PhD, University of Cincinnati, 2017, Business: Business Administration

This dissertation contains three essays examining new Bayesian econometric methodologies. The first develops a heterogeneous Spatial Autoregressive Model by integrating a finite mixture model structure into the traditional homogeneous specification. The second essay builds upon the first by extending this Spatial Mixture Model structure to the more general Spatial Durbin and Spatial Durbin Error specifications. Additionally, this essay covers the interpretation of these new model specifications. Finally, the third essay develops a predictive based model selection process by integrating cross-validation algorithms into standard Bayesian sampling methods with a focus on explicit out-of-sample prediction. 1.0.1 Embracing Heterogeneity: The Spatial Autoregressive Mixture Model In this essay, a mixture distribution model is extended to include spatial dependence of the autoregressive type. The resulting model nests both spatial heterogeneity and spatial dependence as special cases. A data generation process is outlined that incorporates both a finite mixture of normal distributions and spatial dependence. Whether group assignment is completely random by nature or displays some locational "pattern", the proposed spatial-mix estimation procedure is always able to recover the true parameters. As an illustration, a basic hedonic price model is investigated that includes sub-groups of data with heterogeneous coefficients in addition to spatially clustered elements. 1.0.2 Spatial Durbin Mixture Models This essay extends the finite mixture model structure to include Spatial Durbin and Spatial Durbin Error model specifications. The partial derivatives of this heterogeneous spatial model structure are shown to differ between border and interior agents; the designation of which is based on group assignment and first order neighbor designation. As an illustration, individual income based on data from the Panel Study of Income Dynamics (PSID) is examined using the Spatial (open full item for complete abstract)

Committee: Olivier Parent Ph.D. (Committee Chair); David Curry Ph.D. (Committee Member); James LeSage Ph.D. (Committee Member); Jeffrey Mills Ph.D. (Committee Member) Subjects: Economics

Doctor of Philosophy, University of Toledo, 2023, Spatially Integrated Social Science

The COVID-19 pandemic ranks as the deadliest on the list of disasters in the United States. After community transmission was confirmed in early 2020, federal and state governments introduced varying restrictions on public mobility through various Non-Pharmaceutical interventions (NPIs) in response to the growing pandemic. Through a spatial and temporal perspective, this study aims to investigate the impact of changing mobility patterns, NPI measures, and local weather on the spread of COVID-19 during the early years of the pandemic at the county level. The primary goals of the dissertation were 1) identify the influence of different mobility categories, policy indices, and county-specific weekly average temperature data on COVID-19 case positive rate between 2020-2020 by employing spatial analysis and econometric modeling. 2) leverage the identified spatiotemporal relationship to develop a spatial panel data weekly predictive model for COVID-19 case positivity at the county level and publish an ArcGIS online dashboard. Considering the diverse nature of mobility and NPIs, this study incorporates five different mobility categories and two different measures of policy stringency index and county-specific weekly average temperature data. From 2020 onwards, twelve pandemic phases were progressively evaluated for 104 weeks using Spatial-Autoregressive & Spatially Autocorrelated Errors Fixed Effect Panel Data Models for 2380 counties. The spatial spillover effects on how each county was influenced by its neighbors were also evaluated. Results revealed a positive correlation between all the outdoor mobility categories and COVID-19 case positivity with varying levels of confidence at different times during the pandemic, except for parks and recreational visits, which demonstrated a negative correlation. Policy indices of different containment measures and economic supports exhibited negative correlations, indicating the association between lower policy index value and highe (open full item for complete abstract)

Committee: Kevin Czajkowski (Committee Chair); Bhuiyan Alam (Committee Member); Sujata Shetty (Committee Member); Barbara Saltzman (Committee Member); April Ames (Committee Member); Yanqing Xu (Committee Member) Subjects: Epidemiology; Geographic Information Science; Geography; Public Health; Statistics

Doctor of Philosophy, University of Toledo, 2018, Spatially Integrated Social Science

