This paper investigates the impact of conditional cash transfer (CCT) programs on crime. Making use of a unique dataset combining detailed school characteristics with geo-referenced crime information from the city of São Paulo, Brazil, we estimate the contemporaneous effect of the Bolsa Família program on crime. We address the endogeneity of CCT coverage by exploiting the 2008 expansion of the program to adolescents aged 16 and 17. We construct an instrument that combines the timing of expansion with the initial demographic composition of schools to identify plausibly exogenous variation in the number of children covered by Bolsa Família. We find a robust and significant negative impact of Bolsa Família on crime. Incapacitation from time spent in school does not seem to be an important driving force behind the results.

This paper studies the asymptotic behavior of a Gaussian linear instrumental variables model in which the number of instruments diverges with the sample size. Asymptotic efficiency bounds are obtained for rotation invariant inference procedures and are shown to be attainable by procedures based on the limited information maximum likelihood estimator. The bounds are obtained by characterizing the limiting experiment associated with the model induced by the rotation invariance restriction.

We study the medium and long term impacts of Skills for Effective Entrepreneurship Development (SEED), a 3-week entrepreneurship training program for secondary school students in Uganda. The mini-MBA, modeled after business school curricula, was implemented as a randomized field experiment with a nationally representative sample of 4,402 youth. After four years, the training improved both hard and soft skills. SEED graduates became more effective negotiators and communicators and exhibited improved self-efficacy, stability, plasticity, and stress management. In the medium run, treated youth were more likely to start enterprises and more successful in ensuring their survival, thereby gaining greater entrepreneurial experience. Their ventures were also of higher quality: more likely to be formal, have employees, be in collaboration with other entrepreneurs, and use effective business management practices. After nine years, business ownership converged between treatment and control groups as control ownership rates doubled. However, SEED graduates maintained their edge in terms of business quality and operated firms with 20% higher revenues and 16% higher profits, without corresponding increases in capital or labor inputs, consistent with higher total factor productivity. SEED generated high returns on investment: the present discounted values of SEED-induced business and total earnings equal 20 and 27 times program costs, respectively.

Despite the promise of FinTech lending to expand credit access to populations without a formal credit history, FinTech lenders primarily lend to applicants with a formal credit history and rely on conventional credit bureau scores as an input to their algorithms. Using data from a large FinTech lender in Mexico, we show that alternative data from digital transactions through a delivery app are effective at predicting creditworthiness for borrowers with no credit history. Using account-by-month level data on revenues and costs, a machine learning model predicting profits generates similar profits as a model predicting default.

Advances in big data and non-traditional data have grounded many of the innovations in fintech in the past two decades. In this paper, we survey empirical evidence about the benefits of these innovations, as well as the constraints to scaling and the unintended consequences. The benefits of data advances for fintech are numerous. First, they can expand financial access to populations who have been historically excluded due to lack of credit history. Second, large datasets can help financial companies better analyze and mitigate risks, such as through enhanced fraud detection. Third, data can reduce the costs of administration by automating activities like loan approval, insurance payouts, and identity verification. Fourth, data allows for greater personalization, increasing the appeal of financial products to consumers. While these technologies hold great promise, to achieve broad scale they require large sources of data. There are also unintended consequences to consider, including privacy concerns, potential overindebtedness, discrimination baked into algorithms, and consumers gaming algorithmic systems. Further regulation and research can help address many of these concerns.

In this paper, we explore how technologies from new players in fintech differ from traditional financial services in the problems they solve, requirements for success at-scale, and unintended consequences. We focus on applications for consumers in developing countries, while also drawing on insights from developed economies. New players aim to increase access to financial services for those outside of the traditional banking systems, build new technologies and infrastructure to lower costs, and create new markets to solve problems not addressed with traditional financial services. To have any real impact on consumers, these fintech technologies must be brought to scale — a challenge given the fragmented nature of the market and the difficulty of establishing trust, regulatory oversight, and necessary infrastructure. While these technologies have significant potential benefit, increased financial access poses a challenge when bringing in less technically or financially sophisticated consumers. The paper concludes with a discussion of broader impacts on markets, prices, intra-household dynamics, mobility, and the environment.

We study the medium-term impacts of the Skills for Effective Entrepreneurship Development (SEED) program, an innovative in-residence 3-week mini-MBA program for high school students modeled after western business school curricula and adapted to the Ugandan context. The program featured two separate treatments: the hard-skills MBA features a mix of approximately 75% hard skills and 25% soft skills; the soft skills curriculum has the reverse mix. Using data on 4400 youth from a nationally representative sample in a 3-arm field experiment in Uganda, the 3.5 year follow-up demonstrated that training was effective in improving both hard and soft skills, but only soft skills were directly linked to improvements in self-efficacy, persuasion, and negotiation. The skill upgrade was rewarded in substantially higher earnings — 32.1% and 29.8% increases for those who attended hard- and soft-training respectively — most of which was generated through self-employment. Both SEED curricula were very cost-effective; two months of the extra earnings caused by the training alone would exceed the cost of the program.

