ABSTRACT. Following Cripps, Ely, Mailath and Samuelson (ECMA 2008), we seek a Bayesian learning foundation of common knowledge. In a setting with private signals, we find general conditions for ``common learning'' that allow for rich signal spaces and unbounded log-likelihood ratios, extending their result beyond the finite signal case. Examples of signal structures that satisfy our conditions include: (i) those in which the signals take values in compact metric spaces and the signal distributions have strictly positive continuous densities; (ii) the class of multivariate Gaussian location experiments.

ABSTRACT. We study trade in dynamic decentralized markets with adverse selection. Differently from the literature on the topic so far, we assume that the informed sellers make the offers, so that signalling through prices is possible. We establish basic properties of equilibria, provide necessary and sufficient conditions for equilibrium existence, and discuss the two-type case and separating equilibria in detail. Our main result is that market efficiency---measured by the maximum equilibrium welfare---is \emph{invariant} to trading frictions. This shows that signalling and screening have markedly different implications for the relationship between market efficiency and trading frictions.

ABSTRACT. This paper presents a general equilibrium model of price system formation based on the diffusion of information regarding trade opportunities. Decision makers have constrained consideration sets but their consideration sets incorpo rate learning dynamics about opportunities to trade (by consumers and suppliers of factor of production) or to profit (by entrepreneurs). This learning results in the creation of new industries and changes the equilibrium allocation. Con- strained choice by consumers implies in market power by the firms, so market conditions are always imperfectly competitive. Given certain assumptions, in particular that agents have perfect memory regarding their consideration sets and constant technology, the equilibrium of the model converges to a stationary equilibrium that corresponds to a competitive equilibrium with complete markets. This model predicts a “sticky” adjustment to shocks which features cyclical behavior of employment and can produce pricing dynamics where the degree of deviation from competitive conditions varies over time and it can also explain other economic events such as the emergence of superstar firms and the decline in the labor share.

ABSTRACT. I formulate a firm dynamics model with heterogeneous firms and calibrate to the Brazilian Data. Using this model, I investigate the dynamic effects of misallocation on productivity growth that can arise from a government-driven credit expansion. In particular, I argue that the misallocation induced by the credit policy is a contributing factor to the dismal performance of the Brazilian economy in the 2010s. Three conclusion can be drawn from the results. First, subsidized establishments become larger. Second, the productivity cut-off for entering the market is lower for subsidized firms. Last, an increase in the share of subsidized firms and the subsidy benefit puts further pressure on the government budget and makes the policy less sustainable.

ABSTRACT. Monetary policy rules are mostly discussed in terms of Taylor rules, which conjectures that the monetary authority set the short-term interest rate in response to deviations of inflation from its target and of output to some measure of potential output. We show that replacing the output gap with the Wicksellian efficient interest rate as the main indicator of real activity make DSGE models fits the data better, improving their ability to reproduce the Brazilian federal funds rate (Selic rate) time series. Previous result holds for both closed and small open economy versions of the Calvo sticky price model with monopolistic competition.

ABSTRACT. This paper investigates the pass-through from observed and expected policy interest rates to lending rates in the Brazilian economy, accounting for financial institution-specific characteristics, asymmetric behavior and dynamic effects due to potential inertia in loan rates. We use a unique and non-public dataset with identified expectations by financial institutions that avoid loss of information due to aggregation of expectations by the mean or median. The sample is from January 2012 to April 2019, on a weekly basis, and has variability by loan modalities, financial institutions and time. We apply fixed effects estimation and account for asymmetric and dynamic effects in the responses of loan interest rates to changes in either observed or expected Over-Selic rate. We find robust evidence of full and asymmetric pass-through to highly persistent lending rates for both households and non-financial corporations. Asymmetry indicates that downward movements in loan interest rates are always refrained by the financial institutions. In some cases, there is evidence of overshooting pass-through, with estimated pass-through coefficients statistically greater than one. These findings, explained by market imperfections and other frictions, are insightful to understand why lending interest rates are so high in the Brazilian economy.

