(Sep 16, 2003) T030000 Interpreting Dividend Relief Estimates
Estimating the effects of "dividend" tax cuts requires some caution. Currently, "dividends" reported on tax returns can include not only dividends on stocks earned either directly or passed through from mutual funds, but also short-term capital gains and interest earned on money market mutual funds. Draft legislation explicitly states that short-term capital gains do not qualify for the exclusion, but is mute on the question of interest payments. This may be intentional or it may be an oversight. On policy grounds, there is no justification for providing relief from double taxation to interest earned on bonds, because interest is a deductible expense for corporations and clearly not subject to double tax.
Attached are two sets of revenue estimates and distribution tables for the proposal to eliminate the tax on dividends received by individuals. The distribution table shows that cutting taxes on all "dividends" reported on tax forms is regressive. Not surprisingly, 42 percent of the benefits go to the top 1 percent of taxpayers; 53 percent go to the roughly three percent with incomes above $200,000. These estimates probably understate the regressivity of a tax on actual dividend payments because mutual fund holdings are likely to be more concentrated among middle class households than dividend payments. We do not have the data to make this adjustment, though.
We present two revenue estimates, one that limits the tax relief to actual dividends ($278 billion) and a second assuming all currently reported "dividends" would be tax-free($473 billion). WHICH ONE OF THESE ESTIMATES TURNS OUT TO BE CLOSER TO THE JCT'S ULTIMATE ESTIMATES DEPENDS ON WHAT THE ACTUAL DIVIDEND EXCLUSION PROPOSAL TURNS OUT TO BE AND WHAT THE DATA ON THE DISTRIBUTION OF DIVIDENDS, MUTUAL FUND INTEREST PAYMENTS AND SHORT-TERM CAPITAL GAINS TURN OUT TO BE.
Also note that the estimates are static—they do not account for behavioral responses or asset price effects that would be reflected in official revenue estimates. For example, exempting dividends from tax should cause asset values to increase, which would generate taxable capital gains in the near term. If companies pay out more of their earnings in dividends, however, the value of the stock will rise more slowly, which implies less taxable capital gains over the long run. The proposal could also affect the amount of debt held by companies and interest rates. On balance, behavioral responses will tend to increase the revenue cost over the long term, but could reduce it over the near term.
More detail: There is no direct measure for the portion of dividends reported on tax forms that is actually interest. Our first revenue estimate is based on a calculation done by Bill Gale that $142 billion were reported as dividends and $62 billion of that were interest payments from mutual funds in 2000. Thus, we reduced dividends across the board by 62/142 (44 percent) and assume that amount continues to be taxable as interest. This is pretty crude, and still doesn't correct for whatever amount of reported dividends is actually short-term capital gains. We have no data on that right now.
Our second estimate reports the effect of exempting all dividends as currently reported from taxation. This probably overstates the static revenue cost (because we haven't subtracted short-term capital gains), but underestimates the long-term cost after allowing for behavioral response if interest from mutual funds is really excludable. In that case, there would likely be a huge increase in demand for tax-exempt money market funds. Note that this would also wreak havoc on the tax-exempt bond market. For that reason, we think a proposal that excludes interest from mutual funds is a nonstarter, but the draft legislation is unclear on this.