Coase-Sandor Working Paper Series in Law and Economics
Inflation indexing is an important and controversial issue in the design of tax systems and transfer programs. The choice of whether—and how—to adjust policy parameters for inflation carries significant political, distributional, and macroeconomic implications. In recent years, indexing has gained particular attention in three policy contexts: (1) whether to switch from an “unchained” to “chained” inflation index when calculating Social Security benefits; (2) whether to make a similar unchained-to-chained shift when setting federal income tax parameters such as bracket thresholds and deduction amounts; and (3) whether to index basis for inflation when calculating capital gains. Hundreds of billions of dollars ride on the resolution to these three questions.
Across all of these contexts, the debate over inflation indexing is generally framed in terms of “accuracy.” When the National Commission on Fiscal Responsibility and Reform chaired by Alan Simpson and Erskine Bowles recommended in 2010 that Social Security cost-of-living adjustments be calculated using the “chained” Consumer Price Index (CPI), the commission emphasized that chained Consumer Price Index (CPI) is “a more accurate measure of inflation.”1 The Center for a Responsible Federal Budget, a Washington, D.C.-based think tank that picked up the chained CPI mantle after the Simpson-Bowles commission dissolved, likewise listed accuracy as its primary justification for chaining: “[P]olicymakers should ensure that the most accurate measure of inflation is being used,” the group declared in a white paper, and “[a]n overwhelming majority of economists from both parties agree that the chained CPI is a far more accurate measure of inflation than the CPI measurements currently in use.” 2 When the Obama administration proposed a switch to chained CPI across federal tax and transfer programs in 2013, it foregrounded the “accuracy” argument as well.3
The case for capital gains indexing has proceeded on similar premises. For example, Reed Shuldiner—in a comprehensive and insight-packed 1993 article on indexation—argued that computing capital gains without adjusting for inflation produces an “inaccurate” result.4 The congressional Joint Economic Committee, in a 1999 report, similarly said that indexing is necessary in order to “measure capital gains correctly.”5 Lawyers Charles Cooper and Vincent Colatriano—in a 2012 article urging the Treasury Department to index capital gains for inflation via executive action—wrote that capital gains indexation would “more accurately assess the actual increase in a person’s wealth or purchasing power.”6 Accuracy-based arguments for capital gains indexation sprung to life again in 2019 when President Trump asserted that he had the power to index capital gains for inflation of his own accord.7
Critics of chained CPI and capital gains indexation have joined issue on the accuracy point. The AARP, which opposes the use of chained CPI for Social Security cost-of-living adjustments (COLAs), has argued that chained CPI is “even less accurate than the current formula.” 8 Hundreds of economists who signed a letter opposing the use of chained CPI for Social Security in 2012 agreed that the annual Social Security COLA “should be based on the most accurate measure possible of the impact of inflation on beneficiaries,” but disputed that chained CPI was the best way to achieve that goal.9 More recently, in the debate over whether the Trump administration should index capital gains for inflation via executive action, critics of the move have argued that indexing capital gains, but not other elements of the tax code, would lead to the “mismeasurement” of income—a direct counterpoint to the “accuracy” claim pushed by proponents.10
This Article argues that—across all three of these indexing debates (and several more)— the emphasis on “accuracy” misses the mark in two respects. First, inflation is not a quantity that exists in the world apart from how it is measured. It is not like the distance from London to New York, which can be measured accurately or inaccurately. To say that chained CPI is more “accurate” than unchained CPI is something like saying that a U.S. liquid pint measure is more “accurate” than an imperial pint measure. We may have good reasons for using a U.S. liquid pint rather than an imperial pint—or vice versa—but “accuracy” is not one of them. Likewise, we may have good reasons for caring more about the month-to-month change in the price of a fixed basket This Article argues that—across all three of these indexing debates (and several more)— the emphasis on “accuracy” misses the mark in two respects. First, inflation is not a quantity that exists in the world apart from how it is measured. It is not like the distance from London to New York, which can be measured accurately or inaccurately. To say that chained CPI is more “accurate” than unchained CPI is something like saying that a U.S. liquid pint measure is more “accurate” than an imperial pint measure. We may have good reasons for using a U.S. liquid pint rather than an imperial pint—or vice versa—but “accuracy” is not one of them. Likewise, we may have good reasons for caring more about the month-to-month change in the price of a fixed basket.
Second, and more importantly, the adjustment of policy parameters over time is not, at its core, a question of technical accuracy. How Social Security benefits ought to change year to year, how the schedule of tax rates ought to change over time, and whether inflationary gains ought to be included in the tax base are not questions of measurement. They are, instead, value judgments. “Accuracy” in this context turns out to be both an illusion and a distraction.
Seeing through the mirage of “accuracy” is important not only because it offers a clearer-eyed view of the values at stake in indexing debates, but also because it opens up broader vistas for tax and transfer policymaking. For example, rather than focusing on whether “unchained” or “chained” versions of the Consumer Price Index provide more “accurate” measures of inflation, we might ask whether pensioners and disabled adults ought to share in the gains from economic growth. An affirmative answer to the latter question would suggest that Social Security benefits ought to be tied to an index that tracks overall economic changes (e.g., nominal gross domestic product) rather than an index that tracks only price-level changes (e.g., unchained or chained CPI). Likewise, instead of a crimped choice between unchained and chained CPI for tax bracket thresholds and deduction amounts, we might imagine tying tax system parameters to deficit levels or business cycle measures. And instead of an argument about capital gains indexation framed in “accuracy” terms, we might imagine a more direct discussion about whether (and how much) the income tax should operate as a tax on wealth.
Each of these questions will require more than this short Article to answer. The modest goal here is to show why indexing decisions ought to be “unchained,” so to speak, from a narrow focus on “accurate” measures of inflation. Part II introduces the indices according to which tax and transfer parameters are adjusted and the contexts in which indexation questions arise. Part III—the heart of the Article—presents the case against “accuracy” as an objective for parameter adjustment. Part IV considers implications of this argument for participants in policy debates and for scholars engaged in the law and macroeconomics enterprise.
Daniel J. Hemel, "Indexing, Unchained", Coase-Sandor Working Paper Series in Law and Economics, No. 901 (2020).