{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe during the First Years of Transition
Ariane Pailhe´ Abstract. The labour markets of the centrally planned economies of Central Europe (the former Czechoslovakia, Hungary and Poland) were divided into several segments. Using estimates taken from a two-regime model, we show that segmentation has persisted throughout the first years of systemic change. However, labour market segmentation has evolved to some extent. Firms that used to have priority now coexist alongside new activities within the primary segment (in particular, foreign firms and activities in the banking and financial sector). In this way, labour market segmentation results both in the appearance of new, formal institutions and the persistence of informal institutions left over from the past, owing to the growth in market uncertainty.
1. Introduction According to the segmented labour market (SLM) theory, wage disparities do not usually depend on variations in individuals’ abilities but mainly on the job held as well as on the structure of the labour market. This approach argues that the labour market is not a single competitive market, but is composed of two or more non-competitive segments. Hence, workers of equal potential productivity may receive different rewards depending on their job sector. SLM theory was initiated by research work carried out by American institutionalists in the 1940s and 1950s as an alternative
Ariane Pailhe´, Institut National d’Etudes De´mographiques, 133 bd. Davout, 75020 Paris, France. E-mail:
[email protected]. I would like to thank Jiri Vecernik, a senior researcher at the Institute of Sociology at the Czech Academy of Science, for giving me access to the database, and Christine Siegwarth Meyer, Assistant Professor at the Department of Economics at Bentley University, for sending me examples of the program she wrote in collaboration with STATA Corporation. LABOUR 17 (1) 127– 152 (2003)
JEL J42, P31
# 2003 CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
{Journals}Labr/17_1/i189/makeup/I189.3d
128
Ariane Pailhe´
to the standard view of the labour market (Dunlop, 1958; Kerr, 1954). The focus was placed on the demand side, on collective and institutional mechanisms. It developed rapidly during the 1970s (Bluestone, Harrison, 1982; Doeringer, Piore, 1971; Edwards et al., 1975). The theory’s main results were then integrated into the neoclassical framework through efficiency wage theory (Akerlof, 1982; Akerlof, Yellen, 1986; Perrot, Zylberberg, 1989; Shapiro, Stiglitz, 1984), insider– outsider theory (Lindbeck, Snower, 1988) and implicit contracts theories (Azariadis, 1975). Contrary to earlier explanations, they explain the existence of distinct segments on the basis of individual behaviour and imperfect information. Some studies referring to this theory have been carried out on centrally planned economies. The analyses of Stark (1986) and Malle (1986) helped to conceptualize empirical studies, particularly those estimating from Hungary.1 Whereas in a market economy the formation of internal markets2 comes in response to the uncertainty linked to the market itself, in the socialist economies it was aimed at reducing systemic uncertainty caused by administrative regulation. In a context of constant labour shortage, firms had to maintain a reserve stock of manpower, particularly the most qualified, in order to deal with sudden increases in production. However, firms tried to make labour more flexible within their establishments by modifying work rates and dividing posts in order to deal more effectively with the unforeseeable problems of input delivery.3 Firms had every reason to offer qualified workers attractive payment, benefits in kind and=or promotion prospects. Managers in centrally planned economies disposed of a certain amount of freedom as far as wage fixation within their establishments was concerned, even if central government did control the global pay packet. At the macro-economic level, the primary segment presents the same characteristics as a firm’s or an establishment’s internal market; it is formed by a series of internal markets. In this way, it is made up of the most stable and most highly remunerated posts, and those offering the best work conditions and the best promotion possibilities. Labour turnover is, of course, lowest in these jobs. The primary segment thus groups high-status positions, in opposition to the secondary segment (Piore, 1983). In the countries of Central Europe, the state intervened in the structuring of the labour market; the central authorities tried to encourage strategic sectors in order to help firms fulfil the goals set by the Plan. The upper segments comprised large firms and priority # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
129
sectors, such as mining, mechanical and military industries, constituting poles of attraction for labour and subsidy. Firms in these sectors also took advantage of their power to negotiate high salaries with the ministry in charge of their branch. Segmentation thus varied depending on the branch and sector concerned. The change of economic system, especially the suppression of state paternalism and the gradual privatization of the majority of state-run firms, led to major changes in the labour market. Unemployment spread rapidly throughout Central and Eastern Europe, although at varying paces depending on the country (Boeri, 1997) and wage disparities grew sharply (Ko¨llo¨, Kertesi, 1998; Milanovic, 1998; Mu¨nich et al., 1999; Rutkowski, 1997). Radical institutional changes affected the market for firms’ products, as well as their technological levels, their size and their methods of wage determination. Sector transformations, as well as increased labour flexibility and uncertainty, may therefore have caused certain firms and=or branches of the primary segment to shrink. Nevertheless, economic studies have seldom used SLM theory to analyse wage inequalities. Some studies evoke the weak workers’ mobility between firms with different types of ownership (Blanchard, 1997; Boeri, Flinn, 1999). The sole theoretical and empirical study referring to SLM theory is that of Grosfeld et al. (2001). Using