jŕ Labour markets in Eastern Europe Michael Burda Summary K3£ l%- Unemployment is a major concern in the transforming economies of Central and Eastern Europe. The modern 'flowjipproacjťjo labour markets suggests both that unemployment is an important component of the transformation process and that the labour market institutions adopted will influence the rate of unemployment in the long run. To date, Eastern European countries exhibit considerable ALvergences in several of these institutions: especially in unemploy-ment benefit systems, collective bargaining structures, and active labour marketp^icies. '~~~~~~Än aggregate matching function is successfully estimated f or data from Czech and Slovak em^öyinenVöffices. Emerging labour markets in theJEa^j^ictijmjwt^ those in the West. The implied dynamics make both 'big bang' and 'benign neglect' unattractive strategies for transformation: a 'mixed bang' is more appropriate. Quantitative evidence about the effects on unemployment of different labour market institutions in OECD countries is used to make long-term projections of equilibrium rates of unemployment in Central and Eastern Europe, given the labour market institutions now in place there. With the possible exception of the Czech Republic, high unemployment is likely to be pervasive and persistent. Unemployment, labour markets and structural change in Eastern Europe Michael Burda INSEAD, Wissenschaftszentrum Berlin and CEPR 1. Introduction The transition to a market economy remains in its infancy in Central and Eastern Europe. Privatization of large state enterprises, a central aspect of transition and a stated goal of most governments, has been painfully slow. Largely, this is because the associated 'restructuring' entails closing factories and displacing large numbers of workers, many of whom may leave the labour force for ever. Some will find jobs in the expanding private sector, but often oi^)^fter_changing occupations, industries, and locations on a scale rarely seen in advanced Western economies. This paper addresses the mediating role of labour markets and unemployment in the transforming economies. Unemployment is not merely a by-product: it is necessary for the transformation. This is true for several reasons. First, emerging unemployment will offset a growing imbalance in bargaining power of workers over managers in the aftermath of central planning. Unemployment provides the 'disciplining device', as in Shapiro and Stiglitz (1984), to raise effort and productivity 1 J This research was conducted while a Research Fellow of the German Marshall Fund of the US and was also supported by grants to the CEPR by the Ford Foundation (no. 910-0380) and the SPES Program of the Commission of the European Communities (no. E/90100033/PRO). I am grateful to INSEAD and the Wissenschaftszentrum Berlin for logistical support; to Donata Hoesch for research assistance; to the editors, Nick Barr, Charlie Bean, Olivier Blanchard, Tito Boeri, Saul Estrin, John Flemming, Richard Freeman, Olaf Hübler, Hartmut Lehmann, Chris Pissarides, David Soskice, Jan Svenjar, and seminar participants at INSEAD, the LSE and Hannover for valuable comments and discussions. Special thanks for useful conversations and invaluable help with the data are due to Josif Bati, Christo Christov, Daniel Daianu, Gheorghe Damian, Milcho Dimitrov, Elka Dimitrova, Vladislav Flek, Mária Frey, Marie Frýdmanova, Marek Góra, Krzysztof Hagemejer, K. Hampartsumian, Lajos Héthy, Milan Horálek, Julliana Ilieva, Kamil Janáček, Vassil Kountchev, Mária Ladó, Terez Laky, Zbislav Moravek, Sabina Nákova, Ovidiu Nanes, Nikolai Nikolov, Pavol Ochotnický, Nikolai Pavlov, Steliana Perz, Zbigniew Polanski, Cristian Popa, Dimiter ll**vvvS mumcu' in Quälern Europe 103 jjjpb Western levels. Second, unemployment may be necessary to check jHfthe growth of real wages (Blanchard, 1991). Indeed, open unemploy-i||[rnentf may be the only effective means to this end in the presence of :^jjecalcitränt or nonexistent coljecdye bargaining partners. jBf A third reason is offered by the new macroeconomics literature on jBf the matching or flow approach to labour markets, in which job creation ^9|is'seen as a stable function of the stock of unemployed workers and JRoE611 Vf^cancies.1 Unemployment is needed to allow the emergence of l|§fa new private sector and to force the acquisition of new skills by workers, ^^^gf? restructuring requires new job" matches* wrhich in turn requires .fjpijore unemployment; At the same time, more unemployment at given lllyeveis of vacancies makesthe latter 'too efficient', causes labour-market IjSoongestion and leads joJónj^tejinjLinem|^yment. To the extent the Ifiříng^ decision is~in public hands, the state can influence this process greatly. Postponing layoffs reduces congestion, but at a cost: less job greation in the higher-productivity private sector. Is there an optimal jtpeed at which to release workers into the unemployment pool? ipvTo manage labour relations and mediate unemployment resulting Brom the transformation, labour markets need institutions; these must joften be developed de novo with structures different from those in advanced industrial economies. During radical structural change, what Ifs the correct level of unemployment benefit, severance protection, itrade union representation or centralization in collective bargaining? l^hat 'active employment measures' should be offered, or forced upon, I' employment benefit recipients, and under what conditions? How 1 these programmes be financed? i examine both sets of questions. First, I confirm the usefulness of ^matching function approach, using data from the Czech and Slovak publics. A dynamic analysis of Eastern European labour markets ses questions about the rate of closure of state firms, many with gative value-added at world prices and incapable of profitable Operon. A key finding is that neither 'shock treatment' nor 'g;o slow' is ; best approach: the right policy is"a step increase in unemployment igkick-startthe matching process in tee"private sector, then a slow, ^ntrqlled release of workers into unemployment. This conclusion has ffiralíel^implications for the pace of privatization. |||Second, I discuss labour jnarket policy and institutional design in gastern Europe using the experience of OĚČD countries. Roughly 75% m cross-sectional variance in standardized OECD unemployment rates IgFór recent reviews see Pissarides (1990) and Blanchard and Diamond (1992). For empirical ^evidence on gross flows in labour markets, see Blanchard and Diamond HPQrn tt! r.,t-^ -.^j q. tiaei da can be explained by three things: the degree of corporatism in collective bargaining, the generosity of unemployment Fenefitsi and per capita ji spending on active employment measures. Similar results obtain for the fraction unemployed in long spells. I use these results to forecast equilibrium unemployment rates in Eastern Europe. Section 2 summarizes the current labour market situation and the debate on the nature of unemployment in these countries,2 I stress the importance oflHTmätffi^T&nctiön in getting the private sector off the ground. SectionTanalysesTqullibrium unemployment and vacancies and interprets their evolution in Eastern Europe in the past three years. Section 4 reviews labour market institutions, making crosscountry comparisons. Section 5 explores policy implications. I estimate a matching function with flow data from Czech and Slovak labour markets, then draw these results into a normative analysis: how rapidly should the state sector be shrunk? Finally, I present tentative forecasts fol^equllibřium unemployment rates and long-term unemployment. Section 6 offers conclusions. 2. The labour market and unemployment during the transformation 2.1. The current situation: recession, supply shock or restructuring? Table 1 shows that the transformation has been tough going. Measured GDP fell in 1991 and will do so again in 1992 (except possibly in the CSFR). Declines in industrial output were dramatic but independent of the pace of reform (CEC, 1992b). The collapse was a direct consequence of the end of COMECON and the shift of economic activity from a lattice of rigid trading relationships of central planning to a market environment of imperfect information placing value on search and brokering of business contacts?Conyentipnal supply and demand analysis fails to capture this key aspect of economic transformation. This diagnosis applies a fortiori to the labour market. Firms slashed output and, after some delay, bejajito sh.ed labour. Despite considerable interfirm mobility under communism, labour markets were not ready for" unemployment and became very congested. A crude indicator is the ratio of registered vacancies to" unemployed shown in Table 2. Despite the common view that 'there is no new job creation in Eastern Europe', labour markets exhibit both job seekers and openings to an I----------------1 2 For practical reasons, this study is restricted to the 'Visegrad 5' (Poland, CSFR, Hungary, Romania and Bulgaria). Many of my findings and conclusions are relevant for the republics of the former ur 7 'ts 11 :,tern___ope I Table 1. Economic indicators for Central and Eastern Europe, mid-1992 Bulgaria CSFR Hungary Poland Romania 1991 % Change in 1992 Unemployment Industrial Change in % of labour Real GDP production Employment thousands force -17 -16 -10 -9 -14 -27.8 -21.2 -19.0 -11.9 -18.7 14.5 ť7-4 -6.0 -5.5 ■11.6 +210 + 154 +207 +693 +563 11.8 5.6 10.1 12.3 6.2 Source: OECD (1992), CEC (1992a, b), national labour ministries. Notes: Unemployment change in thousands is year to March (Bulgaria), to May r(Hungary), to June (Poland), to July (CSFR and Romania); % of labour force is at June 1992 except Romania (May). l§ Table 2. Registered vacancies and unemployment in mid-1992 Month Vacancies (v) (000s) Unemployment (u) v/u (000s) (%) Czech Republic June Slovak Republic June Poland May Hungary June Bulgaria March Romania June For comparison: UK June France June Germany (West) June ex-GDR May 85.0 13.2 31.5 25.0 12.0 3.3 128.0 59.0 355.5 30.2 142.0 282.3 2,228.0 546.7 453.0 675.0 2,678.0 2,753.0 1,716.0 1,149.1 59.4 4.7 1.4 4.6 2.6 0.5 4.8 2.1 20.7 IS Source: National labour ministries and statistical offices; OECD Main Economic Indicators, August 1992. I extent comparable with the West; and reported vacancies are only a JT3^9IL9l-3X^}l^lS. Job^ppenings. Is Eastern Europe in the throes of a recession (OECD, 1992) or a bout of severe structural change? Unemployment has structural aspects that cannot be overlooked. First, the recent sharp rise in unemployment I is concentrated in agriculture and. heavy industry. Light manufacturing, construction and services will eventually absorb these workers (OECD, 1992; CEC, 1992a). Second, trade in Eastern Europe must be reoriented ■aet a and Winters, 1992; Rodrik, 1992). Third, the practice of labour hoarding still survives from the shortage economy under central planning (Góra and Rutkowski, 1990; Góra, 1992). Except in Hungary and to some extent Poland, a new private sector mustbe builtjrom scratch. Siebert (1991) stresses this aspect of the traHsíořSSSSír f or" Eastern Germany, where the private sector, predominantly engaged in services and light manufacturing, is now growing rapidly (Johnson, 1992; CEC 1992a; OECD, 1992). Ladó etal. (1991) note that ä third of all employment in Hungary is the second (i.e. -entrepreneurial) economy. _ Eastern Europe's labour markets suffer hu^geographic rmsmatch (OECD 1992). For example, in the CSFRJnJune Í9?21rie registerecJ unemployment rate was 5.8%; yet in Prague it was only 0.4%. Assuming steady-state conditions, the average prospective duration of unemployment was 3.7 months. In Northern Moravia unemployment was 4.3%, the prospective duration of unemployment 6.6 months; in Eastern Slovakia, unemployment was 12.6%, and prospective duration 22.1 months. That's mismatch! Table 3 shows that mismatch also extends to the availability of jobs, and this is evident both where unemployment is lo"wlCSFR)'and high (Bulgaria). The within-country variance of the vacancy-unemployment ratio is as high as that between countries; it therefore seems unlikely that measurement error can account for this observed mismatch. It is often argued that the macroeconomic tools of aggregate demand and supply are sufficient for analysing Eastern Europe. Yet it is hard to imagine that price and wage rigidities, nominal or real, caused a slump in which high and accelerating inflation has been so prevalent. Even if unemployment is the product of a postponed oil shock and a demand contraaion (Jackman et aL, 1992), the concurrent 'cleansing effects' of recession (Davis and Haltiwanger, 1990) are significant: after four decades of socialism, a severe bout of spring cleaning was sorely needed! I thus focus on the longer-run determinants of average unemployment over the next five to 10 years. 2.2. The key role of the matching function n The dynamic view of labour markets stresses the role.ofune^^lcyment in job creation: with higher unemployment or more vacancies, more job "matches occur. Since most flows into unemployment result from _______________i 1 See Ball el aL (1988) for evidence that the sensitivity of output to nominal demand is negatively m IřSgSČ'- ms: .vui ►«.:(..> wie/ .; Opt Table 3. Unemployment and vacancies in Bulgarian oblasts and CSFR regions Vacancies Unemployment v/u (000s) (000s) (96) Bulgaria (March 1992) Sofia 3.6 Burgas region 1.1 Varna district 1.0 Lovetch 1.0 Michailovgrad 0.6 Plovdiv 1.2 Russe 0.5 Sofia district 1.6 Haškovo 1.4 CSFR (May 1992) Czech Republic Prague 17.8 Central Bohemia 8.8 West Bohemia 8.2 South Bohemia 4.9 North Bohemia 9.8 East Bohemia 9.7 North Moravia 10.4 South Moravia 8.6 Slovak Republic Bratislava 3.8 West Slovakia 2.3 Central Slovakia 2.8 East Slovakia 3.7 40.4 8.7 44.6 2.4 40.9 2.5 43.8 2.3 38.5 1.6 89.4 1.3 39.7 1.4 60.2 2.7 55.0 2.5 3.7 485.0 16.9 52.3 9.7 84.7 10.3 47.9 21.2 46.3 19.2 50.7 48.9 21.2 39.2 22.1 19.2 19.6 94.2 2.5 83.8 3.3 86.7 4.2 Source: Bulgarian Ministry of Labour, Czech and Slovak Federal Ministry of Labour. involuntary separation, Jbr new jobs to be created others must have been destroyed. Summarizing US evidence, Blanchard and Diamond (1989) write: 'both unemployment and vacancies matter in hiring. The ratejDfJ^nng appears to be determined by both sides of the labour Ü market, not only by the demand side, as is often assumed in macro-economic models.' (p. 29) The centrepiece of the analysis in the next section is the jnatching function, the process by whichvjobs and workers come into contact to form employment relationship^. The number of matches x depends both on the unemployment rate ju and the vacancy rate v. The matching functio^xfWjj;) captures sgatial aspects of the search process, imperfect information, and all types of ěccupational and industrial mismatch | between jobs and workers?1 It also subsumes potentially endogenous C 4 See Hal! (1979) or Pissarid"* (1979) for mmnl« of Ar search activities of workers and firms. In conventional analysis, xjv^ increasing in both u and v, with a diminishing marginal product for each, capturing the notiorTof congestion in labour markets. More controversially, it is often assumed to have constant returns to scale: doubling unemployment and vacancies doubles the flow of job finds. While available evidence supports constant returns in matching for industrial countries, the implications of increasing returns continue to fascinate (see Diamond, 1982, 1984; Howitt, 1985). Under constant returns to scale, the vacancy-unemployment ratio v/u is a sufficient statistic for the tightness of the labour market, which determines both the Hiring rate h — xjv and the job finding rate f=x/u.5 Job-to-job transitions are not modelled by the matching function (but see Pissarides, 1992; Blanchard and Diamond, 1992). However, job-to-job switching is not very relevant during the hard times of economic transformation and slack labour markets. 