Professional Documents
Culture Documents
Kertesi Köllő Economic Transformation and The Revaluation of Human Capital Research in Labor Econ Vol21 2002 PDF
Kertesi Köllő Economic Transformation and The Revaluation of Human Capital Research in Labor Econ Vol21 2002 PDF
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
ECONOMIC TRANSFORMATION
AND THE REVALUATION OF HUMAN
CAPITAL HUNGARY, 19861999
Gbor Kertesi and Jnos Kll
ABSTRACT
The paper analyses skills obsolescence during transition to the market
economy, using individual wage data from Hungary, 19861999. The link
between workers age composition and firms productivity is also examined
using firm-level information. Transition started with the collapse of demand
for unskilled labor and the concomitant improvement in the relative position
of skilled workers. At later stages of the transition, when technological
change gained impetus, general appreciation of skilled labor stopped. Since
1992 the market value of skills acquired under communism has been falling.
Consistent with the wage data, the productivity estimates suggest the
devaluation of skills acquired under communism.
1. INTRODUCTION
Basic measures like enrolment in education or completed school years
suggested, some ten years ago, that Central and Eastern Europes transition to
the market economy would be promoted by a valuable and transferable stock
of human capital. Optimistic and proud references to a highly skilled labor
force were made in government manifestos and the country reports of
235
236
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
237
238
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Section 6 draws attention to the limits of the OLS model, presents instrumental variable estimates for panels of firms, and discusses alternative
explanations of how skilled-young labor and high productivity can be correlated.
Section 7 concludes.
Table 1.
Specification
Key variables
Controls
1. Benchmark
Mincerian
gender dummy,
educational grade dummies,
exp, exp2, exp3, exp4
26 interactive dummies of
education, experience and
gender
gender dummy,
unskilled, skilled-youngb,
skilled-oldb
occupational grade
productivity,
capital/labor ratio,
firm size,
local unemployment
rate,
industry dummies,
region dummies,
constant
2. Augmented
interactive
3. Simplified
interactivea
a
b
Detailed information
Appendix 2
Appendix 3
Appendix 4
239
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
240
241
with a university diploma and 1525 years of experience was lower by about
20 percentage points in 1999 than 1986.
Figure 2b draws attention to the relevance of an interactive model, which
allows for the fact that different educational and experience groups were exposed
to skills obsolescence of greatly varying degree. The interaction effects are
tested with an equation using 25 interactive dummies5 of education, experience
and gender, treating unskilled labor (08 completed classes irrespective of
experience and gender) as the reference category. In order to be able to control
the gender division of the occupational structure, and the profound gender
differentials in the extent of specialized knowledge in the post-primary
education6, Fig. 3 and its underlying equation reports the coefficients for males
and females separately. Regression parameters of the year 1986 are chosen as
uniform (zero value) starting points. Changes in wage returns relative to the
1986 values are plotted in the graphs. Panels ab repeat the results of the
benchmark Mincerian model by gender.
(i) The benchmark regression models (panels ab) hint at the stabilization of
returns to education following 1992/1993, the worst years of the transformation crisis. By contrast, the more precise interactive model (panels ch)
reveals profound differences between young and old cohorts. The experience-related gap of the pre-transitional years (19861989) was in large part
unaffected by 1992/19937, and started to narrow rapidly after 1992/1993,
and even more markedly after 1995 when the economy started to recover
from the post-1989 crisis.
(ii) Workers with college or university backgrounds, both male and female,
improved their position across all experience groups but the value
of education increased at substantially higher rates in the young
cohorts. As a result, the experience-related wage gap between the oldest
and the youngest college cohorts decreased by 2025% by 1999 (panels
gh).
(iii) The returns to secondary school increased in, and only in, the younger
cohorts of men and (particularly) women. The youngest female cohort
managed to improve its position by almost 20%, followed by workers with
610 years of experience with a 15% increase, and older females whose
market skills kept their modest value with no further appreciation. This
kind of imbalance can be a sign of change in the patterns of demand for
non-manual female employees resulting from the expansion of the tertiary
sector.
(iv) Workers who completed vocational training school did not get ahead in
general (neither the females nor the males improved their position relative
241
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
242
GBOR KERTESI AND JNOS KLL
Fig. 3. Changes in Returns Relative to the 1986 returns (Augmented Interactive Model).
i = iti86, it -s are regression parameters of the model (see Appendix 3).
Fig. 3. Continued.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
243
243
244
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
245
246
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
for top managers payment in kind was restricted to meal vouchers, holidays in
company resort houses, and similar benefits of marginal importance.
Selectivity Bias.
Comparisons of wages before and after the transition can be biased by change
in the strength of selectivity bias. Unfortunately we lack data for a proper test
but Appendix 7 presents an admittedly second-best method focusing on some
observable implications of selection bias. Since the start of transition the composition of wage earners has been changing in favor of workers with lower risk
of unemployment and, predictably, higher wages. The data suggest that unskilled
and elderly labor was more strongly affected by this kind of selection than other
groups. Wage models with sample selection (Heckman, 1979) would probably
detect a steeper rise in the wage advantage of skilled and young workers than
ordinary regression.
Ability Bias.
