Effects of taxation on household portfolio choice and risk taking

Abstract

Tax effects on household portfolio behavior have been a controversial issue. The intuition behind this controversy is that taxes reduce not only expected return but also risk. Since expected return and risk have opposite effects on asset demand, the net tax effects would depend on the relative sizes of these conflicting effects. Therefore, this paper aims at estimating the relative sizes of these conflicting effects. This paper also attempts to estimate a direct empirical relationship between taxes and household risk taking.First, this paper derives a system of portfolio share equations through a standard utility maximization with respect to mean and variance, utilizing a homothetic utility function. Then, the share equations are used to estimate utility parameters based on historical returns and flow of fund data for household sector. Then, using the estimated utility parameters, this paper calculates the relative elasticities of asset demands with respect to expected return and risk, and finally tax effects on share demands.Second, this paper develops a direct empirical model on portfolio risk in parallel to one of the benchmark models by Feldstein (1976). This paper estimates an overall measure of the household portfolio risk using the historical variance-covariances of asset returns and household-specific portfolio compositions. Then, this paper estimates tax effects on portfolio risk using the 1983 Survey of Consumer Finances (SCF) data.(1). The results of the estimated portfolio choice model suggest that taxation would encourage household to invest in equity assets and tax-favored assets. On the other hand, taxation would discourage household to invest in corporate bonds, federal bonds, and bank deposits. These results are consistent with the Domar and Musgrave proposition that taxation would encourage investors to invest in riskier assets.(2). The estimated portfolio risk model suggests that an increase in federal income taxation would encourage households to take more portfolio risk regardless of individual risk preferences. However, the sensitivity of household risk taking in response to a tax change increases as the household risk aversion increases. It implies that marginal disutility of risk is greater for risk-averse households than for risk-love households. Overall, the empirical evidence suggests that taxation would encourage the US household to invest more in riskier assets. It also suggests that taxation would induce US households into holding riskier portfolio.