Why has Factor Investing Failed?
“The Case for Causal Factor Investing,” authors Marcos Lopez de Prado, Alex Lipton, and Vincent Zoonekynd critically evaluate the shortcomings of traditional factor models used in financial economics and propose a shift towards causal factor investing. Moreover, detailing how the prevalent mis-specifications in factor models, traditionally driven by associational rather than causal reasoning, contribute to the underperformance of investment strategies.
Main Findings and Theoretical Contributions?
The researchers argue that the fundamental flaw in the current factor investing paradigm lies in its reliance on associational models that fail to accurately specify the underlying causal mechanisms.
As a result of not aligning model specifications with the actual causal structures, most factor investing models likely misestimate risk premia, leading to biased outcomes. Speaking to the potential dangers of such specification errors, which can result in strategies underperforming or even systematically lose money. Contrary to the belief that a misspecified model can still find itself effective if risk premia become estimated correctly.
De Prado, Lipton, and Zoonekynd advocate for the application of causal inference methodologies, including do-calculus and controlled experiments, to uncover and confirm the causal relationships that should guide model specification.
They highlight how advancements in causal inference have enabled a more scientific approach to exploring complex systems, an approach that remains underutilized in factor investing. The authors convincingly argue that adopting these causal techniques can prevent the typical errors that arise from conventional econometric methods, which often favor models with higher explanatory power without considering whether such models accurately reflect the underlying processes.
The treatise challenges the long-standing practices in the multi-trillion-dollar factor investing industry, calling for a foundational revision of how models are constructed. By integrating causal discovery and causal inference into factor investing, the authors believe that the industry can achieve more robust, theoretically sound investment strategies that are less prone to the pitfalls of mis-specification.
Read the full paper:
