Use an Alternative Measure of Inflation to Index Some Parameters of the Tax Code

Each year, the Internal Revenue Service (IRS) adjusts some parameters of the tax code on the basis of changes in the prices of goods and services, which generally increase over time, using the consumer price index for all urban consumers (CPI-U). The CPI-U, which is produced by the Bureau of Labor Statistics (BLS), is a monthly price index that is based on average prices for a broad basket of goods and services (including food and energy). It is designed to approximate a cost-of-living index. Adjusting, or indexing, certain tax parameters every year by the percentage change in the CPI-U is intended to keep their values relatively stable in real (inflation-adjusted) terms. (Inflation—an increase in the average price level—is a significant component in changes in the cost of living.) Among the tax parameters adjusted are the amounts of the personal and dependent exemptions; the size of the standard deductions; the income thresholds that divide the rate brackets for the individual income tax; the amount of annual gifts exempt from the gift tax; and the income thresholds and phaseout boundaries for the earned income tax credit and several other credits. Parameters for the individual alternative minimum tax (AMT) are also adjusted, including the exemption amounts, the income thresholds at which those exemptions phase out, and the income threshold at which the second AMT rate bracket begins.

Indexing is accomplished by adjusting a tax parameter's value in a base year by the percentage change in the CPI‑U between that base year and the most recent year for which the CPI-U is available. The annual period used for the calculation is not a calendar year but the 12 months that elapse from September to August. The value of the CPI-U in August becomes available in September, which allows the IRS enough time to index the tax parameters and prepare the necessary forms for the coming tax year. Adjustments in parameters of the tax code are calculated as follows: In the base year of 1987, for example, the standard deduction for a single tax filer was $3,000. Between 1987 and 2015, the CPI-U increased by 111.4 percent; correspondingly, the standard deduction (rounded to the lowest $50 increment) increased to $6,300 for 2016.

The CPI-U, however, overstates changes in the cost of living by not fully accounting for the extent to which households substitute one product for another when the relative prices of products change. To address that "substitution bias," BLS created another measure of changes in consumer prices—the chained CPI-U. Whereas the standard CPI-U uses a basket of products reflecting consumption patterns that are as much as two years old, the chained CPI-U incorporates adjustments that people make in the types of products they buy from one month to the next (thus "chaining" the months together). In addition, the standard CPI-U overstates increases in the cost of living because of a statistical bias related to the limited amount of price data that BLS can collect, which is known as "small-sample bias." The chained CPI-U does not have the same statistical bias. However, even though the chained CPI-U corrects for the substitution bias in the standard CPI-U and does not suffer from small-sample bias, neither the chained nor the standard CPI-U perfectly captures changes in the cost of living because neither fully accounts for increases in the quality of existing products, the value of new products entering consumers' baskets, or changes in where consumers make their purchases.

Under this option, the chained CPI-U would be used instead of the standard CPI-U to adjust various parameters of the tax code. The Congressional Budget Office estimates that the chained CPI-U is likely to grow at an average annual rate that is 0.25 percentage points less than the standard CPI-U over the next decade. Therefore, using the chained CPI-U to index tax parameters would increase the amount of income subject to taxation and result in higher tax revenues. Furthermore, the effects of instituting such a policy would grow over time. The net revenue increase would be about $3 billion in 2017 but would reach $32 billion in 2026, the staff of the Joint Committee on Taxation estimates. Net additional revenues would total about $157 billion from 2017 through 2026.

An argument in favor of using the chained CPI-U to adjust tax parameters is that this approach would more accurately reflect changes in the cost of living and modify each taxpayer's liability accordingly. The chained CPI-U provides a better measure of changes in the cost of living in two ways: by more quickly capturing the extent to which households adjust their consumption in response to changes in relative prices and by using a formula that essentially eliminates the statistical bias that can occur when estimates of aggregate price changes are calculated on the basis of relatively small samples of prices.

An argument against implementing this option is that only an initial estimate of the chained CPI-U is available on a monthly basis; a final and more accurate estimate is delayed because it is more complicated and time-consuming to compute than the standard CPI-U. (Details of that approach are available in a web-only technical appendix that CBO released with its February 2010 issue brief Using a Different Measure of Inflation for Indexing Federal Programs and the Tax Code.) At the start of every year, all of the initial estimates for the prior year are revised, and one year later those interim estimates are further revised and made final. Because of those delays, the initial and interim estimates of the chained CPI-U, which typically contain errors, would need to be used to index the parameters in the tax code. Since the chained CPI-U was first published in 2002, however, the changes between the initial and final values have been relatively small. If the adjustment for each year was based on the index value from an earlier base year, those small errors would not accumulate beyond the current year. Furthermore, because the initial and interim estimates of the chained CPI-U have been closer to the final version of the chained CPI-U than the standard CPI-U has been, those estimates still reflect the basic improvement attributable to the chained CPI-U.