We should try to quantify, in one way or another, the potential changes in
returns to skill components which are not directly observable, that is, not
captured by simple variables like schooling or experience. A possible measurement strategy is to study wage residuals controlled for observed skill
characteristics and different sort of rents that affect the wage distribution. In a
model where these fundamental wage determinants are controlled for the
residual wage dispersion mostly reflects unobservable skill differentials across
individuals. The question is whether these residuals are stable over time. In
lack of longitudinal data there is no way to observe changes in the residuals
directly. An indirect way of measurement proposed by Juhn, Murphy and Pierce
(1993) takes the residual wage percentiles as the units of observation and
measures wages in real terms to ensure comparability over time. Mean residual
real wages are then compared over time, by percentile. This test was completed
in a former paper (Kertesi & Kll, 1997) for the periods 19861989,
19891992, and 19921995. Residual wages were remarkably stable over time,
all over the residual wage distribution12. Stability in the residual wage
distribution encourages us to rely on the measurable proxies of education and
experience in this paper.
Supply-side effects
Absolute outflows from secondary and higher education (Fig. 4) rose substantially between 1986 and 1999. Annual graduations from secondary schools grew
by more than 1/3 and outflows from colleges and universities nearly doubled.13
The concomitant increase of relative wages and the supply of skilled-young
Fig. 4.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
247
247
248
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
labor leads us to believe that the revaluation of human capital was predominantly
demand-driven.
Fig. 5.
249
As Table 2 shows the early years were also characterized with large scale job
destruction. Workers with different skills were unevenly affected but the
available figures suggest net job destruction even in the skilled part of the market.
As much as 48% of unskilled jobs disappeared between 19901995 but the
market for skilled LABOR also had to face a 11% contraction in that period.14
The years of general decline were followed by an era of divergent evolutions in 19951999. Skilled jobs were created at a similar rate to those that had
been lost during the transition15, and real wages in the upper range of the wage
distribution started to rise. This was not the case with unskilled jobs. The number
of unskilled jobs stagnated after 1995 and real wages decreased further in the
lower ranges of the wage distribution.
With the passing of the transformational recession, which brought about the
collapse of demand for unskilled labor, substantial changes took place in the
evaluation of human capital. As shown in the previous section, the skills
premium of older workers failed to increase further while the appreciation of
new skills gained an impetus that continued until recently. In Sections 5 and 6
we present evidence suggesting that the widening gap between the value of old
and new skills was consistent with their relative productivity. These differences
gathered importance when the market for skilled labor started to grow and new
technologies appeared in the economy.
Female
Year
unskilled
skilled
1990
1992
1995
1999
1,803
1,358
1,225
1,228
845
860
824
875
unskilled
1,380
929
759
702
skilled
unskilled
skilled
1,055
936
869
1,006
3,190
2,287
1,984
1,930
1,900
1,864
1,693
1,881
249
Total
250
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(1)
y being firm level productivity (value added per worker), li the share of the
different types of skilled labor within the firm (skilled-young (l2) and skilledold (l3), the base category being the share of unskilled labor: l1)16, k stands for
the capitallabor ratio approximated with the net value of fixed assets per
worker. Parameters of particular interest are the productivity elasticities with
respect to l2 and l3, that is, the differences between skill groups defined on the
basis of educational background and experience. The estimation results and
descriptive statistics are presented in Appendix 6.
In order to see how the parameters change over time we start with crosssection ordinary least squares (OLS) estimates of equation [1]. OLS can produce
biased and/or inconsistent estimates if firm-specific residuals are correlated with
the explanatory variables, or, if the models variables are endogenously
determined. These problems will be addressed in Section 6 where the OLS
results will be compared with panel estimates using instrumental variables (IV)
and alternative explanations will be briefly explored.
Errors in measuring capital may also bias the coefficients. There is little doubt
that the value of assets was imperfectly measured under socialism when no
capital market existed. The coefficients of the capital/labor ratio are most
probably downwards-biased in 1986 and 1989 due to pure measurement error.
(Other coefficients would be affected only if measurement errors were systematically related to productivity or skills composition). By the end of the transition
period the capital market was in place. More than 80% of the large firms in
the sample were private in 1999, more than 50% incorporated. Many of them
went public and those which did not are also interested in presenting meaningful
data on their assets for creditors and investors. Despite problems with the capital
term its inclusion in all equations can be justified in several ways. First,
including ki to some equations and dropping it from others would make our
estimates non-comparable over time. Second, the bias from dropping the capital
term should be predictably larger than the one from measurement error. Third,
a comparison of OLS and IV estimates (Section 6) can signal whether severe
measurement problems exist.
The equations are estimated for a restricted sample of medium-sized and
large firms. Restriction is required because we shift from individual to firmlevel observations. Information on the internal composition of a firms labor
force is available on the basis of a ten% random sample of its employees, so
we have to restrict our attention to enterprises where the number of workers in
the Wage Survey is sufficiently large.17 The sample thus excludes small and
medium size firms but covers the vast majority of firms employing 300 or more
251
workers. The response rate of the Wage Survey has been deteriorating since
1992 but this size category was largely unaffected. A comparison with firm
counts published in the statistical yearbooks suggested a response rate of 75.8%
in 1992 and 86.3% in 1998.
Figure 6 shows the time paths of the coefficients estimated for young and
old skilled labor. The productivity yield attributed by the model to skilled-young
labor input was rapidly growing in 19861999 while the productivity of skilledold labor input was declining in 19921999 to a point that in the latter year it
did not differ significantly from the productivity yield of unskilled labor (this
was chosen as the base category).
