Software is Eating the World : Using Machine Learning to Analyze Private Equity Style Drift Toward Software

Private equity (PE) firms increasingly drift toward investing in software companies. In this paper, we empirically examine this trend and compare software drifts to general style drifts. We use an unbalanced panel of more than 7,000 PE firms that made investments between 1985 and 2020. Our analysis exploits a machine learning (ML)-based classification approach to identify software transactions. By doing so, we circumvent problems with conventional software classifications that see software as a subvertical in the information technology industry. Our findings are three-fold. First, we document that the share of software transactions increased greatly during the past three decades and across all industries. Second, we find that general style drifts and software drifts represent structurally different investment decisions: style drifts occur during expansionary periods when PE competition levels are low, and software drifts are more likely to occur when competition levels are high. Both drifts negatively affect investment performance and potentially pose agency conflicts. These are more pronounced for software drifts. Third, we show differences contingent on measurement. Our findings contribute to PE style drift literature and a growing body of research applying ML techniques in finance and economics