Eventos

Seminario CEDE - Daniel Jaramillo

Do released offenders influence the criminal behavior of individuals in the neighborhoods they rejoin? Using a unique dataset on arrests, prison releases, and places of residence for the universe of men in Ecuador and exploiting a mass pardon in a difference-indifference design, I find evidence that released offenders contribute to increased criminal activity among their neighbors. On average, one additional release leads to an increase of 0.85 arrests, excluding the released offenders themselves. First-time offenders account for 42% of this effect, with the primary mechanism being the spread of criminal behavior through peer and family networks. These peer effects are larger for defendants who served longer portions of their sentences, suggesting that time spent in prison may intensify criminal behavior. Finally, I show that access to job training programs during incarceration can help mitigate these effects.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Jorge Bonilla

Short-term exposure to air pollution increases the risk of respiratory diseases. This risk is higher during critical air quality events when episodes last for several days and are harmful to the most vulnerable groups. A common strategy in Latin America is temporarily adjusting transport policies to reduce emissions and the impact on people's health. Medellín, Colombia's second-largest city, and the Aburrá Valley Metropolitan Area have been implementing driving restrictions since 2017, which, unlike other cities, not only banned cars but also restricted the circulation of motorcycles and trucks. Restrictions that quickly became stricter for old vehicles. Using a two-way-fixed effects model, we empirically evaluate the effectiveness of the implemented plan between 2017 and 2019 on pollutant concentrations and examine the impact on vehicle trips, noise, and possible changes in regular on-road tests of vehicle environmental performance. Our results show that the implemented plan has reduced air pollution by 2%-24% for several pollutants. The red alerts, introduced with the strictest bans, showed the largest reduction. Particulate matter showed the most significant improvement in air quality. In general, reductions in air pollutant concentrations tend to be greater when restrictions apply to trucks or older vehicles and when bans cover more license plates. Vehicle trips and noise also declined, and there is some evidence of an increase in on-road vehicle testing. These findings suggest that introducing differentiated restrictions focalizing the most stringent bans on the most polluting vehicles is an effective policy to reduce air pollution during short-term episodes of poor air quality.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Hoover Quitian

El artículo explora el efecto del consumo de tecnologías sanitarias sobre la probabilidad de morir, a diferentes niveles de deterioro del estado de salud de los individuos. Se emplearon registros de la base de datos de suficiencia, de la Unidad de Pago por Capitación, que cuenta con información de atenciones de aproximadamente 28 millones de colombianos, entre los años 2013 y 2019. Como estrategia de identificación, se utilizó el cuasi experimento generado por el cambio de los individuos de Entidad Aseguradora, realizado por el Sistema de Salud cuando decreta el cierre de la entidad por insolvencia. Los resultados sugieren que el efecto del consumo de tecnologías se reduce cuando los individuos no tienen afectaciones muy graves sobre su estado de salud y cuando su riesgo de morir es considerable.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Víctor González-Jiménez

This paper shows that well-established biases in decision making under uncertainty generate poverty traps. A theoretical framework is developed to show that: i) probability weighting and ambiguity attitudes can lead individuals to systematically undervalue profitable investments, and ii) poverty exacerbates the consequences of these investment errors. The model predicts that poverty persists because these biases disproportionately discourage investment among poor individuals, while rich individuals continue to invest regardless. These theoretical predictions are validated using data from two experiments conducted on representative samples of American households.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Danilo Aristizabal

There is a close relationship between poverty and social and labor exclusion. I evaluated two ALMPs that targeted people who have difficulties in finding a formal job in a developing country. First, I analyze the effect of a training component of a labor intermediation policy (LIP) called Boost to Employment (BE) on the probability of finding a formal job for vulnerable job seekers in a developing country. I exploit the variation in the leniency among labor counselors as an instrument to approximate the probability that job seekers receive the training component. I also test which courses are the ones that helped job seekers to find formal jobs. I find that those workers who received the training component increased the probability of working in the formal sector by 21 percentage points one year after the implementation of the program compared to those workers who did not. The formal job benefits course seems to be driving my results. Second, I study the effect on labor market outcomes of a payroll tax cut for new hires of young workers under the age of 28 in an economy with a high binding minimum wage. I use exposure to wage rigidities to identify the effect. I measure an individual's exposure to wage rigidities as the gap between the median salary, in the city in which the individual lives, and the minimum wage set at the national level. I use a difference-in-difference model. I find that the effect of a payroll tax cut is asymmetric for youth who face labor markets with a binding minimum wage and those who do not.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Nicole Fleskes

