Defence and Peace Economics
ISSN: 1024-2694 (Print) 1476-8267 (Online) Journal homepage: http://www.tandfonline.com/loi/gdpe20
An analysis of the public–private wage differential
in the Palestinian labour market
Sami H. Miaari
To cite this article: Sami H. Miaari (2018): An analysis of the public–private wage differential in the
Palestinian labour market, Defence and Peace Economics, DOI: 10.1080/10242694.2018.1473137
To link to this article: https://doi.org/10.1080/10242694.2018.1473137
Published online: 09 May 2018.
Submit your article to this journal
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=gdpe20
Defence anD Peace economics, 2018
https://doi.org/10.1080/10242694.2018.1473137
An analysis of the public–private wage differential in the
Palestinian labour market
Sami H. Miaari
Department of Labor studies, Tel-aviv University, Tel-aviv, israel
ABSTRACT
ARTICLE HISTORY
This paper measures and analyzes the dynamics of the public–private wage
differential in the West Bank and Gaza for the period before and during
the ‘second Intifada’ using data from the Palestinian Labour Force Survey
(PLFS) of the Palestinian Central Bureau of Statistics (PCBS). Because the
distribution of workers’ individual characteristics, such as skills, and the
‘returns’ to these characteristics may differ across workers, the wage
differential is decomposed into two components: an ‘endowment’ effect
and a ‘returns’ effect. The results show that in the pre-Intifada period, the
wage gap between the public and private sectors narrowed in both the West
Bank and Gaza. However, a sharp increase is seen after the outbreak of the
Intifada. Moreover, most of this increase comes from an increase in ‘returns’
to skills composition in the public sector, (unexplained effect), rather than a
change in the skills composition of public sector workers, (explained effect).
Using recent econometric quantile regression techniques, the analysis of
the public–private sector wage gap from 1998 to 2006, at various points
along the wage distribution, shows that the wage premium, (penalty), for
the public sector varies across the distribution, being higher, (lower), at the
lowest end of the wage distribution and decreasing (increasing) along the
wage distribution; it becomes negative in the top percentiles.
Received 28 December 2017
accepted 3 may 2018
KEYWORDS
conflict; decomposition;
intifada; Palestine; public
sector; quantile; wage gap
JEL CLASSIFICATION
J21; J31; J61; J45; c14; c24
Introduction
Public sector salaries have attracted much attention in the West Bank and Gaza in recent years. Salaries
are by far the largest component of the Palestinian Authority’s (PA’s) budget, accounting for 62% of
total expenditure in 2005. Labour market data from the last quarter of 2006 suggest that public sector
employment accounts for nearly 25% of all full-time employment. As the largest employer in Palestine,
the PA has both political and economic influence on Palestinian labour markets.
Historically, in the West Bank and Gaza, as in other developing countries, the processes of recruitment,
promotion and wage determination have differed substantially between the public and private sectors.
Following the second Intifada, the Palestinian public sector was perceived as a ‘buffer’ to absorb the
private sector job losses resulting from the impact, on product and labour markets, of increased closures
and movement restrictions imposed by Israel. (Cali and Miaari, 2018; Miaari and Sauer 2011). Latterly,
there were also increases in public sector salaries, such that in 2005 the wage bill averaged $US 85 m per
month.1 Given that the net revenues of the PA were approximately $US 101 m per month in 2005 (Figure 1),
CONTACT sami H. miaari
samimiaari@post.tau.ac.il
© 2018 informa UK Limited, trading as Taylor & francis Group
2
S. H. MIAARI
Figure 1. net Revenue and Wage Bill, 1998–2005. source: international monetary fund data files.
the commitment to public sector salaries and employment crowded out other forms of expenditure,
including spending on the maintenance and development of critically important infrastructure.
This situation has created a serious dilemma for a Palestinian Government faced with the unenviable
task of cutting expenditure at a time when the economy has been stagnating as a result of the closure
regime; reducing its workforce and/or reducing public sector salaries carries the considerable risk of
dangerously high levels of unpopularity, at a time when the PA needs to convince the populace of its
ability to carry out its electoral mandate. On the one hand, high public sector salaries can represent
unsustainably high costs for the PA, while on the other, excessively low salaries can be expected to
have a negative impact on employee motivation and productivity, also making it difficult for the PA to
attract skilled professionals.
Theoretical and empirical research points to a number of reasons for the emergence of public–private
wage differentials, the most obvious being the difference in the economic, political and institutional
environment surrounding these sectors. The public sectors of most countries are not bound by the
profit-maximizing concerns of private sector firms and the consequent cost-consciousness that prevails
in a competitive market. In most cases, particularly in developing countries, the public sector must compete with the private sector to attract the top professionals. This situation suggests that while there is
a ‘floor’ to public sector wages―often dictated by private sector wages―there might not be a ‘ceiling’.
Finally, wage differentials can also emerge as part of the ‘electoral wage-cycle process’, or because of
the collective bargaining strength of the health care and education sectors (Disney and Gosling 1998).
International data on wage differentials are mixed, and it is not easy to draw generalizations based
on country patterns.2 In some developing and developed countries, such as Australia (Cai and Liu
2011) Brazil (Emilio, Ponczek, and Botelho 2012), Greece (Papapetrou 2006), Haiti (Terrell 1993), India
(Glinskaya and Lokshin 2005; Azam and Prakash 2015), Ireland (Foley and O’Callaghan 2011) Italy (Carlo,
Lucifora, and Origo 2005; Depalo and Giordano 2011), Pakistan (Naser 2000; Hyder 2002; Hyder and
Reilly 2005; Aslam and Kingdon 2009) and Romania (Voinea and Mihaescu 2012), there is evidence of
a significant wage premium in favour of the public sector. On the other hand, in cases such as Estonia
(Leping 2005, 2006), Germany (Dustmann and van Soest 1998) and Poland (Adamchik and Bedi 2000)
the bias seems to be in the other direction, with a negative public wage premium (or a positive private
wage premium). France has mixed results, with Giordano et al. (2011) and Lucifora and Meurs (2004)
showing a negative public sector premium, and Bargain and Melly (2008) finding both penalties, (for
male employees), and premiums, (for females), to be the consequence of selection, and the gap itself
DEFENCE AND PEACE ECONOMICS
3
negligible. Some recent studies find a gradual narrowing of the public sector premium in many OECD
countries, as market forces have begun to influence public sector performance and decision-making.
Christofides and Michael (2013) find the public–private wage premium to be generally positive
for a 2008 cross-sectional sample of workers from 27 European states, but also some inter-country
heterogeneity, with Luxembourg, Cyprus, Greece, Hungary and Estonia3 having the largest gaps and
Belgium, Germany and Norway with small negative gaps. Giordano et al. (2011) also find the gap to be
generally positive based on data from 10 European countries over the years 2004–2007, with Greece,
Ireland, Italy, Portugal and Spain exhibiting a larger gap and France exhibiting a negative one. Panizza
and Qiang (2005) find positive public sector wage premiums for most of 13 Latin American countries
studied, and Mizala, Romaguera, and Gallegos (2011) further examine such gaps using 1992–2007
panel data on 11 Latin American countries, (not all of which are studied by Panizza and Qiang), finding
a positive premium for all of them, that increased over the study period. Differences between these
two studies can be attributed to the narrower nature of the data used by Panizza and Qiang (2005),
which consists of a single-year set for each country. A review of empirical wage gap studies finds the
premium to be negative in eastern European economies that transitioned from communism to capitalism during the 1990s, although it became less negative with time, and a zero to positive premium
in developed countries (Lausev 2014).
The public–private wage gap was found to decrease with wage distribution: high, positive gaps
for lower payed workers and small positive, or even negative, gaps for higher paid workers (Azam and
Prakash 2015; Cai and Liu 2011; Christofides and Michael 2013; Depalo and Giordano 2011; Foley and
O’Callaghan 2011; Mizala, Romaguera, and Gallegos 2011; Papapetrou 2006; Poterba and Rueben 1994;
Saha, Roy, and Kar 2014). However, two papers show a different result: Maczulskij (2008) finds that
Finnish private sector males at the lower end of the income distribution enjoyed higher premiums, in
relation to several factors, such as education, which decreased with income deciles. Voinea and Mihaescu
(2012) find that in Romania, the public sector wage premium increases across wage percentiles, then
decreases for the top few percentiles of workers.
An accurate understanding of public sector salaries vis-à-vis the private sector and how they vary
across the wage distribution can greatly assist in setting correct wage and employment policies. A
sizeable public sector wage premium has been observed to lead to aberrant labour market behaviour,
as individuals might prefer to ‘queue, or wait in unemployment’ for a stable public sector job with an
appealing pension plan, and eschew low-paid and/or uncertain jobs in the private sector. This tendency
has immediate relevance to Palestine, where a growing number of young people are entering the pool
of the unemployed.
The aim of this paper is to estimate the public–private wage differential in the West Bank and
Gaza Strip, and to describe its dynamics between 1998 and 2006 using labour force surveys from the
Palestinian Central Bureau of Statistics (PCBS). It is important to discover how much of the wage variation can be explained by differences in individual characteristics in the two sectors, and how much by
differences in the returns to these characteristics across sectors.
The econometric results are interesting for a variety of reasons: this is the first serious assessment
of public–private wage differentials in Palestine, and the data are rich enough to allow a dynamic
analysis of the evolution of differentials over time and by geographical area. The results show that in
the pre-Intifada period the wage gap between the public and private sectors narrowed in both the
West Bank and Gaza. However, a sharp increase was seen after the outbreak of the Intifada. Moreover,
most of the increase in the wage gap comes from an increase in the ‘returns’ to skill composition in
the public sector, (unexplained effect), rather than a change in the skills composition of public sector
workers, (explained effect). These findings have implications for the incentives presented to a relatively
youthful population and a rapidly growing work force, when deciding whether to aim for a job in the
public or in the private sector.
Because of the possibility that the distribution of salaries in the public sector may differ from that in
the private sector due to compression, focusing exclusively on the mean salary levels in the two sectors can be misleading. Instead, using recently developed regression techniques, it has been possible
4
S. H. MIAARI
to compare the wage differential at various points along the distribution of salaries. This comparison
provides a much richer description of the wage differential along the salary scale, and allows testing
for whether there is a decrease or an increase in the differential at upper or lower levels of income,
respectively.
The wage premium, (penalty), for the public sector varies across the distribution, being higher,
(lower), at the lowest end of the wage distribution and decreasing, (increasing), along it; it becomes
negative in the top percentiles. Over time, the lower quantiles of West Bank public sector wage earners
have continued to earn a significant (log) wage premium, which increased over time, particularly after
the outbreak of the Intifada. On the other hand, those in the very top income percentile, (95th), continue
to face a wage penalty that has attenuated over time. In Gaza, the wage premium has increased over
time, (especially after the outbreak of the Intifada), for both low and high wage earners in the public
sector. Comparing 2000 to 2006, the estimates indicate that there was a steady increase in the wage
premium for workers in the West Bank and Gaza in all percentiles.
Given the inseparability, and thus complexity, of the political and economic challenges facing the
PA, and the need for objective analysis to inform policy, this paper offers an interpretation of the wage
gap phenomenon in terms of the changing demographic composition of the workforce over the period
under study. During this period, the public sector was perceived as a ‘buffer’ to absorb private sector job
losses, resulting from the impact, on product and labour markets, of increased closures and movement
restrictions imposed by Israel following the second Intifada. However, these results have important
policy implications, and could explain whether or not public sector workers are underpaid, why they
are reluctant to leave their jobs, and why the PA finds it consistently difficult to fill, and retain the staff
in, top-level administrative and managerial positions.
The rest of this paper is organized as follows: the next section presents features of the data available on the Palestinian labour market. Methodology section describes the methodology, econometric
specification and techniques used to measure wage gaps. The main results of the study are reported
in Results section. Estimating the wage gap at various percentiles section provides further evidence
on the wage gap using quantile regressions. The conclusion summarizes the findings and caveats.
Data
This paper’s data are drawn from the Palestinian Labour Force Survey (PLFS) of the West Bank and
Gaza Strip, which is administered by the Palestinian Central Bureau of Statistics (PCBS). The PLFS was
established in 1995, following the signing of the Oslo Accords and the creation of the Palestinian
Authority (PA). In the PLFS, the same household is surveyed four times over six quarters. Two surveys
are conducted during two consecutive quarters, and then two more are conducted after a break of two
quarters, after which the household is dropped from the sample. From 1998 each yearly survey round
contains approximately 7600 households, containing 22,000 individuals aged 15 years and above,
residing in the West Bank or Gaza. Nomads and persons living in institutions such as prisons or shelters
are not included in the survey.
