TY - JOUR
T1 - Asset pricing with neural networks
T2 - significance tests
AU - Fallahgoul, Hasan
AU - Franstianto, Vincentius
AU - Lin, Xin
N1 - Funding Information:
☆ An earlier version of this paper has been circulated under the title ”Towards Explaining Deep Learning: A Variable Significance Test for Multi-Layer Perceptrons”. We very grateful to the Editor Torben G. Andersen, the Associate Editor, and the two Referees for their invaluable comments. Their feedback has played a pivotal role in effecting significant enhancements to both the content and presentation of our work. We also thank Daniele Bianchi, Daniel Buncic, Ivan Guo, Loriano Mancini, Juan-Pablo Ortega, Farshid Vahid, and seminar participants at UNSW, Monash University, 4th conference on “Non-traditional Data, Machine Learning, and Natural Language Processing in Macroeconomics. Monash Centre for Quantitative Finance and Investment Strategies has been supported by BNP Paribas.
Funding Information:
An earlier version of this paper has been circulated under the title ”Towards Explaining Deep Learning: A Variable Significance Test for Multi-Layer Perceptrons”. We very grateful to the Editor Torben G. Andersen, the Associate Editor, and the two Referees for their invaluable comments. Their feedback has played a pivotal role in effecting significant enhancements to both the content and presentation of our work. We also thank Daniele Bianchi, Daniel Buncic, Ivan Guo, Loriano Mancini, Juan-Pablo Ortega, Farshid Vahid, and seminar participants at UNSW, Monash University, 4th conference on “Non-traditional Data, Machine Learning, and Natural Language Processing in Macroeconomics. Monash Centre for Quantitative Finance and Investment Strategies has been supported by BNP Paribas.
Publisher Copyright:
© 2023 The Author(s)
PY - 2024/1
Y1 - 2024/1
N2 - This study proposes a novel hypothesis test for evaluating the statistical significance of input variables in multi-layer perceptron (MLP) regression models. Theoretical foundations are established through consistency results and estimation rate analysis using the sieves method. To validate the test's performance in complex and realistic settings, an extensive Monte Carlo simulation is conducted. Results of the simulation reveal that the test has a high power and low rate of false positives, making it a powerful tool for detecting true effects in data. The test is further applied to identify the most influential predictors of equity risk premiums, with results indicating that only a small number of characteristics have statistical significance and all macroeconomic predictors are insignificant at the 1% level. These findings are consistent across a variety of neural network architectures.
AB - This study proposes a novel hypothesis test for evaluating the statistical significance of input variables in multi-layer perceptron (MLP) regression models. Theoretical foundations are established through consistency results and estimation rate analysis using the sieves method. To validate the test's performance in complex and realistic settings, an extensive Monte Carlo simulation is conducted. Results of the simulation reveal that the test has a high power and low rate of false positives, making it a powerful tool for detecting true effects in data. The test is further applied to identify the most influential predictors of equity risk premiums, with results indicating that only a small number of characteristics have statistical significance and all macroeconomic predictors are insignificant at the 1% level. These findings are consistent across a variety of neural network architectures.
KW - Asset Pricing
KW - Neural Networks
KW - Risk Premium
KW - Variable Significance Test
UR - http://www.scopus.com/inward/record.url?scp=85175615182&partnerID=8YFLogxK
U2 - 10.1016/j.jeconom.2023.105574
DO - 10.1016/j.jeconom.2023.105574
M3 - Article
AN - SCOPUS:85175615182
SN - 0304-4076
VL - 238
JO - Journal of Econometrics
JF - Journal of Econometrics
IS - 1
M1 - 105574
ER -