Women and Minorities in Federal Government Agencies: Examining New Evidence from Panel Data

Abstract

The employment of women and racial/ethnic minorities in the public work force has grown considerably in the past 20 years, and projections have suggested that such growth can be expected to continue into the immediate future (Johnston et al., 1988). Already, die federal civil service has, in the aggregate, become remarkably reflective of the nation's population. Women comprise more than 43 percent of the total federal civilian work force while African Americans hold 16.6 percent of those positions, and Hispanic representation stands at 5.4 percent (U.S. Office of Personnel Management, 1990a). This level of integration has, however, failed to reach all segments of the public bureaucracy. Women and minorities remain concentrated in lower-echelon jobs. Much attention has focused recently on that issue and on ways to ensure that equal opportunity reaches the higher grades in the public service;(1) yet, there is considerable variation across federal agencies in the overall percentage of jobs held by women and minorities and the rate of growth of minority and female employment. These differences are illustrated in Tables 1 and 2, which report employment shares for women, African Americans, and Hispanics in 30 selected agencies for the year 1988, and the percentage increase in employment shares for each group from 1982-1988, respectively. (The term is used to refer to departments as well as independent agencies.) [TABULAR DATA OMITTED] The empirical issue addressed in this article is the extent to which interagency differences in female and minority employment shares, as well as share changes, are systematically related to the distribution of occupations within an agency and an agency's demographic and organizational characteristics. A distinctive feature of this analysis is the use of panel data, a pooled cross-section/time-series see grey box on page 266.) Previous empirical work has relied amost exclusively on crosssectional econometric approaches. The availability of panel data allows for control of the passage of time and of unobserved time-invariant agency characteristics that may be correlated with explanatory variables in subsequent regressions. Such control is not possible in standard cross-sectional analysis, and as a result, coefficient estimates from such models may be inconsistent. Agency Characteristics and the Employment of Women and Minorities In any study of determinants of government employment integration along gender, racial, and ethnic lines, attention should be directed to the distribution of occupations across agencies. Traditionally, women and minorities have been over-represented in clerical positions and under-represented in professional, administrative, and technical job categories (Saltzstein, 1986; Mladenka, 1989; Kellough, 1990; and Kellough and Elliott, 1992). As a result, it can be expected that the estimated effect of the percentage of clerical positions on the employment of women and minorities will be positive. Similarly, because of the over-representation of African-American and Hispanic employees in blue-collar positions, the prevalence of blue-collar positions in an agency should be positively associated with African-American and Hispanic employment. By contrast, female and minority employment should vary inversely with the percentage of professional, administrative, and technical occupations in an agency. Apart from occupational variation, however, federal agencies also differ dramatically in terms of their size or total employment. Grabosky and Rosenbloom (1975) provided early evidence that minority employment shares are negatively related to agency size. One explanation for this relationship is that each additional minority (or female) hire makes a smaller contribution toward diversity in large agencies; such agencies must employ a larger number of minorities to achieve the same racial/ethnic composition as smaller agencies. …