We show that particularly weak German banks used the European Central Bank’s very long-term refinancing operations (VLTROs) to evergreen exposures to zombie firms. Zombie firms that obtain more credit after the introduction of the VLTROs had a substantially elevated subsequent default probability. This suggests that either positive private information did not materialize or the decision to engage in zombie lending was not driven by any such private information. At the time of loan extension, banks that resort to the VLTRO report lower borrower-specific probabilities of default, reduce loan loss provisions and lower their credit standards relative to banks who lend to the same firm but have not resorted to the VLTRO. Zombie firms, which obtained additional funding from banks relying to a larger extent on VLTRO funding, in turn increase their accounts payable and advance payments received from downstream and upstream firms. This suggests that suppliers relying on banks’ lending decisions as a signal about borrowers’ credit quality might be misled by banks’ zombie lending to extend more trade credit to zombie firms exposing suppliers to elevated contagion risk.
Interest rate spreads vary widely across time and countries and are a central driver of business cycles in emerging market economies (EMEs). Since 2008, advanced market economies (AMEs) have exhibited persistently higher and more volatile spreads, alongside increased macroeconomic volatility and stronger co-movement with EMEs. We document six facts showing that AMEs have become more similar to EMEs along key dimensions. We interpret these patterns through a small open economy model and uncover a stark dichotomy: higher spreads reflect greater indebtedness and lower debt tolerance, whereas greater macroeconomic volatility and co-movement are driven by stronger global shocks.
Control problems frequently arise in scientific and industrial applications, where the objective is to steer a dynamical system from an initial state to a desired target state. Recent advances in deep learning and automatic differentiation have made it increasingly practical to apply these methods to control problems. This paper provides an accessible overview of how neural networks and modern machine learning frameworks can be used to parameterize control inputs for both discrete-time and continuous-time systems, encompassing deterministic and stochastic dynamics. Applications are highlighted across multiple domains, including biology, engineering, physics, and medicine. For continuous-time dynamical systems, neural ordinary differential equations (neural ODEs) provide a useful approach to parameterizing control inputs, while in discrete-time settings, custom control-input parameterizations can be implemented and optimized using automatic differentiation. Overall, the methods presented offer practical solutions for control tasks that are computationally demanding or analytically intractable, making them valuable for complex real-world applications.
Background:Macrolide antibiotics have immunomodulatory activity and when taken chronically reduce exacerbations of COPD. However, chronic use can cause bacterial resistance. EP395 (glasmacinal), a novel macrolide, is being developed as a treatment to reduce exacerbations of COPD without inducing antimicrobial resistance. Methods:In this double-blind, placebo-controlled, phase 2a trial (NCT05572333), patients (≥45 years old, diagnosed with COPD for ≥2 years and stable on at least one maintenance inhaled therapy) were randomised (2:1) to EP395 or placebo daily for 12 weeks. The primary objective was safety, with key secondary objectives assessing pharmacodynamic effects of EP395. Results:A total of 61 patients were randomised (42 EP395, 19 placebo). A 12-week course of EP395 was well tolerated: no serious adverse events were considered related to EP395, and adverse events occurred in similar proportions in both groups (64.3% EP395, 63.2% placebo). Four patients were withdrawn due to adverse events (three EP395, one placebo). Sputum neutrophil elastase and myeloperoxidase, mediators of neutrophil activation, were reduced with EP395 (treatment difference (log scale): neutrophil elastase -0.415 ng·mL-1, 95% CI -0.787 to -0.043 ng·mL-1, p=0.030; myeloperoxidase -0.282 ng·mL-1, 95% CI -0.640 to 0.076 ng·mL-1, p=0.119). Relative changes in neutrophil elastase and myeloperoxidase from baseline with EP395 were 66% and 75%, respectively, of those observed with placebo. Exploratory 16S rRNA sequencing of sputum showed EP395 had no detectable effect on the lung microbiome, including the proportion of pathogenic Proteobacteria species. Conclusion:In patients with stable COPD, EP395 for 12 weeks was well tolerated, demonstrated selective anti-inflammatory activity and had no detectable effect on the lung microbiome.
BACKGROUND:Patients undergoing orthotopic transcatheter tricuspid valve replacement (TTVR) frequently present with a cardiac implantable electronic device (CIED) lead traversing the tricuspid valve. OBJECTIVES:This study sought to investigate the clinical, procedural, and lead-related outcomes of orthotopic TTVR in patients with transvalvular CIED leads. METHODS:All consecutive patients enrolled in the multicenter TRIPLACE (Global Multicenter Registry on Transcatheter Tricuspid Valve Replacement) registry were included for analysis. Patients were stratified based on the presence of a CIED lead traversing the tricuspid valve. Changes in lead function parameters were assessed after TTVR in a subset of these patients who had pacemaker lead parameter data recorded. Lead failure was defined as structural or electrical malfunction requiring new lead or CIED insertion. RESULTS:Among 395 patients, 104 (26.3%) had transvalvular CIED leads. Procedural success, symptomatic improvement, and 30-day mortality were comparable between those with and without CIED. Patients with CIED leads had lower rates of mild or less residual tricuspid regurgitation (82.6% vs 91.4%; P < 0.041) and higher rates of moderate or greater paravalvular leak (17.1% vs 7.1%; P < 0.017). Lead failure occurred in 5.8% over a median follow-up time of 183 days, with modest changes in pacing thresholds. No significant increase in adverse events or mortality was observed at 30 days. CONCLUSIONS:Orthotopic TTVR in patients with transvalvular CIED leads can be safely and effectively performed with low rates of lead failure. Significant paravalvular leak and residual tricuspid regurgitation is more common with a jailed lead. These patients require close CIED follow-up with alternative pacing strategies in place, particularly when pacing dependent. (Global Multicenter Registry on Transcatheter TRIcuspid Valve RePLACEment [TRIPLACE]; NCT06033274).