Prtyage). Party strength captures the idea that stronger parties are more representative of stakeholders' interests, regardless of political instability (see Enikolopov and Zhuravskaya (2007)). Finally, columns [10]–[12] control for the youth of democracy, proxying for the quality of democratic institutions (Ydem). The youth of democratic institutions controls for any direct effect of the enforcement of democracy on labour market institutions (Alesina and Perotti, 1996). Results in Table 2 indicate that the magnitude and significance of the indicators of political instability are not altered by the additional set of controls. In particular, having a left-wing government, or being at the start of the legislature, shows no statistically significant effect on labour market institutions, although approaching the end of the political cycle is marginally significant. Consistent with the idea that party systems in modern democracies protect the interests of insiders, we find that, beside political instability, the

Table 2 Political instability and labour market institutions: Additional political controls. Political orientation

p γ p*γ Left

Phase of the legislature

Party strength

Young democracy

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

μ

τ

ρ

μ

τ

ρ

μ

τ

ρ

μ

τ

ρ

1.03*** (0.13) 0.14*** (0.04) 0.20*** (0.04) −0.00 (0.05)

−2.08*** (0.34) −0.81*** (0.17) −0.78*** (0.16) 0.08 (0.17)

−8.27*** (1.41) −2.83*** (0.37) −1.37*** (0.39) −0.36 (0.39)

−2.08*** (0.34) −0.79*** (0.17) −0.76*** (0.16)

−8.20*** (1.40) −2.82*** (0.37) −1.29*** (0.39)

0.96*** (0.12) 0.14*** (0.04) 0.19*** (0.03)

−2.05*** (0.39) −0.80*** (0.17) −0.77*** (0.16)

−6.50*** (1.08) −2.76*** (0.36) −1.10*** (0.33)

1.01*** (0.12) 0.12*** (0.03) 0.19*** (0.03)

−2.03*** (0.34) −0.73*** (0.16) −0.75*** (0.16)

−8.14*** (1.42) −2.79*** (0.38) −1.34*** (0.39)

0.00 (0.17) −0.12 (0.16)

−0.34 (0.41) −0.66 (0.44) 0.10* (0.05)

−0.06 (0.23)

−2.47*** (0.61) 0.44*** (0.08) 0.94 456

−1.59*** (0.33) 0.96 456

−1.92** (0.83) 0.93 456

Yrcurnt ≤2

1.04*** (0.12) 0.14*** (0.04) 0.21*** (0.04)

0.06* (0.03) −0.01 (0.04)

Yrsoffc ≤2 Prtyage Ydem R sq. N

0.93 456

0.95 456

0.93 456

0.93 456

0.95 456

0.93 456

0.93 456

0.95 456

0.93 456

Notes: μ, τ and ρ stand for labour market regulation, the labour tax, and the unemployment benefit replacement rate, respectively. Political turnover and Preference heterogeneity are standardised measures with mean 0 and standard deviation 1. OLS estimates with robust standard errors in parentheses. In Columns [1]–[3], Left = 1 is a dummy capturing a left-wing government political orientation. In Columns [4]–[6], Yrcurnt ≤ 2 = 1 is a dummy for less than two years left to the end of the legislature, and Yroffc ≤ 2 = 1 is a dummy for less than two years of office by the current government. In Columns [7]–[9], Prtyage is the average age of the main parties in parliament. In columns [10]–[12], Ydem is a categorical variable for the youth of democratic institutions. All specifications also include controls as in Table 1 plus country and time fixed effects. Significance levels: *: 10% **: 5% ***: 1%.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

213

Table 3 Heterogeneity: Political instability, labour market institutions, and legal origins.

p*Civil law p*Common law γ*Civil law γ*Common law R sq. N

[1]

[2]

[3]

μ

τ

ρ

1.16*** (0.18) 0.59*** (0.19) 0.13*** (0.04) 0.11 (0.13) 0.93 456

−1.56*** (0.43) −2.05*** (0.67) −0.86*** (0.19) 0.21 (0.52) 0.95 456

−13.50*** (1.82) −0.54 (0.93) −2.88*** (0.42) −3.08*** (0.85) 0.94 456

Notes: OLS estimates with robust standard errors in parentheses. All specifications also include controls as in Table 1 plus country and time fixed effects. Significance levels: *: 10% **: 5% ***: 1%.

