Wage Determination in Rural Russia: A Stochastic Frontier Model

Abstract

This article examines wages in rural Russia after the first decade of economic transition using data from a nationally representative household survey. The stochastic frontier analysis reveals that Russia's rural labour markets place high value on human capital. The overall level of rural wages, however, is very low, with the median wage 10% below the official subsistence level. The gender pay gap severely depresses women's wages. A woman with the same skills as a man is paid only 47% of the man's wage. Rural workers who receive income from their personal plots accept significantly lower wages. Private firms pay considerably higher wages than state or collectively owned firms, but account only for one fifth of rural workers.

Full text

Wage Determination in Rural Russia: A Stochastic
Frontier Model
CONSTANTIN OGLOBLIN & GREGORY BROCK
Abstract
This article examines wages in rural Russia after the first decade of economic
transition using data from a nationally representative household survey. The
stochastic frontier analysis reveals that Russia’s rural labour markets place high
value on human capital. The overall level of rural wages, however, is very low, with
the median wage 10% below the official subsistence level. The gender pay gap
severely depresses women’s wages. A woman with the same skills as a man is paid
only 47% of the man’s wage. Rural workers who receive income from their personal
plots accept significantly lower wages. Private firms pay considerably higher wages
than state or collectively owned firms, but account only for one fifth of rural workers.
While much has been written about wages in the Russian transition economy (e.g.
Brainerd, 1998 and 2002; Kapelyushnikov, 2003; Newell & Reilly, 1996; Ogloblin,
1999, 2005a and 2005b; Ogloblin & Brock, 2005; and Reilly, 1999), the specifics of
wages in its rural sector are studied to a much lesser degree. What determines Russia’s
rural wages? To what degree does their variation reflect workers’ productivity-related
characteristics? Our study attempts to answer these questions. We examine Russia’s
rural wages after a decade of transition with the economy relatively stable plus new
labour relations and wage setting practices being developed.
To examine factors that determine wages, we use the stochastic frontier model,
which was originally proposed by Aigner et al. (1977) and since then has been
extensively used to analyse production and cost efficiency. Recently several studies
have applied it to analyse wages as well (Hofler & Murphy, 1992 and 1994; Hofler &
Polachek, 1985; Polachek & Yoon, 1987 and 1996; Polachek & Robst, 1998). In a
previous article we used the stochastic frontier method to examine wages in urban
Russia (Ogloblin & Brock, 2005). The present study is the first in the literature to
apply the stochastic frontier analysis to Russia’s rural wages.
In the first section we describe the data and sampling procedures. The next section
portrays Russia’s rural labour market. Then we explain the model we use and discuss
ISSN 1463-1377 print/ISSN 1465-3958 online/06/030315-12
q2006 Centre for Research into Post-Communist Economies
DOI: 10.1080/14631370600881945
Dr. Constantin Ogloblin, School of Economic Development, PO Box 8152, Georgia Southern
University, Statesboro, GA 30460-8152, USA. Email: coglobli@georgiasouthern.edu. Dr Gregory
Brock, Associate Professor, School of Economic Development, PO Box 8152, Georgia Southern
University, Statesboro, GA 30460-8152, USA. Email: gbrock@georgiasouthern.edu
Post-Communist Economies, Vol. 18, No. 3, September 2006