an indirect estimate, they show that the Russian labour market is segmented because of the new regime’s shaky credibility and the soft budget constraint. Their division is based on the sole workers’ position, i.e. white-collar versus blue-collar workers. The objective of this paper is to test labour market segmentation in countries with a stronger institutional environment, and to take into account various factors of segmentation such as the type of ownership, the sector of activity or the workers’ qualifications. The estimate is based on a two-regime model, using the Social Stratification Survey database on four Central European countries: the Czech Republic, Hungary, Poland and Slovakia. The remainder of the paper is organized as follows: the next section presents the estimation procedure and the Social Stratification Survey database. Section 3 gives the empirical results; we show that labour market segmentation continued to exist in Central Europe throughout the first years of systemic change and we determine the characteristics of the labour market’s different segments. Then we discuss how these segments came into being (Section 3). Finally, Section 4 summarizes the paper’s main results. # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
130
Ariane Pailhe´
2. Model specification and data description There are several ways of estimating labour market segmentation (Leontaridi, 1998). The idea behind the main econometric estimate models is that two (or more) distinct earning functions would indicate that the labour market is segmented. Most models predetermine segments arbitrarily on the basis of the jobs’ or the industry’s characteristics (technology, unionism, organizational structure, nature of product demand, size, and so on). Then they proceed to test for differences in the wage equations for each segment (Beck et al., 1978, for instance). Some estimates have been corrected for selectivity bias4 (Heckman, Hotz, 1986). However, this OLS estimate has two main problems. Firstly, the construction of segments is arbitrary, which could lead to unstable results. For example, segmentation based on industries runs the risk of differing from that based solely on occupations since many occupations are common to many industries. Then, some tests could validate labour market segmentation whereas other do not.5 Moreover, both primary and secondary workers can work in the same firm. Secondly, those studies, with the exception of that by Heckman and Hotz, fail to account for the endogeneity of an individual’s labour market segment, which could generate a truncation bias (Cain, 1976). Some researcher have tempted to overcome the arbitrary assignation of individuals to the segments by using factor or cluster analysis (Sloane et al., 1993). The technique developed by Dickens and Lang (1985, 1988, 1992) overcomes both arbitrary assignment and endogeneity bias problems.6 They take up a structural model that attempts to incorporate endogenous selectivity (Maddala, Nelson, 1975). According to this method, the chances of an individual belonging to a certain segment are not calculated a priori, but are based on an estimate made by the model. The position assigned to an individual depends not on one or two criteria but on several. The two-regime model can be expressed as follows: Regime 1: ln ypi ¼ Xi p þ "pi
[1]
Regime 2: ln ysi ¼ Xi s þ "si ;
[2]
where yi is person i’s wage, and p and s refer to the primary and secondary sectors, respectively. The choice between both regimes is # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
given by the condition: ( ypi if I* ¼ Zi þ "wi > 0 ln yi ¼ ysi if I* ¼ Zi þ "wi ˘ 0
131
[3]
The endogenous variables are ypi , ysi and I*. X and Z are the vectors of explanatory variables, and p , s , and are the vectors of estimated parameters. I* is the latent variable which measures an individual’s tendency to belong to the primary sector. The disturbance terms "pi , "si and "wi are assumed to be normally distributed. The two wage equations (equations with X) and the switching equation (equation with Z) are estimated simultaneously. The model’s likelihood function is: Pr("wi > Zi =Zi ; Xi ; "pi ) f("pi ) þ Pr("wi ˘ Zi =Zi ; Xi ; "si ) f("si );
[4]
where f( ) is a density function for the errors "p or "s . The loglikelihood function is thus: 82 13 0 > > >6 B Z pw " C 7 > > > 6 B i pi C 7 > C7 B X 6 C7 B u > > 6 B u 2pw C 7 > t > 4 A5 @ > 1 > : pp 9 1 0 > > > B Z sw " C > > > B i si C > = C B ss C B þ B vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi C ’("si ; "ss ) ; [5] > C B u > > B u 2sw C > > A @ t1 > > ; ss where ’( ) and ( ) are respectively the density and distribution functions of the standard normal distribution, and 2pp and 2ss are the variances of "p and "s ; we know that ww is not identifiable (Maddala, 1983), so 2ww is normalized and equal to 1 (Maddala, Nelson, 1975). pw and sw are the covariances of "p and "w and of "p and "w . # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
132
Ariane Pailhe´
In order to carry out this test, we used a full information maximum-likelihood (FIML) program. The program was applied to data from the 1993 Social Stratification Survey in Central Europe (see Box 1). We have retained two specifications of the switching equation. The variables chosen to determine the allocation of an individual to one segment or to the other are the number of years of education, a dummy for the worker’s sex and his or her region of residence (whether he or she lives in the capital or in the central region in Poland). The variables in the segmentation equation that are not in the wage equation are professional variables:7 the sector of activity and the type of ownership under which the establishment falls (model 1). The managerial position and the worker’s occupational class categories were added to these variables in a second specification of the model (model 2). As the structure of the labour market is based on the characteristics of the jobs or the industry in the SLM approach, the demand-side variables are key variables in the switching equation. The variables in the wage equations are gender, the length of education, professional experience and tenure in the establishment.
