Ethnicity, Class and the Earning Inequality in Israel, 1983-1995

by Nabil Khattab
University of Bristol

Sociological Research Online, Volume 10, Issue 3,

Received: 10 Sep 2004     Accepted: 29 Jun 2005    Published: 30 Sep 2005


This paper focuses on the role of ethnicity and class in generating earnings inequality in Israel. Unlike previous studies on inequality of opportunities in Israel, in this paper I compare the earnings of five ethnic groups: European Jews (Ashkenazi), Asian-African Jews (Sephardi), Muslim Palestinians, Christian Palestinians and Druze Palestinians. In addition, both men and women are taken into account. The analysis, which is based on data obtained from the 1983 and 1995 Israeli population censuses, has revealed that in Israel, class variations resulting from the differentiation of employment contracts in the labour market, appear to have played a much more important role over time in producing earnings inequality. However, at the same time, it was found that class in this context is highly related to ethnicity, thereby suggesting that class and ethnicity are interwoven. While it seemed that to some extent, class plays a similar role among men and women, the role of ethnicity among men was much more central than it was among women, in the allocation of people into class positions.

Keywords: Class, Discrimination, Earning Inequality, Ethnicity, Israel, Palestinians


1.1The ethnic dimension of earnings inequality has been the focus of many investigations over the last three decades. It is the common wisdom of these investigations that ethnicity plays an important role in determining earnings inequality between various groups in society, and that subordinate ethnic minorities earn far less than superordinate ethnic groups. For example, Blacks and Hispanics in the US earn less than Whites (Tienda and Lii, 1987), Pakistani immigrants in the UK earn less than Whites (Blackaby et al, 1998) and Palestinian-Arabs in Israel earn less than Jews (Semyonov and Cohen, 1990). In order to explain ethnic inequality in the labour market, most of the studies in this area have focused on human capital differences (individual-level attributes such as education and experience) and discrimination. As for the latter, two aspects of discrimination are considered: lack of employment opportunities resulting either from the disproportionate concentration of ethnic minorities in some localities or from social and structural barriers that constrain these minorities' access to high prestige occupations, and differential rewards of dominant and subordinate groups with similar characteristics (Cancio et al, 1996; Lewin-Epstein and Semyonov, 1992; Tienda and Lii, 1987). However, explaining ethnic inequality, as a function of human capital characteristics and the local opportunity structure, leaves much of the ethnic gap in labour market outcomes unexplained.

1.2 Other scholars of social stratification have pointed out the importance of class as a key factor of stratification and the determination of social power, and life chances as well as socio-economic rewards (Reid, 1998; Erikson and Goldthorpe, 1992;Goldthorpe, 2000; Sorensen, 1996; Winn, 1984; Yaish, 2001). For example, previous research has shown that class positions, as conceptualised in terms of employment relations, influence political behaviour (Evans, 1999), educational attainment (Goldthorpe, 2000) and health behaviour (Blane et al, 1998). If class, in the above scenario, is so important, then it is reasonable that the earnings of a person will also be influenced by his or her class position, and therefore earnings inequality should be examined with reference to class affiliation or membership (Evans, 1996; Ginn and Arber, 1991; Winn, 1984; Wright, 1978). Nevertheless, most recent studies in the area of social stratification and earnings inequality, not only in Israel, have not paid enough attention to this issue (Lewin-Epstein and Semyonov, 1992).

1.3 This paper aims to evaluate the relative importance of class and ethnicity in generating earnings inequality between various ethnic groups in Israel between two points of time, 1983 and 1995. Previous studies of income and class inequality in Israel have treated the Arabs in Israel as a homogeneous ethnic group without considering the internal ethnic/religious dimension (Khattab, 2002a). In this study I compare five groups in Israel: the two major Jewish groups - European-American Jews (Ashkenazi) and Asian-African Jews (Sephardi) and the three major Arab groups - Muslims, Christians and Druze[1]. It is worth mentioning here that although I refer to these five groups using the term ethnic for the sake of simplicity, in reality the division between them is also national and religious. For example the division between the Jewish groups and the Arab groups is national and religious. Ashkenazi and Sephardi groups are Jews and speak mainly Hebrew. The three Arab groups belong to three different religions (Islam, Christianity and Druze) but all of them speak Arabic and nationally are considered Palestinians (with the exception of Druze). However, the analysis in this study excludes all Palestinians living in the West Bank and Gaza Strip including those who work inside Israel due to the lack of data but also because they are not citizens of the state of Israel[2]. Likewise, foreign workers were excluded due to lack of data and Jewish immigrants from the former Soviet Union and from Ethiopia, as these two latter groups were not included in the 1983 census.

1.4 Ashkenazi Jews are clearly the dominant group in Israel (Kraus and Yonay, 2000). The other groups suffer discrimination and are located far below the Ashkenazi Jews in the socio-economic ladder. However, Sephardi Jews are more likely than the Arab ethnic minorities to have access to socio-economic resources because of their Jewish background. The three Arab groups, not only differ from the dominant group, but also differ from each other. The Christians and the Druze are expected to do better in the labour market and to have higher earnings than the Muslims - the Christians, because of their better human capital resources (Khattab, 2002b), and the Druze because of their presence in the military and security forces, which entitles them to more social and economic opportunities. This unique ethnic/national composition gives a good opportunity to assess the effect of ethnicity, and the combined effect of ethno-nationality on earnings inequality, along with human capital and class.

1.5 Additionally, and unlike most previous studies regarding socio-economic inequality in Israel that excluded women from their analyses (Kraus and Yonay, 2000; Yaish, 2001;Yonay and Kraus, 2001), I will extend this study to include both sexes. For this end, separate models for men and women will be used. Including women in the analysis will significantly contribute to exploring the gender dimension of earnings inequality and will illuminate whether ethnicity and class play similar roles in determining the earnings of men and women in Israel.

Theoretical Considerations

2.1 According to the liberal theory (status attainment approach) (Blau and Duncan, 1967; Treiman, 1970), in modern societies, the criteria of achievement such as education, will replace the criteria of ascription such as ethnicity and sex, in determining the life chances and the socio-economic rewards of individuals. That is to say, in industrial societies, as a result of educational expansion and an increasingly 'open' and 'meritocratic' form of society, access to socio-economic positions and socio-economic rewards depends on educational attainment, skills and qualification, namely human capital. This suggests that earnings inequality results from differences in human capital (Tienda and Lii, 1987) regardless of ethnicity, gender, religion, age and other such featuress. The importance of the latter factors should decline as societies move on towards industrialisation, openness, democracy and liberalism.

