ABSTRACT. Moriarty Graduate Paper Competition

Residential location indirectly affects many other vital economic outcomes, such as school quality and access to employment opportunities. A theoretical tool to better understand the occurrences of residential segregation is the checkerboard model first developed by Thomas Schelling. This model predicts that even when the majority group is not exhibiting explicitly discriminatory behavior, residential segregation still occurs. As a result of residential segregation, the minority group might then be subject to lower quality economic outcomes related to their residential location. This paper will use a simple set of modifications to the general theoretical model to motivate the empirical estimation strategy undertaken to illustrate homophilic preferences, using an exogenous demographic shock, the opening of a new meatpacking plant. A difference-in-differences estimation technique will be utilized to show that the creation of a meatpacking plant caused an influx of low-skilled Hispanic immigrants into previously homogeneous rural communities, which then results in an outflow of white residents from their current neighborhood and into proximate neighborhoods. Alternative hypotheses that housing prices are driving this outflow will be ruled out, and sensitivity analysis will be undertaken on manufacturing industries that experienced similar growth during this period, illustrating that these results are unique to the meatpacking industry. All of these empirical results align with the theoretical predictions developed from the modified model. Given that the case being examined in this study is a subset of high growth Hispanic communities, the results of this paper have timely implications, both for academic researchers and policymakers.

ABSTRACT. Moriarty Graduate Paper Competition Since the advent of the Tennessee Valley Authority (TVA) and the Appalachian Regional Commission (ARC), there has been substantial amount of government funding that has gone towards regional commissions. Despite this trend, little is known about the economic benets of the smaller programs. In this paper, I attempt to expand the literature on regional commissions by analyzing the economic gains to the Delta Regional Authority (DRA). The DRA was founded in 2000 to provide government resources to counties in the Lower Mississippi Valley in order to improve economic conditions within the eligible 252 counties. The research period from 1997 to 2016 is used to measure the benets over time. My focus is on the impact on employment, income, migration, and poverty measures collected from various government sources at the county level. One-to-one propensity score matching is used to generate a counterfactual. Due to the endogenous nature of treatment, I instrument at the county level of being included in the DRA with a dummy of whether the county is within the Lower Mississippi Watershed or not. With these corrections, my results should be an accurate estimation of the intent to treat benets of the DRA. My research finds modest gains in income and decreases in unemployment; however, there is no measurable impact to poverty nor migration.

ABSTRACT. The United States experienced a massive and unexpected influx of Cuban refugees in 1980; called Mariel Boatlift. This paper re-visits Mariel Boatlift to see the impact of the influx of refugees. According to the “Public Law 1980”, the US government provided financial assistance to Marielito for their stable resettlement in a new destination. This paper contributes on Mariel Boatlift literature, considering Marielito as the refugee, which has been ignored by previous Mariel Boatlift economic studies. Also, this paper utilizes a new econometric model, Synthetic Control Method (SCM), to deal with concerning of the previous study on violation of pre-trend parallel assumption for Difference-in-Difference estimation and lack of yearly trend. My SCM results show that Mariel Boatlift sharply increased the proportion of Hispanic in Miami and resulted in a corresponding decrease in White ratio. The primary results of this paper finds rents increase in Miami in line with the previous study. Moreover, this paper provides more clear rents evolution during subsequent years and higher rent increase among better quality units which is different from the results found in the previous study. SCM results on welfare income versus social security income and the evolution of White ratio provide suggestive evidence on the mechanism of government assistance to refugees and the internal migration of white to avoid the refugee shock.

ABSTRACT. Over time, the phenomenon of generating waste by society has increased. It has caused an increase in number of locations where the waste can be disposed of and managed. For residents living nearby, waste disposal in landfills gives rise to nuisances like odor and unattractiveness while potentially causing health concerns like toxic water run-off and methane gas emissions. Reichert et al. (1991) indicated in their study that the residents interpret odor from the landfill as a signal of potential health hazards. Along with odor, unattractiveness, blowing trash, truck noise, toxic water run-off, methane gas were also found as significant concerns among the respondents of their study’s survey. Among all these nuisances, bad odor coming from the landfills is something that people are believed to notice quickly about an area. Again, Resistance to an unwanted facility among residents of a given area increases as the distance to the proposed facility location and the residents’ homes gets smaller (Dear et al. ,1980). The negative features of areas near landfills could decrease property values, making landfills subject to the “not in my backyard” or NIMBY phenomenon. Our study acknowledges the importance of this ongoing issue. This paper examines how landfill sites affect nearby property values in the USA. In order to investigate the causality of the impacts of landfill sites on property values we assume that closed and open landfills have different impacts on peoples’ attitude towards surrounding areas. To establish this statement, we compare the property values before and after the landfill closure using a Difference-in-Differences (DID) model. In this paper, we consider the landfills that got closed during 2004, 2005 and 2006 in USA using data from EPA (Environmental Protection Agency). In our identification strategy, instead of using a certain radius, we use the wind direction, as a major atmospheric information, around the landfill sites to determine our location of interest. This is important as it gives us a better indicator of the direction of bad odor and other toxic materials to be carried by wind. The data for constructing this variable comes from the Remote Sensing Systems Cross-Calibrated Multi-Platform (CCMP) vector wind analysis product (Version 2.0). For the property value data, Zillow through the Zillow Transaction and Assessment Data (ZTRAX) provides not only transaction prices, structural and locational characteristics, but also specific addresses of individual properties. We use the data for different counties of the USA from 1996 to 2017. With the help of GIS software, we determine the exact location using the addresses of the properties and the landfill sites and the distances between them. To estimate the impacts of landfill sites on property value in the framework of a hedonic pricing model, the main concern is biased results because of model specification. The factors that explain variation in property prices could be unobservable. Again, simple regression models are not capable of considering the potential endogeneity problem and so the estimated coefficients might not reflect the exact causal relationship between the environmental attribute and the property prices. Situations where an experiment with randomization cannot be conducted to do an impact evaluation, quasi-experimental methods can be used to address the existing causality. Therefore, we propose to use a Difference-in-Differences (DID) model in this research.

