A Rejoinder to Rector et al.'s "The Extent of Material Hardship and Poverty in the United States"

Abstract

Marc I. Roemer [1] In their article Extent of Material Hardship and Poverty in the United States, Rector et al. make some compelling points about the difficulties defining and measuring income and poverty. [2] To their credit, they recognize some of the differences between two alternative sources of income data, the March Current Population Survey (March CPS) of the Census Bureau and the National and Product Accounts (NIPAs) of the Bureau of Economic Analysis (BEA). However, the authors mislead the reader when they characterize the discrepancy between March CPS Income and the NIPAs as indicating the level of accuracy of the March CPS income data. They refer to the discrepancy as the of the survey, and conclude from the discrepancy that the March CPS dramatically and consistently under reports the economic resources of households. Reporting is an incorrect characterization of the discrepancy, because the income measures are not directly comparable. The March CPS attempts to measure cash income that people received regularly during the previous year. The Personal and related series of the NIPAs reflect a more comprehensive definition of income, and include noncash payments such as employer health insurance contributions and health insurance benefits. As the March CPS does not aim to measure many of the components of income contained in the NIPAs, it is misleading to assert on the basis of unadjusted NIPA figures that the March CPS contains extreme reporting shortfalls. Among the components of Personal that are not included in the Census Bureau's Money concept are employer contributions to private pension and welfare funds; capital consumption and inventory valuation adjustments to farm and nonfarm self-employment income; the rental value of owner-occupied homes; imputed interest from banks, credit agencies, investment companies, life insurance carriers, and private noninsured pension plans; benefits from hospital and medical insurance; public assistance medical care; business transfer payments; interest, dividends, rent, proprietorship income, and partnership income paid to fiduciaries and nonprofit institutions; unredeemed interest on U.S. savings bonds; small corporation income; and lump sum payments. Some of these items are quite large. The coverage universes differ substantially as well. Some population groups are included in the NIPAs but not the March CPS, such as people living in institutions, those who received income in the preceding calendar year but die before the interview date, those living overseas, and military personnel living on a U.S. post without family. The purpose of this rejoinder is not to argue whether these components and population groups should be included in an income concept for measuring poverty as Rector et al. suggest, but rather to present a more accurate assessment of the quality of the March CPS income data than they provide in their article. Table I shows a more correct comparison of the NIPA and Census Bureau income measures. Here, the NIPA figures are adjusted to fit the March CPS income definitions and coverage universe. The resulting differences between the income measures are only a fraction of the shortfalls reported by Rector et al. Of particular concern to Rector et al. is the shortfall in government transfers of US$579 billion that they report. Taking the differences in income definition and coverage universe into account decreases the shortfall to only US$51 billion. The smaller shortfall results primarily from excluding medical and hospitalization benefits, food stamps, the earned income tax credit, payments to nonprofit institutions, energy assistance, business transfer payments, and the income of people who are outside the March CPS sampling frame. …