3. A simple model of unemployment and vacancies in equilibrium The matching function gives the per period flow of exits from unemployment to jobs, given a stock of unemployment and vacancies. Both unemployment and vacancy rates are in fact endogenous. The former depends on the evolution of employment over time, itself the cumulated reflection of past matching and other factors* The latter reflects the incentive to firms to create and offer jobs, which depends on the return from employment and the costs associated with posting vacancies. A number of tractable matching models of equilibrium unemployment exist (Pissarides, 1985, 1990; Blanchard ancFDiamond, 1989). More recently, these models have been applied to such diverse issues as economic growth and regional decline.6 The central model of my paper, similar to Pissarides (1985), is set out in the Appendix, but can be represented graphically in Figure 1, whose axes are the unemployment rate u and the vacancy rate v. The labour force is fixed, so u and v also represent the numbers of unemployed and vacancies. The model has two relationships. The first is the downward-sloping UV curve.7 Given a matching function and a separation rate 5 (the fraction of jobs terminating each period), the UV 5 The matching function is potentially subject to the Lucas critique: changes in the policy regime (unemployment benefits, income and indirect taxes, etc.) might affect its properties. 6 See Aghion and Howitt (1991) and McCormick and Sheppard (1992). 7 The 17V curve should not be confused with the 'Beveridge curve', the negative empirical relationship between vacancies and unemployment. The UV curve is only one blade of the rhT^r^rir"! *r?scrirs that crenpraff" f.he observed data. 'ť" Vacancy rate UV curve (unemployment constant) VS curve (supply of vacancies) Unemployment rate Figure 1. Flow equilibrium in the labour market curve describes the flow equilibriumbetween unemployment and vacancies. Unemployment is constant when the inflow into unemployment 5(1 —w) equals the outflow from unemployment x(u, v). With^high unemployment, employment is low and there are fewer workers to lose their job; thus the inflow to unemployment is low. For the outflow from unemployment (matHung)xt(o'^ě_'ei^iv^u«fťlý low, vacancies must be low. Hence jwhen unemployment is high vacancies are low: the UV curve slopes down. The second relationship is the supply of vacancies by firms. Suppose for example that each firm employs at most dně"wöTkefrWhen it does, output is y. When it does not, it posts a vacancy at cost k, the constant cost of creating and maintaining an unfilled job. Workers who are unemployed receive benefits b plus leisure with a value /. T is the severance payment to a worker, i7 the cost then incurred by the firm, which need not be equal. The supply of vacancies depends on how firms value the job matches that vacancies help generate. This value in turn depends partly on how production and sales revenue is divided between firms and workers. The appendix analyses a Nash bargain between a firm and its worker, and hence determines both the wage rate and the level of vacancies the firm wishes to supply. Figure 1 displays all this as the upward sloping VS line. Along VS, the expjected_ ^enefit to the firm of creating a vacancy is equal to jts (constant) costjf The higher the unemployment rate, the lower is the bargaining power of workers and the lower the wap;ě> the firm wishes i tel 1 Table 4. Comparative statics in the matching model Effect on equilibrium value of: Effect of an increase in: u V v/u Union power + - - Unemployment benefit (b) + — — Net severance burden (F— T) + — — Vacancy cost {k) + — — Interest rate (r) + — — Separation rate (5) — ? — Efficiency of match (x) — ? + Value of match (y) ~ + + to supply more vacancies. Vacancies create successful hires only through the matching function, in which they run into diminishing returns; so higher^unemployment induces only a finite increase in the vacancies it is profitable to supply. 3.1. Comparative statics and institutions Equilibrium unemployment and vacancy rates occur where UV and y^ intersect in Figure 1. Any exogenous change shifting either schedule affects both unemployment and vacancies. Many such changes are in^itutional. For example, an increase in bargaining strength of workers will rotate J^S tothe:. right: it is now less profitable for firms to supply vacancies. From Figure 1 vacancies fall but unemployment rises. Other interesting and relevant changes are those in government policy, including the level of unemploymentjbenefit b, mandated severance-benefits T, or firing costs F. Increases in b and F-T again tilt the wage bargain in favour of workers, reduce firms' incentives to create vacancies, rotate VS, and lead to higher equilibrium unemployment but lower equilibrium vacancies. One might interpret active labour maj^kej^poficjes as .reducing the cost of vacancies, rotating the VS curve to the left. Changes in the rate of separation s and the efficiency of the matching function affect both curves. I summarize the comparative static effects of these changes in Table 4. 3.2. The rise of unemployment in Eastern Europe This analysis illuminates labour market developments in Eastern Europe. Before the revolution, firing or quits into unemployment were rare. While there was severe geographic immobility for the unemployed, job-to»job switching was significant, with annualjurnover in the labour -tr~r.T"r;r~ ~^:~^~~~~~ -o^ r^£*-nr* -,n 1-,' rrl-. or T£ ö£ r, ,rppr Also f fa f1 ropt of it ágll*: :^© ::^3Í5 ■W ms Vacancies li Unemployment Figure 2. Labour market developments in Eastern Europe posting a vacancy was quite low relative to the shadow value of having a"^wFRH^rounB when scarce supplies arrived. Third, unions in the communist economies generally worked closely with the state to keep wages low rather than to maximize the workers' rents in wage bargains. Fourth,_uneinployment was actually illegal! ( All these changecl for the worse after the revolution. Workers dis-covere^TtEeir bargaining power vis-a-vis employers who turned out to be inadequate stewards of the state's productive capital. Unemployment benefits eased the pain of job loss. The cost k of searching for workers rose. The collapse in demand and change in the terms oflirade also reduced the~gam to firms from a match; the rise in the interest rate also reduced the value of an employed worker, as in Calvo and CÓrícelIi (1992). All these factors rptateJĚ^cjockwise. Involuntary separations rose dramatically as the private sector expanded, but also as state firms began to behave like capitalist enterprises, which also shifted the UV schedule outwards. Figure 2 shows the outcome: an unambiguous rise in unemployment and an ambiguous change in long-run vacancies (although a decline seems likely). If vacancies adjust immediately, they will decline at the outset, with dynamics for both u and v shown by the arrows in Figure 2. 4. Labour market institutions in Eastern Europe: an overview The preceding section established the theoretical relevance of institu-IH tionsand policies for labour market outcomes. Generous nriprnr»lovmp»nf benefits, bargaining strength of workers, higher severance costs for firms than severance benefits for workers, or fewer active labour policies all lead to higher u and lower v at any stage of the business cyclj*). Higher benefits may also shift the UV curve outwards if workers seek fewer contacts after becoming unemployed. I now examine the actual state of these labour market institutions, and attempt to rank the countries, countries, at least on a qualitative basis. 4.1 Unemployment benefits: how generous in Eastern Europe? The transforming economies quickly adopted systems to protect individuals from the risk of unemployment. All five countries have EC-style systems of unemployment benefits, funded by payroll taxation, based olTän in_sürähce principle but not experience-rated, and not means-tested (unlike follow-up social assistance, which is). The criterion of income replacement (as a % of previous net income) determines the amount of benefit, usually with minimal and maximal amounts. Initially, , these programmes were quite generous (OECD, 1992; CEC, f992a, b) especially in the potentiaLduration of benefit; for example, unemployment benefit in Poland was originally available wi^qutdme limit. More recently, eligibility, duration and income repjacement have_been tightened after^pressure frörrFnatiqna international organizations (The World Bank and IMF are now in the labour market policy business!). Details of systems as of July 1992 are provided in TableS.8 The value of unemployment benefits to eligible individuals is hard to assess: systems differ along many dimensions. The last column of Table 5 presents a crude measure developedin Burd.a (1988): it is the & present value of thé^ěnefTfs'package for a fully eligible claimant as a percentage of average^weekly earnings, which is then multiplied by the coverage ratio, the ratio of insured unemployment to total.9 The measure ignores taxes (in the Visegrad countries, benefit is taxed but at negligible rates); assumes all recipients of benefit earn the average wage; and assumes the coverage ratio is a good indicator of ex ante eligibility. Despite these problems, the results of Table 5 are revealing. As in Western Europe (Burda, 1988), the generosity of unemployment 8 Since the systems are changed frequently, the descriptions differ slightly from those in the OECD (1992) and CEC (1992a), written earlier in 1992. 9 For example, suppose the average worker in Transylvania eligible for benefit was entitled to draw 75% of salary for 26 weeks, and 50% of unemployed were eligible. Discounting at 20% r,PT. -,r-,r,„rr, m ar,ior>/ npr v.Ä„v> fhf indrv wohM he 930 (930% of weekly salary). ,uUu 6 of last 12 37% months, job losers only; special programme for school leavers CSFR Employed> 12 months 38% in last 3 years; quitters and new entrants ineligible Hungary All eligible according to 78% contribution (360 day minimum not strict); quitters and school leavers eligible after 90 days unemployment Poland Employed>180daysin 73% last_year, except school leavers, disabled, mass layoffs; moonlighting permitted in some cases Romania School leavers ineligible 64% until unemployed 5 months 60% of gross AW in last 671 6-12 months (6 months at 80% of MW for school leavers) 3 months at 60%, then 3 522 at 50% (can be topped up to means-tested 'social minimum') X Duration depends on 3,388 employment in last 4 years (max 18 months, no min); First 2/3 of duration: 70% of AW over last 4 years; then 50% 36% of last quarter's A W, 1,240 for up to 1 year (recently extended) Depends on service and 1,286 education: workers 60%, graduates 70%; for up to 270 days (recently extended) ^ Source: Interviews with labour ministry officials. :: Notes: MW = minimum wage, A W = average wage. ^?i?^^__ľa?Íe-§-.sign^cantly across countries, a variation caused more I by diffe^ejnc^e^Jri^jduj^tion and coverage than in the rate of income ^replacement. Hungary is by far the most generous, which explains the huge pressure on the government budget and recent criticism by the í ILO (1991). At the other end, the C§FR and Bulgaria are pretty frugal: ! the former by policy design^thé Tatter due to fiscal constraints. Poland [and Romania are in the middle of the pack (though both have recently ; adopted an emergency extension of unemployment benefit not included in the calculations behind Table 5). ||||r 4.2. Labour union strength and organization lifcThe trade union movement in post-communist Eastern Europe is i£g|f viewed with ambivalpnrf p^wurii- i^k^,,.. .,..,:.— extensions ot trie communist party; they can seem out of place in a transformation to capitalism. Yet these mammoth organizations managed considerable resources that in part have been retained (Freeman, 1992). Similarly, many union leaders have retained their positions by merit of their organizing skills at the rank-and-file level. Union organization rates and membership figures often mislead by including pensioners and inactive members; membership has dropped sharply in many countries since 1989 (Eberhardt and Heinen, 1992). In Poland two large confederations (OPZZ and Solidarity) dominate the union scene. Solidarity continues to struggle with its identity as a union and political movement; the ex-communist OPZZ häslrhäintäTned its membership at roughly twice that of Solidarity (Freeman, 1992). Membership of the two is about 7 mn., nearly 45% of totalemployment. Inter-union rivalry has enabled the government to get its way in tripartite disputes, although an explicit bargaining structure is lacking. The largest factor influencing bargaining power in Poland is the strength of the works councils, which since Í982_hay_e the power to hire and fire~management, making Polish firms in many respects like worker-managecTenterprises (Schaffer, 1991). Tri the CSFR unions lost credibility by standing by the communists to the bitter end. A state-dominated tripartite commission (seven members each from unions, management and the government) now sets lower bounds for nominal wage growth and decides on indexation of the minimum wage. By law, trade unions only have the right to be consulted on management decisions (c.f. morTpěTv^íverightš in other Eastern European countries). Membership is still high at 65-70% but growth of the private sector will erode this. As in Germany, annual branch bargaining sets a further floor on wages which are binding on firms in the same sector. Industrial relations in the CSFR are remarkably peaceful, due largely to limitations o nthe^ right to strike. In firms with over 200 employees, a system of codetermination existsby which workers can electlip to a third of the supervisory board. Significantly,jworks councils remain_absent in the CSFR. In Hungary, Romania and Bulgaria, collective bargaining occurs through multi-union competition for the position held by the dominant commuriisTörganizations. In Hungary seven ma jor confederations cut across industry and political lines, the largest (MSZOSZ: National' Confederation of Hungarian Trade Unions) having about 2 mn. mem-., bers. Total membership is nearly 60% of employment. Ineffective * employer associations have led to asymmetry in theemerging sectoral-lěvér wage bargaining. Provision for worker codetermination exists as in the CSFR. The burden of wage restraint devolves on the National Council for Reconciliation of Interests, a tripartite body which sets the ^Uit j. 15 |jt minimum wage (and was instrumental in settling the taxi-drivers' strike j§f.