ABSTRACT. Since China's reform and opening-up, with the rapid development of economy, the interregional development gap is expanding, which is mainly caused by the aggregation of economic activities. This study uses the Theil index to measure the degree of aggregation of economic activities from four aspects as economy, population, employment and life, aiming at analyzing the regional economic differences in China comprehensively.

ABSTRACT. Recently, the issue of growing income inequality among individuals has gained considerable attention in the media, in academia, and in government. In the U.S., the share of income received by the top 10% increased from 25.3 percent in 1980 to 30.6 percent in 2016. At the same time, the share of income of the top 1 percent increased from 8 percent of total income to over 22 percent.

At the regional level the literature on state PCPI has focused on the question of PCPI convergence over time. Papers by Barro and Sala-i-Martin (1992), Levernier, Partridge, and Rickman (1995), and Bernat (2001) established the convergence of PCPI among the states through 1990. Since these papers were published there appears to have been a change in the PCPI trend among states. In recent years a number of studies have indicated that the gap between higher income groups and lower income groups has been widening.

This paper looks at these two issues from a local perspective. We analyze the trends in U.S. MSA PCPI from 1969 to 2017 and investigate the period on convergence (through the 1990s) and the period of divergence (the recent years). In addition, we address the question of income inequality and the recent trend of MSA PCPI divergence.

John E. Connaughton, UNC Charlotte Caroline Swartz, UNC Charlotte

ABSTRACT. Since 1980, life expectancy in the United States has increased by roughly 5 years, and by 2016 the average American was expected to live 78.6 years at birth. At the same time, spatial inequality also grew, such that the standard deviation of county-level life expectancy increased from 1.8 to 2.4 years. To explore reasons for this trend, some researchers have focused on morbidity factors such as lack of access to health care and increasing rates of obesity or opioids overdoes, while other researchers focused on how mortality trends differ by characteristics such as race, age, and income. However, the effect community characteristics may play in this growing spatial inequality has not yet been explored.

Current maps of county-level life expectancy suggest the presence of high and low life expectancy clusters, indicating there is likely spatial autocorrelation in the data, which may also be linked to community-level factors. Therefore, our research question is: which community factors affect county-level life expectancy at birth, particularly the increasing variation across counties? Using a cross-sectional dataset of the contiguous United States, we develop a spatial Durbin error model to explore how lagged personal and community factors influence life expectancy in 2014, analyzing men and women separately. In addition to spatial econometrics, we use spatial statistics to detect the presence of general clustering as well as any regional hot and cold spots. We hypothesize that community and social interaction factors will affect county-level life expectancy.

ABSTRACT. As the opioid epidemic has expanded across the U.S., states have intervened by enacting legislation and establishing programs aimed at curtailing the crisis. As of 2018, 49 of the 50 states have devised prescription drug monitoring programs (PDMPs) in an effort to thwart prescription opioid abuse. PDMPs are statewide electronic databases that store information on prescriptions for controlled substances, with data coming from the dispensing pharmacies. The information included in the state PDMPs is useful in identifying patients who are misusing prescription opioids or “doctor shopping,” and in recognizing doctors who are overprescribing opioid medications. Some have voiced concern, however, that such programs may encourage individuals who cannot obtain prescription opioids to turn to illegal alternatives such as heroin.

This paper expands upon a previous study (Pardo, 2017) which used a series of fixed effects models to examine how PDMP robustness affected the prescription opioid overdose death rate at the state level. To create the measure of PDMP strength, Pardo aggregated the number of state PDMP regulations, with specific weights assigned to different legislative components. As an alternative to that approach, this study uses multiple correspondence analysis (MCA) to generate the score of PDMP robustness. MCA is an extension of principal component analysis, and is used to identify latent variables of state PDMP strength by analyzing the same PDMP regulations assessed by Pardo. Incorporating MCA maximizes the correlations between the variables that went into the formation of Pardo’s score, and will provide another measure of the intensity of a state’s PDMP throughout the years.

An important feature of the current study is that it distinguishes between heroin overdose deaths and prescription opioid deaths as outcome variables. Additionally, we use a larger and more current sample of data. In his work, Pardo used a panel dataset that covered PDMP legislation, prescription overdose death rates, and state demographic variables from 1999 to 2014. Our panel dataset covers the years 1999 to 2016, which notably includes two additional years of data as the opioid crisis continued to grow. The rationale behind including heroin deaths as a dependent variable is the hypothesis that a more stringent PDMP may cause opioid misusers to shift from abusing prescription opioids to illicit substances such as heroin.

Two series of fixed effects models are conducted: one using our MCA-generated score as the measure of PDMP strength, and the other using Pardo’s score for comparative purposes. Continuous measures of PDMP vigor indicate that more state regulations are associated with higher levels of heroin-related deaths - thus supporting our hypothesis. When PDMP scores are broken down into quartiles as explanatory variables, there is a decreasing returns to scale effect. Findings show a statistically significant negative effect on prescription and heroin-related overdose deaths for the first (lowest) quartile of scores using both the MCA- and Pardo-generated measures of PDMP strength. However, higher quartiles of scores do not indicate a statistically significant impact on either heroin or prescription overdose deaths. These results indicate that states outside of the least stringent PDMP cohort may not see any impacts to their prescription or illicit opioid death rates.

