Working Paper: NBER ID: w23602
Authors: Catherine Hausman; David S. Rapson
Abstract: Recent empirical work in several economic fields, particularly environmental and energy economics, has adapted the regression discontinuity (RD) framework to applications where time is the running variable and treatment begins at a particular threshold in time. In this guide for practitioners, we discuss several features of this “Regression Discontinuity in Time” (RDiT) framework that differ from the more standard cross-sectional RD framework. First, many applications (particularly in environmental economics) lack cross-sectional variation and are estimated using observations far from the temporal threshold. This common empirical practice is hard to square with the assumptions of a cross-sectional RD, which is conceptualized for an estimation bandwidth shrinking even as the sample size increases. Second, estimates may be biased if the time-series properties of the data are ignored (for instance in the presence of an autoregressive process), or more generally if short-run and long-run effects differ. Finally, tests for sorting or bunching near the threshold are often irrelevant, making the framework closer to an event study than a regression discontinuity design. Based on these features and motivated by hypothetical examples using air quality data, we offer suggestions for the empirical researcher wishing to use the RD in time framework.
Keywords: regression discontinuity; timeseries; empirical applications; environmental economics; energy policy
JEL Codes: C14; C21; C22; Q53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| RDiT designs rely on observations far from the temporal threshold (C22) | biased estimates (C51) |
| Ignoring the timeseries properties of the data (C22) | biased estimates (C51) |
| The McCrary density test (C52) | irrelevant McCrary density test (C52) |
| RDiT (O39) | RDiT is conceptually closer to event studies (C22) |
| Many papers utilizing RDiT (O32) | estimating treatment effects related to environmental and energy policies (Q51) |