Measuring Market Integration: US Stock and REIT Markets

Abstract

Tests of financial market integration traditionally test for equality of risk premia for a common set of assumed risk factors. Failure to reject the equality of market risk premia can arise because the markets do not share a common factor, i.e. the underlying factor model assumptions are incorrect. In this paper we propose a new methodology that solves this joint hypothesis problem. The first step in our approach tests for the presence of common factors using canonical correlation analysis. The second step of our approach subsequently tests for the equality of risk premia using Generalized Method of Moments (GMM). We illustrate our methodology by examining market integration of US Real Estate Investment Trust (REIT) and Stock markets over the period from 1985 to 2013. We find strong evidence that REIT and stock market integration varies through time. In the earlier part of our data period, the markets do not share a common factor, consistent with markets not being integrated. We also show that during this period, the GMM tests fail to reject the equality of risk premia, highlighting the joint hypothesis problem. The markets in the latter half of our sample show evidence of both a common factor and the equality of risk premia suggesting that REIT and Stock markets are integrated in more recent times.