ABSTRACT. To fight against the opioid-epidemic, several US states have introduced Prescription Drugs Monitoring Programs (PDMP). Except for the state of Missouri, all the US states have enacted voluntary PDMP, while few other states have enacted a so-called “Must Access” PDMP. Unlike voluntary PDMP, the “Must Access” PDMP states abide by the law to collect data on controlled substance prescriptions that doctors have written for patients. This allows authorized individuals to view a patient’s prescription history to facilitate detection of suspicious prescription and utilization behaviors. Academic discussions on the impact of “Must Access” PDMP have recently started, but these limited previous studies don’t converge to a consensus. This paper disentangles reasons why studies don’t have a convergence to a consensus and offer several new results. The results comprise comparison of the prescription opioid over-dosage death rates among “Must Access” PDMP states with voluntary PDMP states. This paper uses panel data (1999-2015) to implement double post LASSO method of Belloni, Chernozhukov, & Hansen (2013) to properly select on observable confounders and Generalized Synthetic Control Method of Xu (2017) to absorb the unobserved time-varying confounding effect to maintain the “parallel trend” assumption and develop the counterfactuals with confidence intervals. This paper identifies that the heroin/synthetic fentanyl epidemic and interstates highway effects (I-65, I-70, I-75, and I-80) are important unobserved factors that must be controlled for proper inference. Further, the average of observed prescription opioid over-dosage death rate is higher than that of the counterfactuals, suggesting that the PDMP policy may have unintended consequences to increase the death rates. Finally, I present the heterogeneous treatment effect of “Must Access” PDMP policy.

ABSTRACT. With the explosive rise in opioid overdose mortality in the United States, efforts to curb demand and supply of prescription and non-prescription opioids have been on the rise. One such program aimed at curbing mainly the supply-side of prescription drugs in states is the Prescription Drug Monitoring Program (PDMP). According to the Department of Justice, all U.S. states had fully operational PDMPs as of April 2017. Theoretically, PDMPs can have a positive or a negative relationship with overdose deaths. Impacts could be positive because PDMPs restrict doctor or pharmacy shopping, restricting the supply of prescription opioids in the market, hence the possibility of death by overdose. Impacts could, however, have a potential negative effect on overdose deaths due to the effects from users substituting to cheaper non-prescription drugs like heroin. While research aimed at assessing the effectiveness of PDMPs has begun, few have adequately addressed the heterogeneity in PDMPs, and none have accounted for spatial dependencies and spillover effects to and from neighboring states. The aim of this research is to remedy these gaps in existing research. We use the Spatial Durbin Model to quantify empirically the direct effects and indirect effects of PDMP on overdose deaths. The direct effects estimate the impact of PDMP on overdose deaths within a particular state, while indirect effects estimate the impact of PDMP on overdose deaths in neighboring states. The results indicate that the implementation of PDMPs in states have not helped to decrease the overdose death rates. Mandatory access to PDMPs shows an increase in illicit drug overdose death rates, prescription drugs overdose death rates, and the total drug overdose death rates.

ABSTRACT. TBA

ABSTRACT. TBA

ABSTRACT. TBA

ABSTRACT. TBA

ABSTRACT. Low-density polyethylene (LDPE) plastic is used extensively in agricultural production, including as bale and greenhouse covers, fumigation film, irrigation tubing, and mulch. The United States Department of Agriculture has identified single-use LDPE as a common agricultural aid with a negative impact on the environment. This research assesses the use of LDPE in agricultural production, the cost to recycle different forms of LDPE used in agriculture, as well as the potential for creating a market for recycled LPDE materials used in agriculture. Specifically, preliminary results will be presented to (1) provide an inventory of how LDPE is currently used in agriculture across crops and production practices; (2) estimate the amount of LDPE used per acre across farm types to compute total LDPE used across regions; (3) map the data to show where LDPE use in agriculture is concentrated; (4) identify the potential uses of recycled agricultural plastics, as well as barriers to collecting plastic from farms and preparing plastic for recycling; (5) analyze the viability of creating collection networks and potential markets for recycling LDPE used in agriculture; and (6) identify existing agricultural plastics recycling programs.

ABSTRACT. Ethiopia is a small open economy that has experienced rapid growth over the last decade. Between 2010 and 2017, GDP increased at an average 10.2 percent annually, while CO2 emissions grew at 12.1 percent. This growth is likely to be accompanied by a rise in emissions which can have significant effects on the region. Understanding the role that energy plays in its growth is important for Ethiopian decision makers in addressing issues related to energy security. In this study, various time series models are employed to investigate the long-run equilibrium and temporal dynamic relationships between energy and various macroeconomic variables, and CO2 emissions in Ethiopia. Preliminary findings from a VECM reveal long run causality running from energy, GDP and exports to CO2 emissions. Short run causality among the variables is absent. The error correction term indicates that CO2 emissions will converge towards long run equilibrium at a slow pace following shocks.

ABSTRACT. State-level electricity net metering policies, which generally allow distributed solar generators to sell excess electricity generation to utilities at full retail rates, have been shown to be an important factor promoting adoption of solar photovoltaic panels among US consumers (Darghouth, Barbose, and Wiser 2011, Matisoff and Johnson 2017). Existing literature treats net metering policies as uniform across states, typically accounting for the policies with a simple indicator variable if a state has a net metering law in place. However, in reality, net metering policies are quite heterogeneous between states, with significant variation in policy design. In this study I examine the effects on distributed solar PV adoption of net metering policy design in four areas: applicability to individual utilities, system-wide capacity caps, and reimbursement rates for direct net metered generation and net excess generation. I employ a unique dataset of net metering policy characteristics created by the author, combined with utility-level data derived from US Energy Information Administration sources. To my knowledge, this paper is the first to examine the differing effects of net metering policy design on solar adoption at the individual utility level.

ABSTRACT. Despite repeated calls for a thorough cleanup of water pollution in the Ganges river, there are only two papers by Batabyal and Beladi (2017, 2018) that have shed theoretical light on this cleanup problem and its connection to the sustainability of tourism in Varanasi. Therefore, in this paper, we extend the analyses in Batabyal and Beladi (2017, 2018) and concentrate on two specific questions. First, we introduce the notion of a safe minimum standard (SMS) into the analysis and then show how to construct and analyze a probabilistic model of the Ganges cleanup problem when the SMS is explicitly accounted for. Second, for a representative Varanasi citizen, we study how the magnitude of the elasticity of substitution between a composite consumption good and water quality in the Ganges---modeled by the SMS---affects the tradeoff between consumption and water quality maintenance.