average age of the main parties in parliament has an additional (weakly significant) positive association with labour market regulation, and a negative effect on the unemployment benefit replacement rate. Finally, the youth of democracy in a country suggests that a low quality of democratic institutions has an additional effect, which goes in the same direction as political instability (see Aidt et al. (2006)). In Table 3 we investigate whether the impact of political instability on labour market institutions is shaped by the legal tradition, and interact our indicators of political instability with a country's legal origins (Civil Law and Common Law dummies. See Glaeser and Shleifer (2002); Botero et al. (2004)). The results show a statistically significant larger association of political instability with labour market institutions in civil law relative to common law countries.28 This is in line with the idea that common law countries rely more on the functioning of markets, while civil law countries rely more on market regulation (see Glaeser and Shleifer (2002); Botero et al. (2004)). While it is often assumed that the political process is exogenous with respect to the determination of labour market institutions (see, e.g. Nunziata (2005)), it may be argued that governments jointly choose both labour market institutions and the political setting, so that the same factors that affect political instability are also correlated with the level of labour market institutions. In Table 4, we address the simultaneity problem by replacing the current values of our indicators of political instability by their predetermined values: i.e. we use five year lags in columns [1]–[3] and ten year lags in columns [4]–[6].29 The results show that the predetermined values of political instability affect labour market institutions, though the statistical power of the lagged indicators is somewhat weaker (in one case, the ten years' lagged preference heterogeneity indicator has the wrong sign). In Table 5 we report alternative specifications and estimation methods to control for unobserved country specific shocks and omitted variable bias related to unobservables in the economic and political cycles. In particular, in column [1], we add country specific time trends. In column [2], we implement the Correlated Common Effect Pooled (CCEP) estimator (Pesaran, 2006), which takes into account the possible presence of cross-sectional correlated error terms. Lastly, in column [3], we replace the economic crisis dummy with other business cycle variables such as the output gap and the real exchange rate. The overall results support the previous findings, suggesting that our estimates are not distorted by unobserved shocks or omitted cyclical factors. In columns [4] and [5], we add the lagged unemployment rate and the lagged government deficit to control for the effect of labour market performance and government expenditures on labour market institutions. The inclusion of these (potentially omitted) variables does not change the effect of the political instability indicators. In columns [6]–[8], we check the robustness of the results against the exclusion of selected countries (e.g. New Zealand, Italy and Portugal), which had significant changes in political instability over the period considered. In column [9], we exclude the Nordic countries whose political setting may not be characterised by such high political instability as the indicators may indicate. Finally, in column [10], in order to investigate the medium-run properties of the estimates, we run our regressions using three-year averages. Finally, we enter separately each single indicator we used to construct the aggregate measure of political instability (results are reported in Table 2-B in the Appendix). Note that some caution is needed when interpreting these results since single indicators are likely to be collinear.30 Also, note that the use of closed list and the degree of party fractionalisation in terms of political instability can only be interpreted within a proportional electoral system (see, e.g. Alesina and Glaeser (2004); Persson and Tabellini (2004)). Conditional on the above, lack of a single-winner voting electoral rule, and the degree of dispersion of the government and the opposition in parliament seem to matter, in line with results from Aidt et al. (2006) and Persson et al. (1997).

28 Note that the effect of political turnover on the tax wedge is not statistically different between common and civil law countries. Similarly, the effect of preference heterogeneity on the unemployment benefit replacement rate is not statistically significant between the two law regimes. 29 Note that the use of lagged indicators also mitigates any problem of reverse causality, which may run from labour market institutions to political instability. 30 For this reason we used a principal component analysis to extract the maximum variance out of the pool of indicators.

214

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

Table 4 Predetermined levels of political instability and labour market institutions. 5-year lag

p

−5

γ

−5

p−5 * γ−5 p

10-year lag

[1]

[2]

[3]

[4]

[5]

[6]

μ

τ

ρ

μ

τ

ρ

0.58*** (0.14) −0.01 (0.05) −0.02 (0.04)

−1.90*** (0.32) 0.07 (0.17) −0.06 (0.14)

−8.35*** (1.47) −1.44*** (0.43) −1.01*** (0.35) 0.80*** (0.14) −0.17*** (0.06) −0.01 (0.05) 0.92 451

−1.77*** (0.38) 0.33* (0.19) 0.06 (0.15) 0.95 451

−6.89*** (1.73) 0.40 (0.61) −0.96* (0.52) 0.90 451

−10

γ−10 p−10 * γ−10 R sq. N

0.91 454

0.95 454

0.92 454

Notes: Robust standard errors in parentheses. All specifications include controls as in Table 1 plus country and time fixed effects. Significance levels: *: 10% **: 5% ***: 1%.

6. Concluding remarks There is wide consensus in the public debate that the lack of commitment to reform extensive labour market regulation could be responsible for the low resilience to negative economic shocks and poor labour market performance in OECD countries. In this paper, we focus on the relation between political instability and labour market institutions. We develop a theoretical model in which labour market regulation may arise as a result of political instability. In particular, we show that political instability creates incentives to regulate the labour market and create wage rents. The distortionary effect of regulation induces the government to set the fiscal and public policy to alleviate unemployment.

Table 5 Political instability and labour market institutions: Sensitivity analysis. [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Country spec.

CCEP

Economic

Unemployment

Government

NZL

ITA

PRT

Nordic

3 years

time trends

estimator

cycle

deficit

excl.

excl.

excl.

excl.

average

Panel A — Labour market regulation (μ) p 0.60*** 1.01*** (0.19) (0.12) γ 0.02 0.14*** (0.04) (0.04) p*γ 0.07** 0.20*** (0.03) (0.04) Panel B — Labour tax (τ) p −0.78* (0.47) γ −0.62*** (0.15) p*γ −0.40*** (0.12)

−1.96*** (0.32) −0.80*** (0.17) −0.81*** (0.17)

[10]

1.03*** (0.14) 0.15*** (0.04) 0.22*** (0.04)

0.97*** (0.14) 0.18*** (0.04) 0.22*** (0.04)

0.95*** (0.15) 0.16*** (0.04) 0.22*** (0.04)

1.18*** (0.18) 0.16*** (0.04) 0.17*** (0.04)

0.64*** (0.14) 0.14*** (0.04) 0.13*** (0.03)

1.03*** (0.13) 0.14*** (0.04) 0.20*** (0.04)

1.09*** (0.12) 0.14*** (0.04) 0.21*** (0.04)

1.11*** (0.16) 0.19** (0.07) 0.25*** (0.07)

