62880 research outputs found
Sort by
Systemic risk under the radar: evidence from building societies and challenger banks
Full text linkThis paper provides the first comprehensive quantification of the systemic risk posed by non-listed financial institutions in the UK, focusing on building societies, digital-only challenger banks, and foreign-owned retail banks. Using an indirect estimation approach, systemic risk is measured through
balance sheet characteristics, calibrated against listed institutions' SRISK values. The findings reveal that Nationwide ranks among the top ten systemically important institutions, while several other building societies contribute significantly to aggregate systemic risk. In contrast, digital-only challenger
banks exhibit low systemic risk due to high equity ratios and limited interconnectedness, despite rapid growth and persistent financial losses. Santander, an foreign-owned retail bank, emerges as the ninth most systemically important institution, with risk levels comparable to systemically-important domestic banks. We conduct extensive robustness checks, including alternative predictors and SRISK
specifications, out-of-sample forecasting, and Principal Component Analysis, which confirms the strong co-movement between building societies and the largest UK banks. Finally, we compare SRISK with traditional Z-score metrics to highlight their complementary nature. These findings underscore the need to extend systemic risk frameworks beyond listed entities and support calls to expand the stress testing perimeter to include large non-listed and foreign-owned firms
Driving dynamical inner‐heliosphere models with in situ solar wind observations
No full textAccurately reconstructing the solar wind throughout the inner heliosphere is essential for understanding solar–terrestrial interactions and improving space-weather forecasts. Conventional reconstruction methods rely on photospheric magnetic field observations and coronal models to estimate solar wind conditions near the Sun, typically at 0.1 AU. This introduces substantial uncertainty in the background flow used by heliospheric models through which coronal mass ejections (CMEs) propagate. Here we present a new approach that instead derives the inner-boundary conditions directly from in situ solar wind observations, typically obtained near 1 AU. These observations are ballistically backmapped to 0.1 AU while accounting for both solar wind acceleration and solar rotation, and then corrected for stream-interaction effects using a convolutional neural network trained on synthetic model data. The resulting 0.1 AU boundary conditions are used to drive the Heliospheric Upwind eXtropolation with time dependence (HUXt) model. Applied to the highly geoeffective May 2024 CME interval, this method reproduces solar wind conditions at Earth and at Solar Orbiter—on the far side of the Sun—with speed errors reduced by around 50% relative to traditional coronal-model approaches. Although this represents a post-event reconstruction rather than an operational forecast, the approach provides a fast, accurate, and magnetogram-independent means of reconstructing the inner heliosphere, paving the way for improved CME analyses and future forecasting applications
The prohibition on forced displacement, the right to leave, non-refoulement, and the right to return: four sides of the same coin?
No full textThe paper is set against the near absence of external protection responses to the humanitarian catastrophe in Gaza. Querying the interplay between four recognised international legal norms in the context of armed conflict, it seeks to provide doctrinal clarity in a context where the range and interaction of diverse legal standards may generate uncertainty or claims of apparent norm conflict: the prohibition on forced displacement, the right to leave any territory, non-refoulement, and the right to return to one’s ‘own country’ including as part of the realisation of a collective right to self-determination. The paper posits that a future realisation of the Palestinian people’s right to self-determination has been co-opted by external actors as a justification for infringing, in an immediate and tangible sense, the individual right of Gazans to leave the strip in order to seek and to enjoy elsewhere protection from rights violations, some of which breach jus cogens norms. This latest manifestation of ‘Palestinian exceptionalism’ has had dire consequences for individual Palestinians and, unless unwaveringly rejected, could detrimentally affect those fleeing future armed conflicts
Intra-Caribbean diplomacy and imperial negotiation; Edward Trelawny, the Marquis de Larnage, and Anglo-French relations in the West Indies, 1720–1748
Full text linkDuring the early modern period, concepts of power and authority were heavily contested between rival nations and nascent colonial empires. However, this negotiation was also present in the dealings between colonial officials and agents of the metropolitan government. As well as dealing with each other, the British governor of Jamaica and the French governor of southern Saint Domingue had to constantly negotiate with their superiors in Europe, and with Navy officers sent to enforce the metropoles’ agenda in the region. By analysing the correspondence exchanged between these actors, we can understand how intra-Caribbean diplomacy worked and how specific individuals in the region, could and did influence the outcome of Atlantic conflicts
Depth of Processing, learner aptitude, and the acquisition of L2 English grammatical structures
No full textDepth of Processing (DoP) is central to several models of instructed second language
acquisition (ISLA) (Leow, 2015). While DoP has been extensively examined in vocabulary
learning (e.g., Laufer & Hulstijn, 2001), its role in the acquisition of grammatical structures,
particularly through oral and audio-visual modalities remains comparatively underexplored.
Drawing on the Levels of Processing framework (Craik & Lockhart, 1972), this study
investigated whether task-induced processing depth influences the acquisition of third
conditionals and comparatives, and whether language aptitude (LLAMA B, D, F) predicts
learning gains.