For the purpose of analyzing the public–private sector wage gap, the sample is restricted to male
wage earning employees, working in the domestic, public and private sectors of the West Bank and
Gaza, aged between 18 and 64, and reporting positive net hourly wages and positive days worked per
month in the 1998 and 2006 surveys.4 This excludes Palestinian workers employed in the Israeli labour
market.5 Palestinian women are excluded because their labour force participation rates have traditionally been low.6 Workers’ hourly wages are calculated by dividing daily income by hours worked per
day.7 Because a ‘simple average’ can be strongly affected by large or extreme values, outliers in terms
of hourly wage are dealt with by excluding observations below the 1st and above the 99th percentile
of the log hourly wage distribution for each year.8
Rounds of the survey prior to 1998 are not considered because in 1995 the survey was conducted
in one quarter only, and it was an experimental sample. In 1996, the survey was conducted over three
DEFENCE AND PEACE ECONOMICS
5
quarters. It was not until 1998, after the Palestinian census of 1997, that the survey was conducted in
all four quarters of the year.
The Local Palestinian Labor Market after the Intifada
The share of West Bank male workers employed in the private sector, as a proportion of all employees in
the West Bank labour market, decreased sharply between 1998 and 2006. This share fell by 10 percentage
points, from a high of 67% in 1998 to 57% in 2006. Note the sharp drop in the share of people employed
in the private sector, out of all employed individuals, in 2002 (Figure 2). This is a consequence of the
second Intifada, which began in September 2000, and the onerous system of checkpoints imposed in
the West Bank in 2002. In 1998, 45% of Gaza’s male employees worked in the private sector; by 2006,
this proportion had dipped to 30% (Figure 3).9 As the share of workers in the private sector declined
over the period 1998–2006, the role of the public sector, in absorbing a growing Palestinian labour
force, grew, especially following the second Intifada. In 1998, 30% of West Bank male employees, and
48% of Gazan male employees were working in the public sector; by 2006 the proportions had reached
39% for West Bank males and 60% for Gazan males (Figures 2 and 3). It should be noted that the overall
size of the Palestinian public sector workforce, (as opposed to the percentage of workers employed by
the public sector), is not very large by international standards: based on 2013 World Bank estimates,
Figure 2. share of West Bank male employees across various sectors, 1998–2006. source: author calculations using Palestinian Labor
force surveys (PLfs).
Figure 3. share of Gaza male employees across various sectors, 1998–2006. source: author calculations using Palestinian Labor
force surveys (PLfs).
6
S. H. MIAARI
(which include civil servants and the security services, but which exclude employees of state enterprises, autonomous public bodies not financed by the national government’s budget, and municipal
government employees), public sector employment in the West Bank and Gaza comprised just 4.6%
of the Palestinian population, while the average for the 75 developed and developing countries in the
World Bank’s database was about 5.4%10 (World Bank 2016).
Two factors contributed to the growth of the Palestinian public sector: first, the external and internal closure regimes, imposed by the Israeli authorities, restricted the mobility of Palestinians, leading
to a decline in the number of Palestinians employed in the Israeli labour market, (thus increasing the
supply of employees to the local private sector), and to a fall in local economic activity in the West
Bank and Gaza, (and thus lower demand for employees in the local private sector).11 Consequently,
the share of employees in the local (Palestinian) private sector decreased (Figures 2 and 3).12 As a
result of the closure regime, the unemployment rate in the West Bank increased from 11% in 1998 to
28% in 2002, and in Gaza from 20% in 1998 to 38% in 2002. Given this increase the PA felt compelled
to absorb a large number of the unemployed―despite the unfavourable economic circumstances
―in order to minimize the negative effect of unemployment. Second, faced with political instability,
the PA sought to control both individuals and security in general, and this led it to employ a greater
number of workers in its military and law enforcement services/agencies. One might expect that the
monopsonistic power of the PA, derived from the decline of the private sector, coupled with possible
reductions in revenue, would translate into lower wages in the public sector. As we see throughout
this paper, the opposite is true; public sector wages have risen. It is possible that this increase was, in
part, a response to political pressure on the PA to provide ‘pane e circus’, (‘bread and entertainment’),
to keep the population satisfied during a period of upheaval. The fact that much of the growth, both
in jobs and wages, was in the security sector (World Bank 2006b), required to maintain public order,
strengthens the hypothesis that the increase was a response to political pressures on the PA to keep
the population in order.
Descriptive Statistics
Table 1 summarizes the mean values of the labour force attributes observed in the period 1998–2006
for West Bank male workers in the public and private sectors, respectively. The labour force profile of
workers in the samples changed over the study period. There was a small increase in the average age
of private sector workers and their average years of schooling. However, for public sector workers, the
average years of schooling remained unchanged over the same period, with only a slight increase in
average age. Comparing 2006 to 2000, there was an increase in the proportion of married workers, and
in average tenure in both the public and private sectors. Compared with private sector employees, the
figures in Table 1 also show that public sector employees were on average better educated, older, more
likely to be married and more likely to be tenured.
However, the trend of real hourly wages differed considerably between the public and private sectors
over the period in question. In the public sector, real hourly wages increased over the period 1998–2000,
in 2001 they decreased, and from 2002 they again increased as a result of legislation and administrative
decisions (World Bank 2006b). On the other hand, in the private sector there was a decrease in salaries
over the period 2001–2006, as a result of a sharp decrease in the number of Palestinians employed in
the Israeli labour market and low demand for employees in the local private sector.13
A comparison between real hourly wages in the public and private sectors reveals that the unadjusted wage gap between the two sectors decreased between the years prior to the second Intifada,
from 3.4% in 1998 to −4% in 2000. However, it increased after the beginning of the second Intifada,
from −1.7% in 2001 to 28% in 2006.
Table 2 describes the mean characteristics of the sample of male workers in Gaza. The labour force
profile of males in Gaza changes over the period with a slight increase in the average age of private sector
workers and in their average years of schooling. For public sector workers, these averages remained
Table 1. summary statistics of variables by sector: West Bank males, 1998–2006.
Hourly Wage
Year
1998
1999
2000
2001
2002
2003
2004
2005
Total
Private
6.336
(3.50)
6.809
(3.49)
7.205
(3.65)
6.604
(3.15)
6.608
(3.96)
6.028
(3.61)
5.753
(3.64)
5.520
(3.44)
5.588
(3.69)
6.315
(3.60)
Public
12.399
(3.90)
12.499
(3.91)
12.466
(3.87)
12.420
(3.74)
12.605
(3.80)
12.755
(3.73)
12.627
(3.63)
12.442
(3.59)
12.331
(3.67)
12.492
(3.76)
Private
9.581
(3.51)
9.598
(3.51)
9.719
(3.54)
9.858
(3.38)
10.144
(3.55)
10.058
(3.55)
10.157
(3.57)
10.069
(3.45)
10.133
(3.46)
9.877
(3.51)
Age
Public
35.052
(10.87)
35.289
(10.37)
34.297
(10.95)
35.424
(10.87)
37.764
(11.11)
37.625
(10.93)
37.155
(11.19)
35.828
(10.74)
35.782
(10.69)
35.878
(10.90)
Private
29.996
(9.83)
30.279
(9.81)
30.473
(9.92)
30.809
(9.68)
31.103
(9.84)
31.698
(10.31)
31.546
(10.15)
31.902
(10.20)
32.067
(10.16)
30.994
(10.00)
Married
Public
0.695
(0.46)
0.734
(0.44)
0.664
(0.47)
0.695
(0.46)
0.753
(0.43)
0.749
(0.43)
0.745
(0.44)
0.733
(0.44)
0.708
(0.45)
0.716
(0.45)
Private
0.585
(0.49)
0.597
(0.49)
0.585
(0.49)
0.595
(0.49)
0.588
(0.49)
0.589
(0.49)
0.592
(0.49)
0.619
(0.49)
0.611
(0.49)
0.596
(0.49)
Tenure
Public
80.677
(96.80)
139.394
(166.50)
81.462
(92.16)
85.572
(91.19)
99.474
(98.74)
98.574
(94.64)
98.048
(92.52)
92.021
(85.24)
95.003
(82.85)
96.050
(103.81)
Private
58.937
(78.78)
116.169
(166.71)
69.454
(77.75)
67.891
(75.76)
69.916
(78.61)
70.825
(82.06)
70.301
(81.02)
69.595
(78.45)
70.668
(82.81)
74.664
(97.96)
City
Public
0.378
(0.48)
0.401
(0.49)
0.401
(0.49)
0.376
(0.48)
0.411
(0.49)
0.355
(0.48)
0.362
(0.48)
0.344
(0.48)
0.344
(0.48)
0.374
(0.48)
Private
0.437
(0.50)
0.404
(0.49)
0.414
(0.49)
0.430
(0.50)
0.426
(0.49)
0.416
(0.49)
0.458
(0.50)
0.436
(0.50)
0.422
(0.49)
0.426
(0.49)
Village
Public
0.543
(0.50)
0.447
(0.50)
0.412
(0.49)
0.430
(0.50)
0.437
(0.50)
0.515
(0.50)
0.507
(0.50)
0.508
(0.50)
0.525
(0.50)
0.480
(0.50)
Private
0.490
(0.50)
0.451
(0.50)
0.449
(0.50)
0.432
(0.50)
0.450
(0.50)
0.447
(0.50)
0.410
(0.49)
0.428
(0.49)
0.454
(0.50)
0.448
(0.50)
Full-time
employment
Public
0.908
(0.29)
0.919
(0.27)
0.889
(0.31)
0.954
(0.21)
0.907
(0.29)
0.910
(0.29)
0.936
(0.24)
0.927
(0.26)
0.926
(0.26)
0.919
(0.27)
Private
0.894
(0.31)
0.901
(0.30)
0.882
(0.32)
0.899
(0.30)
0.813
(0.39)
0.862
(0.35)
0.883
(0.32)
0.849
(0.36)
0.786
(0.41)
0.867
(0.34)
N
Public
2098
Private
4385
2208
4393
2527
4237
2112
2954
1495
1925
1791
2497
1873
2554
2357
3510
2338
3466
18,799
29,921
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs), 1998–2006.
note: The sample includes West Bank salaried, prime-aged (18–65) males. The income variable is the real hourly wage in constant 1996 new israeli shekels (₪). in 1996, 1₪.00 equalled approximately
$Us 0.33. The married variable takes on the value 1 if the person is married, and 0 otherwise. The Tenure variable is in months. standard deviations in parentheses.
DEFENCE AND PEACE ECONOMICS
2006
Public
6.555
(3.41)
6.967
(3.40)
6.913
(3.33)
6.490
(2.77)
6.681
(3.38)
6.643
(3.42)
7.093
(3.54)
6.929
(3.33)
7.207
(3.60)
6.844
(3.37)
Schooling
7
8
S. H. MIAARI
Table 2. summary statistics of variables by sector: Gaza males, 1998–2006.