the variables in the frontier equation. The following section presents the stochastic
frontier estimates. The final section concludes.
The Sample
The study uses individual-level data from the latest three rounds (Rounds 11 –13) of
the Russia Longitudinal Monitoring Survey (RLMS), a household-based nationally
representative survey designed and implemented by an interdisciplinary partnership of
leading Russian and American experts.
1
The survey draws a multi-stage probability
sample. First, 1,850 consolidated raiony (administrative subdivisions similar to
counties in the United States) serve as primary sampling units (PSUs). The number of
households drawn into the sample is 4,718. Then, within each selected PSU the
population is stratified into urban and rural substrata, and the target sample size is
allocated proportionately to the two substrata.
The data for Rounds 11, 12, and 13 were collected in the autumn of 2002, 2003
and 2004 respectively. Interviewers were required to visit each selected dwelling up
to three times to secure the interviews. They were not allowed to make
substitutions of any kind. The household response rate exceeded 80%, and
individual questionnaires were obtained from over 97% of the individuals listed on
the household rosters. Our sample includes women aged 18–54 and men aged 18 –
59, which are considered the normal working ages for women and men in Russia,
who were able to work.
The Labour Market in Rural Russia
The RLMS data show that work for a wage plays a very important role in rural
Russia (Table 1). Although the rural labour force participation rate is noticeably below
that in urban areas, it is rather high by international standards.
2
The ratio of wage-
employed to labour force in rural areas is only slightly below that in urban areas.
3
This
difference comes entirely from the higher rate of rural unemployment. The percentage
of wage-employed in total employment in rural areas is virtually the same as that in
urban areas, with about 91% of the employed working-age population working for a
wage in either area.
The wage measure used is the after-tax average monthly wages received at the
primary place of employment in the last 12 months before the interview.
4
That is, we
examine rural workers’ effective monthly wages smoothed over time.
5
Real wages are
Table 1. Labour force participation and employment in Russia, 2002 –04
a
(%)
Rural Urban
Labour force participation rate
b
72.5 83.6
Unemployment rate 12.0 8.8
Wage employment rate
c
80.0 83.2
Wage employment as % of total employment
d
90.8 91.2
Notes:
a
Calculated from the RLMS data, Rounds 11–13 for women aged 18–54 and men aged 18 –59.
b
Calculated using the ILO definition of unemployment (not employed and actively looking for a job).
c
Percentage of labour force whose primary place of employment is an enterprise, organisation, collective
farm, state farm or cooperative.
d
Other forms of employment include farmers, entrepreneurs and other self-employed.
316 Constantin Ogloblin & Gregory Brock

calculated in December 2002 rubles using the official CPI (Goskomstat, 2005) and
described in Table 2.
The overall level of wages in rural Russia is very low, with the median wage 10.2%
below the official subsistence level and the third quartile wage only 53.2% above that
level.
6
But this depressed wage distribution shows broad variation, which indicates
considerable wage inequality. Within one standard deviation from the mean, the wage
falls to only 57 rubles per month, and the third to first quartile wage ratio is 2.95 to 1.
One factor that clearly contributes to this wage inequality is the gender wage
differential. On average, men are paid 48.5% more than women. Women’s median
wage is 24% below the subsistence level, while men’s is above that level. And this is
despite the fact that rural female workers are far better educated than their male
counterparts. As shown in Table 3, 47.8% of wage-employed women have specialised
secondary or higher education, with 16.1% holding a university degree, while for men
these percentages are only 19.5 and 8.8 respectively. On the other hand, only 14.4% of
women have not completed secondary education, while for men this figure is 28.2%.
7
Table 2. Average monthly wages of Russian rural workers by gender, 2002 –04
(December 2002 rubles)
a
All workers (100%) Women (50%) Men (50%)
Mean 2,362 1,906 2,830
Standard deviation 2,305 1,688 2,722
Coefficient of variation, %
b
97.6 88.6 96.2
First quartile 982 875 1,199
Median 1,700 1,439 2,000
Third quartile 2,900 2,398 3,517
Quartile ratio
c
2.95 2.74 2.93
Notes:
a
The after-tax average monthly wages received at the primary place of employment in the last 12
months divided by the CPI. Calculated from the RLMS data, Rounds 11–13 for wage-employed women
aged 18–54 and men aged 18 – 59.
b
(Standard deviation / Mean) £100.
c
Third quartile / First quartile.
Table 3. Rural workers’ human capital characteristics, 2002 – 04
a
Women Men
Education (% of workers):
b
Incomplete secondary 14.4 28.2
General secondary 19.1 26.7
Secondary with vocational 18.8 25.6
Specialised secondary 31.7 10.8
University 16.1 8.8
On-the-job training (% of workers) 5.1 3.4
Mean years with the current employer 7.7 7.4
Age groups:
18–24 15.4 14.8
25–34 27.1 27.9
35–49 44.7 40.8
50–59 12.9 16.5
Notes:
a
Calculated from the RLMS data, Rounds 11 – 13 for wage employed women aged 18 – 54 and men
aged 18–59.
b
The highest level obtained.
Wage Determination in Rural Russia 317

