This dissertation assesses policy effects of conservation efforts adopted within the scope of the federal action plan to combat Amazon deforestation in Brazil. Chapter 1 provides a description of key policy changes and surveys the associated effectiveness literature. It finds evidence that supports the action plan’s efficacy in reducing aggregate deforestation levels, but notes that indirect impacts of conservation policies have received little attention.

The remaining chapters explore direct and indirect impacts of action plan policies using a georeferenced ten-year panel dataset to account for spatial dynamics. Chapter 2 tests whether legal territorial protection grants actual protection against advancing deforestation. Using a measure of neighboring clearing activity to capture local deforestation risk, the analysis compares forest clearing outcomes in unprotected and protected territory under equivalent deforestation pressures. The empirical strategy draws on the dataset’s raster structure to mitigate concerns of potentially confounding unobservables via the use of raster cell fixed effects. Results document protection’s efficacy in a high-risk context, with significantly less forest being cleared in protected cells than in unprotected ones. Yet, although protected territory effectively shields vegetation under its domain from advancing deforestation, it appears to deflect clearings to unprotected areas. Protection therefore affects regional forest clearing dynamics, but not the overall level of deforestation. Chapter 3 investigates whether changes in tropical regeneration constituted a spillover effect from law enforcement targeting forest loss. Secondary vegetation was vulnerable during the first decade of the action plan, which neither promoted tropical regeneration nor sought to conserve existing secondary vegetation. Moreover, regeneration remained undetected in satellite-based forest monitoring systems. Still, during this period, the extent of Amazon secondary vegetation increased by nearly 7 million hectares. The final part of this dissertation examines whether law enforcement contributed to this growth, albeit unintentionally. The empirical strategy uses a ten-year cross-sectional difference in observed regeneration outcomes to address the intrinsically time-consuming nature of this phenomenon. Results are shown to be robust to the inclusion of a host of raster cell-level controls, mitigating concerns about omitted variable bias. 

Findings indicate that the intensity of enforcement in a location’s close surroundings is associated with both increased probability of secondary vegetation expansion and increased area of secondary vegetation in that location. This lends support to the hypothesis that environmental offenders, once faced with a higher perceived cost of engaging in illegal deforestation, abandoned the area they were operating in and thereby allowed a natural process of forest regrowth to occur. The spillover effect of enforcement on regeneration appears largest in places that have undergone neither too much nor too little deforestation: in the former, forest clearings and non-forest land use are probably more consolidated, and regrowth is therefore less likely; in the latter, there is still relatively little area for the forest to grow back in.

Counterfactual exercises shed light on the magnitude of this effect. An enhanced satellite-based monitoring system for targeting enforcement would have resulted in nearly 300 thousand additional hectares of secondary vegetation.

This paper examines the extent to which government audits of public resources can reduce corruption by enhancing political and judiciary accountability. We do so in the context of Brazil’s anticorruption program, which randomly audits municipalities for their use of federal funds. We find that being audited in the past reduces future corruption by 8 percent, while also increasing the likelihood of experiencing a subsequent legal action by 20 percent. We interpret these reduced-form findings through a political agency model, which we structurally estimate. Our results suggest that the reduction in corruption comes mostly from the audits increasing the perceived nonelectoral costs of engaging in corruption.

Seguindo modelo desenvolvido por Prescott (2002), este estudo questiona qual seria hoje a renda per capita do brasileiro em idade ativa se o país adotasse os sistemas tributários de três países desenvolvidos: EUA, França e Inglaterra. Esta pesquisa mostrou que, no Brasil, a tributação sobre consumo é superior à francesa, enquanto que os impostos sobre renda são praticamente equivalentes aos cobrados nos EUA. E apresentou, para comparar, o total de horas trabalhadas por indivíduo em idade ativa de cada país. Por fim, apresentou o resultado de uma possível aplicação deste modelo para o Brasil, concluindo que o PIB por pessoa em idade ativa seria superior ao atual com o sistema americano, inferior com o francês e praticamente equivalente com o britânico.

In this paper we study the impact of Bolsa Família’s CCTs on the number of formal jobs at the municipality level. We find a highly heterogeneous effect across municipalities. Previous works on this topic considered only the effect on the average municipality. We show evidence that the program’s effect can be five times higher on the poorest municipalities, corresponding to 1% increase in formal jobs for a 10% increase in the number of beneficiaries. To accomplish so, we combine administrative data on the program enrollment with data on the universe of formal jobs. This allows us to overcome the problem of measurement error in the treatment variable. Robustness checks indicate that the effect remains significant after controlling for health, educational, climatic shocks, financial and political factors.