Hourly Wage
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
Total
Public
6.474
(3.44)
6.586
(3.12)
6.780
(3.12)
6.742
(2.63)
6.771
(2.97)
6.520
(3.22)
6.733
(3.23)
7.592
(3.46)
8.370
(3.70)
6.990
(3.29)
Private
3.737
(1.99)
4.009
(2.35)
4.353
(2.69)
4.600
(2.49)
4.294
(3.09)
4.006
(2.28)
3.973
(2.64)
3.659
(2.59)
3.566
(2.53)
3.982
(2.51)
Schooling
Public
12.564
(3.72)
12.197
(3.74)
12.416
(3.60)
12.486
(3.50)
12.587
(3.53)
12.414
(3.78)
12.554
(3.71)
12.464
(3.64)
12.474
(3.53)
12.456
(3.64)
Private
8.894
(3.64)
9.435
(3.60)
9.439
(3.88)
10.046
(3.80)
9.946
(3.77)
9.513
(3.67)
9.948
(3.81)
9.725
(3.63)
9.996
(3.56)
9.597
(3.71)
Age
Public
34.517
(11.03)
35.219
(10.84)
35.135
(10.58)
34.917
(10.37)
36.178
(10.37)
36.291
(10.84)
36.214
(10.75)
35.936
(10.15)
35.390
(10.14)
35.537
(10.56)
Private
29.364
(9.59)
28.997
(9.02)
30.046
(10.02)
31.145
(9.69)
31.068
(9.89)
31.854
(9.35)
32.362
(9.76)
32.367
(9.90)
33.185
(10.39)
31.001
(9.81)
Married
Public
0.755
(0.43)
0.830
(0.38)
0.808
(0.39)
0.784
(0.41)
0.825
(0.38)
0.800
(0.40)
0.781
(0.41)
0.820
(0.38)
0.807
(0.39)
0.802
(0.40)
Private
0.683
(0.47)
0.636
(0.48)
0.640
(0.48)
0.698
(0.46)
0.684
(0.47)
0.722
(0.45)
0.708
(0.45)
0.702
(0.46)
0.724
(0.45)
0.685
(0.46)
Tenure
Public
62.604
(79.68)
148.098
(191.51)
62.169
(59.91)
64.196
(58.55)
71.197
(68.22)
75.157
(59.33)
80.825
(64.26)
84.698
(60.69)
87.886
(65.27)
82.506
(91.93)
Private
38.635
(49.34)
94.277
(151.05)
44.866
(57.98)
49.934
(60.06)
42.714
(52.33)
44.426
(48.52)
47.424
(62.32)
39.389
(56.85)
37.895
(55.60)
50.481
(80.00)
City
Public
0.436
(0.50)
0.533
(0.50)
0.547
(0.50)
0.553
(0.50)
0.542
(0.50)
0.546
(0.50)
0.520
(0.50)
0.502
(0.50)
0.494
(0.50)
0.520
(0.50)
Private
0.558
(0.50)
0.613
(0.49)
0.590
(0.49)
0.593
(0.49)
0.614
(0.49)
0.603
(0.49)
0.599
(0.49)
0.614
(0.49)
0.575
(0.49)
0.595
(0.49)
Village
Public
0.144
(0.35)
0.088
(0.28)
0.067
(0.25)
0.064
(0.25)
0.068
(0.25)
0.065
(0.25)
0.077
(0.27)
0.086
(0.28)
0.099
(0.30)
0.083
(0.28)
Private
0.143
(0.35)
0.089
(0.28)
0.094
(0.29)
0.089
(0.28)
0.090
(0.29)
0.110
(0.31)
0.096
(0.29)
0.121
(0.33)
0.137
(0.34)
0.109
(0.31)
Full-time
employment
Public
0.891
(0.31)
0.903
(0.30)
0.878
(0.33)
0.874
(0.33)
0.875
(0.33)
0.868
(0.34)
0.890
(0.31)
0.875
(0.33)
0.862
(0.35)
0.879
(0.33)
Private
0.947
(0.22)
0.941
(0.24)
0.894
(0.31)
0.870
(0.34)
0.850
(0.36)
0.854
(0.35)
0.925
(0.26)
0.926
(0.26)
0.861
(0.35)
0.903
(0.30)
N
Public
1781
Private
1513
2296
1762
2354
1428
2241
745
1903
849
2089
1304
2186
1141
2454
1418
2450
1193
19,754
11,353
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs), 1998–2006.
note: The sample includes Gaza salaried, prime-aged (18–65) males. The income variable is the real hourly-wage in constant 1996 new israeli shekels (₪). in 1996, 1₪.00 equalled approximately $Us
0.33. The married variable takes on the value 1 if the person is married and 0 otherwise. The Tenure variable is in months. standard deviations in parentheses.
DEFENCE AND PEACE ECONOMICS
9
unchanged over the same period. The proportion of married workers increased in both sectors. A significant increase in the average tenure of public sector employees is observed, while there is no clear
pattern among private sector employees. The figures in Table 2 also show that public sector employees
are on average better educated, older, more likely to be married and more likely to be tenured. The
unadjusted wage gap between the public and private sectors decreased between 1998 and 2000, from
73% in 1998 to 55% in 2000. However, it increased from 46% in 2001 to 134% in 2006.
It is evident from Tables 1 and 2 that over the sample period real wages in the private sector remained
below those in the public sector. The only exceptions to this are the years 2000 and 2001 in the West
Bank male sample, where the wage gap was in favour of the private sector. Moreover, the decrease
in the real hourly wage in the West Bank was faster than that in Gaza, because internal and external
closures imposed in the West Bank, after the beginning of the second Intifada, were more widespread
than those in Gaza.14
Methodology
Basic Specification of Model
A convenient starting point for estimating the magnitude of the public–private wage gap is to use
ordinary least squares with a dummy variable for public sector participation on a pooled sample of
workers. In this approach, an individual i has (log) real hourly wage ln Wi, conditional on observed
characteristics Xi and a dummy variable Di that takes the value of 0 or 1 depending on whether the
individual works in the private sector or public sector. Adding an error term εi distributed with a mean
of zero leads to the least squares (OLS) specification:
LnWi = X � i 𝛽 + Di 𝛿 + 𝜀i ,
(1)
where β is a vector of unknown parameters whose estimates would provide the influence or ‘returns’ of
the observed qualitative variables Xi on lnWi, and δ is the unknown parameter whose estimates provide
the ceteris paribus impact of working in the public sector.
While simple and intuitive, the foregoing approach is problematic for several reasons. The OLS estimate of δ captures only a pure ‘shift’ effect of working in the public sector, and ignores the fact that
salaries could well differ because of differences in observed characteristics such as education and age
across the two sectors. The foregoing OLS estimation assumes that error terms are homoskedastic and
identically distributed across individuals. This may not be an appropriate assumption, especially when
dealing with survey samples and two quite distinct population groups.
Secondly, the pooled OLS dummy variable specification assumes that the earnings are distributed
identically across both the private and public sectors. To the extent that the public sector compresses
the distribution of earnings of those employees who work in that sector relative to the private sector,
the least squares estimates are likely to be biased and produce an incomplete picture of the conditional
distribution of lnWi (see Disney and Gosling 1998; Nielsen and Rosholm 2001).
Finally, the least squares procedure, as specified above, does not control for endogenous selectivity
bias; that is, the distribution of workers between the public and private sectors may not be completely
random in the West Bank and Gaza.
The approach employed to correct for these problems was to adopt the Oaxaca–Blinder decomposition method with modifications: (a) for decomposing the wage gap according to observed characteristics; and (b) implementing a correction for endogenous selectivity bias. Moreover, a quantile
regression framework that examines the wage gap at various points along the wage distributions is
appropriate to accommodate differing distributions of wages between the two sectors. It is important
to correct for these in any econometric specification within the context of Palestine, where the public
sector continues to attract workers in an environment where the private sector is buffeted by considerable uncertainty and exogenous shocks.
10
S. H. MIAARI
Decomposing the Wage Gap within the OLS Framework
While the preceding section calculates the wage premium/penalty for working in the public sector, it
does not make it possible to control for the fact that individual attributes or their returns might vary
across workers and sectors. Thus, the dummy variable estimates of the preceding section show the
‘shift’ or ceteris paribus effect of working in the public sector, and provide no information as to whether
the observed differentials are due to differences in attributes or to differences in the returns to these
attributes. The focus of this section is on decomposing the observed wage differential in order to better
understand how much of it is caused by differences in the distribution of attributes, and how much is
due to differences in the returns to these attributes.
Within the OLS framework, a convenient way of decomposing observed pay gaps is to run separate regressions for each sector. Letting the subscript j (j = 1, 0) denote the public and private sectors,
respectively, and i individuals, the following regression specification is estimated for each sector:
LnWi = X � ij 𝛽j + 𝜀ij
(2)
where Wij is the hourly wage, Xij is a vector of worker characteristics, and εij a zero-mean constantvariance error term. Then, the wage equations, estimated by OLS at the mean point, will be:
LnWj = X � j 𝛽̂j
for
(3)
j = 1, 0
where 𝛽̂j is the OLS estimate of the marginal effects, or returns of observed characteristics Xj, on salaries,
and Xj is the mean level of observed characteristics across sector j. The regressors vector x ′ ij includes age,
age squared, years of schooling, tenure, (total months in the same workplace), marital status, (dummy
which takes value 1 if the worker is married and zero otherwise), full-time employment, (dummy which
takes value 1 if the worker works at least 35 h a week and zero otherwise); set of location of residence
dummies, (urban area and refugee camp), occupational dummies and district fixed effects.15 The average public–private gross wage gap ln(1 + G) is the difference between the average salaries in the two
sectors.16:
ln W 1 − ln W 0 = ln(1 + G) = (X̄ 1 − X̄ 0 )� 𝛽̂∗ + X̄ 1� (𝛽̂1 − 𝛽̂∗ ) + X̄ 0� (𝛽̂∗ − 𝛽̂0 ),
����������� �����������������������������������
E
(4)
R
where 𝛽̂∗ is the estimate of the non-discriminatory wage coefficients. Equation (4) is the general Oaxaca
decomposition, as per Oaxaca and Ransom (1994). It disentangles the average gross wage differential ln(1 + G) across the two sectors into two terms: the first term, E, is the explained component of
the overall wage gap: the differential due to differences in the distribution of average characteristics,
(the endowment effect). The second, R, is the unexplained component of the overall wage gap: the
differences in the estimated coefficients or ‘returns’ between the two sectors (the ‘returns’ effect). If we
assume that 𝛽̂∗ = 𝛽̂1, then the general Oaxaca decomposition reduces to the classical Oaxaca-Blinder
(1973) decomposition. In this case, the first term will be evaluated at the returns in the public sector,
and the second at the mean set of private sector characteristics.
The study was conducted under two differing assumptions regarding the non-discriminatory wage
coefficients𝛽̂∗. First, the estimated wage structure of the public sector was adopted as the non-discriminatory standard, i.e. 𝛽̂∗ = 𝛽̂1 as per Blinder (1973). These are the main results, reported in Tables 4–7.
Second, 𝛽̂∗ is assumed to be equal to the estimated wage coefficients from a pooled regression that
includes both public and private sectors, as per Neumark (1988), and Oaxaca and Ransom (1994).17
These results are reported in the Appendix 1. Moreover, since the decomposition results for categorical predictors depend on the choice of the omitted ‘base’ category (Jones 1983; Jones and Kelley
1984; Oaxaca and Ransom 1999; Nielsen 2000; Horrace and Oaxaca 2001; Gardeazabal and Ugidos
2004; Polavieja 2005; Yun ), we estimate the model with the conventional dummy arrangement, but
transform the coefficient vectors so the usually omitted coefficient for the base category is included
DEFENCE AND PEACE ECONOMICS
11
and deviations from the grand mean are expressed. That keeps the result independent of the choice
of base category (Suits 1984; Yun 2005).
Selectivity-corrected Wage Gap Decomposition
Estimates of wage gaps are potentially afflicted by sample selection bias arising from self-selection into
sector (Heckman 1979). This section explores the possible impact of this source of bias on the estimate
of the public–private wage differential. In order to obtain selectivity-corrected decompositions, the
selectivity effect as a whole was calculated. The corrected gross wage gap was then decomposed into
an endowment effect, a return effect, and a selectivity effect. In other words, the decomposition in
Equation 4 is generalized as follows18:
ln(1 + G) = (X̄ 1 − X̄ 0 )� 𝛽̂∗ + X̄ 1� (𝛽̂1 − 𝛽̂∗ ) + X̄ 0� (𝛽̂∗ − 𝛽̂0 ) + (𝜃̂1 𝜆̂ 1 − 𝜃̂0 𝜆̂ 0 ),
����������� ����������������������������������� �������������
E
R
(5)
Selection
where 𝜃̂ is the coefficient of the Inverse Mills Ratio (𝜆̂) in the modified wage equation. The selectivity-corrected wage equations are estimated by the Heckman two-step procedure.
The probability of working in the public or private sector depends on the individual’s profile and a
number of factors may constitute the costs and benefits of employment in a particular sector, such as
job security and working environment. Thus, the explanatory variables in the selection equation are:
age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee
camp residence, occupational dummies, district fixed effects, number of jobholders in the same household, number of adults aged 70 or more in the household and share of public sector employees as a
proportion of total labour force in locality. The number of jobholders (Lokshin and Javanovic 2003),
number of adults aged 70 or more in a given household (Aslam and Kingdon 2009) and share of public
sector employees (Voinea and Mihaescu 2012) are the additional variables included only in the selection
equation. These variables affect a worker’s decision to seek a secure job, in either the public or private
sector, yet do not affect his or her wage. That is, they account for the importance of a secure job and
its associated benefits in sector choice.19
Results
Table 3 summarizes the OLS results from estimating Equation (1) for each year of the PCBS labour force
survey in two specifications both for the West Bank and Gaza. Specifications 1 and 3 include only the
public dummy variable; the reported coefficient therefore measures the overall unadjusted (logarithmic)
wage gap. In Specifications 2 and 4, in addition to the public dummy variable, all of the explanatory
variables, as described in the previous section, are included, hence the reported coefficient measures
the adjusted wage gap.