ABSTRACT. To what extent data support the Middle Class Economics? What do we know about the relationship between the share of the middle class household in a region and the region’s growth rate? Unlike inequality - which has been the hotly debated subject among policymakers and researchers, the subject of the middle class lacks empirical data. One of the reasons behind a dearth of empirical research is because there is no consensus on how to define the middle class. This paper asks whether the argument of the Middle Class Economics is backed up by data: does the region with a larger share of the middle class experience higher growth rate? Some research using national-level data suggest that a larger middle class may provide a positive economic outcome (e.g., higher income, higher poverty reduction rates), researches on the effect of the size of the middle class at the sub-national level are rare. This paper tries to provide empirical data both on the people share and income share of the middle class household and to examine whether these variables affect the future economic performance of the region – measured by per capita gross metropolitan product (PCGMP) and annualized average growth rate of PCGMP. Using panel data from U.S. Census and Moody’s analytics, the author finds that the positive effect of the size of the middle class at the national and U.S. state level does not hold between 1980 and 2015 at the U.S. metropolitan statistical area level. While the relationship between the share of the middle class household and economic performance of the region cannot be confirmed, the relationship between the share of the upper class households in a region seems to have a positive effect on future growth. The result suggests that these relationships are sensitive to econometrical approach as well as different definitions of the middle class.

ABSTRACT. While metropolitan technology growth exerts a positive effect on nonmetropolitan development through knowledge spillovers, metropolitan technology also raises the competitive advantage of metropolitan firms over nonmetropolitan firms in product market competition-especially if there is a metropolitan bias in the production of innovation. The progress in metropolitan technology also affects the nonmetropolitan labor market through brain drain. Rural brain drain, which is often considered to be a negative effect, actually does have an ambiguous effect on nonmetropolitan labor markets. To test this effect of metropolitan technology, I employ county-level data from the contiguous United States. There is evidence that a 1 percent increase in metropolitan technological stock, which is constructed from the perpetual inventory method and the basic inverse distance matrix, is associated with 0.1-0.2 percent higher unemployment rate in a short or medium run. A back-of-the-envelope calculation translates these numbers into about two and a half million nonmetropolitan job losses between 2005 and 2015 by metropolitan technical progress. After examining alternative methods and hypotheses, I find that the benchmark finding is highly robust. I also perform a simple assessment of nonmetropolitan welfare and find that metropolitan technology has a statistically significant negative impact on average wages but not average incomes of the nonmetropolitan population. This crude assessment suggests that employed workers in nonmetropolitan regions might suffer from welfare loss caused by the progress of metropolitan technology, although government transfer might alleviate some of this loss.

ABSTRACT. We analyze the impact of preference matching and income on the distribution of the population in an aggregate economy consisting of an urban and an adjacent rural region. It costs more (less) to live in the urban (rural) region. Individuals choose freely to live either in the urban or in the rural region. They differ in their incomes. These incomes are uniformly distributed on the unit interval. Our analysis leads to four results. First, when the cost differential parameter satisfies a condition, both regions are occupied in the equilibrium. Second, when this parametric condition holds, in any equilibrium in which the mean income of individuals varies across the two regions, every resident of the rural region has a lower income than every resident of the urban region. Third, there exists an income threshold and all individuals with higher (lower) incomes choose to live in the urban (rural) region. Finally, in the equilibrium with income sorting, it is possible to make everyone better off by slightly modifying their residential choices.

ABSTRACT. This study seeks to analyze pedestrian fatality patterns within the state of Ohio over a ten-year period (2007-2016) using spatial statistical methods including nearest neighbor statistic, Moran’s I and General G statistics. Identifying patterns of fatal pedestrian crashes can help to focus safety efforts and encourage walking as a travel mode through elimination of safety hazards. The study also seeks to understand the effects of aggregated data across differing spatial scales on the outcome of the analysis and determine the most useful spatial scale at which to study pedestrian crashes. After analyzing these fatalities across three different spatial scales, counties, census tracts, and traffic analysis zones, it concludes that local spatial analyses at census tract scale are most informative. It goes on to recommend locations within Ohio for future analysis based on the resulting maps of the spatial analysis. The results of this study identify areas for further focus, including areas in Ohio with both higher and lower-than-expected clusters. Further exploration of these areas can illustrate best practices, as well as systematic safety issues, the results of which can be used to inform planners, decision-makers, and researchers.

ABSTRACT. This work is an empirical study of the relationship between the availability of firearms and crime levels at the US state and county level. Existing work in this area has had to overcome the fact that reliable gun ownership data are available only at the national level. Thus, a wide variety of proxies for gun ownership have been used in the literature. Some examples are: gun magazine subscriptions (Duggan (2001)), percent of suicides or homicides involving a gun (Cook (1979), Azrael et al (2004)), the percent of robberies and other crimes in which guns are used (McDowall (1991)) and the General Social Survey (GSS) (Glaeser and Glendon (1998)). The paper’s contribution is a new measure of firearms prevalence that is based on the number of firearms that are listed for sale on several online classified advertisement websites and that are used exclusively for firearms sales. This has the advantage of allowing for a distinction in the data between different types of firearms, such as rifles (which are often used for hunting) and NFA firearms (a category that includes machine guns). The measure also varies across states, regions, and time. These data are linked to state and county-level violent crime data and are used to estimate the linear relationship between the availability of firearms and crime, after controlling for relevant socioeconomic, public safety, and legislative conditions.

ABSTRACT. Bertram (2004) establishes that islands exhibit strong income convergence towards the income level of the patron economy to which they are tied either politically or economically. This paper uses a world sample of small island economies and OECD countries to explore the connections between the trade relationships and the degree of income convergence present in the sample attempting to discover some of the reasons for the result shown by Bertram (2004).