This dissertation is concerned with the spatial dynamics of the U.S. economy. Spatial dynamics is a termcoined in this dissertation to define the geo-spatial aspects of an observed natural process, particularly changes in its spatial relations over time. Geographic inquiry considering spatial dynamics requires an unassuming examination of spatial panel data, an approach that facilitates the discovery of new regularities and tendencies in spatial data and necessitates the development of more flexible tools and methods tailored to the peculiarities of the observed natural process. This dissertation demonstrates the practicality of spatial dynamics as a promising framework with the discovery, description, and analysis of two spatial economic paradoxes, which impelled the creation of several new tools and methods. The dissertation is composed of three essays linked by the exploration and analysis of the spatial dynamics of the U.S. economy, specifically its metropolitan statistical areas (MSAs). The first essay develops two new statistics that quantify physical and human capital accumulation in MSAs. These statistics are used to calculate the classical production function and derive the percent contribution of physical and human capital to average establishment size and Gross Domestic Product by MSA (MGDP). The results conformtomacroeconomic expectations and are spatially distributed according to the familiar economic geography of the United States, rendering the statistics usefulfor spatial economic analysis. The second essay explores the observation that MGDP growth rates are spatially clustered and MGDP levels are uniformly distributed (i.e., exhibit no spatial correlation). This finding is paradoxical because the level of economic activity is the aggregation of previous growth patterns and, if economic growth in the spatial economy is persistently clustered, the location of economic activity should follow the same pattern. The essay seeks (open full item for complete abstract)

Committee: Oleg Smirnov Dr. (Committee Chair); Olugbenga Ajilore Dr. (Committee Member); Peter S. Lindquist Dr. (Committee Member); David J. Nemeth Dr. (Committee Member); Neil Reid Dr. (Committee Member) Subjects: Economics; Geography; Regional Studies; Statistics

Doctor of Philosophy, The Ohio State University, 2010, City and Regional Planning

This dissertation presents five essays dealing with the utilization and abandonment of a city's assets, in particular two key assets: agglomeration economies and legacy capital. The first essay traces out the causes and effects of agglomeration economies by disentangling economies of agglomeration. It disentangles amenity and productivity effects of agglomeration; it decomposes aggregate scale effects into agglomeration factors of interest to policy makers; and it estimates own effects and spillovers to neighbors. It proposes a spatial simultaneous equations model in a spatial equilibrium framework with three agents – worker consumers and producers of traded goods and housing. Results for Ohio counties estimate economies resulting from population size, agglomeration causes, and public service quality and cost on each of the three agents in own and neighboring counties. The second essay theoretically models the abandonment and reuse of legacy capital in the process of industrial restructuring. It aims to identify the conditions for abandonment and the factors that determine the length of abandonment. The model is based on investment theory and game theory. It shows that abandonment is impacted by conversion costs of legacy capital, the rate of growth of industries involved in the restructuring, and policy variables such as tax rate. The third essay empirically verifies the theoretical model developed in the second essay, using data of industrial and commercial properties (ICPs) in the Cleveland city-region in Ohio. It shows that in declining industries or regions, ICPs experience tax delinquency of longer duration and are more likely to be abandoned than elsewhere. Also, ICPs with higher conversion costs are more likely to experience longer spells of tax delinquency and are more likely to be abandoned than others. Abandoned ICPs are spatially concentrated either as a result of negative spillovers or shared history. The fourth essay theoretically models the extern (open full item for complete abstract)

Committee: Burkhard von Rabenau (Committee Chair); Jean-Michel Guldmann (Committee Member); Philip Viton (Committee Member) Subjects: Economics; Urban Planning

Doctor of Philosophy, University of Toledo, 2021, Spatially Integrated Social Science

This dissertation bridges the gap between spatial econometrics and machine learning under the theoretical banner of spatial data science. Methodologically, it uses the spatial error model, spatial lag model, and the randomForest algorithm in order to predict Human Development Index (HDI) values within Morocco at the commune scale. This prediction task is done using the Moroccan censuses of 2004 and 2014. The results of this process show that randomForest can outperform the traditional spatial econometric models in terms of numeric accuracy within this specific case. Since spatial thinkers are just as concerned with spatial accuracy as they are with numeric accuracy, post-estimation procedures were developed in order to assess the spatial accuracy of the spatial error model, spatial lag model, and randomForest in the Moroccan case. These post-estimation procedures were developed for both the global and local levels. In both cases, it is shown that randomForest outperforms both of the spatial econometric models in terms of spatial accuracy within the Moroccan case. With the Morocco specific results complete, the dissertation moves to simulated data experiments in order to assess different properties of randomForest vs. the spatial lag model, and randomForest vs. the spatial error model. The simulation experiments are carried out using five different data generation processes. Throughout the experiments bias, consistency, efficiency, and spatial prediction performance are evaluated and compared. These experiments show that when either the spatial lag model or spatial error model are the correct model specification, randomForest is unable to outperform either of them in terms of bias, consistency, efficiency, or spatial prediction performance. Therefore, it is concluded that if randomForest does outperform the traditional spatial econometric models, as happened in the Moroccan case, neither the spatial lag model nor the spatial error model are the correct m (open full item for complete abstract)

