The Heterogeneous Effect of Alternative Data on the Capital Market: Evidence from Online Recruitment Big Data

Modern information systems enable the collection and analysis of non-traditional information sources, commonly referred to as alternative data. This information is context-rich and has predictive efficacy, making it particularly valuable for investment decisions. However, previous studies have divergent findings on the question of whether alternative data can affect corporate performance. Using the Causal Forest Double Machine Learning model, we analyze approximately 7 million job postings from 1079 Chinese listed companies, including three dimensions: the source, receiver, and content characteristics. Our findings indicate that (i) recruitment data can affect corporate performance, with heterogeneity explaining the inconsistent impacts; (ii) the treatment effect is smaller for enterprises with higher state ownership; (iii) firms in the introduction and growth stages exhibit relatively higher treatment effects, whereas those in the shake-out and decline stages show relatively lower treatment effects. Also, the effects of recruitment positions and salary level are examined.