- in October 1990). Considerably weaker workers councils (with consulta- |gf:tive rights on dismissals and management of the social funds) have g|| replaced the enterprise councils that had powers similar to their Polish IE counterparts. Jjp. In Romania, labour relations remain antagonistic. There is no effec-IBi'tive tripartite relationship. There are 15 major trade union organiz-|§| äíiónšTséven major confederations in the CNSLR (the communists), SE three of which cut across industry lines, and eight independent break-jjH'away unions. Infighting is common, and in late 1991 the CNCS (National ji— Consultative Trade Union Council) was founded tojimprove cooper-sgyition and improve the Bargaining position of organized labour. Total j|j|membership is roughly 30-35%CUnions are sufficiently intrusive in the jjj|management of enterprises to make works councils unnecessary. jjjj" In Bulgaria the Nationwide Commission for Social Partnership, under lj|the Council of Ministers, coordinates the setting of the minimum wage fgjgänd the degree of indexation. The Confederation of Free Independent I Trade Unions is a legacy of the communist regime; Podkrepa was the first liberal, democratic alternative, and has been joined by three other independent unions. Unions represent 90% of the state industrial sector and about 45% of the whole economy, including agriculture. Bulgaria Has no system of works councils. P This section summarized factors that influence the strength of workers in bargaining: I judge it lowest in the CSFR, followed by Bulgaria ancTHungary, then Romania, then Poland. Section 5_will look at cor-Ijjbratism (the degree^ to which collective bargaining is managed by "groups and institutional structures) and the degree of centralization |the concentration of market power for both unions and employers). jpLS. Severance regulations and prior notice ÍĎiscussion in Western Europe has often focused on the employment leffects of job protection and severance pay provisions. Lazear (1990) Ends that tighter regulations and employment growth are negatively Brrelated in OECD countries. In the modei I develop in the Appendix, ghat matters for wage determination is F-T, the difference between pyerance costs faced by firm and the severance payments received by jfie worker: t^iejnexg^ejcij^^ se has no effect / pllg.g.^ (see also Burcta, 1992). jjFmore general models severance payments reduce both firing and plw hiring. Table 6 gives details oTthese reguläüon?Tneastern Europe: |Bey appear modest bv Western stn-nA-arA«: 116 Michael Burda Table G. Severance benefits and notification requirements Bulgaria CSFR Hungary Poland Romania Severance benefits 2_mp_nihs' gross pay; or may be increased through collective bargaining 2..m_o.Oth&Lgross pay; or may be bargained up toJ3jrionths (extra counts against claim on UB) Minimum of J.,^6 months, depending on service (<3 yrs: 1 month's pay; >_25jrs: 6 months' pay); extra counts against UB. claim <10,yrs: 1 month's pay; 10-20 yrs: 2 "]months; > 20 yrs: 3 months •' None Prior notice rules lmonth (for bankruptcies or mass layoffs) Pnorjiotice is means of paying sjyerance. benefit Obligatory consultation with elected works councils for mass layoffs Works councils consulted Not enforced must be Source: Interviews with labour ministry and research institute officials. / 4.4. Active labour market policies The experience of Western Europe, especially the high equilibrium unemployment in France, Italy and Spain, has not been lost on policymakers in Eastern Europe. On paper at least, they understand that deep recessions with poorly planned and passive systems of support for the unemployed can lead to long-term unemployment and politically irrevocable commitments to support-these Individuals. The buzzword is now 'active employment measures' (AEMs). fin theory AĚMá must reduce equilibrium unemployment. First, in Figure 1 they rotate the VS schedule leftwards. Publicworks, soft loans for job creation or self-employment, wage subsidies, tax incentiygsjor privat e firms^and unemployment benefit conditionalit^r^V^pjatorg^-Second, (ÄEMs^ enhance. Jnf;prmatipn exchan of unemployed workers' prospects!) while retraining programmes eiirrii-natejLhe mismatch between ya<^na«^ These yield 'ťFc^iicaljjĽogress' in the matching function, shijFtirjg_the_J7V curye inwards and the VŠ schedule leftwards, as the profitability of vacancies áfany given level of unemployment rate increases. Yet AEMs are now a response to a situation that had already deteriorated: high unemployment has overwhelmed the new labour ministries and employment offices of Eastern Europe. Often without centralized information processing, job matching j^^ reciuients often take moonlighting jobs (explicitly legal in Hungary and H juaoour martlets in Lastern Europe 117 H of beneJ^onreadiness to accept alternative job offers is rarely enforced fe except in the CSFR. The dramatic rise in obligations for passive it measures (i.e. benefits), inevitably crowds out funds available for active It measures. In Bulgaria, for example, ambitious programmes encompass-iLing 35% of the 1992 budget of the 'Fund for Unemployment and jf: Requalification' included abwäge subsidy along the lines of the successful §jT Czech and Slovak programing yet the rapid rise in unemployment in N recent months means that 80-90% of these funds will be paid out as El, unemployment benefits. 0, where V is the steady-state asset value of having an unfilled vacancy derived in the Appendix. Charlie Bean has suggested a 'large firm' model with costs of adjusting vacancies yielding multiple equilibria which are globally or saddle-stable. 1 In decentralized economies with search or matching, multiple equilibria are possible, and an economy may settle at an inferior equilibrium. This possibility is generally discounted since increasing returns in the matrbiniT fuTi'-t.v--./»*-~-*-:-----' (.Uta uaa 5. Empirical analysis and policy implications for Eastern Europe 5.1. Does the matching function work in the East? Extimates from the CSFR My analysis has assumed a stable matching function in Eastern Europe with the same properties as in the West. To test this assumption, I estimate matching functions using monthly registered unemployment and vacancy data from the CSFR, which include gross exits from unemployment to^employment during October 1990-May 1992 (with considerable gaps for Slovak data). The data come from 76 Czech and 38 Slovak employment offices. The remarkable quality oTthese data should not belie the usual limitations of registration data: omission both of discouraged workers and of informal employment of benefit recipients. I assume the matching function is log-linear: in logarithms, matches depend on a constant, on the number of unemployed and on the number of vacancies (and possibly also on some fixe^::ej^^diimmie.s, for example capturing changes over time). This function was estimated, in different months during 1991-92, on cross-section data from 76 employment districts including Prague, and in a sample of the same districts pooled over the entire period November 1990-May 1992, with and without fixed effects by region and time.13 The results are shown in the top part of Table 8. In the cross-sections, the hypothesis that the matching function has constant returns to scale in unemployment and vacancies is not rejected by the F-tests, despite the precision with which the coefficients are estimated. In contrast, in the pooled sample constant returns are rejected in favour of decreasing returns. The pattern of the estimated time dummies hints at technical progress (growing efficiency) in the Czech matching function (as well as some seasonality). To The extent that time dummies may be 'overcorrecting'. I therefore place more weight on the cross-section resulfs7 which tend to favour constant returns but also suggest that devdoping^labour market^takes time:14 the sum of the two coefficients is higher than at the outset of reform:. 13 These results resemble findings of Pissarides (1986) and Layard et al. (1991, ch. 5) for the UK. The Cobb-Douglas functional form seems acceptable for modelling the matching process; in no case did a CES approximation yield evidence against the Cobb-Douglas specification. 14 For example, time may be proxying both technical progress in the matching function and a downward treňčTTn reporting vacancies at employment offices.lTicancies are also measured with error, which also biases downwards the estimated coefficient on vacancies. Regional migration or commuting across districts may also help explain the results: exits can occur into other regions, so that high unemployment in one region may affect the matching process in adjacent regions. aboi ■rket East luro i21 [ Table 8. Estimates of the matching function, Czech and Slovak Republics, 1991-92 (Dependent variable: logarithm of exits from registered unemployment to jobs) taggaiĽ = Unemployment previous month, in logarithms Vacancies previous month, in logarithms Date coefficient /-stat. coefficient F statistic ■ significant i-stat. at 1% level Czech Republic: Cross-section, 76 districts, including Prague Same districts, pooled over 10/90-5/92, 18 monthly dummies "Slovak Republic: Cross-section, 38 districts Same districts, pooled over 12/90-9/91, 9 monthly dummies 1/91 7/91 1/92 5/91 9/91 0.42 (3.5) 0.60 (6.4) 0.75 (10.3) 0.44 (3.3) 0.27 (4.2) 0.24 (4.3) 0.65 0.61 0.62 0.53 (25.1) (3.0) (1.9) (6.2) 0.24 0.10 0.08 0.23 (12.9) (0.9) (0.5) (4.9) no no no yes no no yes gŕ Notes: All regressions have intercepts not shown above; /-statistics are heteroscedasticity-consistent; ||F statistic tests whether constant returns (two coefficients sum to 1) can be rejected; without Wpooling, failure to reject occurs at both 1% and 5% levels. jfe; The Czech lands enjoy one of the lowest unemployment rates in j| Eastern (and Western!) Europe. Do the results for the matching function (extend to high unemployment Slovakia? Slovakia has essentially the fesäme policies as in the Czech lands for unemployment benefits, several ance, trade unions and AEMs.15 Estimates of the Slovakian matching § function for two cross-sections and the pooled sample for the period i December 1990-September 1991 are shown at the bottom of Table 8. p. The Slovak results are remarkably similar to those for the Czech lands. Sg" 5.2. A normative model of closure of the public sector JU-The existence of a stable matching function canjprovidé guidance for H policy design in the transforming economies. Growth of the_p_riya_te i sector will require time, and will depend on the efficiency with which H workers and .firms can be brought. tpgethěř; on how quickly vacancies |; can be created; and, perhaps decisively, on the availability of labour Í resources (unemployed individuals). |i How rapidly should the state free up that labour? This is a question I of how cjuickly to privatize or close the state sector (still over 90% of T?;.rW;- ■^7/ >c f7'ic 15 According to anecdotal evidence, Slovakian administration of benefits and severance pay before ___ael___„a employment in Romania and Bulgaria, about 80% in Hungary and Czechoslovakia and 50% in Poland). For political reasons, wages in the state sector are well above the income of the unemployed, so workers are unlikely to quit.16 This is why unemployment is necessary for structural change. Yet too much unemployment carries its own social costs. Consider how tö~rnäxiffiize the present value of social output in an economy with a largestate sector with low (constant) labour productivity ys and a more productive but initially negligible private^ sector with ~highexj(c.onstant)^productivity yp. Gross hiring is assumed to be nonpositive in the state sector; vacancies there are zero. Public sector employment is a policy variable, private sector employment evolves according to the dynamics of unemployment, vacancies, matching and separations. Given the productivity advantage of the private sector, how rapidly should the state sector be shrunk and to what level? Provided a social cost is attached to unemployment, with increasing marginal cost, k 'cannot be optimal to have Jull andjmmediate closure, for then unemployment, the stepping stone to employment, would temporarily be enormous. In the Appendix I derive the best policy usingJo:pümal control theory.;; An extra worker employed in the private sector has a shadow value z, the appropriately discounted value of the differential productivity of the two sectors. If productivity levels in both sectors are exogenously constant, z must also be constant at its new long-run equilibrium value _ \tj ; (yp-ys)/{r + s), where r is the realjnterest rate and s the separation rate. The marginal benefkjj£ji^^ is thus (fz + l-kv) where / is the rate_of job finds (x]u),Jj.s the valujij[>ndsjui^ and kv the cosjt^f_posting vacancies. If (f>(u) is the spjriaJLcost of unemployment per se and '{u) the marginal cost, then social marginal cost of unemployment_isJ/(u) + y\ which also allows for the opportunity cost of not producing in the state sector. There is a unique (constant) rate of unemployment u that makes the ^ marginal benefit (fz + l + kv) equal to the marginal cost (f?'(u) + ys. Hence the optimal policy begins with an instant shakeout of state employees up to this critical level of unemployment.17 Thereafter, state employment is wound down just as quickly as the labour market can absorb the unemployed into the new, growing private sector, leaving unemployment constant thereafter. Optimal policy is neither 'big bang' nor 16 Blanchard (1991) explores the possibilities of bleeding the state sector by freezing public sector wages. The experience in Poland suggests this may not be politically feasible. 