4760 research outputs found
Sort by
Latent structure estimation for high-dimensional dependent data via eigenanalysis
Full text linkWith advancements in technology, the collection and storage of high-dimensional data have become increasingly common, necessitating tools to analyze such data effectively. Recovering the latent structure has gained popularity as an approach to dimensionality reduction, as the latent processes typically have lower dimensionality, making them easier to analyze and interpret. This thesis explores methods for uncovering the latent structure in high-dimensional data across three domains. Chapter 2 proposes a novel estimation method for the blind source separation model, as introduced in Bachoc et al. (2020). The new method leverages eigenanalysis of a positive definite matrix constructed from multiple normalized spatial local covariance matrices, enabling the handling of moderately high-dimensional random fields. The consistency of the estimated mixing matrix is established with explicit error rates, even under slowly decaying eigen-gaps. Chapter 3 examines the factor model framework for time series Lam and Yao (2012a), with a focus on estimating the number of latent factors. Traditional methods struggle with varying factor strengths, limiting their applicability. To address this, a non-parametric hypothesis testing procedure is proposed, capable of identifying the correct number of factors even when factor strengths differ. The proof on significance level of the test is provided, and its effectiveness is demonstrated through comparisons with existing methods on both simulated and real-world datasets. Chapter 4 addresses the challenge of electricity load forecasting, starting with Generalized Additive Models (GAM) provided by Électricité de France. The residuals from GAM forecasts is analyzed and modeled to uncover its latent structure, which not only simplify the modeling process but also enhance GAM estimations. Two approaches are explored: latent segmentation using TS-PCA Chang, Guo, and Yao (2018a) and Matrix Time Series Decorrelation Han et al. (2023), and dimensionality reduction with factor models using the procedure developed in Chapter 3. Applied to national and regional electricity load data in France, both methods enhance forecast accuracy, as measured by Root Mean Squared Error (RMSE)
Essays in asset management
Full text linkThis thesis contains three essays on the trading behavior and product market innovation of asset managers. The first chapter finds that political beliefs shape professional investors’ interpretations of fiscal policy and, consequently, their asset allocation decisions. Using daily data on Congressional Budget Office cost releases for U.S. legislative proposals, I find that fixed-income mutual fund managers who oppose the current president’s party trade more aggressively and pessimistically in response to news of anticipated rising federal budget deficits. Compared to politically aligned managers, they reduce their positions in long-term Treasury and corporate bonds more significantly while increasing their holdings in Treasury Inflation-Protected Securities. This partisan-driven trading exerts a temporary price impact on Treasury securities and long-term corporate bonds. The second chapter (with Li An, Shiyang Huang, and Dong Lou) documents that despite the removal of all regulatory barriers by 1997, long-short equity mutual funds have seen disappointing growth over the past two decades. We shed new light on this puzzle by documenting a novel set of facts: long-short mutual funds: 1) hold a substantial amount of cash (in excess of cash-collateral requirements) and have an average market beta of 0.6; 2) generate a 5% annual alpha on risky holdings but do not outperform their long-only peers in total returns; and 3) face much higher flow-performance sensitivities and more volatile flows, and use cash buffers more aggressively. These findings challenge prevailing explanations for this puzzle—such as client restrictions, lack of short-selling skills, or high short-selling costs and risks—and motivate a new framework centered on investor clientele and flow responses. The third chapter (with Jiaxing Tian) investigates the decline of traditional mutual funds, focusing on the substitution of mutual funds by collective investment trusts (CITs) in the 401(k) pension investment, and offers insights from both demand and supply sides. Employing several datasets, we demonstrate that CITs are adopted due to their lower fees, comparable returns, and customized nature, aligning with investor preferences sensitive to cost rather than financial transparency. Moreover, mutual fund companies with positive signals, such as past returns and ratings, are incentivized to introduce CITs to reduce auditing costs and gain market shares in 401(k)s. The surge of CITs in 401(k) menus has implications for pension plan governance, with better-governed plans more likely to incorporate CITs. Our findings suggest potential welfare improvements in this delegated asset management model, with investors benefiting from lower total investment costs and mutual fund companies gaining inflow stability. Overall, our research contributes to understanding the dynamics of non-mutual fund investments and their implications for financial markets and investors
Resilience in refugee economies in practice – empirical evidence from the UK, Germany and Sweden