The percentage of women who received on-the-job training, however, is very low, but
is even lower for men.
Since labour force participation rates are traditionally high in Russia, workers’
ages are likely to be closely correlated with their labour market experience. The
distributions of wage-employed women and men by age shown in Table 3 are almost
identical. The only difference is that the men’s distribution is somewhat shifted from
ages 35– 49 to ages 50 – 59 compared with the women’s distribution. This is explained
by women’s younger legal retirement age (55 vs. 60 for men). Workers’ firm-specific
experience is reflected by their tenure with the current employer, which does not differ
significantly between women and men.
Given that women, despite their far better human capital endowments, receive
much lower wages than men do, the question arises whether Russia’s rural labour
market places a low value on human capital or whether it discriminates against
women, possibly taking advantage of their willingness to work for less. Our stochastic
frontier analysis attempts to answer this question.
Another interesting observation from the RLMS data is that wages of those rural
workers who receive supplemental income from their personal plots are considerably
lower than those of workers who have no such income (Table 4). The mean wage of
workers who sell products from their plots is 31.3% below that of workers who do not,
and for the median wage this difference is 20.1%. The human capital endowments of
the two categories of workers are about the same, and even somewhat better among
workers who receive income from their personal plots. That is, the latter group’s lower
wages are hardly a result of their lower skills.
Wages in Russia’s rural sector also differ depending on the type and size of firm.
As shown in Table 5, 61.6% of rural workers are employed at state-owned enterprises
and organisations, and only 20.1% are employed by private firms. Despite the fact that
the latter are significantly less well endowed with human capital, they enjoy
considerably higher wages. The mean wage in the private sector is 62.8% higher and
the median wage is 25.1% higher than in the state sector. Again, from these descriptive
statistics we cannot clearly see whether it is human capital that is given a low value by
Russia’s rural labour market or firm type by itself is an important factor influencing
wages. We leave this question to the stochastic frontier analysis. The size of the firm
also appears to influence wages. Both mean and median wages increase monotonically
with firm size. However, it is medium-sized firms (20 – 200 employees), not large
Table 4. Income from personal plots, wages and human capital endowments,
2002– 04
a
Sell product from personal plots?
Yes (33.3%) No (66.7%)
Wage
b
Mean 1,813 2,640
Median 1,439 1,800
Workers with specialised secondary or higher education (%) 35.5 32.7
Notes:
a
Calculated from the RLMS data, Rounds 11 – 13 for wage-employed women aged 18 –54 and men
aged 18–59.
b
The after-tax average monthly wages received at the primary place of employment in the last 12 months,
in December 2002 rubles.
318 Constantin Ogloblin & Gregory Brock

firms, where the percentage of workers with specialised secondary or higher education
is the highest.
The Stochastic Frontier Wage Equation
To see more clearly how Russia’s rural labour markets determine wages (and in
particular how valuable human capital is), we employ the stochastic frontier
technique. Our analysis is based on the classical human capital earnings function
originally proposed by Mincer (1974), which can be written as follows:
wi¼
b
0þ
b
SSiþ
b
XXiþ1ið1Þ
or
wi¼E½wijSi;Xiþ1ið2Þ
where w
i
is worker i’s wage, S
i
is the set of variables that reflect her level of education,
X
i
is the vector of variables reflecting her labour market experience,
b
0
,
b
S
and
b
X
are
the parameters to be estimated, and 1
i
is the error term. So E½wijSi;Xiis worker i’s
expected (mean) wage, given her human capital endowments S
i
and X
i
. But unlike the
classical earnings function, our model of wage determination does not seek to estimate
this mean conditional wage. Instead, it allows us to find a wage frontier determined by
human capital endowments, i.e. the expected minimum level of wage that a worker
with given skills can earn. Then it allows us to see how and to what extent wage
determinants other than human capital cause actual wages to deviate from this frontier.
Table 5. Wages by firm type and size, 2002 – 04
a
Firm type
State
b
Private Other
c
Percentage of workers employed 61.6 20.1 18.3
Wage
d
Mean 2,009 3,271 2,534
Median 1,599 2,000 1,607
Percentage of workers with specialised secondary
or higher education
38.6 28.9 22.2
Firm size
e
Small Medium Large
Percentage of workers employed 33.6 50.0 16.4
Wage
d
Mean 1,990 2,356 2,680
Median 1,380 1,786 2,000
Percentage of workers with specialised secondary
or higher education
34.8 42.0 32.2
Notes:
a
Calculated from the RLMS data, Rounds 11 – 13 for women aged 18 – 54 and men aged 18 – 59.
b
State-owned enterprise or organisation.
c
Firms with mixed or undefined ownership.
d
The after-tax average monthly wages received at the primary place of employment in the last 12 months,
in December 2002 rubles.
e
Small firms are defined as those with fewer than 20 employees, and large firms are those with more than
200 employees.
Wage Determination in Rural Russia 319

