Measurement Timing and Endogeneity in EPC-Based Policy Classification: Regression Discontinuity Evidence from High-Floor Residential Properties in England and Wales

Vollständiger Abstract

Worum geht es in dieser Arbeit?

This study examines EPC Band C threshold status and subsequent recorded EPC-score changes in high-floor residential buildings in England and Wales (2013–2023), using Energy Performance Certificate (EPC) data. A previously unreported panel-construction problem, Unique Property Reference Numbers are occasionally recorded at a building rather than unit level, is identified and addressed using a conservative same-unit filtering procedure. A threshold-based eligibility construction is shown to be mechanically related to recorded EPC improvement when measured from a dwelling’s post-improvement rather than baseline score: this timing choice alone accounts for the entire positive association found in a naive baseline specification. Measured instead from the baseline score, the association reverses its sign under every panel-construction and threshold combination tested, a reversal corroborated by a regression discontinuity design using a different estimation strategy on the same score variable, though the estimated magnitude is sensitive to bandwidth choice. Two secondary strategies examining whether interest rates moderate this relationship produced no credible estimates, so this question remains open. The findings indicate that eligibility indicators built from the same data used to measure outcomes require explicit attention to measurement timing, with implications for evaluations using EPC thresholds as proxies for financial or policy eligibility.