Before taking these first results as proof, let us examine the productivity
yields in more detail. If the appreciation of new skills is explained by the efficient matching of new technologies and young workers, one would expect
younger employees to be more productive and better paid in firms applying
advanced equipment and new work standards. Ideally, one would like to study
the experience-related differences in productivity and wages by comparing a
modern and an obsolete sector distinguished using firm-level information.
As a second best solution, since no enterprise-level information is available on
251
252
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Value added/workera
Year
domestic
1992
1993
1994
1995
1996
1997
1998
1999
528
769
1,085
1,555
1,855
2,804
2,602
3,368
foreign F-ratio
1,008
1,540
2,065
3,118
4,942
6,022
6,445
7,529
33.1
63.5
26.8
27.2
12.1
25.2
27.3
28.6
foreign
F-ratio
1,202
1,889
2,627
3,150
3,250
4,815
4,033
6,308c
1,151
2,524
2,769
3,955
5,820
7,804
7,766
7,253
0.1n
2.3n
0.1n
0.9n
6.9
7.4
7.9
0.3n
Depreciation rateb
domestic foreign F-ratio
0.113
0.093
0.085
0.096
0.109
0.115
0.134
0.151
0.132
0.109
0.119
0.131
0.141
0.149
0.153
0.181
1.2n
2.3n
20.4
16.7
11.0
19.5
3.8
6.4
Million forints.
Firms reporting a rate higher than one excluded. The number of excluded firms: 1 in 1994, 2 in
1992, 3 in 1996 and 1999.
c
Three domestic firms reported extremely high capital/labor ratios (28 times the average) in this
year. If these firms are omitted the domestic mean becomes 4,410 and F = 10.8.
n
Insignificant at the 0.05 level. The F-ratios test the equality of the means by one-way analysis
of variance.
b
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
253
253
254
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Fig. 9. The Distribtuion of Skilled Labor by Experience in Foreign and Domestc Firms (1992, 1999).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
255
255
256
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
6. ISSUES OF SPECIFICATION
AND INTERPRETATION
The results presented earlier were based on cross-section OLS models which
rely on restrictive assumptions, do not fully utilize the information available in
the database, and relate to an ever-changing population of firms. In this section
we estimate IV panel models, which allow for correlation between residual
productivity and the firms skill composition on the one hand, and endogeneity
on the other. To start with, we write equation (1) in a more general form:
y = + Xit +
i + it
(2)
257
Dependent:
log(value added/worker)
(A) Firm-level variablesa
logk (capital/labor)
logl2 (skilled-young)
logl3 (skilled-old)
constant
aR2
Cross-sections
1996
1997
Panel 19961999
1998
1999
fixed
randomb
0.2219*
0.1893*
0.0296
0.1600
0.3741d
0.2657*
0.3088*
0.0651
0.6230
0.3261
0.3205*
0.3521*
0.0590
0.7890
0.3902
0.2623*
0.4674*
0.0736
1.100
0.4020
0.2664*
0.4714*
0.0283
0.9610
0.4031
0.1342*
0.0477
0.0477
0.5181
0.0296c
0.3836*
0.3724*
0.1181
0.0390
0.3460
0.3506*
0.3695*
0.0371
0.0444
0.3560
0.3189*
0.3835*
0.0529
0.0249
0.3431
0.3509* 0.3618*
0.0752 0.2168*
0.0851
0.0092
0.1759 0.1631
0.0847c 0.2656d
Hausman specification tests for the equality of the fixed and random effects estimates: (chi-squared):
(A) 119.7 (0.0000), (B) 59.9 (0.0000)
a
Equations (A) use the firm-level variables, (B) use the log deviation of firm level variables from
variable means in the respective two-digit industries.
b
Generalized least squares.
c
Within R2.
d
Overall R2.
*
Significant at 0.01 level.
257
258
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
logl2
(skilled-young)
logl3
(skilled-old)
Balanced panel
275 firms, 1,100 obs.
0.2746
(5.3)
0.4759
(2.8)
0.1341
Unbalanced panel
433 firms, 1,495 obs.
0.2344
(5.3)
0.4796
(5.1)
0.1694
(1.1)
Dependent:
log (value added
per worker)
a
b
constant
aR2
0.5477
0.3672
1.409
0.3623
(0.5)
259
is a robust fact which, irrespective of how it came into being, implies that older
workers are less likely to learn how to use modern technologies. This will
inevitably erode their skills (further), and establish a causal link between
workers age composition and firms efficiency. We believe that the correlation
observed today is more than a statistical artifact resulting from compositional
effects and unobserved selection processes skills obsolescence already had a
major contribution. Future research will be able to explore the process easily
and more convincingly.
6. CONCLUSION
Like most studies on Central and Eastern Europe we found a general rise in
returns to education between 1989 and 1992, when technological change was
minimal, and the forces of the market had just started to work. This, we believe,
was simply a mirror image of the collapse of demand for unskilled labor. When
market institutions were already at work, and modern technologies started to
flow in, the general appreciation of education stopped but the returns to
experience continued to decline. Technological renewal apparently contributed
to the appreciation of young and educated labor in this period. We found that
these workers are paid increasing wages and presented evidence suggesting that
their skills are actually worth more in a modern environment. By contrast,
neither productivity nor wages grew for the older cohorts of educated workers
after 1992. The obsolescence of skills acquired under state socialism had, and
will continue to have, far-reaching implications for transition countries ones
ranging from reduced growth potential to disappointment with market-based
liberal democracy.