Full text linkRefugee employment outcomes vary widely across high-income economies. In some countries, refugees are able to access more and better jobs. Countries in which this is the case can be called resilient. Existing theories fail to account for variation in refugee employment outcomes and hence resilience. This thesis argues that cross-country differences in refugee employment and resilience are the result of different socio-economic models and the institutions associated with them. Hence, this thesis proposes an institutional account of resilience. Qualitative comparative case studies are used to explain three different patterns of labour market resilience, each of which is associated with a different socio-economic regime. The Liberal Model, represented by the UK, is associated with a relatively fast labour market integration and a high concentration of refugee entrepreneurs. A large low-skilled sector facilitates waged employment, while entrepreneurship can act as a vehicle for socio-economic integration in a labour market characterized by few regulations. The Conservative Continental Model, represented by Germany, is associated with slow, but steady labour market integration and a high concentration of refugees in medium-skilled jobs that require vocational qualifications. Finally, the Nordic Model, represented by Sweden, is associated with slow, but steady labour market integration and a high concentration of refugees in jobs that require post-secondary education. There are two important findings that follow from this study. Firstly, there is not a single pathway to enhance labour market resilience. Resilience is context-specific. Secondly, advanced market economies have the real capacity to integrate refugees into their labour markets successfully
Rooting for culture: a study of the agency of the governed in international organisations
No full textHow can states conventionally characterised as “weak” shape international policy and legal frameworks within international organisations? Based on a qualitative case study analysis of two case studies using causal process tracing, this thesis explores developing states’ strategies to institutionalise the protection and promotion of their local cultural assets at the international level. The first case study examines the World Bank’s commitment to preserving and promoting developing countries’ cultural assets beginning in the 1990s. The second case study explores the WIPO’s commitment to the protection of traditional knowledge and cultural expressions beginning in the late 1990s, and the consecration of this process with the adoption of the WIPO Treaty on Intellectual Property, Genetic Resources and Associated Traditional Knowledge in May 2024. By exploring two original historical episodes based on extensive archival research, this thesis draws attention to the creative potential and activism of a large number of states that have too often been portrayed as passive actors on the international scene, at the mercy of more powerful member states and the dictates of the Secretariats of major international organisations. In choosing to focus on how developing states have institutionalised the protection and promotion of their local cultural assets, this thesis not only sheds light onto a policy domain that is of common interest to many of the world’s countries but that has escaped the radar of IR scholars, but also portrays the institutionalisation of novel and controversial ideas in contestation of long-standing norms, policies and practices. This contrasts with a large portion of the specialised literature that focuses on the co-constitution of norms, the examination of negotiations on discrete items, the resistance against the progress and implementation of existing norms and policies, or their local adaptations. More specifically, by focusing on the agency of the governed within international organisations, this thesis contributes to deepening our understanding of the specific strategies used by relatively weak states, and in particular how they take advantage of their institutional environment to propose new initiatives that serve their interests, offering an account that transcends state-to-state level interactions
Scales of dispossession: policing the threshold of Syrian lives in Istanbul
No full textThis thesis offers a critical mixed-method account of a few Syrian migrant friends living in Istanbul in the last decade. From their gaze I situate Istanbul within a loose carceral landscape, at once grounded within the city streets and situated within a wider geographical arc, encompassing European asylum policies and the Turkish government’s occupation of North Syria. I examine their lives through foregrounding three frames in Istanbul – the policing of space and legal categorisations, the practices of property relations, and the situated Syrian tourist worker. Through close ethnographic engagement with their lives, combined with interviews with journalists, lawyers and human rights workers, critical discourse analysis of newspaper reports, and historical analysis, I analyse each frame as sets of practices and relations which are entangled in wider geographical and temporal scales. I deploy two central concepts in order to critically interrogate these further – dispossession and carcerality – and draw out the multiplicity of power, working across disciplinary, biopolitical, spatial and sovereign registers, through which they operate. Drawing on theories of political belonging, urban space, bordering, black and feminist geography, and racialisation, this thesis is an attempt to map some of the contours through which the figure of “the migrant” is being produced, criminalised and punished across multiple scales, grounded in the streets of Istanbul