ABSTRACT. While direct-to-consumer (DTC) agricultural marketing has been conducive for metropolitan farms historically due to low transportation costs, recent efforts to expand broadband access in rural America may have helped develop a new market channel for rural farms to undertake DTC sales. We find that farms distant from metropolitan counties that are new to DTC marketing are 6 percentage points more likely to have online marketplaces, ceteris paribus. Our results imply that if food e-commerce increases in the future, an increasing proportion of DTC sales could be of value-added goods produced at greater distances from cities.

ABSTRACT. Some rural regions of the U.S. face significant net out-migration, consistent with ongoing trends of rural-to-urban migration and migration of high skilled individuals (college and high school graduates). For example, West Virginia, which is predominately rural, was one of three states to experience negative population growth, and one of fifteen states to experience net out-migration over the 2010-2016 period. Additionally, many rural regions are affected by other economic trends that favor urban areas, leading to economic distress. These same regions are suffering from significant impacts due to increases in drug addiction and death, primarily due to opiate abuse. All of these are headwinds to rural regions, which are struggling to retain workforce talent. Quantifying the effects of these factors would help policymakers develop strategies to mitigate net out-migration and increase the broader socioeconomic well-being of rural communities.

ABSTRACT. This study attempts to explain the presence of attribute non-attendance (ANA) in the discrete choice experiment through previous purchasing experience. Attribute non-attendance arises in the DCE framework if decision-makers do not attend to all of the attributes presented in the choice exercise and here is inferred via the method developed in Hess and Hensher (2010). Preliminary results show the presence of ANA in this choice experiment, though after controlling for previous purchasing experience, mean measures of ANA fall between 53 and 92%. These results suggest that inferred measures of ANA are sensitive to model specification.

ABSTRACT. In 2018, California held a ballot on proposition 8, the limits on dialysis clinics’ revenue and require refund initiative. The proposition requires clinics to issue refunds to patients or their payers for revenue that exceeds 115% of the cost. In this paper, we explain the counties regional dynamics of the California proposition 8’s failure. First, we develop a theoretical model on how voters weigh their cost and benefit to the voting outcome. Then, we estimate determinants of such voting behavior by implementing elastic-net penalized binomial logit regressions. This is a machine learning method and helps assess the best model. Then we estimate a spatial panel model to account for spatial effects controlling for county-specific characteristics, median voter preferences, and special interests. We use high-dimensional data retrieved from American Community Survey, referendum data from California Secretary of State, and data on licensed facilities in California. This study contributes to explain California healthcare and business regulation dynamics in regional setting.

ABSTRACT. This paper studies whether Spanish individuals move from one region to another because of the huge regional differences in the Inheritance Tax or if they are driven by alternative motives. This is an important issue in Spain since the Inheritance Tax is a regional tax. Regions set their regulations in terms of tax rates and exemptions. In some regions the acceptance of an inheritance is not very costly, but in others people cannot afford the Inheritance Tax and are therefore unable to accept the inheritance. In our work, we analyze the effect that the interregional disparities in the Inheritance Tax have on the borders between regions. For this purpose, data from the Municipal Register (Padrón Municipal) for more than 15 years (1996-2011) and several Censuses from 1991 to 2011 (National Institute of Statistics) are used to estimate econometric models in which the endogenous variable is the growth rate of the population at the municipal level and the main explanatory variable is a variable that considers the spatial differences in the Inheritance Tax, assuming two scenarios. The first is that the municipality belongs to a region with a lower marginal tax rate and the second is a region with a lower marginal tax rate than its neighboring region (geographically). The samples used also allow us to explore the behavior of the population by age group and to examine if the possible effect is asymmetric, affecting individuals more. This study contributes to a very scarce literature, and has social, economic, and demographic consequences.

ABSTRACT. I study the contribution of unemployment inflows and outflows to the dynamics of the unemployment rate for different age groups, skill level, and recipients of benefits in accordance with the Social Code Book III and II at the regional level in Germany. Using administrative data provided by the German Federal Employment Agency, prime age and older workers as well as low and high skilled worker differ significantly in their labor market flow rates. In contrast to the recent literature I provide results for flow contributions using regional level data. I also provide an alternative solution to consider the flows from/to the non-labor market to/from unemployment that allows a measurement of their relative contribution. With respect to the literature on the relative importance of separation and job finding rates for the dynamics of the unemployment rate, Hall (2005) and Shimer (2005, 2012) conclude for the US labor market that the job finding rate is more relevant, while Fujita and Ramey (2009) and Elsby et al. (2009) come to the opposite conclusion and find evidence for a relative larger contribution of job separation. Smith (2011) find evidence for UK that increases in the unemployment rate come along with rising separations. Petrongolo and Pissarides (2008) conclude that both flow rates are of similar importance for the UK labor market, while job finding rates contribute relative more to the French and Spanish unemployment fluctuations. Shimer (2001) argues that a high proportion of young workers provides an incentive for firms to create new jobs because younger workers undertake more search activities, which reduce the firms' recruitment costs. Burgess (1993) and Pissarides and Wadsworth (1994) find evidence in Great Britain that rates of job separation are higher for young workers because a higher proportion of such workers engage in on-the-job search activities. Lower job finding rates for older workers can result from age discrimination (Johnson and Neumark 1997, Charness and Villeval 2007, Langot and Moreno-Galbis 2008) and assumed or actual productivity differentials (Haltiwanger et al. 1999, Hellerstein et al. 1999, Daniel and Heywood 2007). Productivity may increase with age if job experience is important (Autor et al. 2003, Nordström Skans 2008) or decline if human capital depreciates over time, particularly due to technological change or a loss of manual abilities (Bartel and Sicherman 1993, Börsch-Supan 2003, Autor and Dorn 2009). A change in the relative demand for low and high skilled worker can also cause different job finding and separation rates. Finally, negative long-term effects that unemployment has on future labor market outcomes, the scarring effect (Arulampalam et al. (2000), Gregg (2001), Biewen and Steffens (2010)), may significantly affect transition rates. Considering these findings, I argue that it is not evident which implications the increasing relative appearance of older job seekers and job candidates, the share of low skilled worker, and the changing share of basic security recipients may have on unemployment dynamics. This study adds to the literature as follows. First, I provide an alternative solution for both the theoretical and empirical analyses of flow contributions on unemployment dynamics in a three-state model (flows between employment, unemployment, and inactivity). Second, I analyze the flow contributions to unemployment fluctuations at the regional level for all unemployed as well as for different age cohorts, three different education levels, and the groups of unemployment benefit and basic security recipients. Third, I consider panel data with regional flow rates in the empirical analysis. During the period considered the annual unemployment rate has declined from 9.0% to 5.7%. Based on monthly administrative data for the period January 2007 to December 2017 and 402 districts (Kreise) in Germany I receive the following preliminary results: Separation and job finding flows account for roughly 80% of unemployment fluctuations for prime age and high skilled workers while for older and low skilled workers for about 60%. The relative contributions of separation rates to explain unemployment dynamics are more important than those from job finding rates. On average, for all unemployed the contribution of the separation rate is about twice as large as the unemployment dynamics due to the job finding rate. Separation and job find are more important for unemployment dynamics of unemployment benefit recipients compared to those receiving basic security. Older workers unemployment dynamics is more driven by separation than unemployment dynamics of prime age workers. The same applies to low skilled workers compared to the high skilled. The results also differ across the level of rurality. In general, in rural areas the unemployment dynamics are lower. In some cases exiting unemployment has in these regions a larger contribution to unemployment dynamics.