NOTES
1. Their choice of controlling the wage equation for tenure may have affected their
pre-unification estimates because labor turnover was particularly low in the GDR, calling
into question the distinction between general and firm-specific experience.
2. An introduction of data sources and descriptive statistics are presented in
Appendix 1.
3. We use higher than second order experience terms in order to be able to follow
cohort specific changes in the earnings profiles over time.
4. For the sake of consistency the charts and the text always refer to the estimated
coefficients () as presented in Appendices 24. The precise measure of the effect of
education (and other dummy variables) on earnings should be expressed as e which
differs from the coefficient at high values of .
5. The 26 groups are defined by combining gender, three levels of education
(vocational, secondary, higher) and four experience groups (05, 610, 1120, 2130
259
260
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
years). Two additional categories comprise old workers (more than 30 years of experience, without distinguishing by gender and schooling) and unskilled labor (without
distinguishing by gender and experience). The latter category is the group of reference
in the regressions. In the augmented interactive model the same controls are used as in
the benchmark specification.
6. Female workers typically graduate from general secondary schools (37% in 1999)
as opposed to vocational training schools (19%). The opposite holds for male workers
(24 vs. 42%).
7. Exceptions are female college graduates and workers with completed secondary
school. Even in these groups about half of the narrowing of the base years differential
took place after 1992/93.
8. The underlying assumption that the experience-earnings profiles of 1989 properly
approximated the expected life-cycle paths seems justified in view of the stability of
relative wages under socialism.
9. The relatively high gains of this group may be explained by early retirement of
many high-educated males, aged already 4550 at the start of transition. A precise
clarification of the reasons would require a selectivity bias correction model.
10. Authors calculations based on the Labor Force Survey, available on request.
11. Cukor (1989) provides ample evidence of how firms reclassified their workers in
response to labor shortages between 1973 and 1976.
12. The only exception was the 100th percentile where there was an upward jump
in the mean residual real wage between 1989 and 1992. We suspect that the spectacular wage rise of the best-paid employees in this period represented a sort of
privatization gain, which is an exceptional event from the point of view of the present
paper.
13. Outflows from vocational and secondary schools were also affected by demography. The share of secondary schools in total outflows increased sharply but the cohorts
reaching graduation age (1718 years) were smaller and smaller in subsequent years of
the transition period. In evaluating the magnitudes of possible supply side effects the
absolute numbers in graduations are relevant.
14. Consistent time series on employment are available only from 1990 therefore we
cannot have 1989 as a starting date.
15. The net skilled job creation rate was 11% in 19951999.
16. Skilled means completed secondary school or incomplete or completed college
or university, unskilled means incomplete or completed primary or incomplete
secondary school. Young means experience less than the median experience, old
means median experience or more.
17. The critical sample size on the level of a particular firm was set at 30 workers.
By this restriction firms with less than 300 workers are excluded. Those employing
slightly more than 300 are more likely to drop out than larger firms. To reduce the
resulting bias the computations in the restricted sample were weighted. Weights were
defined as the ratio of firms in the original and the restricted samples within groups
formed by firm size and ownership.
18. A firm was defined majority foreign in case foreigners share in equity exceeded
50%.
19. According to a recent survey comprising 264 domestic and 78 foreign-owned
firms the former spent 2.4% of their total investment on training while the latter spent
14.2% in 2000. A difference of similar magnitude was observed in the Czech Republic
261
by Filer et al. (1995) For an introduction of the Hungarian survey see EBRD (2000).
The quoted figures have been calculated from the original data file.
20. These results stem from the simplified interactive regression models presented in
Appendix 4.
21. Capital-abundant modern firms are likely to start businesses in highly profitable
sectors, can afford to pay high wages, and attract job seekers who tend to be younger
and better educated than the average incumbent employee. Firm creation can lead to
spurious correlation between age and productivity in this way.
22. The annual industry means were calculated for 42 industries using data of all
firms observed in the Wage Survey excluding those in the panels. Monetary aggregates
were discounted using producer price indices for three, two and one digit industries.
23. The estimates also suggest that the link between share of skilled-young labor and
productivity is dominated by the between effect, that is, the linkage between intertemporal firm-specific means of the variables.
24. Note that the IV coefficients are different from those of the non-instrumented
panel model. (Compare the estimates for the balanced panel in Tables 4 and 5). The
equality of the key parameters of k, l1, and l2 is rejected at conventional levels of significance.
25. The aforementioned survey of the EBRD suggested lower risk of dismissal in
foreign firms (average annual rate of 3.4% in 19971999 as opposed to 6.9% in domestic
enterprises).
ACKNOWLEDGMENTS
The authors are grateful for comments by Gbor Krsi (Institute of Economics
and CEU, Budapest); an anonymous referee; and participants of the conference
Understanding Skills Obsolescence: Theoretical Innovations and Empirical
Applications organized by the Research Center on Skills, Knowledge and
Organizational Performance (SKOPE), University of Oxford, Department of
Economics and the Research Center for Education and the Labor Market (ROA),
Maastricht University, May 1112, 2001.