ABSTRACT. The main debate on affirmative action in higher education focuses on mismatch, a situation where students benefiting from preferential admissions struggle with their college-level work because they lack prior academic preparation. In the United States, these students may be able to switch majors if they underperform in the originally intended major. Only in the extreme they might drop out. What happens when major switching is not allowed? In this paper, we examine the margins of adjustment for beneficiaries of affirmative action in a top university in Brazil, where prospective students must choose a major prior to the entrance test, and cannot switch it while in college. Not surprisingly, we find a relatively high dropout rate, but we also provide evidence of strong catching-up. Because they fail in more subjects early on, to successfully graduate, students benefiting from preferential admissions end up reducing the number of credit hours taken in the first and second years of college, but compensate by taking more credit hours in the third and fourth years.

ABSTRACT. We investigate how job flexibility impacts the gender wage gap (GWG) in Brazil. As women are still viewed as the main providers of child care and household chores, they are more disadvantaged in less flexible workplaces. We explore two dimensions of job flexibility: one related to the characteristics of the occupation itself and another specific to the firm. To that end, we create an inflexibility index for each occupation, and we identify exporters as firms with, on average, less flexible workplaces. Our results reveal a wage premium for occupation inflexibility, and this premium is greater among exporting firms. We also find that the GWG increases with the occupations' degree of inflexibility, and this effect is even stronger among exporters.

ABSTRACT. This paper studies the long-run consequences of a major trade liberalization reform in Brazil on infant mortality. By exploiting geographic variation in pre-existing industrial composition, baseline tariff levels, and the timing of the reform, we document that the policy led to substantial and persistent declines in infant mortality. We show that these results are not driven by pre-existing trends in infant mortality rates, and, that the coefficients are robust to controls for microregion-level characteristics. A variety of evidence suggests that reductions in the opportunity cost of time-intensive childcare investments are an important source of the gains in infant health. In relation to this, we highlight that (1) the health effects of trade policy coincide with a deterioration in the mothers' labor market conditions, and (2) the roll-out of a major healthcare program that arguably improved the access to free preventive and primary health services magnifies the effects of the trade policy, providing further evidence on the mechanisms.

ABSTRACT. This paper investigates mental health consequences of Fundao dam rupture, in Mariana, Brazil, in 2015. Unlike previous studies that analyze the affected region as a unit, we investigate each affected region separately hence accounting for heterogeneous effects due to the vast spectrum of the disaster. We use a difference-in-differences framework to investigate mental disorders hospitalizations and use the approach of Ferman and Pinto (2019) to correct our standard errors estimations due to the few number of treated units in each affected group. We show a sizable effect in mental disorders hospitalizations in Minas Gerais state, especially at municipalities located near Fundao dam (upstream Candonga reservoir). Our estimations indicate that the disaster induced at least a two-fold increase in mental disorders hospitalizations in this region with considerable impacts in a broad set of mental disorders. We also find effects for municipalities in Minas Gerais state located downstream Candonga reservoir (a four-fold increase in hospitalizations due to disorders with onset on childhood and adolescence) and indirectly affected ones (a two-fold increase in stress-related hospitalizations). Finally, we bring evidences of the late and persisting effects of the disaster in mental health, with an increased number of hospitalizations persisting (if not starting) years after the dam collapsed.

ABSTRACT. This paper presents evidence that the exogenous release of information on mismanagement of federal resources to public healthcare in Brazil led to lower funding to this sector in following years. I leverage from an inspection policy based on randomized audits to the public accounts of Brazilian municipalities, which was implemented nationwide by a federal agency. Using data from reports generated from these audits, combined with administrative data on federal transfers to health and healthcare programs, I show empirical, reduced-form estimates that transfers are significantly reduced to municipalities where a higher incidence of irregularities was revealed through the audits. Preferred estimates suggest that procured transfers to infrastructure maintenance in public healthcare establishments (which are made through contractual agreements with local vendors or governments called "partnerships") in those municipalities are reduced in at least 58%, on average. Back-of-the-envelope calculations indicate such figure should correspond to total cuts of about 5.57 billion Reais (roughly 1.4 billion US Dollars). These reductions, however sizable, do not appear to impact provision at the endpoints of such transfers, as additional estimations show that audits are inconsequential to observable measures of public healthcare infrastructure. Both results combined suggest that, at the observable extensive margin, supervision through random audits may help improve efficiency in public healthcare spending.