In other words, instead of the usual assumption that the error term 1
i
is normally
distributed with a mean of zero, we treat it as the sum of two components:
1i¼viþuið3Þ
where v
i
is a random disturbance distributed as Nð0;
s
2
vÞ, but u
i
is assumed to have a
normal distribution truncated at zero, with a mean of m$0 and variance of
s
2
u.
To analyse factors that might cause systematic deviations of wages from the human
capital frontier, we model the mean parameter mas a linear combination of variables
that reflect those factors:
m¼
d
Di;ð4Þ
where D
i
is the vector of variables that reflect wage determinants other than worker i’s
skills, and
d
is the vector of parameters to be estimated. If the parameter is positive, then
the corresponding factor increases the difference between the actual wage and the
human capital frontier wage, and if it is negative, then it decreases that difference.
8
The dependent variable used in the wage equation is the natural logarithm of the
after-tax average monthly wages received at the primary place of employment in the
last 12 months. The Svector in equation (1) includes a set of dummy variables
signifying workers’ level of education (the highest degree obtained) and whether or
not they had on-the-job training. General labour market experience is reflected by the
age variable, which together with its square term models a typical concave
experience– earnings profile. Firm-specific experience is reflected by tenure with the
current employer. Our wage equation also includes controls for part-time work status
(less than 35 hours per week), macro region (to account for regional differences in cost
of living and wages) and year.
Variables included in equation (4) attempt to capture the influence of factors other
than workers’ skill-related characteristics on the mean deviation from the wage
frontier (the uterm). The gender variable, for example, is included in equation (4) with
the assumption that men and women with equal human capital characteristics are
equally skilled. Other variables included in vector Drepresent the factors discussed in
the previous section. These are dummy variables that represent additional income
from the worker’s personal plot, the type of firm and its size. We also include variables
that signify possible workers’ influence on the wage decision: supervisory
responsibility and co-ownership of the firm. Finally, variables that may influence
workers’ reservation wages are marital status and whether or not they have a second
job. All variables are described in Table A1 in the Appendix.
Results
This frontier equation is estimated by maximum likelihood. The results are shown in
Table 6. The wage frontier is profoundly influenced by the worker’s level of
education. The returns to education are high and follow the pattern predicted by
human capital theory. A university degree raises the frontier wage by 99.2% compared
with general secondary education
9
and the wage premium for a specialised secondary
qualification is 32.1% over general secondary and 52.6% over incomplete secondary
education. On-the-job training also significantly influences wages, shifting the wage
frontier up by 28.6%. Thus, Russia’s rural labour market values workers’ skills highly.
Russia’s rural workers’ experience– wage profile is also consistent with human
capital theory. As can be derived from the coefficient estimates, the wage frontier
320 Constantin Ogloblin & Gregory Brock

