48670 research outputs found
Sort by
Comparative Analysis of Risk in Euro-Area Sovereign Yields
Get PDFWhat risk premia exist in Euro-area sovereign yields, and how big are they? What are the similarities and differences across countries? This paper aims to answer these questions for eight Euro countries by using Kalman filtering to decompose their sovereign yields into the following premia: a short-rate and term premium, a default premium, a liquidity premium, and a segmentation premium. The main finding is that countries with similar credit ratings and yield magnitudes tend to exhibit similar patterns, and that on average, countries that were especially affected by the European Debt Crisis (Italy, Portugal, Spain) tend to have default premia that account for more than 50% of their sovereign yields
Virtualizing Reconfigurable Architectures: From FPGAs to Beyond
No full textWith field-programmable gate arrays (FPGAs) being widely deployed in data centers to enhance the computing performance, an efficient virtualization support is required to fully unleash the potential of cloud FPGAs. However, the system support for FPGAs in the context of the cloud environment is still in its infancy, which leads to a low resource utilization due to the tight coupling between compilation and resource allocation. Moreover, the system support proposed in existing works is limited to a homogeneous FPGA cluster comprising identical FPGA devices, which is hard to be extended to a heterogeneous FPGA cluster that comprises multiple types of FPGAs. As the FPGA cloud is expected to become increasingly heterogeneous due to the hardware rolling upgrade strategy, it is necessary to provide efficient virtualization support for the heterogeneous FPGA cluster. In this dissertation, we first identify three pairs of conflicting requirements from runtime management and offline compilation, which are related to the tradeoff between flexibility and efficiency. These conflicting requirements are the fundamental reason why the single-level abstraction proposed in prior works for the homogeneous FPGA cluster cannot be trivially extended to the heterogeneous cluster. To decouple these conflicting requirements, we provide a two-level system abstraction. Specifically, the high-level abstraction is FPGA-agnostic and provides a simple and homogeneous view of the FPGA resources to simplify the runtime management and maximize the flexibility. On the contrary, the low-level abstraction is FPGA-specific and exposes sufficient low-level hardware details to the compilation framework to ensure the mapping quality and maximize the efficiency. This generic two-level system abstraction can also be specialized to the homogeneous FPGA cluster and/or be extended to leverage application-specific information to further improve the efficiency. We also develop a compilation framework and a modular runtime system with a heuristic-based runtime management policy to support this two-level system abstraction. By enabling a dynamic FPGA sharing at the sub-FPGA granularity, the proposed virtualization solution can deploy 1.62x more applications using the same amount of FPGA resources and reduce the compilation time by 22.6% (perform as many compilation tasks in parallel as possible) with an acceptable virtualization overhead, i.e., \u3c10% degradation in single application\u27s performance. Finally, we use Liquid Silicon as a case study to show that the proposed virtualization solution can be extended to other spatial reconfigurable architectures. Liquid Silicon is a homogeneous reconfigurable architecture enabled by the non-volatile memory technology (i.e., RRAM). It extends the configuration capability of existing FPGAs from computation to the whole spectrum ranging from computation to data storage. It allows users to better customize hardware by flexibly partitioning hardware resources between computation and memory based on the actual usage. Instead of naively applying the proposed virtualization solution onto Liquid Silicon, we co-optimize the system abstraction and Liquid Silicon architecture to improve the performance
Three Essays on Internal Migration and Risk Factors for Non-communicable Diseases (NCDs) in Low- and Middle-Income Countries (LMICS)
No full textLow- and middle-income countries (LMICs) have been experiencing extensive internal migration, which is closely associated with the increasing prevalence of risk factors for non-communicable diseases (NCDs). In this dissertation, I study the impact of internal migration on main NCD risk factors across three diverse LMIC contexts: China, Indonesia, and Malawi. In Chapter 1, I introduce the background, motivation, and research goals of this dissertation. In Chapter 2, I use data from the 2011 China Health and Retirement Longitudinal Study (CHARLS) to examine the associations between rural-urban migration and three main NCD risk factors, hypertension, obesity, and abdominal obesity, among older adults aged 45 or above. I find that rural-urban migrants have significantly higher chances of getting all three risk factors than rural non-migrants, suggesting a negative impact of rural-urban migration on health outcomes. Meanwhile, the number of years lived in cities significantly predicts being hypertensive, implying a “years since migration (YSM)” effect. In addition, health-related behaviors examined play a very limited role in mediating the