Committee: Oleg Smirnov Dr. (Advisor); Neil Reid Dr. (Committee Member); Sujata Shetty Dr. (Committee Member); David Nemeth Dr. (Committee Member); Jack Kalpakian Dr. (Committee Member) Subjects: Geographic Information Science; Geography

Master of Arts, University of Toledo, 2019, Economics

A 5.2% assessed value premium is estimated for homes with an unobstructed view of a natural meadow floodplain in an affluent residential community in Northwest Ohio.

Committee: Kevin Egan PhD (Committee Chair); Kristen Keith PhD (Committee Member); Yanqing Xu PhD (Committee Member) Subjects: Economics; Environmental Economics

Master of Arts, University of Toledo, 2017, Economics

Little is known about the interaction of spatial dependent models and random decision forests. Most previous research has not implemented modern machine learning techinques with economics let alone spatial econometrics. In this paper we apply random forest analysis with a spatial dependent component to hedonic pricing models. This paper sought to improve parameter identification, prediction performance, and bridge the gap between spatial economics and machine learning. The data provided details 45,381 residential real estate sales in Lucas County, Ohio between 2001-2016. Evaluation by log-linear and spatial log-linear models shows that random forests can make comparatively accurate model predictions using less indicators than models selected by conventional methods. While the spatially dependent random forest models did not produce the lowest root mean square error compared to the spatially dependent models, reducing the number of parameters by 35\% only marginally increased error compared to other models. The results have implications for improving understanding of components used real estate appraisal as well as construction or investment.

Committee: Oleg Smirnov (Committee Chair); Aliaksandr Amialchuk (Committee Member); Kristen Keith (Committee Member) Subjects: Economics

Doctor of Philosophy, The Ohio State University, 2016, Economics

Spatial econometrics has been obtained more and more attention in the recent years. The spatial autoregressive (SAR) model is one of the most widely used and studied models in spatial econometrics. So far, most studies have been focused on linear SAR models. However, some types of spatial or network data, for example, censored data or discrete choice data, are very common and useful, but not suitable to study by a linear SAR model. That is why I study an SAR Tobit model and an SAR binary choice model in this dissertation. Chapter 1 studies a Tobit model with spatial autoregressive interactions. We consider the maximum likelihood estimation (MLE) for this model and analyze asymptotic properties of the estimator based on the spatial near-epoch dependence (NED) of the dependent variable process generated from the model structure. We show that the MLE is consistent and asymptotically normally distributed. Monte Carlo experiments are performed to verify finite sample properties of the estimator. Chapter 2 extends the MLE estimation of the SAR Tobit model studied in Chapter 1 to distribution-free estimation. We examine the sieve MLE of the model, where the disturbances are i.i.d. with an unknown distribution. This model can be applied to spatial econometrics and social networks when data are censored. We show that related variables are spatial NED. An important contribution of this chapter is that I develop some exponential inequalities for spatial NED random fields, which are also useful in other semiparametric studies when spatial correlation exists. With these inequalities, we establish the consistency of the estimator. Asymptotic distributions of structural parameters of the model are derived from a functional central limit theorem and projection. Simulations show that the sieve MLE can improve the finite sample performance upon misspecified normal MLEs, in terms of reduction in the bias and standard deviation. As an empirical application, we examine the school (open full item for complete abstract)

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

Master of Arts, University of Toledo, 2005, Economics

We examine the spatial autoregressive relationship between county-level voting outcomes in the 2000 Presidential election and a host of candidate explanatory variables taken from the year 2000 census. These include: measures of past voting behavior, indicators of socioeconomic demographic status of the population, and economic variables that reflect recent economic conditions. Using a recently developed spatial econometric extension of least-squares regression-based Markov Chain Monte Carlo model composition methodology (often labelled MC3) by LeSage and Parent(2004), we present evidence on which explanatory variables are important in explaining voting outcomes. The LeSage and Parent (2004) methodology deals with cases where the number of possible models based on different combinations of candidate explanatory variables is large enough that calculation of posterior probabilities for all models is difficult or infeasible. In addition, we produce estimates using a spatial autoregressive seemingly unrelated regression methodology developed in LeSage and Pace (2005), that takes into account cross-equation error covariance between the Bush and Gore equations in the model.