Box 1. The data The data come from the Social Stratification Survey in Eastern Europe. This survey was part of an international research project coordinated by the Institute for Research in Social Sciences at the University of California, carried out in 1993 in nine Eastern European countries, The Netherlands and the USA. The survey provides information concerning income collected in 1992, goods, housing, lifestyle, political participation and opinions. It also offers a retrospective view of the activity and education of the interviewee and his or her family. From these original samples we extracted the people in employment during the period of income declaration. We have limited our study to the wage-earning population from this group, the income of independent workers being likely to be underestimated for fiscal reasons. We have also eliminated part-time workers from the sample, being as they represent a majority of women. The samples retained number 2,895 cases for the Czech Republic, 1,868 for Hungary, 1,848 for Slovakia and 1,214 for Poland. # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
133
Labour Market Segmentation in Central Europe
The first variable allows us to measure the return on initial training. The second and the third variables allow us to measure general and specific qualifications, respectively. 3. A segmented labour market We will show that there are two differentiated segments in the labour markets of Central Europe, and that the coefficients of the wage equations differ significantly from one segment to the other. We will then study the characteristics that favour entry into the first segment. The estimates for a single labour market (OLS) and for a dual market (models 1 and 2) are presented in Tables 2– 5. 3.1 Two differentiated segments Through the use of a likelihood-ratio test we can test the hypothesis that the two-equation model fits significantly better than the single wage equation model. To do this, twice the difference between the log-likelihood of the two-regime model and the OLS [2(L* L)] is compared with a 2 with n degrees of freedom, n being equal to the sum of the number of constraints and of the number of unidentified parameters (Monte Carlo test). A single labour market model (OLS) can thus be rejected at the 1 percent level; two wage equations give a better representation of the labour market than one (see Table 1). 3.2 Distinct wage structures Moreover, the coefficients of the wage equations differ significantly from one segment to the other, particularly with regard to the return on general and specific qualifications. Table 1. Likelihood-ratio test Hungary
2(L * L) n 2n at 99%
Czech Republic
Poland
Slovakia
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
134 15 30.6
381 18 34.8
1,100 16 32.0
296 19 36.2
390 17 33.4
397 20 37.6
220 16 32.0
803 18 34.8
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Model 1
Constant Woman Education Experience Tenure
State-run establishment Mines and energy
Primary segment
Secondary segment
8.974 * * (193.7) 0.203 * * (11.9) 0.062 * * (23.4) 0.008 * * (8.1) 0.001 (0.5)
9.156 * * (81.9) 0.184 * * (3.6) 0.074 * * (11.5) 0.008 * * (4.2) 0.001 (0.3)
8.879 * * (199.7) 0.172 * * (10.3) 0.056 * * (22.1) 0.006 * * (7.6) 0.002 * * (2.0)
Switching equation 0.180 (0.6) 0.118 (0.9) 0.016 (0.9)
0.240 * * (2.3) 0.213 * * (2.1) 0.910 * * (3.5)
Primary segment
Secondary segment
Switching equation
9.009 * * (50.4) 0.095 (1.4) 0.072 * * (7.8) 0.010 * * (2.6) 0.004 (0.8)
8.979 * * (168.3) 0.207 * * (10.5) 0.055 * * (17.0) 0.004 * * (5.6) 0.004 * * (3.5)
0.627 * * (3.9) 0.038 (0.7) 0.038 * * (4.1)
0.330 * * (6.3) 0.015 (0.3) 0.665 * * (7.8)
Ariane Pailhe´
Budapest
OLS
Model 2
{Journals}Labr/17_1/i189/makeup/I189.3d
134
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Table 2. Estimations of a two-regime model in the set of Hungarian employees
Construction Manager (10 people and more) Manager (1– 9 people)
Standard deviation of error
Log-likelihood
0.37 (61.1) L ¼ 778
pp ¼ 0:28 (26.9) pw ¼ 0:02 (0.5)
ss ¼ 0:22 (38.2) sw ¼ 0:48 (39.8) L ¼ 711
* Significant at the 10 percent level; * * significant at the 5 percent level. t-statistics in parentheses. N ¼ 1;868. Source: Social Stratification Survey in Eastern Europe, author’s computation. Notes:
1
pp ¼ 0:52 (22.3) pw ¼ 0:05 (0.9)
ss ¼ 0:29 (39.7) sw ¼ 0:52 (75.9) L ¼ 397
135
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Senior executive
{Journals}Labr/17_1/i189/makeup/I189.3d
Industry
0.446 * * (5.3) 0.087 * (1.7) 0.083 (1.2) 0.417 * * (6.5) 0.266 * * (3.8) 0.591 * * (7.9) 1
Labour Market Segmentation in Central Europe
0.829 * * (3.0) 0.250 * * (2.5) 0.378 * * (2.0)
Finance
Model 1
Constant Woman Education Experience Tenure
State-run establishment Mines and energy
Primary segment
Secondary segment
7.700 * * (117.9) 0.326 * * (14.0) 0.059 * * (14.9) 0.010 * * (6.6) 0.0001 (0.3)
7.660 * * (47.5) 0.373 * * (6.2) 0.095 * * (9.2) 0.017 * * (5.6) 0.0001 (0.1)
7.652 * * (132.6) 0.241 * * (11.8) 0.050 * * (14.8) 0.006 * * (5.1) 0.001 (0.7)
Switching equation 0.127 (0.3) 0.127 (1.1) 0.025 (1.1)
0.238 * (1.8) 0.325 * * (5.6) 0.698 * * (26.1)
Primary segment
Secondary segment
Switching equation
7.683 * * (40.4) 0.369 * * (6.2) 0.096 * * (8.2) 0.016 * * (5.2) 0.0001 (0.0)
7.642 * * (126.8) 0.246 * * (12.0) 0.049 * * (13.4) 0.007 * * (5.0) 0.002 (1.5)
0.087 (0.3) 0.081 * * (2.0) 0.036 * * (2.0)
0.274 * (1.6) 0.352 * * (2.1) 0.641 * * (24.5)
Ariane Pailhe´
Centre
OLS
Model 2
{Journals}Labr/17_1/i189/makeup/I189.3d
136
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Table 3. Estimations of a two-regime model in the set of Polish employees
Construction Administration and police Transport
Senior executive Standard deviation of error
Log-likelihood
0.40 (49.3) L ¼ 624:6
pp ¼ 0:36 (17.6) pw ¼ 0:03 (0.5)
ss ¼ 0:30 (40.6) sw ¼ 0:54 (18.9) L ¼ 429:7
1
pp ¼ 0:36 (13.9) pw ¼ 0:03 (0.6)
ss ¼ 0:31 (42.1) sw ¼ 0:54 (19.1) L ¼ 426:0
* Significant at the 10 percent level; * * significant at the 5 percent level. t-statistics in parentheses. N ¼ 1;214. Source: Social Stratification Survey in Eastern Europe, author’s computation. Notes:
137
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Manager (10 people and more) Manager (1 – 9 people)
