Richard Rubin of the Wall Street Journal reports that the Trump administration is considering changing tax law so that capital gains would be adjusted for inflation. Under current policy, households owe taxes on the full nominal value of certain capital gains; this proposal would index the asset basis to inflation, leaving only the real value of any capital gain as taxable income. Our analysis suggests that this policy would cost $102 billion dollars over the next decade. While high-income households would benefit most, the share of taxes paid by AGI would not change meaningfully.
We use data from the IRS to impute length of time held for realized capital gains in the PWBM tax microsimulation module.1 This imputation allows us to disaggregate nominal capital gains into real and inflationary components. We then model a policy scenario where capital gains are indexed to chained CPI and compare the results to our baseline projections.
We estimate that such a policy would reduce individual tax revenues by $102 billion during the next decade, from 2018 - 2027. This estimate is likely a lower-bound cost estimate since we use conservative assumptions when imputing asset basis year. 2
Because income from capital gains is concentrated among high-income households, the benefits of this change would accrue primarily to the upper end of the income distribution. Table 1 shows that the top one percent of tax units would receive more than 86 percent of the tax cut, and that after tax-incomes would increase most for the top 0.1 percent. Overall, the policy, however, would not meaningfully change the distribution of tax burden.
|AGI percentile||Share of tax cut received||Percent change in after tax income||Share of federal tax burden|
|Current law||Tax cut|
Source: PWBM tax microsimulation model. Federal taxes include individual income taxes, corporate taxes, and estate taxes. Corporate income taxes are allocated under the assumption that owners of capital bear 75 percent of the tax burden.
This 10-year cost estimate of $102 billion ignores potential behavioral responses by investors. In response to the tax change, investors might also change the timing and frequency of capital gains realizations, and they might take advantage of new tax arbitrage opportunities created by leaving other forms of capital income unindexed.3 Calculating this behavioral impact rigorously would require keeping track of “vintages” of capital, which can quickly lead to a computational “curse of dimensionality.” PWBM is currently working on solving this type of high dimensional computing problem, and so future updates incorporating behavioral changes could lead to a larger or smaller score.
For example, length of time held asset data is top-coded at 20 years, so assets are never held for longer than 20 years in the model. ↩