−2.26*** (0.34) −0.75*** (0.17) −0.75*** (0.17)

−2.30*** (0.33) −0.69*** (0.16) −0.79*** (0.16)

−1.81*** (0.31) −0.82*** (0.18) −0.71*** (0.16)

−1.81*** (0.43) −0.74*** (0.17) −0.89*** (0.17)

−2.17*** (0.47) −0.74*** (0.16) −0.77*** (0.16)

−2.26*** (0.35) −0.95*** (0.19) −0.93*** (0.18)

−2.22*** (0.34) −0.85*** (0.18) −0.92*** (0.18)

−2.41*** (0.51) −0.82** (0.32) −0.88*** (0.33)

−8.48*** (1.42) −2.73*** (0.36) −1.50*** (0.39)

−8.35*** (1.47) −2.92*** (0.37) −1.26*** (0.38)

−13.20*** (1.79) −3.09*** (0.40) −0.21 (0.37)

−1.01 (0.74) −2.68*** (0.35) −0.03 (0.30)

−8.56*** (1.47) −3.09*** (0.38) −1.63*** (0.44)

−8.49*** (1.44) −3.04*** (0.39) −1.68*** (0.43)

−9.03*** (2.50) −3.53*** (0.68) −1.97** (0.81)

Panel C — Unemployment benefit replacement rate (ρ) p −2.36** −8.08*** −8.28*** (1.48) (1.40) (1.01) γ −0.74** −2.88*** −2.89*** (0.33) (0.37) (0.37) p*γ 0.25 −1.37*** −1.38*** (0.30) (0.40) (0.39)

Notes: All specifications include controls as in Table 1 plus country and time fixed effects. In columns [4] and [5] the unemployment rate and government deficit are lagged by one year. OLS estimates with robust standard errors in parentheses. Significance levels: *: 10% **: 5% ***: 1%.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

215

We empirically evaluate these predictions using panel data for 21 OECD countries. Our results are consistent with the view that political instability is associated with higher labour market regulation, lower labour taxation, and a lower unemployment benefit replacement rate. We find that, for the average OECD country, an increase by one standard deviation in political instability is associated with: an increase of 1.2 points in the indicator of labour market regulation (which is about 90% of the total variation occurring over the whole period in the OECD countries), a decrease in both labour taxation and unemployment benefit replacement rate by 3 and 11 percentage points, respectively. Notice that this empirical evidence may also be consistent with alternative views in the political economy literature, regarding the links between the political process and labour market institutions (see e.g. Alesina and Glaeser (2004); Boeri et al. (2012)). Indeed, while the idea that the political process is an important determinant of labour market institutions is not new, the political instability channel has not received enough attention so far. This paper tries to fill this gap. One relevant implication of our findings is that the political process can be an important factor in explaining the heterogeneity in unemployment patterns across OECD countries. This suggests that attempts made by international organisations to urge governments to reform their labour market institutions to alleviate long-term unemployment should take into account their political systems. Political systems characterised by higher turnover and polarisation are likely to experience greater resistance to reforming their labour market institutions. This may also explain why reforms that minimise the opposition from social groups with vested interests, such as two-tier reforms, tend to be more successful. The main findings of the paper may be relevant for a number of extensions and future research. First, this paper applies the general view that taxation and regulation are two alternative instruments a government may handle to reach its policy objectives (Oates, 2002). While we focus on the labour market, it would be interesting to extend our study to other specific sectors or markets– e.g. the banking sector, the environmental sector, or the product market. Second, while our analysis neglects strategic interactions between countries in an open economy, it would also be worth assessing the effect of political instability when governments compete on fiscal and regulatory policies. Finally, while in this paper we assume that labour is the only production factor (there is no capital), and that the workforce is homogeneous in terms of skills, it would be interesting to consider the implications for taxation and regulatory activities of having more than one factor of production and heterogeneity in skills. All these important and interesting topics are left for future research. Acknowledgements We are particularly grateful to the editor, Prof. Toke Aidt and two anonymous referees for their valuable comments. We also received useful comments and suggestions from Andrea Bassanini, Kristian Behrens, Michel Beine, Georges Bressons, Eve Caroli, Quentin David, Domenico Delli Gatti, Damien Gaumont, Etienne Lehmann, Michele Longo, Marco Lossani, Maurizio Motolese, Yasusada Murata, Yasuhiro Sato, Bruno van der Linden, David Wildasin and Skerdilajda Zanaj. A previous version of this paper was presented at various international conferences: ‘Economic Integration, Trade, and Spatial Structure’ (Nagoya University, Japan), 8th Journées Louis-André Gérard-Varet (Marseille, France), 2009 Annual Conference of the European Association of Labour Economists (Tallin, Estonia), REPLHA International Conference (Catholic University of Milano, Italy) and in seminars at Kyoto University, Università Cattolica, Université de Paris 2, Université du Luxembourg and Université de Paris X. We are grateful to Andrea Bassanini and Romain Duval for providing the ‘OECD Labour Market Institutions Data’. Access to World Bank political indicators' data is also gratefully acknowledged. Part of this paper was completed when Claudio Lucifora was visiting the School of Economics at UNSW and when Simone Moriconi was visiting the University of California Davis. The usual disclaimer applies. Appendix A. Theoretical model Appendix A.1. Expected utility from public goods From Eq. (6), the constituency B derives utilities (1 − γ)hi and γfi when party i = a, b sets hi, fi respectively. If party b is in office at stage 1, this implements the provision of public good preferred by constituency B B

g b ¼ min½ð1−γÞhb ; γf b 

with hb þ f b ¼ z; :