Estudei a transmissão da política monetária via canal de crédito com uma base mensal dos balanços dos bancos brasileiros de 2003 a 2017. Concluí que bancos que ofertam mais crédito subsidiado pelo governo e proveniente de capta-ções compulsórias são menos sensíveis a variações da política monetária. Dessa forma, ao manter uma política de incentivo econômico via expansão do crédito, o governo está obstruindo esse canal de transmissão e consequentemente reduzindo a potência da política monetária. Adicionalmente, concluo que os bancos continu-am rebalanceando sua carteira de crédito meses após o choque monetário, o que corrobora a visão de Bernanke & Blinder (1992) de que os bancos não desfazem seus contratos de crédito vigentes, mas apenas se recusam a fazer novos. Por fim, verifico um efeito de flight-to-quality em períodos de aperto monetário e menos liquidez na economia. Bancos tendem a reduzir mais seus créditos de alto risco do que os créditos de baixo risco. Entretanto, ao manter um alto volume de crédito subsidiado e proveniente de depósitos compulsórios esse efeito é mitigado.

We consider a new, flexible and easy-to-implement method to estimate thecausal effects of an intervention on a single treated unit when a control group is not available and which nests previous proposals in the literature. It is a two-step methodology where in the first stage, a counterfactual is estimated based on a large-dimensional set of variables from a pool of untreated units by means of shrinkage methods, such as the least absolute shrinkage and selection operator (LASSO). In the second stage, we estimate the average intervention effect on a vector of variables, which is consistent and asymptotically normal. Our results are valid uniformly over a wide class of probability laws. We show that these results hold even when the exact date of the intervention is unknown. Tests for multiple interventions and for contamination effects are derived. By a simple transformation of the variables, it is possible to test for multivariate intervention effects on several moments of the variables of interest. Existing methods in the literature usually test for intervention effects on a single variable and assume that the time of the intervention is known. In addition, high-dimensionality is frequently ignored and inference is either conducted under a set of more stringent hypotheses and/or by permutation tests. A Monte Carlo experiment evaluates the properties of the method in finite samples and compares it with other alternatives. As an application, we evaluate the effects on inflation, GDP growth, retail sales and credit of an anti tax-evasion program.

This thesis is composed of three papers. The first one studies the relation between monetary policy power and the availability of earmarked loans. To that end, we estimate the responses of sectoral macroeconomic variables to monetary policy shocks identified through sign and equality restrictions in a factor-augmented VAR (FAVAR). We find that monetary policy loses power in sectors with a larger share of earmarked loans among its bank debt. The second paper proposes the introduction of new restrictions to identify monetary policy shocks in SVARs. In particular, besides standard sign restrictions on interest rates and inflation, we propose to add as an identification restriction the inability of monetary policy to have real effects ten years after the shock. This paper presents evidence of the model consistency of this neutrality restriction both for the canonical 3- equation new keynesian model and the Smets and Wouters (2007) model. In a simple empirical application, this paper shows that this restriction may be important to recover real effects of monetary policy. The third paper shows that foreign banks can mitigate informational barriers vis-a-vis private domestic banks by observing their peers’ behavior. Conditional on a loan application being filed by a SME firm, we find that the existence of past loans of this firm with private domestic banks constitute a more important predictor that a loan will be granted by foreign banks in comparison to private domestic banks. Our results are compatible with the view that the higher ability of private domestic banks to access informationally opaque business credit risks makes past loans with these banks a more valuable signal for foreign lenders

Este trabalho investiga os efeitos da incerteza sobre a atividade econômica no Brasil. Para isso, são construídas diversas proxies que buscam capturar o nível de incerteza vigente na economia brasileira (incerteza doméstica) e em vários de seus principais parceiros comerciais (incerteza externa). Em seguida, são estimados modelos de vetores autorregressivos (VAR) estruturais, cujas funções de resposta ao impulso sugerem efeitos contracionistas significativos da incerteza sobre a atividade, em particular sobre o investimento. Além disso, as estimativas indicam que os efeitos da incerteza doméstica são mais acentuados do que os da incerteza externa. Pode-se afirmar, portanto, que os níveis de incerteza vigentes no Brasil desde as eleições presidenciais de 2014 representam importante fator por trás da recessão subsequente. Estima-se que caso não houvesse a expansão de incerteza doméstica observada a partir do segundo semestre de 2014, a produção industrial em 2015 teria sido, em média, entre 0,9% e 3,9% maior, dependendo da variável proxy de incerteza utilizada. No caso do IBC-Br, este teria sido entre 0,4% e 1,3% maior. Os resultados encontrados são robustos para diversas alterações no modelo.

Revisto em fevereiro de 2018