2.2 However, recent studies have indicated that in some multi-ethnic societies, such as in Israel, America and Britain, ethnicity is still playing an important role in the process of stratification and in determining the life chances of individuals (Cancio et al, 1996; Heath and McMahon, 1997; Modood et al, 1997; Khattab, 2002a; Kraus and Yonay, 2000). For example, Khattab (2002), has found that ethnicity (religion) within the Palestinian minority in Israel is a major factor in determining the labour market participation of women. Cancio et al (1996), using data from the Panel Study of Income Dynamics for successive cohorts of young workers in 1976 and 1983 in the US have found that the effect of race on earnings has increased rather than decreased. Using data from the National Survey of Ethnic Minorities conducted in 1994, Modood et al (1997) have found large ethnic differences in educational and economic attainment, with Whites and Indians being the most advantaged groups, and Pakistanis and Bangladeshis, the least advantaged. Thus, it is reasonable to hypothesise that in Israeli society, ethnicity would play a significant role in generating earnings inequality.

2.3 The literature on social stratification has suggested that in modern societies class structure is the major determinant of wealth, powers and prestige in society. Differences between people in each one of these (wealth, powers and prestige) result from their differential location in the class structure. There are two main classical approaches for measuring class: Marxist and Weberian (Crompton, 1998: 55). Even the most contemporary theories of class can be categorised along these lines such as Wright's approach (1985) (Marxist), and Goldthorpe's approach (1992) (new Weberian). For example, according to Wright's approach that follows the self-consciously Marxist project, classes are defined in terms of their location within the social relations of production (Wright, 1985) and not in terms of occupational positions in the marketplace. He argues that occupations or occupational aggregation cannot produce classes as these are defined within the technical relations of production. Contrary to Wright's approach, the new-Weberian approach of Goldthorpe suggests that occupations (the market and work situations) are important in defining classes. In this paper I adopt Goldthorpe's model for measuring class, as I believe that the occupational status is what determines the amount of wealth and prestige that one can obtain. Indeed, the division of labour in the marketplace has become the main source of inequality of opportunities and rewards. Erikson and Goldthorpe (1992) (see also Goldthorpe 2000: ch.10), have developed a theoretical framework for the formation and existence of class division in industrial countries with the aim to differentiate position within labour markets and product units ... in terms of the employment relations that they entail (Goldthorpe 1992: 37). This notion of employment relations is represented in the first instance by the distinction between employers, the self-employed, and employees, which is in line with the Marxist and Weberian traditions. However, Erikson and Goldthorpe make a further distinction within the category of employees, based on the employment relations. In the Goldthorpe class schema there are two forms of employment relations. On the one hand, the employment of manual and lower-grade non-manual workers typically involves a form of labour contract, while that of professional, administrative and managerial employees in organisational bureaucracies is usually characterised by a form of service relationship. The essence of the distinction between a labour contract and a service relationship can be derived by considering their defining characteristics. The labour contract entails: "a relatively short-term and specific exchange of money for effort. Employees supply more-or-less, discrete amounts of labour, under the supervision of the employer or of the employer's agents, in return for wages which are calculated on a 'piece' or time basis" ( Erikson and Goldthorpe 1992: 41-42). In contrast, service relationships: involve a longer-term and generally more diffuse exchange. Employees render service to their employing organisation in return for 'compensation' which takes the form not only of reward for work done, through a salary and various prerequisites, but also comprises important prospective elements - for example, salary increments on an established scale, assurances of security both in employment, and through pension rights after retirement, and above all, well-defined career opportunities (1992: 41-2).

2.4 In addition, they identify several intermediate (mixed forms of employment) types of job referred to as classes - routine non-manual workers, supervisors of manual workers and technicians. Typically, these positions are located between bureaucratic structures and rank-and-file workforces. Those forms of employment "in various ways, combine elements of both the labour contract and the service relationship" (Goldthorpe 2000: 221).

2.5 Individuals who share the same social class (as a matter of position in a system of employment relations) are expected to demonstrate similar attitudes, views and behaviour and to have similar rewards and returns for their human capital or effort. Most relevant to this study are the earnings of individuals, and the extent to which these earnings vary across classes. According to Erikson and Goldthorpe (1992) (see also Evans, 1996), earnings are likely to be empirically related to the distinction between a service and a labour contract, and therefore should vary across classes. Thus, it can be assumed that earnings and other employment relations characteristics, such as autonomy and authority, are given differentially to employees of different employment contracts in order to ensure economic efficiency and to achieve their loyalty.

2.6 The most extended version of Goldthorpe's schema includes eleven classes (Erikson and Goldthorpe, 1992). However, it is actually the seven-class version of the schema that was used for the empirical analysis, because some combination of categories was necessary in order to maintain standards of cross-national comparability in the data. In this study, it seems that further merging is needed, as I do not have sufficient cases for such detail for the Arab population, and therefore I have combined Class IVab and IVc, and Class VIIa and VIIb (Kraus et al, 1998).

Israeli Society

3.1 Israel is a multi-ethno-national society comprised of two major Jewish groups (Sephardi and Ashkenazi) and three major Palestinian religious groups (Muslims, Christians and Druze). While the main cleavage that differentiates Jews and Palestinians is the national one, the secondary division among each group is also important. Within the Palestinian population, the Muslims are in the vast majority accounting for 82%, Christians number approximately 10%, and Druze constitute around 8%. The Palestinian population as a whole makes up approximately 19% of the population in Israel (Central Bureau of Statistics, 2002: Table 2.1 at Within the Jewish population it is common to distinguish Jews of European or American descent (Ashkenazi Jews) from those of Asian or African origin (Sephardi Jews) (Lewin-Epstein and Semyonov, 1992). Both groups have similar proportions (Yaish, 2001). Indeed, in the Jewish Israeli society in Israel the ethnic cleavage is considered the most important as it affects and shapes some of the other important divisions such as the religious and the political ones. For example, most Israeli sociologists have taken the ethnic and national demarcations to be the main axes of stratification in Israel. However, recently some Israeli sociologists have begun to consider class alongside the ethnic and national cleavages (Kraus, 1998: 298; Yaish, 2001: 574; Yaish, 2000: 595).

