Some Income-Measurement Issues and Their Policy Implications

Abstract

This paper discusses some major categories of nonpecuniary income: noncash benefits for lower-income households and for the elderly, and the imputed rent of owner-occupied housing. These are three of the 15 categories in the comprehensive income definition developed by Timothy M. Smeeding and Daniel H. Weinberg (1998). They are certainly important, and their measurement has been controversial. Government outlays on Medicare in 1998 totaled $190 billion; outlays on the three major low-income benefit programs (Medicaid, food stamps, and housing subsidies) were $145 billion. The rental value of owneroccupied housing is harder to measure, but larger; as of 1995 the market sales value of the stock may have been over $10 trillion (Arthur B. Kennickell and R. Louise Woodburn, 1997), and conventional rules of thumb in the housing industry yield an annual rental value of at least $1 trillion. The paper considers both conceptual issues and their policy significance. Economists and policymakers are most interested in three statistics of income: the well-being of the average American, the well-being of those at the bottom of the income distribution, and the overall distribution of income, usually measured as median household or family income, the poverty rate, and the Gini ratio. As many economists have noted, we as a society are most interested in how these measures change over time and differ between groups. Such comparisons provide the context for the current numbers. When the data are announced, the media immediately compare the current income and poverty figures to last year, the last cyclical peak or trough, or the all-time best or worst; and also compare households by race, ethnicity, and gender of household head. (The Gini ratio attracts less attention because it has no intuitive explanation for the layman.) In this paper, I focus on how the measurement issues affect these statistics.