Table 6. Frontier equation estimates
a
Variable
b
Coefficient St. error
Wage frontier
Education:
Incomplete secondary 20.144*** 0.043
Secondary vocational 0.037 0.041
Specialised secondary 0.279*** 0.043
University 0.689*** 0.050
On-the-job training 0.251*** 0.067
Experience:
Age 0.023** 0.011
Age
2
/100 20.041*** 0.015
Years with current employer 0.006*** 0.002
Part-time work 20.302*** 0.045
Macro region:
Caucasus 20.214*** 0.050
Eastern Siberia 0.022 0.053
North 0.525*** 0.061
Ural 20.043 0.056
Volga 20.422*** 0.054
Western Siberia 20.354*** 0.058
Year:
2002 20.141*** 0.034
2004 0.139*** 0.033
Constant 6.861*** 0.221
Deviation from the frontier
Man 0.760*** 0.131
Income from personal plot 20.334*** 0.074
Single 20.224*** 0.074
Worker co-owns firm 20.667*** 0.161
Second job 20.236** 0.093
Supervisory responsibility 0.146** 0.074
Firm size:
Small 20.218*** 0.081
Large 0.188** 0.074
Undefined 0.107 0.065
Firm type:
Mixed ownership 0.527*** 0.083
Private 0.543*** 0.068
Undefined ownership 0.231** 0.093
Constant 20.311*0.170
Number of observations 2610
Wald x
2
793.2***
g
c
0.098***
Mean deviation from the frontier
d
0.308
Notes:
a
The dependent variable is the logarithm of average monthly wage after taxes
received in the last 12 months before the interview, in December 2002 rubles.
b
The baseline categories are general secondary education, Central macro region,
year 2003, medium-sized firm, and state-owned enterprise or organisation.
c
Calculated as
g
¼
s
2
u=
s
2where
s
2¼
s
2
vþ
s
2
u.
d
Calculated as PN
i¼1ui=Nwhere Nis the number of observations.
*Statistically significant at the 0.1 level; ** at the 0.05 level; *** at the 0.01 level.
Wage Determination in Rural Russia 321

initially rises with the worker’s age (which is a proxy for general labour market
experience), reaches its maximum at 28 years and then declines. That is, the labour
market favours relatively young workers. The wage profile is rather flat for younger
ages, so that, for example, the wage differential between 28 year old and 18 year old
workers is only around 4%. But for older ages the profile becomes steeper, so that the
differential between 28 year old and 55 year old workers is about 26%.
10
The wage
frontier is also significantly positively influenced by the worker’s firm-specific
experience. This effect, however, is very small: the wage rises only 0.6% per
additional year of tenure.
The wage equation estimates show considerable regional wage differentials. For
example, the wage level in the Northern macro region is 2.6 times that in the Volga
region. And over the three-year period, real wages in rural Russia show a time trend
with a constant 15% rate of growth.
The estimates from the ‘u’ part of the equation indicate that the average deviation
of wages from the human capital-determined frontier is 36.1% and that it is
statistically significant.
11
That is, factors other than human capital also significantly
influence rural workers’ wages. More importantly, the coefficient estimates in this part
allow us to see the role of each of those factors. Perhaps the most important among
them is gender. The frontier analysis allows us to conclude that in rural Russia a man
with a given level of human capital is paid 2.1 times as much as an equally skilled
woman. This is consistent with the descriptive statistics showing that despite their far
better human capital endowments women receive much lower wages than do men.
What can explain such a large gender pay gap? We believe two factors contribute
to it. First, women agree to accept lower wages, i.e. women have lower ‘reservation
wages’ than men do. According to the RLMS (2003), 66.4% of rural women and
74.1% of men agree that it is a husband’s responsibility to earn money and a wife’s
responsibility to take care of the house and children. Only 16.7% of women and 8.0%
of men disagree with this statement (the rest are not sure).
12
This attitude on both
sides, women and their employers, leads to a gender pay gap.
13
Second, the same
social attitude results in employer prejudice against women as less motivated and less
productive workers, leading to all sorts of direct and indirect discrimination against
them (see Katz, 2001, pp. 11– 37). Lower reservation wages of single workers coupled
with employers’ compassion for workers with families lead to wages of single workers
being 20.1% lower than those of married workers.
Another important factor that influences wage deviation from the human capital
frontier is income from workers’ personal plots. The wages of workers who sell
products from their plots are on average 28.4% lower than the wages of those who do
not, and the difference is highly statistically significant. Workers with an additional
source of income agree to accept lower wages from their employers (i.e. have lower
reservation wages) than workers for whom those wages are the only source of income.
We can also hypothesise that with income from personal plots, wages become less
important relative to non-wage benefits from formal employment, such as the official
employment record (which influences the state pension), social benefits provided by
the organisation, and perhaps also access to the firm’s inputs and outputs that workers
could use for their private production. A similar explanation may be suggested for the
finding that the expected wage at the primary place of employment is 21.0% lower if
the worker has a second job.
Private firms and firms with mixed ownership
14
pay significantly more (72.1%
and 69.4% respectively) to a worker with given skills than do state enterprises and
organisations. One possible explanation for this is that workers at firms with private
322 Constantin Ogloblin & Gregory Brock

