Full Abstracts & Presentations
Abstracts & Presentations
Edson Severnini (Carnegie Mellon University)
Basedosdados: a new package to access and manipulate more than 60 clean Brazilian databases directly from Stata
Isabella H. Claudino (IPEA) e Ricardo Dahis (Base dos Dados e PUC Rio)
This presentation aims to show the applicability of the new basedosdados Stata package under development. The package enables access to more than 60 Brazilian databases already cleaned and compatibilized, available in the public datalake BD+ from the Base dos Dados organization. The datasets available range from 250 GB of RAIS, Censo's, PNAD, to meteorology, health, politics, and more than 20 topics.
There are packages already available to the public in R and Python that allow, in seconds, to access and manipulate these data with a few lines of code. Now, we have developed a new Stata package basedosdados that allows users of the community to perform this same process. The package consists of 7 commands. The commands give the possibility to list all available datasets from BD+ datalake and downloading or analyzing them directly from Stata. The basedosdados package aims to make the distance between anyone and an analysis just a good issue, by providing access to clean, compatible and high-quality public data.
Political environmental cycles: how election incentives affect environmental policy and forest conservation outcomes
Patricia Ruggiero (FEA/USP) e Paula Pereda (FEA/USP)
Sarah Ferreira (UNICAMP), Renato Garcia (UNICAMP) e Veneziano Araújo (UNIFESP)
The effects of road on trade and migration.
Jaqueline Oliveira (Rhodes College)
Taís Maria G. de Lima (UNIFESP), Solange Ledi Gonçalves (UNIFESP) e André Roncaglia de Carvalho (UNIFESP)
This paper aims to investigate the poverty duration of female-headed households between 2012 and 2019. Our focus is on comparing the duration of poverty in households headed by women and men. The article uses longitudinal microdata from PNADC (IBGE). The main contribution of the article is to apply an analysis of poverty dynamics through the use of survival models to the poverty situation of families. In addition, heterogeneity analysis are performed for four different groups of families, combining information on gender and the presence of a spouse in the household. The results found indicate that the access to the labor market and work in the formal sector reduce the
duration of poverty for families, these findings are in agreement with the literature on poverty incidence. The main result of the study indicates that female-headed families without a spouse and with children in the household, tend to be more penalized and remain in poverty longer, with a probability thirteen percentage points higher than the male counterpart. Finally, the results of this dissertation suggest that the design of government policies against poverty should consider the different household characteristics, as well as the duration of the phenomenon.
Gender Gap and Leisure Time: Evidence from Brazilian Couples
Ana Luiza N. de Holanda Barbosa (IPEA, GeFam e IBMEC-Rio)
Pink Tech: Did Computers and the Internet Reduce the Gender Wage Gap? Evidence from Brazilian Data
Ana Abras , Giovana C. Bigliazzi e Mônica Y. Kuwahara (Universidade Federal do ABC)
Despite the increase in female participation in the Brazilian labor market and a reduction in the average wage gap between men and women, there are gender inequalities in the distribution of workers within sectors and occupations and different degrees of formalization of employment (Bruschini 2007; Itaboraí and Ricoldi 2016). Some studies associate the reduction of the wage gap with a set of skills held by women in certain occupations, making them more able to face technological challenges (Black and Spitz-Oener 2010). We study the factors that are related to the reduction in the wage gap by turning our attention to the job tasks performed by men and women. Could tasks associated with occupations with a greater female presence allow women better reactions in terms of wage growth and formal employment in the face of a series of technological challenges? The results indicate that in markets more specialized in routine tasks the growth of wages and job formality for women is higher than for men. This evidence is
robust to the inclusion of regional internet density as an independent variable.
Multinomial logit models for longitudinal data
Joerg Luedicke, Social Scientist and Software Developer (Stata Corp)
The multinomial logit model is a popular tool in the toolkit of many social scientists and applied statisticians. It is used to model categorical outcome variables where the categories have no natural ordering. Stata 17 introduced the new command xtmlogit, which fits multinomial logit models to panel/longitudinal data. In this presentation, I will discuss the theoretical underpinnings and corresponding assumptions of two commonly used estimators: the random-effects and the (conditional) fixed-effects estimators. I will also discuss some potential practical pitfalls related to these estimators, and I will show how to use the new xtmlogit command to fit both random-effects and fixed-effects multinomial logit models.