REFERENCES
Baltagi, B. H., & Chang, Y. (1994). Incomplete panels: A comparative study of alternative estimators for the unbalanced one-way error component regression model. Journal of
Econometrics, 62, 6789.
Burda, M., & Schmidt, C. M. (1997). Getting behind the East-West German wage differential.
Discussion Paper No. 250. University Heidelberg.
Chase, S. R. (1997). Markets for communist human capital: Returns to education and experience
in the Czech Republic and Slovakia. The Davidson Institute Working Paper Series, No. 109,
Ann Arbor.
Cukor, E. (1989). Wage differentials among blue collar workers, (Kereseti klnbsgek a fizikai
dolgozk krben). Kzgazdasgi Szemle, 36(3), 346362.
EBRD (2000). Transition report 2000 (Chapter 5). EBRD, London.
261
262
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Fazekas, K. (Ed.) (2000). LABOR market report, 2000 (Munkaerpiaci Tkr, 2000), Institute of
Economics, National Employment Fund, Budapest.
Filer, R., Schneider, O., & Svejnar, J. (1995). Wage and non-wage LABOR cost in the Czech
Republic: The impact of fringe benefits. CERGE-EI, Working Paper Series 77, Prague.
Flanagan, R. J. (1995). Wage structures in the transition of the Czech economy. IMF Working
Paper WP/95/36, March.
Franz, W., & Steiner, V. (1999). Wages in the East German transition process Facts and explanations. ZEW Discussion Paper No. 9940, Mannheim.
Juhn, C., Murphy, K. M., & Pierce, B. (1993). Wage inequality and the rise in return to skill.
Journal of Political Economy, 101(3), 410442.
Kertesi, G., & Kll, J. (1997). Real wages and earning inequalities: 19861996 (Relbrek s
kereseti egyenltlensgek). Kzgazdasgi Szemle, 44(78), 612634.
Kertesi, G., & Kll, J. (1999). Economic transformation and the return to human capital. Budapest
Working Papers on the Labor Market, 1999/1996. Institute of Economics, Budapest
University of Economics, Budapest.
Krueger, A. B., & Pischke, J. S. (1995). A comparative analysis of East and West German labor
markets before and after unification. In: R. Freeman & L. Katz (Eds.), Differences and
Changes in Wage Structures (pp. 405445). Chicago: University of Chicago Press.
Munich, D., Svejnar, J., & Terrell, K. (1999). Returns to human capital under the communist wage
grid and during the transition to a market economy. CERGE-IE Discussion Paper No.
199929 (November), CERGE-IE, Prague.
OECD (1993). Education at glance. Paris: OECD.
Puhani, P. (1997). All quiet on the wage front in Poland? ZEW Mannheim, mimeo.
Rutkowski, J. (1996). High skills pay off: The changing wage structure during economic transition in Poland. Economics of Transition, 4(1), 89112.
Sakova, S. (1998). Changes and differences in earnings structures, unpublished thesis, Central
European University, Economics Department, Budapest.
Steiner, V., & Bellmann, L. (1995). The East-German wage structure in the transition to a market
economy. Labor, 9(3), 539560.
Steiner, V., & Wagner, K. (1997). East-West German wage convergence How far have we got?
ZEW Discussion Paper No. 9725, Mannheim.
Vecernik, J. (1995). Changing earnings distribution in the Czech Republic: Survey evidence from
19881994. Economics of Transition, 3(3), 355371.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Variables
1986
263
log(gross earnings/m):
mean
8,74
st. dev.
0,37
Gender:
male
60,07
female
39,93
Education:
primary
48,65
vocational
24,41
secondary
21,31
college
5,63
Experience:
mean
22,22
st. dev.
11,63
log(value added/w):
mean
1,65
st. dev.
0,61
log(net fixed assets/w):
mean
1,36
st. dev.
0,89
Firm size:
1120
2150
0,48
51300
11,23
3011,000
29,35
1,0013,000
27,17
3,001 +
31,77
3,244
116,205
1989
1992
1993
1994
1995
1996
1997
1998
1999
9,18
0,45
9,84
0,51
10,02
0,52
10,27
0,56
10,40
0,55
10,55
0,57
10,75
0,61
10,93
0,63
11,07
0,62
61,60
38,40
60,11
39,89
58,96
41,04
58,48
41,52
59,99
40,01
61,99
38,01
62,92
37,08
60,72
39,28
61,30
38,70
44,45
27,94
22,17
5,43
32,03
33,45
27,07
7,45
31,85
34,70
27,06
6,39
26,08
32,74
31,33
9,85
25,03
33,82
31,33
9,82
26,18
38,91
27,17
7,74
24,96
37,07
29,14
8,83
23,65
37,19
29,48
9,68
22,75
37,99
29,94
9,32
22,68
11,42
22,02
10,58
21,90
10,63
21,84
10,42
21,73
10,60
21,52
10,65
21,28
10,69
21,49
10,69
21,88
10,95
1,13
0,64
0,75
0,82
0,48
0,76
0,18
0,81
0,06
0,80
0,32
0,79
0,48
0,89
0,66
0,92
0,78
0,92
0,99
0,93
0,50
1,27
0,33
1,29
0,39
2,40
0,00
1,40
0,11
1,39
0,32
1,49
0,38
1,63
0,54
1,46
1,85
14,14
28,72
28,04
27,25
6,20
30,00
26,36
16,86
20,58
8,39
36,32
24,90
14,97
15,41
9,93
36,80
25,72
12,38
15,17
7,34
11,25
35,46
19,85
11,12
14,98
8,13
10,99
38,48
17,77
12,13
12,50
9,59
13,51
25,55
19,57
15,35
16,43
7,62
11,90
30,77
21,05
14,50
14,16
9,31
8,93
29,90
22,30
13,40
16,17
4,082
111,293
6,498
86,935
7,243
85,833
7,776
94,639
7,456
90,717
7,796
97,918
7,922
88,208
7,909
102,102
8,541
102,547
263
No. of firms
No. of workers
Table A1.
