We propose a large-scale Bayesian VAR model with factor stochastic volatility to investigate the macroeconomic consequences of international uncertainty shocks on the G7 countries. The factor structure enables us to identify an international uncertainty shock by assuming that it is the factor most correlated with forecast errors related to equity markets and permits fast sampling of the model. Our findings suggest that the estimated uncertainty factor is strongly related to global equity price volatility, closely tracking other prominent measures commonly adopted to assess global uncertainty. The dynamic responses of a set of macroeconomic and financial variables show that an international uncertainty shock exerts a powerful effect on all economies and variables under consideration. / Series: Department of Economics Working Paper Series

The generation of vector boson mass via the Higgs mechanism in the Standard Model has been confirmed by the 2012 discovery of a candidate Higgs boson in the H&rarr;WW, H&rarr;ZZ, and H&rarr;&gamma;&gamma; decay channels. In contrast, the Yukawa couplings hypothesized to provide the mass of fermions in the Standard Model have yet to be observed. The H&rarr;&tau;&tau; decay channel currently provides the best opportunity for observing these couplings. This thesis describes two separate but related searches for Higgs boson decays in the H→&tau;&tau; decay channel using proton-proton collisions recorded by the ATLAS detector. The first analysis is a general search for all Higgs boson production mechanisms leading to a H&rarr;&tau;&tau; decay using 4.5 fb⁻¹ of 7 TeV and 20.3 fb⁻¹ of 8 TeV proton-proton collision data. A deviation from the background-only hypothesis is observed with a significance of 4.5&sigma; for a hypothetical Higgs boson mass of m_H = 125 GeV — a strong indication of a H&rarr;&tau;&tau; signal. For the same mass point, the best fit value for the signal strength is found to be 1.43 ^&plus;0.43<sub style='position: relative; left: -2.4em;'>−0.37</sub> x the Standard Model expectation. The second analysis is a search for Higgs boson production in association with a vector boson using 20.3 fb⁻¹ of 8 TeV proton-proton collision data. Results in the Z_llH&rarr;_&tau;l&tau;h channel indicate limits of 9.14 x the Standard Model expectation for VH signal production at m_H = 125 GeV. In addition, two studies on enhancement of computing performance in the ATLAS trigger and data analysis pipeline are presented.

Active learning is a subfield of machine learning that has been successfully used in many applications. One of the main branches of active learning is query synthe- sis, where the learning agent constructs artificial queries from scratch in order to reveal sensitive information about the underlying decision boundary. It has found applications in areas, such as adversarial reverse engineering, automated science, and computational chemistry. Nevertheless, the existing literature on membership query synthesis has, generally, focused on finite concept classes or toy problems, with a limited extension to real-world applications. In this thesis, I develop two spectral algorithms for learning halfspaces via query synthesis. The first algorithm is a maximum-determinant convex optimization method while the second algorithm is a Markovian method that relies on Khachiyan’s classical update formulas for solving linear programs. The general theme of these methods is to construct an ellipsoidal approximation of the version space and to synthesize queries, afterward, via spectral decomposition. Moreover, I also describe how these algorithms can be extended to other settings as well, such as pool-based active learning. Having demonstrated that halfspaces can be learned quite efficiently via query synthesis, the second part of this thesis proposes strategies for mitigating the risk of reverse engineering in adversarial environments. One approach that can be used to render query synthesis algorithms ineffective is to implement a randomized response. In this thesis, I propose a semidefinite program (SDP) for learning a distribution of classifiers, subject to the constraint that any individual classifier picked at random from this distributions provides reliable predictions with a high probability. This algorithm is, then, justified both theoretically and empirically. A second approach is to use a non-parametric classification method, such as similarity-based classification. In this thesis, I argue that learning via the empirical kernel maps, also commonly referred to as 1-norm Support Vector Machine (SVM) or Linear Programming (LP) SVM, is the best method for handling indefinite similarities. The advantages of this method are established both theoretically and empirically.

New donor acquisition and current donor promotion are the two major programs in fundraising for higher education, and developing proper targeting strategies plays an important role in the both programs. This thesis presents machine learning solutions as targeting strategies for the both programs based on readily available alumni data in almost any institution. The targeting strategy for new donor acquisition is modeled as a donor identification problem. The Gaussian na ̈ıve bayes, random forest, and support vector machine algorithms are used and evaluated. The test results show that having been trained with enough samples, all three algorithms can distinguish donors from rejectors well, and big donors are identified more often than others.While there is a trade off between the cost of soliciting candidates and the success of donor acquisition, the results show that in a practical scenario where the models are properly used as the targeting strategy, more than 85% of new donors and more than 90% of new big donors can be acquired when only 40% of the candidates are solicited. The targeting strategy for donor promotion is modeled as a promising donor(i.e., those who will upgrade their pledge) prediction problem in machine learning.The Gaussian na ̈ıve bayes, random forest, and support vector machine algorithms are tested. The test results show that all the three algorithms can distinguish promising donors from non-promising donors (i.e., those who will not upgrade their pledge).When the age information is known, the best model produces an overall accuracy of 97% in the test set. The results show that in a practical scenario where the models are properly used as the targeting strategy, more than 85% of promising donors can be acquired when only 26% candidates are solicited. / Graduate / liangye714@gmail.com