ownership are more productive because those firms are usually better organised and
better managed, and their wage-setting practices (formal and informal) provide
greater incentives for productive work. This may also explain why the wages of
workers who consider themselves to be co-owners of their firm are 48.7% lower than
the wages of those who do not. That is, firms with worker ownership are apparently
less productive and less financially sound than other firms, which seems to depress
wages more than the influence of worker-co-owners can push them up.
15
Also, state
enterprises and employee co-owned firms usually provide more social benefits to
their workers than do private firms, which must be compensated by a higher level of
wages at the latter.
Firm size can influence wages in two ways. First, bigger firms may pay higher
wages because they are more likely to have more financial resources.
16
Second, large
firms may possess monopsony power in the local labour market and hence tend to pay
lower wages than do smaller firms. As follows from the estimates in Table 6, the first
tendency prevails. Wages rise monotonically and statistically significantly from the
lowest at small firms to the highest at large firms.
Conclusions
Wage employment plays a very important role in rural Russia, with 91% of the
employed working-age population working for a wage. And the ratio of wage-
employed to labour force in rural areas is only slightly below that in urban areas. The
overall level of rural wages, however is very low, with the median wage 10% below
the official subsistence level.
Our stochastic frontier analysis reveals that the expected minimum level of wage
that a rural worker with given skills can earn is profoundly influenced by the worker’s
level of education. A university degree almost doubles that wage compared with
general secondary education. On-the-job training also raises wages significantly.
Rural workers’ experience– wage profile is typical, concave, reaching its maximum at
28 years of age, i.e. favouring fairly young workers. Firm-specific experience also
significantly positively influences wages, but this effect is rather small in magnitude.
The frontier results show that the gender differential in rural wages, net of worker
productivity-related characteristics, is very large. A woman with a certain level of
skills is paid only 47% of what an equally qualified man would be paid. Rural Russia’s
patriarchal social attitude toward women’s role in society, which is prevalent among
both men and women, leads to a labour market where women ‘willingly’ accept lower
wages and where employer prejudice against them causes direct and indirect pay
discrimination. Low wages in rural Russia are to a large extent a result of these labour
market practices, as they considerably depress wages of the higher skilled half of the
rural labour force.
Our results also show that rural workers with additional income, the most
important source of which is personal plots, accept significantly lower wages. This
supports the hypothesis that non-wage benefits from formal employment, such as the
official employment record, social benefits, and perhaps also access to the firm’s
inputs and outputs, are more important to this category of workers.
Private and partly private firms pay considerably higher wages than state or formally
worker-owned enterprises and organisations, which may suggest that the former are
more productive and financially sound. Private firms, however, account for only one
fifth of rural wage-employed workers, and this is another explanation of why Russia’s
rural wages are so low, and an argument for further privatisation of Russia’s rural sector.
Wage Determination in Rural Russia 323

Notes
1. The survey has been coordinated by the Carolina Population Center (CPC) at the
University of North Carolina at Chapel Hill in collaboration with Paragon Research
International and the Russian Academy of Sciences. Detailed project descriptions
including the sampling techniques and the RLMS datasets are available from the RLMS
website, http://www.cpc.unc.edu/projects/rlms/rlms_home.html.
2. For example, the overall labour force participation rate in the US is about 66% (BLS,
2006).
3. We define wage-employed as those whose primary place of employment is an enterprise,
organisation, collective farm, state farm or cooperative.
4. We restrict our sample to workers whose tenure with the current firm is at least one month.
5. Another measure of wages reported by the RLMS respondents is the amount received in
the last month before the interview. This measure, however, has several disadvantages
with regard to rural wages. First, given the persistent wage arrears, it is unclear whether
the money received last month reflects the contractual monthly wage, back wages finally
being paid or some combination of both. Second, owing to the seasonality of agricultural
production and related rural industries, with most interviews being conducted in late
autumn, one cannot be sure that wages received in the last month before the interview are
workers’ typical monthly wages.
6. The official subsistence level in the fourth quarter of 2002 was 1,893 rubles (Goskomstat,
2005).
7. General secondary education in Russia is 10– 11 years of a general school. Secondary
education with vocational training can be obtained at vocational schools, which offer
vocational training either along with complete general secondary education (for those
who enter them with only 7–8 years of general schooling) or in addition to general
secondary education. Students of specialised secondary schools acquire some special
(technical, medical, pedagogical, art) knowledge along with or in addition to general
secondary education.
8. This is essentially a wage equation analogue of the Battese & Coelli (1995) stochastic
production frontier model (with the reverse sign of the uterm). The idea is essentially the
same, but instead of factors of production we use human capital endowments, and instead
of the production frontier we estimate the wage frontier.
9. The marginal effects of the variables shown by the coefficient estimates in Table 6 are in log
points. The formula to convert them to percentage point effects is p¼½exp ð
b
Þ21£
100 where pis the percentage point effect and
b
is the log effect.
10. The mathematics here is asfollows. Given the estimates in Table 6, the marginal effect of age
(a)is
›
E½wja

