The effects of macroeconomic variables on stock prices: conventional versus news models

Stock prices are usually analysed and explained in terms of underlying financial indicators, such as earnings per share or dividend payout ratios. Nevertheless, fluctuations in the conditions of the economy can result in changes in demand, which can impact on profits and dividends. Since macroeconomic variables affect financial indicators it follows that macroeconomic variables affect stock prices. If markets are rational and efficient, then stock prices will reflect all known information regarding macroeconomic factors that are perceived to affect stock prices. It follows that stock prices should not change significantly unless there is a surprise or news about the state of the economy (as reflected in unexpected changes in macroeconomic variables). Intuitively, this implies that models of stock price determination based on news ought to be superior to conventional models that use the levels or changes in variables. The utilisation of news in research on stock prices is very limited. Two approaches have been traditionally used to represent the news in the absence of surveys of expectations: either by assuming announcements are news such as those in event studies or by using an econometric time series approach to extract the news components from total changes in the variables, as is the case with the news model. The majority of studies involving news models have been in the foreign exchange market using news estimated econometrically—very little has been done in estimating and testing a macro news model of stock prices and certainly nothing has been done on stock prices in developed economies such as Australia. Thus this research is motivated by the significant gaps in the literature with respect to the development, estimation and testing of a news model of stock prices. Most of the studies that investigate the relations between macro variables and stock prices have been carried out using conventional approaches by estimating models that use the variables in their levels. Some of the multivariable models of stock prices arise as a result of anomalies found in implementing the capital asset pricing model. Other multivariable approaches such as the arbitrage pricing theory (APT), due to Ross (1976), suggest that macro variables are useful, but APT is silent on the appropriate macroeconomic explanatory variables. Furthermore, there have been limited attempts to examine macroeconomic variables collectively, but not with the aim of developing a macro model of stock prices. This thesis presents the results of research that uses comprehensive econometric procedures to investigate which macroeconomic variables have significant effects on Australian stock prices and whether news about such variables can enhance the performance of conventional stock price determination models. Seven macroeconomic variables are examined: interest rates, inflation, the money supply, economic activity, commodity prices, exchange rates and a foreign stock market index to account for spill-over effects. This provides a valuable contribution to the understanding of the individual effects of macroeconomic variables on stock prices and adds to the limited literature regarding the usefulness of news in models of stock price determination. The results from this research demonstrate that although news is a theoretically sound and intuitively plausible basis for improving macro models of stock prices, in practice there is no ex-ante exploitation possible by estimating news utilising econometric methods. Simply put, news cannot be predicted—this is established by using three comprehensive methods of estimating news, which is the residual of a model fitted to the time series data of a particular variable