17 The optimality of a constant profile for unemployment depends of course on the simple linearities of the model: in a more plausible model this would no longer be so; however, the 123 Private employment e° Unemployment rate u State employment 0s jj| Figure 4. Optima! paths of private and public sector employment 'go slow', but a 'mixed bang' large enough to get to long-run unemployment immediately, but proceeding thereafter only at the rate at which the labour market can match or digest further layoffs. Figure 4 summarizes these conclusions. What speeds of private sector growth are implied by the matching function? Imposing constant returns in matching, Table 8 implies a matching function like x = Auav1~a. Based on the Czech estimates .in JableQ8 (including unreported intercepts), I assume a = 0.74 and A-e ■ .1 also assume unemployment levels and v/u ratios remain constant at the levels shown in Table 2 for each country. I can then compute the implied rate of monthly layoffs from the state sector to keep unemployment constant along the optimal path suggested above; and the corresponding matching and new job creation in the private sector. Results are shown in Table 9. They are comparable to current rates of inflow into unemployment in Eastern Europe, which have already reached Western levels: in the Czech Republic the inflow rate as^a fraction of employment was 0.5%; in Slovakia, 1.4% (both June 1992); in Bulgaria 0.9%, and in the ex-GDR, 1.0% (both March 1992) Corresponding monthly ratesin the JJK, France and Germany in recent years were about LJ.5 %. This analysis suggests that a 'mixed-bang' approach to the state sector ^feasible, but the simplicity of theln^del igríor^iveral issues beyond the scope of this paper. For example, feedback from lavoffs to aggregate m^ ,j__ .o u tut raa Table IL Unemployment forecasts for Eastern Europe Assumed values, by country Forecasts Unemploy- Long-term ment unemployed as % of as % of all ÜBEN CORP CENT AEM labour force unemployed Bulgaria 0.671 Belgium (7) Belgium (8) 1.13 7.2 28.3 CSFR 0.522 Sweden (3) Germany (6) 3.03 3.0 C26\5 Hungary 3.388 Holland (6) France (11) 3.42 8.2 42.5 Poland 1.240 Spain (12) France (11) 1.70 12.2 Ho Romania 1.286 UK (11) Canada (17) 0.23 11.6 15.8 Memo: OECD average 3.476 (6.6) (9.5) 6.5 r 8.4 35.6 -:' Notes: Penultimate column computed using estimated coefficients (and unreported constants) from first regression in Table 10; final column uses final regression of Table 10. centralization and its square, severance benefits, and prior notice were insignificant in the presence of these variables. Note that expenditure on active employment programmes is robustly estimated to yield a 0.2% reduction in unemployment for each additional 1 % of per employed worker output spent. This coefficient is insignificant when spending is measured as a fraction of GDP, confirming the relevance of expenditures per person. I examine long-term unemployment, a chronic..problem in OECD countries yet with considerable cross-country variation. The results are again shown in Table 10. The Calmfors-Driŕnll centralization measure is now significant, whereas the AEM variable is now insignificant. One interpretation is that long-term unemployment consists of 'outsiders' alienated from the labour market. Highly centraližedand^eceritřalížed labour markets may^ allow_swifter_..reintegration of outsiders in the wage-setting r^rjocess. "~"~ """"""" \j Frömlnese OECD estimates and attributed values for the Visegrad countries one c^ifor^castayerage standardized unemployment rates and the percentage öTühempIÖyment in long-term unemployment. To do so, I also have to assign comparator Western countries whose indices of corporatism and centralization can be assumed by Eastern counterparts; necessarily, this~matching is subjective. Moreover, exogenous variables we observe today can change, sometimes rapidly and radically (e.g. unemployment benefits in Poland and the CSFR, or the development of collective bargaining in Hungary and Bulgaria). My forecasts must be taken with a grain of salt. Even so, Table 11 shows my 'best guesses' for the exogenous variables anH t h p associatptl forecasts of the average unemployment rate and the wua in.tu sunt uruj 2/ long-term unemployment rate for the five countries. In general the predictions are not terribly encouraging. The CSFR apart, Eastern Europe has a long period of high unemployment ahead. However, the predictions are often several percentage points below current observed unemployment rates; this deviation can be interpreted as Keynesian unemployment. High unemployment is usually associated with high rates of long-term unemployment: for both, benefits and corporatism are major determinants. The forecast of low long-term unemployment in Romania is due to a combination of decentralized collective bargaining and low benefits (as in Japan or the US). The simultaneously high unemployment forecast reflects low spending on AEMs and noncorporatist union-management relations. Ig: 6o Conclusions What have we learned from this study of the transforming Central and Eastern European economies? In its labour market institutions, Eastern Europe is moving rajjidlyjowarcls West European norms, although cross-country differences in unemployment benefits systems, collective bargaining and active employment policies are significant. Labour market institutions influence equilibrium unemployment in theory, and appear to explain much of the cross-country differences within the OECD in practice. The flow or matching approach to labour markets has concrete lessons for managing the transformation process. OjJtimal control suggests a I mix of shock.^he^apj^and^go-slow that reflects-the^ tirn^elementjpf the matching function. Estimates or the matching function support the hypothesis that Eastern European economies behave like those in the West and that unemployment will be a necessary ingredient for the transformation. At the same time, conclusions about the optimal rate of privatization or closure of enterprises must be drawn with caution. Portes (1992) has arguecTtnat many of the firms in sectors designated as value-added subtractors by Hughes and Hare (1991, 1992) could have proved viable, reorganized by new management(JReestablishing industries after closure may be associated with significant fixed cbsti. The option value of waiting (Dixit, 1992) may be high, especially given the big variation in enterprise profitability, both at firm and aggregate levels (see Hughes and Hare, 1992; Kolanda and Kubista, 1990). Yet the calibration exercises using estimated parameters for the match-ing function imply that selectixe_s!a;yjng_of the most_inefficient dinosaur enterprises (Siebert, | 1991) is feasible and perhaps the optimal poliev. 24 lc^c urát Table 9. Implied rate of layoff for the public sector Implied severance rate per month: Implied job Absolute % of employment finding rate Bulgaria 47,900 1.38 0.106 Czech Republic 33,700 0.90 0.238 Hungary 45,200 1.04 0.102 Poland 200,200 1.24 0.090 Romania 46,200 0.42 0.068 Slovakia 34,600 0.90 0.123 Notes: Assuming constant v/u ratio in Table 2; no initial private sector employment; and constant returns matching function estimates. Middle column data for Czech and Slovak Republics as % of combined CSFR employment. in both sectors, magnified by the multiplier effect on demand. A further problem is related to the multiple equilibria of Section 4.5 which emphasized fiscal effects on the incentive to create vacancies. That discussion should now'Se generalized: the cost of privatization includes a loss of revenues to the government that would have helped finance unemployment benefits (this effect is stressed in Bolton and Roland, 1992). Third, increases in unemployment may destroy human capital or lead to wage settirlgThät permanently increases the naturalIrate ]^ le^dmg^to^path-derjendence. Even without formal analysis, it seems likely that the existence of multiple equilibria will put more weight on caution and 'go slow' to avoid moving to a high unemployment trap,. 5.3. The equilibrium rate of unemployment: lessons from the West The previous section explored the p_rofile. of layoffs in transforming economies, taking the vacancy-unemployment ratio (and hence /, the rate of job finds) as given. A natural extension of the analysis is to address the determinants of v/u within the model itself, and thereby the equilibrium or natural rate of unemployment. The discussion of Sections 3 and 4 emphasized the role of labour market institutions and policies in determining the equilibrium rate of unemployment. To learn about their relative importance, I estimated a parsimonious model for a cross-section of 18 OECD countries.18 I regress either the average standardized unemployment rate during jlc ma in l .-.n E—cz Table 10. Unemployment rates in OECD countries 125 /'-CP('wSffficienton: ÜBEN CORP CENT CENT5 AEM R* Dependent variable: Average standardized (1) unemployment rate (%) 1986-90 (2) (3). (4) (5) (6) (7) 0.85 (5.0) 0.95 (1.9) 0.96 0.93 (3.8) 1.06 1.03 -0.047 (1-3) (-1.0) -0.57 0.027 -0.18 (-2.8) -0.24 (-1.9) -0.20 (3.6) (3.6) (-1.4) (1.4) (-2.6) 1.13 1.10 -0.66 0.039 (3-5) (3.7) (-1.3) (1.4) Dependent variable: % of total 1988 unemployed in a current spell of over 12 months unemployed 8.61' (4.8) 5.96 (2.8) 6.15 (3.0) 5.93 (3.2) 3.50 (2.9) 2.31 (1.9) 2.20 (2.0) 12.50 (3.7) 9.49 (2.6) 9.22 (2.6) -0.62 (-3.4) -0.483 (-2.6) -0.479 (-2.6) 0.04 (0.1) 0.09 (0.2) 0.29 (0.6) 0.79 0.49 0.81 0.75 0.79 0.83 0.89 0.88 Sources: Dependent variable data from Bean (1992); ÜBEN = Burda's (1988) measure of unemployment benefit generosity (divided by 1,000); CORP = Tarantelli's (1986) corporatism ranking (1 =most corporatist); CENT = Calmfors and Driffilľs (1988) measure of centralization in wage setting (1 = most centralized); AEM = spending per unemployed as fraction of output per employee (Jackman et al, 1990). Notes: Sample: Belgium, Denmark, France, West Germany, Ireland, Netherlands, Spain, UK, Australia, New Zealand, Canada, US, Japan, Austria, Finland, Norway, Sweden and Switzerland. Sample sizes of 18 (regressions 1-4) and 14 (regressions 5-8). Heteroscedastic-consistent ť-statistics in parentheses). Estimated intercepts not shown in table. || 1986-90 or the long-term unemployment rate in 1988 on unemploy-| ment benefit generosity in 1985 (Burda, 1988), Tarantelli's (1986) index ^ of corporatism, Calmfors and Driŕfilľs (1988) measure of wage centralization and Jackman et al.'s (1990) measures of expenditure on AEMs. IFor a somewhat smaller sample, I also included as repressors unioniz-fation rates (OECD, 1991) and severance benefit and notice rules (Lazear, |1990). „ L j£" The results are shown in Table 10. Benefits, corporatism and spend-ling on,active emplgxrnejbt-measur-e^XQbustly explain .75-80% of cross-|sěttional_variance of average OECD standardized unemployment in f hf> 1 9 roimfrtmc *=>-<.••->■> 28 icha irda I have made the case that conventional demand and supply factors, despite their importance, are only part of the transformation story. New policies and institutions will be necessary to mediate and accommodate the special aspects of labour markets: search by employers and employees, the special role of information, the highly specialized nature of the 'commodity' that is traded. Some of these institutions which are widespread in advanced Western economies - collective bargaining, unemploymentj.nsurance and the social net, re^gulajdojis^^cr^dtive IaBô^r^äŕTcei^oličy"-will have profound effects on the ultimate outcome of the transformation process. Simple regression results from OECD countries suggest that with the exception of the CSFR, high unemployment will be a feature of Eastern Europe well into the foreseeable future, and that long-term unemployment will_be an important and durable component. i. t^^v^,! :"=v-. >■-,-•' To counteract this tendency, the new market economies must quickly implement corjDorai^ and tighter admini- stration of unemployment benefits, perhaps combined with a concerted e^prtJ:oja^3^^ described in Section 4. While the evidence shows that the benefits from active labour markets are fairly high, so are the costs. In this context foreign assistance, at least at the technical level, may have high social returns. Early successes in the CSFR suggest that emphasis on j^_jmtching and information exchange, active promotion of entrejgnsneurm and subsidiz- ation of job creation in both public and private spheres may be the right approach. Discussion Charles Bean London School of Economics This is a very ambitious paper, which as well as containing a vast amount of information about the emerging labour market institutions of Eastern Europe, also seeks to quantify the likely impact of these new institutions on the equilibrium level of unemployment in the medium term. Most, economists are reluctant to pin themselves to tight predictions under any conditions, so Burda's willingness to do so in the context of major structural change is bravery (or foolhardiness?) of the highest order. My critical comments below should therefore not detract from what is a very useful paper. My first worry is the application of the standard matchingapproach to labour markets, already applied fruitfully to Western economies by TV-----.-.' 1.-.- T» 1-----. ....1. .. ... ,..1 „, ,,^ J T~i ľ „ ,.^-j ^.-, ._1 f.r, f-T-ii> r>T-n ovrriri >~r 1-il^rMir* rmrrptC §111 ._ bour ... kets___astei,* j^dropt, 129 jljjt- of Eastern Europe. While this is likely to constitute a perfectly good im~- paradigm for the post-reform economy, it seems less plausible as a story of how labour markets operated in the ancien regime. The way in which labour was allocated under communism was fundamentally different, and it is not clear that one can really describe the status quo ante as being given by the intersection of the lower UV curve and the higher VS curve in Figure 2, and reform in terms of a few parameter shifts. Surely the labour market did not really exist in any meaningful sense pre-reform, but rather has slowly emerged since the. end pf communism? This in turn~wííľ have implications for the pooled time-series cross-section estimates of the matching function, although the presence Jjl of a few time dummies should help to control for it. |gi|;--"" My second observation relates to the possibility of multiple equilibria, lip which here relies on 'fiscal increasing returns' rather than 'thick market *fe externalities', which in any~case seem to be empirically unimportant. However, Burda does not give us any idea whether multiple equilibria due to fiscal increasing returns are likely to be any more relevant empirically. It would seem natural, therefore, to use the estimated matching function, together with some appropriate characterization of lj the supply of vacancies and the tax-benefit system, to identify the || likelihood of such high unemployment equilibria emerging. ^.:--r*TEirHTÍét me take issue with the analysis of the optimal speed of I restructuring. This makes matching functions the only obstacle to struc-| tural change, but surely such frictions are a secondary (or even tertiary) St issue. The model treats the decision to open a firm and to open a ff vacancy as synonyms, and assumes that there are no impediments or p| frictions involved in doing this. In^reality, surely, the limited supply of % entrepreneurial j^lis and questions of how the requislt^^Bysical^and 1 financial capital are obtained, are of much greater significance. I do !' gnot see howöne can seriously evaluate the optimal speed of restructuring Iwithout taking account of these issues. ä Finally, let me note that Burda's diagnosis of the structural factors lying behind high unemployment in the OECD is entirely consistent with the work that I and others have carried out. This indeed points to the important role played