I. Introduction

This report summarizes findings from an analysis of income disparities between whites and nonwhites[1] in the United States. It deals with present and past relative money income of whites and nonwhites at various points in the distribution of each. The empirical results are themselves of interest. They contradict many of the most familiar statements of the last few years on these subjects, in particular statements about the low and high ends of the nonwhite and white income distributions and their changes. Further, they provide a concrete context for clarifying and illustrating the meanings of some related but distinct goals in an area of urgent policy concern. These goals are frequently confused. Yet programs directed to one of these ends may do little or nothing to achieve another. Clarification here, then, can have a pragmatic value.

Section II treats current income disparities; Section III, changes over time. Section IV has to do with three major components of the income disparities: differences in age structure, years of schooling, and job distribution.

Recent Views

Some useful empirical studies have been done of relations between white and nonwhite income and some suggestive theoretical models of discrimination proposed. We have benefited from them. Nonetheless, not only the popular, but also the professional literature abounds in statements about the relations of nonwhite and white income distributions and their changes over time that do not seem to square with our own results (as summarized in Part V). Sometimes the contradiction is only apparent, but more frequently it is genuine. It is sometimes said, for example, that the income gap between the races is greatest at the lowest levels of the income distribution. And there has been up until very recently a rather widespread impression of decline or stagnation, especially at the bottom. Such impressions have been common not only among militants but among moderate civil rights advocates, and not only among polemicists and politicians in ephemeral pamphlets, they occur rather often in the writings of able economists and sociologists in the course of serious (and generally illuminating) studies. The following quotations are more or less typical:[2]

Current Race Differences Smallest at Top of the Distribution

"The degree of inequality of incomes of whites and nonwhites is related to position in the relative income distribution. Inequality is less at higher levels of the income distribution than at lower levels."
Guthrie, 1969

Inequality Among Nonwhites Increasing More than Inequality Among Whites

"The reality seems to be that some Negroes, especially those in the middle and upper income brackets, are gaining rapidly on whites, while others, particularly those in the slums, are losing ground in relative terms..."
Faltermayer, 1968

"...in terms of employment and income occupational status it is quite possible the Negro community is moving in two direc tions, or rather that two Negro communities are moving in oppo site ones. Obviously such a development would be concealed -- cancelled out -- in aggregate statistics that list all 'non- whites' together."
Moynihan, 1967

Absolute Decline

"the economic position of the Northern Negro deteriorates rapidly..."
Bayard Rustin, 1966

"It is a stark reality that the black communities are becoming more and more economically depressed."
Carmichael and Hamilton, 1967

"...there has been almost no change or change for the worse in the daily lives of most blacks."
Martin Duberman, 1968

"Although it is true that the income of middle-class Negroes has risen somewhat, the income of the great mass of Negroes is declining."
Whitney Young, Jr., 1968

Absolute Stagnation of Lowest Fifth

In talking about the period 1947-1966, the Kerner Report says, "About two-thirds of the lowest income group -- or 20 percent of all Negroes -- are making no significant economic gains despite continued general prosperity."
Kerner Report, 1968

Widening Gap Between White and Nonwhite or Lower Nonwhite to White Ratios

Title of an article comparing 1949 and 1959 incomes, "Decline in the Relative Income of Negro Men."
Batchelder, 1964

"The income gap between this country's whites and nonwhites is wide and getting wider."
McGraw-Hill's special report #2005, 1968

The Kerner Commission phrases its main comprehensive comparison between Negro (actually nonwhite) family income and white, "...Although it is growing, Negro family income is not keeping pace with white family income growth. In constant 1965 dollars, median nonwhite income in 1947 was $2,174 lower than median white income. By 1966, the gap had grown to $3,036."
Kerner Report, 1968

Not Closing, Stable Ratio

"[It is] startling...that there has been very little change in the ratio of nonwhite family income to white family income over the last decade and a half."
Rashi Fein, 1967

"Nor has discrimination declined. The average Negro family income has consistently remained near 55 percent of the white income. For a given average income level the white and black distributions have similar shapes, but the Negro distribution lags approximately thirty years behind the white. On the basis of relative measures such as these, discrimination has neither declined nor increased. In absolute terms, however, between 1947 and 1967 the average income gap between white and Negro families widened from $2,300 to $3,100 (in 1967 dollars). Relative measures indicate a stable pattern of discrimination, absolute measures a more intense pattern."
Thurow, 1969

