Files in this item



application/pdf9625183.pdf (6MB)Restricted to U of Illinois
(no description provided)PDF


Title:The effect of personal income taxes on labor supply in Brazil: An application of quantile regression
Author(s):Ribeiro, Eduardo Pontual
Doctoral Committee Chair(s):Leuthold, Jane H.
Department / Program:Economics
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Economics, General
Economics, Labor
Abstract:The objective of this dissertation is to measure the welfare burden and labor supply effects of the personal income tax in Brazil. Labor supply functions are derived from utility maximization and applied to a sample of prime age (22 to 55 years old) urban male employees in Brazil, using the Linear, LES and CES functional forms. The data are a subsample taken from the PNAD 1990 household survey, conducted by the Instituto Brasileiro de Geographia e Estatistica, IBGE.
The Linearized Budget Constraint method is used for the empirical model, since the budget constraint is piecewise linear under a progressive income tax. Different assumptions are made regarding the tax liability imputation. It is seen that the empirical results vary significantly across tax hypotheses, with the assumption of married men acting as if they were single yielding the best results.
The estimation is carried out using semi-parametric estimation techniques, namely least squares and quantile regression. The variability of responses in the sample is investigated using quantile regression. Instrumental variables estimators are employed since specification tests indicate that the net wage and virtual income are correlated with the labor supply function error term.
Results suggest that the labor supply functions are backward bending for small changes in the net wage rate. From 2SLS estimates, the elasticity of work hours to changes in the wage rate is small, between $-$.01 to $-$.09, and the income elasticities range between 0 and $-$.06, across functional forms. The compensated elasticities, on the other hand, are markedly different between specifications. While the LES model yield a compensated elasticity for representative agents from.05 to.10, the CES and Linear models imply a compensated elasticity from.42 to.68.
Quantile regression estimates suggest considerable variation in the above estimates. Interestingly, the choice of hours of individuals working the standard workweek of 40 hours does not seem to be a function of wages and income, since the calculated income and wage elasticities were all zero.
Finally, the deadweight loss generated by the income tax, compared to an equal yield lump-sum tax, is calculated for the sample using the estimated coefficients and equivalent variation measures. Mean and median estimates of the deadweight loss per unit of revenue in the sample are between 9% to 15%, for the Linear and CES models. For those at the 25% tax rate bracket, the losses are much larger, between 49 and 79%. On the other hand, sample deadweight losses measures for the LES model indicate a much smaller burden, from 1% to a maximum of 10% of collected revenues.
The J test for non-nested models estimated by 2SLS is used to obtain partial evidence on the more appropriate model. The tests decisively reject the Linear model, indicating that the LES model is more appropriate in a pairwise comparison. It follows that the welfare burden of the 1990 Brazilian personal income tax is small, making this tax an efficient (second-best) source of revenues.
Issue Date:1996
Rights Information:Copyright 1996 Ribeiro, Eduardo Pontual
Date Available in IDEALS:2011-05-07
Identifier in Online Catalog:AAI9625183
OCLC Identifier:(UMI)AAI9625183

This item appears in the following Collection(s)

Item Statistics