ABSTRACT. The link between artistic pursuits and scientific and engineering discoveries—best exemplified by Leonardo da Vinci—is so common in the pantheon of Nobel Laureates (Root-Bernstein; Csikszentmihalyi) as to seem unremarkable. However, there is at least strong anecdotal evidence to suggest spillovers from the arts to less stellar talent is not motivating public debate on the best way to prepare students for the innovation economy. Finding evidence of a general arts-innovation nexus in the population has been stymied by the lack of widespread data on amateur pursuit of the arts needed to detect an association with rare innovation events. We address this deficiency by using the web behavior from millions of geo-located devices to identify where internet traffic to websites catering to arts hobbyists is highest. Using the NEA’s Survey of Public Participation in the Arts we are able to model the individual and place characteristics that are most strongly associated with having an arts avocation. This information allows us to use the web based arts avocation index as an endogenous treatment affecting the patent productivity rate among STEM occupations likely to participate in patent applications. Our central research question can be simply summarized as follows: if an engineer can be expected to contribute to the patent productivity of a region, does an engineer who plays blues guitar contribute even more?

ABSTRACT. William Nordhaus, co-winner of the 2018 Nobel Memorial Prize in Economics, was commended for his path breaking research on climate change. What made this research uniquely significant in the ever-growing realm of climate change research was his understanding of the relationships between wellbeing, investment, wealth and sustainability. His papers on non-market accounting and expanded measures of national wealth were critical to his analysis of climate change. The Rural Policy Research Institute (RUPRI) takes a similar approach to understanding community and economic development. The Comprehensive Wealth Framework (CWF) builds on the concept of Fisherian wealth as described by Nordhaus and identifies eight types of public and private capital that comprise true wealth. One of the components of comprehensive wealth is cultural capital. With initial funding from the National Endowment for the Arts, RUPRI has recently established the Rural Cultural Wealth Lab, the foundation of which is the CWF. The mission of the Lab is to explore the intersection of rural arts and culture, entrepreneurship and innovation, and the role of rural culture in contributing to the nation’s comprehensive wealth. In this paper we describe how cultural wealth will be measured and how it will be modeled as a dynamic system. This issue is relevant to the field of geography because cultural capital involves characteristics of place, immobile assets, and spatial distribution of asset ownership.

ABSTRACT. We study the impact that the provision of a local public good (LPG) by two cities has on their ability to attract and retain members of the creative class. This creative class consists of two types of members known as engineers and artists. Engineers are wealthier than artists and they also value the LPG more. We first focus on each city in isolation. We compute the marginal value and the marginal cost of the LPG and then determine the provision of this LPG when the provision is determined by uniform contributions and majority voting. Next, we allow the creative class members to migrate between the two cities and analyze whether engineers or artists migrate, the equilibrium distribution of the creative class, and the efficiency of the LPG provision. Finally, we consider the situation in each city just before migration and study how much of the LPG is provided when proportional contributions and majority voting determine this provision. A related question we address is whether engineers or artists now have an incentive to migrate and, if yes, we identify who would like to migrate and to which city.

ABSTRACT. Spatial Econometrics is often use to estimate peer effects in network models. Exogeneity of the spatial weight matrix is an unrealistic assumption when it comes to model interactions between individuals. In this paper we relax the assumption of exogeneity of the weight matrix and extend the exponential random graph models (ERGMs) by including nodal-level attributes that would directly affect the probability of links between individuals. ERGMs are some of the most popular models in network analysis but are in practice very difficult to estimate. We propose a new estimation procedure based on the Double Metropolis Hastings to estimate direct and indirect peer effects.

ABSTRACT. Monte Carlo estimates for the log-marginal likelihoods arising in a set of space-time dynamic panel data models involving space-time lags of the dependent variable vector, as well as space-time lags of the disturbances are set forth. The log-marginal likelihoods can be used to compare these model specifications allowing practitioners to find the specification most consistent with the sample data. Model comparison can also shed light on appropriate spatial weight matrices as well as time lags incorporated in alternative specifications. The approach taken is computationally efficient for very large samples, and produces estimates of the model parameters along with Monte Carlo estimates for the log-marginal likelihoods.

Our Markov Chain Monte Carlo estimation procedure uses: 1) a Taylor series approximation to the log-determinant based on traces of matrix products calculated prior to sampling, 2) block sampling of the spatiotemporal parameters, which allows imposition of the stability restrictions, and 3) a Metropolis-Hastings guided Monte Carlo integration of the log-marginal likelihood.

