Pareto models for top incomes and wealth

Abstract

Quantiles and Expectiles (invited speaker)2017

Declining inequality in Latin America? Robustness checks for Peru

“…Ruiz and Woloszko (2016) propose corrections with a similar spirit by fitting a Pareto distribution to the top income data. A different but related approach is advanced in Alfons et al (2013), who propose weighted estimators to compensate for the under-reporting of top incomes (see also Charpentier and Flachaire, 2019). Eckerstorfer et al (2016), who focus on wealth data, also correct the data after interpreting the downward bias as a natural consequence of non-observability in small samples drawn from a skewed distribution.…”

“…the distribution of all incomes greater than the threshold Y P . Despite its simplicity, this distribution is appealing from a theoretical viewpoint (Reed, 2003) and has received enormous empirical support (see Clementi and Gallegati, 2005;Cowell and Flachaire, 2007;Charpentier and Flachaire, 2019) both as an exact model for the upper tail, or as a useful approximation. 6…”

“…Thus, when simulating inequality measures across time we will fix the same top income 13 The estimators of α exploit that the top P observations do belong to the Pareto tail of the true income distribution (see Ruiz and Woloszko, 2016;Ibragimov and Ibragimov, 2018). Charpentier and Flachaire (2019) review how the 'Rank-Size' equation derived from (1), log(i) = c − α log(Y (i) ) where Y (i) > YP is the income in the i-th position of the upper group, can be used to estimate α. The estimation faces a bias-variance trade-off, with the choice of P being crucial, as the bias is reduced at the cost of a higher variance as P → 0 (see Eckerstorfer et al, 2016).…”

The Impact of Top Incomes Biases on the Measurement of Inequality in the United States

“…The two‐parameter generalized Pareto type 2 and the generalized beta type 2 (GB2) distributions have been proposed as providing a better fit for the top tail of incomes in the United Kingdom (Jenkins et al ., 2011; Atkinson, 2017; Jenkins, 2017) and the United States (Charpentier and Flachaire, 2019). We consider these two alternative distributions for their simplicity and well‐known properties, as well as the improved flexibility they provide (estimations available on request).…”

Composite distributions in the social sciences: A comparative empirical study of firms' sales distribution for France, Germany, Italy, Japan, South Korea, and Spain

“…More recent publications dealing with firm sizes' distribution and related matters could be, for example, Reed (2001Reed ( , 2002Reed ( , 2003; Reed and Jorgensen (2004); Di Giovanni et al (2011); Guo et al (2013); Ioannides and Skouras (2013); Toda (2017); Cortés et al (2017); Kwong and Nadarajah (2019); Bǎncescu et al (2019); Su (2019); Toscani (2018, 2019b,a); Puente-Ajovín et al (2020a,b); Campolieti and Ramos (2021); Reginster (2021); Ishikawa (2021). Recent articles on the issue about whether the upper tail is lognormal or Pareto could be, e.g., Perline (2005); Clauset et al (2009); Bee et al (2011Bee et al ( , 2013; Bee (2015); Bee et al (2017); Chu et al (2019); Schluter and Trede (2019); Bee (2022); Charpentier and Flachaire (2022). Also, there are studies aiming at the explanation of economically important quantities by mixtures of distributions (Belmonte et al, 2014;Kuninaka, 2020), being quantities for which sub-populations of it may exist (McLachlan and Peel, 2003).…”