2015-2016

Title,Abstract

The informational content of aggregate discrete choice models

Structural empirical IO typically relies on a mixed multinomial logit model. Relying on small-sigma expansions around the homogeneous model, we derive the statistics that identify the main parameters. This allows us to propose pre-estimation diagnostics of the informational content of the model. In addition, it naturally yields approximate estimators of "macro-BLP" models that only use simple 2SLS. Finally, arbitrarily precise estimators can be obtained by constrained 2SLS.

Commodity price shocks and violence in Colombia: A story told by pipelines

The main line of research in literature on conflicts was concentrated on finding the relationship between natural resources and conflict. In this paper, I argue that natural resources have limited value if they cannot be delivered to the consumer. The most common and effective means of delivering oil or gas from the oilfields are the pipelines, and anecdotal evidence suggests that they are targeted a lot by different armed groups worldwide. In particular, this is true for Colombia where the leftist guerrillas are constantly reported to target the energy infrastructure. I study how pipelines shape the civil conflict in Colombia. Presumably, the placement of the actual pipeline system is endogenous to the conflict, and to overcome this problem I build a hypothetical pipeline system. The distance from a municipality to this hypothetical pipeline system is used as an instrument for the actual distance, and I show that in response to oil prices shocks, the municipalities located closer to the pipelines experience a differentially larger activity of the FARC rebels, while the distance is not important in explaining the behaviour of the ELN or the paramilitaries groups.

Keywords: Civil war, Pipelines, Economic shocks, Natural resources

Automobile Emissions Test Manipulation: Implications for Consumer Welfare and the Environment

Mounting evidence suggests that automakers have systematically manipulated fuel economy tests in Europe in order to comply with strict carbon regulations. In this project, we use highly detailed data to quantify the degree of manipulation and the consequences of this manipulation for environmental policy and consumer welfare.

Identifying the Effects of Scientific Information and Recommendations on Physicians Prescribing Behavior

We investigate how prescription behavior of physicians react to scientific information released by public authorities. Taking the example of antidepressant drugs, we use French panel data on exhaustive prescriptions of a representative sample of 386 general practitioners to more than 170,000 depressed patients, between 2000 to 2008. Changing scientific evidence on the efficacy and side effects of antidepressants during that period resulted in new official warnings and recommendations. Examples are the new results on the increase of suicidal thinking in children reported in 2004 for selective serotonin reuptake inhibitors (SSRI). After the warnings, physicians must update their risk perception on different drug treatments for the patients and physicians may react differently to these warnings. We use the official warning to identify how they affected physicians' drug choices on the first visit the patient is diagnosed by depression. We investigate identification conditions depending on the unobserved heterogeneity of patients and physicians. We find that physicians respond to the new information heterogeneously, providing evidence that heterogeneity of prescribing behavior is not simply due to heterogeneity of preferences towards different side effects but heterogeneity of knowledge or information acquisition. We find that antidepressant prescriptions decrease after 2004 for kids and adolescents. Even though the warning is only for kids and adolescents physicians change their prescription behavior for other age groups too

The Benefits of Alternatives to Conventional College: Labor Market Returns to Proprietary Schooling

This paper provides novel evidence on the labor-market returns to proprietary (also called for profit) postsecondary school attendance. Specifically, we link administrative records on proprietary school attendance with quarterly earnings data for nearly 70,000 students. Because average age at school entry is 30 years of age, and because we have earnings data for five or more years prior to attendance, we estimate a person fixed-effects model to control for time invariant differences across individuals. By five years after entry, quarterly earnings returns are around 26 percent for men and 21-22 percent for women. Average returns are quite similar for associate’s degree programs and certificate programs, but vary substantially by field of study. Differences in return by gender are completely explained by differences in field of study.

Should Off-Label Drug Prescriptions Be Prevented? Empirical Evidence from France

We investigate physicians' prescription decisions over approved vs. off-label drugs in treatment of alcoholism, anxiety, bipolar, and depression. We use French panel data on exhaustive prescriptions of a representative sample of 386 general practitioners to more than 1.5 million patients between 2000 to 2008. The rate of off-label prescriptions is at least 18 % in each of these diseases. All of the physicians prescribe off-label drugs; however, there is some heterogeneity across physicians in their off-label prescriptions. Counterfactual simulations show that removing the off-label drugs from the choice set of physicians would lead to a substantial increase in the cost of prescription drugs, whereas it does not lead to an improvement in terms of treatment outcomes. The estimations taking into account the unobserved heterogeneity at patient level reveal that there is selection into treatment. Controlling for this selection, the preliminary results show that off-label drugs are not necessarily less effective than the approved alternatives.

Can Centralized Allocation Beat the Market? Evidence From 22,000 Land Consolidations

A central tenet of economic theory is that centralized allocation of goods or factors of production is less efficient than the market. We test this assumption by using land consolidation in France as a natural experiment. Between 1942 and 2008, half of the French agricultural usable area has changed hands as part of a major land consolidation program called "Remembrement." Land consolidation aimed at decreasing the morceling of land by creating new, bigger plots and by concentrating them around each farmer's house with the aim of triggering the adoption of new technologies, among them chiefly the tractor. Our tentative preliminary results are that centralized allocation has made no detectable changes to the evolution of farm size, and thus that centralized allocation has not improved on the land market, a surprising result considering the huge frictions characterizing this market. If confirmed, this result is extremely relevant since many developing countries consider implementing similar programs.

Searching for services: Ranking and waiting time

We analyze client's decision to contact home service professionals an online platform: plumbers, mechanics, builders, and so on, using private data. The nature of the transaction and the organization of the platform make this setting interesting and rich for analysis potential. Firstly, there is informational asymmetry between the client and the expert, who provides both the diagnosis and the treatment, and it often persists after the transaction. Secondly, the client receives offers which arrive in the order and at the time of submission by the professionals, relating our setting to the literature on ordered search and rankings. As we have both the search cost as a waiting time and the order of search as the ranking of offers, we are able to separately analyze these two aspects for the first time. The relevant literature has varied predictions on how clients consider the time and rank of the offers when deciding which professionals to contact, thus the question becomes empirical. Our preliminary results indicate that clients react negatively to the waiting time, but positively to lower ranking. As a next step we explore endogeneity and differential effects with respect to the search cost and the search horizon of the client.

TBA