Recent empirical evidence based on a linear framework tends to suggest that a Markov-switching version
of the consumption-aggregate wealth ratio (cayMS), developed to account for structural breaks, is a better
predictor of stock returns than the conventional measure (cay)—a finding we confirm as well. Using
quarterly data over 1952:Q1–2013:Q3, we however provide statistical evidence that the relationship
between stock returns and cay or cayMS is in fact nonlinear. Then, given this evidence of nonlinearity,
using a nonparametric Granger causality test, we show that it is in fact cay and not cayMS which is a
stronger predictor of not only stock returns, but also volatility.