Multicountry Comparisons of the Consumption Based Capital Asset Pricing Model: Germany, Japan and USA

Publication Date
Financial Markets Group Discussion Papers DP 209
Publication Authors

This paper presents a series of empirical tests of the traditional Consumption Based Capital Asset Pricing Model using data from the German, Japanese and US Economies. This family of models (designated as CCAPM) was developed to explain the serial correlation properties of asset returns and consumption. The present paper is concerned with testing the CCAPM model's performance across countries and describes to that effect two new data sets on German and Japanese aggregate consumer expenditures and asset returns.

The basic structure of a CCAPM model is as follows: a representative consumer chooses a stochastic consumption plan to maximize his expected intertemporal utility. The utility at each period is of constant relative risk-aversion type. The budget constraint includes two possible types of asset returns, value weighted dividend adjusted stock index returns, and returns on short term debt instruments.

The quarterly data series used to test the model ranged from 1957 to 1991 and included the following:

  • Aggregate private consumer expenditure and population estimates obtained from the Japanese Economic Planning Agency's Expenditure Division and the Ministry of Welfare, the Statistiches Bundesamt and Citibase tapes for Japan, Germany and the US respectively.
  • Stock index returns obtained from the Japanese Securities Research
  • Institute, Morgan Stanley Germany and CRSP tapes for the US.
  • Short term interest rates from the respective Central Banks.

The testing procedure, identical for all three countries, is based on the General Method of Moments, and consists in constructing a criterion function from orthogonality conditions, the latter using the original model's first order condition and a set of instruments. The minimum of this criterion function yields an estimate of the parameters (namely relative risk aversion and the discount factor) and its asymptotic properties yield a simple test of the model.

The results show that the model is rejected in all three countries when multiple assets are used (namely value weighted dividend adjusted stock index return and return on short term debt instruments). The model is not rejected in any of the three countries when stock returns are used alone in the budget constraint and it is strongly rejected in all three countries when short term interest rates are used as the asset returns. The estimation procedure also results in similar estimates for the coefficient of relative risk aversion and the discount factor in all three countries and reveals the existence of an equity premium for all three nations (equity premium calculated as the difference of the annualized real return on equity and the annualized real return on short term debt) and of an equity premium puzzle in that the values of the coefficient of relative risk aversion and of the discount factor needed to explain such a value for the equity premium are too high to be believable.

Download is not available