Occupational and Residential Segregation: Volume 17

Although the measurement of segregation by gender or ethnic group in the labor force has long been of interest to both sociologists and economists, the sociology and economics literatures on this topic have evolved in different ways and remained largely separate. This has also been the case to some extent with research on the measurement of residential segregation. Although much of the segregation measurement literature is in sociology and geography, economists have contributed to this field as well, particularly in the development of measures of residential income segregation. Again, however, the economics literature has remained largely separate from that in geography and sociology.

Purpose – We propose the Information Theory of Segregation, which holds that all measures of segregation and of inequality are united within a single conceptual framework. Accepting this framework implies that all measures of inequality can also be used to measure segregation and that all measures of segregation are fundamentally based on measures of inequality.

Methodology – We state several propositions that follow from the information theory perspective, and show mathematically that many common measures of inequality and segregation satisfy the propositions.

Findings – We show that all common measures of inequality can be used to form measures of segregation and that the resulting measures can be applied to binary, polytomous, and continuous variables. Further, we develop several new measures, including a Gini Segregation Index (GS) for continuous variables and Income Dissimilarity Index (ID), a version of the Index of Dissimilarity suitable for measuring economic segregation. We show that segregation measures can easily be adapted to handle persons of mixed race, and describe the Non-Exclusive Index of Dissimilarity (NED) and the Non-Exclusive Entropy Index of Segregation (NEH). We also develop a correction for structural constraints on the value of segregation measures, comparable to capacity constraints in a communications channel, which prevent them from reaching their theoretical maximum or minimum value.

Originality – Placing inequality and segregation measures in a common framework is useful for several reasons. It highlights a common mathematical structure shared by many different segregation measures, and it suggests certain useful variants of these measures that have not been recognized previously.

In the context of educational segregation by ethnic group, it has been argued that rigorous pairwise segregation comparisons over time or across space should be invariant in two situations: when the ethnic composition of the population changes while the distribution of each ethnic group over the schools remains constant (invariance 1), or when the size distribution of schools changes while the ethnic composition of each school remains constant (invariance 2). This paper makes two contributions to the segregation literature. First, it argues by means of the Mutual Information or M index, which is neither invariant 1 nor 2, that both properties have strong implications, and it provides reasons to defend that the overall segregation index need not satisfy either one. Second, nevertheless, it is shown that in pairwise comparisons this index admits two decompositions into three terms. In the first decomposition, a term is invariant 1 and also satisfies a weak version of invariance 2. In the second decomposition, a term is invariant 2 and also satisfies a weak version of invariance 1. It is shown that these decompositions can be used to reach the analogous ones obtained in Deutsch et al. (2006).

Purpose – The Gini coefficient is a widely used measure of income inequality. It has also been used as a segregation measure, but only in the case of binary variables, for example race or gender. We develop a general version of the Gini Segregation Index (Gs) that can accommodate either continuous or binary variables, and discuss its relationship to existing measures.

Methodology – The Gini Index of Segregation is developed graphically and derived mathematically, illustrating the relationship between Gini's use in segregation and inequality applications.

Findings – Using the Public Use Microdata Sample for 25 U.S. metropolitan areas from the 2000 Census, we illustrate the calculation of the index and show that it is highly correlated with an existing measure of economic segregation.

Originality – This paper develops and illustrates a measure of segregation for continuous variables, a task for which there are few alternative measures.

This article axiomatically derives a class of numerical indices of integration (equality) in the distribution of different types of workers across occupations. The associated segregation (inequality) indices parallel one form of multidimensional generalized Gini inequality indices. A comparison is made with the other Gini-related segregation indices. A numerical illustration of the family of indices is also provided using US occupational data.

Purpose – This paper considers methods for decomposing indexes that incorporate economic disadvantage into a measure of segregation. According to such indexes, segregation in high-economic-status occupations is worse than similar segregation in low-economic-status occupations. The paper presents three decompositions of these indexes.

Methodology/Approach – The paper first characterizes a class of segregation indexes that include economic disadvantage in the index. It then develops mathematical methods for decomposing a change in such an index. The change is decomposed into two or more components: components that indicate either the effect of changes in economic disadvantage or the effect of changes in a standard measure of segregation – a measure that essentially ignores economic disadvantage. The paper then implements the decompositions using data on U.S. occupational segregation by gender between 1970 and 2000.

Findings – The primary finding is that a segregation index that incorporate economic disadvantage can be decomposed in interesting ways. A secondary finding is that such indexes indicate reduced segregation between 1970 and 2000. The dominant forces associated with the reduction were (a) the convergence of occupational gender ratios and (b) the movement of women out of less advantaged occupations and into the comparatively well-compensated professional and managerial occupations.

Research limitations/Implications – The 1970–2000 results are mainly illustrative. They are based on three broad occupational categories for which there were compatible earnings data, and the analysis could quite feasibly be done with more detailed occupational categories.

