It is generally accepted that household surveys fail to accurately portray the top tail of the income distribution. Indeed, studies based on tax data challenge the credibility of surveys as a source to study inequality. To date, there is no broad consensus on how to best reconcile these two datasets. This paper presents a novel method that enables a consistent combination of these two sources of data, under the assumption that tax data sets a credible lower bound on the amount of people with given levels of income. The resulting micro-dataset preserves the consistency of other socio-demographic variables at both the individual and aggregate levels. It thus allows researchers to analyze dimensions of social inequality under a more representative distributive framework. Our procedure is illustrated by empirical applications to five countries, covering both developed and less-developed contexts over numerous years.

The Weight of the Rich: Improving Surveys Using Tax Data

Household surveys often fail to capture the top tail of income and wealth distributions, as evidenced by studies based on tax data. Yet to date there is no consensus on how to best reconcile both sources of information, given the multiple biases at play. This paper contributes a novel method, rooted in standard calibration theory, to directly confront the problem of survey non-response between survey micro-data and anonymous tax data under reasonable assumptions. Our key innovation is to endogenously determine a “merging point” between the datasets, above which we start to incorporate information from tax data into the survey, under the assumption that the rate of representativeness is constant, then decreasing with income. This is followed by a “reweighting” and a “replacing” step, which preserves the microdata structure of the original survey, assuming no re-ranking of observations. We illustrate our approach with simulations, which show that our method is robust to the existence of income misreporting, and performs better than alternative methods. We also apply it to real data from five countries, both developed and less developed, finding changes to the levels and trends in income inequality. We discuss several limits to our approach and suggest some guidelines for future research. 

The weight of the rich: improving surveys using tax data

(1968). The Rich and the Super Rich: A Study in the Power of Money Today. Journal of Economic Issues: Vol. 2, No. 4, pp. 455-457.

The Rich and the Super Rich: A Study in the Power of Money Today

In this paper, we estimate the recent evolution of global interpersonal inequality and examine the effect of omitted top incomes on the level and direction of global inequality. We propose a methodology to estimate the truncation point of household surveys by combining information on income shares from household surveys and top income shares from tax data. The methodology relies on a flexible parametric functional form that models the income distribution for each country-year point under different assumptions on the omitted information at the right tail of the distribution. Goodness-of-fit results show a robust performance of our model, supporting the reliability of our estimates. Overall, we find that the undersampling of the richest individuals in household surveys generate a downward bias in global inequality estimates that ranges between 15 per cent and 42 per cent, depending on the period of analysis, and the assumed level of truncation of the income distribution.

Global Inequality: How Large is the Effect of Top Incomes?

/pdf/o-que-faz-os-ricos-ricos-2np4fg5t6k.pdf

O que faz os ricos ricos

To deal with the problem of underestimation of top incomes in household surveys, we propose a methodology to combine the income distributions of the Brazilian 2010 Census (survey) and of 2010 DIRPF (personal income tax reports). The method consists in estimating a system of non-response weights that uses as frame the tax register and is applied to the top of the distribution. After applying this calibration methodology, we decompose inequality income sources. Correcting survey distributions with tax data increases the contribution of non-labor income to inequality, as the case of the Brazilian Census shows. Changes in the methodology do not affect the results substantially.

Correcting the Underestimation of Top Incomes: Combining Data from Income Tax Reports and the Brazilian 2010 Census

El articulo examina en que medida la educacion puede ser considerada uno de los principales determinantes de la riqueza en Brasil. Valiendose de los datos de formacion universitaria especifica del Censo de 2010, el foco se restringe al 1% mas rico de la distribucion de rendimientos de trabajo. La principal conclusion es que la educacion puede ser importante para explicar la desigualdad total, pero no hay evidencias de que la educacion en masa sea uno de los factores mas relevantes para explicar las diferencias entre los ricos y el restante de la poblacion brasilena. Ni siquiera la educacion de elite puede ser tomada como uno de los principales determinantes de los actuales niveles de pobreza. Por lo tanto, hay una parte significativa de la desigualdad total que no sera reducida por politicas educativas.

/pdf/educacao-e-rendimentos-dos-ricos-no-brasil-ussnup25o5.pdf

Educação e Rendimentos dos Ricos no Brasil

To deal with the problem of underestimation of top incomes in household surveys, we propose a methodology to combine the income distributions of the Brazilian 2010 Census (survey) and of 2010 DIRPF (personal income tax reports). The method consists on estimating a system of non-response weights that uses as frame the tax register and is applied to the top of the distribution. After applying this calibration methodology, we decompose inequality income sources. Correcting survey distributions with tax data increases the contribution of non-labor income to inequality, as the case of the Brazilian Census shows. Changes in the methodology do not affect the results substantially.

We propose a methodology to conciliate the distributions of incomes of Census 2010 and of DIRPF 2010 (personal income tax reports). Applying this methodology, we decompose inequality by classes, regions and sex, race and age groups. We use annualized incomes and detailed educational information about university degrees obtained. Our results suggest that the underestimation of incomes in household surveys lead to an underestimation of the role of elites in Brazilian inequality. These elites contribute disproportionally to inequality and the correction resulting from conciliation increase even more this contribution.

https://www.econstor.eu/bitstream/10419/129895/1/840780818.pdf

A composição da desigualdade no Brasil: Conciliando o censo 2010 e os dados do imposto de renda

Portuguese Abstract: Avaliamos em que medida a educacao pode ser considerada um dos principais determinantes da riqueza no Brasil. Nosso foco e nos trabalhadores no 1% mais rico da distribuicao da renda do trabalho. Fazemos isso utilizando os dados de formacao universitaria especifica da Amostra do Censo 2010. Nossa principal conclusao e que a educacao pode ser importante para explicar a desigualdade total, mas nao ha evidencias de que a educacao de massa seja um dos fatores mais relevantes para explicar as diferencas entre os ricos e o resto da populacao brasileira. Nem mesmo a educacao de elite pode ser tomada como um dos principais determinantes dos niveis atuais de riqueza. Ha, portanto, uma parte importante da desigualdade total que nao sera reduzida por politicas educacionais.English Abstract: We examine the extent to which education is a main determinant of affluence in Brazil. We focus on workers in the top 1% of the labor earnings distribution. We use university courses data from the Sample Questionnaire of the 2010 Census. Our main conclusion is that while education may be important to explain total inequality, there is no evidence that mass education is a major factor explaining the differences between the rich and the rest of the population. Not even elite education can be assumed to be a main determinant of the current levels of affluence. Consequently, an important part of total inequality cannot be reduced by educational policies.

Juliana de Castro Galvão

Papers

Correcting the Underestimation of Top Incomes: Combining Data from Income Tax Reports and the Brazilian 2010 Census

Educação e Rendimentos dos Ricos no Brasil

Correcting the Underestimation of Top Incomes: Combining Data from Income Tax Reports and the Brazilian 2010 Census

A composição da desigualdade no Brasil: Conciliando o censo 2010 e os dados do imposto de renda

Educação e o Rendimento dos Ricos no Brasil (Education and the Income of the Rich in Brazil)