Publications

Because they play an increasingly important role in determining access to credit, credit scoring models are under growing scrutiny from banking supervisors and internal model validators. These authorities need to monitor the model performance and identify its key drivers. To facilitate this, we introduce the explainable performance (XPER) methodology to decompose a performance metric (e.g., area under the curve (AUC), [Formula: see text]) into specific contributions associated with the various features of a forecasting model. XPER is theoretically grounded on Shapley values and is both model-agnostic and performance metric-agnostic. Furthermore, it can be implemented either at the model level or at the individual level. Using a novel data set of car loans, we decompose the AUC of a machine-learning model trained to forecast the default probability of loan applicants. We show that a small number of features can explain a surprisingly large part of the model performance. Notably, the features that contribute the most to the predictive performance of the model may not be the ones that contribute the most to individual forecasts (Shapley additive explanation). Finally, we show how XPER can be used to deal with heterogeneity issues and improve performance. This paper was accepted by Kay Giesecke, finance. Funding: The authors thank the Institut Universitaire de France, the Autorité de contrôle prudentiel et de résolution Chair in Regulation and Systemic Risk, the HEC-Deloitte Chair on Artificial Intelligence for Business Innovation, the Excellence Initiative of Aix-Marseille University [A*MIDEX], and the French National Research Agency [Grants AMSE ANR-17-EURE-0020, Ecodec ANR-11-LABX-0047, and MLEforRisk ANR-21-CE26-0007] for supporting our research. Supplemental Material: The online appendices and data files are available at https://doi.org/10.1287/mnsc.2023.02025 .

This paper provides benchmark experimental evidence on how individuals trade off financial performance against the risk of disclosing their own financial information when privacy risks are explicit, and outcomes are immediate. In a laboratory investment task with repeated decisions, participants choose between two risky options that differ in expected return and in an explicitly stated probability that choosing one option triggers a pseudonymous disclosure of a limited subset of their pre-elicited financial information to other participants. Participants respond systematically to both return spreads and leak probabilities, but choices are substantially more elastic to financial incentives than to changes in leak risk. In dynamic analyses, experiencing an own leak has modest and short-lived effects, whereas higher leakage among other participants is followed by a lower propensity to select the privacy-risky option. Standard measures of risk and loss aversion and most economic characteristics explain little of the heterogeneity in choices, while context-relevant privacy attitudes are associated with more privacy-protective behavior. Overall, in this transparent-probability benchmark environment, monetary incentives dominate leak risk on average, while social information about others' leaks meaningfully shapes behavior.

Environmental concerns are becoming more pressing as the climate emergency intensifies, posing a major challenge for many governments: increasing green investments to promote better adaptation and resilience to climate events, while maintaining fiscal discipline. This raises the question of whether governments that operate under fiscal rules tend to safeguard environmental spending in light of the climate emergency, or whether they are more inclined to scale it back to meet their fiscal targets, given that such investments require substantial public funding. Using data covering 31 advanced economies between 1995 and 2021, we find robust evidence that the strengthening of fiscal rules significantly reduces environmental spending, in particular debt rules and expenditure rules. Moreover, the adverse impact of fiscal rules on environmental expenditures is amplified during election periods, whereas it is mitigated in the presence of sound past fiscal conditions, the Kyoto Protocol, and stringent environmental policies. Further analysis reveals that although fiscal rules tend to reduce environmental spending, they are associated with greater efficiency in such spending.

This study examines how epidemic outbreaks influence foreign direct investment (FDI) inflows in developing countries, with a particular focus on whether healthcare systems can act as buffers during such shocks. Using a panel dataset of 98 countries from 2000 to 2022, the analysis combines two-way fixed effects (FE) models and the Local Projection Method (LPM). The analysis is structured in three parts: First, fixed-effects regressions assess the average effect of 20 major epidemics on FDI, revealing that diseases such as Ebola, MERS, Lassa Fever, and Leptospirosis significantly reduce investment inflows. Second, local projection methods trace the short-and medium-term responses of FDI to health shocks by transmission type. The results show varying recovery patterns: while FDI tends to rebound after direct contact or mosquito-borne outbreaks, airborne diseases cause more persistent declines. Third, the study explores whether stronger healthcare systems can mitigate these negative effects. Results suggest that countries with a higher density of nurses experience less severe FDI losses during outbreaks, particularly for diseases transmitted through direct contact or bodily fluids. These findings underline the importance of healthcare investment not only for public health but also for economic resilience. By distinguishing effects across disease types and highlighting the moderating role of health infrastructure, this study offers practical insights for policymakers seeking to safeguard investment flows during times of crisis.

Based on the 2020 International Social Survey Program (ISSP) IV Environment Module, this paper examines how individual characteristics influence the willingness to pay to protect the environment. We contribute to the literature by highlighting the role of social status: the higher respondents place themselves on the social scale, the more they are willing to pay; and of prosocial orientation: the more willing respondents are to take action for the environment independently of others, the more they are willing to pay. We underline that the impact of social status, i.e., respondents' subjective perception of their position on the social scale, remains significant even after controlling for income and education levels. Our results suggest that, in order to strengthen individuals' willingness to pay for the environment, two types of policies may be effective: (1) policies aimed at influencing subjective social status and its determinants -for example, exposing professional groups to narratives framing their position as socially valued, reducing perceived inequalities, enhancing individuals' sense of control over life events, and fostering orientation toward long-term planning; and (2) policies aimed at shaping prosocial orientation, that is, convincing individuals that personal actions are meaningful even when they are not widely adopted by others. Besides, based on our results, climate policies can be successful when imposing net costs on the following groups: individuals who rank themselves in the top 40% of the population (social status); the 50% of individuals who agree to act without waiting for others to do so (prosocial orientation). By contrast, climate policies that impose net costs on the complementary segments of the population are likely to face disapproval.