Real estate and stock returns: a multivariate VAREC model

Abstract

In this paper, we examine dynamic relationships among three housing market variables and a stock market index in a multivariate vector autoregressive error correction (VAREC) model. It is first found that, in the USA, sales and the median sales price of the existing single‐family homes and the 30‐year mortgage rate have unit roots, while the New York Stock Exchange (NYSE) value‐weighted portfolio returns appear random. Moreover, it is found that not only are three real estate variables cointegrated with one another but that they are also cointegrated with the stock index returns. After controlling for the unit root problem and cointegration, a multivariate VAREC model is further developed to examine dynamic relationships among the four variables using Johansen’s approach. It is found that the price, mortgage rate, and stock returns affect sales. It is found that the mortgage rate and stock returns affect the price. The 30‐year mortgage rate is affected by sales and the stock returns. Except for the mortgage rate which is negatively correlated with the stock returns, significant evidence is not found that sales and the median sales price affect the stock returns directly.