Is There a Replication Crisis in Finance?

✍🏻️ Author: THEIS INGERSLEV JENSEN, BRYAN KELLY, LASSE HEJE PEDERSEN

📖 Source: Journal of Finance Volume 78 Issue 5, October 2023

Introduction

The article "Is There a Replication Crisis in Finance?" published in the October 2023 issue of The Journal of Finance, addresses the question of whether there is a replication crisis in finance research. The authors, Theis Ingerslev Jensen, Bryan Kelly, and Lasse Heje Pedersen, argue that the majority of asset pricing factors can be replicated, and that there is no replication crisis in finance.

Discussion

The replication crisis is a phenomenon that has been observed in several research fields, including medicine, psychology, management, and experimental economics. It refers to the difficulty of reproducing the results of scientific studies, which has led to concerns about the credibility of research findings. In finance, the replication crisis has been attributed to the large number of factors that have been identified in the literature, and the use of multiple testing procedures that increase the likelihood of false positives.

To address the question of whether there is a replication crisis in finance, the authors develop and estimate a Bayesian model of factor replication. The model is based on the idea that factors should be understood in light of economic theory, and that they can be clustered into themes that capture the underlying economic intuition. The authors use a hierarchical Bayesian approach that allows them to estimate the hyperparameters and the posterior alpha distributions of the model, and to account for the uncertainty in the estimates.

The authors apply their model to a sample of 153 factors across 93 countries, and find that the majority of factors can be replicated. They also find that the factors can be clustered into 13 themes, the majority of which are significant parts of the tangency portfolio. The authors argue that this provides evidence that the factors are not spurious, but capture important economic information that can be used to construct investment portfolios.

To address the challenges to the replicability of finance research, the authors propose several solutions. First, they argue that factors should be constructed in a consistent and implementable way, and that the data used to construct the factors should be publicly available. Second, they propose the use of Bayesian methods that allow for the joint modeling of all factors, and that provide a more powerful multiple testing adjustment than common frequentist methods. Third, they propose the use of out-of-sample tests that evaluate the performance of the factors in new data sets, and that provide evidence of their external validity.

The authors provide several pieces of evidence to support their conclusion that there is no replication crisis in finance. First, they find that the majority of factors can be replicated, and that they can be clustered into themes that capture the underlying economic intuition. Second, they find that the factors work out-of-sample in a new large data set covering 93 countries, which provides evidence of their external validity. Third, they find that the evidence on factor replication is strengthened (not weakened) by the large number of observed factors, which suggests that the factors capture important economic information that can be used to construct investment portfolios.

Conclusion

In conclusion, the authors argue that the replication crisis in finance is a myth, and that the majority of asset pricing factors can be replicated and used to construct investment portfolios. They propose several solutions to address the challenges to the replicability of finance research, and provide evidence that their Bayesian model of factor replication provides a more powerful multiple testing adjustment than common frequentist methods. The authors hope that their findings will help to advance replication in finance, and that their data set and code will be useful to other researchers in the field.