For most years in the West Bank and Gaza Strip, the estimated coefficients of δ are highly significant
at the 95% confidence level, and increase over time. It is apparent from Table 3 that introducing the
productivity-related variables into the wage equations greatly reduces the measured wage gap. This
fact highlights the magnitude of the explained components of the wage gap.
The estimates for West Bank male workers suggest that in the earlier phases of the second Intifada,
public sector male workers in the West Bank faced a wage penalty vis-à-vis their private sector counterparts, (controlling for differences in attributes or ceteris paribus), yet this effect disappears, then eventually reverses over time. From the beginning of the second Intifada, the PA systematically addressed
the issue of a negative public sector wage gap through salary increases, such that by 2003 there is
considerable evidence of an increase in the wage gap. In 1999, the public–private log wage differential was estimated at –0.10. However, by 2003 a positive public sector pay premium emerged. In
2005, the public sector pay premium rose significantly as another salary increase was implemented,
(column 2). This increase may represent an effort to appeal to public sector workers in the run-up to
12
S. H. MIAARI
Table 3. estimated unadjusted and adjusted wage differential.
West Bank males
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
(1)
0.0648***
(0.0128)
0.0385***
(0.0123)
−0.0156
(0.0117)
0.0137
(0.0123)
0.0760***
(0.0183)
0.1487***
(0.0159)
0.2764***
(0.0154)
0.2870***
(0.0133)
0.3310***
(0.0141)
Gaza Males
(2)
−0.0544***
(0.0137)
−0.1028***
(0.0139)
−0.0965***
(0.0128)
−0.0341**
(0.0135)
−0.0296
(0.0201)
0.0207
(0.0164)
0.0883***
(0.0148)
0.1252***
(0.0133)
0.1944***
(0.0154)
(3)
0.5462***
(0.0156)
0.5292***
(0.0140)
0.4906***
(0.0156)
0.4332***
(0.0186)
0.5410***
(0.0205)
0.5049***
(0.0154)
0.5837***
(0.0182)
0.7904***
(0.0158)
0.9182***
(0.0174)
(4)
0.2655***
(0.0187)
0.2269***
(0.0187)
0.2380***
(0.0173)
0.2367***
(0.0176)
0.2742***
(0.0233)
0.2604***
(0.0187)
0.3005***
(0.0196)
0.4266***
(0.0180)
0.5515***
(0.0210)
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: samples include salaried, prime-aged (18–65) males. The reported gap is measured by the coefficient of sector dummy variables (that take on the value 1 if the worker is employed in the public sector, and 0 if employed in the private sector) in a pooled
wage regression that includes both public and private sector employees. The dependent variable in all specifications is the log of
hourly wage. specifications 1 and 3 include only public dummy variable. specifications 2 and 4 include public dummy variable,
age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, and district fixed effects. Robust standard errors are in parentheses. The symbols *, **, *** represent statistical
significance at the 10, 5, and 1 percent levels.
the national elections, held in January 2006. Among male workers in the Gaza Strip, the public sector
pay premium declined between the years 1998 and 1999, while the period 2000–2006 witnessed an
increase (column 4).
Alongside the fiscal burden of providing the public sector with wage premiums in relation to the
private sector and the equity issues it raises, this trend is also worrying for other reasons. Within an
environment of high unemployment, particularly among the young, the trend creates an incentive for
job seekers to ‘wait’ for a secure job in the public sector, (or engage in civil strife to acquire it), rather
than to strive to find employment in a highly fragmented and uncertain private sector. This may be
described as rent seeking.
Although the simple empirical exercise can be revealing, the objective is to decompose the observed
wage differential into two components: one is due to differences in individual worker characteristics,
and the other is due to differences in returns to those characteristics. The econometric results for the
Oaxaca decomposition are shown in Table 4, under the assumption that 𝛽̂∗ = 𝛽̂1. The analyses of Tables
5–7 are also performed under that assumption. For comparison, Tables A1–A4 report analyses made
under the assumption that 𝛽̂∗ = 𝛽̂pooled, i.e. 𝛽̂∗ is equal to the estimated wage coefficients from a pooled
regression that includes both public and private sectors. The results under the two assumptions are
reported for the sake of completeness, although the description of the data findings will be confined
to the first assumption.
The figures in Table 4, column (G), indicate the mean (log) wage differential between the public and
private sectors. It is the sum or aggregate of the endowment effect (E) and the returns effect (R). An
interesting pattern emerges among West Bank male workers: during the period before the Intifada,
between 1998 and 2000, the gross wage gap decreased. This decrease is explained by the returns effect
decreasing during this period. During that time, the endowment effect increased. After the outbreak
of the Intifada, the period between 2001 and 2006, the gross wage gap increased. During that time,
both the returns effect and the endowment effect increased, with the returns effect responsible for the
DEFENCE AND PEACE ECONOMICS
13
Table 4. Decomposing the wage differential over time: β* = βpublic.
West Bank males
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.0664***
(0.0129)
0.0336***
(0.0129)
0.0277**
(0.0120)
0.0318**
(0.0125)
0.0868***
(0.0187)
0.1490***
(0.0160)
0.2776***
(0.0155)
0.2884***
(0.0134)
0.3296***
(0.0142)
E
0.1405***
(0.0165)
0.1772***
(0.0174)
0.1878***
(0.0153)
0.1250***
(0.0153)
0.1502***
(0.0220)
0.1725***
(0.0204)
0.2118***
(0.0190)
0.1815***
(0.0158)
0.1575***
(0.0197)
Gaza males
R
−0.0741***
(0.0178)
−0.1436***
(0.0182)
−0.1602***
(0.0161)
−0.0932***
(0.0168)
−0.0634***
(0.0241)
−0.0235
(0.0220)
0.0658***
(0.0202)
0.1069***
(0.0171)
0.1721***
(0.0214)
G
0.5483***
(0.0156)
0.5248***
(0.0142)
0.4911***
(0.0156)
0.4333***
(0.0188)
0.5410***
(0.0207)
0.5049***
(0.0155)
0.5832***
(0.0183)
0.7902***
(0.0158)
0.9182***
(0.0175)
E
0.2587***
(0.0191)
0.2768***
(0.0186)
0.2493***
(0.0160)
0.1423***
(0.0146)
0.1881***
(0.0177)
0.2152***
(0.0180)
0.2244***
(0.0192)
0.3445***
(0.0154)
0.3638***
(0.0169)
R
0.2896***
(0.0203)
0.2479***
(0.0202)
0.2418***
(0.0176)
0.2910***
(0.0183)
0.3529***
(0.0222)
0.2897***
(0.0200)
0.3587***
(0.0210)
0.4457***
(0.0164)
0.5545***
(0.0194)
note: The dependent variable in the wage equations is the logarithm of hourly wage; the independent variables are age, age
squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, and district fixed effects. G refers to the gross wage gap (or ln(1+G)), e refers to the explained component of the wage gap
(endowment effect), and R refers to the unexplained (return effect) component of the wage gap. Robust standard errors are in
parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 percent levels.
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs), 19982006.
majority of the wage gap increase. In other words, the positive endowment impact in the public sector
is offset by the large negative return effect in the early years, resulting in a negative wage differential
or an overall wage penalty for working in the public sector by the year 2000. However, a rapid rise in
the returns effect over time generates a positive wage gap in favour of the public sector from 2001. In
2006, both the returns effect and the endowment effect are positive and work in the same direction to
produce a substantial wage premium for the public sector, about 48% of which is due to the endowment
effect, and 52% to the so-called return effect. Note that throughout 1998–2006, (with the exception of
2006), the endowment effect constituted the majority of the gross wage gap’s size.
Among male workers in Gaza, the gross wage gap experienced the same general trend as the West
Bank – decreasing in 1998–2000 and increasing in 2001–2006. However, unlike for the West Bank, both
the endowment effect and the returns effect decreased, in the period before the Intifada. As in the
West Bank, after the outbreak of the Intifada, both effects increased. The decline in the endowment
and returns effects in the public sector in the early years of the study, in turn result in a significant
decline in the public–private wage differential. A rapid rise in the returns and endowment effects over
time generates a concurrent increase in the wage gap from 2001. In 2006, both the returns effect and
the endowment effect are positive and work in the same direction to produce a wage premium for
the public sector, about 40% of which is due to the endowment effect and 60% to the returns effect.
Unlike for the West bank, it was the returns effect that constituted the majority of the wage gap over
the entire period, with the exception of 1999–2000.
The results reported so far do not take into account the possible selection bias caused when workers
with different characteristics choose different sectors. Table 5 reports the selectivity-corrected decomposition results. We see that taking into account selectivity correction does not change the general trend
of the gross wage gap, but does affect the trends of its components. The corrected gross wage gap, Ĝ,
still declines in the period before the outbreak of the Intifada, (1998–2000), and increases thereafter,
(2001–2006). There are however some differences. In the West Bank, the gross wage gap decreases in
the period before the outbreak of the Intifada (1998–2000). This is explained by a decrease in the returns
14
S. H. MIAARI
Table 5. selectivity-corrected wage gap decomposition: β* = βpublic.
West Bank males
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.0656***
(0.0129)
0.0291**
(0.0129)
0.0277**
(0.0120)
0.0318**
(0.0125)
0.0868***
(0.0187)
0.1490***
(0.0160)
0.2776***
(0.0155)
0.2884***
(0.0134)
0.3296***
(0.0142)
E
0.1690***
(0.0409)
0.2310***
(0.0497)
0.1747***
(0.0357)
0.1009**
(0.0430)
0.1778***
(0.0520)
0.1513***
(0.0461)
0.2406***
(0.0370)
0.2290***
(0.0385)
0.2066***
(0.0527)
R
−0.1284*
(0.0742)
−0.2703***
(0.0833)
−0.1375**
(0.0623)
−0.0783
(0.0750)
0.0489
(0.0890)
0.0525
(0.0827)
0.0198
(0.0685)
0.0007
(0.0722)
0.0654
(0.0925)
Gaza males
Selection
0.025
Ĝ
0.041
0.068
−0.039
−0.009
0.037
0.009
0.023
−0.140
0.227
−0.055
0.204
0.017
0.260
0.059
0.230
0.058
0.272
G
0.5483***
(0.0156)
0.5248***
(0.0142)
0.4911***
(0.0156)
0.4333***
(0.0188)
0.5410***
(0.0207)
0.5049***
(0.0155)
0.5832***
(0.0183)
0.7902***
(0.0159)
0.9182***
(0.0175)
E
0.4576***
(0.0845)
0.5803***
(0.0707)
0.4926***
(0.0630)
0.3792***
(0.0510)
−0.0305
(0.0575)
0.1822***
(0.0539)
0.1386**
(0.0585)
0.2055***
(0.0429)
0.2136***
(0.0457)
R
−0.0070
(0.1298)
−0.1866*
(0.1041)
−0.1214
(0.1101)
0.3609**
(0.1409)
0.7707***
(0.1351)
0.3956***
(0.0952)
0.5440***
(0.1038)
0.6968***
(0.0779)
0.8617***
(0.1084)
Selection
0.098
Ĝ
0.451
0.131
0.394
0.120
0.371
−0.307
0.740
−0.199
0.740
−0.073
0.578
−0.099
0.683
−0.112
0.902
−0.157
1.075
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. G refers
to the gross wage gap (or ln(1 + G)), e refers to the explained component of the wage gap (endowment effect), R refers to the
unexplained (return effect) component of the wage gap. selection is the component of the wage gap attributed to self-selection
into sector, and Ĝ is the selectivity-corrected gross wage gap. The dependent variable in the wage equations is the logarithm of
hourly wage; the independent variables are age, age squared, years of schooling, tenure, marital status, full-time employment;
urban area/refugee camp residence, occupational dummies, and district fixed effects. The selection equation includes as explanatory variables, age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp
residence, occupational dummies, district fixed effects, number of jobholders, number of adults aged 70 or more in a given household, and locality’s share of public sector employees. Robust standard errors are in parentheses. The symbols *, **, *** represent
statistical significance at the 10, 5, and 1 per cent levels.
effect and in the selection effect. During the same period, the endowment effect increases, but this
effect is much smaller. After the outbreak of the Intifada (2001–2006), the gross wage gap increases. The
majority of this increase is explained by an increase in the returns effect. The endowment effect also
increases during the same period, but this effect is smaller. Throughout this period, the endowment
effect is dominant in explaining the gross wage gap itself, (though not the change).
