What have we learned from 25 years of cross-sectional asset pricing research, and how can methodological improvements enhance the robustness of return predictors in global equity markets?

Cross-sectional asset pricing studies have developed over the last 25 years to address the initial shortcomings of the Capital Asset Pricing Model (CAPM) by Fama and French (1992) through the development of a so-called factor zoo of more than 316 verifiable return predictors (Harvey, Liu, and Zhu, 2016). This growth, although showing important patterns in the expected returns, has brought up serious questions regarding data mining, publication bias, and decay of anomalies; most predictors have dropped their returns on average by 58% after publication (McLean & Pontiff, 2016). The momentum anomaly is perhaps the best example of both persistence of behavioral and risk-based models and the difficulty of replication in the non-U.S. markets (Jegadeesh and Titman, 1993). The thesis considers four main research questions: (1) what has followed the empirical and theoretical developments of cross-sectional asset pricing (2) how the improvement of techniques by making corrections to multiple testing (exceeding t-statistic thresholds), out-of-sample validation, and Bayesian shrinkage, can increase the strength of predictors (3) do U.S. centric forms persist in the European equity markets; and (4) how could modern techniques help to mitigate the problem of data mining biases? The research, based on a complete sample of 2001-2023 of European equities, forms a four-factor model (market, size, value, momentum) with strict portfolio sorts and Fama-MacBeth regressions. The major findings are a high significance level of value premium (2.321-26.4% monthly, t=4.61, annualization=26.4) that is significantly higher than the international rates, and a small but significant momentum premium (0.091% monthly). Momentum is time varying, whereby it takes a positive premise in expansions and a negative premise during a crisis, whereas ESG conditioning does not produce significant interaction effects. Ways to improve the methods used (60-month rolling regressions and multiple testing consciousness) verify the strength and reduce replication crisis fears (Hou, Xue, and Zhang, 2020; Chen and Zimmermann, 2022). These findings confirm standard contributions to the European market, invalid weak-form efficiency, and the importance of methodological rigour in supporting sound international asset prices. The results help to close the U.S. literature gap and provide an implication of global factor investing and future investigation of ESG-factor interactions.