3.2 Since the establishment of the State of Israel in 1948, Palestinian-Arabs as a whole have continually been Israel's most oppressed group, facing wide systematic discrimination in every aspect of social, political and economic life (Haidar, 1994). For example, many highly educated Palestinians men and women are denied access to the most desirable jobs within the general Jewish controlled labour market forcing them to seek jobs within the local ethnic labour market where jobs opportunities are very limited. This has resulted in high unemployment rate amongst Palestinians in general and highly qualified people in particular (Kraus and Yonay, 2000). Politically, the Palestinian minority has been marginalised during public debates over the country's future, and Palestinian political parties have been excluded from government coalitions (Barzilai, 2001). Under these circumstances, Palestinians could neither convert their electoral power into better access to social and economic resources, nor exert any real pressure on the Israeli government to redistribute resources equally between both communities.

3.3 The majority of Palestinian workers (62%), depend on the Jewish-controlled labour market, and consequently their place of employment differs from their place of residence (Lewin-Epstein and Semyonov, 1992), resulting from the high degree of residential segregation between both communities (Yaish, 2001).

3.4 Previous studies have indicated that the spatial segregation of the Palestinian community plays an important role in developing the local ethnic labour market and encourages the gradual spread of public services provided by the state to the Palestinian community, such as public health clinics, welfare services, educational institutions and municipal administration (Kraus and Yonay, 2000; Lewin-Epstein and Semyonov, 1994; Shavit, 1992). According to these studies, those who find employment within the Arab ethnic enclave enjoy advantages in the conversion of educational resources into occupational outcomes (Lewin-Epstein and Semyonov, 1994). Semyonov (1988: 264) for example, argues that Israeli Arabs benefit from their residential segregation. By analysing the 1983 Israeli census data, he shows that Arabs working in mixed or Jewish communities suffer the detrimental consequences of occupational discrimination, while Arabs working in Arab towns and villages are shielded from Jewish competition.

3.5 The three major religious (ethnic) groups within the Palestinian population, mentioned earlier, also differ from each other culturally, residentially and in their social and economic characteristics. Christians, who tend to reside in urban localities, have a lower fertility rate, are provided with better educational opportunities, and have an occupational status that is considerably higher than that of the Muslims and Druze. Since 1956, Druze males have been conscripted into the Israeli military, which theoretically entitles them to better educational resources, better access to the labour market, more government investments and other financial benefits (Khattab, 2002a).

3.6 Likewise, the Jewish population is also ethnically deeply divided. During the pre-state period, a dominant Jewish capitalist class comprised of Jewish settlers arriving from Europe (Ashkenazi Jews) with many more resources than the local Palestinians, and motivated by Zionist ideology, had begun to crystallise (Kraus and Yonay, 2000; Yaish, 2001). The Ashkenazi Jews played a key role in establishing the state of Israel, and then in taking control and dominating its resources and institutions, thereby enhancing the inferior position of Sephardi Jews who were in Palestine long before the establishment of the new state. In addition, by applying a population dispersal plan and creating residential differences between the Ashkenazi population and the new Sephardi immigrants, the Ashkenazi group seeks to maintain and shelter its dominance and control over the social and economic resources (Yaish, 2001). Most Sephardi Jews who arrived from the Middle East and North Africa were regarded as primitive, lacking any kind of qualification, and thus required educating in order to fit modern life (Kraus and Yonay, 2000). Consequently, these new immigrants were more likely to join the labour market as manual workers. Nonetheless, it should be noted that they were regarded as part of the Jewish community, and automatically received more benefits in comparison with the Palestinian community which became the most oppressed minority after the majority of them were forced to leave their homes during the war, and found themselves outside of what had become Israel.

3.7 What has been said so far reveals that ethnicity in Israel is a major determinant of attainment. The Jewish population has, in general, higher attainment levels than Arabs in every social and economic field. Furthermore, Jews enjoy higher returns on every kind of individual resource (Lewin-Epstein and Semyonov, 1993). For instance, in terms of income, Lewin-Epstein and Semyonov (1992) have found that Jews earn much higher incomes than Palestinian workers, as a result of different opportunity structures faced by Jews and Arabs, in which the latter are more likely to be employed in far less desirable and lower-paid jobs.

3.8 Furthermore, it is not obvious that all Jewish sub-groups have higher attainment than all Palestinian sub-groups. Previous studies that examine the differences in educational and occupational attainment of Jews and Palestinians have suggested that Palestinian Christians have higher educational and occupational attainment than Sephardi Jews, whereas the latter have similar rates in achieving higher education to those achieved by Muslims (Kraus and Yonay, 2000; Shavit, 1990).

3.9 Thus, it is quite clear that there is an evident ethnic basis for inequality of opportunities in Israel and that ethnicity/nationality play an important role in the stratification system. Yet, there is an additional axis of stratification which influences the level of inequality of opportunities for the majority of Israeli society, across ethnic groups and over time, and that is the class structure (Yaish, 2001). In this study, I will examine the extent to which class structure, independently of ethnicity, affects the inequality in earnings in Israel. In what follows, I will describe the data and the variables used in the analysis.

Data and Variables

4.1 The analysis in this paper is based on data from the 1983 and 1995 Israeli censuses conducted by the Israel Central Bureau of Statistics (CBS). A 4% sample of individuals aged 15+ (N=120,936) in Israel's population who participated in both censuses provided detailed information on education, employment, earnings and other socio-demographic variables. The sample of 1983 comprised 24,538 persons (14,609 men and 9,794 women), while the 1995 sample included 33,693 persons (18,549 men and 15,144 women). All persons were employees, aged 25-64 in respect of men and 25-60 in respect of women. In both samples, all ethnic groups were well represented, with the exception of Druze women in the sample of 1983, whose number in the labour market was extremely low, and therefore they have been excluded from the analysis concerning the 1983 data. Furthermore, it is worth mentioning that all new immigrants who arrived during the 1990s are not included in the following analyses.

4.2 In both censuses, there was no difference in measuring the variables used in this study, with the exception of earnings (the dependent variable). Between 1983 and 1995, the Israeli currency changed to become the New Israeli Sheqel instead of the old Sheqel used until September 1985. The new Sheqel is equal to 1,000 old Sheqels (three zeros were dropped from the old Sheqel). However, this change does not impose any constraint on the analysis, since in the regression models, earnings are measured in both cases as the natural logarithm of the gross monthly income from employment. Moreover, for descriptive statistics and comparison purposes, earnings are presented in American US Dollars. Additionally, both datasets (the 1983 data and the 1995 data) are analysed separately (for a description of the variables see Appendix II).


5.1 Table 1 displays means and standard deviations for the variables used in the analysis, by gender and year. Because the interest of this paper is the impact of ethnicity/nationality and class on earnings, I present the mean and standard deviation of the monthly wage (presented in US Dollars) by ethnicity, year and sex in Table 2, and by class, year and sex in Table 3. Table 2 shows that in 1983 the Ashkenazi Jews have the highest earnings among both men and women, followed by Sephardi Jews among men and by Christians among women. Amongst men, Muslims have the lowest earnings with $508 per month. The other two Palestinian groups are somewhere in the middle between Sephardi Jews and Muslims, with Christians having higher earnings than Druze ($675 and $637 respectively). Like Muslim men, Muslim women have the lowest income at $429 per month. Sephardi women are in the middle between Muslim and Christian women. As already indicated, Druze women were excluded from the analysis of the 1983 data, due to their very small number. In 1995 the distribution of earnings among men is much the same as it was in 1983 with the exception of Druze having slightly higher earnings than Christians ($1522 versus $1508 respectively). Among women, the Ashkenazi Jews have the highest earnings levels, earning $407 per month on average more than Sephardi Jews, $509 per month more than Christian women, $654 more than Muslim women, and $753 more than Druze women. These figures suggest that the most advantaged group among men and women are the Ashkenazi Jews. While Muslims were the most disadvantaged group among men in both years and among women in 1983, Druze women were the most disadvantaged group among women in 1995. The figures in Table 2 also suggest that Druze men succeeded in closing the gap between themselves and Christians, and Sephardi women succeeded in establishing an advantage over Christians.

Table 1. Means and Standard Deviations of the Variables Used in the Analysis.

Table 2. Means and SD's of income* by ethnic group, sex and year of census.
* US Dollars.
** In 1983 there were a very small number of Druze women (9) in the sample who were in paid jobs.

5.2 Additionally, it is clear that the monthly income of men is much higher than the monthly income of women, a fact that is particularly noticeable if we compare the earnings of men and women within the same ethnic group. This also holds true regarding the earnings of Ashkenazi, Sephardi and Christian men against all women's groups. However, Table 2 shows that Ashkenazi women earned more than Muslim men in both years, and more than Druze men in 1983 only.

5.3 Table 2 also presents the ratio between the income in 1983 and 1995 across the ethnic groups. While the monthly income among male Ashkenazi Jews, Muslims and Christians in 1995 is around 2.22-2.23 times bigger than it was in 1983, it is 2.29 and 2.39 times bigger among Sephardi and Druze respectively. These figures suggest that the income of the latter (especially for the Druze) has increased more rapidly than the income of the former. Unlike men, women in the two Jewish groups have a similar income ratio (2.20), which is much higher than the comparable ratio among the Palestinian groups (1.92 and 1.68 respectively for Muslim and Christian women), suggesting that the income inequality between Jewish women and Palestinian women has increased.

Table 3. Means and SD's of income* by class, sex and year of census.
* US Dollars.

5.4 Table 3 presents the distribution of the monthly gross income (US Dollars) by class, year and sex. Service class employees (I+II) have the highest monthly income among both men and women, in both years, followed by skilled workers (V+VI), with the exception of men in 1995, where routine non-manual workers (IIIab) earned more than skilled workers (V+VI). The latter earned more than the petty bourgeoisie (IVabc), with unskilled manual workers (VIIab) being the most disadvantaged group earning the lowest monthly income. Concerning the differences between men and women, Table 3 clearly shows that men earn much more than women within each of the classes. Additionally, this table shows that the income ratio for classes I+II and IIIab is higher than for the other classes (especially classes IVabc and VIIab), suggesting that the earnings inequality between the former classes and the latter has increased between 1983 and 1995. It also seems that earnings inequality between men and women in classes IVabc and VIIab has increased as well, whereas in classes I+II and V+VI it has decreased somewhat.

5.5 From Tables 2 and 3 we can only learn about the income distribution by ethnicity and by class, noting for example, that earnings vary across ethnic groups and across classes. However, we are not in a position to predict the relative strength of both ethnicity and class in influencing earnings. In order to explore the relative impact of these variables, including other background and human capital variables and how this impact has changed between 1983 and 1995, I turn now to report the results generated from the regression analysis. Separate models for men and women at the two points of time (1983 and 1995) will be presented. Tables 4 and 5 display the results for men and women respectively. In each table there are four models: the first reports the effect of background variables; the second presents the impact of human capital variables; the third demonstrates the effect of ethnicity; and the fourth examines the influence of class.

5.6 Model 1 in Table 4 shows that all background predictors are statistically significant and explain 5% of the total variance in 1983, but only 1% in 1995. This suggests that the total impact of living in an urban locality, living in a metropolitan area and being married has somewhat declined between 1983 and 1995. Model 2 shows that human capital predictors have significant influence on the monthly (ln) income in both years (1983 and 1995) and in the expected direction. The only exception here is experience and experience (squared) that are insignificant in 1995. While Model 2 in 1983 explains around one quarter (23%) of the variance in earnings, in 1995 it explains only 4% suggesting that human capital predictors have become far less important than before. This result contradicts the arguments based on the liberal theory and the human capital theory that human capital predictors will play a greater role in determining earnings over time.

Table 4. Unstandardised Regression Coefficients Predicting Income (ln) of Men.

+ The coefficient and its standard error are multiplied by 100.
* P < .05
** P < .01

5.7 In Model 3 the ethnicity factor is included showing that, holding other factors constant, all ethnic groups earn less than Ashkenazi Jews. The only exception here is the coefficient for Druze in 1995 that is not significant. This means that the monthly (ln) income of Druze is not statistically different from the equivalent income of Ashkenazi Jews. However, this model contributes 3% to the total explained variance in 1983, whereas in 1995, interestingly, it does not contribute in any way to explain the variance. This may support the hypothesis mentioned earlier that over time, the influence of ethnicity will be on the decline. Alternatively, it may indicate that the effect of ethnicity has not vanished, but simply shifted from influencing earnings directly to affecting the allocation to social classes. This hypothesis will be examined later.

5.8 Model 4, which examines the effect of class, contains the most interesting result. While in 1983, class improves the ability to explain the variance in earnings by only 3%, in 1995 it dramatically improves the explained variance from 4% to 24%. This means that class is the most powerful factor in determining one's monthly income in 1995, whereas in 1983 the most powerful factor was human capital. This may suggest, as with the case of ethnicity, that in 1983 human capital variables were directly influencing earnings, and that over time their influence has shifted from determining one's earnings to determining one's social class. This explanation will be examined empirically later. However, in 1983 classes IIIab, IVabc, V-VI and VIIab were significantly disadvantaged in terms of earnings relative to class I+II, with classes IVabc and VIIab being the most disadvantaged. In contrast, in 1995 only the coefficients concerning classes of IVabc and VIIab were significant, suggesting that, holding all other factors constant, the workers in these classes earned far less than those in class I+II. It is worth mentioning that even after controlling for class positions, the coefficients for ethnicity were still significant. From Model 4 it can be seen that relative to Ashkenazi men, Muslim men are the most disadvantaged group (-0.46 and -0.29 in 1983 and 1995 respectively), whereas Sephardi men are the least disadvantaged group (-0.14 and -0.12 in 1983 and 1995 respectively). Druze and Christians were somewhere between Sephardi and Muslim men. Moreover, controlling for class positions in 1983 seems only to slightly decrease the coefficients of ethnicity and education. However, controlling for class positions in 1995 has dropped the coefficients of ethnicity and education by approximately half. This might be a clue that over time the relationship between ethnicity and class has been reinforced.