Empowering adolescent girls through education has become a priority among numerous stakeholders. However, recent evidence suggests that education alone may not be sufficient if women remain in a low-empowerment equilibrium and face internal constraints as they relate to aspirations, self-efficacy, leadership, and other life (soft) skills. We study the long-term impacts of a school-based upper-secondary intervention, the Educate! Experience, designed to enhance adolescents' leadership and social entrepreneurship skills in Uganda. The program was implemented as a cluster randomized controlled trial (RCT) in 48 schools. Four years post-intervention, we document lasting impacts on a wide array of leadership and soft skills. Overall, Educate! graduates developed skills that are traditionally associated with greater focus on long-term goals; they reported being more in control of aspects of their lives (self-efficacy and grit) and more empowered to implement actions towards their plans. Young women in the treatment group are also more likely to complete secondary education, delay family formation, enroll in tertiary education, and pursue STEM and Business majors relative to their counterparts in the control group. The program yielded socially desirable and gender relevant spillovers, including expansions in women's agency. Both male and female Educate! graduates embraced more progressive views concerning women's standing in the society and women's ability to exercise their agency to engage in the labor market and refuse sex. The incidence of intimate partner violence (IPV) also improved among Educate! graduates, as did their attitudes toward IPV social acceptability.

Between 2007 and 2014, more than 164,000 civilians were victims of homicide in Mexico. During the same period, roughly 103,000 civilians died in the Afghanistan and Iraq wars combined. These spikes in violence occurred during an intensification of large-scale efforts by the Mexican government to cripple organized crime by targeting its leadership, calling into question the efficacy of the so-called kingpin strategy. Exploiting information on the presence of cartels in municipalities and the locations of neutralizations of cartel operatives since 1995, we document the geographic and temporal spillovers of violence that result from power vacuums following the capture or killing of key drug cartel operatives. With access to high-frequency data on labor market outcomes, we then study the short and medium term impacts of the kingpin strategy on labor markets in the context of dynamic panel data models. The evidence suggests that neutralizations have large and persistent effects on the level of violence, as well as on earnings, hours worked, and quality of attachment to the labor force. Violence acts as a shock to the demand for low-skilled labor in the formal sector. Low-skill workers adjust by substituting into lower quality jobs; e.g., by joining the informal sector and working part-time.

In the context of the instrumental variables model we focus on tests that are similar conditional on the first stage F statistic. We argue that in some economic applications, it is desirable to conduct inference conditionally on the first stage F statistic. Assuming homoskedastic Gaussian errors and known covariance matrix, we derive the power envelopes for conditionally similar and conditionally unbiased tests for the coefficient on the endogenous regressor. Making use of Staiger and Stock (1997) asymptotics, we show that the efficiency bounds derived under the assumptions of Gaussian errors and known covariance matrix can also be attained in large samples when the reduced form covariance matrix has to be estimated and the errors are nonnormal. A Monte Carlo study is conducted to assess the performance of the conditional testing procedures.

This paper proposes two new identification schemes for estimating the parameter of intertemporal allocation, exploiting the equilibrium condition delivered by the rational expectations consumption model. The two identification schemes constitute an alternative to the existing methods of estimating Euler equations in the presence of measurement error. The small sample performance of the proposed identification schemes as well as that of existing approaches are evaluated by way of a MonteCarlo experiment.

As a means to reduce delays in public works implementation, a number of Brazilian states have recently reformed their procurement rules allowing contractor price proposals to be assessed before the technical evaluation of submitted bids is undertaken (inversao das fases). This paper adopts a difference-in-differences methodology to compare the procurement performance of São Paulo state (a reformer state) and Minas Gerais (a non-reformer state) along three efficiency dimensions: (i) procurement process duration; (ii) likelihood of complaint resolution litigation; and (iii) prices paid to contractors. The analysis finds that the reform is associated with a 24 day reduction in the duration of procurement processes for large projects and a 7 percentage point drop in the likelihood of court challenges irrespective of project size. The paper finds no evidence of an effect of the procurement reform on prices paid.

This paper proposes some basic elements of a conceptual framework to help organize the thinking about policies that can strengthen the knowledge mission of the World Bank. It argues that the Bank occupies a unique and prominent subset of the social and economic development "knowledge space" that ranges from abstract basic research to codified knowledge solutions. The range of products coupled with an increasing emphasis on just-in-time advisory services dictates the need for not more generalists, but rather an increased range and depth of very specific and high quality human capital. However, this increased specialization in turn creates the need for "hinge" actors who can communicate and operate well across different knowledge communities — academics, policy makers, practitioners, etc.