{Journals}Labr/17_1/i189/makeup/I189.3d
Industry
0.536 * * (17.3) 0.129 * * (3.5) 0.118 (0.1) 0.073 (0.2) 0.063 (0.1) 0.402 * * (4.5) 0.141 * (1.7) 0.180 * (1.6) 1
Labour Market Segmentation in Central Europe
0.624 * * (2.2) 0.126 (0.6) 0.108 (1.1) 0.048 (0.3) 0.062 (1.5)
Finance
Model 1
Constant Woman Education Experience Tenure
State-run establishment Mines and energy
OLS
Primary segment
Secondary segment
Switching equation
Primary segment
Secondary segment
Switching equation
7.926 * * (186.5) 0.261 * * (20.1) 0.052 * * (21.8) 0.002 * * (3.2) 0.000 (0.2)
8.063 * * (47.2) 0.085 (1.5) 0.040 * * (4.1) 0.002 (0.6) 0.002 (0.7)
7.863 * * (170.0) 0.279 * * (18.8) 0.052 * * (18.2) 0.003 * * (5.5) 0.001 (1.3)
0.866 * * (7.2) 0.022 (0.5) 0.005 (0.8)
8.236 * * (28.1) 0.307 * * (3.9) 0.063 * * (4.7) 0.001 (0.6) 0.004 * (1.7)
7.894 * * (87.3) 0.272 * * (9.2) 0.046 * * (14.7) 0.001 (0.9) 0.005 * * (4.0)
0.842 (0.8) 0.141 (0.5) 0.014 (0.3)
0.197 * * (4.9) 0.243 * * (4.9) 0.393 * * (7.7)
0.454 * * (4.4) 0.245 * * (2.2) 0.223 (1.1)
Ariane Pailhe´
Prague
Model 2
{Journals}Labr/17_1/i189/makeup/I189.3d
138
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Table 4. Estimations of a two-regime model in the set of Czech employees
Finance Construction Transport
Senior executive Standard deviation of error
Log-likelihood
0.34 (76.1) L ¼ 1;011:4
pp ¼ 0:56 (28.6) pw ¼ 0:52 (68.1)
ss ¼ 0:28 (51.2) sw ¼ 0:06 (1.2) L ¼ 461:3
* Significant at the 10 percent level; * * significant at the 5 percent level. t-statistics in parentheses. N ¼ 2;895. Source: Social Stratification Survey in Eastern Europe, author’s computation. Notes:
1
pp ¼ 0:32 (31.6) pw ¼ 0:39 (9.9)
ss ¼ 0:24 (20.4) sw ¼ 0:02 (0.7) L ¼ 863:2
139
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Manager (10 people and more) Manager (1– 9 people)
0.109 (1.0) 0.860 * * (3.3) 0.462 * * (2.8) 0.364 * (1.8) 0.599 * * (3.3) 0.298 * * (2.1) 0.144 (1.0) 1
{Journals}Labr/17_1/i189/makeup/I189.3d
0.014 (0.4) 0.535 * (1.6) 0.165 * (1.7) 0.248 * * (20.2)
Labour Market Segmentation in Central Europe
Industry
Model 1
Constant Woman Education Experience Tenure
State-run establishment Mines and energy
Primary segment
Secondary segment
7.949 * * (159.4) 0.274 * * (18.0) 0.045 * * (16.3) 0.005 * * (4.7) 0.002 * (1.9)
8.109 * * (65.9) 0.205 * * (3.6) 0.048 * * (6.2) 0.004 * (1.8) 0.002 (0.8)
7.857 * * (148.9) 0.244 * * (16.6) 0.042 * * (15.7) 0.003 * * (3.1) 0.002 * * (2.0)
Switching equation 0.152 (0.4) 0.165 (1.0) 0.012 (0.5)
1.118 * * (4.1) 0.243 * (1.8) 0.612 * * (2.9)
Primary segment
Secondary segment
8.214 * * (31.4) 0.211 * * (2.1) 0.046 * * (2.7) 0.002 (0.3) 0.010 * (1.7)
7.872 * * (162.1) 0.256 * * (17.6) 0.045 * * (16.2) 0.004 * * (3.3) 0.000 (0.1)
Switching equation 0.177 0.163 * (1.7) 0.041 * * (3.8)
0.676 * * (2.4) 0.207 * * (2.6) 0.289 (0.8)
Ariane Pailhe´
Bratislava
OLS
Model 2
{Journals}Labr/17_1/i189/makeup/I189.3d
140
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Table 5. Estimations of a two-regime model in the set of Slovak employees
Administration and police Transport
Standard deviation of error
0.32 (60.8)
Log-likelihood
L ¼ 489
pp ¼ 0:29 (30.1) pw ¼ 0:04 (1.1)
ss ¼ 0:20 (36.6) sw ¼ 0:25 (64.2) L ¼ 379
* Significant at the 10 percent level; * * significant at the 5 percent level. t-statistics in parentheses. N ¼ 1;848. Source: Social Stratification Survey in Eastern Europe, author’s computation. Notes:
1
pp ¼ 0:55 (17.0) pw ¼ 0:01 (0.1)
ss ¼ 0:23 (37.2) sw ¼ 0:47 (51.6) L ¼ 88
141
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Manager (10 people and more) Manager (1– 9 people)
{Journals}Labr/17_1/i189/makeup/I189.3d
Construction
1.488 * * (5.5) 0.237 * * (2.3) 0.355 * (1.6) 0.323 * * (3.3) 0.424 (1.5) 0.117 (1.0) 1
Labour Market Segmentation in Central Europe
1.270 * (1.6) 0.325 * (1.9) 0.774 * * (3.2) 0.091 (0.5)
Finance
{Journals}Labr/17_1/i189/makeup/I189.3d
Ariane Pailhe´
142
The return on education and total experience is higher in the primary segment, whatever the specification retained.8 Thus, general qualifications significantly increase wages in the first segment, in far greater proportions than in the second, particularly in Poland. Whereas the variable coefficient measuring seniority is significantly positive for the secondary segment, it is never significant for the primary segment, whichever country is under consideration; it may even take on a negative sign. These results illustrate the specific characteristics of labour market partitions in economies undergoing systemic transition; it appears that employment positions in the primary segment belong either to newly created firms or to establishments giving priority to new labour market entrants during the hiring process, except in the Czech Republic. Consequently, seniority has no significant effect on wages. On the other hand, specific qualifications have an important impact on wages in the secondary segment. The primary segment is then able to attract the most qualified workers, whereas the secondary segment conserves the least mobile workers and values seniority. By measuring discrimination against women by the coefficient of the ‘woman’ variable, it appears that discrimination penetrates both market segments. But, above all, gender discrimination manifests itself by differing allocations to the two sectors, depending on sex. 3.3 Factors contributing to the adherence to the primary segment The probability that a worker i works in the primary segment, given his wage and his personal characteristic, is: Prprim ¼ Pr("wi > Zi =Zi ; Xi ; "pi ) f("pi ) Pr("wi > Zi =Zi ; Xi ; "pi ) f("pi ) þ Pr("wi ˘ Zi =Zi ; Xi ; "si ) f("si )
:
[6] This ex ante probability is calculated for each individual (Dickens, Lang, 1992). The probability of a certain category of workers belonging to the primary segment is estimated by calculating the average probability of working in the primary segment for all workers holding the same characteristic (see Table 6). # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
143
