Spatial Panel Data Models Using Stata

Abstract

xsmle is a new command for spatial analysis using Stata. We consider the quasi-maximum likelihood estimation of a wide set of both fixed-and randomeffects spatial models for balanced panel data. Of special note is that xsmle allows to handle unbalanced panels thanks to its full compatibility with the mi suite of commands, to use spatial weight matrices in the form of both Stata matrices and spmat objects, to compute direct, indirect and total marginal effects and related standard errors for linear (in variables) specifications, and to exploit a wide range of postestimation features, extending to the panel data case the predictors proposed by Kelejian and Prucha (2007). Moreover, it also allows the use of margins to compute total marginal effects in presence of nonlinear specifications obtained using factor variables. This paper describes the command and all its functionalities using both simulated and real data.

Who Creates Jobs? Econometric Modeling and Evidence for Austrian Firm Level Data

“…The time period from 1993 to 2013 are provided by Belotti, Deb, Manning and Norton (2015). Again the estimated standard errors of the predictions become larger during the crisis period, although net job creation is still an increasing function of firm size with similar magnitudes as obtained from the full sample.…”

Impacto de Mudanças no Emprego no Setor Público sobre o Mercado de Trabalho Local: evidências para as mesorregiões brasileiras de 2003 a 2010

“…Para tanto, acrescentamos à especificação original do modelo na Equação (1) um termo formado pela multiplicação da matriz de pesos espaciais W (de dimensão 274×274) pela matriz de dimensão 274×2 dos dados das variáveis mudança no emprego do setor público e a interação (dummy de tempo×mudança no emprego do setor público), que é também multiplicada pelo vetor de parâmetros teta de dimensão 2×1. Vide detalhes sobre o modelo EF-SLX em Belotti et al (2017). Em seguida, criamos, com o comando spmat do Stata (Drukker 2013), uma matriz W contígua 12 de ponderação ou pesos espaciais, utilizando os centroides das mesorregiões obtidos com o shapefile das mesorregiões brasileiras no software IPEAGEO (Carvalho et al 2011).…”

“…Em seguida, criamos, com o comando spmat do Stata (Drukker 2013), uma matriz W contígua 12 de ponderação ou pesos espaciais, utilizando os centroides das mesorregiões obtidos com o shapefile das mesorregiões brasileiras no software IPEAGEO (Carvalho et al 2011). Finalmente, estimamos por máxima verossimilhança um modelo Efeitos Fixos Spatial Lag of X (EF-SLX) com o comando xsmle do Stata (Belotti et al, 2017 Com base nos resultados na Tabela 4, concluímos que não há efeito espacial da mudança no emprego do setor público na economia do Brasil como um todo, e nem no setor de comercializáveis e no setor de não comercializáveis da economia brasileira, o que corrobora os resultados e conclusões apresentados nas seções anteriores do presente trabalho.…”

Benefits and Costs of Digital Consulting in Clinics Serving Young People With Long-Term Conditions: Mixed-Methods Approach (Preprint)

“…A two part model [29] was employed to analyse the data collected from the clinical staff to identify any factors associated with the time spent on digital consulting. This model was chosen to reflect the fact that the time spent on using digital consulting is rightly skewed and contains many zeros [29].…”

“…A two part model [29] was employed to analyse the data collected from the clinical staff to identify any factors associated with the time spent on digital consulting. This model was chosen to reflect the fact that the time spent on using digital consulting is rightly skewed and contains many zeros [29]. The Logit model was used for the first part while the generalised linear model (GLM) with the Poisson family was used for the second part after conducting the modified links test [30].…”