264
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
size and two-digit industry weights were attached (w2) to each size-industry
cell. Calculations were made using the compound weights (w1 w2).
Highest educational attainment can be primary school (8 years in school),
vocational (11 years, provides no certificate required at entry to higher
education), secondary (12 years), college (16 or 17 years depending on the type
of institution). Experience was approximated on the basis of age and the level
of education (age years in school 6). Unemployment was measured at the
level of 170 labor office districts. The earnings figures comprise all work-related
payments made by the enterprise in May and 1/12 of the premia, bonuses and
rewards paid in the preceding year. All statistics refer to pre-tax earnings.
265
Estimation results.
1986
Constant
8,0504
Gender:
Male
0,2838
Schooling:
Vocational training school
0,1203
Secondary school
0,1359
College
0,3592
Experience:
Linear
0,0522
Quadratic/100
0,2187
Cubic/10000
0,4654
Quartic/1000000
0,4149
Occupation:
Non-manual
0,0957
Managerial
0,5436
Productivity:
log(value added/worker)
0,0565
Negative value added
0,0601
Capital/labor ratio:
log(net fixed assets/worker) 0,0350
Firm size: (No. of workers)
1020
2150
0,0005n
3011,000
0,0312
1,0013,000
0,0502
3,001 +
0,0772
Unemployment:
log(unemployment rate)
Industry:
49 dummies, F-test
108,25
Region:
15 dummies, F-test
131,47
No. of observations
Adjusted R2
F- test
Heteroscedasticity
Omitted variables
Normality of residuals
116,205
0,4588
1,033,11
710,191
148,08
1,579,83
1989
1992
1993
1994
8,4343
8,9185
9,0968
9,1676
0,2995
0,2234
0,2317
0,2378
0,1157
0,1460
0,4410
0,1339
0,2197
0,5597
0,1323
0,2308
0,5971
0,1288
0,2194
0,5981
0,0514
0,0538
0,0482
0,0481
0,2034
0,2554
0,2226
0,2285
0,4130
0,6259
0,5416
0,5760
0,3579
0,6000
0,5182
0,5569
0,1757
0,8700
0,2218
0,7505
0,2480
0,7039
0,2451
0,8334
0,0808
0,1177
0,1615
0,1270
0,0681
0,1099
0,0903
0,1068
0,0225
0,0200
0,0168
0,0131
0,0142n
0,0478
0,0764
0,0900
0,0546
0,0669
0,0382
0,0404
0,0716
0,1294
0,0582
0,1059
0,1558
0,1294
0,1822
0,2318
0,0553
0,0714
0,0811
105,97
93,46
56,03
80,64
152,39
57,39
35,78
30,55
86,935
0,5275
940,88
574,001
269,20
5,397,33
85,833
0,5058
825,08
858,773
234,17
7,523,75
94,639
0,5324
898,97
702,04
122,85
3,833,55
111,293
0,4639
881,07
848,551
251,54
3,259,07
265
266
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Table A2.
Independent variables
Continued.
1995
1996
1997
1998
1999
Constant
9,4365
Gender:
Male
0,2196
Schooling:
Vocational training school 0,1108
Secondary school
0,1882
College
0,5461
Experience:
Linear
0,0454
Quadratic/100
0,2187
Cubic/10000
0,5394
0,4985
Quartic/1000000
Occupation:
Non-manual
0,2142
Managerial
0,7453
Productivity:
log(value worker/worker)
0,1698
Negative value added
Capital/labor ratio:
log(net fixed assets/worker) 0,0174
Firm size: (No. of workers)
1020
0,2114
2150
0,1142
3011,000
0,0795
1,0013,000
0,1176
3,001 +
0,1673
Unemployment:
log(unemployment rate)
0,0839
Industry:
49 dummies, F-test
68,61
Region:
15 dummies, F-test
15,13
9,5614
9,7322
9,7852
9,6720
0,2025
0,1929
0,1921
0,2088
0,1303
0,1999
0,6004
0,1297
0,2101
0,6263
0,1220
0,2154
0,6342
0,1183
0,2069
0,6272
No. of observations
Adjusted R2
F test
Heteroscedasticity
Omitted variables
Normality of residuals
90,717
0,5201
835,44
3,551,24
235,60
3,547,12
0,0549
0,0593
0,0646
0,0636
0,3004
0,3262
0,3883
0,3834
0,8134
0,8705
1,0912
1,0734
0,8038
0,8466
1,1013
1,0806
0,2389
0,8461
0,2281
0,7979
0,2334
0,8291
0,2491
0,8021
0,2170
0,1962
0,0319**
0,2256
0,0543
0,2301
0,1064
0,0243
0,0072
0,0068
0,2506
0,2868
0,2926
0,2749
0,1408
0,1692
0,1808
0,1901
0,0784
0,1085
0,1397
0,0796
0,1405
0,1202
0,0981
0,1710
0,1545
0,1242
0,1884
0,2179
0,0843
0,0725
0,0831
0,0970
68,61
69,81
69,59
71,34
15,13
16,27
18,90
14,90
97,918
0,5472
917,92
5,688,65
178,08
3,494,94
88,208
0,5591
906,14
3,435,75
174,96
3,515,77
102,102
0,5710
1,192,02
4,923,05
258,54
3,607,09
102,547
0,5771
1,243,58
4,894,68
277,51
4,534,56
0,0995
0,0058*
Unmarked coefficients are significant at 0,0001. Significant at * 0,001 ** 0,05 n not significant.
