Replication data for: Lipsmeyer & Zhu (2011)

Replication Article: Lipsmeyer, Christine S., and Zhu, Ling (2011) “Immigration, Globalization, and Unemployment Benefits in Developed EU States”, in American Journal of Political Science, Vol. 55, Issue 3, pp. 647-664 [DOI: 10.1111/j.1540-5907.2011.00509.x]. ************************************************************************** Co-authors: Diana Draghici ([e-mail]) and Jerome Hughes ([e-mail]). **************************************************************************************** In their 2011 AJPS article, Christine Lipsmeyer and Ling Zhu address highly controversial and intensely debated aspects pertaining to the interplay between the political and institutional configuration of developed EU countries, and the labor market policies these countries devise under economic globalization.We highlight a number of readily apparent errors at several stages in the data analysis process: (1) The data compilation from various sources has produced a joint dataset with unbalanced panels and missing data even where data were not missing in the original files (as explicitly referenced by the authors in the article, and available online); additionally, treatment of missing data is performed using less principled approaches such as list wise deletion and interpolation. (2) The estimated model relies on a Stata command (-xtpcse-) currently unsupported by the -Clarify- package, and the algorithm for simulating model parameters and quantities of interest, based on .ado files created by the authors, contains errors that inadvertently generate missing values for the ancillary parameters, thereby eliminating an important source of uncertainty surrounding the estimation process. (3) The article rests on some methodological confusions: most notably, the authors conflate the features of two time-series techniques (Prais-Winsten regression, which acc ording to the .do file is the method actually implemented by the authors), and error-correction model, which would indeed enable the estimation of both short-run and long-run effects (but which is clearly not implemented by the authors, as evident from both .do file and the model specification equation on p. 654 in the article). As a consequence, the computation of short-run and long-run effects is based on incorrect parameterizations, as evident on p. 659. We show that a rectification of all these methodological shortcomings can potentially alter the authors’ substantive conclusions to a significant extent . ****************** Note: This replication project is still work in progress. Please check back later for updates.