Applications of optimal transport in multivariate statistics
No full textFinding a desirable generalisation of rank-based statistical methods to multivariate case has been a timeless statistical endeavor. While various concepts of multivariate rank/quantile have been proposed in the past decades, most of them do not maintain crucial properties enjoyed by the traditional univariate rank/quantile, such as distribution-freeness and strong consistency. A novel multivariate rank/quantile from the perspective of optimal transport (OT) was proposed by Chernozhukov et al. [Che+17] and Hallin et al. [Hal+21]. This OT-based concept extends most of desirable properties of traditional rank/quantile on real line to multidimensional space, thus has drawn an increasing attention in recent years. In this essay, we apply the OT-based multivariate rank/quantile on two statistical domains: multiple-output quantile regression and nonparametric independence testing between random vectors. On the first direction, we introduce a robust estimation of multiple-output linear model coefficient by extending the traditional univariate composite quantile regression to the case of multivariate response variable through OT-based techniques. Both the consistency and the convergence rate of the proposed estimator is established under multivariate heavy-tailed random error case. For the second direction, we proposed an geometrically intuitive correlation coefficient for random vectors utilising the OT-based multivariate rank. The proposed coefficient enjoys an entirely distribution-free asymptotic theory under the independent assumption, thus avoiding any permutation-based p-value calculations. Moreover, unlike many existing measurement, the proposed coefficient is capable of detecting not only functional dependency but also spurious correlation via confounders
Essays on outsourcing, automation and economic justice
Full text linkThis thesis examines how domestic outsourcing and automation shape labour market outcomes, and how the philosophy of John Rawls can inform normative economics and economic policy. It is structured around three distinct essays. The first essay provides the first empirical analysis of the impact of domestic outsourcing on workers in the UK, using matched worker-firm data and focusing on cleaners and security guards. I estimate wage penalties of 2.8% and 5.6% respectively, and show that contractor firms pay systematically lower AKM wage premiums than in-house employers. These differences reflect both lower rents per worker and, for security workers, weaker rent-sharing—highlighting the role of firm-level practices in driving wage inequality. The second essay uses a novel proprietary survey of UK manufacturing sites between 2005 and 2023 to examine the employment effects of CNC machine tools and industrial robots. I show that both technologies are associated with significant increases in employment at adopting plants – between 6% and 9% after four years – compared to a control group of non-adopting plants. I also find positive spillovers on employment among industry competitors and a positive aggregate impact on industry-level employment, challenging the view that automation necessarily displaces workers. The final essay argues that the standard economic interpretation of John Rawls—as an advocate of redistribution justified by a maximin social welfare function—misrepresents the spirit and substance of his theory. I argue that Rawls offers a richer conception of justice, grounded in reciprocity and focused on access to economic power and the social bases of self-respect. I outline the core features of a Rawlsian normative economics—plural in values, grounded in measurable primary goods, and psychologically realistic; and propose a policy agenda centred on predistribution and meaningful work. Together, these essays contribute new evidence on how firms and technologies shape labour market inequality, and offer a rich non-welfarist framework for rethinking the goals and design of economic policy
The hidden hand of politics in education markets: how intergroup conflict & everyday choices shape school practices in Delhi, India
No full textIn fragmented and politically charged societies, what do parents seek from schools, and to whom are education providers accountable? Global education reforms increasingly promote accountability measures and expanded parental choice as mechanisms for improving school quality, positioning these policies as a means of empowering communities and reducing state inefficiencies. India's education system mirrors these global trends, yet in a country where politics permeates everyday life and influences 1.4 billion socially, religiously, and economically diverse citizens, can education ever remain separate from broader political discourses? This study examines how quasi-market reforms in the Indian capital’s education system privilege certain voices whilst marginalising others, shaping schools in ways that reflect existing social hierarchies. Drawing on tools from political science and political anthropology, it interrogates how seemingly apolitical mechanisms – such as parental voice and choice - are deeply embedded in struggles over power, identity, and exclusion. Through in-depth interviews with 230 parents and school representatives across Delhi and an analysis of 192 school disclosure documents, this research reveals how market-driven competition creates conditions of hyper-accountability, forcing schools to align with dominant group preferences rather than educational principles. Unexpectedly, the findings also reveal a form of religious nationalism that emerges from below - what this study terms 'bottom-up Hindutva' - rather than through state policy as existing scholarship suggests. This operates through parental demands for Hindu-first curricula, pressure to dismiss Muslim staff, and the careful crafting of socio-religious homogeneity in school communities. With this, the research challenges the assumption that parental agency is a neutral or depoliticised force, arguing instead that it serves as a mechanism for entrenching socio-political divisions. By questioning what and whom we consider 'political' in education research, this study calls for a critical rethinking of policies that fail to recognise how quotidian schooling decisions reproduce broader systems of power and exclusion in deeply stratified societies
Social media and democratic deliberation
Full text linkThe internet continues to have a profound influence upon political communication. Have social media changed the language we use to talk about politics, made particular voices more prominent in online discussions, or even led people to turn away from democratic deliberation entirely? In three distinct papers, this thesis explores each of these questions in turn, making the following theoretical, methodological, and practical contributions. First, using a novel dataset of 4 million tweets, paper 1 shows how political elites in the United Kingdom have used increasingly emotive rhetoric over time on Twitter. I argue this changing rhetorical behaviour has been influenced by, and rewarded with, greater engagement. Paper 2 also employs new data, linking a nationally-representative survey of the UK population with Twitter accounts, to show that people who discuss politics on Twitter are more ideologically and affectively extreme than those who do not. These findings have important implications for understanding who we are most likely to hear in online political discussion, and for increasingly-polarised online political debate. Finally, in a pre-registered lab-in-the-field experiment on Facebook, I argue that people disengage from discussing contentious political issues in groups of people with contrasting opinions and beliefs. In conducting a real-world experiment which attempts to causally identify why people turn away from political discussion, this study represents a methodological advance on existing survey-based research, and understanding the key drivers of online self-censorship may help mitigate some of its most negative consequences. Overall, these findings contribute to our understanding of how social media is influencing the scope and tone of democratic debate
Essays in finance
Full text linkThis thesis examines the strategic and market implications of open market repurchase (OMR) programs through three interconnected studies that advance our understanding of corporate payout policy and market dynamics. The first chapter provides causal evidence that firms significantly adjust repurchase activity in response to undervaluation. Using price pressure induced by mutual fund flows and leveraging the 2003 mutual fund trading scandal as a natural experiment, I establish that flow-induced valuation shocks causally drive repurchase decisions. Instrumental variable estimates reveal substantially stronger effects than standard regressions, suggesting that the non-fundamental component of fund flows is potent but rare, with its true impact typically masked in conventional analyses. Analysis of long-run stock performance reveals that the well-documented buyback anomaly is primarily driven by repurchase announcements following periods of negative fund flows, highlighting how flow-driven price distortions meaningfully impact corporate payout policy and create predictable patterns in subsequent stock performance. The second chapter examines the puzzling heterogeneity in completion rates of open market repurchase programs, where some announcing firms execute zero repurchases while others complete their programs rapidly. I propose that managers strategically balance duration-dependent costs of undervaluation against immediate costs of share repurchases, with completion decisions signaling expected timelines of information asymmetry resolution. Using hand-collected SEC filing data spanning 2004-2019, I find that low-completion firms significantly outperform analyst forecasts in years one and two post-announcement, while high-completion firms excel in years three and four, with corresponding patterns in market reactions and long-run returns. The third chapter examines how OMR programs affect firms’ exposure to systematic liquidity shocks. I find that repurchasing firms act as buyers of last resort, experiencing significant but temporary declines in liquidity commonality during programs. This reduction in liquidity commonality is accompanied by decreased liquidity risk, highlighting firms’ role in stabilizing against institutional demand and market maker supply variations