Alexis Akira Toda and Yulong Wang, "Efficient Minimum Distance Estimation of Pareto Exponent from Top Income Shares", Journal of Applied Econometrics, Vol. 36, No. 2, 2021, pp. 228-243. There are two data files, which are ASCII files in DOS format. They are zipped in the file tw-data.zip. In addition, there are six Matlab code files, which are zipped in the file tw-code.zip. Unix/Linux users should use "unzip -a". ** Description of files ** -- "TopShareUS.csv" CSV file containing top income shares for US. The source data are available as an Excel spreadsheet (Table A.3) at https://eml.berkeley.edu/~saez/TabFig2017prel.xls -- Variable list * Year: the year of the income data * p90p100: top 10% income share (from 1917 to 2017, total 101) * p95p100: top 5% income share (from 1917 to 2017, total 101) * p99p100: top 1% income share (from 1913 to 2017, total 105) * p99.5p100: top 0.5% income share (from 1913 to 2017, total 105) * p99.9p100: top 0.1% income share (from 1913 to 2017, total 105) * p99.99p100: top 0.01% income share (from 1913 to 2017, total 105) -- "TopShareFrance.csv" CSV file containing top income shares for France. The source data are available at https://wid.world/ * Year: the year of the income data * p90p100: top 10% income share (from 1900 to 2014, total 102) * p95p100: top 5% income share (from 1900 to 2014, total 102) * p99p100: top 1% income share (from 1900 to 2014, total 102) * p99.5p100: top 0.5% income share (from 1900 to 2014, total 102) * p99.9100: top 0.1% income share (from 1900 to 2014, total 102) * p99.99p100: top 0.01% income share (from 1900 to 2014, total 102) Matlab code files: -- Chowtest.m conducts Chow test -- getCI.m computes confidence interval using the approach of Ibragimov & Mueller -- MDobj.m defines objective function of minimum distance estimator -- ParetoCUMDE.m implements continuously updated minimum distance estimator (CUMDE) -- test_MDobj.m example of how to use MDobj.m -- test_ParetoCUMDE.m main file to replicate results in paper ** Instructions for replicating results in paper ** 1. Open "test_ParetoCUMDE.m" in Matlab (file works in version 2019b) and run. By setting "country = 1" on line 22, you get the results for US. Setting "country = 2" is for France. 2. After running "test_ParetoCUMDE.m", run "Chowtest.m" and "getCI.m".