ABSTRACT. Good science needs good measurement. A good measurement involves three aspect: a clear concept of what we are trying to measure, a measuring device or methodology, and understanding of how the results from the instrument can be interpreted or applied in practice. Important applications of good measurements are description of observed phenomena, analysis, and forecasting. Moran's I is a widely used spatial statistic that is used to diagnose the presence of spatial dependence in the data, that is, to infer whether spatial data appear to be statistically independent or not. As a diagnostic device, Moran's I is a direct computation statistic, so it is easy to compute and has well-established properties under the null hypothesis of no spatial dependence. However, Moran's I fails to be useful in forecasting and analysis: forecasts based on Moran's I would be biased and inconsistent, whereas numerical value of Moran's I is useless for answering questions that we normally expect from a good measurement: is one dataset contain more spatial dependence than another, or how much spatial dependence there is a dataset, or even if one dataset has more spatial dependence than another. The existing workarounds involve method of moments or maximum likelihood estimation of models with spatial dependence, with both approaches involving an indirect definition of the measurement -- through a numerical solution of a non-linear constrained optimization problem. This paper proposes an easy-to-compute spatial statistic that is expressed as a closed-form direct computation and provides unbiased, consistent, and easy-to-interpret measurement of spatial dependence.

ABSTRACT. Focus is on efficient estimation of a dynamic space-time panel data model that incorporates spatial dependence, temporal dependence, as well as space-time covariance and can be implemented where there are a large number of spatial units and time periods. Quasi-maximum likelihood (QML) estimation in cases involving large samples poses computational challenges because optimizing the (log) likelihood requires: 1) evaluating the log-determinant of a large matrix that appears in the likelihood, 2) imposing stability restrictions on parameters reflecting space-time dynamics, as well as 3) simulations to produce an empirical distribution of the partial derivatives used to interpret model estimates that require numerous inversions of large matrices. A Markov Chain Monte Carlo (MCMC) estimation procedure is introduced that produces estimates equivalent to those from QML along with a Monte Carlo integrated estimate of the log-marginal likelihood, useful for model comparison. An applied illustration is based on over six million observations. The MCMC estimation procedure uses: 1) a Taylor series approximation to the log-determinant based on traces of matrix products calculated prior to MCMC sampling, 2) block sampling of the spatiotemporal parameters, which allows imposition of the stability restrictions, and 3) a Metropolis-Hastings guided Monte Carlo integration of the log-marginal likelihood. In addition, an efficient approach to simulations needed to produce the empirical distribution of the partial derivatives for model interpretation is set forth.

ABSTRACT. This paper develops a spatial equilibrium model, which combines housing, population, human capital, and asymmetric labor demand. The focus of this structure lies in testing the symmetry between labor demand shocks, building upon the static model of Glaeser and Gyourko (2005). We report significant asymmetry in both shocks and their effect, which has significantly varying affects in declining and growing counties. This has implications regarding housing, education and place based development policies.

ABSTRACT. In the United States, there is a great deal of focus on affordable housing. As real wages have stagnated, home prices have continued to increase. According to a monocentric city model, the inner cities should be the areas most likely to have access to affordable housing. Despite this, many rural areas are experiencing difficulties with providing affordable quality housing. Rural communities tend to have two types of housing options for its residents: 1) low-cost housing that lacks quality or 2) quality housing that is unaffordable. Using data from the American Housing Survey, I plan on researching housing quality and affordability in rural America. This study will look at how widespread these issues are and where and what kind of communities are most impacted.

ABSTRACT. This paper shows that neighborhood identity serves as a principal determinant of housing prices. Casting a new light on the standard hedonic housing model, we view neighborhood identities as heuristic cues that reflect a complex set of local attributes. Market participants organize and simplify information gathering around commonly accepted neighborhood definitions. Searching for housing by neighborhood identities reduces the effort needed to scan and evaluate a plethora of properties and a multitude of possible local amenities.

Our database contains more than 50,000 housing market transactions in the Charleston, South Carolina metropolitan region. In a log-linear, hedonic model, we estimate neighborhood fixed effects, along with the other fundamental determinants: housing size, age, and structural characteristics. These determinants are what buyers and realtors typically use to evaluate properties for sale and make decisions. Neighborhood fixed effects capture time-invariant characteristics of the local area. We then add covariates that potentially change the value of the house during the period and contrast their influence with the fundamentals of size, structural characteristics and neighborhoods.

Overall, our regression fits the data well, with an adjusted R2 of 0.92 when the housing size, structural characteristics, and the neighborhood identity are included. The neighborhood fixed effect had a pronounced influence on regression results. Neighborhoods defined by the real estate market outperform alternative local area identities that could be used to simplify information gathering and decision-making (zip codes and elementary school boundaries). We find that hedonic estimates of the covariates when with different local area fixed effects. When we add other covariates to the neighborhood fixed effect and other fundamental determinants of housing to our model, the estimates remain stable. Even so, we do find that a major infrastructure may have affected housing prices in the Charleston region. In addition, we uncover evidence that buyers paid more for homes zoned for schools that had received positive local school quality ratings.

Finally, we tackle an important statistical issue in housing market research: possible spatial dependency stemming from unobserved neighborhood market conditions. To ensure that our estimates are not subject spatial dependence, we devised a new approach to testing for spatial autocorrelation using the Moran’s I, which is computationally advantageous for large housing market datasets lie ours. Interestingly, when the neighborhood fixed effects are added to the model, spatial autocorrelation is no longer detected.

ABSTRACT. While metro rail access is associated with benefits from proximity to public transportation, it also has costs in terms of noise and congestion that nearby residents may not value. Previous research has found that there is an average net value of access to public transit. However, the results in one region may not hold up in other regions. Additionally, the net value of metro rail access may vary by income levels or other factors. Using an array of econometric techniques, we use housing data to examine how residents value the proximity to metro rail lines in Los Angeles. The complex housing markets of Los Angeles allow us to assess how property values are impacted in different neighborhoods that may not value access to metro lines the same. We also explore whether, overall, the results from previous studies hold in this region. With metro lines requiring large financial investments, understanding how residents in different communities value access to metro lines is of considerable policy importance.

