Using ridge regressions, we introduce a novel methodology to estimate a time-varying version of the market risk-return trade-off. Our model improves available techniques since it allows for flexible patterns in the relationship and does not need a long span of data or additional state variables to accurately estimate the trade-off. We find that this relationship is positive during almost all the sample but it occasionally turns out negative during deep recessions. Our results may help solve the controversy in the previous literature. Investors, policymakers, and regulators must monitor this relationship to optimise investment, manage risks, and prevent financial instability.

Funding for open access charge: CRUE-Universitat Jaume I