"Negro incomes are growing as fast as white, but while Negroes are increasingly well-off absolutely, the gap remains nearly constant in relative terms."
Banfield, 1970

"...while the absolute condition of the Negro has improved markedly in the last decade, his relative position has improved barely perceptively..."
John Kain, 1969

Increase or No Decline after World War II, then Decline after Korea

[There was a]..."failure of the economy in the second postwar decade to match its performance in the first."
James Tobin, 1967

"...[the Negroes] maintained their wartime gains in the immediate postwar period..." "The Korean War prosperity... represented the apex of Negro prosperity, at least relative to white prosperity."
Arthur M. Ross, 1967

Not all of these views are based on studies and some of the studies have been affected by rather severe limitations. For the most part:
  1. Even where these writers talk of income distributions or make inferences that depend on the distributions, they have actually compared only medians or means; or where they have used distributions they have selected only a few points or used summary statistics (such as the Gini coefficient) that essentially enable comparisons only of inequality among nonwhites with inequality among whites, which is not the same as comparing inequalities in income between nonwhites and whites.

  2. They have focused on one or a few income series -- for example, family income or male income -- have sometimes drawn invalid inferences from these magnitudes to others -- for example, total money income to persons.

  3. They have sometimes drawn conclusions about personal income that really apply only to places or to job categories -- conclusions that do not apply to net figures covering both the starting and end points of migrations from one place or occupational category to another.

  4. In estimating changes over time they have sometimes used rather short time spans, or pairs of years sharply affected by the cyclical variability of nonwhite absolute and relative income.

  5. Some inferences that depend implicitly on assumptions about trends have been based on cross-sectional analysis at an arbitrary point of a cycle -- a point poorly located to sustain such inferences.

  6. For very early periods before income data by color are available, changes in nonwhite to white income ratios have been estimated by very coarse occupational indices that neglect nonwhite-to-white real income changes within occupations. This is sometimes done by analogy with production indices where prices are held constant; but in the present case changes in real income are a product of changes in occupational distribution and in real income within occupations. This second factor in the product has been ignored in these studies.

  7. They have sometimes used absolute dollar differences to draw welfare conclusions more appropriately based on percentages or logarithms.

  8. They have ignored sampling errors, even though these can be estimated from much of the data and are relevant for inference.

What This Part of Our Study Does

The part of our study summarized here has:
  1. Compared white and nonwhite income along the entire distribution of each and developed several statistical measures and graphic devices for this purpose.

  2. Used data, including unpublished data from the Current Population Surveys, to construct distributions of wage and salary income, of total money income to families and total money income to persons 14 years old and over, with several socioeconomic breakdowns of each; extended the personal-earnings and family income series of distributions back to 1945, and the wage and salary distributions to 1939; extended the continuous series of income to persons 14 years and over from 1953 back to 1948.

  3. Systematically used data on sampling errors varying for different points on a given distribution and at different calendar dates to estimate errors in our income ratios and invarious functions of them. These error estimates permit inference about the probable longitudinal and cross-sectional relations of various distributions.

  4. Estimated the effects on nonwhite relative income of differences in the distribution of whites and nonwhites by age, years of schooling, and major occupational categories.

  5. In the course of this work, explored briefly the relations to the data of various economic theories of discrimination in the marketplace, genetic explanations of income differences between whites and nonwhites, queuing models of the relative instability of nonwhite and white income during the business cycle, human capital theories of formal schooling and on-the-job-training as an investment process, a model of the dynamics of relative improvement in terms of migrations from one place, job category, or industry to a higher income place, job category, or industry (in which the movers come from the top half of the distribution at the point of origin but, to begin with, are in the lower half of the distribution at the point of destination).

    Other parts of this study completed or in process but not summarized here:

  6. Use unpublished Current Population Survey or Survey of Economic Opportunity data tapes to estimate the joint effects of various components of the disparity.

  7. Analyze the effects of redistributing nonwhites among detailed occupations.

  8. Use a new, generally useful, and powerful statistical measure of the relative internal inequality of two distributions to compare inequalities among nonwhites with inequalities among whites, and to estimate changes over time in such internal inequalities. (The measure used to compare inequalities within groups is a parameter of the same ratio-at-quantiles function useful for comparing inequalities between the groups.)