We analyze the interaction between monetary policy in the US and the global economy proposing a new class of Bayesian global vector autoregressive models that accounts for time-varying parameters and stochastic volatility (TVP-SV-GVAR). Our results suggest that US monetary policy responds to shocks to the global economy, in particular to global aggregate demand and monetary policy shocks. On the other hand, US-based contractionary monetary policy shocks lead to persistent international output contractions and a drop in global inflation rates, coupled with rising interest rates in advanced economies and a real depreciation of currencies with respect to the US dollar. We find considerable evidence for heterogeneity in the spillovers across countries, as well for changes in the transmission of monetary policy shocks over time. (authors' abstract) / Series: Department of Economics Working Paper Series

This paper examines the Environmental Kuznets Curve (EKC) hypothesis and tests for causality using Dynamic Ordinary Least Squares (DOLS) and the Vector Error Correction Model (VECM). There is evidence of long-run relationships in the three models under consideration. The Dynamic Ordinary Least Squares (DOLS) finds no evidence to support the existence of an environmental Kuznets curve (EKC) hypothesis for Zambia in the long-run. The evidence from the long-run suggests an opposite of the Environmental Kuznets Curve (EKC), in that the results indicate a U-shaped curve relationship between income and carbon emission. The conclusion on causality based on the VECM is that there is evidence of neutrality hypothesis between either total electricity and income or between industrial electricity and income in the short-run Additionally, there is evidence of conservation hypothesis in the context of residential and agricultural electricity consumption.

A search for vector-like quarks (VLQs) decaying to a Z boson using multi-stage machine learning was compared to a search using a standard square cuts search strategy. VLQs are predicted by several new theories beyond the Standard Model. The searches used 20.3 inverse femtobarns of proton-proton collisions at a center-of-mass energy of 8 TeV collected with the ATLAS detector in 2012 at the CERN Large Hadron Collider. CLs upper limits on production cross sections of vector-like top and bottom quarks were computed for VLQs produced singly or in pairs, Tsingle, Bsingle, Tpair, and Bpair. The two stage machine learning classification search strategy did not provide any improvement over the standard square cuts strategy, but for Tpair, Bpair, and Tsingle, a third stage of machine learning regression was able to lower the upper limits of high signal masses by as much as 50%. Additionally, new test statistics were developed for use in the Neyman construction of confidence regions in order to address deficiencies in current frequentist methods, such as the generation of empty set confidence intervals. A new method for treating nuisance parameters was also developed that may provide better coverage properties than current methods used in particle searches. Finally, significance ratio functions were derived that allow a more nuanced interpretation of the evidence provided by measurements than is given by confidence intervals alone.