association between migration and health. In Chapter 3, using data from the fourth and fifth waves of the Indonesia Family Life Survey (IFLS), I study the impact of rural-urban migration on overweight status in Indonesia. I find that rural-urban migration is significantly associated with being overweight, and the association is significantly stronger among women than men, demonstrating a gender disparity in health. Moreover, the number of years lived in cities does not predict overweight status, and health behavioral factors still explain little of the association between migration and health. In Chapter 4, I use data from the 2008 and 2019 waves of the Malawi Longitudinal Study of Families and Health (MLSFH) to investigate the impact of internal migration, not only rural-urban but also rural-rural, on weight status in Malawi. I find that rural-urban and rural-rural migration are both significantly associated with increased Body Mass Index (BMI) and that rural-rural migration significantly predicts being overweight. Meanwhile, the impact of rural-rural migration is significantly stronger among women than men, showing a gendered impact of migration on health. In Chapter 5, I summarize findings from the three main chapters and discuss their implications for policymaking and future research
Complex Systems Engineering: Designing Advanced Functions in Dynamical and Mechanical Systems
No full textFrom computation in neural networks to allostery in proteins, numerous natural and artificial systems are comprised of many interacting parts that give rise to advanced functions. To study such complex systems, a diverse array of interdisciplinary tools have been developed that relate the interactions of existing systems to their functions. However, engineering the interactions to perform designed functions in novel systems remains a significant challenge due to the nonlinearities in the interactions and the vast dimensionality of the design space. Here we develop design principles for complex dynamical and mechanical systems at the lowest level of their microstate interactions. In dynamical neural systems, we use methods from control theory and dynamical systems theory to mathematically map precise patterns of neural connectivity to the control of neural states in human and non-human brains (Chapter 2) and to the learning of computations on internal representations in artificial recurrent neural networks (Chapter 4). In mechanical systems, we use methods from algebraic geometry and dynamical systems to mathematically map precise patterns of mechanical constraints to design shape changes as a minimal model of protein allostery and cooperativity (Chapter 6) and to engineer mechanical metamaterials that possess arbitrarily complex shape changes (Chapter 8). These intuitive maps allow us to navigate previously unexplored design spaces in nonlinear and high-dimensional regimes, enabling us to reverse engineer form from function in novel complex systems that have yet to exist
Semantic Simultaneous Localization and Mapping
No full textTraditional approaches to simultaneous localization and mapping (SLAM) rely on low-level geometric features such as points, lines, and planes. They are unable to assign semantic labels to landmarks observed in the environment. Recent advances in object recognition and semantic scene understanding, however, have made this information easier to extract than ever before, and the recent proliferation of robots in human environments demand access to reliable semantic-level mapping and localization algorithms to enable true autonomy. Furthermore, loop closure recognition based on low-level features is often viewpoint dependent and subject to failure in ambiguous or repetitive environments, whereas object recognition methods can infer landmark classes and scales, resulting in a small set of easily recognizable landmarks. In this thesis, we present two solutions that incorporate semantic information into a full localization and mapping pipeline. In the first, we propose a solution method using only single-image bounding box object detections as the semantic measurement. As these bounding box measurements are relatively imprecise when projected back into 3D space and difficult to associate with existing mapped objects, we first present a general method to probabilistically compute data associations within an estimation framework and demonstrate its improved accuracy in the case of high-uncertainty measurements. We then extend this to the specific case of semantic bounding box measurements and demonstrate its accuracy in indoor and outdoor environments. Second, we propose a solution based on the detection of semantic keypoints. These semantic keypoints are not only more reliably positioned in space, but also allow us to estimate the full six degree-of-freedom pose of each mapped object. The usage of these semantic keypoints allows us to effectively reduce the problem of semantic mapping to that of the much more well studied problem of mapping point features, allowing for its efficient solution and robustness in practice. Finally, we present a method of robotic navigation in unexplored semantic environments that robustly plans paths through unknown and unexplored semantic environments towards a goal location. Through the use of the semantic keypoint-based semantic SLAM algorithm, we demonstrate the successful execution of navigation missions through on-the-fly generated semantic maps