This paper studies how dealers provide liquidity in over-the-counter (OTC) markets during an acute crisis when they can choose their capacity: how fast they can execute orders and their inventories. I develop a search model of an OTC market in which trade between homogeneous dealers is centralized, but trade with investors is subject to search frictions proportional to dealer order execution rates. Dealers have positive bargaining power and charge intermediation fees to investors: dealers face a trade-off when choosing their capacity between their income from intermediation fees and capital gains, and the flow costs of their order execution rate and inventories. The model features a one time aggregate adverse liquidity shock to investors' private valuations. The more severe the liquidity shock, the lower the order execution rate dealers choose and the longer they wait to begin taking on inventories. Compared to a constrained planner, the existence of dealer bargaining power results in a double-sided hold-up problem that can be resolved in the steady state in a knife-edge case, but not out of the steady state following an aggregate liquidity shock. A calibration to the U.S. secondary corporate bond markets suggests that inefficiencies in dealer capacity results in larger welfare losses following a shock compared to the steady state.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Brigitte Castañeda

Do the macroeconomic effect of a carbon tax differ between countries, according to the primary energy source? I answer this question with a theoretical model of directed technical change and test empirically the main results. I find four main results: (i) In the absence of subsidies, carbon taxes have a negative effect on economic growth, (ii) this negative effect is a decreasing function of the proportion of clean energy sources. (iii) subsidies for clean inputs have a positive effect on economic growth, and (iv) the magnitude of this effect grows with the proportion of clean energy sources. The empirical results are consistent with the predictions of the theoretical model and suggest that policymakers could consider this relationship between the energy mix and the economic effect of a carbon tax when creating environmental regulations.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Juliana Jaramillo Echeverri

Can radio campaigns affect fertility? This paper examines the impact of a national radio campaign promoting family planning clinics in late 1960s Colombia on the country's rapid fertility decline. The campaign, initiated by Profamilia in 1969, provided information about the location of clinics without providing detailed contraceptive information. Using data from the full count 1973 census and information on clinic locations and radio programs, the study leverages exogenous variation in radio signal strength to estimate the campaign's effect on fertility. I follow a difference-in-differences strategy to compare fertility before and after the start of the radio campaign at the individual level. The findings indicate that the radio campaign reduced between 4% to 14% the probability of having a child one year after the campaign. The effects are stronger for women living close to a Profamilia clinic. However, by 1973 motherhood was still almost universal. This research contributes to the understanding of the effects of family planning programs and media exposure on fertility decline, highlighting the role of media in overcoming geographical barriers and driving social change.

12:30 pm

Universidad de los Andes

Activo

Seminario CEDE - Paul Soto

This paper examines the link between industrial production and the sentiment expressed in natural language survey responses from U.S. manufacturing firms. We compare several natural language processing (NLP) techniques for classifying sentiment, ranging from dictionary-based methods to modern deep learning methods. Using a manually labeled sample as ground truth, we find that deep learning models--partially trained on a human-labeled sample of our data--outperform other methods for classifying the sentiment of survey responses. Further, we capitalize on the panel nature of the data to train models which predict firm-level production using lagged firm-level text. This allows us to leverage a large sample of "naturally occurring" labels with no manual input. We then assess the extent to which each sentiment measure, aggregated to monthly time series, can serve as a useful statistical indicator and forecast industrial production. Our results suggest that the text responses provide information beyond the available numerical data from the same survey and improve out-of-sample forecasting; deep learning methods and the use of naturally occurring labels seem especially useful for forecasting. We also explore what drives the predictions made by the deep learning models, and find that a relatively small number of words--associated with very positive/negative sentiment--account for much of the variation in the aggregate sentiment index.

12:30 pm

Universidad de los Andes