The probability of belonging to the primary segment is relatively low since it does not exceed 25 percent for any of the countries studied; the probability is higher for men. Likewise, living in the region surrounding the country’s capital improves the chances of belonging to the primary segment. People holding a diploma of higher education always have a higher probability of working in this segment, as do those with technical training, except in Poland. Experience acquired before entering the establishment increases the probability of belonging to the primary segment, particularly in Hungary. Workers with a greater number of general qualifications have more chance, therefore, of working in the primary segment. On the other hand, the likelihood of belonging to the primary segment varies little based on the number of years of tenure, and is much higher, whichever country is considered, for newly arrived workers in the establishment. This result seems to go against the tenets of the theory. But this contradiction is only apparent, not real, reflecting the impact of systemic transition; the most qualified workers are those who have easily found work opportunities in alternative, better paid jobs, especially in the new private sector. The growth in labour mobility is typical of the period of change, during which employment stability is not a characteristic of the primary segment. The type of ownership under which the establishment falls plays an important role in the chances of belonging to the primary segment. Thus, workers in private firms are favoured. The probability is particularly high for workers in foreign firms, especially in Poland and Slovakia. While in Poland and Hungary workers from recently privatized firms have more chance of belonging to the primary segment than those in newly created private firms, the opposite is observed in the Czech Republic and Slovakia. These results illustrate the role of the type of privatization chosen as well as the restructuring undertaken (Carlin et al., 1994). Nevertheless, the probability of belonging to the primary segment varies greatly depending on the worker’s sector of activity. Those areas that bring together the most highly structured internal markets are both the former priority branches (such as mines and energy) and some service-sector activities enjoying a period of rapid growth throughout systemic change (such as finance).9 Workers from the agricultural sector, education, health and social services still have little chance of reaching this level. # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Hungary
Slovakia
Poland
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
25 29 20 39
22 26 18 35
19 21 16 27
24 25 23 42
26 30 20 85
18 21 13 32
27 32 20 37
25 30 19 34
26 26 22 23 23
24 24 21 21 21
22 20 16 16 17
27 26 22 21 22
32 31 21 25 24
21 20 15 16 17
30 22 29 22 27
29 25 27 20 21
22 23 27
20 20 25
17 19 19
20 24 27
25 25 29
16 18 19
27 26 28
24 23 26
29 21 27 27 24 23
31 20 23 22 21 20
24 20 20 16 15 17
31 31 29 19 22 19
30 28 28 28 26 23
20 20 18 16 17 17
28 24 26 24 23 31
25 22 24 23 21 29
47
51
22
47
36
34
45
48
Ariane Pailhe´
Total Men Women Capital Tenure Less than 1 year From 1 to less than 5 years From 5 to less than 10 years From 10 to less than 15 years 15 years and more Experience at entry in the establishment Less than 5 years From 5 to less than 10 years 10 years and more Education level Tertiary education Secondary general school Secondary technical school Incomplete secondary school Primary Apprentice training Status Manager (10 people and more)
Czech Republic
{Journals}Labr/17_1/i189/makeup/I189.3d
144
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Table 6. Probability of belonging to the primary segment for different categories of workers (percent)
33 19
22 18
37 20
33 24
21 15
31 24
32 21
21 44 38 32
21 39 29 27
16 33 24 28
22 49 24 38
25 56 33 43
17 39 21 26
23 67 43 40
22 63 39 35
35 26 26 11 59 63 15 27 23 26
25 24 25 13 45 47 18 23 25 24
23 21 23 14 44 27 13 14 20 21
38 18 26 17 58 34 13 18 42 26
38 23 22 19 93 49 17 23 60 25
25 19 14 13 61 29 15 15 30 17
19 20 28 10 63 71 21 27 25 23
18 20 25 09 57 70 19 20 27 21
40
48
26
41
35
22
39
38
19
17
18
24
27
18
25
22
21 25 24
19 20 22
20 17 16
26 21 17
25 25 24
16 17 16
22 25 29
19 23 28
{Journals}Labr/17_1/i189/makeup/I189.3d
34 22
Labour Market Segmentation in Central Europe 145
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
Manager (1 –9 people) Subordinate Type of ownership State-run establishment Foreign establishment Privatized establishment Private establishment Branch Construction Transports and communications Retail trade, restaurants Education, health, social services Finance Mines and energy Agriculture Industry Civil service, police Other Occupational class categories Higher-grade professionals, administrators and officials Lower-grade professionals, administrators and officials Non-manual employee Skilled manual worker Semi-skilled and unskilled manual worker
{Journals}Labr/17_1/i189/makeup/I189.3d
146
Ariane Pailhe´
The chances of working in the top layer of the labour market also vary with the type of post occupied and managerial position. It is quite usual for workers who direct other workers. Like in Russia, the chances of this occurring are also relatively high for manual workers in Hungary, and even more so in Poland, despite the fact that they are unskilled; indeed, unskilled workers enjoy ‘insider’ power (the influence of workers’ councils within firms is very apparent in Poland). It seems, therefore, that whichever specification is retained, the chances of working in the primary segment depend on the managerial position attained, the firm’s ownership characteristics and the sector of activity. On the other hand, due to systemic changes, professional experience and seniority play an opposite role to that advanced by the theory, with the more mobile workers, like in Russia, holding the best jobs. 4. Mechanisms leading to segmentation Labour administration practices were modified profoundly during the first years of systemic change, particularly due to the hardening of budget constraints, the increase in unemployment and the growth of a small but dynamic private sector. Motives for forming internal markets have also changed over this period. The motives for constituting an internal market vary from firm to firm, depending, in particular, on economic performance, level of uncertainty and type of ownership. Workers, especially the