Using a pretest–posttest design, four intact classes involving 108 Grade 8 EFL learners in India
were assigned to one of the three experimental conditions differing in level and type of
processing: Low DoP, High DoP Explicit, and High DoP Implicit. The treatment task required
the participants to listen to an audio-visual story containing the target structures (third
conditionals and comparatives) and complete analytic tasks (varying in different groups) before
listening to the story for a second time. The oral performance of the participants was analysed
for accuracy, and frequency of use of the target structures. A grammaticality judgement test
(GJT) and an elicited imitation task (EIT) were used to assess the participants’ gains in the
knowledge of the target structures. In addition, language aptitude (LLAMA B, D, F) was
measured and modelled as a predictor of gain scores.
Results showed a significant effect of processing condition on GJT performance with the High
DoP Implicit group outperforming the High DoP Explicit group on total scores and
comparatives. In contrast, ANCOVA analyses indicated no significant group differences on EIT
outcomes once pretest performance was controlled, with pretest scores strongly predicting
posttest performance. In oral production, processing effects were structure-specific: the Low
DoP group showed greater accuracy gains for comparatives, while both High DoP groups
revealed increased frequency of use of third conditionals. However, across all four measures,
multiple regression analyses indicated that LLAMA B, D, and F did not significantly predict
gain scores.
Overall, the findings support a processing-by-structure pattern: deeper processing benefits were
most evident for the more salient structure (comparatives) on an explicit, written measure,
whereas complex, low-salience forms (third conditionals) showed limited accuracy change but
increased frequency of use under deeper processing
Artificial Intelligence and the prohibition on the use of force: intention and causation
No full textThis article explores the application of Article 2(4) of the United Nations Charter to AI-enabled systems that carry out unintended engagements involving the use of force. First, it analyzes whether State responsibility for a breach of the prohibition on the use of force is defined in subjective or objective terms. Most commentators maintain that a State must intend to use force against the victim State in order for the prohibition to apply. However, through an examination of State practice, this article demonstrates that the prohibition is based on objective responsibility. Second, this article assesses whether a State must cause the resulting use of force in order for responsibility to ensue. After determining that causation is a condition precedent for establishing a breach of the prohibition on the use of force, this article explains that causation comprises two elements: factual causation asks whether the harmful effects would have occurred but for the impugned conduct, while legal causation asks whether the use of force was reasonably expected when the operation was launched. This article then examines how these elements apply to AI-enabled systems that engage in unintended uses of force and offers illustrative examples
A frequency stability predictive approach with PV integration. A case study for the Republic of Mauritius
No full textFacial emotion recognition from feature loss media: human versus machine learning algorithms
No full textThe automatic identification of human emotion, from low-resolution cameras is important for remote monitoring, interactive software, pro-active marketing, and dynamic customer experience management. Even though facial identification and emotion classification are active fields of research, no studies, to the best of our knowledge, have compared the performance of humans and Machine Learning Algorithms (MLAs) when classifying facial emotions from media suffering from systematic feature loss. In this study, we used singular value decomposition to systematically reduce the number of features contained within facial emotion images. Human participants were then asked to identify the facial emotion contained within the onscreen images, where image granularity was varied in a stepwise manner (from low to high). By clicking a button, participants added feature vectors until they were confident that they could categorise the emotion. The results of the human performance trials were compared against those of a Convolutional Neural Network (CNN), which classified facial emotions from the same media images. Findings showed that human participants were able to cope with significantly greater levels of granularity, achieving 85% accuracy with only three singular image vectors. Humans were also more rapid when classifying happy faces. CNNs are as accurate as humans when given mid- and high-resolution images; with 80% accuracy at twelve singular image vectors or above. The authors believe that this comparison concerning the differences and limitations of human and MLAs is critical to (i) the effective use of CNN with lower-resolution video, and (ii) the development of useable facial recognition heuristics
The importance of accounting for stakeholder values, power relationships and language in constructing relevant and trustworthy climate information
No full textFacing increasing risks from climate change, governments at all levels have started to mainstream the use of climate information. It has been widely acknowledged that the inclusion of stakeholder knowledge and needs, for example, in a co-design and co-production process, is important for producing user-relevant information. Here we start from a hypothetical example and two real-world case studies from South America and West Africa to discuss the role of user values, power relationships and language in the construction of climate information. While these aspects have been discussed individually in several papers, we focus on the mutual influences of these aspects in the information construction and argue that, therefore, they cannot be considered separately. We identify five dimensions—the level of risk, the complexity of the scientific problem, user values, power relationships and language—to characterize the complexity of a given user context. Analyzing these dimensions can guide the choice and design of user engagement in a given situation. In particular, even basic research may benefit from such an engagement. Regularly accounting for these aspects in research projects may require substantial changes in the way research funding is organized and how the work of researchers is rewarded
Hamiltonian dynamics
No full textHamiltonian dynamics describes the evolution of conservative physical systems. Originally developed as a generalization of Newtonian mechanics, it represents a core component of any undergraduate physics curriculum. What is not so widely recognized is that the ideal (i.e. conservative) form of the governing equations used in dynamical meteorology are also Hamiltonian dynamical systems. This chapter explains how this is so, and some of the consequences that follow from this fact. It is important to be able to connect theoretical results across the hierarchy of various models used in dynamical meteorology, from the simplest to the most complex. Hamiltonian dynamics is what allows one to do precisely that