ABSTRACT. We study how the introduction of a new information and communication technology affects the organizational and educational structure of firms. Using Brazilian data, we circumvent endogeneity issues by exploiting quasi-random variation in the roll-out of fast Internet and a physical constraint in ADSL connectivity. Our results show that establishments respond to broadband availability by expanding both the shares and absolute numbers of leadership positions (high-educated employees) to the detriment of other occupational layers (types of employees) and reducing overall employment. These impacts vary by industry, location, and firm age. The increase in leadership positions is driven by promotions instead of new hires, and is not restricted to the promotion of high-educated individuals or employees previously working in positions just below leadership. While results show labor market polarization, they also suggest that these structural changes mediate firm survival in the market.

ABSTRACT. Lending cyclicality may impede economic growth during booms or exacerbate depression in burst periods. In this paper, I explore branch-level cyclicality in lending and its role in the transmission of monetary policy. While most of the literature relies on bank-level data, I use the branch’s balance sheet to ascertain whether local markets may impose restrictions on monetary policy conduction. My results indicate that an unpredicted increase of 0.25pp in short-term interest rates leads to a sizable decrease in the branch's lending. But this reduction is roughly 1.3pp smaller if branches are on their highest month of historical lending volume. Thus, I show lending cyclicality seems to act as a shield for increases in interest rates. These results suggest that policymakers can be more effective by considering an optimal market composition and timing when deciding for interest rates. JEL Codes: E32, E44, E52, G21

ABSTRACT. This paper studies the links between competition in the lending market and spreads of bank loans in Brazil. Evidence from a dataset of more than 13 million loan-level observations from private banks shows a positive relationship between market power, measured by the Lerner Index, and the cost of finance, measured by spreads over the treasury curve. Furthermore, there is evidence of the holdup problem, originated from informational switching costs faced by firms. Private banks engage in a strategy of first competing fiercely for clients by offering a lower loan interest rate and later increasing interest rates as the bank-firm relationship duration increases. Both results are stronger for micro and small firms than for medium and large firms.

ABSTRACT. This paper sets up a link between central bank credibility and attention allocation in the context of investment funds. Dividing the portfolio into inflation-linked assets and equity, we are able to see how the level of central bank credibility affects the first, and how the state of the economy affects the second. From there, the main findings of the model are obtained. First, there is evidence that attention is cyclical since there is a growing incidence of updates to inflation forecast when central bank credibility is falling, alongside with net purchases of NTN-B in those moments. Second, there is more cross-sectional dispersion in inflation-linked assets holdings and returns across funds in low central bank credibility periods and recessions, respectively. Third, there is higher average fund outperformance in recessions. These findings reinforce the fact that attention is allocated optimally and that investors have different skill levels.

ABSTRACT. We introduce speculation, in the spirit of Harrison and Kreps (1978), into a standard real business cycle model. Investors (speculators) hold heterogeneous beliefs about firm growth. Firm ownership, and thus, the firm's discount factor, varies with waves of optimism and leverage. These waves ripple into the firm's investment in hours. The firm's discount factor links the equity premium and labor volatility puzzles. We find that the amplification that can be generated by speculation for any model of beliefs, for a standard calibration, has an analytic upper bound of 1.5. When we specialize to a model of diagnostic beliefs, the volatility of hours is amplified by a factor of 1.15 and a bubble component of 20 percent in asset prices emerges.