5.9 Table 5 presents the regression analysis of women's monthly (ln) income against the independent variables. The results for Model 1 show that in both years, background variables do not contribute to explaining the variance in women's income. More interesting are the results of Model 2 which show that human capital factors, especially education, have a very strong influence on women's monthly (ln) income, explaining 28% of the earnings variance in 1983. This pattern is very similar to that found regarding the effect of human capital in the men's model. In 1995, model 2 explains 9% of the variance, indicating that the explanatory power of this model among women is higher than the equivalent model among men, which has contributed only 3% to the explained variance.

Table 5. Unstandardised Regression Coefficients Predicting Income (ln) of Women.

+ the coefficient and its standard error are multiplied by 100.
* P < .05
** P < .01

5.10 In contrast, it seems that ethnicity does not have a major effect on women's monthly income, since in 1983 it does not contribute at all to explaining the variance, and in 1995 it only adds 1%. Additionally, the coefficients show that only the data for Sephardi Jews in 1983 and of Muslims in 1995 were statistically significant. Finally, adding the class factor contributed 2% more to the explained variance in 1983, and 5% in 1995, indicating a somewhat similar, but much more modest, pattern to the one found among men. In other words, it seems that over time, class position has become a major factor in determining earnings among women, and the most important factor among men.

5.11 From the results we can see that the coefficients for education in 1995 are bigger than those for education in 1983, but have a lower contribution to the explained variance. Furthermore, the coefficients for education (among men and women), and to a lesser extent for ethnicity among men, have largely decreased when we control for class. This may lead to the conclusion that human capital (education) and ethnicity have become insignificant dimensions of earnings inequality in Israel. However, this may also lead us to the assumption that the influence of human capital and ethnicity on earnings has become much more indirect over time, although still substantial, and is mediated through differentially distributing people into class positions. In other words, it is possible that education and ethnic origin play an important role in sorting people across social classes. In order to examine the latter argument, I regressed class distribution against education and ethnicity, controlling for background variables. The results of this analysis are presented in Table 6 and from the results it can be seen that the effect of education on the log odds contrasting entry classes VIIab, V+VI, IVabc and IIIab with class I+II are much stronger in 1995 than in 1983 for both men and women. This means that education played a greater role in sorting people into class positions. For example, it is quite clear that people with less than academic education (men and women) are less likely than academics to have occupations which fall into class I+II. This can also be seen by the effect of post-secondary education on the log odds of women being in classes VIIab, V+VI, IVabc and IIIab rather than in class I+II. In 1995 women with post-secondary education were more likely than academic women to be in classes VIIab, V+VI, Ivabc and IIIab rather than in class I+II, whereas in 1983 the situation was reversed. This would indicate that academic education defines the boundaries between Class I+II and other classes.

5.12 Like education, ethnicity would seem to play an important role in allocating men and women in Israel into class positions. However, its effect among men is slightly higher than among women, and it follows a different pattern. Generally speaking, the influence of ethnicity on class has increased between 1983 and 1995, especially among men, indicating that all ethnic groups (except for Druze in 1995) were disadvantaged relative to Ashkenazi Jews. As far as men are concerned, Muslims were the most disadvantaged group in both years, with a clear sign that over the period under study, their class position has worsened. For example, in 1983, Muslims were less likely than Ashkenazi Jews to be in Class V+VI, and more likely to be in classes VIIab and IVabc than in class I+II. In 1995, the coefficient involving classes VIIab and IVabc became higher, and the coefficient involving class IIIab became significant, indicating that Muslims were less likely than the other groups to be in Class IIIab. Sephardi Jews and Christians became closer to each other, with Sephardi Jews being less handicapped than Christians in 1995. Finally, Druze men in 1983 were less likely than Ashkenazi Jews to be in class V+VI, but more likely than the latter to be in class IIIab. By contrast, in 1995 it seems that Druze were not statistically different from Ashkenazi Jews regarding classes VIIab, IVabc and IIIab, yet less likely to be in class V+VI.

Table 6. The effect of Ethnicity and Education on Class Entry Contrasted with Class I+II, Controlling for Background Variables, 1983 and 1995 (Multinomial Logit).

* P < .05
** P < .01

5.13 Regarding the effect of ethnicity on class positions among women, it can be seen that unlike men, not all ethnic groups were disadvantaged relative to Ashkenazi Jews, and that the coefficients in 1995 were somewhat smaller than they were in 1983. This tendency reveals that over the period investigated, the role of ethnicity among women in sorting individuals into social classes has somewhat decreased. Sephardi women were more likely than Ashkenazi women to be in class VIIab in both years rather than in class I+II. In 1983, they were less likely to be in IVabc, whereas in 1995 they were less likely to be in class V+VI. Muslim women in 1983 were less likely than Ashkenazi women to be in each one of the classes other than class I+II, but in 1995 they were more likely to be in class VIIab. For the other three classes, there was no change in the direction of influence, yet the size of the coefficient was slightly smaller. The class position of Christian women did not change relative to Ashkenazi women between 1983 and 1995. In both years, they were less likely than Ashkenazi to be in classes V+VI and IIIab, with the coefficients of 1995 being smaller than in 1983. Finally, Druze women in 1995 were less likely than Ashkenazi women to be in class IIIab.

Discussion and Conclusions

6.1 The main purpose of this study was to examine the role of ethnicity and class in generating earnings inequality in Israel. The analysis has focused on how earnings inequality has changed between 1983 and 1995, and the role played by ethnicity and class in this process. It was proposed that although the effect of ethnicity should be on the decline over time, ethnicity would still be an important factor in generating earnings inequality. Further, it was suggested that over time (in modern societies), class as measured by Goldthorpe's class schema, would become the main source of inequality between people.