less qualified and the more risk averse, are now confronted with uncertainty and have more reason to want to belong to an internal market. We have shown that the primary segment covers both the firms belonging to the socialist period’s primary segment and firms undergoing expansion in certain service sectors. It thus seems that old habits have helped to maintain highly structured internal markets in sectors which today are in decline. Uncertainty has been so all-encompassing during the transition period that agents have protected themselves by retaining old habits. Institutional and legislative changes, structural evolution and economic fluctuations have perturbed the internal coherence achieved during the socialist period. Firms’ provisioning circuits remain uncertain due to the break in relations with former CMEA (Council for Mutual Economic Assistance) suppliers, companies’ financial difficulties # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
147
and the risk of seeing partners going into bankruptcy. Bumpy production rhythms have not totally disappeared and a certain amount of labour hoarding still seems necessary.10 Furthermore, managers in those sectors in decline try to hoard qualified labour by stopping workers from looking for employment elsewhere. Qualified workers ensure high production quality and therefore the firm’s competitiveness. These same workers are those most likely to be capable of adapting to new production conditions. The need for teams of polyvalent maintenance technicians is still apparent in many establishments due to the slow pace of dismantling the old production lines.11 But new firms and=or firms enjoying expansion, particularly companies with foreign participation, are also those with the greatest means and the best reasons for forming highly structured internal markets.12 Wage rises and promotion prospects offer a means of attracting the most qualified workers,13 as well as increasing worker productivity and discipline, especially in an environment where there is an abundance of work on offer (Dabrowski et al., 1993). Likewise, certain foreign firms aim to increase workers’ specific qualifications.14 Privatized companies engaged in an offensive restructuring process in Hungary and in Poland (more so than in the former Czechoslovakia), tend to favour a highly structured internal market. A survey by Roberts et al. (1997) reveals that in Poland private entrepreneurs have begun to use old methods to obtain trained workers by subsidizing schools and offering work experience to the students. On the other hand, state-run firms, held back by national policy on wage controls during the first years of transition, have had trouble maintaining their former highly structured internal markets. Strict rules controlling wage increases left them with little scope for offering higher wages to attract or retain qualified labour. This wage regulation was successful in all the countries under consideration, with the exception of Poland, where strong labour unions, friendly with management, pushed for standards to be changed. For this reason, an internal coalition between directors and workers persists in state-run and newly privatized firms. Management has every reason to engage in contractual relationships with workers in order to maintain its position (as the representatives of the personnel constitute one-third of administrative councils). The common goal is the firm’s survival, particularly when its existence is threatened. The continued # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
148
Ariane Pailhe´
presence of old managers contributes to behavioural inertia, all the more so because these managers are risk averse, and, therefore, maintain the internal markets. Thus there are several dichotomies in the labour market during this transition period, depending on the sector of activity, the size and the type of ownership. The primary segment does not, therefore, appear homogeneous, but is itself subdivided into several layers. 5. Conclusion This study makes it clear that allocation of labour has differed from one segment to another in the Central European labour markets during the first years of transition. The first segment concentrates a greater proportion of men, people residing in the capital, occupying a management position and working in private establishments, as well as in the mining, energy and finance sectors. Education levels, on the other hand, have not played a significant role in the allocation of workers during the transition period. This result signals both the obsolescence of former training circuits and the fact that the productive characteristics on entering the labour market do not determine the occupational distribution of workers. Wage structures in the two segments are significantly different. In particular, the return on general qualifications is higher in the primary segment whereas that on specific qualifications is higher in the secondary segment. This last element reflects systemic transition: establishments in the first segment are those which attract fresh manpower, newly arrived in the labour market, as well as the most qualified workers who have left their former employment in order to enjoy higher wages. For these reasons, ‘pure’ segment characteristics, such as defined by Doeringer and Piore, are not to be found. Labour market structure is clearly marked by structural characteristics inherent in each of the transition economies. Labour market segmentation results from the growth of market uncertainty, a combination of new formal institutions and the persistence of old, informal institutions. These results confirm the importance of labour market structure in wage determination and reinforce the idea that remuneration is not determined solely by competitive mechanisms. Coordination is the product of the firm, too. The sector of activity catalyses the set of internal rules used by # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
149
the firms it encompasses by promoting collective knowledge, common values, specific professional relations and thus encouraging a certain behavioural stability. This explains why labour market segmentation is based, in large part, on branches of activity.
