Doctor of Philosophy (Ph.D.), Bowling Green State University, 2023, Data Science

Deep Gaussian Mixture Models (DGMMs) are probabilistic models that combine multiple layers of latent variables (Viroli and McLachlan, 2019). DGMMs adeptly capture intricate, non-linear interactions among variables, facilitating efficient unsupervised learning. This study strives to improve DGMMs by introducing a method to systematically choose optimal covariance structures for each DGMM layer. Our proposal involves a closed multiple-testing procedure that utilizes the likelihood-ratio test to select the most suitable covariance structure from a set of candidate structures. Our proposal is inspired by the likelihood-ratio method proposed by Greselin and Punzo (2013). Typically, information criteria are widely used in the context of model selection, but they have different properties and may be more appropriate for different types of data or modeling situations. Additionally, the selection of a specifc information criterion is subjective and many practitioners tend to use a particular method routinely, which can limit the potential for discovering the best covariance structure for the data at hand (Greselin and Punzo (2013), Punzo et al. (2016)). The proposed method draws inspiration from McNicholas and Murphy (2008), in the context of mixture factor analyzers, where constraints are applied to covariance structures. In Deep Gaussian Mixture Models (DGMMs), these covariance structures can be defned at each layer, creating a range of complexities. To aid covariance structure selection in DGMMs, it is assumed that each cluster within a layer shares the same covariance structure. The chosen structure achieves a balance between model complexity, enhancing performance and predictive accuracy. The method employs a closed multiple-testing approach based on the likelihood ratio test, comparing likelihoods of different covariance structures for the DGMM. We conduct a series of simulations considering multiple heteroscedasticity configurations that represent different cova (open full item for complete abstract)

Committee: Junfeng Shang Ph.D. (Committee Chair); Lauren Maziarz Ph.D. (Other); Hanfeng Chen Ph.D. (Committee Member); Rob Green Ph.D. (Committee Member) Subjects: Statistics

Master of Arts, Miami University, 2017, Economics

Significant attention in the financial economics literature is given to the usage of aggregated factors in their ability to explain variability in asset returns. Whereas the Capital Asset Pricing Model (CAPM) considers the excess return on the market portfolio as the dominant source of systematic variability in asset returns, the framework of Arbitrage Pricing Theory (APT) suggests that systematic risk can be further decomposed into numerous common risk factors that underlie co-movement in asset returns. Chen, Roll, and Ross (1986) popularized empirical evaluation of macroeconomic indicators in their relation to asset returns, finding that macro-economic indicators can be useful to price assets and carry statistically significant risk premiums in sample. Following the intuition of the Roll (1977) critique, I consider the pricing of risk derived from unexpected shocks, or innovations, to a wider set of macroeconomic and capital market variables. I find that information contained in shocks to common risk factors is significantly priced in the cross-section of asset returns and differs from information contained in the Fama-French-Carhart factors.

Committee: Thomas Boulton Ph.D. (Advisor); George Davis Ph.D. (Committee Member); Tyler Henry Ph.D. (Committee Member) Subjects: Economics; Finance

Doctor of Philosophy, The Ohio State University, 2017, Agricultural, Environmental and Developmental Economics

This dissertation consists of three essays on trade, market structure and agricultural cooperative financing. In my first essay, we explore how public investment in commercial infrastructure affects the composition of trade between countries. We find that, ceteris paribus, greater public investment in commercial infrastructure raises general labor productivity, leading to gains in workers' real income. However, workers must be taxed to finance the infrastructure improvements, so that the net benefits of increased public investment are rendered ambiguous. In my second essay, we explore the relationship between wage premia and market structure. We find that the wage premium is not the result of relative labor endowments between the two trading countries, as past literature has suggested, but is the result of differences in market structure. More specifically, the wage premium is higher under oligopolistic competition than under monopolistic competition. In my third essay, we examine how capital constraints affect the growth of agricultural cooperatives and whether external sources of capital relieve capital constraint problems. We find that long-term debt use, cash flow, unallocated equity and long-term debt financing are critical contributors to asset growth for small and medium-sized cooperatives.