by the duration of unemployment benefits, the degree of coordination in wage bargaining, and the benign influence ^of actiye job market programmes. I am struck by the diversity of labour "market institutions in Eastern Europe that is nevertheless emerging, which suggests that perhaps Eastern Europeans have not learned as Jmuch from Western experience as they might. It is to be hoped that I the transition process, which will be uncomfortable enough, is not '-exacerbated further by the application of misguided labour market uiui Ida Jan Svejnar University of Pittsburgh and CERGE, Prague This is a timely and competent paper, which consists of three parts: (a) an overview of the main labour market institutions that have emerged in CEE during the transition; (b) a theoretical model of equilibrium with unemployment, vacancies and a matching function between the unemployed and jobs; and (c) an empirical analysis which uses monthly district-level data from the Czech and Slovak republics to estimate loglinear (Cobb-Douglas and Kmenta CES) matching functions. The main contribution of the paper lies not in its description, which is reasonably familiar, but in its analysis. The author reviews and further develops the macroeconomic literature on unemployment, vacancies and matching in the labour market and he stresses its relevance for the transforming socialist economies. He brings up the important and frequently neglected fact that the transition has brought about significant job creation and not just job destruction. He points to significant mismatch between the geographic location of workers and jobs and he arg;ues that unemployment is a necessary part of the job creation process. The simple theoretical model presented in Section 3 shows how equilibrium unemployment and vacancies are related to institutional variables reflecting workers' bargaining power (e.g. corporatism, laws on trade unions and the presence of workers' councils) as well as government policies (e.g. unemployment benefits, severance pay and firing costs). The empirical analysis focuses on estimating one component of the theoretical model, namely the matchingjunction that relates the number of job matches created to unemployment and vacancies. As Michael Burda points out, the matching function, like the production function, is3^Säck.t>ox'./It reflects the spatial and informational aspects of the 1 matching process äTwell as the institutional features of the labours market and the search activities of workers and jfirms. 6| The author concludes that the empirical results point to the existence of a stable matching function and the importance of both vacancies and unemployment in job creation. This leads him to argue that the growth of the private sector will require time and will depend on the speed with which vacancies can he created and on the availability of unemployed individuals. Overall, this is a fine paper with important policy implications. My comments below reflect a concern that future research should attempt to broaden both the theoretical model and the empirical analysis. A particularly troublesome feature of Burda's approach is that it focuses exclusivelv on transitions from unemployment into employment ^. )uur lets istei :ropi lol ME. ■•• jJU and ignores job-to-job switching. While it is true that job-to-job switching ljjb was mučHjmôFejímpořfařiT Tinder the full employment communist HI system that now, it st^rejnams an extrem^ jjl TáTte~theexample of the districts comprising Prague. The growth of 'Ü* private sector employment has been phenomenal there and yet the I1ÉJ unemployment rate has been under 0.5%. T^ew"matches have been Ä massive and mostly of a job-to-job nature. From a policy standpoint, HI? this type of labour market transition has great appeal and ought to be Hfe fadlitatejd by government policies. There is the additional consideration SÍL tHätTwith one-half oHarge Czechoslovak enterprises having been swiftly |||T privatized in 1992fthe most rapid private job creation has involved no Wm job switching or matching.^ :' iljg^ Analytically, one would also like to see the empirical analysis linked HC to the theoretical model beyond the simple estimation of the matching 9p function. The matching function is an important building block of the (Wkoverall model, but it does not capture many of the aforementioned IB behavioural and institutional features whose empirical estimates would j|§ be useful for public policy. |§r With respect to the econometric work itself, it is worth pointing out f. that the Slovak cross-sectional regressions have a poor fit and that the f estimated coefficients on vacancies are both small and statistically insig-|nificant. In Slovakia, the support for the matching model is weak and I stems primarily from the pooled regression. Both here and in the rest lpi Eastern Europe, the question of how important are issues of matching [ | relative to those of conventional demand and supply factors remains a ; | very open one. [General discussion I Discussion in the Panel focused both on theoretical and empirical aspects lof the paper. Olivier Blanchard said that, while sympathetic to the I general approach of the paper, he had some concerns about the realism |of some of the theoretical assumptions. First, much hiring was done fdireclly_from employment to employment without passing through ^unemployment in between-a criticism echoed by Willem Buiter, who f pointed out that on-the-job search would make decreasing returns quite ^plausible in the matching function. Second, the reductions in employ-Iment were often achieved by a hiring freeze, which meant that first-time |workers_were dispr-oportionately.. represented in the pool, of unemployed. Third, the real bottle-neck in job creation was notjhe cost of a vacancy, but <^stacles^to_^Ci:reation of new firms. And fourth he .^wasconcerned that wagesTmöüTcTnöt be ccoicělved as being determined 0��331864542633594 1 Mi ßu given that much of the rent in the state sector would be appropriated by workers. (This point was disputed by the author, who said that rent capture in the state sector did not affect the nature of matching, which was primarily a private sector phenomenon.) Blanchard also wondered ? whether the quality of the data on vacancies in Eastern Europe was í good enough for the sophistication of the exercise in hand. Willem Buiter added to these observations a concern that a model formulated in levels would not capture non-stationarities that could be important. He was also worried that a partial equilibrium model might be inadequate, given the endogeneity of other variables such as the * interest rate and the productivity of a successful job match. John Flemming was worried by the assumption of constant and uniform productivity of state employees. It was an essential feature of economies in transition that productivity was not uniform, because of changes in relative prices. Then the immediate elimination of value-subtracting sectors, followed by a more gradual convergence to the long-run level, would imply a different, and lower, set of transition costs than that described by the model of the paper. Empirically, he was also concerned that lags in eligibility meant that apparently generous benefits would in fact have been eroded by inflation_by the time they ! were received; this could significantly affect some of the results reported in the paper. Various suggestions were made for improving the precision of the empirical analysis. Paul Seabright suggested using data on new firm registrations to test whether there appeared to be a new firm bottle-neck effec^ilclitional to the cost of creating a vacancy. Georges de Menil said there were a number of other ways of measuring structural change that could be employed to test more directly the competing hypotheses of structural shock and demand-induced recession. Alan Manning was less convinced than some of the Panel that the effect of trade unions and of benefits had convincingly explained the rise in unemployment in recent years in Western Europe; he wondered why they should be any more significant in the East. Appendix: Equilibrium unemployment and vacancies, and the optimal closure of the state sector A.l. Equilibrium unemployment and vacancies The UV curve in Figure 1 solves x(u, v) = s(l - it). Total differentiation rr;rior v 4.- /V7-,M,,\ — _c i,.rhe>r>rf* JvMit <0 f.nnvpvitv of "-^p T7V CÜTVC Lal^uu, ,narkci* %a Eastern Europe 133 to the origin follows from the diminishing marginal rate of substitution between u and v in the matching function. Off the UV curve, unemployment obeys it = s{\ — u) — x(u, v). Now for the supply of vacancies. Let / and V denote respectively the values to the firm of employing a worker and posting a vacancy (being without a worker). Assuming risk-neutrality, and letting 6 denote v/u, the return in each of these states equals the interest rate r times the value of the state: ■V = -k + h(d)(J-V) and rj = y w HJ-V + F) (1) t where w is the wage, F a once-off severance cost in the event of a separation which occurs with probability s each period, k the cost of a gjggr vacancy and h the hiring rate x/v which itself depends on 0. Thus S J— V = (y — w — sF-hk)/(r + s + h). A similar calculation can be made for workers. If unemployed they receive benefits b and leisure valued at I. Ä The probability of a job find f(6) = x/u — h8, where f'(6)> 0. Placing a Sworker value E or U on states of employment or unemployment, and S letting T be the once-off severance benefit for a worker: %rU = b + l+f{E-U) and rE = w + sT-s(E- U) "Hence E-U = (w + sT-b-/)/(?- + s +/). ~l Assume a Nash bargain over wages to maximize k = (E-Uf(J-V) 1-/3 (2) (3) Jsubject to the conditions that E—XJ and/— V are both positive. Clearly Wy — sF + k> w> (b + l — sT) if a match is to be attractive to both parties, i/j measures the relative bargaining power of workers. Logarithmic fdifferentiation of (3) with respect to w shows that for Biff/dw = 0, m==(l-ß)(b + l-sT) + ß(y-sF + k) (4) lÜEntry occurs until the value of a vacancy V = 0. (1) then implies f£=k/h = (y-w-sF)/(r + s) (5) |Frorn (5), k(r + s) iW= y-w-sF (6) ys.a ray through the origin in v-v *™r<* Tí^ ™ • i 34 Michael Burda A.2» Optimal closure of the state sector Productivity in the state and private sectors is ys and yp, {yp>ys)\ employment is es and ep. The value of leisure of the unemployed is ul, the cost of posting vacancies is kv or ku6. The social cost of unemployment is (f)(u) with (u)} dt (7) subject to three constraints. The first is (6), the private incentive to supply vacancies. Note that (6) implies 6 and hence f(6) and h{6) are all constant. The second is u = 1 — es — ep, and the third is ep = u6h(6)-sep = (l-es-ep)f-sep (8) The Hamiltonian is H = e~ri{yses + ypep + (l-es-ep)(l-kd)-(})(l-es-ep) + z[(l-es-ep)f-sep]} where the multiplier z is the shadow value of private sector employment. Necessary conditions for a maximum are: ys-l + ek + '-zf=0 (9) yp-l + 6k + '-z(f+s)=rz-i (10) whence z = (r + s)z-(yp-ys) (11) (9) and (10) also imply both the optimal level of unemployment u = '-\zf-ys-k0 + l) (12) which on substitution into (8) yields ép=fu-sep=f(f>'-1(zf-ys-kd + l)-sep (13) Figure A.l draws the phase diagram for (11) and (13) in (z, ep) space. The saddlepath is the locus along which z is constant at its steady-state value. The jump variable z immediately assumes this value; the state variable ep is predetermined and converges to its steady-state value according to (13). From (12) it is evident that u jumps immediately to its steady-state value and remains at that level from then on. Finally, from all this we infer the optimal path for the control variable es, which shrinks as ep grows, such that the identity 1 — es + ep + u continues to Labour markets in Eastern Europe 135 ep = 0 -< «S£---------«^------------ Z= 0 Figure A.l. Phase diagram for z and ep values of u and ep sum to less than unity and the state sector does not vanish even in the long run. Although the private sector is more productive, I have assumed that there are no quits in the state sector: the only separations are optimally chosen layoffs. In contrast, private employers must take into account the prospect of future separations. It is this that allows the possibility of permanent survival of some state sector production. p References ■^5.- I Aghion, P. and P. Howitt (1991). 'Growth and Unemployment', CEPR Discussion Paper No. 577. Ball, L., G. Mankiw and D. Romer (1988). 'The New Keynesian Economics and the Output-Inflation Trade-ofF, Brookings Papers on Economic Activity. ' Bean, C. (1992). 'European Unemployment: A Survey', Centre for Economic Performance Work-\ ing Paper No. 71, LSE. Blanchard, O. (1991). 'Notes on the Speed of Transition, Unemployment and Growth in Poland', mimeo. Blanchard, O. and P. Diamond (1989). 'The Beveridge Curve', Brookings Papers on Economic Activity. ------(1990). 'The Cyclical Behavior of the Gross Flows of US Workers', Brookings Papers on Economic Activity. .------(1992). 'The Flow Approach to Labor Markets', American Economic Review, Papers and Proceedings. Blanchard, O. and L. Summers (1987). 'Fiscal Increasing Returns, Hysteresis, Real Wages and Unemployment', European Economic Review. Bolton, P. and G. Roland (1992). 'Privatization Policies in Central and F.asrpm Fnmn»1 f-„-,---•- ��51//+� 54 The Social Challenge of Job Creation 25 /Vn innovative suggestion has been voiced recently by Belgium's Finance Minister Maystadt. The goal is to transfer into jobs some of the gains accruing to prosperous firms thanks to the 1995-96 wage freeze. Exceptions to the freeze would be authorised if they took the form of "service-vouchers", usable by households to purchase proximity services (see footnote 18 above). The face value of the vouchers would be treated on par with wages for assessment of labour taxes and income taxes. The vouchers would be tradable. The idea is to generate a demand for proximity services high enough to induce a corresponding supply by non-profit organisations. 26 My views on the fiscal guidelines are given in "1 Market + 1 (tight) Money = 2 Rules of Fiscal Discipline: Europe's Fiscal StanclTDeserves Another Look", in European Economic integration: A Challenge in a Changing World, M. Dewatripont and V. Ginsburgh, (eds), North-Holland, Amsterdam, 1994. I touch on some aspects of monetary policy in Money and Uncertainty: Inflation, Interest, Indexation, Banca d'ltalia, Lezioni Paolo Baffi di Moneta E Finanza, Edizioni Dell' Elefante, Roma. 27 Perhaps against the advice of an impressive group of MIT economists? 28 Under that regime, domestic money creation is subject to 100% reserves in the reference currency. 29 Cf. "Priorite ä l'EmpIoi", a manifesto by 72 French-speaking Belgian economists in January 1987. 3 . Preventing Long-Term Unenployment: An Economic Analysis Richard Layard* 3.1. INTRODUCTION AND SUMMARY The European Union has set the target of halving unemployment by the year 2,000 (EU, 1994). How can it be done without increasing inflation? The strategy must be to reduce those kinds of unemployment which do little to restrain inflation. The most obvious such category is long-term unemployment. 