ABSTRACT. This paper studies the role of slums in shaping the economic and health dynamics of pandemics. Using data from millions of mobile phones in Brazil, an event-study analysis shows that residents of overcrowded slums engaged in less social distancing after the outbreak of Covid-19. We de- velop a choice-theoretic equilibrium model in which individuals are het- erogeneous in income and some people live in high-density slums. The model is calibrated to Rio de Janeiro. Slum dwellers account for a dispro- portionately high number of infections and deaths. In a counterfactual sce- nario without slums, deaths increase in non-slum neighborhoods. Policy simulations indicate that: reallocating medical resources cuts deaths and raises output and the welfare of both groups; mild lockdowns favor slum individuals by mitigating the demand for hospital beds, whereas strict con- finements mostly delay the evolution of the pandemic; and cash transfers benefit slum residents to the detriment of others, highlighting important distributional effects.

ABSTRACT. We consider a price taking model of insurance provision where consumers are privately informed about their risk level and preferences. The presence of two-dimensional heterogeneity introduces novel distribution effects absent from the one-dimensional model typically studied in the literature. Equilibria necessarily involve pooling, mixing agents with identical willingness to pay for coverage in equilibrium. Focusing on the case of small preference heterogeneity, we obtain explicit formulas for equilibrium prices that allow us to study two comparative static exercises. First, the release of a risk-informative signal leads to interim and ex-ante welfare improvements if, and only if, the signal structure satisfies a certain monotonicity condition, while non-monotonic signals may harm some consumers and be overall welfare reducing. Second, we study the effect of changes to the risk distribution and show that increases in the risk distribution, according to the monotonic likelihood ratio property (MLRP), lead to higher prices and lower welfare, while shifts in the sense of first order stochastic dominance ∗We thank seminar participants at the 2017 and 2019 SAET meetings, INFORMS 2019, FGV/EPGE, UBC, Penn State, University of Pittsburg/Carnegie Mellon, 2019 Southern Economic Association meeting and the 2019 LACE-LAMES. We are also in debt with Andrea Attar, Eduardo Azevedo, Carlos da Costa, Daniel Gottlieb, Li Hao, Pierre-André Chiappori, Wei Li, Michael Peters, Bernard Salanié, Sergei Severinov and André Trindade for useful discussions. Moreira thanks CNPq and Faperj for financial support. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. This research was also supported by the Social Sciences and Humanities Research Council of Canada. †University of British Columbia. email: vitor.farinhaluz@ubc.ca ‡University of Essex. email: piero.gottardi@essex.ac.uk §FGV EPGE Brazilian School of Economics and Finance. email: humberto.moreira@fgv.br 1 can be beneficial for some consumers. These two sets of results are relevant to the study the effect of allowing the use of demographic characteristics in pricing insurance policies and policies targeted at changing the risk distribution in a given market, respectively.

ABSTRACT. A biased sender acquires a signal about an object’s quality and commits to a rule to disclose its realizations to a receiver, who then chooses whether to buy the object. Optimal disclosure rules typically conceal negative signal realizations when the object’s sale is very profitable to the sender and positive signal realizations when the sale is less profitable. I show that, despite this strategic concealment of some signal realizations, the receiver may prefer being informed by a more biased advisor. Specifically, more biased senders can be more informative because their bias produces an additional incentive to invest in acquiring a precise signal of the object’s quality.

ABSTRACT. This article presents identification results for the marginal treatment effect (MTE) when there is sample selection. We show that the MTE is partially identified for individuals who are always observed regardless of treatment, and derive uniformly sharp bounds on this parameter under four increasingly restrictive sets of assumptions. The first result imposes standard MTE assumptions with an unrestricted sample selection mechanism. The second set of conditions imposes monotonicity of the sample selection variable with respect to the treatment, considerably shrinking the identified set. Third, we incorporate a stochastic dominance assumption which tightens the lower bound for the MTE. Finally, we provide a set of conditions that allows point identification for completeness. Our analysis extends to discrete instruments and distributional MTE. All the results rely on a mixture reformulation of the problem where the mixture weights are identified. This extends Lee (2009) trimming procedure to the MTE context. We propose nonparametric estimators for the bounds derived, provide a numerical example and simulations that corroborate the bounds feasibility and usefulness as an empirical tool. We highlight the practical relevance of the results by analyzing the impacts of managed health care options on health expenditures, following Deb, Munkin, and Trivedi (2006).