ðA  1Þ

From Eq. (A-1), and since γ ∈ (1/2, 1], party b sets a composition of public good that makes the utility obtained by constituency B from consumption of h at least as high as the utility it gets from consumption of f i.e. ð1−γÞhb ¼ γf b ;

ðA  2Þ

which implies party b's optimal allocation of spending on public goods z = hb + fb:



f b ¼ ð1−γÞz;



hb ¼ γz:

ðA  3Þ

216

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

With γ N 1/2, party b allocates a higher share of z to public good h. If we substitute Eq. (A-3) into Eq. (5), we obtain the utility of constituency A from (fb⁎, hb⁎): A



Gb ¼

   h i 1−γ 1 1 A 2 2 gb ¼ min γ z; ð1−γÞ z ¼ z: γð1−γÞ γð1−γÞ γ

ðA  4Þ

It readily follows that GAb b GAa because γ ∈ [1/2, 1]. Appendix A.2. Proof of Proposition 1 The second order condition for the concavity of the welfare function is: ″

0



V ττ ¼ 2δlc þ ½w−v þ δðτ þ vÞlcc b 0:

ðA  5Þ

The negativity of the second order derivative confirms that V is strictly concave in the labour tax, thus V 0τ is invertible. Let T := −1 be defined for any μ N 0 and vNv such that V 0τ = 0, i.e. the maximisation problem (11) admits an interior solution τ* = V 0τ T(μ , v) for any μ N 0 and vNv as stated by Proposition 1. When μ = 0 and v ¼ v, from Eq. (1) we have w ¼ v ¼ v, and from the first order condition (13) there follows



0

δl þ δðτ þ vÞlc ¼ 0;

ðA  6Þ

which implicitly defines τmax as in statement (i) of Proposition 1. From Eq. (A-6), when τ = τmax the following holds  max 0 l¼− τ þ v lc :

ðA  7Þ 0

Now, define the elasticity of labour demand to labour costs as ϵlc = − lc * c/l. Evaluate ϵlc at τ = τmax. In this case, μ = 0 and v ¼ v. Therefore, total labour costs are c ¼ τ max þ v. From Eq. (A-7) we have  ϵlc jτmax ¼ −

 τmax þ v l0c ¼ 1: l

ðA  8Þ

This implies that any tax increase evaluated at lðτmax þ vÞ causes a proportional decrease in the tax base, which leaves tax revenues unchanged. Evaluate ϵlc at τ = τ*. In this case, μ N 0 and vNv, accordingly wages result from Eq. (1), and labour costs are c = τ* + w. So we have: ϵlc jτ ¼ −

ðτ þ wÞl0c ; l

ðA  9Þ

where at μ N 0 and vNv, from Eq. (13) l¼−

0    lc  w−v þ δ τ þ v : δ

ðA  10Þ

Insert Eq. (A-10) in Eq. (A-9), after some substitutions obtain ϵlc jτ ¼ 1 þ

ðδ−1Þðw−vÞ N ϵlc jτmax ; 2δ þ w þ vðδ−1Þ

ðA  11Þ

which implies τ* b τmax, as in statement (ii) of Proposition 1. We finally apply the implicit function theorem: 0

Tμ ¼ − 0 Tv

∂V ″τμ ∂V ″ττ ″

¼−

∂V τμ ∂V ″ττ

0 ðδ þ 1Þl0c þ ½w−v þ δðτ  þ vÞl″cc −W μ ½ðδ þ 1Þ þ Ω ¼ b 0; 2δ þ Ω 2δl0c þ ½w−v þ δðτ þ vÞl″cc h i 0 0  ″ ″ W v ðδ þ 1Þlc þ ðw−v þ δðτ þ vÞÞlcc −ðδ−1Þlcc W 0 0 δ−1 ¼− ¼ 0v T μ − b 0; Wμ 2δ þ Ω 2δl0c þ ½w−v þ δðτ þ vÞl″cc 0

¼ −W μ



where Ω ¼ ½w−v þ δðτ þ vÞ llcc0 N0, which proves statement (iii) of Proposition 1. c

ðA  12Þ

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

217

Appendix A.3. Proof of Proposition 2 When v ¼ v, party a's maximisation problem is h i A max E V ≡ v þ ðw−vÞLðμ; vÞ þ ωðδ; p; γ ÞZ ðμ; vÞ

ðA  13Þ

μ

where Lðμ; vÞ ¼ lðw þ T ðμ; vÞÞ

ðA  14Þ

Z ðμ; vÞ ¼ T ðμ; vÞLðμ; vÞ−½1−Lðμ; vÞv: The first derivative of the objective function (A-13) is 0

0

0

0

V μ ¼ W μ L þ ðw−vÞLμ þ ωðδ; p; γÞZ μ :

ðA  15Þ

From Eqs. (A-14) and (A-12), the derivatives with respect to μ of the left hand sides of the two equations in (A-14) are h i 0 0 0 0 Lμ ¼ lc W μ þ T μ b 0; 0

0

ðA  16Þ

0

Z μ ¼ T μ Lðμ; vÞ þ ½T ðμ; vÞ þ vLμ b 0:

Inserting Eqs. (A-16) and (A-12) into Eq. (A-15) and evaluating this at the optimal tax rule (13) yields, after some simplifications, Eq. (17). When ϕ(p, γ) = 1, Eq. (17) becomes 0