3.1.1 Effects of long-term unemployment Let us examine the evidence. In wage equations long-term unemployment is usually found to have a very small (or zero) effect in reducing wage pressure.1 The reasons for this are obvious: long-term unemployed people are not good fillers of vacancies. This can be seen from data on exit rates from unemployment: exit rates decline sharply as duration increases. Equally, aggregate time-series show that, for a given level of unemployment, vacancies increase the higher the proportion of unemployed who are long-term unemployed. If long-term unemployment is an optional extra, depending on social institutions, it is not surprising that there are striking differences in its prevalence across countries. As Table 3.1 shows, in the 1980s the majority of countries had between 3 and 6 per cent of the labour force in short-term unemployment (of under a year). But there were huge differences in long-term unemployment. It was under 1 per cent in the US, Japan, Canada and Sweden and over 8 per cent in Spain, Belgium and Ireland. Clearly some short-term unemployment is necessary in any economy, to avoid the inflationary pressure which would develop in an over-tight labour market. But long-term unemployment is not needed for this purpose. s^ 56 The Social Challenge of Job Creation 3.1.2 Causes of long-term unemployment So how can it be prevented? To consider this we need to know under what conditions it occurs. Figure 3.1 provides a striking clue. It shows on the vertical axis the maximum duration of benefit in each country and on the horizontal axis the percentage of unemployed people in long-term unemployment (over a year). In countries like the US, Japan, Canada and Sweden benefits run out within a year and so unemployment lasting more than a year is rare. By contrast in the main EU countries benefits have typically been available indefinitely or for a long period, and long-term unemployment is high. The relationship shown in Figure 3.1 is of course a partial correlation. But if one allows for multiple causation, the effect of benefit duration upon the aggregate unemployment rate remains strong and clear.2 The effect of unemployment benefit availability upon unemployment is not surprising. Unemployment benefits are a subsidy to idleness, and it should not be surprising if they lead to an increase in idleness. In principle of course the benefits are meant to protect individuals against an exogenous misfortune and there is meant to be a test of willingness to work. But in practice it is impossible to operate a "work test" without offering actual work. So after a period of disheartening job search, unemployed individuals often adjust to unemployment as a different life-style. 3.1.3 Preventing long-term unemployment What should we do about the situation? One possibility would be to reduce the duration of benefits to say one year and put nothing else in its place. This would be the American-style solution. But we know this only works because people thrown onto the labour market accept an ever-widening inequality of wages. A much better approach would be to help people to become more employable so that they would justify a better wage. This leads to our central proposal. After 12 months the state should stop paying people for doing nothing. But at the same time it should accept a responsibility to find them temporary work for at least 6 months.3 In return the individual would recognise that if he wishes to receive income, he must accept one of a few reasonable offers. These offers would be guaranteed through the state paying to any employer for 6 months the benefit to which the unemployed individual would otherwise have been entitled. This would have huge advantages: (i) After the 12th month, it would relieve the public finances of any responsibility for people who are already in work.4 It is very Preventing Long-Term Unemployment: An Economic Analysis 57 difficult to prevent fraud without being able to offer full-time work. (ii) Between months 12 and 18, people would be producing something rather than nothing. (iii) But the biggest effect would come after the 18th month. Provided the temporary work had been real work with regular employers, unemployed people would have re-acquired work habits plus the ability to prove their working capacity. They would have a regular employer who could provide a reference - or (even better) retain the individual on a permanent basis. The main justification for the proposal is not that it employs people on a subsidised basis but that, by doing so, it restores them to the universe of employable people. This is an investment in human capital. That is the central objective of the exercise. Job creation schemes in the past have often failed because the jobs have been marginal and have failed to make the individual more employable thereafter. The job subsidy should therefore be available to any employer (private or public). There should also be the least possible restrictions on the kind of work that could be done. Clearly no employer should be allowed to employ subsidised workers if he was at the same time dismissing regular workers. But there should be no condition (as there was in the UK's former Community Programme) that the work done should be work that would not otherwise be done for the next two years. Such a requirement is a formula for ineffectiveness. The reason why job creation schemes have so often had these disastrous limiting conditions is the fear of substitution and displacement. This fear is understandable but misplaced. 3.1.4 Substitution and displacement Most opposition to active labour market measures is based on fears of displacement and substitution. In their extreme form these derive from the "lúmp-of-läbour fallacy": there are only so many jobs, so, if we enable Mr. X to get one of them, some other person goes without work. This is a complete fallacy. However it is easy to see how it arises. In the most immediate sense, the proposition is true. If an employer has a vacancy and, due to a job subsidy, Mr. X gets it rather than Mr. Y, Mr. Y remains temporarily unemployed. But by definition Mr. Y is inherently employable. If he does not get this job, he will offer himself for others. Employers will find there are more employable people in the market and that they can more easily fill their vacancies. This increases downward pressure on wages, making possible a higher level of employment at the same level of inflationary pressure. On average over the cycle the level of unemployment is determined at the level needed to hold inflation stable. Active labour market policy 58 The Social Challenge of Job Creation increases the number of employable workers, and thus reduces the unemployment needed to control inflation. Equally, in the short-run a government that has a given inflation target (or exchange rate target) will allow more economic expansion if it finds that inflationary pressures are less than would otherwise be expected. Many people find it difficult to believe that (inflationary pressure equal) jobs automatically expand in relation to the employable labour force. So we devote the whole of Section 3.2 of the paper to that issue. 3.1.5 Benefits and costs We can now proceed to sum up the effects of the scheme and its impact on human welfare. In a formal sense it would abolish long-term unemployment. However this is to overclaim since someone who reverts to unemployment after 18 months (after his temporary job) is not really short-term unemployed, even though this would be his classification in the statistics. So let us consider the impacts on the flow of a cohort entering unemployment. During the first 12 months, some people may, it is true, delay taking a job because their potential employer has an incentive to wait for the subsidy. But more people will take a job who would not otherwise have done so because they would not like to end up on the programme. The hope is that a completely new climate would develop in which neither individuals nor the Employment Service accept the idea that someone should reach the humiliating position of being confronted with temporary work as the only possible source of income. In Sweden in the 1980s typically about 3 per cent of the workforce reached the 14th month of unemployment (when benefit ran out): in Britain the figure was about five times larger. Going on, between the 12th and 18th months all the cohort is now employed. After the 18th month the proportion employed should be very much higher than it would have been, due to the employability of those concerned. Thus it is reasonable to suppose that unemployment would fall by roughly the same size as the stock of long-term unemployed, leading to a substantial increase in production. Suppose average European unemployment fell to 5 per cent compared with a counterfactual rate of say 9 per cent. Output would be at a minimum 2 per cent higher. This is the social gain (not to mention an additional non-income related gain in psychic well-being among those affected). What is the social cost? Very little. The employment service would need more administrative staff, bul: this is a tiny cost compared with the gain.5 (The typical EC country spends only 0.1% of GNP on its employment service.) The balance is also favourable if we focus exclusively on the benefits and costs to the public finances: Preventing Long-Term Unemployment: An Economic Analysis 59 (i) After the 12th month the taxpayers stop supporting those who are already fraudulently in work. (ii) Between the 12th and 18th month, the taxpayers keep paying benefit but now it goes to employers not workers. However an employer who would anyway have hired somebody unemployed between 12 and 18 months will of course claim the subsidy, so that there would on this account be some deadweight - i.e. extra expenditure. (iii) After the 18th month, there will be major savings on benefits and extra taxes received. On any reasonable estimate the total of all these will be a positive saving to the government, and a saving higher than the extra cost of the Employment Service. 3.1.6 Carrot and stick Why does this analysis seem so much more cost-effective than most existing active labour market policy? Because it is much more drastic. Job subsidies without compulsion to accept an offer can easily be ineffective. Consider for example the proposal put forward by Snower (1994) which has inspired a recent British government initiative. The idea here is to make possible the conversion of a person's unemployment benefit into an employment subsidy, but not to make it mandatory. While the social net benefits should be positive, they may well be small. Major falls in unemployment are unlikely down this route. What is needed is a shift of regime.6 No one would now design a system like the existing one. But it requires courage and commitment to change it. One thing however is sure. Unless it is changed, we shall be almost as far from the EU's target early next century as we are now. In the rest of the paper, we first discuss the issue of substitution and displacement (Section 2). We then in Section 3.3 review the effects of existing work-based policies in Sweden and the US, as a basis for evaluation of our own proposal. 3.2. SUBSTITUTION AND DISPLACEMENT Programmes to help unemployed people have always been subject to two types of criticism. First, they may help people to do things they would have done anyway. Such expenditure is called "deadweight" since it has no effect but involves a public outlay. The social cost of this public outlay is the excess burden of the tax that financed the outlay. While this can be an important issue, it is not the main criticism. 60 The Social Challenge of Job Creation The second and more serious objection is that, if unemployed workers get jobs they would not otherwise have got, this may not increase total employment but simply deprive other workers of jobs. This can happen either if each firm employs the same number of people as before but just substitutes one lot of workers for another, or if some firms expand employment and output but displace employment in other firms. 3.2.1 No job fund Such arguments taken to the limit are based on the idea that the total number of jobs is somehow fixed, presumably by the level of aggregate demand. But there is no reason to suppose that demand is ever the main constraint in an economy. The monetary and fiscal authorities can always generate more demand. The constraint is the inflation constraint. This is illustrated by the Phillips curve A0A0 in Figure 3.2. When the employment rate is above (l-u0*) inflation tends to rise, and vice versa. Most governments and electorates seem to have some kind of inflation objective. Given this objective, the level of employment depends on u*. Only policies which alter u0* will change the actual level of unemployment. But, conversely, if a policy reduces u0*, it will reduce u. This is illustrated by the new inflation constraint AjAj. There is no fixed number of jobs to be done. Given the inflation target, the number of jobs is fixed entirely on the supply side of the economy. 3.2.2 Employabiiity The main thing that determines the number of jobs is the number of "employable" people in the economy. Economists generally take for granted the idea that ceteris paribus the number of jobs rises in proportion to the labour force, so we will for the moment take that as read. The more difficult issue is the notion of "employabiiity". People clearly differ along a wide spectrum of employabiiity. Near one end is Mr. A: a skilled worker who is willing to take any job and searches every day. Near the other is Mr. B: unskilled worker with an excessive reservation wage who only samples the job market once a month. If there are vacancies, Mr. A will probably be hired soon and Mr. B after a longer spell of unemployment. More specifically, we can denote the "employabiiity" of an individual c: and the average employabiiity of all unemployed people c. Then the total number of unemployed people hired in a given period (H) will depend on the number of vacancies (V) and on the number of unemployed people (U) weighted by their average employability(c).7 Hence H=f(V,cU) fpf2>0 (1) Thus our concept of employabiiity refers to the capacity to fill vacancies. Preventing Long-Term Unemployment: An Economic Analysis 61 How then does the employabiiity of the unemployed affect the number of jobs - for a given inflation path? The path of inflation is given by the wage-price spiral, which we shall depict in the simplest possible form. Prices (p) are a mark-up on expected wages (we) so that, using small letters for logarithms: p-we = b0 (2) Wages (w) are a mark-up on expected prices (pe), and this mark-up is affected by "inflationary pressure", denoted by 0 and defined below. Thus w-pe =J>o + 0 (3) Substituting expected prices from (2) we have w -we =ßß + y q + 0 Jf price inflation is perceived as a random walk, then when w=we inflation is stable; when w>we inflation rises; and when w0+yj (HI ell) If unemployment is constant, hires equal separations i.e. employment (N) times the separation rate (s). So w - we =ßß +yo+yjs I (cli IN) Hence for a given inflation path, unemployment is inversely proportional to average employabiiity (c).9 The basic concept of this paper is that cU is a constant. More generally, if Uj is the number of unemployed of type i, ZcjUj=constant. Going on, we could for simplicity assume that there are only two types of unemployment, short-term and long-term, and that long-term unemployment causes people to be less employable (cLsN while unemployment falls. 