Publications
Objective This study aims to develop high-performing Machine Learning and Deep Learning models in predicting hospital length of stay (LOS) while enhancing interpretability. We compare performance and interpretability of models trained only on structured tabular data with models trained only on unstructured clinical text data, and on mixed data. Methods The structured data was used to train fourteen classical Machine Learning models including advanced ensemble trees, neural networks and k-nearest neighbors. The unstructured data was used to fine-tune a pre-trained Bio Clinical BERT Transformer Deep Learning model. The structured and unstructured data were then merged into a tabular dataset after vectorization of the clinical text and a dimensional reduction through Latent Dirichlet Allocation. The study used the free and publicly available Medical Information Mart for Intensive Care (MIMIC) III database, on the open AutoML Library AutoGluon. Performance is evaluated with respect to two types of random classifiers, used as baselines. Results The best model from structured data demonstrates high performance (ROC AUC = 0.944, PRC AUC = 0.655) with limited interpretability, where the most important predictors of prolonged LOS are the level of blood urea nitrogen and of platelets. The Transformer model displays a good but lower performance (ROC AUC = 0.842, PRC AUC = 0.375) with a richer array of interpretability by providing more specific in-hospital factors including procedures, conditions, and medical history. The best model trained on mixed data satisfies both a high level of performance (ROC AUC = 0.963, PRC AUC = 0.746) and a much larger scope in interpretability including pathologies of the intestine, the colon, and the blood; infectious diseases, respiratory problems, procedures involving sedation and intubation, and vascular surgery. Conclusions Our results outperform most of the state-of-the-art models in LOS prediction both in terms of performance and of interpretability. Data fusion between structured and unstructured text data may significantly improve performance and interpretability.
The relationship between entry-regulation, corruption, and entrepreneurship is controversial in the literature. Using a broad cross-country dataset to deepen the investigation, this paper distinguishes opportunity and necessity-motivated entrepreneurship in different development contexts. Corruption might grease the wheels of ineffective administrative machinery in developing countries with heavy entry-regulation. Yet, the marginal effect of corruption will generally be non-significant in other developing countries and in developed countries. Moreover, our results suggest that corruption deters opportunity-motivated entrepreneurship—the type of entrepreneurship that may contribute the most to productivity, economic growth, and development—in developed countries.
We rationalize the observed short-run differences in corporate and long-term government bond yields in an financial-accelerator model with frictions that restrict changes in portfolio shares. We estimate the model on quarterly data for the Euro Area from 1999 to 2019, and show that the portfolio friction parameter is positive and significant. Portfolio frictions not only generate a time-varying wedge between the two returns that fits the data, but also raise the volatility of return differentials, and the precautionary motive of savers. As a result, the macroeconomic effects of uncertainty shocks are amplified by portfolio frictions.
Many people obtain job information from friends and acquaintances. However, one factor influencing labor-market outcomes that is ignored in the literature is the presence of overlapping friendship circles in social networks. We find that overlapping friendship networks produce correlated information flows, resulting in an increased probability of two events: either receiving redundant job offers or receiving no job offers at all. Consequently, people with common contact networks exhibit worse employment prospects even if they have the same number of information providers and compete with the same number of people for vacancies. In quantitative terms, the impact of overlapping friendship circles rivals that of the number of direct contacts and contacts' contacts. This implies that the results in Calvo-Armengol (2004) only apply for networks where people's friends are neither connected nor have common contacts. Because overlapping friendship circles are a crucial aspect of strong relationships, our findings uncover an alternative mechanism behind "The Strength of Weak Ties"(Granovetter, 1973): their ability to maintain independence in job information flows. We further show that people with common job contacts earn lower incomes on average. However, conditional on being employed, their expected wage is higher because they can take advantage of the multiple job offers received by selecting the one with the highest pay.