=
›
awhere E½wja¼w0þ0:023a20:041 a2=100

,andw
0
is the level
of wage determined by factors other than age. Then,
›
E½wja

=
›
a¼0:023 20:00082a,
and the age– wage profile reaches its maximum when this marginal effect is zero (the first-
order condition), i.e. when a¼28. Further, assuming that w
0
equals the mean log wage,
which is 7.445, we can calculate the expected log wages and their differentials.
11. Following Battese & Corra (1977), we define
g
¼
s
2
u=
s
2, where
s
2¼
s
2
vþ
s
2
uand 0 #
g
#1, and use Coelli’s (1995) third-moment test to see whether
g
is statistically different
from zero.
12. This patriarchal attitude is more pronounced in rural than in urban Russia. The RLMS
data show that in urban areas the percentages of women and men who conform to it are
53.2% and 63.6%, respectively, and the percentages of those who do not are 22.9% and
13.5%.
13. As Ogloblin (1999, 2005a and 2005b) shows, this occurs mostly through job segregation
by gender rather than through direct offers of lower pay for the same job.
14. That is, partly privately owned.
15. When interpreting this result, one should keep in mind that the variable in question
reflects workers’ perception of whether they co-own the firm, rather than the fact that they
324 Constantin Ogloblin & Gregory Brock

actually own a share in their firm. According to the RLMS 2002, 71.2% of rural workers
who claimed they co-owned their firm could not say how big their share was. And out of
those who could, 42.5% owned less than 1% of the firm, and only 18.4% owned more than
5% (unfortunately, this question was not asked in RLMS 2003 and 2004). Thus, much of
the reported worker ownership is only formal and is likely to signify weak enforcement of
property rights and hence low efficiency and productivity. We suspect, for example, that
many of the workers who call themselves co-owners are simply members of the former
kolkhozy, which have not changed much with respect to their ambiguous definition of
property rights.
16. Kapelyushnikov (2003) finds evidence of this in Russia’s industrial sector.
References
Aigner, D. J. et al., ‘Formulation and Estimation of Stochastic Frontier Production Function
Models’, Journal of Econometrics, 6, 1977, pp. 21 – 37.
Battese, G. E. & Coelli, T. J., ‘A Model for Technical Inefficiency Effects in a Stochastic
Frontier Production Function for Panel Data’, Empirical Economics, 20, 1995,
pp. 325–332.
Battese, G. E. & Corra, G. S., ‘Estimation of a Production Frontier Model: With Application to
the Pastoral Zone of Eastern Australia’, Australian Journal of Agricultural Economics, 21,
1977, pp. 169– 179.
Brainerd, Elizabeth, ‘Five Years After: The Impact of Mass Privatization on Wages in Russia,
1993– 1998’, Journal of Comparative Economics, 30, 2002, pp. 160 – 190.
Brainerd, Elizabeth, ‘Winners and Losers in Russia’s Economic Transition’, American
Economic Review, 88, 5, 1998, pp. 1094 – 1116.
Bureau of Labor Statistics (BLS), 2006, http://www.bls.gov
Coelli, T.J., ‘Estimators and Hypothesis Test for a Stochastic Frontier Function: A Monte Carlo
Analysis’, Journal of Productivity Analysis, 6, 4, 1995, pp. 247 – 268.
Hofler, Richard A. & Murphy, Kevin J., ‘Estimating Reservation Wages of Employed
Workers Using a Stochastic Frontier’, Southern Economic Journal, 60, 4, 1994, pp.
961–976.
Hofler, Richard A. & Murphy, Kevin J., ‘Underpaid and Overworked: Measuring the Effects of
Imperfect Information on Wages’, Economic Inquiry, 30, 3, 1992, pp. 511 – 529.
Hofler, Richard A. & Polachek, Solomon W., ‘A New Approach for Measuring Wage
Ignorance in the Labor Market’, Journal of Economics and Business, 37, 3, 1985,
pp. 267–276.
Kapelyushnikov, R.I., Mekhanizmy formirovaniya zarabotnoi platy v Rossiiskoi promyshlen-
nosti, WP3/2003/07, Moscow State University, Higher School of Economics, 2003.
Katz, Katarina, Gender, Work and Wages in the Soviet Union: A Legacy of Discrimination
(New York, Palgrave, 2001).
Mincer, J., Schooling, Experience and Earnings (New York, National Bureau of Economics
Research, 1974).
Newell, Andrew & Reilly, Barry, ‘The Gender Wage Gap in Russia: Some Empirical
Evidence’, Labour Economics, 3, 3, 1996, pp. 337 – 356.
Ogloblin, Constantin G., ‘The Gender Earnings Differential in the Russian Transition
Economy’, Industrial & Labor Relations Review, 52, 4, 1999, pp. 602 – 627.
Ogloblin, Constantin, ‘The Gender Earnings Differential in Russia After a Decade of
Economic Transition’, Applied Econometrics and International Development,5,3,
2005a, pp. 5– 26.
Ogloblin, Constantin, ‘The Sectoral Distribution of Employment and Job Segregation by
Gender in Russia’, Regional and Sectoral Economic Studies, 5, 2, 2005b, pp. 5 – 18.
Ogloblin, Constantin & Brock, Gregory, ‘Wage Determination in Urban Russia: Underpayment
and the Gender Differential’, Economic Systems, 29, 3, 2005, pp. 325–343.
Wage Determination in Rural Russia 325