6.2 The analyses lend support to both propositions. With respect to class, it was found that class in Israel appears to have assumed a much more important role over time in producing earnings inequality. This indicates that a class structure resulting from the differentiation of employment contracts in the labour market has crystallised over time, and this seems to determine inter alia how much a person will earn. It seems that the most important differentiation in this regard is between class VIIab and IVabc on the one hand, and classes I+II, IIIab and V+VI on the other hand. While workers (men and women) in the former classes earn far less than workers in the latter classes, it was also observed that the earnings gap between these classes has increased over the period under discussion. The differences between these classes are expected and can be explained by the concept of economic efficiency, and level of qualification, knowledge and skills required for the employees in performing their work (Goldthorpe 2000: ch.10). According to this rationale, earnings (and other employment relations characteristics such as autonomy and authority) are given differentially to employees of different classes (employment contracts) in order to ensure economic efficiency, and to achieve their loyalty and commitment. Manual unskilled employees (class VIIab) can be easily supervised, have no authority and autonomy at work, and are not expected by employers to have any level of qualification. Additionally, those employees can be relatively easily replaced without causing any particular damage to the employer. Therefore, the inducements given to them will be low, reflecting the nature of their labour contract. Concerning the petty bourgeoisie, it was found in previous studies (Yaish, 2000; Yaish, 2001) that in Israel, this consists mainly of self-employed (class IVb) rather than small employers (class VIa). The former were more likely to have high mobility into the unskilled working class. While this trend has been reduced over time, it does nevertheless suggest that the self-employed and the unskilled workers may have some common characteristics. Although workers in this class enjoy substantial independence, other previous studies have shown that they do not hold any advantaged economic position compared to employees (Kraus, 1992, cited in Yaish, 2000). The results of this study lend high support to this conclusion, at least based on the earnings of the disadvantaged classes IVab and VIIab relative to other classes.

6.3 By contrast, professionals and managers, and to some extent supervisors of manual workers and technicians, are expected by employers to be economically efficient, loyal and highly committed to their employment. In order to achieve all of that, employers offer much better employment relations conditions (including salary) to these workers, creating different levels of earnings (Evans, 1996) (see also Erikson and Goldthorpe, 1992).

6.4 However, the analysis has revealed that the role of class in generating earnings inequality in Israel is highly related to ethnicity and education. On the one hand, the findings demonstrated that over time, class, independently from other factors, has become much more important in determining earnings inequality, whilst on the other hand, the findings have evidently indicated that a large part of the class influence on earnings has resulted from ethnic and educational differences.

6.5 My analysis of the effect of ethnicity and education on class position showed that the role of both factors in determining class position has increased from 1983 to 1995. Within the same period, the direct influence of both factors on earnings decreased, and the role of class in determining earnings dramatically increased, especially among men. Thus, it is safe to conclude that the influence of ethnicity and education on earnings has not vanished, but has become more indirect, influencing earnings through the allocation of people to class positions.

6.6 The finding concerning the influence of education on earnings and on the distribution of men and women in Israel into class positions (which was slightly stronger than the effect of ethnicity) may lend support to the human capital and the liberal theories. However, concerning the influence of ethnicity, a recent study on the class structure of the Israeli society has concluded that "the ethnic/national cleavage in Israel appears to have played a less important role over time in the allocation of Israeli men to class positions" (Yaish, 2001: 434). The findings of the present study are inconsistent with that conclusion. It was found that for men, but not for women, ethnicity did play an increased role over time in the class structure. In fact, it seems that to a great extent, the class structure has been formatted along ethnic lines creating a situation in which class and ethnicity are complemented or even interwoven. This finding is highly supported by a recent study on the labour force participation of Palestinian men in Israel (Sa'di and Lewin-Epstein, 2001). According to that study, "Class seems to be the main social mechanism through which ethnic inequality is reproduced" (2001: 799). Thus, the type of employment relations a person may obtain is very likely to be affected by his or her ethnicity, which in turn may affect his or her earnings. For example, Muslim men have the lowest level of earnings, and this can be simply explained by their over-concentration within the unskilled manual class and the self-employed class. As mentioned earlier, those were the most disadvantaged classes in terms of earnings. Indeed, Muslim men were more likely than other groups to become unskilled manual workers, 48% of them being in this class in 1983 and 44% in 1995 (See Appendix 1). The difference between the two figures over time is explained mainly by increased mobility from the unskilled manual class into self-employment as mentioned by Yaish (2001: 431). Although becoming self-employed within the Palestinian ethnic labour market (Yaish, 2001; Yonay and Kraus, 2001) has not improved their earnings, it has at least provided them with employment independence, and some protection from job competition. Christian men follow this pattern somewhat, relative to Ashkenazi Jews, yet in terms of earnings they are far ahead of Muslims, being more likely to be skilled and non-manual workers, and to enjoy better access to educational resources (Khattab, 2002b; Kraus and Yonay, 2000).

6.7 Sephardi men earn more than all other Palestinian groups making them the least disadvantaged group in terms of earnings relative to Ashkenazi Jews. However, from previous studies we know that that their educational attainment is similar to that of Muslims, but lower than the attainment of Christians, and their occupational distribution is less favoured than that of Christians (Kraus and Yonay, 2000). Their earnings nonetheless, are higher than both groups, a situation explained by the religious-national ideology of the state of Israel that views all Jews, regardless their ethnic origins as part of the nation (the Jewish collective), whereas other groups are, by definition, excluded. By their inclusion within the 'right' collective, Sephardi Jews become entitled to better and higher earnings since they are able to enter labour markets with better opportunities, regardless of their human capital resources. This finding is consistent with a study conducted by Lewin-Epstein and Semyonov a decade ago in which they concluded that "earnings of Arabs (as compared to Jews) are more dependent on the characteristics of the local labor market and less on resources of individuals" (1992: 1115). Based on their individual resources, we would expect Sephardi Jews to have lower earnings than at least Christians, but because of their access to labour markets with better opportunities than Muslims and Christians, Sephardi Jews can receive higher earnings.

6.8 The Druze represent an interesting case. In terms of earnings, Druze males have succeeded in locating themselves above both Muslims and Christians, and in terms of class positions they appear to be slightly higher than Sephardi Jews and the other Palestinian Groups (see Tables 2, 6 and Appendix 1). However, when class and education are controlled for, it seems that they are less advantaged than Christians, but still far ahead of Muslims. The Druze seem to enter labour markets with better employment opportunities than other Palestinian groups as a consequence of their special status in the state of Israel and their special relations with the Jewish community arising from the fact that they are the only Palestinian group who serve within the Israeli military. This finding provides more support to the argument mentioned earlier concerning the effect of the local employment opportunities available for the group, versus the effect of individual attributes (Lewin-Epstein and Semyonov, 1992).

