The Lives We Buy: Investigating the Effects of Stock and REIT Returns on Home Values in the U.S. Using Vector Autoregressive Modelling

Abstract

History has proved that the relationship between the asset classes is tied to social and economic mobility. One of the most critical issues in the U.S. is the growing wealth disparities among different income classes. Like many countries around the world, homeownership has served as a wealth building tool for many Americans. However, access to a home first requires sufficient disposable income that is sometimes acquired through earnings from asset investments. This thesis investigates the relationship between stock and REIT returns and home values of the three most expensive (Atherton, CA, Ross, CA, and Beverly Hills, CA) and least expensive (Marion, IN, Danville, IL, and Elmira, NY) in the United States from April 1996 to April 2019. By applying Vector Autoregression (VAR), a popular time series analysis method, this thesis forecasts the home values of these different cities using past stock and REIT returns as predictor variables. Before building the VAR model, we applied several econometric tests to measure the impact of these different home values on total stock and REIT returns. The results show that the most expensive zip codes were stationary after the second difference I(2) and the least expensive after the first difference I(1). Johansen’s cointegration test validated the existence of a long-term relationship between the home values of Elmira and total stock and REIT returns. The remaining five cities had a short-run relationship of city’s home value with stock and REIT returns. The Granger causality test implied that stock and REIT market returns Granger-causes all of the home values except for Ross, CA. The REIT returns do not Granger-cause Elmira, NY’s house values. Moreover, we cross-examined results of the seven forecasting models with diagnostic accuracy assessments to ensure the results were consistent with the preliminary econometric tests and previous academic literature. Based on these forecasting models, we conclude that, on average, stock and REIT returns are better predictors for homes in less expensive neighborhoods compared to more expensive neighborhoods.