ABSTRACT. In this paper we develop theoretical criteria and econometric methods to evaluate policy interventions from a welfare standpoint when it is assumed that agents are loss-averse. The criteria define a weak partial ordering over distributions of policy-induced gains and losses, that is, they apply over a class of social value functions that model individual preferences with increasing, concave and loss-averse attitudes towards outcomes. We show that these social preferences imply a criterion based on distribution functions that is similar to the condition implied by first order stochastic dominance. We develop tests using this loss averse-sensitive dominance criterion using nonparametric plug-in estimates to provide a practical decision rule using observed data. We establish the limiting distributions of test statistics by showing that, as maps from distributions, they are directionally differentiable, and this condition implies that inference can be conducted by a special resampling procedure. The fact that gains and losses are considered means that outcomes, representing the difference between two random variables, may not be identified without strong assumptions. Therefore, we extend comparison criteria, test statistics and resampling procedures to the partially-identified case. Finally, we illustrate our methods with a welfare comparison of two aid programs.

ABSTRACT. We analyze how peers affect the participation in a family allowance for poor families in Chile called Subsidio Unico Familiar (SUF) using a regression discontinuity design. To identify the spillovers effect, we exploit variation in the information about social programs due to a home-visitation program for families in extreme poverty introduced in 2002 called Chile Solidario (CS). Conditional on an index of wealth, eligibility to receive the home-visits are random around municipality level cutoffs. We find that not only individual participation in CS increases the take-up of SUF, but that participation in SUF also responds to neighbors' participation in CS. In particular, we estimate network effects on participation in social programs of about .5. To put this figure in context, if neighbors' participation in SUF increases by 3.7pp, then the participation of the household increases by 1.5pp. To understand the mechanisms through which peer effects operate, we allow for heterogeneity for different types of networks.

ABSTRACT. We propose a simple way to assess the quality of asymptotic approximations required for inference methods. Our assessment can detect problems when the asymptotic theory that justifies the inference method is invalid and/or provides a poor approximation given the design of the empirical application. It can be easily applied to a wide range of applications. If widely used by applied researchers, this assessment has the potential of substantially reducing the number of papers that are published based on misleading inference. We analyze in detail the cases of differences in differences with few treated cluster, stratified experiments, shift-share designs, and matching estimators.

ABSTRACT. Our objective is to test the rationality of the forecasters for Brazilian inflation and we analyzethe relationship of rationality with the macroeconomic and electoral variables. We use Survey ofProfessional Forecasters (SPF) for next month’s inflation with monthly data. We consider times seriesand panel data traditional tests as Mincer and Zarnowitz (1969) and West and McCracken (1998)to verify if forecast errors have zero mean and are uncorrelated with variables available at the timethe forecast is made. We also use Rossi and Sekhposyan (2016) test with their asymptotic criticalvalues and finite sample adjusted distribution critical values of El-Shagi (2019). We reject forecastrationality with panel data or time series (consensus) using traditional tests. We do not reject thenull hypothesis of rationality for the consensus inflation forecast if we use fluctuation rationality test.We obtain that forecasters have bias in inflation forecasts in the easing and tightening periods ofmonetary policy or election periods with panel data. But we have that economic cycle and monetarypolicy do not affect the rationality test with panel data. The consensus forecast seems to neutralizethe bias of individual forecasts comparing with panel data and it reduces irrationality only for periodsof recession, monetary policy tightening and without election.