  9. Analyze and illustrate the relations and the differences among programs to reduce unemployment or to eliminate poverty or to reduce internal inequalities in society as a whole on the one hand and, on the other, programs addressed to the problem of eliminating inequalities of opportunity and income disparities between whites and the nonwhite races.

  10. Explore various political and sociological theories of relative deprivation and discontent.

Qualifications

Several disclaimers are in order. Although our study has been almost entirely concerned with income, this does not imply that we believe income disparities are the only important problem of race discrimination; nor that the only important problem of race discrimination; nor that the only such problems of importance are economic; nor even that they are all quantifiable. But clearly a broad range of social, political, and economic problems is reflected in income disparities, and income data are among the most complete and tractable of the available quantitative measures. Moreover, income is important. Among other things, nonwhites do want more income, and more income relative to whites.

Our primary data source has been the Current Population Survey (CPS), a monthly survey conducted by the Bureau of the Census. The CPS tabulations, currently based on a stratified sample of about 50,000 households, contain the most complete figures currently available. However, some qualifications regarding these data are still in order. First, there is the problem of under-reporting; whites have more of a principal kind of income that is under-reported -- self-employment income. On the other hand, some hold that nonwhite income is under-reported because more nonwhites, especially men, are missed by census interviewers. Then there is the problem of price differences in and out of the ghetto: these seem to favor whites, but there are no reliable estimates of the magnitudes of these differences. The partial exclusion of members of the armed forces introduces another unknown bias. We have considered at least in a gross way a variety of other possible sources of error (variation in access to public services, differences in imputed income from property, differing family structure and dependency ratios, differing age distributions of nonwhites and whites, the relative instability of nonwhite income, and, since the data used are money income before taxes, possible differences in effective taxes). On the whole, a rough evaluation suggests that the data understate the current nonwhite relative disadvantage; and also appear to understate the relative improvements over time. Inevitably, however, there are substantial uncertainties. These are problems that are introduced by the income data, and inferences about income inequality have to be so qualified.

There are three main income concepts used in the CPS tabulations: (1) wage and salary income, (2) earnings, and (3) total money income. Wage and salary income includes the income of employees, that is, regular pay, bonuses, tips, commissions, and so on. Earnings include wage and salary income plus self-employment income. Finally, total money income includes earnings plus rental income, interest, dividends, and transfer payments of all sorts such as welfare, social security, alimony, and others. We have dealt primarily with total money income in our study, since this is probably the best readily measurable income indicator of one's command over goods and services. Other types of income are important for certain classes of questions regarding racial income differences, but except where otherwise indicated, all of the references to income in this report mean total money income.

We shall refer mainly to two sorts of income receiving units: families on the one hand, and persons on the other. Income to families[3] is of substantial interest in considerations of welfare and consumer behavior. It has, however, a good many limitations, especially for comparing race differences. Family units at any given time vary greatly in the number of persons that compose them, in number of earners, and in the amount of doubling up (the combination of young married couples or of the elderly with family heads in the prime of life). Moreover, family structures vary as between nonwhites and whites. Family structures also vary over time for both nonwhites and whites. And these changes over time have differed as between nonwhites and whites. For example, though white women participate less in the labor force, since World War II they have increased their participation more than nonwhites. For such reasons it is important to deal not only with families as receiving units, but also with persons.[4]

[1]For the most part we have used annual data for whites and nonwhites since these are available since the 1940s. For purposes of the Census, the nonwhite population is primarily the black population (over 90 percent). Other nonwhites are mainly Orientals and American Indians. Orientals receive more and Indians less income than blacks. Both an upward and a downward bias then affect inferences from white-nonwhite income disparities to white-black differences. The biases do not substantially affect the largest aggregates but are important, for example, in regional comparisons: Orientals and Indians make up nearly half the nonwhites in the West, while in the South almost all nonwhites are black.

[2]For full citations, see bibliography.

[3]The term "family" as used here refers to a group of two or more persons related by blood, marriage, adoption, or residing together.

[4]See T. P. Schultz, 1965, p. 7, which stresses the variable size, structure, and wants of the family unit and therefore the greater ambiguity of family income data compared with data on personal income.

Contents
Next Chapter