Vμ ¼

0   0 ðw−vÞðδ−1Þlc þ ðδ−1Þ τ þ v lc b 0; δ

ðA  17Þ

that is, labour market regulation has a negative impact on welfare, thus the government chooses the corner solution μ* = 0, which proves statement (i) in Proposition 2. When ϕ(p, γ) b 1, then, provided that V ″μμ b 0, the function V 0μ is invertible, and there exists a function M := (V 0μ )−1 in a contour of p b 1, γ b 1 such that V 0μ = 0. Then the maximisation problem (A-13) admits an interior solution μ* = M(p, γ) N 0 for any pNp and γNγ, which proves statement (ii). Finally, applying the implicit function theorem yields " 0

Mk ¼ −

V ″μ j V ″μμ

ω0j ðτ þ vÞ−ϕ0j ¼−

# ðw−vÞT 0μ 0 lc w0μ

V ″μμ

N 0;

k ¼ p; γ;

ðA  18Þ

which proves statement (iii) of Proposition 2. Appendix A.4. Proof of Proposition 3 Rewrite the optimal tax rule (13) when v = ρ w: 0

δl ¼ −½wð1−ρÞ þ δðτ þ ρwÞlc ; which defines the equilibrium labour tax T(μ, ρ). Applying the implicit function theorem yields −Ψ0μ ½ðδ þ 1Þ þ ρðδ−1Þ þ Ω b 0; 2δ þ Ω 0 Ψρ 0 wðδ−1Þ 0 Tρ ¼ 0 Tμ − b 0: Ψρ 2δ þ Ω 0

Tμ ¼

The government at stage 0 chooses μ and ρ to maximise the objective function (21) subject to its budget constraint and the optimal tax rule at stage 1. After some simplifications,

0

Vμ ¼

  0 wT 0μ l0c ð1−ωÞρ ð1−α Þð1−ρÞwl0c 0 þ −ð1−α−ωÞ τ þ ρw lc þ ð1−α−ϕÞ ; þ απ0 þ ð1−α−ρϕÞwlc − δ Ψ0μ 1−ρ

ðA  19Þ

218

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221 0

0

0

V ρ ¼ Ψρ V Iμ − 0

Ψ0





ðω−1Þð1−l Þw αwðπ0 −l Þ 0 0 þ ð1−α−ϕÞwT^ ρ lc − : 1−ρ 1−ρ

ðA  20Þ

ðδ−1Þ . Provided that the second order conditions are satisfied, when ϕ(p, γ) b 1, the first order condition where T^ ρ ¼ T ρ − Ψρ0 T μ ¼ − w2δþΩ μ 0 ″ b 0, it V μ = 0 identifies an interior solution, the first term in Eq. (A-20) disappears, and we obtain Eq. (22) in the main text. As Vρρ 0  follows that Vρ′ is invertible and there exists a function ϒ := ðVρ0 Þ−1 a contour of p b 1, γ b 1 such that V ρ μ  ¼ 0. Applying the implicit function theorem yields

0

0

ϒk ¼ −

V ″ρk V ″ρρ

0

¼−



 0 0  w δð1−l Þ wðδ−1Þlc ðπ −l Þ 0 0 wðδ−1Þlc þ αk − 0 ϕk − þ ; ″ 2δ þ Ω 2δ þ Ω 1−ρ 1−ρ V ρρ

k ¼ p; γ:

It readily follows that 0

ϒk b 0

if

0

0

α k N ϕk Θ where Θ ¼

 0 −δð1−l Þð2δ þ ΩÞ þ ð1−ρÞwðδ−1Þlc b 0;  −ðπ0 −l Þð2δ þ ΩÞ þ ð1−ρÞwðδ−1Þl0c

which proves Proposition 3. Appendix B. Data Appendix The following variables are drawn from the World Bank DPI (capital letters denote the original variables in the DPI. See Beck et al. (2001) for more details). Pr = 1: if a system of proportional representation is adopted (OUSESYS = 0). ProPlu = 1: if a mixed plurality and proportional system is adopted but the majority of seats in the House and the Senate are proportional (HOUSESYS = 0 and SENSYS = 0). CL = 1: if closed lists are used. Nowtal = 1: if a winner-takes-all system is not adopted (PLURALITY = 0). Invmaj: inverse margin of majority of the executive over the opposition in parliament (1/MAJ). Stabs: percentage of veto players who drop from the government in a given year. Applying a principal component analysis to these variables, we extract the eigenvector associated with the first eigenvalue (43% of the total variance is explained by the first component), to obtain the following index of political turnover: 0:55  Pr þ 0:54  PrPlu þ 0:39  CL þ 0:48  Nowtal þ 0:12  Invmaj þ 0:04  Stabs:

ðB  1Þ

As for preference heterogeneity, the variables used in the principal component analysis are the following. Maxpolar = 1: if there is a maximum ideological distance between the executive party and the four main parties in the legislature (POLARIZ = 2). Govdisp, Oppdisp: reciprocal of the Herfindal index of the government, respectively, the opposition; that is, the inverse of the sum of the squares of the shares of the seats of each governing, respectively, opposition, party (1/HERFGOV, 1/HERFOPP). Govfrac, Oppfrac: probability that two deputies picked at random from among the parties in parliament will be from different parties in the government, respectively, the opposition. The score drawn out of the first component (i.e. accounting for 57% of the total variance) leads to the following index of preference heterogeneity: 0:22  Maxpolar þ 0:52  Govdisp þ 0:46  Oppdisp þ 0:51  Govfrac þ 0:45  Oppfrac:

ðB  2Þ

Both variables were rescaled to have zero mean and unit standard deviation as follows: X std ¼ X−X σ X where X and σX are, respectively, the average and standard deviation of X = p, γ over the sample including the 21 OECD countries for the period 1975–2006. Note that the length of the sample period used for the normalisation is motivated by the use of lagged variables throughout the analysis and ensures that the unit of measure used for the political variables is fully consistent with the institutional variables (see below). Ydem: categorical variable for the youth of the democratic institutions. Equal to 3 if the democracy has been in place for less than 20 years (TENSYS b 20); equal to 2 if the democracy has been in place for 20–40 years (20 ≤ TENSYS b 40); equal to 1 if the democracy has been in place for more than 40 years (TENSYS ≥ 40). Prtyage: categorical variable for the age of the parties in parliament. Equal to 1 if PARTYAGE b 20; equal to 2 if 20 ≤ PARTYAGE b 40; equal to 3 if PARTYAGE ≥ 40. Yrcurnt, Yrsoffc: respectively years remaining of the current legislature and years of office of the current government.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

219

Civil, Common: dummy variable that takes the value 1 if the origin of the legal system is civil law and 0 otherwise, and a dummy variable that takes the value 1 if the origin of the legal system is common law and 0 otherwise. The definitions are based on Botero et al. (2004). The information on labour market institutions is drawn from the OECDLabour Market Institutions Database (LMID, see Bassanini and Duval (2009)). Since the relevant time series stop in 2003, we updated them to 2006 using original OECD data sources. The synthetic indicator of labour market regulation is derived from the following indicators for regulation in the area of employment and collective relations law: Eplr: 1–5 summary indicator of the stringency of employment protection legislation of regular workers (Sources: LMID; OECD Employment Outlook, 2004 and 2009). LackCorp: indicator for lack of corporatism in the wage bargaining process, which takes the value 1 for high levels of corporatism, 2 for intermediate levels, and 3 for a low level of corporatism (LMID, OECD Employment Outlook, 2004 and 2009). Undens: Trade union density rate, that is, the percentage share of workers affiliated with a trade union (Sources: LMID; OECD Employment Outlook, 2004 and 2009). These three indicators are standardised to have zero mean and unit standard deviation and summed to obtain the aggregate indicator of labour market regulation. Ltax: marginal effective tax rate on labour, Carey and Rabesona (careyrabesona) (Sources: LMID; OECD National Accounts and the OECD Revenue Statistics). Arr: average unemployment benefit replacement rate, which describes the replacement rate during the first year of unemployment, and the duration of the monetary transfers aggregated over family types (Sources: LMID; OECD Benefits and Wages Database, 2007). EU, Euro: dummies equal to 1 if a country is a member of the European Union and the European Monetary Union, respectively. Crisis = 1: if effective output falls 4 standard deviations below its potential level. Additional controls used in the analysis include standardised unemployment rate (number of unemployed persons as a percentage of the civilian labour force for all countries except Austria, where we use the commonly used definition; Sources: LMID; OECD Main Economic Indicators); the current deficit (difference between cyclically adjusted current disbursement and cyclically adjusted current receipts of the general government; sources: LMID; OECD National Accounts 2010); the output gap (percentage deviation of effective output from its potential level; sources: LMID; OECD Economic Outlook, 2009); the real effective exchange rate (Source: OECD, Main Economic Indicators) and economic shocks to total factor productivity, terms of trade, real interest rate, and labour demand (Source: LMID). Table B-1 Means and standard deviations of political instability indicators by country: OECD, 1985–2006. Country

p

γ

H

Australia

−0.14 (0.02) 0.93 (0.05) 0.96 (0.03) −1.72 (0.05) −0.23 (0.02) 0.38 (0.03) 0.58 (0.06) 0.41 (0.05) 0.44 (0.04) −1.72 (0.07) −1.73 (0.04) −0.10 (0.05) 0.51 (0.04)

−0.75 (0.30) 0.07 (0.24) 2.36 (0.55) −0.66 (0.25) 2.04 (0.96) 0.10 (0.15) 0.67 (0.12) −0.60 (0.27) 0.74 (0.33) 0.05 (0.60) −1.01 (0.11) −1.08 (0.15) −0.51 (0.36)

−0.89 (0.30) 1.00 (0.21) 3.32 (0.53) −2.38 (0.29) 1.81 (0.96) 0.49 (0.15) 1.25 (0.10) −0.19 (0.29) 1.18 (0.30) −1.66 (0.61) −2.74 (0.10) −1.18 (0.16) −0.01 (0.36)

Austria Belgium Canada Switzerland Germany Denmark Spain Finland France UK Greece Ireland

(continued on next page)

220

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

Table B-1 (continued) Country

p

γ

H

Italy

0.26 (0.64) −0.34 (0.05) 0.98 (0.05) 1.15 (0.13) −0.82 (0.69) 1.02 (0.09) 1.02 (0.04) −1.65 (0.06) 0.00 1.00

−0.66 (0.45) −0.49 (0.43) 0.61 (0.40) 0.57 (0.32) −0.49 (0.71) −0.48 (0.40) 0.54 (0.21) −0.83 (0.31) 0.00 1.00

−0.41 (1.05) −0.83 (0.44) 1.59 (0.40) 1.71 (0.40) −1.31 (1.33) 0.54 (0.46) 1.56 (0.20) −2.48 (0.36) 0.00 1.69

Japan The Netherlands Norway New Zealand Portugal Sweden US Total Notes: Standard deviations in parentheses.

