Measuring Tax Fairness The Problems with Tax Distribution Tables Presentation to the President’s Advisory Panel on Federal Tax Reform New Orleans March 23, 2005 William W. Beach Director, Center for Data Analysis The Heritage Foundation Washington, D.C. What is tax fairness? • In many respects, tax fairness is similar to the concept of fairness in other aspects of U.S. law. – Equal treatment – Transparency – Continuity of rules and practices • Fairness in the tax realm certainly means that everyone pays their fair share. That could mean that taxes are proportional to consumption, to income, or to some other factor that measures our use of government. • Fairness also implies “forward equity,” since tax policy today frequently shapes the future. 2 Why consider “forward equity” • Economists frequently talk about horizontal equity (that equals will be treated equally) and vertical equity (that, for example, tax burden rises with income). • Lawmakers, however, need to recognize that most of their decisions will affect relatively distant future acts rather than today’s activities or today’s income or tax distribution. • That being so, lawmakers should consider whether policy change facilitates individual economic, social, and personal choices that set in motion a sequence of activities that lead to goals a person sets for him or herself. – For example, do tax policy changes made today raise barriers to women re-entering the workforce years from now after raising a family, or to immigrants starting micro-businesses, or to retirees pursuing part-time work? – Do policy changes make it more or less difficult for young people to achieve their goals? 3 How would you measure fairness? • One of the goals of distribution analysis is to show how policy change affects the economic well-being of taxpayers and non-taxpayers. The problem, however, is deciding how to measure the relationship between tax policy and economic well-being. • Unfortunately, we cannot measure all of the things that affect a taxpayer’s well-being. Thus, we settle on proxies for those data we cannot obtain or activities we cannot measure. – The most common way of measuring fairness is to analyze changes in income. – However, what is income? • What is spent on all goods and services including leisure? • Net worth? • Cash and non-cash compensation? – Even if we settle on a concept, how good are the data? 4 The neat classification of taxpayers by income breaks down when we look at their tax liability in each quintile. 5 Why does the neat classification of taxpayers by income so easily break down? • The previous chart shows that taxpayers who are grouped into quintiles by income have vastly different tax liabilities. • Millions of taxpayers in the third quintile pay more in taxes than those in the fourth quintile; there are millions in the fourth that pay more than those in the fifth. • These differences stem from many factors that are not revealed by simple income distribution: family businesses, differences in family size, investments, access to itemized deductions, and so forth. 6 What about distributing tax policy effects by consumption? • Avoids many of the definitional problems surrounding “income,” even though short-run and long-run consumption (housing, education) muddy the data waters. • Over the course of an individual’s life, consumption tends to follow income change: incomes are low in youth, rise to a peak in middle age, and fall again in retirement. Generally, consumption follows a similar pattern. – However, hard to study an income tax by looking at consumption. – Significant data problems, especially with the Consumer Expenditure Survey. 7 How much can we learn about the complexity and equity of our tax system from marginal tax rates ? • If a primary fairness goal is steadily higher tax rates as income rises, then looking at marginal tax rates after a policy change may be a good metric for fairness. • However, targeting tax policy at specific groups in order to achieve fairness often frustrates the analysis of whether the overall tax system is more just. • As Kevin Hassett of AEI shows in the following dramatic graphic, current marginal tax rates on the income of a family of four shows distinctly unequal treatment as their income rises.* *Further information can be found at http://www.aei.org/publications/pubID.22160/pub_detail.asp 8 Marginal Tax Rates for 2004 (Drawing by Marina Sagona based on graph from Kevin Hassett, American Enterprise Institute) Is this just? The marginal tax rates of a family of four as income rises. (Graph courtesy of Dr. Kevin Hassett of the American Enterprise Institute) 10 Snapshot Analysis • When economists create a cross-section of taxpayers, they have grouped these people by income and, usually, family type at a moment in time. It truly is a snapshot of our tax system. • This snapshot analysis is done differently at Congress’s Joint Committee on Taxation and the Treasury Department’s Office of Tax Analysis, the two agencies most responsible for tax policy analysis. • These two agencies use snapshot analysis of the policy change implementation using projected historical data. 11 Snapshot Analysis (continued) • The groupings are based on historical data that have been changed to reflect the increase in income and taxpayers that forecasting models predict. But, these groupings commonly are not affected in all of the ways that tax policy changes can impact the economy. • The next three graphs show how JCT and OTA distribute tax changes across income. JCT uses an income concept most taxpayers would recognize. OTA used to employ an income concept that only an economist would appreciate. It has recently used an income concept more like that of the JCT. 12 The JCT distributes taxes against a measure based on Adjusted Gross Income (an expanded version of AGI) Analysis of the Bush 2001 tax cut proposal 13 The OTA, however, used to adjust the JCT income concept to include “imputed income” from things not commonly taxed. 14 The new OTA distribution method uses an income concept closer to the Joint Committee’s Analysis of the Bush 2001 tax cut proposal 15 But, do these approaches tell us what we really want to know about tax policy change? • Answering fairness questions by looking at snapshot changes in tax liability does not tell us what happens to individual taxpayers living in a society reshaped by tax policy change. • What we really want to know is how certain types of tax policy changes affect long-term social and economic outcomes. • To answer those types of questions, we need “longitudinal” tax analysis. An example of excellent longitudinal work is the Congressional Budget Office’s “Effective Tax Rate: Comparing Annual and Multiyear Measures” (January, 2005). • In the following graph, the number in bold along the diagonal represents the percentage of taxpayers who were at that tax rate in 1987 and remained there ten years later. The chart shows significant income mobility. 16 Follow taxpayers not classifications over time 17 The Center for Data Analysis is taking longitudinal analysis one step further • The CDA uses a rich database of taxpayers, taxpaying families, and households to analyze tax policy change. • We employ standard projection techniques to create a ten-year panel of taxpayers. • However, we also shape each year’s cross-section a dynamic simulation of how tax policy affects those economic factors that shape income, job creation, and other key indicators. 18 Summary • Fairness questions are natural and inevitable features of tax policy debate. • Tax and economic data as well as definitions of income poorly serve those seeking answers. • Snapshot analysis probably produces more confusion and bad information than clarity. • Following taxpayers over time should provide more insight on fairness than other analytical approaches. It is important to – Include more than one year of income in our analysis – Continue to move beyond snapshot analysis – And, wherever possible, incorporate the impact of tax policy on the economy and, thus, on the pool of income and economic decisions that compose the base from which taxes are drawn. • Note: Many of the slides used in this presentation were taken from Jason Fichtner’s excellent essay, “A Comparison of Tax Distribution Tables: How Missing or Incomplete Information Distorts Perspectives,” The Heritage Foundation Center for Data Analysis Report CDA04-13, November 9, 2004. 19