ABSTRACT. Unemployment insurance aims to maintain workers’ consumption while they are unemployed, but may have undesirable effects on the beneficiaries’ labor supply. Theoretical and empirical studies have provided robust evidence of a reduction in the job search effort and an increase in the duration of unemployment, i.e., ex-post moral hazard. By contrast, this paper analyzes the influence of unemployment insurance on the behavior of workers while employed, i.e., ex-ante moral hazard. We use regression kink design and differences-in-differences to estimate the higher probability of termination after workers’ ensure eligibility for the benefit, and after increases in the potential duration of unemployment insurance by one month. Additionally, this paper analyzes the interaction of business cycles with ex-ante moral hazard. Contrary to what one would expect, the probability of termination has a procyclical behavior. The insurance is more widely used in periods of economic expansion than in recessions. According to the theoretical model, this occurs because, in expansion, it is easier to move out of unemployment after the insurance payout, whereas in recession, heading back to the job market is more challenging and, consequently, ex-ante moral hazard is less likely.

ABSTRACT. An important stylized fact about public sector employment is that it predominantly hires skilled and more experienced workers. In this paper, we consider a search and matching model with public sector and on-the-job human capital accumulation that incorporates this stylized fact to study how the public sector employment affects the labor market volatility. In the model, public sector employment affects aggregate fluctuations by changing the composition of workers employed in the private sector. Because workers accumulate human capital and become more productive when employed, the flows of benefits from forming a match are spread over time. In this environment, if the flow into the public sector increases with human capital, then the government hiring policy decreases the firm’s benefit of hiring and the matching surplus, increasing the responsiveness of labor market tightness to shocks. We calibrate the model for the Brazilian economy and show that this mechanism amplifies the effects of public employment on vacancy creation and private sector employment volatility.

ABSTRACT. We study an equilibrium occupational choice model where heterogeneous agents decide to become either workers or entrepreneurs in the formal or informal sector. Informal output is subjected to taxation determined by a combination of managers capital choice and the society’s tolerance of informality. The model is consistent with empirical evidence for the Brazilian informal sector. The counterfactual analysis shows substantial heterogeneity of policy effects on occupational choices (entrepreneur-worker) and within the entrepreneurial choices (formal-informal). Changes in the society’s tolerance of informality lead agents to shift between the two entrepreneurial choices rather than in the entrepreneur-worker dimension.

ABSTRACT. Over the past two decades, the level of international reserves in emerging economies increased significantly. In Brazil, the 2019 level of around 360 billion dollars was considered high by some metrics, based on the precautionary motive. On the other hand, in addition to the opportunity cost, Latin America is also characterized by historically high costs of holding reserves, due to the payment of positive and high interest rates. Behind any model in the literature that studies the optimal level of reserves, there is a trade-off between the insurance benefits and the costs associated with the accumulation of reserves, so that a significant change in this rate is expected to be relevant in the optimization made by the Central Bank. In that sense, recently, the policy-related interest rate in Brazil (Selic) decreased considerably, from 14.25% until October 2016 to 2.25% until June 2020, an all-time low. Addressing this issue, this thesis studies the effect of this change in the direction of Brazil's monetary policy in the management of foreign exchange reserves. Our counterfactual results show that the net FX reserves level - a benchmark adopted by the Central Bank since August 2019-, would have fallen in this period, but the decline in the interest rates made it possible for the Central Bank to keep a roughly stable level until 2019.

ABSTRACT. This paper investigates how technologies that use different combinations of artificial and human intelligence are incorporated into classroom instruction, and how they ultimately affect students' outcomes. We conducted a field experiment to study two technologies that allow teachers to to outsource grading and feedback tasks on writing practices. The first technology is a fully automated evaluation system that provides instantaneous scores and feedback. The second one uses human graders as an additional resource to enhance grading and feedback quality in aspects in which the automated system arguably falls short. Both technologies significantly improved students' essay scores, and the additional inputs from human graders did not improve effectiveness. Furthermore, the technologies similarly helped teachers engage more frequently on nonroutine tasks that support pedagogy individualization.

ABSTRACT. We examine how exposure to a church-affiliated TV channel affects religiosity and compliance with modes of economic and social behavior prescribed by the church. The empirical strategy exploits the expansion of RecordTV, a Brazilian TV channel that started broadcasting religious content over the 1990s. Our results show that an increase of one standard deviation in RecordTV signal strength leads to an increase of 1.1 p.p. in the share of Pentecostals, an increase of 0.95 p.p. in the fertility rate and a reduction of 1.1 p.p. in female labor force participation. We also find an increase of 1.3 p.p. in the share of votes for Pentecostal candidates. The results are robust to an IV strategy where we exploit the placement of transmitters prior to RecordTV being religiously affiliated.

