We consider the optimal taxation of capital if government is allowed to differentiate taxes on the basis of an asset’s risk class. We separately consider wealth taxes and the joint taxation of both wealth and capital income. We find that taxes on normal returns (be it wealth or income taxes) should be higher for risk-bearing assets if individual risk aversion is declining with earnings ability. We show that the insurance motive to tax excess returns to risky assets does not interfere with the redistributional motive to differentiate taxes on normal returns. However, a tax on excess returns is shown to also have a redistributive role in our setting, as long as taxes on normal returns are differentiated.

The answer is: we should tax capital income, because taxes on capital income impose a non-distorting tax on the risk premium, whereas taxes on wealth do not. This conclusion is derived by analyzing taxes on capital income and wealth in a Merton-Samuelson multiple-period portfolio model with safe and risky assets. Tax reforms are analyzed where taxes on capital income are increased and taxes on wealth are lowered. Such reforms are unambiguously welfare improving with idiosyncratic risk and also welfare improving with aggregate risk if public goods provision is not too inefficiently large. Optimal taxes on capital income and wealth are also derived. Taxes on capital income are used to tax the risk premium, while (negative) taxes on wealth should ensure intertemporal efficiency, which would boil down to a tax a rate of return allowance, joint with a tax on capital income.

Recently, there have been calls for the introduction of a wealth tax. However the major concern is that a wealth tax induces evasion. This paper theoretically characterises the optimal wealth tax in a general equilibrium model with tax evasion and financial frictions deriving a new optimal tax formula consisting of three components. First, a redistribution term, reflecting the welfare gains from transferring resources toward agents with lower. Second, a general equilibrium-effects term. Third, a behavioural-effects term including the key elasticity of evasion with respect to taxes which depends on the curvature of the evasion costs and on the extent to which hidden wealth can be collateralised. The numerical exercise quantifies two main results. First, the higher the share of hidden wealth that cannot be collateralised, the lower the elasticity of evasion. Second, as the share of hidden wealth that cannot be collateralised increases, the planner chooses a higher tax rate.

Entrepreneurial taxation differs from standard Mirrlees income taxation because the entrepreneur’s effort is an input into her own production. As a result, taxing entrepreneurial income distorts not only effort but also the firm’s production plan and the demand for other inputs. This paper studies optimal nonlinear taxation when entrepreneurs differ in privately observed skill and the government observes only entrepreneurial income and capital. I derive conditions under which production efficiency with respect to capital is desirable and when a second-best capital tax (or subsidy) should accompany the income tax to offset incentive-driven distortions. With effort-augmenting skill and weak separability, a production-side analogue to Atkinson–Stiglitz implies that capital should not be taxed when capital demand does not vary across types at a given income. With capital-augmenting skill, the sign of the optimal capital instrument depends on technology curvature: declining capital elasticities favor a capital subsidy.