Committee: Mario Miranda (Advisor); Ani Katchova (Committee Member); Brian Roe (Committee Member) Subjects: Agricultural Economics

Doctor of Philosophy, The Ohio State University, 2014, Business Administration

This dissertation aims at understanding the organizational antecedents and performance consequences for hospitals of combining conformance quality and experiential quality when delivering care. Conformance quality reduces variance around a set of technical guidelines and has been a long standing priority in health care delivery. Experiential quality requires a shift in culture and focuses on the quality of interactions between caregivers and patients and by definition enhances variance. In the context of health care delivery, reconciling both quality dimensions has proven a challenge given the need for the same caregiver to simultaneously focus on two dimensions that trigger different learning mechanisms. Existing theories on reconciling dual learning goals are not adapted to the particularities of this setting. This dissertation develops a framework on combining conformance and experiential quality in the health care delivery context through three inter-related studies that use multiple methods. The first study, “The Impact of Combining Conformance and Experiential Quality on Health Care Clinical and Cost Performance”, investigates whether overcoming the tensions between conformance quality and experiential quality has an impact on the effectiveness of care delivery. The study builds on the quality management and organizational learning literature to investigate the impact of combining experiential and conformance quality dimensions on clinical and cost outcomes. Hypotheses are tested using secondary data from 9 distinct sources on 3474 U.S. acute care hospitals over a six-year period. Econometric analyses indicate that combining conformance and experiential quality promotes clinical outcomes but at the expense of cost efficiency. The second study, “The Effects of Coordination Mechanisms on Combining Conformance and Experiential Quality in the Delivery of Care: A Multi-Method Study”, investigates the effect of top-down control versus bottom-up decision-makin (open full item for complete abstract)

Committee: Aravind Chandrasekaran PhD (Advisor); Peter Ward PhD (Advisor) Subjects: Health Care Management; Organizational Behavior

Doctor of Philosophy, The Ohio State University, 2010, Economics

The first essay, “Pricing Housing Market Returns,” finds the housing premium to be smaller than the equity premium. Using state-level data that spans the 1983 to 2006 period, I estimate the asset pricing Euler equations from the intertemporal consumption problem faced by a representative consumer with Epstein-Zin (EZ) preferences. The EZ Capital Asset Pricing Model captures a large proportion of the variation in housing returns over the sample period, and I find there to be heterogeneity in the structural parameter estimates across geographies. Controlling for the risk priced by the model and the consumption value of housing, I find that the housing premium is smaller than the equity premium. This result is surprising given that frictions, such as high transaction costs and borrowing constraints, affect the investor in housing more than the investor in equities. I examine institutional differences between the asset classes and find that some of the difference between the two premia may be related to differences in the tax treatment between the two asset classes. The second essay, “Non-durable Consumption Volatility and Illiquid Assets,” finds that factors beyond the volatility of asset payoffs may significantly affect the volatility of the agent's consumption stream. The empirical failure of consumption-based asset pricing models is often attributed to the lack of volatility in aggregate measures of consumption. However, I illustrate in this paper that frictions faced by agents may lead to much higher levels of volatility in individual consumption than we observe in the aggregate data. I develop a life-cycle model of in which the consumer derives utility from non-durable consumption and stock in a risky asset: housing. Non-convex adjustment costs generate lumpy changes in the stock of the risky asset over the life-cycle. The model predicts that non-durable consumption volatility is increasing in both the ability to borrow against the assets held in the consumer's (open full item for complete abstract)

Committee: Pok-sang Lam PhD (Committee Chair); Donald Haurin PhD (Committee Member); Mario Miranda PhD (Committee Member) Subjects: Economics; Finance

Doctor of Philosophy, The Ohio State University, 2007, Economics

The ancient joke in real estate is that the three most important criteria for selecting a house are location, location, and location. This explains the great emphasis of a household on residential location choice when he/she is buying a home. Driven by households' demand on location, it should also play an important role in determining house prices. As a key determinant in household consumption behavior, neighborhood effects are worth investigation. This dissertation examines neighborhood effects in the housing market using spatial econometric methods. The first essay studies the importance of social interactions in a household's location decision. I argue that individuals prefer interacting with others who have similar socioeconomic backgrounds. This hypothesis suggests that a household desires to find a good community match. An unwritten rule in real estate is that one should buy the cheapest house in an expensive neighborhood, which is formally the Tiebout hypothesis that households search for fiscal surplus. Community matching implies households will prefer similarity, while the Tiebout hypothesis implies households will prefer neighborhoods with richer neighbors. I use a nested logit (NL) regression to analyze a household's residential decision within Franklin County, OH. The results support the hypothesis that a household prefers neighbors with like socioeconomic characteristics in almost all of the similarity dimensions and only prefers an affluent neighborhood to a moderate degree. The second essay employs a spatial autoregressive model (SAR) to estimate housing asset prices. Applying the rational expectations hypothesis, this essay models the current value of a housing unit as the conditional expectation of the discounted stream of housing services accruing to the owner of the house. Based on the importance of location, the value of housing services is determined by neighborhood effects as well as the physical attributes of the property itself. In the exist (open full item for complete abstract)