Committee: James LeSage (Advisor) Subjects:

Master of Arts, University of Toledo, 2004, Economics

We examine the relationship between fourth grade student proficiency scores, various educational spending categories, student and teacher characteristics as well as population socioeconomic characteristics using a building-level database containing 1,965 Ohio elementary schools. While most economic studies of the relationship between student performance and educational resources show a weak link, we find a strong link between student performance and resources. We argue that use of the appropriate spatial scale is an overlooked issue in previous economic studies. Furthermore, we provide an alternative method of capturing variation over geographical space through the estimation of locally linear models. By estimating a separate model for each individual building, we see that the global estimates provided by previous estimation methodologies may not be provide valid inferences. The locally linear estimates and inferences suggest that location-special forces exist in the relationship between resources and student performance.

Committee: James LeSage (Advisor) Subjects:

Doctor of Philosophy, The Ohio State University, 2010, Environmental Science

Communities living in close proximity to livestock operations may be subjected to economic externalities (third party or spill over effects) like pollution or odors originating from these facilities. Direct environmental contamination is of considerable concern since waste is frequently stored in stacks or pits and may leach or spill into the surrounding environment. There is also the danger that waste may be transported even further across the landscape via runoff. But adequate information about the social costs associated with livestock operations is scarce. Addressing this informational gap constitutes the first objective of this work. Here, economic modeling combined with GIS (Geographic Information Systems) based mapping has been used to analyze the impact of animal agriculture on the environment and local communities. Several hedonic pricing models were developed for three selected counties in the state of Ohio across two different time periods (2000 to 2001 and 2003 to 2004). Hedonic regressions explain the value of a house in terms of its characteristics; therefore they can be used to ascertain whether or not an environmental variable has a statistically significant impact on housing values. Because of the spatial nature of the data a neighborhood sampling method as well as explicit spatial modeling was used to derive value estimates. The impacts estimated by both models in terms of prices were quite similar thereby increasing confidence in the robust nature of the results. Marginal prices of $0.51 to $0.57 per foot revealed the impact of dairies to be negative and highly significant. These results indicate that property values may decrease by 2% per mile (on average) as proximity to a dairy operation increases, meaning that dairies function as disamenities. The second stage of the estimation revealed that decreasing these impacts by 10% to 25% (by employing anaerobic digestion for example) would result in welfare gains of over $500 to $1,100 per househol (open full item for complete abstract)

Committee: Fred Hitzhusen (Committee Chair); Floyd Schanbacher (Committee Member); Elena Irwin (Committee Member); Brent Sohngen (Committee Member) Subjects: Environmental Science

Doctor of Philosophy, The Ohio State University, 2008, Agricultural, Environmental and Development Economics

This dissertation analyzes whether inequality in the distribution of income influences real per capita GDP growth in the provinces of Argentina, while taking into account spatial autocorrelation. The primary objective is to decouple this influence into within effects, the inequality from the own province, and spillover effects, the inequality from the neighboring provinces. These influences are examined in the long and short run.The dissertation identifies significant clustering, with relatively high (low) income provinces located next to high (low) income provinces. The coefficient of variation of provincial per capita GDP and Moran I statistic for spatial autocorrelation suggest an interruption of the process of convergence among provinces during the 1990s, when rapid growth was followed by a major crisis and increasing inequality. Weaker clusters are identified for inequality. Following Partridge (2005), the dissertation first considers parsimonious models with a few key variables. Next, it adds several control variables, to get more fully specified base models and to explore spillover effects. To control for spatial autocorrelation, spatial lag and spatial error model specifications are used. Pooled OLS models with a fixed time effect and fixed- and random-effects panel specifications confirm the robustness of the results. Income inequality might have a separate effect at the middle versus the tails of the distribution in the regressions, the provincial Gini coefficient controls for the overall distribution, especially at the tails, while the share of the third quintile controls for a potential middle-class consensus and role of the median voter. The dissertation finds robust evidence that both the own province inequality and inequality in the neighboring provinces negatively affect per capita GDP growth in the provinces of Argentina for the 1991-2002 period. There is also robust evidence that the share of the third quintile negatively affects growth. This resu (open full item for complete abstract)

Committee: Claudio Gonzalez-Vega PhD (Advisor); Mark Partridge PhD (Committee Member); Joseph Kaboski PhD (Committee Member) Subjects: Economics

Doctor of Philosophy, The Ohio State University, 2007, Agricultural, Environmental and Development Economics

