Adrien Pacifico, Ph.D.
Economics of Taxation, Fiscal Microsimulation, Data Science
How does the French income tax work:
An Openfisca Notebook Story (in French).
OpenFisca is an OpenSource microsimulation software that can be used to explore the French tax-benefit system. In this notebook, I try to explain how does the French income tax work in some of its dimensions. Simply click on the "launch binder" icon to start to explore the French income tax.
Execute binder:
Travis build:
Reproducible Science (in economics)
When data are accessible, less than 2/3 of the economics article are reproducible (Chang & Li, 2015). This could be due to many reasons, the most critical are the unavailability of the code used to produce the study, and errors in the code made by the authors (e.g the Reinhart & Rogoff case).
The use of administrative data is more and more frequent in the economic literature. It often contains sensitive information that researcher access through confidentiality agreements. Researchers are thus unable to give access to the data they used to the general public.
I plan write about Jupyter Notebooks as a good solution to allow other researchers to investigate how exactly results have been produced (with limited possibilities of alteration), without entailing anonymity of the sampled observation.
For the time being, here is some links that nourrish my reflexions on the subject.
Display links
-
Economics research is not replicable 2/3 of the time:
Chang & Li, 2015
-
A great book on how to do reproducible research with exemples:
The Practice of Reproducible Research
-
The usage of Binder to communicate code and make easily reproducible examples:
Binder 2.0 - Reproducible, interactive, sharable environments for science at scale (2018)
-
How Notebooks are used at Netflix:
Beyond Interactive, Notebook Innovation at Netflix
-
The Berkeley Initiative for Transparency in the Social Sciences:
https://www.bitss.org/
-
Open Science Framework:
https://osf.io/
-
Statistics Done Wrong:
https://www.statisticsdonewrong.com/
, and especially the Everybody makes mistakes/ reproducibility section
-
Big Data in economics An IZA working paper to translate computer scientist doing "machine learning" to economist doing "regressions" and vice versa.
-
Reproducibility and exploratory computing with a Jupyter-based workflow
-
Reproducible analysis and Research Transparency (with R)
-
Transparency, Reproducibility, and the Credibility of Economics Research, video of the presentation.
-
A blogpost that advocates for a progressive consolidation effort of scientific code, rather than putting too high a bar on code release
- Wiki of the OSF, that contains a list of empirical and theoretical papers about open science.
More miscellaneous links
https://www.nature.com/news/interactive-notebooks-sharing-the-code-1.16261
https://web.stanford.edu/~gentzkow/research/CodeAndData.pdf
Economist in tech: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3247794
https://www.aeaweb.org/about-aea/committees/economic-statistics/administrative-data
Mail:
Twitter:
GitHub:
StackOverflow: