ABSTRACT. In this paper, we propose a novel rational-inattention approach to model forecasting behavior of professional forecasters in the \textit{Focus Survey}, gathered by the Brazilian Central Bank (BCB). This is a world-class data base which has a few unique features that sets it apart from other well-known expectation surveys. Our framework uses the recent theoretical progress in the rational-inattention literature while keeping the model simple enough that it could be taken to the data and tested using standard econometric techniques. This is important, since theory is ahead of measurement in this field, where empirical testing and validation has been relatively scarce. In our modelling strategy, we recognize the dynamic nature of the agent's rational-inattention optimization problem. Indeed, we find an optimal sequence of decisions instead of a sequence of optimal static solutions, which are prevalent in this literature. Our dynamic optimal solution is based on the Kalman-Filter approach to rational inattention proposed by \citet{MACKOWIAK_ETALL_2018}. We advance with respect to their setup because we recognize and model the mixed-frequency nature of the agent's decision -- forecast on a daily basis a random variable that is sampled at the monthly frequency. We believe this is important from a practical point-of-view, where quite often agents must face a daily attention problem predicting random variables that are sampled at a much lower frequency, solving a distribution ("interpolation") problem to obtain an optimal forecast.

Our model and our empirical strategy entail the following ingredients. First, on a daily basis, agents solve a cost-benefit problem where their decision to update information (extensive margin) and how much information to use in updating (intensive margin) essentially depend on their individual cost-benefit ratio. Second, given the well documented heterogeneity of updating behavior in the Focus data base -- either across agents or across time (days) -- the model assumes that individual attention is characterized by an i.i.d. draw for a given day. However, its mean is allowed to vary across days. Third, we compute the moments implied by the rational-inattention model on three dimensions: the daily mean-squared forecast error (MSFE), the daily probability to update the forecast in the data base, and the daily \textit{ARMA} structure implied by the best \textit{ARMA} model that fits the monthly observations. These implied moments are then confronted with moments computed using the data, with structural parameters estimated using the simulated method of moments. Correct specification of the model is tested by means of an over-identifying-restriction test.

Our empirical results do not reject the structural rational-inattention model proposed here. Moreover, using model estimates, a number of counter-factual exercises are proposed and executed. Using them, we are able to propose improvements in the updating incentive scheme currently used by the Brazilian Central Bank.

ABSTRACT. We provide a new way to select the regularization parameter for the LASSO and adaLASSO. It is based on the theory, as Bickel et. al. (2009) and incorporates an estimate of the variance of the noise. We show theoretical properties of the procedure and Monte Carlo simulations showing that it is able to handle more variables in the active set than other popular options for the regularization parameter.

ABSTRACT. Estimating patterns of occurrence of cases and deaths related to the COVID-19 pandemic is a complex problem. The incidence of cases presents a great spatial and temporal heterogeneity, and the mechanisms of accounting for occurrences adopted by health departments induce a process of measurement error that alters the dependence structure of the process. In this work we propose methods to estimate the components of trend and seasonality in the cases of COVID-19, controlling for the presence of measurement error. This decomposition is presented in Bayesian time series and spatio-temporal models for counting processes with latent components, and compared to the empirical analysis based on moving averages. We applied time series decompositions for the states of São Paulo, Rio de Janeiro, and Amazonas, and a spatio-temporal analysis for all occurrences of deaths at the state level in Brazil.

ABSTRACT. This paper examines the degree of risk sharing within Brazilian regions. It also investigates their economic integration with MERCOSUL countries. Our findings suggest a ``border effect'' in which risk sharing is higher at the regional by comparison to the international level irrespective of the geographical distance. We also document the importance of fiscal federalism on smoothing income volatility. Our results point out to a share of around 13% of income shocks being smoothed by the flow of regional taxes and transfers. We also find evidence of a redistributive role of the federal government.

ABSTRACT. This paper investigates the importance of the age composition in the Covid-19 pandemic. We augment a standard SIR epidemiological model with individual choices on work and non-work social distancing. Infected individuals are initially uncertain unless they are tested. We find that older individuals socially distance themselves substantially in equilibrium. Confining the old even more reduces their welfare. Confining the young extends the duration of the epidemic, with negative consequences on the old if the epidemic cannot be controlled after confinement. Testing and quarantines save lives, even if conducted just on the young, as does separation of activities by age. Combining policies can increase the welfare of both the young and the old.

ABSTRACT. Several countries of Sub-Saharan Africa have experienced episodes of long lasting droughts throughout the 1970-2014 period. I examine the impacts of rainfall in the reallocation of labor from agriculture towards non-agriculture, and on aggregate labor productivity. I use a structural transformation model with agricultural and non-agricultural sectors, whose production functions use only labor as input, and rainfall impacts both the level and the growth of labor productivity. I find a significant negative impact of rainfall shortage on agricultural labor productivity, that translates into a slower process of labor reallocation out of agriculture. This negative impact was particularly severe in the countries of Western Africa. For example, if there were no droughts in Burkina Faso, the share of labor in agriculture would be 4.5% lower than observed in 2014 and aggregate labor productivity would be 22.4% higher.

ABSTRACT. In this paper, we study the time variation of the risk premia in U.S. Treasuries bonds. We propose a novel approach for deriving a single state factor consistent with a dynamic term-structure with unspanned risks theoretically motivated model. Using deep neural networks to uncover relationships in the full set of information from the yield curve, we derive a single state variable factor that provides a better approximation to the spanned space of all the information from the term-structure. We also introduce a way to obtain unspanned risks from the yield curve that is used to complete our state space. We show that this parsimonious number of state variables have predictive power for excess returns of bonds over 1-month holding period. Additionally, we provide an intuitive interpretation of derived factors and show what information from macroeconomic variables and sentiment-based measures they can capture.

ABSTRACT. A monthly tail risk measure is estimated for Brazil based on the risk neutral excess expected shortfall of a set of size, book-to-market and momentum portfolios. The tail risk measure peaks at stock market crashes, financial crises and episodes related to political uncertainty. Furthermore, the tail risk index provides relevant information about contemporaneous and future market returns, even after controlling for risk factors, past returns and other uncertainty measures. In order to investigate the origins of tail risk variation, daily news from the front-page of the largest financial newspaper in Brazil, Valor Econômico, are extracted using Machine Learning techniques. The co-movement between news and tail risk suggests that the tail risk measure captures concerns related to both economic/financial and government/policy uncertainty, while the major perception of the Brazilian investor regarding tail risk pertains to financial and political crises. Moreover, the tail risk capacity in explaining current market returns and its short term predictive power seem to be related to the economic concerns captured by the measure, while the predictive capacity for longer horizons is associated to government/policy and crisis concerns. Overall, the evidence supports that the estimated tail risk measure is a comprehensive measure of uncertainty and market fears.

ABSTRACT. We study the effects of a price transparency shock in the Brazilian OTC equity lending market. Previously, a publicly available stock-specific loan fee benchmark was the average fee of the past 15 trading days. On March 1, 2011, this interval was reduced to 3 days, significantly improving short-sellers' ability to predict current loan fees. We find that loan fees fell, lending volume increased, total lending revenue remained stable, high-cost lenders lost market share, and price efficiency increased after the benchmark change. Only a few countries benefit from loan fee benchmarks and our results may be relevant to regulators.

ABSTRACT. This paper studies how decentralization affects local development. We exploit an episode of a remarkable increase in the number of new municipalities in Brazil to understand the long run impacts of municipal splits. Using a rich panel of spatial and administrative data, we first show that splits are initiated by poor and rural districts motivated by fiscal incentives and neglect from their parent municipality. When using areas whose request to split was never approved as a control group, we find that splitting increases economic activity and public service delivery in new municipalities. Parent municipalities remain unaffected. We document that increases in fiscal revenues and reduced physical remoteness from their administrators are the main channels explaining the results. Our findings suggest that decentralization in the presence of equalizing intergovernmental grants may effectively promote local development.

ABSTRACT. This paper uses a novel dataset on Brazilian bankruptcy requests to investigate whether the congestion of courts affects a creditor's decision to request or not the liquidation of a defaulted debtor. Exploiting the large variation in the level of congestion of the courts across the state of São Paulo, we find that banks are more passive with defaulted debtors that are located in municipalities with more congested courts, waiting longer since default until requesting the liquidation. Implementing an instrumental variable strategy based on the laws that rule the judicial organization, we find results in the same direction. With a simple theoretical framework, we argue that the possible mechanism is that creditors' recovery in liquidation is lower in less efficient courts, potentially increasing firms' bargaining power and leading creditors to postpone the decision about requesting the liquidation of a debtor. This evidence suggests that the congestion of the judiciary affects the use by the creditors of their legal protection prescribed by the bankruptcy law.

ABSTRACT. Distortions tied to firms' characteristics are associated to misallocation of resources and to loss of aggregate productivity. Although estimation and identification of production functions received special attention in Economics, a discussion on the performance of available methods under the presence of severe endogenous distortions is still necessary. We show that under those circumstances some assumptions of popular production function estimators are systematically violated, introducing a particularly different bias from mismeasurement. More specifically, endogenous distortions generates conditional biased estimates of firm-level productivity and affect dispersion measures. Ignoring those hidden effects may drive misleading policy conclusions when it comes to productivity analysis, specially when applied to economics plagued by severe distortions.

ABSTRACT. In this work we show that contemporary differences in domestic labor markets within Brazil trace their origins to female slavery’s prevalence more than 120 years ago. We test the hypothesis of historians and anthropologists that the colonial period culture of having women slaves into the house doing domestic work persisted after slavery abolition until modern days. We find that, consistent with existing hypotheses, municipalities that traditionally have a higher share of women domestic slaves in 1872 have also a higher share of female labor force occupied as housemaids and other low skill domestic occupations today. Instrumental variables approach using agriculture suitability for sugar cane enhance the reliability of our results. We also show that these results cannot be fully explained by current development, inequality or human capital accumulation levels and find suggestive evidence that culture may be an important channel of persistence.

ABSTRACT. This paper studies the long run effects of female slavery in the 19th century on violence against black women. I take advantage of the draft of male slaves by the Brazilian empire during the Paraguayan war (1864-1870), and instrument the share of female slaves using the distance to the war front. I found that Brazilian municipalities with a higher share of female slaves in 1872 have more cases of physical and psychological violence against black women today. The same pattern is not observed for cases against white and brown women.

ABSTRACT. This article documents the impact of the Brazilian railway network on structural transformation between the late nineteenth and middle twentieth century. We exploit variation induced by geographic location, where municipalities near the least-cost routes were more likely to be connected to railway system, to identify the causal effects of railroads on structural transformation. We show that the expansion of the transportation infrastructure shifts the workers from agriculture to manufacturing. We provide evidence that market integration and the adoption of new technologies by manufacturing firms were two important mechanisms that explain the change in the occupational structure of the economy.

ABSTRACT. We characterize the effect of short selling costs on interactions between informed and uninformed speculators, showing how this dynamic impacts corporate decisions such as investment and stock repurchases. Low shorting costs allow for manipulation to coexist with informed trading, reducing price informativeness and investment. Manipulation becomes less profitable as shorting costs increase, making prices more informative and boosting investment when speculators are less likely to be informed. At high shorting costs, informed shorting is unprofitable even in the absence of manipulation threats, resulting in low price informativeness and constraining firms' access to financing. Our model shows that the ability to pre-commit funds before prices reflect speculators' information yields a negative relation between investment and shorting costs. Critically, it demonstrates how managers can stop manipulative shorts through stock repurchases, leading to efficient investment. Stock liquidity, cash flow uncertainty, and management--creditor agency problems are shown to shape the impact of short selling costs on corporate policies.

ABSTRACT. We propose a new approach to restrict the set of admissible risk-neutral measures and construct bounds for non-redundant asset prices in incomplete markets. It is based on the idea that there is no a priori reason for a risk-neutral measure to deviate from the physical distribution other than satisfying the no-arbitrage pricing constraints for the basis assets. We select risk-neutral measures that minimize a family of dispersion loss functions each presenting different sensitivities to all higher-order risk-neutral moments. The minimum dispersion admissible set is explicitly characterized, so that it is possible to recover the measure that prices a given asset. We apply our method to the canonical problem of option pricing without dynamic trading. In simulated economies and in a large-scale empirical application pricing 1,817,095 S\&P 500 options, the minimum dispersion bounds contain most of the observed option prices and are tighter than alternative methods. Moreover, our approach suggests an alternative explanation for the implied volatility smile based on a multiplicity of heterogeneous marginal risk-neutral measures.

ABSTRACT. How do political leader's words and actions affect people's behavior? We address this question in the context of Brazil by combining electoral information and geo-localized mobile phone data for more than 60 million devices throughout the entire country. We find that after Brazil's president publicly and emphatically dismissed the risks associated with the COVID-19 pandemic and advised against isolation, the social distancing measures taken by citizens in pro-government localities weakened compared to places where political support of the president is less strong, while pre-event effects are insignificant. The impact is large and robust to different empirical model specifications, and definitions of political support and events. Moreover, we find suggestive evidence that this impact is driven by localities with relatively higher levels of media penetration, municipalities with presence of active Twitter accounts, and municipalities with a larger proportion of Evangelic parishioners, a key group in terms of support for the president.

ABSTRACT. We show that the Brazilian trade liberalization from the 1990s led to a permanent relative decline in the vote share of left-wing presidential candidates in the most affected regions. This happened even though the incumbent party during the reform belonged to the center-right and the left was identified with protectionist policies. To rationalize this response, we consider a new institutional channel for the political effects of trade shocks: the weakening of labor unions due to a reduction in manufacturing and, in Brazil, formal employment. We show that proxies for union strength---such as the number of workers directly employed by unions and union density---indeed declined in regions that suffered relatively larger increases in exposure to foreign competition. Finally, we show that the effect of the trade liberalization on the vote share of the left is stronger in regions that had higher union density before the reform. These findings are consistent with the hypothesis that tariff cuts during the trade reform reduced the vote share of the left partly through the weakening of labor unions. This institutional effect is essentially different from the individual-level responses, motivated by economic and identitary concerns, that have been considered in the literature.

ABSTRACT. We investigate how market power affects wage inequality. Market power is driven by the market structure of the goods market as well as in the distribution of firm-specific technology. We ask how these contribute to the rise in the Skill Premium, the premium of the average college wage relative over the average non-college wage. We estimate the firm-specific technology as well as the market structure and find that the increase in market power contributes over half to the rise in the skill premium. We also find that there is a substantial welfare cost from excess inequality due to market power.