Notes: OLS regressions with heteroscedasticity corrected standard errors.
Base categories: women; completed or incomplete primary school; manual workers; firms with
51300 workers.
267
Estimations Results.
1986
1989
1992
1993
1994
vocational
secondary
college
vocational
secondary
college
05
610
1120
2130
05
610
1120
2130
05
610
1120
2130
0,1004
05
610
1120
2130
05
610
1120
2130
05
610
1120
2130
0,2351
0,1031
0,2088
0,2581
0,1384
0,0997
0,2256
0,3300
0,0112n
0,2553
0,4315
0,5305
MALES
0,1313
0,1009
0,1984
0,2555
0,1456
0,0801
0,2285
0,3448
0,0454***
0,2676
0,4766
0,6014
FEMALES
0,2615
0,1433
0,1682
0,0756
0,1049
0,0078n
0,0198n
0,3709
0,3900
0,2080
0,2571
0,0685
0,0983
0,0805
0,0598
0,0698
0,0927
0,1155
0,1855
0,2808
0,3570
0,4246
0,4782
0,1136
0,0747
0,0665
0,0909
0,1902
0,2498
0,0681**
0,1218
0,2655
0,3607
0,1572
0,3834
0,5579
0,6696
0,0878
0,1951
0,2516
0,0469
0,1280
0,2846
0,3699
0,2406
0,4528
0,5519
0,6886
0,1169
0,1972
0,2492
0,0422**
0,1270
0,2635
0,3600
0,2295
0,4390
0,5619
0,6894
0,2085
0,1969
0,2339
0,1112
0,1244
0,1719
0,0583
0,0677
0,0900
0,0249**
0,2374
0,1052
0,0174**
0,1224
0,1834
0,2896
0,4373
0,5436
0,0263**
0,2243
0,0997
0,0168***
0,1290
0,2101
0,3424
0,4980
0,6070
0,0015n
0,2432
0,1215
0,0344
0,0899
0,1894
0,3419
0,4539
0,6159
0,2892
0,2879
0,3135
0,3200
0,3403
116,205
0,3647
649,39
187,121
52,101
111,293
0,3998
576,21
234,891
29,371
86,935
0,5023
617,41
448,511
59,931
85,833
0,4802
616,97
628,163
85,28
94,639
0,5106
696,18
481,45
80,83
267
268
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Table A3.
Interactive variables
1995
Continued.
1996
1997
1998
1999
05
610
1120
2130
05
610
1120
2130
05
610
1120
2130
0,0745
05
610
1120
2130
secondary
05
610
1120
2130
college
05
610
1120
2130
male and female,
experience > 30
0,1985
0,2997
No. of observations
Adjusted R2
F- test
Heteroscedasticity
Omitted variables
90,717
0,4999
651,44
3,474,02
182,43
vocational
secondary
college
vocational
0,0771
0,1597
0,2199
0,0788
0,1060
0,2233
0,3217
0,1785
0,4097
0,5130
0,6238
0,1142
0,0777
0,0114n
0,2363
0,1231
0,0454
0,0592
0,2001
0,3774
0,4149
0,4988
MALES
0,0605
0,0713
0,1663
0,2309
0,0893
0,1159
0,2291
0,3205
0,1956
0,4736
0,5564
0,6317
0,0577
0,0559
0,0432
0,0564
0,1674
0,2353
0,0816
0,1087
0,2361
0,3301
0,2748
0,5312
0,5750
0,6691
0,0816
0,1668
0,2292
0,0622***
0,1159
0,2507
0,3275
0,2775
0,5273
0,6220
0,6511
0,0808
0,1716
0,2251
0,0306
0,1287
0,2600
0,3309
0,3072
0,5777
0,6214
0,6842
0,1607
0,1513
0,1351
0,1014
0,1065
0,0716
0,0660
0,0655
0,0629
0,0052n
0,2282
0,0693
0,0190+
0,0934
0,2182
0,4161
0,4513
0,5942
0,0076n
0,1879
0,0540
0,0268*
0,0977
0,2345
0,4646
0,5025
0,5546
0,3061
0,3057
0,2907
0,2817
97,918
0,5303
716,36
5,645,25
138,21
88,208
0,5445
737,18
3,357,66
156,04
102,102
0,5570
948,85
4,934,77
237,19
102,547
0,5622
1,006,96
4,909,06
236,07
FEMALES
0,1803
0,1184
0,0718
0,0148n
0,2397
0,0982
0,0556
0,0783
0,1725
0,3679
0,4348
0,5634
0,0077n
0,1857
0,0587
0,0345
0,0815
0,2299
0,4584
0,4735
0,5268
269
Estimation Results.