The principle aim of this thesis is to construct a consumption function for Turkey for policy analysis using the annual State Planning Organisation (SPO) time-series data. This study commences from 1962 and extends until the end of 1994, when a financial crisis occurred in Turkey. It attempts to analyse not only the decline in the private savings rate during the first half of the 1980s, but also the significant rise from 1986 onwards. The thesis starts with an introduction which explodes the main research objectives, considers the existing consumption theories and extentions, records the main data features to be explained, briefly overviews the modelling strategy and discusses the basic considerations of the research and gives the structure of thesis. A literature survey on the theory of consumption is given in Chapter 2. The LifeCycle/ Permanent-Income hypothesis is considered as central to the two mainstream approach. : the Euler approach and the solved-out approach. These approaches are further extended by considering uncertainty and precautionary saving, credit restrictions, saving and leisure. habit or costs of adjustments and the durability of goods, the role of assets and asset prices. financial liberalisation and demographic factors. Finally, comparisons between the two approaches arc made in the conclusion of that chapter. Theory can deliver concepts with permanent relationships in economics, but it should be supported by empirical findings, since theory alone is insufficient to determine the actual economic relationship. Hence, Chapter 3 focuses on theoretical and appl ied modelling issues to construct a theory-consistent, congruent and encompassing consumption function. Congruency implies that the empirical model matches the available evidence in al l measured attributes (i.e., it is consistent with the theory from which it was derived, has unexplained components that arc innovations against available information, has basic parameters that are constant, is data admissible, and where any conditioning variables are weakly exogenous for the parameters of interest). Encompassing denotes that the model of interest can account for the result of rival models of the same phenomena. I also define structure as the set of invariant features of the economic mechanism. A parameter can be structural only if it is invariant for an extension of the sample period (constant), is invariant with respect to changes elsewhere in the economy (regime shifts), and is invariant over extensions of the information set (adding more variables). Chapter 4 examines the small-sample properties of the statistical methods used by means of Monte Carlo simulations. The informativeness of the data is investigated in an unrestricted Vector Auto-regression (VAR) with small-samples of noisy data combined with a high real growth rate and nominal inflation. This is to see how the relative drift dominates in explaining the informativeness of the data. The Monte results are summarised by using response surfaces to relate the biases to sample size. The ratio of standard deviations to standard errors in each equation is also analysed. The strong impacts of the system error variances in these response surfaces indicate the importance of high variances in VA Rs. Furthermore, I found noise, and a function of the signal to noise ratio. and cross-equation correlation had a large impact, but less effect from the relative drift. Chapter 5 presents an overview of the Turkish Economy, particularly during the sample period. by pointing out the lessons to be drawn from the stabilisation experiments and their effect on the private sector saving decision in Turkey. The aim of Chapter 6 is to get nominal housing wealth and housing price data from the available data, such as the nominal private disposable income. nominal private investment in the housing sector and the consumer price index, since housing wealth is claimed to be a major determinant of private savings in Turkey. Chapter 7 aims to reveal the problems of Turkish data by analysing the history of the Turkish a1ional Accounts to construct a data-base for estimating a consumption function for Turkey. GDP by expenditure is constructed from five different sources. Turkish accounting residuals are allocated by applying the linear regression approach. The results show that GDP-by-output is more reliable than the GDP-by-expenditure measure for Turkey. Chapter 8 is devoted to the time series modelling and evidence. Previous findings on consumption for Turkey have been formulated using conventional econometric techniques with a static estimation methodology within the Permanent Income Hypothesis (PIH). I adopted the equilibrium correction model (ECM) solved-out consumption function approach and tried to incorporate the effects of age. precautionary behaviour in the case of uncertainty, credit constraints, habits or costs of adjustments. and the durability of goods for developing belier understanding of private sector savings behaviour in Turkey. The modelling is based on the dynamic econometric methodology that involves the estimation of a general unrestricted model (GUM). a co-integration and long-run analysis, and the simplification of the GUM to a parsimonious dynamic model that is deduced by applying a sequential testing procedure. The final model is congruent: It matches the available evidence in all measured attributes and forecasts well, has white noise errors and constant parameters, and encompasses the VAR model equation as well as other specifications in previous models. Moreover, the model has a structural interpretation. The results of the final model reveal strong positive effects of the real interest rate. inflation and inflation uncertainty, a strong negative effect of population aged 15-44, a positive effect after one lag period of the change in the average propensity to consume. which represents the effects of expectations, habits or adjustment costs, in addition to the significant effect of the inverse of per capita Private Disposable Income and the change in housing wealth to income ra1io on the private average propensity to consume in Turkey. These findings offer an explana1ion for the salient features of the Turkish consumption pattern observed from 1he lime series data. These results also provide some policy implications such that inOation control should be strengthened and improved for consumption stabilisation. Furthermore. interest rate policy also has an important role to play in the savings process in Turkey. The research on small-sample properties of 1hc statistical methods by means of Monte Carlo Simulations strengthens the results of the empirical model. These. confirm the poor determination of intercepts in I(I) VARs, and the corresponding advantages of an equilibrium correction model formulation. Furthermore. the insignificance the of irrelevant dynamics should encourage model builders to use a dynamic econometric methodology to develop parsimonious models, such as used for building a consumption model for Turkey in this thesis.

In this thesis we study various ways in which every two general points on a variety can be connected by curves of a fixed genus, thus mimicking the notion of a rationally connected variety but for arbitrary genus. We assume the existence of a covering family of curves which dominates the product of a variety with itself either by allowing the curves in the family to vary in moduli, or by assuming the family is trivial for some fixed curve of genus g. A suitably free curve will be one with a large unobstructed deformation space, the images of whose deformations can join any number of points on a variety. We prove that, at least in characteristic zero, the existence of such a free curve of higher genus is equivalent to the variety being rationally connected. If one restricts to the case of genus one, similar results can be obtained even allowing the curves in the family to vary in moduli. In later chapters we study algebraic properties of such varieties and discuss attempts to prove the same rational connectedness result in positive characteristic.

This article assesses the empirical relationship between per capita income growth fluctuations and the age-structured human capital variations across four groups of geographically clustered developed and developing countries from spatial perspective. We estimate a spatial Vector Autoregressive (VAR) model of income dynamics where the distance between countries is defined on relational space based on their similarity in appropriation tendency of human capital in the production processes. These distances are computed using a newly developed human capital data set which fully characterizes the demographic structure of human capital, and thus underlines the joint relevance of demography and human capital in economic growth. Spatial effects on growth interdependence and complementarity are then explored with respect to the proposed distance metrics. Our results imply that significant cross-country growth interdependence based on human capital distances exists among defined country groups suggesting the need for a cooperative policy programme among them. We also find that the relationship between economic growth and human capital is highly nonlinear as a function of the proposed human capital distance.