Sensitivity of Individual Income Tax in the United States, 1979-2004

Partly following the research done by Blackburn (1967), Musgrave and Musgrave (1984), Musgrave and Thin (1948), Pechman (1973), Ram (1991), and Tanzi (1969, 1976), sensitivity of individual income tax in the United States is studied for the period 1979-2004. Besides a brief discussion of the major tax-law changes and description of various tax-income ratios, ten points are noted. First, elasticity of tax revenue with respect to adjusted gross income (AGI) shows a sizable fall from 1979 to 1989. Second, there is a fair recovery in the elasticity from 1990 to 2001. Third, these variations reflect the tax-law changes during the Reagan-era of the 1980s and Clinton's initiatives of the 1990s. Fourth, significance of the income distribution variable (Gini) in the tax equation noted by Ram (1991) almost disappeared by 1989, which may also indicate an erosion of the tax progression suggested by fall in the tax elasticity. Fifth, corresponding to the recovery in the size of the tax elasticity, significance of the Gini-parameter tended to improve during 1991-2004, but its magnitude did not reach the levels observed for 1979-1980 and earlier periods. Sixth, the relatively steady magnitude of the tax elasticity from 2001 to 2004 may indicate that the possible erosion of tax progressivity due to the tax laws enacted during 2001-2004 may not yet have been fully captured in the data. Seventh, while estimates for years up to 1992 show that omission of the income-distribution term (Gini) causes the tax elasticity to be understated, the pattern since the mid-1990s is the opposite, apparently due to a change in the sign of income-Gini correlation. Eighth, estimates of Tanzi-type (1976) elasticities of "rate structure" and "tax base" indicate an expected pattern. Ninth, Tanzi-type "built-in flexibilities" also show an expected pattern that corresponds to the ratios of tax to AGI, tax to taxable income, and taxable income to AGI. Tenth, measures of average-rate progression, defined by Musgrave and Musgrave (1984, p. 362), for selected years indicate that while the expected decline with AGI is observed, the progression is negative across the two top AGI groups for 1999 and 2004. Four additional points are also noted. First, as in Ram (1991), the tax-model that is quadratic in income does not do well. Second, estimates based on state-level Gini from census and annual surveys look fairly similar, but Gini parameter tends to be weaker in the latter. Third, cross-section elasticity estimates from AGI classes, of the kind used by Blackburn (1967) and others, seem less stable and informative than Tanzi-type estimates from state-level data. Fourth, along with the well-known increase in inequality since the late 1970s, the correlation between Gini and AGI (and tax) is negative until 1992, but turns positive in later years.