Polachek, Solomon W. & Robst, John, ‘Employee Labor Market Information: Comparing
Direct World of Work Measures of Workers’ Knowledge to Stochastic Frontier
Estimates’, Labour Economics, 5, 2, 1998, pp. 231– 242.
Polachek, Solomon W. & Yoon, Bong Joon, ‘A Two-Tiered Earnings Frontier Estimation of
Employer and Employee Information in the Labor Market’, The Review of Economics and
Statistics, 69, 2, 1987, pp. 296 – 302.
Polachek, Solomon W. & Yoon, Bong Joon, ‘Panel Estimates of a Two-tiered Earnings
Frontier’, Journal of Applied Econometrics, 11, 1996, pp. 169– 178.
Reilly, Barry, ‘The Gender Pay Gap in Russia during the Transition, 1992–96’, Economics of
Transition, 7, 1, 1999, pp. 245 – 264.
Rossiya v tsifrakh (Moscow, Goskomstat, 2005).
Appendix
Table A1. Variable descriptions
Variable Mean St. deviation
Log wage 7.445 0.847
Education:
Incomplete secondary 0.207 0.405
Secondary vocational 0.221 0.415
Specialised secondary 0.221 0.415
University 0.128 0.334
On-the-job training 0.045 0.208
Experience:
Age 38.3 10.3
Age
2
/100 15.7 7.9
Years with current employer 7.99 8.75
Part-time work 0.116 0.320
Macro region:
Caucasus 0.220 0.415
Eastern Siberia 0.176 0.381
North 0.093 0.291
Ural 0.123 0.328
Volga 0.152 0.359
Western Siberia 0.118 0.323
Year:
2002 0.329 0.470
2004 0.333 0.471
Man 0.502 0.500
Income from personal plot 0.341 0.474
Single 0.213 0.410
Worker co-owns firm 0.104 0.305
Second job 0.093 0.290
Supervisory responsibility 0.166 0.372
Firm size:
Small 0.251 0.433
Large 0.130 0.336
Undefined 0.225 0.418
Firm type:
Mixed ownership 0.096 0.294
Private 0.204 0.403
Undefined ownership 0.089 0.285
Number of observations 2,610
326 Constantin Ogloblin & Gregory Brock