Data for: Does inflation worsen income inequality? A meta-analysis

Abstract of associated article: Despite extensive econometric evidence, the research literature has been unable to draw firm conclusions regarding the effect of inflation on income inequality. In this paper, we apply meta-regression methods to a novel data set of 1767 estimates reported in 124 published studies that investigate the effect of inflation on income inequality. We distinguish between estimates that examine the effect of inflation on levels of income inequality and those that examine the effect of inflation on differences of income inequality. For level estimates, not controlling for moderator variables points to mild publication bias in favor of positive estimates (i.e., the current literature favors publishing studies that find that inflation increases income inequality), but publication selectivity does not hold once we control for a set of moderator variables. For difference estimates, mild publication bias in favor of negative estimates is found only once we control for moderator variables. In addition, our results suggest that inflation has a (small-to-moderate) inequality increasing effect for both level and difference estimates. Furthermore, we show that several factors influence reported estimates, including researcher choices concerning the measurement of inflation and inequality, the characteristics of data and estimation methods, and controlling for other components of inequality.

关联论文摘要：尽管已有大量计量经济学证据，但现有研究文献仍无法就通货膨胀对收入不平等的影响得出确切结论。本文采用元回归分析方法，对124项已发表研究中的1767个估计值构成的新型数据集展开分析，所有研究均探讨了通货膨胀对收入不平等的影响效应。本文区分两类估计场景：一类是考察通货膨胀对收入不平等水平的影响，另一类则聚焦于通货膨胀对收入不平等差异的影响。针对水平类估计，未控制调节变量时，存在偏向正向估计结果的轻度发表偏倚（即当前文献更倾向于发表认为通货膨胀会加剧收入不平等的研究）；但在纳入一组调节变量进行控制后，该发表选择性偏倚便不再显著。针对差异类估计，仅在控制调节变量后，才会呈现出偏向负向估计结果的轻度发表偏倚。此外，研究结果表明，无论是水平类还是差异类估计，通货膨胀均存在（小幅度至中等幅度的）加剧收入不平等的效应。本文还证实，多项因素会影响研究报告的估计结果，包括研究者在通货膨胀与不平等度量上的选择、数据特征与估计方法，以及是否控制不平等的其他组成成分。

搜集汇总

背景与挑战

背景概述

该数据集支持一篇元分析文章，旨在探究通货膨胀对收入不平等的影响。数据集包含来自124项研究的1767个估计值，分析发现通货膨胀对收入不平等有轻微至中等的加剧效应，同时研究揭示了估计值受测量方法、数据特征和模型设定等多种因素影响。