6.9 As far as women are concerned, the analysis reveals that women in Israel earn far less than men. However, the ethnic basis for inequality of earnings among women would appear to affect only Muslim and Druze women in Israel. The findings indicate that Muslim women suffer from a double penalty regarding their earnings, being women who earn far less than men, and also being part of a highly-disadvantaged ethnic group. Muslim women are more likely than Muslim men, and women from other ethnic groups, to join the labour market as non-manual employees (classes I+II and IIIab) resulting in a higher occupational status (Khattab, 2002a; Semyonov et al, 1999), but the lower wages in the Palestinian enclave economy relative to the dominant economy (Semyonov, 1988; Yonay and Kraus, 2001), renders their income much lower than Ashkenazi, Sephardi and Christian women. Further, it was observed that over time (between 1983 and 1995), earnings inequality between Muslim women on the one hand, and Muslim men, and women from other ethnic groups on the other hand, has increased. There were two sources of this increase: 1) the income of Muslim women has increased less rapidly than other groups at a lower ratio; and 2) Muslim women have entered class VIIab where wages are very low, in increasing numbers. Muslim women represent a very interesting case that reveals some of the weaknesses and constraints of the ethnic enclave. Based on the ethnic enclave rationale, one may expect their earnings to be as high as the prestige (status) of the occupation. However, whilst some conversion of educational resources to high status jobs within the ethnic enclave labour market does occur, substantially lower economic returns (income) on education are nevertheless received (see also Semyonov, 1988; Yonay and Kraus, 2001).

6.10 The analysis of this study has shown that in order to understand the earnings inequality, and perhaps the inequality in other spheres in Israel, it is crucial to take both ethnicity and class into account. Focusing on ethnicity only may generate a wrong conclusion, while concentrating purely on class will only offer one part of the story. The analysis in this paper has revealed that ethnicity and class are inextricably linked with each other in a way that makes the understanding of them separately, impossible. That is to say, that in the surface processes, class seems to supersede ethnicity in generating social and economic inequalities, but in the underlying processes, class structure is highly-correlated with ethnicity in Israel. In other words, it was revealed that ethnicity strongly influences the class position in Israel, as is believed to be the case in other multi-ethnic societies such as in Britain. However, it is important to emphasise that in addition to ethnicity, it was observed that other factors might have played an influential role in creating the class structure, and as a result, the earnings inequality. For example, the ideology of the Jewish state that provides Sephardi Jews with better opportunities, political and historical circumstances that would seem to assist Druze mobility, and spatial and cultural factors that depress the attainment of Muslims.

6.11 The above analysis and discussion can be beneficial in expanding our understanding of the factors and mechanisms of exclusion that might be used by different societies and groups. For example, ideologies that determine who belongs (is included) and who does not belong (is excluded) on the base of race or ethnic background, such as in Germany where citizenship was, until recently, exclusively determined by belonging to the German ethnic group by blood, which in turn sanctioned access to the different types of resources in society (education, employment opportunities, political influence and so on). This ideology to a great extent may shape the class position of immigrants and ethnic groups in Germany. Similarly, racial discrimination that is practised by the white dominant population in other European countries (e.g. Britain, France or Sweden) against their minority ethnic groups can explain the class position of the latter. While racial discrimination or 'racist' ideologies may closely tie ethnicity with class, political and historical circumstances may differentiate between minority ethnic groups in their class position such as in the case of Indian and Pakistani-Bangladeshi groups in Britain. By considering their political and historical circumstances before their immigration we may be able to generate more understanding of why Indians are more advantaged than Pakistanis and Bangladeshis in terms of their class position. Thus, this present study is not only relevant to understanding the Israeli case, but it goes beyond to shed light on other cases where ethnicity and class go hand in hand and the former is closely tied into the latter.

Appendix 1

The distribution of class positions by ethnicity in 1983 (left) and 1995 (right) for men (top) and women (bottom) aged 25-64.

Appendix II

The following independent variables were used in this study:

Type of locality: a dummy variable indicating whether a person lives in an urban locality or a rural locality (1=Urban, 0=Rural).

Metropolitan area: a dummy variable that indicates whether a person lives in a metropolitan area (1= lives in a metropolitan area, 0= does not live in a metropolitan area).

Marital status: a dummy variable coded 1 if a person is married and 0 if otherwise.

Educational qualification: this variable was measured using the CASMIN scale (Brauns and Steinmann, 1999). The variable has five response categories: elementary education or less (1ab), secondary education but less than matriculation (2ab), matriculation (2c), lower tertiary education (3a) and higher tertiary education (3b). The first category is the references group (1ab).

Weekly hours of work: is measured as the usual number of hours worked per week.

Years of previous work experience: is measured as age minus years of schooling minus six years (Berndt, 1991).

Years of previous work experience (squared): this variable measures the assumption of the human capital theory that earnings should follow a parabolic shape, peaking somewhere in midlife. The coefficient of this variable is expected to be negative (Berndt, 1991).

Schooling X Experience: this interaction term is specified in order to examine the assumption of the human capital theory that individuals with more schooling receive more on-the-job training.

Ethnicity: following Kraus and Yonay (2000), a set of dummy variables was used to represent the different ethnic groups. The categories were: Western Jews (Ashkenazi), Eastern Jews (Sephardi), Muslims, Christians and Druze. The Ashkenazi Jews were the reference group.

Class position: class position is measured according to Goldthorpe's class schema (Erikson and Goldthorpe, 1992). In this study I employ a five class schema as follows: I+II: service class; IIIab: routine non-manual class; IVabc: petty bourgeoisie; V+VI: skilled manual workers; and VIIab: unskilled manual workers.


1 The Druze is a fiercely independent group concentrated in Lebanon around the base of Mount Hermon, and in the mountains behind Beirut and Sidon. A few villages are also located on the Golan Heights, in Syria and just inside the northern border of Israel. Very little information is known about the Druze religion. It started in the 9th century CE as a breakaway group from Islam.

2 In nowadays, there are severe restrictions on the employment of Palestinians from the West bank and Gaza Strip to work inside Israel as a result of the current political situation and on going violence between the Palestinians and Israel.


BARZILAI, Gad. 2001. "Fantasies of Liberalism and Liberal Jurisprudence: State Law, Politics, and the Israeli-Arab-Palestinian Community." Israel Law Review.

BERNDT, Ernst R. 1991. The Practice of Econometrics : Classic and Contemporary. New York: Addison-Wesley.