1992
1993
1994
1995
skilled-young
skilled-old
0,2675
0,5053
0,2707
0,5036
0,3170
0,5586
0,2641
0,4919
81,301
0,3927
9,8293
75,791
0,3872
9,9952
82,768
0,3811
10,2428
65,754
0,3573
10,3331
Interactive variables
1996
1997
1998
1999
skilled-young
skilled-old
0,2517
0,4888
0,2307
0,4714
0,2613
0,4959
0,2579
0,4796
77,733
0,4031
10,5108
68,013
0,4401
10,6838
75,415
0,4424
10,8141
75,321
0,4277
10,9606
No. of observations
Adjusted R2
Mean ln (monthly wage)
No. of observations
Adjusted R2
Mean ln (monthly wage)
FOREIGN FIRMS
Interactive variables
1992
1993
skilled-young
skilled-old
0,3753
0,5089
0,3543
0,4919
0,3745
0,5422
0,3382
0,4849
5,493
0,3992
10,0673
9,815
0,4004
10,2099
11,490
0,3919
10,4685
24,650
0,3952
10,6292
Interactive variables
1996
1997
1998
1999
skilled-young
skilled-old
0,3385
0,5032
0,3363
0,4968
0,3496
0,4723
0,3361
0,4307
19,814
0,4196
10,8283
20,195
0,4092
11,0455
26,687
0,4435
11,2282
27,226
0,4627
11,3634
No. of observations
Adjusted R2
Mean ln (monthly wage)
No. of observations
Adjusted R2
Mean ln (monthly wage)
1994
1995
269
270
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
where the -s measure the regression-adjusted wage advantage of workers with
E level of education, gender G, and exp years of experience over the base
category, in 1989 and 1999. The coefficients were estimated with the augmented
interactive model using 42 dummies for the interactions of two genders, three
levels of education, and seven experience groups. Workers with completed or
incomplete primary school background were treated as the reference.
Table A5. Life-Cycle Gains and Losses from Change in
Experience-Earnings Profiles between 1989 and 1999.
Labor market
entry
19841988
19791983
19741978
19691973
19641968
Experience
in 1989
15
610
1115
1620
2125
Vocational training
school
males females
2,5
2,8
4,0
3,1
2,1
6,0
3,7
1,6
3,7
4,4
Secondary school
males
females
3,1
4,3
0,6
2,2
3,4
12,0
4,5
3,8
0,3
0,4
College
males females
20,1
11,7
12,2
5,3
11,7
15,6
11,4
4,2
8,1
4,0
271
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
1986
1989
1992
1993
1,2928
1,2463
1994
2,2069
1,5558
0,2418
0,1175
0,1822
0,1591
0,1273
0,2428
0,1894
0,0723n
0,2145
0,2180
0,1069**
0,1833
0,4025
0,0170n
0,1354
971
0,2111
748
0,2170
600
0,1241
567
0,1736
506
0,2341
1995
1996
1997
*
0,8274
1998
**
0,4725
1,3857
0,4975
1999
*
0,5156
0,4461**
0,2510
0,0623n
0,2545
0,3227
0,3441
0,4385
0,0180n
0,0111n
0,0291n
0,2834
0,3417
0,2897
0,5355
0,1047**
0,2628
470
0,2666
477
0,3419
400
0,3665
476
0,4045
441
0,4677
DOMESTIC FIRMS
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
1992
1993
1994
1,2976
1,2451
1,4589
1995
0,9548
0,1658*
0,0788n
0,1982
0,1405*
0,1575*
0,1502
0,3346
0,0903n
0,1040
0,1157***
0,1868*
0,1887
543
0,1104
478
0,1324
420
0,2071
312
0,1777
1996
1997
1998
1999
0,6556
0,6073
0,6085
0,6032**
0,2431
0,0978n
0,2257
0,2492
0,0905n
0,2756
0,3103
0,1057***
0,2121
0,4225
0,0418n
0,1818
345
0,2923
276
0,3160
310
0,3518
269
0,3995
271
272
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Table A6.
Continued.
FOREIGN FIRMS
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
Independent Variables
Constant
log share of skilled-young
log share of skilled-old
log capital/labor ratio
No. of firms
Adjusted R2
1992
1993
1994
1995
1,5904*
0,8008*
0,4548*
0,1748n
0,1800n
0,5979*
0,1018n
0,1874
0,6069
0,0461n
0,3077
0,3721
0,0514n
0,2397*
57
0,2601
89
0,4876
86
0,3612
158
0,3607
1996
1997
1998
1999
0,3343n
0,0804n
0,3491n
0,2204n
0,3946
0,0642n
0,3424
0,3015*
0,0308n
0,3961
0,4657
0,0824n
0,3517
0,1265n
132
0,3881
124
0,3245
166
0,3390
172
0,4322
1,7159
1,8830
0,4980
0,3613
273
273
274
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40