In Gaza, the gross wage gap also decreased in the years before the outbreak of the Intifada (1998–
2000). During this period, the endowment effect increases, the returns effect decreases and the returns
effect is more dominant in determining the net change in the wage gap. The dominant component
accounting for the gross wage gap itself remained the endowment effect. In the period after the outbreak of the Intifada (2001–2006), the gross wage gap increases. This increase is explained by an increase
in the returns effect. The endowment effect decreased during this period. Except for 2001, the returns
effect was dominant in explaining the gross wage gap itself.
The results show that in the pre-Intifada period, the wage gap between the public and private
sectors narrowed in both the West Bank and Gaza. However, a sharp increase was seen at the end of
the time period, as a result of the increase in public sector wages brought on by implementation of
new civil service laws and government-mandated public wage increases, and a rapid decrease in private sector wages brought on by a sharp decrease in the number of Palestinians working in the Israeli
labour market and low demand for employees in the local private sector.20 This is not surprising given
the disruptive influence the Intifada had on private sector jobs. Moreover, most of the increase in the
wage gap, in both the West Bank and Gaza after the outbreak of the Intifada, stems from an increase in
the ‘returns’ to skill composition in the public sector, (unexplained effect), rather than a change in the
skills composition of public sector workers (explained effect).
DEFENCE AND PEACE ECONOMICS
15
Palestinians working in Israel were generally less skilled, (and lower paid), workers in the private
sector, or workers with fewer years of schooling, relative to the average for the Palestinian workforce
(Miaari and Sauer 2011). Unskilled workers, returning from Israel, compete with locals largely for private
sector unskilled jobs in the Palestinian labour market, often winning them at the expense of unskilled
local workers. Consequently, for wage earners who continued to be employed in the local market, or
who had obtained jobs in the local market, the impact was exacerbated by the decline in average real
private sector wages over much of the Intifada period, meaning a high gross public–private wage gap
among unskilled workers during this period. Further, since the unskilled workers are homogenous, we
would expect the endowment effect of the wage gap to be low. In turn, we would expect the unexplained component of the wage gap to follow the pattern of the overall gross pay gap.
Table 6 and Table 7 document the selectivity-corrected decomposition of the public–private wage
gap for skilled and unskilled workers in the West Bank and Gaza, respectively. For skilled workers in
the West Bank, the gross wage gap decreased between 1998 and 2002, and increased from 2003 to
2006. The selectivity-corrected gap decreased between 1998 and 2000, and increased between 2001
and 2006. The endowment effect for these workers increased in 1998–2000, decreased in 2001 and
increased in 2002–2006. The returns effect decreased in 1998–2000, increased in 2001, decreased again
in 2002–2003 and increased in 2004–2006. The gross wage gap among unskilled workers in the West
Bank decreased, (became more negative), from 1998 and 2000, and increased thereafter, reaching
0.24 in 2006. The selectivity-corrected gap’s trend was similar, although more volatile. The endowment
effect increased between 1998 and 1999, decreased between 2000 and 2001, and increased again from
2002 to 2006, while the returns effect decreased from 1998 to 1999, and increased over the period
2000–2006. The continued increase during the years of the Intifada is related to the entry of a large
number of unskilled workers previously employed in the Israeli labour market, coinciding with a decline
in the local economic activity. An increase in unskilled labour supply, along with a general decrease in
demand, resulted in a decrease in private sector wages unrelated to skills and other explained variables,
thus increasing the returns effect.
In Gaza, the gross wage gap among skilled workers decreased between 1998 and 2001 and increased
over the period 2002–2006. The selectivity-corrected wage gap was more volatile, but generally
decreased from 1998 to 2002, and increased between 2003 and 2006. The return effect followed exactly
the same pattern as the selectivity-corrected wage gap. The endowment effect steadily decreased over
the period 1998–2006. The gross wage gap among unskilled workers in Gaza decreased between 1998
and 2001 and increased between 2002 and 2006. The selectivity-corrected gap decreased between 1998
and 2000, and increased between 2001 and 2006. The endowment effect increased from 1998 to 2001,
decreased in 2002 and increased again from 2003 to 2006. The returns effect decreased from 1998 to
1999, increased between 2000 and 2002, decreased in 2003 and increased again from 2004 to 2006.
As shown in Tables 6 and 7, for most of the period studied, especially the years after the outbreak of
the Intifada, the returns effect is the dominant component of the gross wage gap for unskilled workers
in both the West Bank and Gaza. Furthermore, the returns effect follows almost exactly the same trends
as the selectivity-corrected wage gap for unskilled workers, in both the West Bank and Gaza.21 Both of
these results confirm our expectations regarding the effect on unskilled workers.
Estimating the Wage Gap at Various Percentiles
The simple OLS analysis has thus far focused exclusively on the average level of earnings differentials
between the public and private sectors. It provides little information on the extent to which this differential varies across the wage distribution. Empirical evidence from many countries suggests that the
pay gap, ln(1 + G), varies across the wage spectrum, and focusing on the mean could be misleading.
Here, the dummy variable approach is used in a quantile regression model, where the estimates of β
and δ are computed at various points along the wage distribution. A series of quantile regressions is
estimated, taking the form22:
Q𝜃 (ln wi |xi ) = xi 𝛽𝜃 + di 𝛿𝜃 ,
(6)
16
S. H. MIAARI
Table 6. selectivity-corrected wage gap decomposition by skill group: West Bank, β* = βpublic.
Skilled
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.0653***
(0.0246)
0.0288
(0.0233)
0.0294
(0.0229)
0.0065
(0.0241)
−0.0732**
(0.0342)
−0.0267
(0.0307)
0.0770***
(0.0297)
0.1417***
(0.0280)
0.1946***
(0.0293)
E
0.0763
(0.0620)
0.1283**
(0.0593)
0.1586***
(0.0389)
−0.0156
(0.0626)
0.0230
(0.0620)
0.0175
(0.0844)
0.0355
(0.0443)
0.0583
(0.0476)
0.1047
(0.0753)
R
0.1668
(0.1169)
−0.0340
(0.1260)
−0.1622*
(0.0970)
0.1777
(0.1446)
0.1514
(0.1425)
−0.0003
(0.1717)
0.1840
(0.1275)
0.4427***
(0.1641)
0.2054
(0.1992)
Unskilled
Selection
0.011
Ĝ
−0.085
0.092
−0.136
0.086
−0.067
−0.116
0.084
0.122
−0.247
−0.044
0.012
−0.027
0.032
0.000
0.161
0.135
0.070
G
−0.1022***
(0.0151)
−0.1728***
(0.0152)
−0.1654***
(0.0141)
−0.1116***
(0.0151)
−0.0531**
(0.0221)
0.0146
(0.0188)
0.1408***
(0.0171)
0.1537***
(0.0144)
0.2415***
(0.0162)
E
0.0244
(0.0423)
0.0627
(0.0562)
0.0322
(0.0431)
−0.0051
(0.0474)
0.0494
(0.0654)
0.0703
(0.0452)
0.1393***
(0.0416)
0.1147***
(0.0415)
0.1559***
(0.0597)
R
−0.2219**
(0.0943)
−0.3806***
(0.1076)
−0.2308***
(0.0848)
−0.2060**
(0.0940)
−0.0988
(0.1214)
−0.1104
(0.0919)
−0.1592*
(0.0835)
−0.1286
(0.0821)
−0.0527
(0.1050)
Selection
−0.216
Ĝ
0.097
0.016
−0.210
0.022
−0.214
0.004
−0.116
−0.072
0.021
0.053
−0.018
0.094
0.060
0.069
0.092
0.186
0.051
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs), 1998–2006.
note: a skilled worker is defined as one with more than 12 years of schooling. Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. G
refers to the gross wage gap (or ln(1 + G)), e refers to the explained component of the wage gap (endowment effect), R refers to the unexplained (return effect) component of the wage gap. selection
is the component of the wage gap attributed to self-selection into sector, and Ĝ is the selectivity-corrected gross wage gap. The dependent variable in the wage equations is the logarithm of hourly
wage; the independent variables are age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, and district fixed
effects. The selection equation includes as explanatory variables, age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational
dummies, district fixed effects, number of jobholders, number of adults aged 70 or more in a given household, and locality’s share of public sector employees. Robust standard errors are in parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 per cent levels.
DEFENCE AND PEACE ECONOMICS
17
Table 7. selectivity-corrected Wage Gap Decomposition By skill Group: Gaza, β* = βpublic.
Skilled
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.4620***
(0.0513)
0.4466***
(0.0433)
0.3236***
(0.0430)
0.2357***
(0.0427)
0.2492***
(0.0611)
0.3297***
(0.0486)
0.2091***
(0.0450)
0.4445***
(0.0435)
0.6790***
(0.0474)
E
0.2317**
(0.0954)
0.2588***
(0.0886)
0.2088***
(0.0521)
0.1411***
(0.0420)
0.0983**
(0.0460)
0.1111*
(0.0596)
0.0757**
(0.0331)
0.0984**
(0.0396)
−0.0739
(0.0619)
Unskilled
R
0.6233
(0.4890)
0.3716
(0.3304)
0.8365**
(0.3592)
0.0134
(0.3418)
−0.8875*
(0.5141)
0.3257
(0.3546)
0.5193
(0.3840)
0.6139***
(0.2236)
1.0955***
(0.2273)
Selection
0.183
Ĝ
0.038
0.314
0.042
−0.053
0.377
0.509
−0.294
0.780
−0.567
−0.176
0.510
0.198
−0.013
0.012
0.401
−0.252
0.856
G
0.4094***
(0.0160)
0.4136***
(0.0143)
0.3883***
(0.0157)
0.3737***
(0.0188)
0.4722***
(0.0194)
0.4032***
(0.0153)
0.5188***
(0.0172)
0.7209***
(0.0150)
0.8574***
(0.0176)
E
0.2115**
(0.0988)
0.4284***
(0.0774)
0.3177***
(0.0721)
0.3595***
(0.0618)
−0.2088***
(0.0645)
0.1448***
(0.0551)
0.1817***
(0.0659)
0.2293***
(0.0515)
0.3445***
(0.0580)
R
0.0477
(0.1585)
−0.1747
(0.1153)
−0.1073
(0.1214)
0.1368
(0.1375)
0.7112***
(0.1211)
0.2179**
(0.0954)
0.3304***
(0.1018)
0.4659***
(0.0870)
0.6376***
(0.1091)
Selection
−0.089
Ĝ
0.495
−0.020
0.436
0.087
0.310
−0.003
0.441
−0.069
0.571
0.043
0.404
0.038
0.483
0.052
0.671
0.136
0.729
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: a skilled worker is defined as one with more than 12 years of schooling. Regressions are estimated using Heckman selection
correction method where the first step is a probit regression. G refers to the gross wage gap (or ln(1 + G)), e refers to the explained
component of the wage gap (endowment effect), R refers to the unexplained (return effect) component of the wage gap. selection is the component of the wage gap attributed to self-selection into sector, and Ĝ is the selectivity-corrected gross wage gap.
The dependent variable in the wage equations is the logarithm of hourly wage; the independent variables are age, age squared,
years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, and
district fixed effects. The selection equation includes as explanatory variables, age, age squared, years of schooling, tenure, marital
status, full-time employment; urban area/refugee camp residence, occupational dummies, district fixed effects, number of jobholders, number of adults aged 70 or more in a given household, and locality’s share of public sector employees. Robust standard
errors are in parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 per cent levels.
where 𝜃 is an arbitrary percentile between (0,1); Q𝜃 (ln wi |xi ) is the ‘𝜃th’ quantile function of wages conditional on observed characteristics or attributes specified by xi; 𝛽𝜃 is the vector specifying the ‘returns’ or
‘effects’ of individual characteristics; and 𝛿𝜃 captures the unexplained gap of log earnings, all at the 𝜃th
quantile. Within the QR framework, for any given 𝜃 and a sample size of n, 𝛽𝜃 is derived as the argmin to
n−1
n
∑
𝜇𝜃 (ln wi − xi 𝛽𝜃 − di 𝛿𝜃 ),
(7)
i=1
where μθ is the ‘check’ function and is defined as μθ = θɛ if ɛ ≥ 0, or μθ = (θ - 1)ɛ if ɛ < 0. ɛ is the ‘error’ term
analogous to the OLS specification. The QR estimates of β at any given quantile can be interpreted as
the ‘returns’ to the attributes if Qθ(wi|xi) is assumed to be linear, (or a linear approximation). Moreover, to
ensure the results are not driven by selection bias, we estimate the QR specifications using the Heckman
selection correction method where the first step is a probit regression.
The usefulness of this technique can be seen from Table 8, which compares the OLS and quantile
regression, (QR), estimates of the dummy variable, or unexplained variation, (or impact), of working in
the public sector. While the OLS estimates only give information near the median or 50th percentile
of the wage distribution, the QR technique also provides estimates at various points along the wage
distribution.
