Alt et al. (2014) Replication
This repository includes a replication of Alt et al. (2014) using corrected election timing data.
27 October 2014
Alt et al. (2014) use the yrcurnt election timing variable from the Database of Political Institution's (2012) in their recent study of fiscal gimmickry in Europe. They find that fiscal gimmickry is more common directly before elections (and in countries with weak fiscal transparency).
However, this variable has a number of problems that reduce its reliability and validity of the variable as a measure of government election timing. See Gandrud for details.
This replication simply uses a corrected version of the yrcurnt variable in the same regression models as those presented in the original paper.
Overall, the original results are robust to models using the corrected variable, though the estimated magnitudes and statistical significance of the effects are reduced somewhat.
The repository contains the following main files:
data_source/sup002_corrected.dta: the original Stata data file with one addition: the corrected yrcurnt variable yrcurnt_corrected
data_source/sup003_updated.do the Stata do-file with source to run both the original and replication analyses.
detailed_results/Alt_et_al_2014_detailed_results.pdf: a write up of the replication with biased and corrected results tables.