ABSTRACT. Using admission lotteries and registry data linking labor market outcomes, we study the effect of a vocational training program focused on disadvantaged individuals in Brazil. The intensive program is an 18-month classroom training coupled with a 6-month on-the-job training provided by government-sponsored training centers. Comparing 7,000 winners and 108,000 nonwinners, we find positive and persistent effects on women’s labor market participation. We also find larger effects for youth workers. Results are driven by high-demand courses and non-health-related courses. Investigating outcomes beyond the labor supply, we do not find an impact on entrepreneurship or university admission.

ABSTRACT. Most recently, business-people have intensively engaged in elections around the world. Besides their claims tied to pro-business policies, little is known about the impact of their elections on the business environment. This paper aims to identify the causal effect of electing candidates with business experience on firm entry/exit, job creation and wages in Brazilian municipalities. Close electoral races were exploited using a regression discontinuity design. Results suggest considerable heterogeneity on the effect, particular with regard to firms size and sectors, and also to municipalities population. Regarding the workforce characteristics, there is an increase of 5-7% in wages, driven by the industry and services sectors, and a small improvement in the educational level of workers in services and agriculture.

ABSTRACT. Governments have allowed "tax deferral" in the worldwide taxation system in the belief that it does not affect the expected revenue coming from multinational firms. Although our dynamic structural model shows that the effective tax rate of the worldwide system is lower than tax rates in the territorial system, we observe that tax deferral is a bad policy instrument because it does not encourage reinvestment in the home country of multinational firms (reducing government revenues). It also contributes to the headquarters of multinational firms relocating to where a subsidiary was previously located (M&A transactions), and the subsidiary relocating to where the headquarters used to be.

ABSTRACT. We aim to assess whether having entrepreneurial peers is positively associated with the likelihood of an individual starting a new business. Further, the second goal of this paper is evaluate whether gender homophily is a driver of peer effect. We applied the Probit Regression model in two Stages (IV Probit) in a unique dataset of 10889 students from a Brazilian Institute of Technology in northeastern Brazil. The dataset covers the period from 2001 to 2010. The dependent variable on this modeling is a dummy representing if the individual has become an entrepreneur within five years after graduating at the institution. Our variable of interest was the fraction of entrepreneurs in the classroom "j," excepting the focal individual. We used variables related to the student personal, social, and economic factors as controls. Our results indicate that having entrepreneurial classmates positively affects the likelihood of an individual to start a new venture. Furthermore, we found that there is a significant effect between males and same-sex peers. We found no evidence that women are influenced by peers in the propensity to undertake, regardless of the gender of the peers. Our findings contribute to the literature on entrepreneurship and bring new evidence on peer effect entrepreneurship and the role of gender in the transmission of entrepreneurial behavior among classmates. Also, our paper sheds light on peer effect and entrepreneurship in one of the largest emerging markets, where 52 million people are involved in some entrepreneurship activity (GEM, 2018).

ABSTRACT. This paper employs a stakeholder-oriented approach to study how complementors’ capabilities may influence platform sales performance and diffusion. Focusing on service providers (SPs) as key platform complementors, the theoretical discussion is guided by a simple Stackelberg game model with endogenous entry. The derived hypotheses are tested using data from a randomized field experiment with poor entrepreneurs who are SPs in a fintech’s platform. The intervention exogenously changed SPs’ platform service provision capabilities by providing platform-focused training to a random sample of SPs. I show that SPs’ platform service provision capabilities improve platform sales performance for SPs with low pre-existing platform sales experience. I also show that these capabilities can induce platform diffusion if SPs with high pre-existing platform sales experience are trained. Yet my results indicate that the net benefits accruing to SPs may be negative, thereby suggesting that platform businesses may generate unintended negative social impact.