12 If s is constant there is a second-order rise in sN and H, due to the rise in N. 13 If the population of working age is used on the horizontal axis, the diagram still works well. Preventing Long-Term Unemployment: An Economic Analysis 69 REFERENCES European Commission, Growth, Competitiveness, Employment. The Challenges and Ways Forward into the 21st Century, Brussels, European Commission, 1994. Gueron, J.M., "Work and Welfare: Lessons on Employment Programs", Journal, of Economic Perspectives, Vol.4, No.l, p. 79-98, 1990. Layard, R., Nickell, S., Jackman, R., Unemployment: Macroeconomic Performance and the Labour Market, Oxford, Oxford University Press, 1991. Layard, R., Nickell, S., Jackman, R., The Unemployment Crisis, Oxford, Oxford University Press, 1994. Snower, D., "Converting Unemployment Benefits into Employment Subsidies", London, Centre for Economic Policy Research, Discussion Paper No.930, 1994. ©Jordi Gual 1996 AU rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior permission of the publisher, Published by Edward Elgar Publishing Limited 8 Lansdown Place Cheltenham Glos GL50 2HU UK Edward Elgar Publishing Company Old Post Road Brookfield Vermont 05036 US British Library Cataloguing in Publication Data The social challenge of job creation: combating unemployment in Europe 1. Job creation - Europe 2. Manpower policy - Europe I. Goal, Jordi 331.1'25'094 Library of Congress Cataloguing in Publication Data The social challenge of job creation: combating unemployment in Europe / edited by Jordi Gual. Includes bibliographical references and index 1. Job creation—European Union countries—Congresses. 2. Full employment policies—European Union countries—Congresses. 3. Job creation—Spain—Congresses. 4. Full employment policies—Spain--Congress. I. Gual, Jordi HD5764.A6S63 1996 96-15145 331.12'042'0942—dc20 CIP 1 85898 459 9 Printed and bound in Great Britain by Biddies Limited, Guildford and King's Lynn Contents Contributors vii Prologue ix Carlos Cavallé INTRODUCTION 1. Four Myths about Employment 3 Jordi Gual PARTI POLICIES AND INSTITUTIONS FOR JOB CREATION 2. Employment in Europe 29 Jacques H. Drěze 3. Preventing Long-Term Unemployment: An Economic Analysis 55 Richard Layard 4. Does it Fit? Drawing Lessons From Differing Labour Practices 70 Richard B. Freeman PART II SPAIN: LESSONS FROM A FAILURE IN JOB CREATION 5. Job Creation in Spain: A Macroeconomic View 91 José Viňals 6. Creating Employment in Spain: Labour Market Imperfections 119 Carlos Sebastian 4 The Social Challenge of Job Creation the multidisciplinary approach of IESE to business education was reflected in the conference. This gathering took place in Barcelona in the autumn of 1994 and brought together economists, philosophers, sociologists and business leaders from several countries to discuss policy proposals that could aid in the generation of employment, particularly in Europe. Some of the papers presented at the conference are collected in this volume. This introductory chapter will discuss four common fallacies about the difficulties of generating employment in industrialised countries. It does not pretend to constitute a survey of all that is known about employment creation.1 Rather, it looks at a few common misunderstandings about the factors that contribute to job creation, as a motivation for presenting the papers included in this volume. Three of the myths to be examined are directly or indirectly discussed in detail in the chapters that follow. We will thus place the contributions of this book in a broader context. The fourth myth, which refers to the relationship between employment and trade, is not covered in this volume, and will therefore be the subject of a more complete discussion. This last section can be seen as a selective survey of a topic which is currently the subject of hot debate. 1.1. EMPLOYMENT AND TECHNOLOGICAL CHANGE There has always been a misunderstanding of the impact of technological change on employment. This erroneous perception was already present during the industrial revolution and has been recurrent in periods of rapid technological progress. Our first myth can be stated as follows: Myth 1. Given the current trends in technological change, there are not going to be enough jobs for the whole population. The citizens of industrial societies will have to accept "technological" unemployment and adapt to an increase in leisure time. The fears of technological unemployment are widespread, particularly among European citizens, as they witness that each new economic expansion over the last 20 years has been unable to bring unemployment back down to the level' attained in the previous cyclical peak. It is certainly true that Western societies will have to adapt to a reduction in work time. However, this is nothing but the continuation of a long-established trend and a reflection of the augmented well-being of society. This betterment stems precisely from the total factor productivity increases which technological change has brought about.This growth in real income has in turn led to an increase in the value that citizens attach to leisure. Four Myths about Employment 5 But despite these arguments, the main problem with this first fallacy is that it constitutes a well-known but still common misconception. It is based upon the idea that the total amount of jobs is fixed, determined by what is required in order to produce the goods and services demanded in the marketplace. If technological change allows the satisfaction of this demand with a diminished use of workers, the story goes, then these are jobs that are lost. Of course, this argument is wrong. As Professor Layard points out in his contribution to this volume, aggregate demand is not the main constraint on job availability. IT necessary, monetary and fiscal authorities can always transitorily generate more demand. Moreover, technological change - with the subsequent increase in total factor productivity - generates real returns, either in the form of lower final prices for goods or in increased wages and profits. These are increases in real income which in the end result in higher demand. Although technological change may be the direct cause of job losses in a particular sector, the increase in income heightens the demand for labour elsewhere in the economy. Indeed, jobs cannot be created simply by artificially increasing the aggregate demand of the economy. The true constraint for employment creatipnjs the-inflation constraint. The inflation target determines the level of aggregate demand, but the number of jobs compatible with that jňflatťon level is determined entirely by the supply side of the economy: whaľProfessor Layard calls the "employability" of the labour force. A worker's employability depends on her willingness to accept a job and on her adequacy to the job market's requirements. A more "employable" labour force will make a higher level of employment compatible with a given inflation target. The explanation is simple. When aggregate demand peaks up, the employable workers limit the reappearance of inflation since they are abfe to compete for the new jobs. Non-employable workers cannot bid for the new jobs, and the expansion of aggregate demand may lead to wage inflation. In his paper, Professor Layard exploits the notion of employability to design a plan to reduce long-term unemployment. This kind of unemployment is the natural "target for measures that attempt to reduce the number of unemployed without increasing the rate of inflation, since long-term unemployed workers exert no downward pressure on labour markets. The notion of employability is intimately linked, to the implications of technological change for labour markets. Rapid technological change displaces workers with outdated abilities and creates demand for workers with a different expertise. Technological change does not create aggregate unemployment; it triggers profound"changes in the structure of labour demand. The difficulties for employment arise due to the inability ojfthe labour supply to adjust quickly to the new demands of the labour market so that enough employable workers are available. The decline in 8 The Social Challenge of Job Creation This will lead to a slow employment reaction in recoveries and to an excessive persistence of high unemployment levels. Other persistence mechanisms are the result of the dual structure of the labour market, with insiders and outsiders exhibiting different behaviour in terms of wage negotiations and attitudes toward work. Recent claims that demand policies should be used in the fight against unemployment (see for example Blanchard et al., 1994 and Alogouskoufis et al., 1995) can therefore be justified on several counts from a theoretical perspective. This, however, is still not a predominant view among practitioners, particularly in terms of the policy prescriptions of most international organisations. Their recommendations grant clear priority to fiscal consolidation and the fight against inflation. While recognising that it is dangerous to advise policy makers to engage in fine-tuning and that there can be no long-term trade-off between inflation and unemployment, the proponents of active demand management emphasise the high costs in terms of employment and output which may result from the combination of sharp drops in aggregate demand and the hysteresis effects "pervasive in employment markets. 1.3. EMPLOYMENT AND THE STRUCTURE OF LABOUR MARKETS A consensus appears to be taking shape among the major international organisations with regard to the need to liberalise labour markets in order to improve the employment situation in industrialised countries.2 This view has been greatly influenced by the relative employment performances of the US and the EU over the last 20 years. As Dr Viňals points out in his essay, both areas grew at an average annual rate of 2-2.2% over the period, but the US was able to increase employment yearly at a rate of 1.6%, whereas Europe managed only 0.5%. If the Spanish case is worth examining at all, it is because not only did Spain fail to create employment, it actually destroyed it, at a rate of -0.4% per year. General political trends, as well as the development of new thinking in economics, have contributed to the increasing popularity of the deregulation of markets, and in particular of the labour market. Although it is true that prices set freely by private economic actors tend to clear markets (as does, in principle, the wage), I believe that the present state of opinion has led to the development of a third myth, which can be stated as follows: Myth. 3. The deregulation of labour markets, in terms of both wage-setting procedures and contractual conditions, will facilitate the creation of employment and thus contribute to an overall improvement in living standards. Four Myths about Employment 9 The fundamental problem with the deregulation of labour markets is not that it might fail to generate employment. It would probably succeed. The key issue is whether such deregulation would lead to employment creation's ultimate goal: namely, the improvement of living standards for a broad majority of the population. There are at least two reasons to doubt that a thorough deregulation of labour markets constitutes the right strategy if one seeks to create jobs that will lead to a widespread improvement in the standard of living. The first argument is based upon the observation of employment trends and real earnings in the United States. Professor Richard Freeman sets it out clearly when he observes that fully employed American workers with low wages have living standards below those of similar workers in Europe, despite the fact that"the US enjoys a higher overall standard of living. The strong US performance in employment creation has been accompanied by an increase in wage inequality among workers with different skills, and by an actual decline in real wages for the low-skilled. Concomitantly, US society has shown disquieting indicators' of the growth of a permanent underclass, with sharp increases in poverty rates and in crime (this point is also put forward by Professor Drěze in his paper). As Professor Freeman writes, "Countries that maintained the earnings of the less skilled seemingly 'paid' in terms of high unemployment; while the US 'paid' for its growth of employment through falling real earnings." The second reason to question full deregulation has to do with the imperfect nature of labour markets. Although it is probably true that most labour markets in Europe are overregulated, it does not follow that the appropriate policy is to dismantle all regulations. There are some sound reasons to regulate-certain features of the labour market. Reforms should scrap unnecessary rigidities and restrictions but preserve those regulations that attempt to correct the imperfections of the market. Professor Drěze argues that full labour market flexibility would subject workers to excessive income uncertainty. On the one hand, this could lead to inefficient levels of volatility in aggregate demand. On the other, more fundamentally, workers invest in human capital which cannot be diversified away as easily as other forms of wealth. This may justify a reduction of income volatility through some degree of rigidity in real wages. There are other possible justifications for some degree of regulation in labour markets. Prominent among these are the existence of an asymmetrical distribution of information in that market and the problem of time inconsistency in the contractual relationship.3 Asymmetrical information can be significant to the extent that workers may have difficulties in assessing the characteristics of the jobs being offered (e.g., in terms of health and safety), whereas employers are likely to be much better informed. The problem of time inconsistency refers to the acquisition of firm-specific knowledge by workers. Once this know-how has been acquired, T'1 ""¥?>■■■. ■■_.- A!V.aŕ%:;.:.