Formation and destruction of coalition groups under economic growth

This paper introduces a model economy in which formation of coalition groups under technological progress is generated endogenously. The coalition formation depends crucially on the rate of arrival of new technologies. In the model, an agent working in the saroe technology for more than one period acquires skills, part of which is specific to this technology. These skills increase the agent productivity. In this case, if he has worked more than one period with the same technology he has incentives to construct a coalition to block the adoption of new technologies. Therefore, in every sector the workers have incentives to construct a coalition and to block the adoption of new technologies. They will block every time that a technology stay in use for more than one period.

Brazil’s 35 years-old quasi-stagnation: facts and theory

Brazil is growing around 1% per capita a year from 1981; this means for a country that is supposed to catch up, quasi-stagnation. Four historical new facts explain why growth was so low after the Real Plan: the reduction of public savings, and three facts that reduce private investments: the end of the unlimited supply of labor, a very high interest rate, and the 1990 dismantling of the mechanism that neutralized the Dutch disease, which represented a major competitive disadvantage for the manufacturing industry. New-developmental theory offers an explanation and two solutions for the problem, but does not underestimate the political economy problems involved

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we can model the heteroskedasticity of a linear combination of the errors. We show that this assumption can be satisfied without imposing strong assumptions on the errors in common DID applications. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative inference method that relies on strict stationarity and ergodicity of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment periods. We extend our inference methods to linear factor models when there are few treated groups. We also derive conditions under which a permutation test for the synthetic control estimator proposed by Abadie et al. (2010) is robust to heteroskedasticity and propose a modification on the test statistic that provided a better heteroskedasticity correction in our simulations.

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we know how the heteroskedasticity is generated, which is the case when it is generated by variation in the number of observations per group. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative application of our method that relies on assumptions about stationarity and convergence of the moments of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment groups. We extend our inference method to linear factor models when there are few treated groups. We also propose a permutation test for the synthetic control estimator that provided a better heteroskedasticity correction in our simulations than the test suggested by Abadie et al. (2010).

Quasi-experimental estimates of fiscal spillovers and spatial interactions among Brazilian cities

Nós usamos a metodologia de Regressões em Descontinuidade (RDD) para estimar o efeito causal do Fundo de Participação dos Municípios (FPM) recebido por um município sobre características dos municípios vizinhos, considerando uma variedade de temas: finanças públicas, educação, saúde e resultados eleitorais. Nós exploramos a regra que gera uma variação exógena da transferência em munícipios próximos às descontinuidades no repasse do fundo de acordo com faixas de população. Nossa principal contribuição é estimar separadamente e em conjunto o efeito spillover e o efeito direto do FPM, considerando ambos municípios vizinhos ou apenas um deles próximos às mudanças de faixa. Dessa forma, conseguimos entender melhor a interação entre municípios vizinhos quando há uma correlação na probabilidade de receber uma transferência federal. Nós mostramos que a estimativa do efeito direto do FPM sobre os gastos locais diminui em cerca de 20% quando controlamos pelo spillover do vizinho, que em geral é positivo, com exceção dos gastos em saúde e saneamento. Nós estimamos um efeito positivo da transferência sobre notas na prova Brasil e taxas de aprovação escolares em municípios vizinhos e na rede estadual do ensino fundamental. Por outro lado, o recebimento de FPM por municípios vizinhos de pequena população reduz o provimento de bens e serviços de saúde em cidades próximas e maiores, o que pode ocorrer devido à redução da demanda por serviços de saúde. A piora de alguns indicadores globais de saúde é um indício, no entanto, de que podem existir problemas de coordenação para os prefeitos reterem seus gastos em saúde. De fato, quando controlamos pela margem de vitória nas eleições municipais e consideramos apenas cidades vizinhas com prefeitos de partido diferentes, o efeito spillover é maior em magnitude, o que indica que incentivos políticos são importantes para explicar a subprovisão de serviços em saúde, por um lado, e o aumento da provisão de bens em educação, por outro. Nós também constatamos um efeito positivo do FPM sobre votos para o partido do governo federal nas eleições municipais e nacionais, e grande parte desse efeito é explicado pelo spillover do FPM de cidades vizinhas, mostrando que cidades com dependência econômica do governo federal se tornam a base de sustentação e apoio político desse governo. Por fim, nós encontramos um efeito ambíguo do aumento de receita devido ao FPM sobre a competição eleitoral nas eleições municipais, com uma queda da margem de vitória do primeiro colocado e uma redução do número de candidatos, o que pode ser explicado pelo aumento do custo fixo das campanhas locais.