Reproduction of "Public Pension Reforms and Retirement Decisions: Narrative Evidence and Aggregate Implications"

Reproduction of "Public Pension Reforms and Retirement Decisions: Narrative Evidence and Aggregate Implications". The full report is available under Files. Authors responded: "Thank you very much for sharing the replication study of our paper “Public Pension Reforms and Retirement Decisions: Narrative Evidence and Aggregate Implications.” We are pleased to see that the replicators are able to computationally reproduce our main results from the raw data and that your robustness exercises broadly support our qualitative conclusions, even though some magnitudes and significance levels vary under alternative specifications. Your extensions using balanced samples, alternative country sets, and Newey–West standard errors are all useful complements that help clarify how sensitive specific estimates are to sample and inference choices. Regarding the use of an unbalanced panel and varying samples across horizons, our original design choice reflects the nature of the data we work with: annual observations and a finite number of pension reforms. Our priority was to exploit all available information at each horizon, especially given the relatively small number of narrative-identified reforms. For this reason, we did not impose a balanced sample across horizons in the baseline. At the same time, we fully agree that enforcing balanced samples is a natural robustness check in this setting, and we view your results as providing a helpful perspective on the role of more recent reforms and sample composition. We also appreciate your exploration of distributional effects through quantile local projections. These results are an interesting extension of our framework and, as you note, they are broadly consistent with our narrative about the effects of immediate versus delayed pension reforms on labor market outcomes and pension spending. Overall, we see your work as a constructive and valuable contribution that reinforces the main messages of our paper while illuminating how certain modeling choices affect specific findings. We are happy for you to post the replication report and only if you want this response on your webpage. We don't intend to write a formal comment. Best, Sarah and Huixin"