Looking at the pay differential across the wage distribution, (Table 8), suggests a distinct trend over
time: the wage premium (penalty) for the public sector varies across the distribution, being higher,
(lower), at the lowest end of the wage distribution and decreasing, (increasing), along the wage distribution. It becomes negative at the top percentiles. By comparison, Azam and Prakash (2015) found
18
S. H. MIAARI
Table 8. selectivity-corrected quantile regression results.
Quantile Regression wage differentials at various percentiles
West Bank males
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
OLS Wage Differential
−0.0544***
(0.0137)
−0.1028***
(0.0139)
−0.0965***
(0.0128)
−0.0341**
(0.0135)
−0.0296
(0.0201)
0.0207
(0.0164)
0.0883***
(0.0148)
0.1252***
(0.0133)
0.1944***
(0.0154)
5th
0.1358
(0.1355)
0.1602
(0.1653)
0.5629***
(0.1644)
0.1463
(0.1863)
0.4545*
(0.2375)
0.2289
(0.1906)
0.0082
(0.2350)
0.2374
(0.1547)
0.3370*
(0.1820)
20th
0.0838*
(0.0503)
−0.0106
(0.0612)
0.2565***
(0.0563)
0.2145***
(0.0722)
0.4999***
(0.1218)
0.2553**
(0.1216)
0.0372
(0.0884)
0.2318**
(0.1004)
0.3497***
(0.0937)
50th
−0.0284
(0.0647)
0.0231
(0.0552)
−0.0260
(0.0403)
0.0771
(0.0784)
0.4361***
(0.0874)
0.2409***
(0.0591)
0.0892*
(0.0529)
0.1086
(0.0749)
0.2833***
(0.0831)
Gaza males
75th
−0.0825
(0.0865)
−0.1090**
(0.0468)
−0.1392***
(0.0484)
−0.1280
(0.1144)
0.0799
(0.0874)
0.0662
(0.0657)
−0.0049
(0.0489)
0.0225
(0.0775)
0.1295
(0.0797)
95th
−0.0621
(0.1712)
−0.5158***
(0.1662)
−0.4870***
(0.0865)
−0.2939**
(0.1200)
−0.2831*
(0.1518)
−0.2039
(0.1438)
0.1334
(0.1788)
−0.0349
(0.1174)
−0.1802
(0.1132)
OLS Wage Differential
0.2655***
(0.0187)
0.2269***
(0.0187)
0.2380***
(0.0173)
0.2367***
(0.0176)
0.2742***
(0.0233)
0.2604***
(0.0187)
0.3005***
(0.0196)
0.4266***
(0.0180)
0.5515***
(0.0210)
5th
0.0008
(0.4103)
0.0350
(0.2199)
0.0687
(0.2360)
0.6550***
(0.1987)
1.0119*
(0.6138)
0.4979**
(0.2262)
0.2681
(0.1978)
0.2904*
(0.1705)
0.4365**
(0.1893)
20th
0.3812**
(0.1713)
0.1630
(0.1083)
−0.0392
(0.1190)
0.4137***
(0.1509)
0.9175***
(0.2148)
0.3197**
(0.1364)
0.3541**
(0.1578)
0.5824***
(0.0836)
0.5400***
(0.1058)
50th
0.2617**
(0.1101)
−0.0521
(0.0976)
−0.1375
(0.1233)
0.4636***
(0.1154)
0.7921***
(0.1155)
0.3769***
(0.0979)
0.5604***
(0.0900)
0.7133***
(0.0708)
0.7678***
(0.0872)
75th
−0.1585
(0.1432)
−0.2419***
(0.0851)
−0.2434*
(0.1324)
0.3330**
(0.1638)
0.7881***
(0.0796)
0.4816***
(0.1063)
0.5677***
(0.1148)
0.6733***
(0.0812)
0.7785***
(0.1112)
95th
−0.0759
(0.3757)
−0.1279
(0.2068)
−0.0971
(0.1965)
0.5337**
(0.2654)
0.4774*
(0.2525)
0.3170*
(0.1741)
0.2195
(0.2266)
0.3793***
(0.1265)
0.4352**
(0.1835)
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs), 1998–2006.
note: Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. The dependent variable in the wage equations is the logarithm of hourly wage;
the independent variables are age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, district fixed effects and
a sector dummy variable (that takes on the value 1 if the worker is employed in the public sector, and 0 if employed in the private sector). The selection equation includes as explanatory variables,
age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, district fixed effects, number of jobholders, number
of adults aged 70 or more in a given household, and locality’s share of public sector employees. standard errors are in parentheses. The symbols *, **, *** represent statistical significance at the 10, 5,
and 1 per cent levels.
DEFENCE AND PEACE ECONOMICS
19
this to be positive for the entire wage distribution in India, but still smaller at the top quantiles, while
Saha, Roy, and Kar (2014) found the premium to be positive at the 25th quantile and negative at the
50th and 75th quantiles. Christofides and Michael (2013), found that the wage gap is negatively related
to income quantiles.
West Bank workers in the lowest 5th percentile earn significant wage premiums in the public sector
compared to their private sector counterparts. West Bank public sector workers in the 20th percentile
also had a positive premium throughout the entire observed period, as did public sector workers at the
median, (50th percentile), for the period 2001–2006. For the 75th percentile the premium was negative
until 2001, and for the 95th percentile it was negative throughout 1998–2006 (except for 2004). Between
1998 and 2000, the wage gap becomes more negative for the higher percentiles, and more positive
for the lower percentiles. Both of these trends reverse between 2000 and 2001. The wage premium
increased for all groups, except for the 95th percentile, between 2001 and 2002, becoming less negative
for high wage earners and more positive for those on lower wages. There was a decrease in the gap
between 2003 and 2004, and then another increase during the period 2005–2006. For workers of the
95th percentile, the gap increased between 2001 and 2004 and decreased in the period 2005–2006.
In Gaza, public sector workers in the 5th, 20th and 50th percentiles have a positive premium in 1998,
while the others have a negative one. Between 1998 and 2000, the wage gap decreases for all groups
except the 5th percentile, for whom in increases. All groups’ wage gaps rapidly rise in 2001, and for all
except the 95th percentile, this trend continues to 2002. For all groups, there is another decrease in
2003, and an increase in 2004–2006. In both the West Bank and Gaza, the wage gap sharply increased
for lower wage workers during the years of the Intifada. This difference in the public sector premium,
between high and low earning workers, mirrors the difference between high and low skilled workers
reported in Results section.
Conclusion
Currently, Palestinian public sector salaries absorb at least 85% of net revenues, crowding out virtually
all other expenditure. The past five years have seen a series of wage increases in the public sector with
the stated intention of bringing public sector salaries up to par with the private sector. This paper has
measured the public–private wage differential in the West Bank and Gaza, and described its dynamics
before and during the second Intifada using data from the PLFS of the PCBS. The paper is the first
systematic analysis of the public–private wage gap in the context of the economic costs of political
instability. While the public sector has become increasingly attractive to workers, the private sector
continues to be buffeted by exogenous shocks and beset by considerable uncertainty. Given that the
public sector was seen as a ‘buffer’ to absorb private sector job losses, due to increased closures and
movement restrictions following the second Intifada, employment in it expanded as the political and
economic situation worsened. Given the inseparability, and thus complexity, of the political and economic challenges facing the PA, and the need for objective analysis to inform policy, this paper offers
an interpretation of the wage gap phenomenon in terms of the changing demographic composition
of the Palestinian workforce over the study period. The paper has used Oaxaca–Blinder decomposition
to estimate the explained and unexplained components of the wage gap.
The estimates for West Bank workers suggest that in the earlier years of the analysis, particularly
in the pre-Intifada period, the public sector suffered a penalty in (log) hourly wages, but this penalty
disappeared, and then reversed over time. In 1999, the public–private log wage differential was estimated at –0.10. However, from 2003 a clear public sector premium emerges, which rises significantly
in 2005, just prior to the Palestine national elections. Among Gaza Strip workers, the public sector pay
premium declined between 1998 and 1999, and then increased between 2000 and 2006. Moreover,
most of the increase in the wage gap, in both the West Bank and Gaza, in the post-Intifada period stems
from an increase in the ‘returns’ to skills composition in the public sector (unexplained effect), rather
than a change in the skills composition of public sector workers, (explained effect).
20
S. H. MIAARI
The massive influx of Palestinians, who had previously worked in the Israeli private sector before the
second Intifada, into the PA local private sector, and a fall in local economic activity in the West Bank and
Gaza, simultaneously reduced the demand for, and increased the supply of, Palestinian unskilled workers. Consequently, the wage for private sector employees decreased over the Intifada period. Moreover,
public sector wages increased several times during the study period. The average PA wage increased
at an annual rate of 3.5% during the years 2000–2003 (World Bank 2006b). Furthermore, public sector
wages were further increased in the period 2004–2005, when the PA implemented its Civil Service Law
(World Bank 2006b) and again in 2006. Consequently, although in the pre-Intifada period the wage gap
between the public and private sectors narrowed, it has widened in the post-Intifada period.
Looking at the pay differential in the West Bank and Gaza across the wage distribution suggests a
distinct trend over time: the wage premium, (penalty), for the public sector varies across the distribution,
being higher, (lower), at the lowest end of the wage distribution and decreasing, (increasing), along the
wage distribution. It becomes negative at the top percentiles. Over time, the lower quantiles of West
Bank wage earners have continued to earn a significant (log) wage premium, and this has increased
after the outbreak of the Intifada. On the other hand, those in the very top income percentiles, (95th),
continue to face a wage penalty, which has attenuated over time. In Gaza, especially after the outbreak
of the Intifada, the wage premium increased for both low and high wage earners in the public sector.
Comparing 2000 to 2006, estimates indicate that public sector workers in the West Bank and Gaza in
all percentiles showed a steady increase in the wage premium.
Besides the fiscal burden of providing the public sector with wage premiums vis-à-vis the private
sector and the equity issues this raises, this trend is also worrying for other reasons. In an environment
of high unemployment, particularly among the young, this trend creates an incentive for job seekers
to ‘sit it out’ and wait for a secure job in the public sector rather than become absorbed in a highly
fragmented and uncertain private sector.
The results of this paper might suggest that the PA was using the expansion of the public sector and
the wage increases to maintain its popular support. Further research is needed to test this hypothesis.
Another interesting hypothesis is that security sector jobs are used to ‘buy out’ various armed groups
or potential dissidents.
Notes
1. Source: International Monetary Fund data files.
2. The Palestinian political and economic situation is unique in many ways, and international experience should be
interpreted with care when applied to the Palestinian context.
3. This is contrary to Leping (2005, 2006). Some of this difference can be attributed to the different timeframe for the
data of these papers, which is 2000 and 1989–2005, respectively, compared to 2008 for Christofides and Michael
(2013). We also know from Leping (2006) that the negative public sector premium became less negative with time.
4. Wage employment represents approximately 55% of total employment in the West Bank, and two-thirds of total
employment in Gaza, percentages that have remained fairly stable since the outbreak of the Intifada. The other
three types of employment are ‘employer’, representing approximately 5% of total employment in the West Bank
and 3% in Gaza; ‘self-employed,’ amounting to roughly 28% in the West Bank and 23 percent in Gaza; and ‘unpaid
family member,’ which accounts for approximately 11% of employment in the West Bank and 9% in Gaza. These
three groups were not asked about their wages in the PLFS.
5. Including Palestinians workers in Israel would give a confused picture of the public-private wage gap in the
Palestinian local market, because Palestinian workers in Israel are generally paid more than those employed in
the Palestinian local market.
6. The female labour force participation rate in 1998–2006 averaged 14.4% for females from the West Bank and 8.4%
for females from Gaza.
7. The PLFS questionnaire on hours worked asks, ‘How many hours did the household member work in all jobs last
week?’ This number was multiplied by 4.35 weeks per month, and then divided by the number of reported workdays
in the month to calculate hours worked per day.
8. Thereby removing the most extreme responses, which in some cases are simply the results of incorrect data entry.
9. According to the World Bank report in 2006a ‘West Bank and Gaza Update’: ‘Between Q3 of 2000 and Q4 of 2000, the
number of wage employees working in the private sector in the West Bank fell by 28,500; by Q2 of 2002, a further
27,900 West Bank private sector wage employees were no longer working, a decline of 48% from the last quarter
DEFENCE AND PEACE ECONOMICS
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
21
prior to the Intifada, when 117,600 workers enjoyed regular wage employment. In Gaza, the reduction was more
sudden: Whereas 43,000 Gazans held regular wage employment in Q3 of 2000, that number fell to 22,600 in Q4
of 2000; a further 2900 were without regular private sector wage jobs by Q3 of 2002, a decline of 59%.’.
