AN INVESTIGATION OF THE SHORT RUN EFFECTS OF CAPITAL GAINS ON HOUSEHOLD CONSUMPTION AND SAVING*

Abstract

THIS DISSERTATION EXAMINES one of the most controversial channels of monetary policy, the wealth effect, insofar as it affects household saving and consumption as modeled by the MPS model. Using cross section survey data, it is possible to provide an independent alternative estimate of the impact of changes in wealth on consumption for the household sector from those estimated exclusively through the use of time series data. The theoretical and empirical literature on the consumption function is reviewed and the shortcomings of the time series approach are examined insofar as they would affect the results reported in the published studies. The short run impact of capital gains on consumption is likely to be biased substantially towards zero due to measurement error. Thus, an alternative approach is justified. This dissertation employed a theoretical model and a cross section survey data set, the latter with some success. The data employed is from the Federal Reserve Board Survey of Changes in Family Finqnces and the Survey of Consumer Characteristics, which provided detailed information for 2159 households on income, saving and wealth (as well as many other variables of interest). The wealth data was highly detailed including the exact number of shares of each stock held by each household. Given such information, it is possible to estimate reliably capital gains (largely unrealized and unanticipated) on a major and the most volatile portion of household wealth. Thus it was possible to estimate the impact of 1963 capital gains, as they were distributed over the year, on 1963 saving. The regression results imply that between 2.5 and 3.2 per cent of 1963 capital gains found their way into 1963 spending streams. This estimate of the impact of unanticipated capital gains is only slightly below the impact implied by the latest unpublished version of the MPS model, the only model which incorporates a wealth effect. It is, however, substantially larger than all the other time series studies which have attempted to estimate a short run capital gains effect. These results were found to be robust with respect to sample size and across numerous regression equations. Many observations were found to be of dubious quality but it was possible to purge the sample of many of these observations to obtain more reliable sub-samples. Although the basic regressions were subject to heteroscedasticity, as is common for cross section saving regressions, the capital gains effect implied by regressions which had been adjusted for heteroscedasticity were insignificantly smaller. An attempt was made to adjust for household tastes which tend to bias the effect of wealth on saving but, unfortunately, this procedure failed. Nevertheless, there is little reason to believe that the capital gains effect is biased significantly in either direction and the findings confirm the sizeable capital gains effect included in the MPS model. Thus, this analysis suggests that the wealth channel of monetary policy is very