The basic premise of Hutchens's paper is that there are cases in which measures of segregation need to take account of the relative status of the groups into which members of a population are segregated. Segregation by occupation, Hutchens argues, is in some sense worse for a group if that group is segregated into lower status occupations. Hutchens proposes several measures that incorporate group status information, shows their properties, and works out their decompositions. I argue in this comment that the measures proposed by Hutchens have questionable utility in that they combine two fundamentally dissimilar types of information: a segregation dimension and a disparity dimension.

I want to thank Paul Jargowsky for his comment. I take it as a serious comment by a thoughtful scholar. I admit, however, that I cannot make sense of much of it.

Purpose – To develop measures of segregation that are appropriate when either the groups or the organizational units are defined by ordered categories. These methods allow the measurement of segregation among groups defined by ordered educational attainment categories or among ordered occupational categories, for example.

Approach – I define a set of desirable properties of such measures, develop a general approach to constructing such measures, derive three such measures, and show that these measures satisfy the required properties.

Originality – Traditional methods of measuring segregation focus on the measurement of segregation among groups defined by nominal categorical variables (e.g., race and gender) among organizational units also defined by nominal categorical units (e.g., schools and neighborhoods). Such methods are not appropriate to the measurement of occupational segregation, for example. The methods developed here are widely applicable and appropriate for such cases.

Purpose – Evidence suggests that during the 1990s, many US metropolitan areas saw fundamental changes in the spatial distribution of household income. Following two decades of increasing economic segregation, many metropolitan neighborhoods saw declines in economic segregation, particularly those neighborhoods located within central cities and rural areas. This paper adapts the Spatial Ordering Index proposed by Dawkins (2007b) to explore these trends.

Methodology/Approach – Using US Census data, I calculate economic segregation indices for a sample of 205 US metropolitan areas in 1990 and 2000 and decompose changes in the indices into portions attributable to changes in the spatial distribution of households and portions capturing changes in the spatial distribution of aggregate income. I also examine regional variations in the decompositions.

Findings – The results suggest that changes in the spatial distribution of households and of income each influenced metropolitan economic segregation in different ways during the 1990s. Furthermore, the spatial dynamics of income segregation exhibited significant regional heterogeneity.

Originality/Value of paper – This paper presents a new approach to measuring the dynamics of economic segregation.

Purpose – The aim of this paper is to analyze the changes that took place in occupational segregation by gender, nationality, and age in Switzerland during the period 1970–2000.

Methodology – The paper starts by using correspondence analysis to detect changes in occupational segregation by gender and nationality. It then generalizes a decomposition procedure originally proposed by Karmel and McLachlan by combining their approach with what is now known as the Shapley decomposition. Such a generalization offers a clear breakdown of the variation over time in occupational segregation into a component measuring changes in net segregation and another one corresponding to changes in the margins, the latter itself including variations in the occupational structure and in the shares of the subpopulations (e.g., the genders) in the labor force.

Findings – Between 1970 and 2000 there was a slight increase in gross segregation by gender but a decrease in net segregation. The change in gross segregation is because the change in the margins more than compensated that in the internal structure. But even the change in the margins is the consequence of opposite forces since variations in the occupational structure would have per se led to a decrease in gross segregation.

Originality – The results of the empirical illustration based on Swiss data for 1970 and 2000 prove the usefulness of the approach. They stress in particular that in several instances, variations in gross and net segregation worked in opposite directions.

Purpose – We analyze segregation between immigrants and natives at the firm level and explore the connection between segregation and wage inequality in Switzerland.

Methodology/Approach – Our approach accounts for the interaction between skill level and immigration status (work permit). First, we calculate exposure rates in order to analyze segregation at the firm level along these two dimensions. Second, we examine the role of segregation in the explanation of wage inequality between different skill–nationality groups. We use data from the Swiss Wage Structure Survey 2002, an employer–employee database that records individual wages among a very large sample of establishments in all industries, covering approximately 42,000 firms and 1 million workers.

Findings – Our results show that interfirm segregation is particularly pronounced for unskilled foreign workers and for recently arrived, highly skilled foreigners. The former earn lower wages than equally skilled Swiss workers, and the latter are paid higher wages than highly skilled Swiss workers. In both cases, interfirm segregation accounts for almost the entire wage differential.

Originality/Value of paper – This paper presents a generalization of the approach used by Groshen (1991) to the multigroup case by defining segregation with respect to the two dimensions of nationality and skill. The use of multigroup exposure rates is common in studies of neighborhood segregation (e.g., Bayer et al., 2004), but our paper shows that they can also be fruitfully applied in the analysis of interfirm segregation and wage inequality.

This paper analyzes the evolution of gender segregation in the workplace in Mexico between 1994 and 2004, using a matching comparisons technique to explore the role of individual and family characteristics in determining gender segregation and wage gaps. The results suggest that the complete elimination of vertical segregation would reduce the observed gender wage gaps by 5 percentage points, while the elimination of occupational segregation would have increased gender wage gaps by approximately 6 percentage points. The results also indicate that the role of occupational segregation in wage gaps has been increasing in magnitude during the period of analysis, while the role of vertical segregation on the determination of wage gaps has been decreasing.