Essays on Technology and Platforms
No full textThis dissertation investigates how technology in different marketing-related use cases may affect consumer behavior, and how firms should respond. In particular, we study two important types of platforms in marketing—social media platforms such as Facebook and shopping platforms such as Amazon Marketplace. For social media platforms, this dissertation focuses on a recent, hotly-debated topic—content moderation. We build a game-theoretical model to study how economic incentives will shape a social media platform\u27s content moderation strategy, as well as how a platform\u27s technology strategy will interact with the way in which it moderates its user-generated content. We find that a social media platform\u27s optimal content moderation strategy depends on its revenue model: a platform under advertising is more likely to moderate its content than one under subscription, but does so less aggressively compared to the latter when it moderates content. We also show that a platform under advertising does not necessarily benefit from a better technology for content moderation, but one under subscription does, which means that platforms under different revenue models can have different incentives to improve their content moderation technology. For shopping platforms, we investigate whether retailers should adopt technology-enabled shopping (TES) platforms, such as Amazon\u27s Alexa, as a new distribution channel. We combine game-theoretical analysis and experiments that show consumer reactions to different shopping technology. We find that consumers with stronger brand preferences are less likely to benefit from decision support (DS) technology, whereas ordering convenience (OC) technology benefits all consumers at a similar level. Such differences in consumer reactions to different technology induces vastly different distribution and pricing strategies in retail markets: the heterogeneous consumer valuation of the DS technology results in a monopolistic retailer adoption and generates Pareto improvements, but OC technology results in competitive retailer adoption and generates a prisoner-dilemma type outcome. Furthermore, we also find that a technology provider may choose not to offer the best possible OC technology to mitigate downstream retailer competition
Removing Strong Data Assumptions in Causal Inference Via Large-Scale Optimization
No full textMany traditional and newly-developed causal inference approaches require imposing strong data assumptions, and if those assumptions were violated in practice, these approaches may be inapplicable, suffer from low statistical power, or lead to misleading causal conclusions. In this dissertation, we present three papers to show how large-scale optimization can sometimes aid in removing strong assumptions about the data generating process or the data collection procedure that are required by some existing causal inference approaches. The first and second papers show how large-scale optimization can sometimes help remove strong assumptions about the data generating process. In the first paper, a new adaptive approach is proposed to combine two test statistics in matched observational studies. The proposed adaptive approach asymptotically uniformly dominates both of the two component test statistics in sensitivity analyses, regardless of the underlying data distribution. In the second paper, a model-free and finite-population-exact framework is proposed to analyze randomized experiments subject to outcome misclassification. This new framework is based on large-scale integer programming and can help researchers analyze a randomized experiment subject to outcome misclassification in a more comprehensive way without imposing any additional assumptions on a randomized experiment. The third paper illustrates how large-scale optimization can help remove strong assumptions about the data collection procedure. Specifically, to study the effect of reducing malaria burden on the low birth weight rate in sub-Saharan Africa, a pair-of-pairs approach to a difference-in-differences study is proposed, which is built on optimal matching (a large-scale network flow problem) and cardinality matching (a large-scale integer programming problem). Unlike the traditional difference-in-differences studies, this pair-of-pairs approach does not require either panel data or repeated cross-sectional data to be collected before the analysis stage
Essays on Economic Growth and Inequality
No full textThis thesis studies the impact of policy changes and technological progress on economic growth and inequality. The first chapter studies the impact of intellectual property rights protection policies on firms\u27 boundary and innovation choices, and economic growth. This chapter shows specialization patterns of US firms in the 1980s and 1990s. Specifically: 1) Firms, especially innovating ones, decreased the number of industries in which they produce. 