Notes 1 Nagy and Sziraczki (1984); Cukor and Kertesi (1985); Kertesi and Sziraczki (1988). 2 The internal market is different from a competitive labour market; it can be defined as ‘an administrative unit, such as a manufacturing plant, within which the pricing and allocation of labour is governed by a set of administrative rules and procedures’ (Doeringer, Piore, 1971, p. 1). 3 Flexibility was even institutionalized in Hungary in 1982 by the legal instigation of the VGMK — teams of autonomous workers employed through sub-contracts. These teams of workers signed contracts with their firms to carry out specific tasks, such as installing new equipment, doing non-routine maintenance and carrying out specialized work. They were allowed to use the firms’ equipment and supplies to do this work. Their creation was intended to allocate work in a more efficient manner, increase productivity and workmanship while diverting workers from parallel activities, often carried out in the workplace. The remuneration of VGMK members was not included in the pay bracket, but negotiated with the directors; earnings were roughly four times higher than the average wage in the industry (Stark, 1986). This type of organization was very successful; in 1982 there were 3,531 associations, 21,490 in 1986, falling to 16,978 in 1988. A similar sort of workers’ association was introduced in Poland in 1984, but it did not encounter the same success. 4 The estimates are biased by selectivity because individuals choose the sector in which they will work based on the different wages offered. 5 For instance, whereas some surveys failed to produce evidence in favour of labour market segmentation, other surveys using the same sample succeeded: see Psacharopoulos (1978); McNabb and Psacharopoulos (1981). 6 Magnac (1991) also developed a labour supply model in a four-sector labour market. 7 We have to impose the usual identifiability conditions in this model, namely that there is at least one variable in the equations containing X that is not included in the equation containing Z, and vice versa (Maddala, 1983). 8 Nevertheless, in the Czech Republic, the experience coefficient is never significant for the primary segment, whereas it is significantly positive for the secondary segment for the first specification. 9 The example of the retail trade and restaurants is more paradoxical. It is difficult to believe that these branches form highly structured internal markets, being by nature highly competitive and made up mainly of small establishments. This incompatibility underlines the limits of our analysis, which does not focus on wages and does not take into account career profiles or work conditions. 10 A survey of 25 firms in the Polish electromechanical sector carried out in 1995 # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
150
Ariane Pailhe´
(Richet, 1997) reveals that 40 percent of them hoarded labour. One of the main reasons given, other than social considerations, employee resistance and other legal obstacles, was the fear of not finding qualified personnel if production picked up again. A survey of 193 firms in the Czech Republic in 1993 showed that 12.4 percent of them had trouble recruiting qualified labour, the percentage being even higher for state-run companies (Koubek, 1994). 11 In the electromechanical sector in Poland, for example, the majority of active workers were first hired back in the 1970s. 12 However, this type of strategy is not systematic and differs from sector to sector. Thus, IBM, based in the Szekesfehervar region of Hungary, practises a policy of hiring inexpensive labour with a high turnover. Qualified workers represent only 20 percent of manpower, which is, in general, overqualified. Lastly, whereas Hungarian law requires the formation of workers’ councils in all firms employing more than 50 workers, one has not been formed at IBM (Mako´ et al., 1997). 13 There is, in particular, a shortage of competencies now essential in the fields of marketing, accounting, finance, legal counsel, and of advanced technicians. It is not so much their specific character as their rarity that makes company directors try to stabilize these workers. 14 The Swiss – Swedish firm ABB is a good illustration of this tendency. It bought the Polish turbine factory Zamech in 1990 and in 1994 founded training centres in Brno and Warsaw, training 17,000 workers in the Czech Republic and Poland.
References Akerlof G. (1982) ‘Labor Contracts as Partial Gift Exchange’, Quarterly Journal of Economics XCII(4): 543– 569. Akerlof G. and Yellen J. (eds.) (1986) Efficiency Wage Models of the Labor Market, Cambridge: Cambridge University Press. Azariadis C. (1975) ‘Implicit Contracts and Underemployment Equilibria’, Journal of Political Economy 83(6): 1183– 1202. Beck E. M., Horan P. M. and Tolbert C. M., II (1978) ‘Stratification in a Dual Economy: A Sectoral Model of Earnings Determination’, American Sociological Review 43: 704– 720. Blanchard O. (1997) The Economics of Post-communist Transition, Oxford: Clarendon Press. Bluestone B. and Harrison B. (1982) The Deindustrialization of America: Plant Closings, Community Abandonment, and Dismantling of Basic Industry, New York: Basic Books. Boeri T. (1997) ‘Heterogeneous Workers, Economic Transformation and the Stagnancy of Transitional Unemployment’, European Economic Review 41: 905– 914. Boeri T. and Flinn C. (1999) ‘Returns to Mobility in the Transition to a Market Economy’, Journal of Comparative Economics 27: 4– 32. Cain G. G. (1976) ‘The Challenge of Segmented Labor Market Theories to Orthodox Theory: A Survey’, Journal of Economic Literature 14(4): 1215– 1257. # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