Table B-2 Political instability and labour market institutions: Single indicators. Political turnover

Stabs (p) Invmaj (p) CL (p) Pr (p) PrPlu (p) Nowtal (p)

Preference heterogeneity

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

μ

τ

ρ

μ

τ

ρ

μ

τ

ρ

−0.01 (0.07) −0.05 (0.06) −0.24 (0.26) 1.21*** (0.34) 0.10 (0.08) 1.79*** (0.25)

−0.02 (0.26) −0.04 (0.28) 9.61*** (0.69) −12.26*** (1.00) 0.15 (0.51) −3.29*** (0.46)

−0.97 (0.66) −0.66 (0.61) 1.80 (1.40) −5.90*** (1.62) 5.14** (2.19) −27.44*** (2.10) −1.72* (0.93) −1.52*** (0.18) −2.36 (1.84) −7.55** (3.11) 0.55 (0.46) 0.92 456

−0.00 (0.07) −0.05 (0.06) −0.25 (0.24) 1.31*** (0.32) 0.02 (0.09) 1.92*** (0.26) 0.16*** (0.06) 0.10*** (0.03) −0.61** (0.28) −0.33 (0.21) 0.02 (0.07) 0.94 456

0.06 (0.26) −0.08 (0.28) 10.72*** (0.76) −12.61*** (1.04) −0.02 (0.44) −1.69*** (0.58) −0.62** (0.31) −0.11 (0.08) −0.93 (0.70) −0.93 (0.96) 0.03 (0.22) 0.95 456

−0.92 (0.65) −0.70 (0.65) 2.69* (1.43) −6.34*** (1.60) 5.36** (2.12) −25.85*** (2.30) −2.66*** (0.79) −1.80*** (0.15) 4.36*** (1.41) 2.74 (2.60) 0.58 (0.45) 0.95 456

Govdisp (γ) Oppdisp (γ) Oppfrac (γ) Govfrac (γ) Maxpolar (γ) R sq. N

Turnover and heterogeneity

[1]

0.94 456

0.95 456

0.94 456

0.05 (0.08) 0.07* (0.03) 0.21 (0.30) 0.76** (0.30) 0.06 (0.08) 0.92 456

−0.47 (0.32) −0.06 (0.08) −2.06*** (0.60) −2.03** (0.90) −0.15 (0.20) 0.95 456

Notes: Robust standard errors in parentheses. All specifications include controls as in Table 1 plus country and time fixed effects. Significance levels: *: 10% **: 5% ***: 1%.

References Abreu, R., David, F., 2010. Socially responsible behaviour: labour market in Portugal. Polytechnical Stud. Rev. 8 (13), 103–127. Aidt, T.S., Dutta, J., 2007. Policy myopia and economic growth. Eur. J. Polit. Econ. 23, 734–753. Aidt, T.S., Tzannatos, Z., 2008. Trade unions, collective bargaining and macroeconomic performance: a review. Ind. Relat. J. 39 (4), 258–295. Aidt, T.S., Dutta, J., Loukoianova, E., 2006. Democracy comes to Europe: franchise extension and fiscal outcomes 1830–1938. Eur. Econ. Rev. 50, 249–283. Alesina, A., Glaeser, E., 2004. Fighting Poverty in the U.S. and Europe: A World of Difference. Oxford University Press, Oxford. Alesina, A., Perotti, R., 1996. Income distribution, political instability and investments. Eur. Econ. Rev. 40 (6), 1203–1228. Alesina, A., Tabellini, G., 1990. Voting on the budget deficit. Am. Econ. Rev. 80, 37–49. Arpaia, A., Mourre, G., 2010. Institutions and performance in European labour markets: taking a fresh look at evidence. J. Econ. Surv. 26 (1), 2–48. Azzimonti, M., 2011. Barriers to investment in polarized societies. Am. Econ. Rev. 101, 2182–2204. Bassanini, A., Duval, R., 2009. Unemployment, institutions and reform complementarities: re-assessing the aggregate evidence for OECD countries. Oxf. Rev. Econ. Policy 25 (1), 40–59.

