Heritage Foundation Attack on Center
Rests on Misrepresentation
by Robert Greenstein
| View PDF version of this report If you cannot access the file through the link, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader. |
The Heritage Foundation recently issued a two-page broadside that attempts to discredit two Center analyses of issues related to the Administration's tax cut proposal. Debate and analysis on such issues is part of what democracy is about. But critiques of analyses that reflect a different point of view should be based on accurate representations of the analyses that are being criticized. The Heritage piece fails this test. It severely misrepresents the Center analyses it attacks. It consists more of ideological statements than serious analysis.
A complete discussion of the misrepresentations and unsupported statements in the Heritage piece is beyond the scope of this brief reply. (Such a discussion would end up being considerably longer than the Heritage paper.) A few examples, however, provide a flavor of the nature of the Heritage piece.
One of the analyses the Heritage piece castigates is a Center report by Isaac Shapiro and Joel Friedman which examines IRS data that show substantial increases in after-tax income disparities between 1992 and 1998 (and also between 1989 and 1998). Heritage writes: "Shapiro and Friedman badly misuse data to create a statistical mirage of growing income inequality in America from 1992 to 1998. The Census Bureau warns researchers not to do this because of major survey changes in 1994. In fact, the Census Bureau (using directly comparable data) reports that income inequality has not increased from 1993 to 1999." To a reader not familiar with the data sources, this sounds like a damaging indictment. To the contrary, in just three sentences, Heritage commits a series of errors any careful analyst would avoid.
- In fact, the Census Bureau cautions researchers about using the Census Bureau's Current Population Survey to measure trends in income disparities over this period, because of changes in that survey in 1994. But the Center's analysis did not use this survey; it used IRS' Statistics of Income data, for which there was not a change in 1994. The IRS data are based on actual income tax returns. It is difficult to understand how Heritage could have been confused about the data the Center used; the Center's analysis begins by explicitly citing its use of IRS data.
- Heritage states that the Census Bureau reports that income
inequality has not increased from 1993 to 1999. The Census Bureau has explained, however,
that it lacks reliable data on the incomes of those at the top of the income scale, both
because its official measure of income does not include income from capital gains and
because (for confidentiality reasons) it records only part — rather than all —
of the income of individuals who are at very high income levels. As a result, the Census
survey is unable to provide reliable data on changes in the incomes of those at the top of
the income scale or changes in income disparities between those at the top of the income
scale and other U.S. households.
Virtually all researchers in the field recognize this weakness in the Census data and concur that IRS data provide the best information on changes in incomes among those at the top of the income scale. Indeed, since the late 1980s, the Congressional Budget Office has been using the same IRS data that the Center's analysis uses to track changes in income at the top of the income scale and merging it with Census data to look at income changes for the entire population. These CBO analyses and projections show the same type of trend of increasing income disparities over the 1990s that the Center's analysis discusses. The Heritage piece fails to mention these CBO analyses.
Heritage is not unaware of CBO's work on this matter. Last March, CBO held a one-day meeting of analysts to discuss methodological issues related to these income measures. Analysts at the meeting agreed once again that the IRS data need to be used and provide the best measure of income and income changes at the top of the income spectrum. A Heritage analyst attended and participated in the meeting.
In short, Heritage knows better. Its broadside — which is both inaccurate in its descriptions of the Center analyses and disingenuous in its discussion of data sources — reads more like the type of material that a political campaign might issue to try to tar an opposing candidate than a serious discussion of analytic or policy disagreements among analysts with differing perspectives. Ideological zeal should not take precedence over careful analysis.