According to the same report, public sector employment increased by 13% through 2007–2013, making it likely
that for the period studied in this paper, the public sector was also not very large by international standards.
For instance, the UNSCO database shows that the number of comprehensive closure days imposed in the West
Bank increased from 53 in 1998 to 260 in 2006; and in Gaza from 28 days to 77 days during the same period.
According to the World Bank report in May 2003, ‘The negative impact on domestic employment of job losses in
Israel was aggravated by the difficulties in conducting business within the West Bank and Gaza: Internal closures
and curfews are attended by significant transaction costs, disruption in production cycles, losses of perishable
output, and lower economies of scale. Regional variations in unemployment and labor participation between the
West Bank and Gaza are significant. By Q3/2002, 51,000 of the 327,000 eve-of-intifada private sector jobs had been
lost in the West Bank (16%), and 54,000 of 164,000 in Gaza (33%).’.
Miaari and Sauer (2011) document the large and statistically significant negative effects of the Israeli-Palestinian
conflict on Palestinian employment rates in Israel and mean monthly earnings, regardless of work location, (Israel
or the West Bank and Gaza), following the outbreak of the Intifada.
External closures consist of restrictions on the movement of Palestinians and Palestinian goods between the
West Bank, Gaza, and Israel (as well as third countries). Internal closures consist of restrictions on the movement
of Palestinians within the territories.
Controlling for occupational affiliation in the wage regressions would eliminate inter-occupational wage gaps.
Adding and subtracting the term (X̄ 1 − X̄ 0 )� 𝛽̂jin Equation (3).
We also use (a) pooled regression non-discriminatory wage coefficients because the source of the wage gap might
stem from both positive discrimination of one sector and negative discrimination of the other, and the two often
occur together (Jann 2008). In our case, the positive ‘discrimination’ is the extra funds allocated towards public
sector employment, while the negative one is the effects of the second Intifada on the private sector.
See Reimers (1983) and Neuman and Oaxaca (2004a) for more details on choosing the ‘correct’ selectivity
decomposition.
A greater number of jobholders in the same household increases the probability of joining the private sector,
without affecting the worker’s wage. A higher share of public sector employees in a locality facilitates access to
jobs in the public sector.
The number of Palestinian labourers in Israel falls from a high of 146,000 just prior to the start of the uprising,
(116,000 from the West Bank and 30,000 from Gaza), to around 50,000 in Q4 of 2004; since then, the number of
Palestinian workers in Israel and in the settlements has been relatively stable, fluctuating with the extent of closure
imposed by Israel.
Note that there is also great similarity between the trends of the returns effect and those of the selectivity-corrected
wage gap for skilled workers.
See Bassett snd Koenker (1978; 1982) for details on quantile regressions.
Acknowledgements
I am deeply grateful to the Palestinian Central Bureau of Statistics for providing the Palestinian Labour Force Survey data; I
owe special thanks to Michael Beenstock, Massimiliano Cali, Esteban Klor, Hani Mansour, Daniele Paserman, Robert Sauer,
Eytan Sheshinski, Avichai Snir, Asaf Zussman, and Noam Zussman for their helpful suggestions and comments.
Disclosure statement
No potential conflict of interest was reported by the author.
References
Adamchik, V. A., and A. S. Bedi. 2000. “Wage Differentials between the Public and the Private Sectors: Evidence from an
Economy in Transition.” Labour Economics 7 (2): 203–224.
Aslam, M., and G. Kingdon. 2009. “Public–Private Sector Segmentation in the Pakistani Labour Market.” Journal of Asian
Economics 20 (1): 34–49.
Azam, M., and N. Prakash. 2015. “A Distributional Analysis of Public-Private Wage Differentials in India.” Labour 29 (4):
394–414.
Bargain, O., & B. Melly (2008). Public Sector Pay Gap in France: New Evidence Using Panel Data. SSRN 1136232.
Bassett, G., and R. Koenker. 1978. “Regression Quantile.” Econometrica 46: 33–50.
22
S. H. MIAARI
Bassett, G., and R. Koenker. 1982. “An Empirical Quantile Function for Linear Models with I.I.D Errors.” Journal of the American
Statistical Association 77: 407–415.
Blinder, A. S. 1973. “Wage Differential: Reduced Form and Structural Variables.” The Journal of Human Resources 8: 436–455.
Cai, L., and A. Y. Liu. 2011. “Public-Private Sector Wage Gap in Australia: Variation along the Distribution.” British Journal of
Industrial Relations 49 (2): 362–390.
Calì, M., and S. H. Miaari. 2018. “The Labor Market Impact of Mobility Restrictions: Evidence from the West Bank.” Labour
Economics 51: 136–151.
Carlo, D. A., C. Lucifora, & F. Origo. 2005. Public Sector Pay and Regional Competitiveness: A First Look at Regional PublicPrivate Wage Differentials in Italy (No. 1828). Institute for the Study of Labor (IZA).
Christofides, L. N., and M. Michael. 2013. “Exploring the Public-Private Sector Wage Gap in European Countries.” IZA Journal
of European Labor Studies 2 (1): 1–53.
Depalo, D., & R. Giordano. 2011. “The Public-Private Pay Gap: A Robust Quantile Approach.” Bank of Italy Temi di Discussione
(Working Paper) No, 824.
Disney, R., and A. Gosling. 1998. “Does It Pay to Work in the Public Sector?” Fiscal Studies 19 (4): 347–374.
Dustmann, C and A. Van Soest. 1998. “Public and Private Sector Wages of Male Workers in Germany”. European Economic
Review 42 (8): 1417–1441.
Emilio, D., V. Ponczek, and F. Botelho. 2012. “Evaluating the Wage Differential between Public and Private Sectors in Brazil.”
Revista De Economia Política 32 (1): 72–86.
Foley, P., and F. O’Callaghan. 2011. “Investigating the Public-Private Wage Gap in Ireland Using Data from the National
Employment Survey 2007.” Journal of the Statistical and Social Inquiry Society of Ireland 39: 23–52A.
Gardeazabal, J., and A. Ugidos. 2004. “More on Identification in Detailed Wage Decompositions.” Review of Economics and
Statistics 86 (4): 1034–1036.
Giordano, R., D. Depalo, M. C. Pereira, B. Eugène, E. Papapetrou, J. J. Perez, A. Aaaa, and M. Roter. 2011. “The Public Sector
Pay Gap in a Selection of Euro Area Countries.” Working Paper Series 1406, European Central Bank.
Glinskaya, Elena, and Michael Lokshin. 2005. Wage Differential between the Public and Private Sectors in India. Mimeo.
Washington, DC: World Bank.
Heckman, James J. 1979. “Sample Selection Bias as a Specification Error.” Econometrica 47 (1): 153–162.
Horrace, W. C., and R. L. Oaxaca. 2001. “Inter-Industry Wage Differentials and the Gender Wage Gap: An Identification
Problem.” ILR Review 54 (3): 611–618.
Hyder, A. 2002. “Public-Private Wage Differentials in Pakistan.” The Bangladesh Development Studies 28 (4): 79–93.
Hyder, A., and B. Reilly. 2005. “The Public and Private Sector Pay Gap in Pakistan: A Quantile Regression Analysis.” The
Pakistan Development Review 271–306.
Jann, B. 2008. “The Blinder-Oaxaca Decomposition for Linear Regression Models.” The Stata Journal 8 (4): 453–479.
Jones, F. L. 1983. “On Decomposing the Wage Gap: A Critical Comment on Blinder's Method.” The Journal of Human Resources
18 (1): 126–130.
Jones, F. L., and J. Kelley. 1984. “Decomposing Differences between Groups: A Cautionary Note on Measuring Discrimination.”
Sociological Methods & Research 12 (3): 323–343.
Lausev, J. 2014. “What has 20 years of public–private pay gap literature told us? Eastern European transitioning vs. developed
economies.” Journal of Economic Surveys 28 (3): 516–550.
Leping, K. O. 2005. Public-Private Sector Wage Differential in Estonia: Evidence from Quantile Regression. University of Tartu
Faculty of Economics and Business Administration Working Paper, (39).
Leping, K. O. 2006. “Evolution of the Public–Private Sector Wage Differential during Transition in Estonia.” Post-Communist
Economies 18 (4): 419–436.
Lokshin, M., and B. Javanovic. 2003. “Public-Private Sector Employment Choice and Wage Differential in Yugoslavia.” Mimeo.
Washington, DC: World Bank.
Lucifora, C. and D. Meurs (2004), “The Public Sector Pay Gap in France, Great Britain, and Italy.” IZA Discussion Paper No. 1041.
Maczulskij, T. 2008. “Public-Private Wage Differentials for Males: Evidence from Finland.” University of Jyväskylä, School of
Business and Economics, Working Paper.
Miaari, S. H., and R. M. Sauer. 2011. “The Labor Market Costs of Conflict: Closures, Foreign Workers, and Palestinian
Employment and Earnings.” Review of Economics of the Household 9: 129–148.
Mizala, A., P. Romaguera, and S. Gallegos. 2011. “Public–Private Wage Gap in Latin America (1992–2007): A Matching
Approach.” Labour Economics 18: S115–S131.
Naser, Z. M. 2000. “Earnings Differential between Public and Private Sectors in Pakistan.” Pakistan Development Review 39
(2): 111–130.
Neuman, Shoshana, and Ronald L. Oaxaca. 2004a. “Wage Decompositions with Selectivity-Corrected Wage Equations: A
Methodological Note.” The Journal of Economic Inequality 2 (1): 3–10.
Neumark, D. 1988. “Employers’ Discriminatory Behavior and the Estimation of Wage Discrimination.” The Journal of Human
Resources 23: 279–295.
Nielsen, H. S. 2000. “Wage Discrimination in Zambia: An Extension of the Oaxaca-Blinder Decomposition.” Applied Economics
Letters 7 (6): 405–408.
DEFENCE AND PEACE ECONOMICS
23
Oaxaca, Ronald L. 1973. “Male-Female Wage Differential in the Urban Labor Market.” International Economic Review 14 (3):
693–709.
Oaxaca, Ronald L., and Michael R. Ransom. 1994. “On Discrimination and the Decomposition of Wage Differentials.” Journal
of Econometrics 61 (1): 5–21.
Oaxaca, R. L., and M. R. Ransom. 1999. “Identification in Detailed Wage Decompositions.” Review of Economics and Statistics
81 (1): 154–157.
Panizza, U., and C. Z. W. Qiang. 2005. “Public–Private Wage Differential and Gender Gap in Latin America: Spoiled Bureaucrats
and Exploited Women?” The Journal of Socio-Economics 34 (6): 810–833.
Papapetrou, E. 2006. “The Public-Private Sector Pay Differential in Greece.” Public Finance Review 34 (4): 450–473.
Polavieja, J. G. 2005. “Task Specificity and the Gender Wage Gap: Theoretical Considerations and Empirical Analysis of the
Spanish Survey on Wage Structure.” European Sociological Review 21 (2): 165–181.
Poterba, J.M., K.S. Rueben. 1994. “The Distribution of Public Sector Wage Premia: New Evidence Using Quantile Regression
Methods”. NBER Working Paper No. 4734 (May).
Reimers, Cordelia W. 1983. “Labor Market Discrimination against Hispanic and Black Men.” The Review of Economics and
Statistics 65 (4): 570–579.
Saha, S., P. Roy, and S. Kar. 2014. “Public and Private Sector Jobs, Unreported Income and Consumption Gap in India: Evidence
from Micro-Data.” The North American Journal of Economics and Finance 29: 285–300.
Nielsen, H. S., and M. Rosholm. 2001. “The Public-Private Sector Wage Gap in Zambia in the 1990s: A Quantile Regression
Approach.” Empirical Economics 26: 169–182.
Suits, D. B. 1984. “Dummy Variables: Mechanics V. Interpretation.” The Review of Economics and Statistics 66: 177–180.
Terrell, K. 1993. “Public-Private Wage Differentials in Haiti Do Public Servants Earn a Rent?” Journal of Development Economics
42 (2): 293–314.
Voinea, L., and F. Mihaescu. 2012. “A Contribution to the Public–Private Wage Inequality Debate.” Economics of Transition
20 (2): 315–337.
World Bank. 2003. “Twenty-Seven Months: Intifada, Closures, and Palestinian Economic Crisis: An Assessment”, May.
Washington, DC: World Bank.