Labour Market Segmentation in Central Europe
151
Carlin W., Van Reenen J. and Wolfet T. (1994) ‘Enterprise Restructuring in the Transition: An Analytical Survey of the Case Study Evidence from Central and Eastern Europe’, European Bank for Reconstruction and Development Working Paper No. 14, July. Cukor E. and Kertesi G. (1985) ‘Differences in Pay and Modes of Earning’ in Galasi P. S. and Sziraczki G. (eds.) Labour Market and Second Economy in Hungary, Frankfurt: Campus: 70 – 120. Dabrowski J., Federowicz M. and Levitas A. (1993) ‘Privatisation of Polish State Owned Enterprises: Progress, Barriers, Initials Effects’, Economic Transformation Papers No. 33, Gdansk Institute for Market Economics. Dickens W. and Lang K. (1985) ‘A Test of Dual Labor Market Theory’, American Economic Review 64(4): 792– 805. Dickens W. and Lang K. (1988) ‘Labor Market Segmentation and the Union Wage Premium’, Review of Economics and Statistics 70(3): 527– 530. Dickens W. and Lang K. (1992) ‘Labor Market Segmentation Theory: Reconsidering the Evidence’, NBER Working Paper Series No. 4087. Doeringer P. B. and Piore M. (1971) International Labour Market and Manpower Analysis, Lexington, MA: D. C. Heath. Dunlop J. (1958) Industrial Relations Systems, Carbondale, IL: Southern Illinois University Press. Edwards R. C., Reich M. and Gordon D. M. (1975) Labor Market Segmentation, Lexington, MA: D. C. Heath. Grosfeld I., Senik-Leygonie C., Verdier T., Kolenikov S. and Paltseva E. (2001) ‘Workers’ Heterogeneity and Risk Aversion: A Segmentation Model of the Russian Labour Market’, Journal of Comparative Economics 29(2): 230– 256. Heckman J. J. and Hotz J. V. (1986) ‘An Investigation of the Labor Market Earnings of Panamanian Males: Evaluating the Sources of Inequality’, Journal of Human Resources 21: 507–542. Kerr C. (1954) ‘The Balkanisation of Labour Market’ in Bakke E. W. (ed.) Labour Mobility and Economic Opportunity, Cambridge, MA: MIT Press. Ko¨llo¨ J. and Kertesi G. (1998) ‘From High Unemployment to Low Wages — Which Way in the Future?’, Budapest Working Papers on the Labour Market 1999=5, IE-Budapest University. Koubek J. (1994) ‘Human Resource Management in the Czech Republic as Reflected by the PriceWaterhouse Cranfield Project 1993’, Working Paper, Prague University of Economics. Leontaridi M. R. (1998) ‘Segmented Labour Market: Theory and Evidence’, Journal of Economic Survey 12: 63 – 101. Lindbeck A. and Snower D. (1988) The Insider – Outsider Theory of Employment, Cambridge, MA: MIT Press. Maddala G. S. (1983) Limited-dependent and Qualitative Variables in Econometrics, Cambridge: Cambridge University Press. Maddala G. S. and Nelson F. D. (1975) ‘Switching Regression Models with Exogenous and Endogenous Switching’ in Proceedings of the American Statistical Association, Business and Economic Statistics Section: 423 – 426. Magnac Th. (1991) ‘Segmented or Competitive Labour Markets?’, Econometrica 59: 165–187. Mako´ Cs., Ellingstad M. and Kuczi T. (1997) Regional Innovation System # CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
{Journals}Labr/17_1/i189/makeup/I189.3d
152
Ariane Pailhe´
(REGIS): Sze´kesfehe´rva´r Region Survey Results and Interpretation, Budapest: Institute for Social Conflict Research — HAS, January. Malle S. (1986) ‘Heterogeneity of the Soviet Labour Market as a Limit to a More Efficient Utilisation of Manpower’ in Lane S. (ed.) Labour Market and Employment in the USSR, Brighton: Harvester Press. McNabb R. and Psacharopoulos G. (1981) ‘Further Evidence on the Relevance of the Dual Labour Market Theory in the UK’, Journal of Human Resources 16(3): 442– 448. Milanovic B. (1998) ‘Explaining the Growth in Inequality during the Transition’, Working Paper Series No. 1935, World Bank. Mu¨nich D., Svejnar J. and Terrell K. (1999) ‘Returns to Human Capital under the Communist Wage Grid and during the Transition to a Market Economy’, CEPR Discussion Paper No. 2332, December. Nagy A. and Sziraczki G. (1984) ‘Labour Market Segmentation in Eastern Europe: The Case of Hungary’, Papers on Labour Economics No. 2, Department of Labour and Education Economics, Karl Marx University of Economics, Budapest. Perrot A. and Zylberberg A. (1989) ‘Salaire d’efficience et dualisme du marche´ du travail’, Revue e´conomique No. 1: 5 – 20. Piore M. J. (1983) ‘Labor Market Segmentation: To What Paradigm Does it Belong?’, American Economic Review 73(2): 249– 254. Psacharopoulos G. (1978) ‘Labour Market Duality and Income Distribution: The Case of the UK’ in Krelle W. and Shorrocks A. F. (eds.) Personal Income Distribution, Amsterdam: North-Holland. Richet X. (Coord.) (1997) Etude sur les entreprises de certains segments industriels en Europe de l’Est, Contrat entre la DGSI-SAEI du Ministe`re de l’Industrie et le CNRS, March. Roberts K., Kurzynowski A., Szumlicz T. and Jung B. (1997) ‘Employers’ Workforce Formation Practices, Young People’s Employment Opportunities and Labour Market Behaviour in Post Communist Poland’, Communist Economies & Economic Transformation 9(1): 87 – 98. Rutkowski J. (1997) ‘Low Wage Employment in Transitional Economies of Central and Eastern Europe’, Moct-Most No. 7: 105– 130. Shapiro C. and Stiglitz J. (1984) ‘Equilibrium Unemployment as a Worker Discipline Device’, American Economic Review 74(3): 433– 444. Sloane P. J., Murphy P. D., Theodossiou I. and White M. (1993) ‘Labour Market Segmentation: A Local Labour Market Analysis Using Alternative Approaches’, Applied Economics 25: 569– 581. Stark D. (1986) ‘Rethinking Internal Labor Markets: New Insights from a Comparative Perspective’, American Sociological Review 51(4): 492– 504.
# CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd 2003.