2) Small firms increased innovation intensity while large firms decreased it. A new hypothesis is proposed, highlighting the role of pro-patent reforms that make firms\u27 innovations more tradable. An endogenous growth model with firm heterogeneity is developed. Calibrating the model suggests that increasing tradability of innovations can explain 25% of the decrease in firms’ number of industries and 58% of the reallocation of innovation activities. It results in a 0.64 percent point increase in the annual economic growth rate. The second chapter explores how the rise of digital advertising technology affects consumption, leisure, and welfare of high- and low-income consumers. An information-theoretic model is constructed where free media goods complement leisure and are financed by two types of advertising that inform consumers about the prices of goods--traditional and digital. Calibrating the model shows that the increasing provision of free media goods, due to the rise in digital advertising, boosts consumer welfare significantly. It also leads to more leisure. The increase in leisure is more pronounced for low-income consumers vis-a-vis the high-income ones. The third chapter studies the role of the venture capital (VC) industry in shaping wealth inequality and mobility in the United States. This chapter develops a model where households endogenously choose entrepreneurship entry and the source of external funds (bank or VC). The model can quantitatively match the wealth distribution in the United States. Calibrating the model generates that the VC sector: 1) increases the wealth share of the top 0.1% households by 1 percent points and the wealth share of the top 1% households by 2.1 percent points, 2) increases the probability that the households at the bottom 99% move to the top 1% after a generation by 1.4 percent points
Essays on General Equilibrium
No full textThis dissertation develops new general equilibrium results on how markets react to risk and aggregate information. The first chapter extends a classical result in portfolio theory about the effect of risk on value functions to its effect more generally on policy functions. If odd moments of shocks are zero up to some order, then the odd order marginal effects on value and policy functions of introducing these shocks are zero as well. Mathematically, all coefficients of corresponding odd order in the perturbation parameter are zero. If shocks are symmetric, e.g. normally distributed, then this holds for all odd orders. The main theorem (1) generalizes past results on perturbations and unifies their economic intuition, (2) improves the computation of stochastic coefficients, and (3) illustrates how to derive properties of high order perturbations through simple induction. The second chapter tries to reconcile classical versions of the efficient market hypothesis with the surveyed level of technical analysis in practice. If past security prices are public information, then any patterns contained within should be approximately accounted for in current prices, and fundamental analysis would be relied on relatively more. While each past security price might individually be public information, the disconnect is that it should not imply their collective patterns and interactions are public information as well. Economics, unlike probability theory, must recognize costs and therefore distinguish between observing pieces of information and analyzing their many interactions. (1) We generalize sigma-fields to families of events and define information more broadly as knowledge about optimization solutions. (2) This provides a new framework for efficiency hypotheses and theorems. (3) We illustrate how complex patterns arise from variably diffuse information that only technical analysts can aggregate indirectly, changing the informational behavior of prices
Gravitational Imagination: Picturing Suspension from Eadweard Muybridge to the Space Age
No full textResisting gravity holds an allure. Situating that appeal within the realm of art history, my dissertation charts modern aesthetic efforts to channel and challenge gravitational force—casting suspension as vital to modernism. I contend that new modes of pictorial time—and, in turn, novel possibilities for embodied engagement—emerged once photographic technology accelerated enough to catch airborne bodies and hold them aloft in the space of an image—documenting a potential which was actualized in the Space Age, when humans first experienced sustained weightlessness. Tracing an ungrounded sensibility that emerged between these nodal points, my project offers a thematic account of how gravitational disruption coheres in pictorial composition and perceptual effects. Drawing upon a range of interdisciplinary sources and period voices, my chapters posit the rise of a form of suspended viewership—which does not presume grounded-ness or fixed coordinates, either within artworks or on our part. From Eadweard Muybridge’s photographs of figures held in momentary flight to artists such as Helen Frankenthaler and Marcel Duchamp enacting an “aerial gesture” that employs and subverts gravity, and from Claude Monet’s “upside down” waterlily paintings to Aaron Siskind’s levitational midcentury imagery, my case studies explore increasingly unbound aesthetic terrain. Once gravity became dislodged in visual representation, I argue, formal axes were opened to more symbolic creative dimensions. With that metaphoric tenor, this dissertation defines a pictorial suspension ripe with potential—and charged with the power to resist seemingly inexorable forces. Materializing a stillness that arose in the face of modern momentum, the objects at its core open space for a “gravitational imagination”—founded in the world but also challenging its limits