World Bank. 2016. “Public Expenditure Review of the Palestinian Authority. Towards Enhanced Public Finance Management
and Improved Fiscal Sustainability.” Washington, D.C: World Bank.
World Bank. 2006a. West Bank and Gaza Update, A Quarterly Publication of the West Bank and Gaza Office, April. Washington,
DC: World Bank.
World Bank. 2006b. West Bank and Gaza Public Expenditure Review: From Crisis to Fiscal Independence.
Yun, M. S. 2005. “A Simple Solution to the Identification Problem in Detailed Wage Decompositions.” Economic Inquiry 43:
766–772.
24
S. H. MIAARI
Appendix 1.
Table A1. Decomposing the wage differential over time: β* = βpooled.
West Bank males
Year
1998
G
0.0664***
(0.0129)
0.0336***
(0.0128)
0.0277**
(0.0120)
0.0318**
(0.0125)
0.0868***
(0.0186)
0.1490***
(0.0160)
0.2776***
(0.0154)
0.2884***
(0.0133)
0.3296***
(0.0141)
1999
2000
2001
2002
2003
2004
2005
2006
E
0.1208***
(0.0121)
0.1364***
(0.0124)
0.1241***
(0.0116)
0.0659***
(0.0118)
0.1164***
(0.0179)
0.1283***
(0.0149)
0.1893***
(0.0142)
0.1632***
(0.0122)
0.1351***
(0.0133)
Gaza males
R
–0.0544***
(0.0137)
–0.1028***
(0.0139)
–0.0965***
(0.0128)
–0.0341**
(0.0134)
–0.0296
(0.0199)
0.0207
(0.0161)
0.0883***
(0.0146)
0.1252***
(0.0131)
0.1944***
(0.0152)
G
0.5483***
(0.0156)
0.5248***
(0.0142)
0.4911***
(0.0156)
0.4333***
(0.0186)
0.5410***
(0.0205)
0.5049***
(0.0154)
0.5832***
(0.0182)
0.7902***
(0.0158)
0.9182***
(0.0174)
E
0.2828***
(0.0170)
0.2979***
(0.0165)
0.2531***
(0.0153)
0.1966***
(0.0150)
0.2669***
(0.0187)
0.2445***
(0.0156)
0.2826***
(0.0171)
0.3636***
(0.0156)
0.3667***
(0.0166)
R
0.2655***
(0.0186)
0.2269***
(0.0186)
0.2380***
(0.0171)
0.2367***
(0.0171)
0.2742***
(0.0225)
0.2604***
(0.0185)
0.3005***
(0.0188)
0.4266***
(0.0173)
0.5515***
(0.0204)
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: The dependent variable in the wage equations is the logarithm of hourly wage; the independent variables are: age, age
squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, and district fixed effects. G refers to the gross wage gap (or ln(1+G)), e refers to the explained component of the wage gap
(endowment effect), and R refers to the unexplained (return effect) component of the wage gap. Robust standard errors are in
parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 percent levels.
Table A2. selectivity-corrected wage gap decomposition: β* = βpooled.
West Bank males
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.0656***
(0.0129)
0.0291**
(0.0128)
0.0277**
(0.0120)
0.0318**
(0.0125)
0.0868***
(0.0186)
0.1490***
(0.0160)
0.2776***
(0.0154)
0.2884***
(0.0133)
0.3296***
(0.0141)
E
0.1100***
(0.0259)
0.1139***
(0.0279)
0.0895***
(0.0239)
0.0534*
(0.0285)
–0.0350
(0.0371)
0.0547*
(0.0294)
0.1959***
(0.0267)
0.1625***
(0.0249)
0.1137***
(0.0307)
R
–0.0694
(0.0581)
–0.1531**
(0.0597)
–0.0523
(0.0516)
–0.0308
(0.0620)
0.2617***
(0.0800)
0.1491**
(0.0667)
0.0645
(0.0624)
0.0671
(0.0594)
0.1583**
(0.0728)
Gaza males
Selection
0.025
Ĝ
0.041
0.068
–0.039
–0.009
0.037
0.009
0.023
–0.140
0.227
–0.055
0.204
0.017
0.260
0.059
0.230
0.058
0.272
G
0.5483***
(0.0156)
0.5248***
(0.0142)
0.4911***
(0.0156)
0.4333***
(0.0186)
0.5410***
(0.0205)
0.5049***
(0.0154)
0.5832***
(0.0182)
0.7902***
(0.0158)
0.9182***
(0.0174)
E
0.3304***
(0.0675)
0.4475***
(0.0613)
0.4020***
(0.0555)
0.0533
(0.0395)
–0.0399
(0.0695)
0.1524***
(0.0422)
0.1889***
(0.0442)
0.3090***
(0.0406)
0.3188***
(0.0456)
R
0.1202
(0.1299)
–0.0538
(0.1092)
–0.0308
(0.1244)
0.6868***
(0.1452)
0.7800***
(0.1846)
0.4254***
(0.0976)
0.4937***
(0.1072)
0.5933***
(0.0870)
0.7565***
(0.1279)
Selection
0.098
Ĝ
0.451
0.131
0.394
0.120
0.371
–0.307
0.740
–0.199
0.740
–0.073
0.578
–0.099
0.683
–0.112
0.902
–0.157
1.075
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. G refers
to the gross wage gap (or ln(1+G)), e refers to the explained component of the wage gap (endowment effect), R refers to the unexplained (return effect) component of the wage gap. selection is the component of the wage gap attributed to self-selection into
sector, and Ĝ is the selectivity-corrected gross wage gap. The dependent variable in the wage equations is the logarithm of hourly
wage; the independent variables are age, age squared, years of schooling, tenure, marital status, full-time employment; urban
area/refugee camp residence, occupational dummies, and district fixed effects. The selection equation includes as explanatory
variables , age, age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence,
occupational dummies, district fixed effects, number of jobholders, number of adults aged 70 or more in a given household, and
locality's share of public sector employees. Robust standard errors are in parentheses. The symbols *, **, *** represent statistical
significance at the 10, 5, and 1 percent levels.
DEFENCE AND PEACE ECONOMICS
25
Table A3. selectivity-corrected wage gap decomposition by skill group: West Bank, β* = βpooled.
West Bank males
Skilled
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.0653***
(0.0244)
0.0288
(0.0230)
0.0294
(0.0227)
0.0065
(0.0237)
–0.0732**
(0.0336)
–0.0267
(0.0303)
0.0770***
(0.0293)
0.1417***
(0.0277)
0.1946***
(0.0290)
E
0.0389
(0.0377)
0.0954**
(0.0403)
0.1222***
(0.0307)
0.1081**
(0.0440)
0.0690
(0.0469)
0.0786*
(0.0439)
0.0790**
(0.0379)
0.1582***
(0.0353)
0.0900*
(0.0468)
R
0.2041*
(0.1143)
–0.0012
(0.1212)
–0.1257
(0.1079)
0.0540
(0.1544)
0.1053
(0.1417)
–0.0613
(0.1520)
0.1405
(0.1533)
0.3429**
(0.1528)
0.2201
(0.1988)
Unskilled
Selection
0.011
Ĝ
–0.085
0.092
–0.136
0.086
–0.067
–0.116
0.084
0.122
–0.247
–0.044
0.012
–0.027
0.032
0.000
0.161
0.135
0.070
G
–0.1022***
(0.0151)
–0.1728***
(0.0151)
–0.1654***
(0.0140)
–0.1116***
(0.0150)
–0.0531**
(0.0219)
0.0146
(0.0187)
0.1408***
(0.0170)
0.1537***
(0.0143)
0.2415***
(0.0161)
E
–0.0071
(0.0250)
0.0096
(0.0310)
–0.0949***
(0.0251)
–0.0943***
(0.0282)
–0.2381***
(0.0422)
–0.1217***
(0.0299)
0.0286
(0.0282)
–0.0382
(0.0257)
–0.0559*
(0.0330)
SelecR
tion
–0.1905*** –0.216
(0.0680)
–0.3275***
0.016
(0.0773)
–0.1036*
0.022
(0.0620)
–0.1168
0.004
(0.0714)
0.1887* –0.072
(0.1040)
0.0815
0.053
(0.0728)
–0.0485
0.094
(0.0736)
0.0243
0.069
(0.0638)
0.1592**
0.186
(0.0769)
Ĝ
0.097
–0.210
–0.214
–0.116
0.021
–0.018
0.060
0.092
0.051
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: a skilled worker is defined as one with more than 12 years of schooling. Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. G refers to the gross wage gap (or ln(1+G)), e refers to the
explained component of the wage gap (endowment effect), R refers to the unexplained (return effect) component of the wage
gap. selection is the component of the wage gap attributed to self-selection into sector, and Ĝ is the selectivity-corrected gross
wage gap. The dependent variable in the wage equations is the logarithm of hourly wage; the independent variables are age,
age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational
dummies, and district fixed effects. The selection equation includes as explanatory variables , age, age squared, years of schooling,
tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, district fixed effects,
number of jobholders, number of adults aged 70 or more in a given household, and locality's share of public sector employees.
Robust standard errors are in parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 percent levels.
26
S. H. MIAARI
Table A4. selectivity-corrected wage gap decomposition by skill group: Gaza, β* = βpooled.
Gaza males
Skilled
Year
1998
1999
2000
2001
2002
2003
2004
2005
2006
G
0.4620***
(0.0499)
0.4466***
(0.0423)
0.3236***
(0.0420)
0.2357***
(0.0415)
0.2492***
(0.0597)
0.3297***
(0.0470)
0.2091***
(0.0439)
0.4445***
(0.0428)
0.6790***
(0.0464)
E
0.2710***
(0.0906)
0.2903***
(0.0679)
0.1817***
(0.0447)
0.1025**
(0.0400)
0.1348***
(0.0508)
0.1590***
(0.0490)
0.0817***
(0.0286)
0.1497***
(0.0351)
0.1288***
(0.0453)
R
0.5839
(0.5121)
0.3401
(0.3801)
0.8636**
(0.3473)
0.0520
(0.3834)
–0.9240*
(0.4898)
0.2777
(0.3444)
0.5134
(0.3643)
0.5626***
(0.1852)
0.8928***
(0.2258)
Unskilled
Selection
0.183
Ĝ
0.038
0.314
0.042
–0.053
0.377
0.509
–0.294
0.780
–0.567
–0.176
0.510
0.198
–0.013
0.012
0.401
–0.252
0.856
G
0.4094***
(0.0159)
0.4136***
(0.0142)
0.3883***
(0.0156)
0.3737***
(0.0186)
0.4722***
(0.0192)
0.4032***
(0.0152)
0.5188***
(0.0171)
0.7209***
(0.0149)
0.8574***
(0.0175)
E
0.1997***
(0.0759)
0.3391***
(0.0698)
0.2675***
(0.0639)
0.0036
(0.0481)
–0.0461
(0.0866)
0.1273***
(0.0471)
0.1417***
(0.0464)
0.2960***
(0.0502)
0.3540***
(0.0542)
R
0.0595
(0.1435)
–0.0854
(0.1225)
–0.0571
(0.1356)
0.4926***
(0.1370)
0.5485***
(0.1716)
0.2355**
(0.1059)
0.3703***
(0.0969)
0.3993***
(0.1013)
0.6281***
(0.1243)
Selection
–0.089
Ĝ
0.495
–0.020
0.436
0.087
0.310
–0.003
0.441
–0.069
0.571
0.043
0.404
0.038
0.483
0.052
0.671
0.136
0.729
source: author calculations using Palestinian Labor force surveys (PLfs) of the Palestinian central Bureau of statistics (PcBs),
1998–2006.
note: a skilled worker is defined as one with more than 12 years of schooling. Regressions are estimated using the Heckman selection correction method where the first step is a probit regression. G refers to the gross wage gap (or ln(1+G)), e refers to the
explained component of the wage gap (endowment effect), R refers to the unexplained (return effect) component of the wage
gap. selection is the component of the wage gap attributed to self-selection into sector, and Ĝ is the selectivity-corrected gross
wage gap. The dependent variable in the wage equations is the logarithm of hourly wage; the independent variables are age,
age squared, years of schooling, tenure, marital status, full-time employment; urban area/refugee camp residence, occupational
dummies, and district fixed effects. The selection equation includes as explanatory variables , age, age squared, years of schooling,
tenure, marital status, full-time employment; urban area/refugee camp residence, occupational dummies, district fixed effects,
number of jobholders, number of adults aged 70 or more in a given household, and locality's share of public sector employees.
Robust standard errors are in parentheses. The symbols *, **, *** represent statistical significance at the 10, 5, and 1 percent levels.