information processing task, while being complements for sequential and more complex tasks.
학회소식
[2019년 제 4차] Machine Invasion: Automation in Information Processing and the Cross Section of Stock Returns
작성자 : 관리자
조회수 : 47
게시일 :
2019-12-13
We separate downloads on the SEC EDGAR database into human and machine actions by the intensity of information retrieval (Ryans, 2017). The split shows that the extent of machine downloads has risen 35 times since 2004, accounting for over 96% of total downloads as of 2016. We formally investigate the relationship of machine automation in information processing and the cross-section of stock returns. We find that stocks in the lowest quintile of machine coverage outperform those in the highest quintile by 6 to 7% annually after adjusting for risk. Our results are further supported by a natural experiment on the phase implementation of XBRL tags that enabled machine readable financial disclosure. Our results are consistent with recent theoretical work on: (1) big data (Begenau, Farboodi, and Veldkamp, 2018) since we show that higher machine coverage is associated with lower expected returns, and (2) with Acemoglu and Restrepo (2018) amongst others as we find that machine and human labor are substitutes for the same
information processing task, while being complements for sequential and more complex tasks.
information processing task, while being complements for sequential and more complex tasks.
