Measuring the Efficiency Cost of Taxing Risky Capital Income

In this paper, we derive a measure of the efficiency cost of taxing risky capital income in an infinite horizon stochastic model. The resulting measure differs from all those that have been proposed in the existing literature. It can be represented by the expression -sigma(s) T(s)c(deltaX(s)), where T(s) measures the present value of the taxes that would be paid on a unit of investment in a riskless project with the same expected depreciation rate and tax treatment as capital invested in period s, X(s), while c(X(s)) represents the certainty equivalent to the representative individual of the lottery where measures the ex post change in investment in period s due to the tax change. The paper then compares this measure with others that have appeared in the literature. We were unable to find support for the argument in Bulow-Suinmers(1984) that the efficiency cost of taxing risky capital income is much larger than that implied by the measure -sigma(s)T(s)E(deltaX(s)). In fact, we show in special cases that our measure implies a smaller efficiency cost than does the measure -sigma(s)T(s)E(deltaX(s)).