The main objective of this research is to reveal the effects of water quality on housing values around Lake Erie. Both the first and the second stage of hedonic price analysis are conducted with identified housing submarkets by using Hierarchical Clustering with quantized similarity measures in the region including Erie, Lorain, Ottawa and Sandusky Counties located along Lake Erie. We use both individual houses and census block groups as the smallest building blocks of the clusters and compare the clustering and hedonic results for both cases. Fecal coliform counts and secchi disk depth readings measuring water clarity are used as water quality variables. In order to overcome the spatio-temporal aspects of secchi depth disk reading data, kriging is used for spatial prediction. Robust Lagrange Multiplier test indicates that spatial error models are appropriate for the estimation of hedonic price functions in each submarket. We found that secchi disk depth readings variables are positive significantly influencing housing prices in most of the clusters while mixed results are found for fecal coliform counts. Demand functions with different functional forms are estimated with two-stage least squares with submarket dummy variables. While computed welfare changes for fecal coliform by using non-linear demand functions are very small, the benefit of the improvement of water clarity by 25 centimeters to be estimated 230 dollars per household. We found that the welfare changes are larger for the degradation of water quality compared to the improvements of water quality in the same amount. We further analyzed the welfare changes by using demand functions derived specifically for each household. Welfare changes based on the individual demand functions were computed by integrating under each demand curve for multiple scenarios. If we consider our SIG Fecal data represents 33 percent of entire population in four counties, the total estimated net benefit was derived as 51,934,180 (open full item for complete abstract)

Committee: Timothy Haab (Advisor) Subjects: Economics, General

Doctor of Philosophy in Urban Studies and Public Affairs, Cleveland State University, 2012, Maxine Goodman Levin College of Urban Affairs

This dissertation presents estimates of the relationship between early investment in broadband infrastructure and a number of local economic indicators using a data set of communities (by zip code) across the U.S. Data is matched from the FCC (Form 477) on broadband infrastructure availability with demographic and other socio-economic data from the U.S. Population Censuses and Business Trends Surveys. Spatial econometric techniques are utilized. Even after controlling for community-level factors known to influence broadband availability and economic activity, it was found that between 1998 and 2008, communities in which broadband was available by 1999, compared to those that did not, experienced a greater difference in the growth of 1) rents, 2) salaries, 3) employment, and 4) overall establishments. In addition, broadband contributed to the share of different industry structures lending support to the GPT hypothesis. This research replicates and extends Lehr et al. (2005).

Committee: Robert Scherer (Committee Chair); Joel Elvery (Committee Member); Haifeng Qian (Committee Member) Subjects: Public Administration

Doctor of Philosophy, The Ohio State University, 2013, City and Regional Planning

Several powerful natural disasters have recently occurred in different countries around the world. Mitigating the impacts of these disasters has been an important component of urban/land-use planning (Burby et al., 1999). Seismic hazard and urban vulnerability are the two major factors in the assessment of seismic risk. The Peak Ground Acceleration (PGA) is the principal measure of seismic hazard, the Peak Ground Displacement (PGD) represents a minor one, and land-use patterns characterize urban vulnerability. A spatial statistical approach is used to better understand the spatial interactions of seismic hazards and their relationships with land-use patterns. In seismic engineering, spatial autocorrelation (SA) has been rarely discussed, but may cause biased estimations due to omitted variables (Esposito and Iervolino, 2011). A LOSS model is also estimated, that links the economic losses incurred by various activities (housing, commerce, industry, etc.) to PGA and PGD. Next, in order to deal with the uncertainty of earthquakes, an optimization approach has been developed, using the LOSS statistical model, to allocate future land uses. A set of pseudo-data is generated with an earthquake-engineering simulation model for the city of Taichung, Taiwan, using as input the characteristics of 22 significant historical earthquakes. A PGA model is first formulated as a spatial lag panel (SLP) model based on earthquake magnitude, epicenter-to-site distances, source depth, and neighborhood effects accounting for site geology. A PGD SLP model is also specified to account for neighborhood effects represented by soil liquefaction. The results demonstrate the need to be concerned not only by the PGA/PGD at a specific location, as caused directly by the earthquake, but also by the PGA/PGD at neighboring locations. Then, a seismic loss model is formulated, relating monetary damages to seismic impacts (PGA and PGD) and land uses (residential, commercial, industrial, etc.). By combin (open full item for complete abstract)

Committee: Jean-Michel Guldmann (Advisor); Philip Viton (Committee Member); Gulsah Akar (Committee Member) Subjects: Urban Planning