The Effect of Dividends on Consumption

In classical models the division of stock returns into dividends and capital gains has no "real" consequence for investor consumption. This paper, using two micro data sets that provide cross-sectional variation in dividend receipts and capital gains, empirically measures the effect of dividends on investor consumption. Analysis of data from the Consumer Expenditure Survey indicates that household consumption is particularly sensitive to realized dividend income, when one controls for total portfolio returns including dividends. Analysis of data from a discount brokerage shows that dividends are withdrawn from household portfolios at a much higher rate than capital gains, further suggesting that the form of returns matters for consumption and that investors pursue a mental accounting strategy to "consume income, not principal." Finally, the paper discusses what these estimates imply for the response of aggregate consumption to the May 2003 dividend tax cuts in the United States.