Estimation and Inference for Actual and Counterfactual Growth Incidence Curves

Ferreira, Francisco H. G. ; Firpo, Sergio ; Galvao, Antonio F.

World Bank, Washington, DC

关键词: growth incidence curve;    potential outcome;    inference;    quantile process;    economic growth;   

DOI  :  10.1596/1813-9450-7933
RP-ID  :  WPS7933

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

Different episodes of economic growthdisplay widely varying distributional characteristics, bothacross countries and over time. Growth is sometimesaccompanied by rising and sometimes by falling inequality.Applied economists have come to rely on the Growth IncidenceCurve, which gives the quantile-specific rate of incomegrowth over a certain period, to describe and analyze theincidence of economic growth. This paper discusses theidentification conditions, and develops estimation andinference procedures for both actual and counterfactualgrowth incidence curves, based on general functions of thequantile potential outcome process over the space ofquantiles. The paper establishes the limiting 0 distributionof the test statistics of interest for those generalfunctions, and proposes resampling methods to implementinference in practice. The proposed methods are illustratedby a comparison of the growth processes in the United Statesand Brazil during 1995-2007. Although growth in the averagereal wage was disappointing in both countries, thedistribution of that growth was markedly different. In theUnited States, wage growth was mediocre for the bottom 80percent of the sample, but much more rapid for the top 20percent. In Brazil, conversely, wage growth was rapid belowthe median, and negative at the top. As a result, inequalityrose in the United States and fell markedly in Brazil.