BLACKABY, D. H., D. G. Leslie, P. D. Murphy, and N. C. O'Leary. 1998. "The Ethnic Wage Gap and Employment Differentials in the 1990s: Evidence for Britain." Economics Letters 58:97-103.

BLANE, David, Scott M. Montgomery, and Lee R. Berney. 1998. "Research Note: Social Class Differences in Lifetime Exposure to Environmental Hazards." Sociology of Health and Illness 20:532-536.

BLAU, Peter Michael and Otis Dudley Duncan. 1967. The American occupational structure. New York,: Wiley.

BRAUNS, Hildgard and Susanne Steinmann. 1999. "Educational Reform in France, West-Germany, the United Kingdom and Hungary: Updating the CASMIN Educational Classification." ZUMA-Nachrichten 23:7-44.

CANCIO, A. Silvia, T. David Evans, and Jr. David J. Maume. 1996. "Reconsidering The Declining Significance of Race: Racial Differences in Early Career Wages." American Sociological Review 61:541-556.

CROMPTON, R (1998). Class and Stratification. Polity Press (2nd edition).

ERIKSON, Robert and John H. Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford, England: Clarendon.

EVANS, Geoffrey. 1996. "Putting Men and Women into Classes: An Assessment of the Cross-Sex Validity of the Goldthorpe Class Schema." Sociology 30:209-234.

EVANS, 1999. The End of Class Politics? Class Voting in Comparative Context. Oxford: Oxford University Press.

GINN, Jay and Sara Arber. 1991. "Gender, Class and Income Inequalities in later life." The British Journal of Sociology 42:369-396.

GOLDTHORPE, John H. 2000. On Sociology: Numbers, Narratives and the Integration of Research and Theory. Oxford: Oxford University Press.

HAIDAR, Aziz. 1994. "The Arab Minority in Israel's Economy: Patterns of Ethnic Inequality." Contemporary Sociology 23:652-653.

HEATH, Anthony and Dorren McMahon. 1997. "Education and Occupational Attainments: The Impact of Ethnic Origins." in Education, edited by A. H. Halsey, H. Lauder, P. Brown, and A. S. Wells. Oxford: Oxford University Press.

KHATTAB, Nabil. 2002a. "Ethnicity and Female Labour Market Participation: A new look at the Palestinian enclave in Israel." Work Employment & Society 16:91-110.

KHATTAB, 2002b. "Fantasy, Rationality or Resistance: The Social and Gender Construction of Future Orientations among Palestinian Youth in Israel." Unpublished Ph.D., The Hebrew University of Jerusalem (in Hebrew), Jerusalem.

KRAUS, Vered, Yossi Shavit, and Meir Yaish. 1998. "Gender and Ethnic Differences in the Transition from School to Work in Israel." Pp. 221-251 in From School to Work: A Comparative Study of Educational Qualifications and Occupational Destinations, edited by Y. Shavit and W. Muller. Oxford: Clarendon Press.

KRAUS, Vered and Yuval Yonay. 2000. "The Power and Limits of Ethnonationalism: Palestinians and Eastern Jews in Israel, 1974-1991." British Journal of Sociology 51:525-551.

LEWIN-EPSTEIN, Noah and Moshe Semyonov. 1992. "Local Labor Markets, Ethnic Segregation, and Income Inequality." Social Forces 70:1101-1119.

LEWIN, 1993. The Arab Minority in Israel's Economy: Patterns of Ethnic Inequality. Boulder: Westview Press.

LEWIN, 1994. "Sheltered Labor Markets, Public Sector Employment, and Socioeconomic Returns to Education of Arabs in Israel." American Journal of Sociology 100:622-651.

LEWIN-EPSTEIN, Noah and Moshe Semyonov. 1994. "Sheltered Labor Markets, Public Sector Employment, and Socioeconomic Returns to Education of Arabs in Israel." American Journal of Sociology 100:622-651.

MODOOD, Tariq, Richard Berthoud, Jane Lakey, James Nazroo, Patten Smith, Satnam Virdee, and Sharon Beishon. 1997. Ethnic Minorities in Britain: Diversity and Disadvantage. London: Policy Studies Institute.

SA'DI, Ahmad H. and Noah Lewin-Epstein. 2001. "Minority Labour Force Participation in the Post-Fordist era: the case of the Arabs in Israel." Work Employment & Society 15:781-802.

SEMYONOV, Moshe. 1988. "Bi-Ethnic Labor Markets, Mono-Ethnic Labor Markets, and Socioeconomic Inequality." American Sociological Review 53:256-266.

SEMYONOV, Moshe and Yinon Cohen. 1990. "Ethnic Discrimination and the Income of Majority-Group Workers." American Sociological Review 55:107-114.

SEMYONOV, Moshe, Noah Lewin Epstein, and Iris Brahm. 1999. "Changing Labour Force Participation and Occupational Status: Arab Women in the Israeli Labour Force." Work, Employment and Society 13:117-131.

SHAVIT, Yossi. 1990. "Segregation, Tracking, and the Educational Attainment of Minorities: Arabs and Oriental Jews in Israel." American Sociological Review 55:115-126.

SHAVIT, 1992. "Arabs in the Israeli Economy: A Study of the Enclave Hypothesis." Israel Social Science Research 7:45-66.

SORENSEN, Aage B. 1996. "The Structural Basis of Social Inequality." American Journal of Sociology 101:1333-1365.

TIENDA, Marta and Ding-Tzann Lii. 1987. "Minority Concentration and Earnings Inequality: Blacks, Hispanics, and Asians Compared." American Journal of Sociology 93:141-165.

TREIMAN, D. J. 1970. "Industrialization and Social Stratification." Pp. 280 in Social stratification: research and theory for the 1970s, edited by E. O. Laumann. Indianapolis,: Bobbs-Merrill.

WINN, Stephen Van Zandt. 1984. "Social Class and Income Returns to Education in Sweden: A Research Note." Social Forces 62:1026-1034.

WRIGHT, Erik Olin. 1978. "Race, Class, and Income Inequality." American Journal of Sociology 83:1368-1397.

WRIGHT, Erik Olin (1985). London: Verso.

YAISH, Meir. 2000. "Old Debate, New Evidence: Class Mobility Trends in Israeli Society, 1974-1991." European Sociological Review 16:159-183.

YAISH, 2001. "Class Structure in a Deeply Divided Society: Class and Ethnic Inequality in Israel, 1974-1991." British Journal of Sociology 52:409-439.

YONAY, Yuval P. and Vered Kraus. 2001. "Strategies of Economic Endurance: Israeli Palestinians in the Ethnic Economy and the Public Sector." Research in Social Stratification and Mobility 18:207-247.