We study a class of location games where players want to attract as many resources as possible and pay a cost when deviating from an exogenous reference location. This class of games includes political competitions between policy-interested parties and firms' costly horizontal differentiation. We find that the introduction of reference locations simplifies the set of pure-strategy equilibrium to a unique candidate which has a strong property: at most four players, the two most-left and two most-right, deviate from their reference locations. We provide necessary and sufficient conditions for the candidate to be an equilibrium. We illustrate our results in particular cases including the duopoly competition where we moderate the principle of minimal differentiation.
A common practice in many auctions is to offer bidders an opportunity to improve their bids, known as a best and final offer stage. This improved bid can depend on new information either about the asset or about the competitors. This paper examines the effects of new information regarding competitors, seeking to determine what information the auctioneer should provide assuming the set of allowable bids is discrete. The rational strategy profile that maximizes the revenue of the auctioneer is the one where each bidder makes the highest possible bid that is lower than his valuation of the item. This strategy profile is an equilibrium for a large enough number of bidders, regardless of the information released. We compare the number of bidders needed for this profile to be an equilibrium under different information structures. We find that it becomes an equilibrium with fewer bidders when less additional information is made available to the bidders regarding the competition. It follows that when the number of bidders is a priori unknown, there are some advantages to the auctioneer not revealing information and conducting a one-stage auction instead.
The article examines stock index price responses in Brazil, Chile, and Mexico to those in the US, Spain, and four European countries during three sub-periods surrounding the neoliberal reforms of the 1990s: 1988 to 1994, 1995 to 1998, and 1999 to 2004, using VAR modeling. It finds that equity markets became more interconnected as countries opened to international trade and capital flows, and that there was an increasing impact of Spain on Latin American equity markets. Stronger economic linkages (more trade and foreign direct investment) between Spain and these countries, especially in Brazil, seem to explain increased equity markets’ interconnectedness.
Household surveys do not capture incomes at the top of the distribution well. This yields biased inequality measures. We compare the performance of the reweighting and replacing methods to address top incomes underreporting in surveys using information from tax records. The biggest challenge is that the true threshold above which underreporting occurs is unknown. Relying on simulation, we construct a hypothetical true distribution and a “distorted” distribution that mimics an underreporting pattern found in a novel linked data for Uruguay. Our simulations show that if one chooses a threshold that is not close to the true one, corrected inequality measures may be significantly biased. Interestingly, the bias using the replacing method is less sensitive to the choice of threshold. We approach the threshold selection challenge in practice using the Uruguayan linked data. Our findings are analogous to the simulation exercise. These results, however, should not be considered a general assessment of the two methods.
The increase in employment polarization observed in several high-income economies has coincided with a reduction in inter-generational mobility. This paper argues that the disappearance of middling jobs can drive changes in mobility, notably by removing a stepping stone towards high-paying occupations for those from less well-off family backgrounds. Using data from two British cohorts who entered the labour market at two points in time with very different degrees of employment polarization, we examine how parental income affects both entry occupations and occupational upgrading over careers. We find that transitions across occupations are key to mobility and that the impact of parental income has grown over time. At regional level, using a shift-share IV-strategy, we show that the impact of parental income has increased the most in regions experiencing the greatest increase in polarisation. This indicates that the disappearance of middling jobs played a role in the observed decline in mobility.
This paper studies differences across genders in the re-contesting decisions of politicians following electoral wins or defeats. Using close races in mixed-gender French local elections, we show that women are less likely to persist in competition when they lose compared to male runners-up, but are equally or more prone than male winners to re-contest when they win. Differences in observable characteristics or in the expected electoral returns of running again cannot fully account for these gender gaps in persistence. In contrast, evidence suggests that results are driven by behavioural explanations such as cross-gender differences in candidates’ attitudes toward competition, or by political parties behaving differently toward female and male candidates for a given electoral outcome. Additionally, we provide evidence that a woman’s victory encourages former female challengers to re-contest but does not trigger the entry of new female candidates.