Committee: Donald Haurin (Advisor) Subjects:

Doctor of Philosophy, The Ohio State University, 2006, Economics

As an effective method in analyzing interdependence among the observations, the spatial autoregressive (SAR) models have witnessed ever-increasing applications. This dissertation intends to enrich both the spatial econometrics theory and the social interaction estimations. In the first essay, a SAR model with group unobservables is applied to analyze peer effects in student academic achievement. Unlike the linear-in-means model in Manski (1993), the SAR model can identify both endogenous and contextual social effects due to variations in the peer measurements, thus resolving the “reflection problem”. The group fixed effects term captures the confounding effects of the common variables faced by the same group members. I use datasets from the National Longitudinal Study of Adolescent Health (Add Health) survey and specify peer groups as friendship networks. I find evidence for both endogenous and contextual effects, even after controlling for school-grade fixed effects. The result indicates that students benefit from the presence of high quality peers, and that associating with peers living with both parents helps improve a student's GPA, while associating with peers whose mothers receive welfare has a negative effect. The second essay considers the GMM estimation of SAR models with unknown heteroskedasticity. We show that MLE is inconsistent whereas GMM estimators obtained from certain moment conditions are robust. Asymptotically valid inferences can be drawn from the consistent covariance matrix estimator. And efficiency can be improved by constructing the optimal weighted GMM estimation. We also propose some general tests for heteroskedasticity. In the Monte Carlo study, 2SLS estimators have large variances and biases in finite samples for cases where regressors do not have strong effects. The robust GMM estimator has desirable properties while the biases associated with MLE and non-robust GMM estimator may remain in large sample, especially, for the spatial effect (open full item for complete abstract)

Committee: Lung-fei Lee (Advisor) Subjects:

Doctor of Philosophy, The Ohio State University, 2013, Economics

This dissertation consists of three chapters covering the following topics in spatial econometrics: estimation, specification and the bootstrap. In Chapter 1, we first generalize an approximate measure of spatial dependence, the APLE statistic in Li et al (2007), to a spatial Durbin (SD) model. This generalized APLE takes into account exogenous variables directly and can be used to detect spatial dependence originating from either a spatial autoregressive (SAR), spatial error (SE) or SD process. However, that measure is not consistent. Secondly, by examining carefully the first order condition of the concentrated log likelihood of the SD (or SAR) model, whose first order approximation generates the APLE, we construct a moment equation quadratic in the autoregressive parameter that generalizes an original estimation approach in Ord (1975) and yields a closed-form consistent root estimator of the autoregressive parameter. With a specific moment equation constructed from an initial consistent estimator, the root estimator can be as efficient as the MLE under normality. Furthermore, when there is unknown heteroskedasticity in the disturbances, we derive a modified APLE and a root estimator which can be robust to unknown heteroskedasticity. The root estimators are computationally much simpler than the quasi-maximum likelihood estimators. In Chapter 2, we consider the Cox-type tests of non-nested hypotheses for spatial autoregressive (SAR) models with SAR disturbances. We formally derive the asymptotic distributions of the test statistics. In contrast to regression models, we show that the Cox-type and J-type tests for non-nested hypotheses in the framework of SAR models are not asymptotically equivalent under the null hypothesis. The Cox test in non-spatial setting has been found often to have large size distortion, which can be removed by the bootstrap. Cox-type tests for SAR models with SAR disturbances may also have large size distortion. We show that the boots (open full item for complete abstract)

Committee: Lung-fei Lee (Advisor); Stephen Cosslett (Committee Member); Robert de Jong (Committee Member) Subjects: Economics