Publications
We show that least squares cross-validation methods share a common structure which has an explicit asymptotic solution, when the chosen kernel is asymptotically separable in bandwidth and data. For density estimation with a multivariate Student t(ν) kernel, the cross-validation criterion becomes asymptotically equivalent to a polynomial of only three terms. Our bandwidth formulae are simple and noniterative thus leading to very fast computations, their integrated squared-error dominates traditional cross-validation implementations, they alleviate the notorious sample variability of cross-validation, and overcome its breakdown in the case of repeated observations. We illustrate our method with univariate and bivariate applications, of density estimation and nonparametric regressions, to a large dataset of Michigan State University academic wages and experience.
Some complex models are frequently employed to describe physical and mechanical phenomena. In this setting, we have an input X\ X \ in a general space, and an output Y=f(X)\ Y=f(X) \ where f\ f \ is a very complicated function, whose computational cost for every new input is very high, and may be also very expensive. We are given two sets of observations of X\ X \, S1\ S_1 \ and S2\ S_2 \ of different sizes such that only fS1\ f\left(S_1\right) \ is available. We tackle the problem of selecting a subset S3⊂S2\ S_3\subset S_2 \ of smaller size on which to run the complex model f\ f \, and such that the empirical distribution of fS3\ f\left(S_3\right) \ is close to that of fS1\ f\left(S_1\right) \. We suggest three algorithms to solve this problem and show their efficiency using simulated datasets and the Airfoil self-noise data set.
The impact of US allocation of family planning aid on other donors is studied in order to gain new insights into donor interactions. Within this context, the dominant player in the sector is the United States, whose policies on family planning undergo changes influenced by domestic debates surrounding abortion. By utilizing the Mexico City Policy and considering exposure to this particular policy as an instrumental factor, it has been observed that other donors do not immediately react to policy changes made by the United States, either contemporaneously or within one year. However, a noticeable shift occurs after a two-year period, indicating that these donors eventually align their allocation strategies with those of the United States. Further analysis of this phenomenon reveals varying patterns among different types of donors. While smaller donors exhibit a clear intention to compensate for US policy changes, larger donors display a mix of competitive tendencies and herding behavior, thereby reinforcing the impact of the Mexico City Policy after the two-year time frame.
Earnings inequality in Germany has increased dramatically. Measuring inequality locally at the level of cities annually since 1985, we find that behind this development is the rapidly worsening inequality in the largest cities, driven by increasing earnings polarisation. In the cross-section, local earnings inequality rises substantially in city size, and this city-size inequality penalty has increased steadily since 1985, reaching an elasticity of .2 in 2010. Inequality decompositions reveal that overall earnings inequality is almost fully explained by the within-locations component, which in turn is driven by the largest cities. The worsening inequality in the largest cities is amplified by their greater population weight. Examining the local earnings distributions directly reveals that this is due to increasing earnings polarisation that is strongest in the largest places. Both upper and lower distributional tails become heavier over time, and are the heaviest in the largest cities. We establish these results using a large and spatially representative administrative data set, and address the top-coding problem in these data using a parametric distribution approach that outperforms standard imputations.
BACKGROUND: Sexual and gender diverse people face intersecting factors affecting their well-being and livelihood. These include homophobic reactions, stigma or discrimination at the workplace and in healthcare facilities, economic vulnerability, lack of social support, and HIV. This study aimed to examine the association between such factors and symptoms of anxiety and depression among sexual and gender diverse people.
METHODS: This study is based on a sample of 108,389 gay, bisexual, queer and questioning men, and transfeminine people from 161 countries collected through a cross-sectional internet survey. We developed a multinomial logistic regression for each group to study the associations of the above factors at different severity scores for anxiety and depression symptoms.
RESULTS: Almost a third (30.3%) of the participants reported experiencing moderate to severe symptoms of anxiety and depression. Higher severity scores were found for transfeminine people (39%), and queer or questioning people (34.8%). Severe symptoms of anxiety and depression were strongly correlated with economic hardship for all groups. Compared to those who are HIV-negative, those living with HIV were more likely to report severe symptoms of anxiety and depression, and the highest score was among those who do not know their HIV status. Transfeminine people were the most exposed group, with more than 80% higher risk for those living with HIV suffering from anxiety and depression. Finally, homophobic reactions were strongly associated with anxiety and depression. The relative risk of severe anxiety and depression was 3.47 times higher for transfeminine people facing transphobic reactions than those with no symptoms. Moreover, anxiety and depression correlate with stigma or discrimination in the workplace and healthcare facilities.
CONCLUSIONS: The strong association between the severity of anxiety and depression, and socioeconomic inequality and HIV status highlights the need for concrete actions to meet the United Nations' pledge to end inequalities faced by communities and people affected by HIV. Moreover, the association between stigma or discrimination and anxiety and depression among sexual and gender diverse people is alarming. There is a need for bold structural public health interventions, particularly for transfeminine, queer and questioning people who represent three communities under the radar of national HIV programmes.
In the follow-up to the 1926 political and monetary crisis in France, a new government led by Raymond Poincaré attempted to restore monetary stability by restructuring public debt. A sinking fund was missioned to withdraw short-term public bills from money markets. This policy disorganized the largest Parisian banks of the time, as they relied on these bills to manage their liquidity. Without developed domestic money markets, no other asset could absorb the excess liquidity freed by the withdrawal of these bills, and these leading banks faced a low-rate environment. In search of yield, they expanded their activities abroad a few months before the 1929 crash. These findings renew our understanding of the expansion of France's banking sector in the 1920s. In addition, they shed new light on the role of public debt in financial stability in an open economy.
Worrisome topics, such as climate change, economic crises, or pandemics including Covid-19, are increasingly present and pervasive due to digital media and social networks. Do worries triggered by such topics affect the cognitive capacities of young adults? In an online experiment during the Covid-19 pandemic (N=1503), we test how the cognitive performance of university students responds when exposed to topics discussing (i) current adverse mental health consequences of social restrictions or (ii) future labor market hardships linked to the economic contraction. Moreover, we study how such a response is affected by a performance goal. We find that the labor market topic increases cognitive performance when it is motivated by a goal, consistent with a ‘tunneling effect’ of scarcity or a positive stress effect. However, we show that the positive reaction is mainly concentrated among students with larger financial and social resources, pointing to an inequality-widening mechanism. Conversely, we find limited support for a negative stress effect or a ‘cognitive load effect’ of scarcity, as the mental health topic has a negative but insignificant average effect on cognitive performance. Yet, there is a negative response among psychologically vulnerable individuals when the payout is not conditioned on reaching a goal.
Despite their high predictive performance, random forest and gradient boosting are often considered as black boxes which has raised concerns from practitioners and regulators. As an alternative, we suggest using partial linear models that are inherently interpretable. Specifically, we propose to combine parametric and non-parametric functions to accurately capture linearities and non-linearities prevailing between dependent and explanatory variables, and a variable selection procedure to control for overfitting issues. Estimation relies on a two-step procedure building upon the double residual method. We illustrate the predictive performance and interpretability of our approach on a regression problem.
What is the role of income polarization for explaining differentials in public funding of education? To answer this question, we provide a new theoretical modelling for the income distribution that can directly monitor income polarization. It leads to a new income polarization index where the middle class is represented by an interval. We implement this distribution in a political economy model with endogenous fertility and public/private educational choices. We show that when households vote on public schooling expenditures, polarization matters for explaining disparities in public education funding across communities. Using micro-data covering two groups of school districts, we find that both income polarization and income inequality affect public school funding with opposite signs whether there exist a Tax Limitation Expenditure (TLE) or not.
