Replication package — Two Pre-Registered Nulls on Sectoral TFP, and Why the Aggregate Cannot See Misallocation

Replication package for the working paper “Two Pre-Registered Nulls on Sectoral TFP, and Why the Aggregate Cannot See Misallocation (The QE Capital Key and the Euro Boom).” Using public data only, it reproduces the two pre-registered designs: (i) the QE design, where ECB Public Sector Purchase Programme purchases anchored on the capital key, interacted with Rajan–Zingales external-finance dependence, pass a randomization-inference pre-trend gate (p_RI = 0.29) and then show a sectoral-TFP null; and (ii) the boom design, where nominal euro convergence (1995–2001) interacted with the same sectoral exposure yields a null at the boom peak (p_RI = 0.73). Inputs: EUKLEMS & INTANProd sectoral TFP, ECB PSPP purchases, Eurostat Maastricht long-term bond yields, and Rajan–Zingales (1998) external-finance dependence. Code in Python (numpy/pandas); randomization inference seeded, B = 1000. No proprietary model is used or required.