C. Lucifora, S. Moriconi / European Journal of Political Economy 39 (2015) 201–221

221

Beck, T., Clarke, G., Groff, A., Keefer, P., Walsh, P., 2001. New tools in comparative political economy: the database of political institutions. World Bank Econ. Rev. 15, 165–176. Blanchard, O., Summers, L., 1986. Hysteresis and the European Unemployment Problem. NBER Working Paper No. 1950. Boeri, T., Conde-Ruiz, J.I., Galasso, V., 2012. The political economy of flexicurity. J. Eur. Econ. Assoc. 10 (4), 684–715. Botero, J., Djankov, S., La Porta, R., Lopez de Silanes, F., Shleifer, A., 2004. The regulation of labor. Q. J. Econ. 119 (4), 1339–1382. Busse, M., Hefeker, C., 2007. Political risk, institutions and foreign direct investment. Eur. J. Polit. Econ. 23 (2), 397–415. Carey, D., Rabesona, J., 2002. Tax ratios on labour and capital income and on consumption. OECD Economic Studies No. 35, 2002/2. Carmignani, F., 2003. Political instability, uncertainty and economics. J. Econ. Surv. 17, 1–54. Castles, F.G., Mitchell, D., 1993. Worlds of welfare and families of nations. In: Castles, F.G. (Ed.), Families of Nations: Patterns of Public Policy in Western Democracies. Dartmouth, Aldershot. Coe, D.T., Snower, D.J., 1997. Policy complementarities: the case for fundamental labour market reform. IMF Staff. Pap. 44 (1), 1–35. Dal Bó, E., Rossi, M., 2011. Term length and the effort of politicians. Rev. Econ. Stud. 78 (4), 1237–1263. Deroose, S., Flores, E., Turrini, A. (Eds.), 2006. Proceedings from the ECFIN Workshop ‘The Budgetary Implications of Structural Reforms’. European Economy—European Commission Directorate-General for Economic and Financial Affairs Publications. Devereux, M., Wen, J.F., 1998. Political instability, capital taxation and growth. Eur. Econ. Rev. 42, 1635–1651. Duval, R., 2008. Is there a role for macroeconomic policy in fostering structural reforms? Panel evidence from OECD countries over the past two decades. Eur. J. Polit. Econ. 24, 491–502. Enikolopov, R., Zhuravskaya, E., 2007. Decentralization and political institutions. J. Public Econ. 91, 2261–2290. European Commission, 2013. Annual Growth Survey 2013: Charting the Course to Recovery. Freeman, R.B., 2008. Labor Market Institutions Around the World. LSE-CEP Discussion Paper No. 844. Galasso, V., 2014. The role of political partisanship during economic crises. Public Choice 158 (1-2), 143–165. Glaeser, E., Shleifer, A., 2002. Legal origins. Q. J. Econ. 117, 1193–1230. Hessami, Z., Baskaran, T., 2014. Has globalization affected collective bargaining? An empirical test, 1980–2009. World Econ. http://dx.doi.org/10.1111/twec.12239. Høj, J., Galasso, V., Nicoletti, G., Dang, T., 2006. The Political Economy of Structural Reforms: Empirical Evidence from OECD Countries. OECD Economics Department Working Paper, No. 501. Holsey, C.M., Borcherding, T.E., 1997. Why does government's share of national income grow? An assessment of the recent literature on the U.S. In: Mueller, D. (Ed.), Perspectives on Public Choice. Cambridge University Press, New York, pp. 562–589. Mortensen, D., Pissarides, C., 2001. Unemployment responses to ‘skill-biased’ shocks: the role of labor market policy. Econ. J. 109, 242–265. Nunziata, L., 2005. Institutions and wage determination: a multi-country approach. Oxf. Bull. Econ. Stat. 67, 435–466. Oates, W.E., 2002. Fiscal and regulatory competition: theory and evidence. Perspekt. Wirtsch. 3 (4), 377–390. OECD, 2012. OECD Employment Outlook 2012. OECD Publishing http://dx.doi.org/10.1787/empl_outlook-2012-en. Pagano, M., Volpin, P., 2006. The political economy of corporate governance. Am. Econ. Rev. 95 (4), 1005–1030. Persson, T., Svensson, L.E.O., 1989. Why a stubborn conservative would run a deficit: policy with time-inconsistent preferences. Q. J. Econ. 104 (2), 325–345. Persson, T., Tabellini, G., 2000. Political Economics: Explaining Economic Policy. MIT Press, Cambridge, Mass. Persson, T., Tabellini, G., 2004. Constitutions and economic policy. J. Econ. Perspect. 18 (1), 75–98. Persson, T., Tabellini, G., 2006. Electoral Systems and Economic Policy. In: Weingast, B.R., Wittman, D.A. (Eds.), Oxford Handbook of Political Economy. Oxford University Press. Persson, T., Roland, G., Tabellini, G., 1997. Separation of powers and political accountability. Q. J. Econ. 112 (4), 1163–1202. Persson, T., Roland, G., Tabellini, G., 2000. Comparative politics and public finance. J. Polit. Econ. 108 (6), 1121–1161. Pesaran, M.H., 2006. Estimation and inference in large heterogeneous panels with a multi-factor error structure. Econometrica 74 (1), 967–1012. Raess, D., 2014. Export dependence and institutional change in wage bargaining in Germany. Int. Stud. Q. http://dx.doi.org/10.1111/isqu.12096. Saint-Paul, G., 1996. Exploring the political economy of labour market institutions. Econ. Policy 23, 265–315. Stern, N., 1987. The effects of taxation, price control and government contracts in oligopoly and monopolistic competition. J. Public Econ. 32, 133–158. Svensson, L.E.O., 1998. Investment, property rights and political instability: theory and evidence. Eur. Econ. Rev. 42, 1317–1341. Tanzi, V., Schuknecht, L., 2000. Public Spending in the 20th Century: A Global Perspective. Cambridge University Press, Cambridge. Tsebelis, G., 1995. Decision-making in political systems: veto players in presidentialism, parliamentarism, multicameralism and multipartyism. Br. J. Polit. Sci. 25, 289–325. Tsebelis, G., 2002. Veto Players: How Political Institutions Work. Princeton University Press, Princeton, N.J. Tsebelis, G., Chang, E., 2004. Veto players and the structure of budgets in advanced industrialized countries. Eur. J. Polit. Res. 43 (3), 449–476. Wright, R., 1986. The redistributive roles of unemployment insurance and the dynamics of voting. J. Public Econ. 31, 377–399.