Spurious Inference Caused by Time-Series Variation in Scaling: Real Estate Shocks Did Not Affect Corporate Investment

Real-estate price shocks did not positively associate with corporate investment from 1992 to 2008, as suggested by Chaney, Sraer, and Thesmar (2012). Instead of changes in the variables of interest (real-estate and investment), their coefficients explain changes in their firm-size scale normalization (property, plant, and equip- ment). Similar panel regression specifications are widespread in empirical corporate economics. My paper proposes a simple alternative specification that can eliminate and thus also diagnose concerns about the role of time-varying denominators. This specification suggests working with ratios of changes, where only the change is normalized, as in (∆ Variable of Interest)/Size Scaling.