Decomposing wage gaps between ethnic groups: the case of Israel.

I. Introduction

The 2000 riots among Israeli Arabs focused public attention on the social and economic conditions of Arabs in Israel. Many commentators pointed to the discrimination against the Arabs in many dimensions, including the labor market, as one of the main explanations for these riots. Previous literature and published statistics show large income differences between Jews and Arabs in Israel, which cannot be explained by differences in individual characteristics.

Past investigations of income gaps between Jews and non-Jews in Israel have treated non-Jews as one group. The main contribution of this paper is to analyze separately the income gaps between Jews and the three main religious minorities: Muslims, Christians, and Druze. Given the history of the Jewish-Arab conflict and the different characteristics of the various religious groups, we have reason to believe that the Christians and the Druze are less influenced by inter-religion hostility than the Muslims and it would be interesting to see if this fact has any effect on labor market outcomes.

In this paper, we are using the 1995 Israeli census. To the best of our knowledge, the income gaps between Jews and Arabs in Israel have never been investigated using this data set; rather, most of the relevant literature focuses on earlier or smaller data sets. The size of this data set allows us to contribute to the discussion in the following respects. First, we have enough observations to separate the different Arab religious groups and to compare them. Second, previous literature pointed to differences in occupations as one of the main explanations for the income gaps between Jews and Arabs in Israel, and such a large sample size allows us to focus on the concentration of each religious group in a specific occupation or industry.

In general, the literature on this topic is quite narrow. All of the research on this topic finds large wage gaps between Jews and Arabs, but when this income gap is decomposed, the results show that, in large part, it is not explained by differences in individual characteristics. Some authors interpret the unexplained wage gap as discrimination against Arabs. Lewin-Epstein and Semyonov (1994), for instance, find that more than half of the income gap between Jews and Arabs in Israel cannot be explained by differences in characteristics, including occupational status. Haberfeld and Cohen (1998) find the unexplained part to be a quarter of the wage gap. Wolkinson (1999) provides some micro-evidence for discrimination. Using a sample of 48 large industrial companies, Wolkinson found that Arabs are limited both in recruitment and in the type of positions they assume within companies.

In addition to differences in wages and income, previous research also documents large differences in the occupational distribution between Jews and Arabs in the Israeli labor force. Lewin-Epstein and Semyonov (1993), for example, show that Arab workers are over-represented in blue-collar jobs, but under-represented in academic jobs. Similarly, Gharrah and Cohen (2001) find that educated Arabs are much less likely to work in management or academic-technical occupations than educated Jews, but are much more likely to be found in the educational field.

Most Arabs in Israel live in Arab-only communities, which are usually much smaller than Jewish communities. Previous research suggests that this residential segregation has both advantages and drawbacks for the Arab population. On the one hand, residence in non-Jewish communities limits labor opportunities and variety, but on the other hand, such communities do not have to contend with heavy competition for job opportunities since Jews are not looking for employment in Arab sectors (Semyonov 1988; Lewin-Epstein and Semyonov 1994).

This analysis focuses on the northern part of Israel, where most Christians, Druze, and Muslims live. Using a sample of men aged 25-64 who are not new immigrants and do not live in a kibbutz, (1) we find significant explained and unexplained income gaps in favor of Jews, which confirms results from previous literature. The unexplained gaps tend to be larger the more educated the individual. In addition, Jews, who have much higher representation in the more lucrative occupations, earn significantly more than other population groups in Israel. The main result of this paper is that, in almost every dimension, Muslims suffer from the largest income gaps. Druze, on the other hand, have the lowest income gaps across most of the income distribution, due in large part to direct and indirect benefits they reap from being the only minority serving in the army. However, among all the minorities, Christians are the most educated and most concentrated in the top occupations, which explains why they have the lowest gaps in the highest percentiles of the income distribution. (2)

This paper is organized as follows: section 2 introduces the primary religious groups in Israel and their statistical characteristics; section 3 describes the data set, the sample being used, and the methodology; section 4 reports the results for the various decompositions performed and the distribution of occupations across religious groups; and section 5 discusses the main results and directions for future research.

2. The Main Religions in Israel

The main religious group in Israel, the Jews, makes up more than 80% of the population. There are three main minorities in Israel: Muslims, Christians, and Druze. In addition, a little more than 1% of the population belongs to some other religion, are new immigrants who are not Jews, or did not report their religion. In this analysis, we are focusing on income gaps between the Jews and the three main minorities. While we realize that there are several subgroups based on ancestry within the Jewish group, in this work we treat all Jews as members of one group. (3)

Muslims are the largest minority population in Israel, comprising more than two-thirds of the entire minority population (Table 1). Christians make up the second largest group, while Druze form the smallest of the three. Despite their different religions, all three minority groups speak the same language--Arabic. The Druze are a Middle Eastern minority group that traces their origins back to 11th-century Cairo, where it began as an Islamic reform movement. Today, most Druze live in the mountainous regions of Lebanon, Syria, Israel, and Jordan.

In many respects, the Muslims and the Druze are very similar, while the Christians in the region more closely resemble the Jews. Muslim and Druze populations are younger and less educated; they have higher fertility rates, live in smaller communities, and have a much smaller rate of female representation in the labor force. Most of the minorities live in the northern part of Israel.

Historically, the relationship between the Jewish and Arab populations in Israel is rife with conflict and tension. Such a climate of hostility on both sides very likely is a factor in labor market outcomes. While we do not know of any research that attempts to measure the different attitudes between the Jewish majority and the Arab minority, we believe that the most hostile relationship exists between the Jews and the Muslims, since a large part of the Israeli-Arab conflict is actually a Jewish-Muslim conflict. The relationship between Jews, Christians, and Druze is probably much less hostile. In addition, the Druze serve in the Israeli army in significant numbers, while members of the other two minorities almost never serve.

3. Data and Estimation

We are excluding women from this analysis because there are enormous differences in female labor force participation rates across religions. Table 1 shows that 71% of Jewish women aged 25-64 are in the labor force, compared to less than 20% among Muslims and Druze. Since the Arab women who decide to work are probably very different from the ones who do not, such a disparity likely would create selection biases that would be very difficult to correct. (4)

Among males, we are focusing on those from the ages of 25-64. Younger males are not included because of marked differences in labor force participation among the groups. Beginning when they are 18, Jewish and Druze males usually serve in the army for 3-4 years, and only then do they enter the job market or institutions of higher education. This career path is very different from the typical course taken by most young Muslims and Christians, who do not serve in the army.

The labor force participation rates across religions are quite similar for males aged 25-40. However, the minority participation rates decline sharply in the older age groups, especially among less educated workers. Because the workers who retire from the labor force are probably quite different from the ones who stay, this, again, can create selection biases. As a robustness check we also used samples with workers younger than 45, the results of which do not change the conclusions presented here.