^v7!'i^ l O The Social Challenge of Job Creation there is no incentive on the part of the firm to provide the workers with an ?PPro£riäIe^wärďm (JThe absence of markets fojJ^waxd^r-caiitingent transactions reinforces ^e persTstence^ôf equilibria with excess capacities and unemployment. An analogyis ihstructiverit comes from' th(^fieórjjž£^jük-föad mcjitg-i ,i,e. pncineDf^ßgiäi^iifed^ffiäft'i- Electricity is thďsTarrrJTmř^x^mplěTFifficiency 'cMß'jOT^EciLe^uaHo marginal, invariable, cost at times of excess capacity (low demand). All the costs due to eapacity investment, or Fixed costs, are covered thanks to theTu^^ . twffen capacity išTuHý ušed).'TITe same logic applied to mäcroeconömtc fluctuations calls for recouping investment costs through a high mark-up ■1 34 The Social Challenge of Job Creation f f z Ň£ f in the case of full use of capacity, while covering variable cp,sts alone in , // the case of exTešrcapäcTfyTJi"'""""""" '''' 9Ho tíT'H/iVf^ n r (J,l?í\ F-oVffľ )"1.....' Because we are now considering mutually exclusive uncertain paths, the aMYVí ^^[y^nc^oť i firms in reeessions_.wapuld require^^ fomsl^rni^ to the sWeToFunc^^ financing can do that - butTfapplies to a/ Vl ( sm^l5lTáT^nnvestment flows.*P F^t^eďeji^^___íV u-\ A instead lead firms to fflairaainjo[yejricy by ccjjecjx^ěl^ > ,.'' ;v excess of variable costs evenjyn recession, which jjm its do w n w ard price x y Vi^^^Y^^^^^^^^^' persistence of undef-activity^viW"excess cabaciües.11 Of course, excess capacities are eventually removed by ridn-Tepíacľenient, in contrast to excess labour supply, which explains the pattern of Figure 2.2: successive returns to high rates of capacity utilisation ' . ' ■ have been associated with successively higher rates of unemployment. .,•., One might hope that business Firms be relieved, at feast partly,.of their finandj^bjjj^en^ Thofsocial cost of additional output ]s"7n~en lower than at future times wheTPéxčěss capacities will have ' disappeared, so that reajjn|ej£sj^^ But nominal lí iJ interest rates cannot become negative. Hence, negative real rates call for ) anticipated inflation exceeding nominal rates. But monetary authorities"~j\ !$j ., f react to inflation forecasts by raising nominal rates, and thus bar the road to financial relief. I conclude that 'demand stabilisation policies are needed,; in decentralised market economies. Such policies are met with' justified scepticism, when they aim at Fine tuning: our tools for measurement and Employment in Europe 35 control are not fine! But I am not advocating fine tuning. I am advocating support of growth over a decadejo reduce mass unemployment by 7 or ^%ľ"13elT^n3™llabilisatiorí"poTicies, however, must be pursued at a Eurojr^an-Vjä^J^eJ. Europe is sufficiently closed to make such policies affectiversTrice foreign trade accounts for only 9% of EU 12 GDP (a percentage that will decrease further with the entry of new members). Demand stabilisation on a narrower basis is illusory, as was revealed for instance by thefperman "_locomp.tiv,e,.Iexpsp or by the French isolated expansion after 1981. 2.3.2 Wage moderation I turn now to wage developments. The level of real wages matters on three counts: (i) The first count is capital-labour substitution - the engine of long-term growth, but a form of waste under unemployment. In gross terms, i.e. ignoring the labour content of capital equipment, this substitution reduces employment in Europe (but not in the US) at the rate of roughly 1 % per year. It is likely - though not documented T' econometrically, to my knowledge - that this process boils down in large part to substitution of skilled labour for unskilled labour. (In small countries, it also amounts to substituting imported capital goods for domestic labour.) (ii) The second count is international competitiveness. High wages curtail exports and encourage imports. That aspect is very important for small open economies like Belgium, much less so for relatively closed areas like Europe. (iii) The third count is profitability, wjthojJt_which investment stalls and reorganisations with" negative: emrjlo^meiit., cgri^^Fňces^are earned out. These three effects would seem to overpower the positive contribution of wages to consumer demand stabilisation. Next to the level of real wages, their evolution matters: any sign of wage inflation triggers - rigrTHyT^fl^r-ongly, meaning sometimes rightly and sometimes wrongly - restrictive monetary policies; these discourage investment, curtail profitability and inducepessmii^dcexpeHällöns^ with ™furtliéTlíegativě~ěTTec^ consumption and investment."" ' "~~ The contrast between wage and employment developments in Europe and in the US is striking, as illustrated by Table 2.1 or Figures 2.3.a-2.3.b. 36 The Social Challenge of Job Creation . Table 2.1 Comparisons Europe 12- - United States Growth rates 1960-1990 % Europe United States Real GDP Employment Real wages 3.3 0.3 3.0 3 2 1 Source: European Economy, 54. 12 10 6- 4- Total Employment Unemployment rate I960 1965 1970 1975 1980 1985 1990 Figure 2.3a Employment and rate of unemployment, Europe 1960-1991 cent oo O Total Employ ment —.......".......... _,.••• yz.. ^ - 120 : 100 § Unemployment rate .............. 1) f. -a. ° •■• '--.. ^^j>^T~~^ '"" \ / -80 4- —-"""^ "......-..../ - fin 19 60 1965 1970 . 1975 1980 1985 199 3 Source: A. Lindbeck, Unemployment and Maci Press, 1993. 'oeconomics, Cambridge, Mass., MIT Figure 2.3b Employment and rate of unemployment, US 1960-1991 Employment in Europe 37 Wage formation in Europe has witnessed some errors - like the rise of real wages in 1974-76 (overJ0%_for_EU 12, 14% for France, 15% for felgíum), at a" time"ofcoTIectJveTrnp^veríshment through terms oTtriHe deterioration. There is also a statistical error. Wages injiurope (but not in the US) incorporate rar^iyand almost fully gams"Tn gross average productivit^l^jd^ without regärďľor the fact that these prodtřctivitý gainsHi) reflect in part capital-labour substitution induced by the wage increases - which creates a spiral: wages-productivity-wages-prices-wages;12 (ii) reflect in some cases scrapping of older idle equipment and not technological or managerial advances. How do we explain the resjstancejofjeal wages in the face of persistent unempIöymenl^We^^ labour markets. At the low end.pfihe^kills ladder, where unemployment is most severe, the wa^efloor comes from unemployment benefits and minimum wages (either legal or negotiated). This floor has no counterpart in the US, where benefit payments last at most six months while minimum wages are either non-existent (in some states) or ineffective(JnJEinjgEe^the^age.vfJ2or_ seems to reflect a broad social consensus, a collective desire to guarantee to every worker a rnirumaTíňč^me, still not generous for families with no other source of income.13"^ most European countries, the wage floor has prevented the rise of poverty witnessed in the US. We must maintain that kind of income protection for low-skilled workers, while continually trying to find more efficient schemes. At the upper end of the skill ladder, on the other hand, wages and salaries are geared to equilibrium of supply and demand. True, markets remain impeŕTectrsaläŕ3TlIiŤf^rehtials between firms exceed plausible compensating "differentials; but there is no evidence of either upward or downward bias, except for narrowly defined jobs where wages come closer to a 'contest outcome than to a competitive outcome. Between these two extremes, there is a broad range of intermediate qualifications, where long-term unemployment is rare, but work^TJefow quálifi cTtTon_is^^ •" 'In that farTg^Twiiere wages are to some extent lifted from above and propped from below, negotiations between employers and trade unions play the major role. The recurrent plea for flexibility is presumably addressed primarily to that middle rangejfr should However be stressed that full wage flexibility išjněffic^^ and long-run labour contracts are limited inscope. If trie wages did adjust continuously to clear ärnátTOTJnriaTlčeTs, tne resulting income volatility and uncertainty would be a hardship to workers, who cannot diversify risks on their human capital to the extent made possible for non-human wealth by investment funds and related assets. That income uncertainty must be curtailed, all the more so as it would exacerbate demand volatility. Constraining real net wages, both upward and downward, is more efficient.14 It also follows that 38 The Social Challenge of Job Creation \ some resistance of j;eal wages in the face of persistent unemployment is ; justified~ofi efficiency grounds. From a positive viewpoint, the dominant opinion among macroeconomists is that wage._inflatioaJs^jiegati^ unemployment: the Phillips curve.rs" Under that relationship, a''fan in unempjoyinent automaticallytriggers some wage inflation^ unlesT'Tfie process of wage formation is tampered" with. That viewpoint, which____ underpins the plea for structuralJQexjbiljty, seems hardly convincing at K. unemployment rates approaching 12%. Yet I haW~"říBlarď""S"eTrous '' macroeconomists advertise -liořrěsco referens! - an equilibrium rate of unemployment of some 10% in Europe today. A more reasonable, and widely accepted, viewpoint suggests le_aving out of these calculations the r7 long-term unemployed andJow:-skilled^ol:leTsľThis provides yet another reason to concentrate on creating low-skill jobs, o Aj! '2lAt■$i- t± !- ?4r. V' tTo COT£[iľďejjpersistent unemployment is accompanied by a major' distortion of the price system. The wage costs of firms do not reflect correctly the scarcity of low-skilled labour. The wedge between private ^_ <^ and social costs includes all labour taxes16 (so~č"íaTinsurance contriButions and income taxes), plus some^uř^mglu^ment beneiTtšTTwo-thírds of the labour costs saved by a firm that does nófTéplače a retiring worker are borne by public budgets. Our economies operate with a fundamental price, the Töw-skiTIed wage, which is grossly biased relative to the underlying reality. That blatant distortion must be corrected, knowing that a return to full employment is ten years distant, at best. 2.4. POLICIES FOR GROWTH How can we bring about a decade of employment-generating growth? By bringing about sustained growth of aggregate demand and reasonable wage developments. Is that possible? Does that correspond to the current orientation of macroeconomic policies in Europe? Let us take a look. 2.4.1 The recommendations In the current publications of international institutions, for instance the OECD jobs study or the annual report of the European Commission, three lines of action are stressed: (i) Budgetary restraint and inflation control, as per the guidelines of theTvIááštricht treaty, should pave the way for lower interest rates, and for realisation of the European Moneľäŕy iJníohľ Lower interest rates will then stjjiLuMe.mveiim^ ........ (ii) Labour market flexibility should spur employment. A long list of Employment in Europe 39 proposals covers flexible wages and hours, part-time work, elimination or flexibility of minimum wages, lower social insurance contributions, especially on low wages, firrn-levej^ rather than i?,K higher level wage bargaining and so on. (iii) Education, training and active labour market policies should ■--■■■"' increase the employment prospects of low-skilled or long-term unemployed workers. These recommendations are generally well-founded (with some exceptions, like elimination of minimum wages). But they fall short of /"",""", target. I am of course in favour of lower interest rates, but_I dojjot think v^J that monetary policy alone can j>tim^ o9rnänd (due lrr^äTtkro1^fTr3~ffie" non-negativity constraint on nominal rates). Fhsxjbiht^ stimulates hirings in a growing economy, but it stimulates "7 y nnngsaurihg recessions, äs was confirmed recently. The link between ' * laBoW^arEéT^TTéxiřjTlTfý and wage moderation remains indirect and perhaps tenuous. Flexibility is not the universal remedy which is sometimes advertised. Training and active labour market policies bring > their beneficiaries closer to the head of the queue.of job-seekers, but they do no^red^cejhj^pjverall length of the queue - not until full employment^;.-. ; ; is'in sight, as might hopefulty be the case in 7 or 8 years. '"" r{", fTKe White Paperjon Growth, Competitiveness and Employment of the Europe1mT!öTnmission is more ambitious. It does not diverge from the above recommendations. Butjt introduces some specifictargets - like a lowering of social insurance contributions on low wages amounting to 1 or J 2% of GDP by year 2000, with substitute funding coming from a European-level tax on energy (C02) or a uniform withholding tax on interest income. The White Paper also contains some investment proposals, extending the Edinburgh growth initiative. These concern in the first place transeuropean networks for transportation, energy and information. They concern new investments related to environmental protection, to reconcile growth and ecology. They also concern - in less specific terms, unfortunately - investment in urban renewal, low-cost housing, urban transportation, and so on. These programmes are not motivated by a demand stimulation goal, but rather as contributions to the growth potential or to social welfare.17 Needless to say, my quotations from these documents are a biased selection. I read them in the wake of the attempt by a group of a dozen economists, convened by Edmond Malinvaud and myself,18 to define the scope of a European growth and employment initiative. That attempt has led to a consistent set of policy proposals,19 ranging through lower interest rates, budget restructuring and wage moderation, with special emphasis on two essential medium-term policies: lower wage costs for unskilled laboljř7~Tfn^gTí elimination of employers, sociaf ínšurähčé contributions on minimum wages (with substitute funding from a tax on 40 The Social Challenge of Job Creation energy or interest income, or from VAT); and demand stimulation through investment programmes inQujbari^j^ney^al, /low cost housing, urban transportation and franseuropean networks, witH empTo^ment subsidies (rěfíěTTroni labour taxes) on the labour content of the investments. We are back to the two pillars of sustained growth: wages and aggregate demand. 2.4.2 Investment Regarding investment, I do naturally deplore the lack of concern for demand stabilisation in the OECD study and in the White Paper. The current state of hiacroeconomic thinking is reflected there. And I regret the vagueness and lack of instruments in the presentation of the White Paper - except perhaps for transeuropean networks and environmental protection. In these two areas, the list of projects is impressive. It adds up to 574 billion ECUs over the period 1994-99,20 or some 1.5% of GDP over the six years. This would undoubtedly make a significant contribution to aggregate demand. The order of magnitude is similar to that advanced by our group before seeing the figures in the White Paper. Our proposal puts more emphasis on projects using extensively low-skilled labour, while .-V-- meeting unfilled needs - namely urban renewal and low-cost housing. Třeaíísě ChHTöü"r proposal to subsidise the labour content of investments calls for preparations, hence delay. But it provides the needed correction to the two price distortions noted above, namely wages and real interest rates. This is in the spirit of