ABSTRACT. A policy to provide higher education is a tool of the local state government to enhance human capital in an area, which is important for regional growth. There is a concern, however, about public universities dramatic increases in resident tuition fees. Public universities have been trying to attract more out-of-state students since they could generate revenue from higher nonresident tuition fee, and they also could increase their reputation from higher abilities of nonresident students. This situation could create negative effect on resident enrollment to access higher public education. Another concern is that educated people are quite mobile. As a result, public investments in higher education may not benefit state if the college students move to another state after their graduation. This paper aims to answer two questions. The first question is whether or not there is a crowding out effect of nonresident on resident enrollment. The main source of data is 4-year public university from the Integrated Postsecondary Education Data System from 2005 to 2014. Based on the categories of Barron’s Profiles of American Colleges, the universities are separated into low, medium, and high quality. Moreover, they also are grouped by average income of state, in which state they are located, which can be low- and high-income state. The second question is who is going to move or stay after graduation. The main source of data is American Community Surveys from Integrated Public Use Microdata Series from 2005 to 2014. Primarily, the results show that there is a crowding out effect in low- and medium-quality universities in low-income states. Resident students may be deterred to access public universities from nonresident enrollment. Since nonresident students have a higher chance to leave the state after they graduate, decrease in resident students enroll in higher education could be a cause of decrease in human capital stock for low-income states.

ABSTRACT. We revisit the issue of whether or not pubic schools are resegregating. The previous research that has examined the possibility of resegregation has focused on the release of schools from desegregation orders in large school districts, frequently in the South. Court ordered desegregation was commonly achieved by merging schools and/or busing students to achieve the unitary status of a single, rather than a segregated school system. In the early 1990s, three Supreme Court rulings allowed for these desegregation plans to be terminated. After being released from a court order, schools were free to remove busing plans and it is possible that districts could resegregate. We expand on this literature in three ways: we allow school district boundaries to change, we include rural areas as well as urban districts, and we reconsider if there are differences between Southern and Non-Southern districts. We find that for those districts released from a court order, the boundaries did not affect resegregation. We also find that the entire country was affected by these desegregation orders, not just urban areas. Finally, we find similar segregation patterns in both Southern and Non-Southern states, which is different from the previous research who has found that these effects are concentrated in the Southern states.

ABSTRACT. Moriarty Graduate Paper Competition

This paper investigates the effect of public library programs and participation on unemployment and labor force participation in Appalachia. Appalachia is an economically distressed area, mostly rural, and with a sustained lower level of labor force participation and a higher level of unemployment. I use public library staff and the amount of print resources and computers available as instruments. The results show that neither adult nor children's programs and participation affect local labor market outcomes. These results are robust across different specifications. Spatial econometric estimates corroborate the main results and provide evidence of spatial spillover effects, especially for children's programs.

ABSTRACT. Arsenic is a naturally occurring element found in ground water, surface water, and many foods. US EPA has classified arsenic as a Group A human carcinogen; consuming water containing arsenic over a long period of time increases risk of bladder, lung, and skin cancer as well as cardiovascular disease, and has been linked to increased risk of adverse birth outcomes and reduced IQ in children. Some regions of the country like New Hampshire have more naturally occurring arsenic in drinking water. In 2001, the USEPA lowered the Maximum Contaminant Level (MCL) for arsenic in drinking water from 50 parts per billion (ppb) to its current level of 10 ppb. In 2018, the New Hampshire Department of Environmental Services (NHDES) submitted a report to the Legislature recommending reducing this standard to 5 ppb. Reduction of arsenic in drinking water may provide a range of economic benefits from reductions in treatment costs of cancer and cardiovascular diseases (CVD), reductions in premature death associated with cancer and CVD, reduced loss of good health associated with cancer and CVD morbidity, reduction in uncertainty about getting cancer or CVD, and other issues such as avoiding reductions in children’s IQ. However, additional treatment costs incurred by water supply systems will impact already stressed infrastructure. This paper provides estimates of the economic value of reducing the MCL allowable in public water systems in NH. The research used secondary data to assess risks associated with arsenic ingestion and paired these with economic data monetizing these adverse health effects. Primary data was collected via survey from New Hampshire households to obtain estimates of the willingness to pay for risk reduction from arsenic ingestion; these results were also used to generate a value of statistical life estimate for comparison to estimates from other studies. We derive a NH specific VSL, which, when coupled with estimates of deaths avoided from the increased water standards, provides a more refined estimate for the benefits of the proposed legislation. Our stated preference survey estimates show that, on average, respondents are willing-to-pay $35.50 per month ($426.00/year) for reduction in lung and bladder cancer risks associated with lowering the maximum allowable level of arsenic in drinking water from 10 ppb to 3 ppb. Using these estimates, we derive a NH specific VSL of $12,425,000. When applying the VSL values to the NHDES-provided estimate for deaths from arsenic-related CVD, lowering the arsenic MCL to 3 ppb is associated with an economic value between $4.2 billion and $7.1 billion. Applying recent estimates of the economic value of children’s IQ points to the expected loss of IQ among children exposed to arsenic above 5 ppb, the economic value of reducing the arsenic MCL to 3 ppb ranges from $26.9 million to $194.3 million. Taken together, these economic values total between $4.256 billion and $7.605 billion. To estimate net benefits of the proposed standard change, we gathered secondary data on the costs of various treatment options. These include ion exchange, reverse osmosis, adsorption, activated alumina, and coagulation/filtration. We use estimates previously generated in a cost-benefit analysis by EPA as well as NHDES, and consider scales of implementation from household to community water system level.

ABSTRACT. Social capital has long been known to have an effect on health and health outcomes. A number of studies have explored how social relationships—or conversely, the lack of social ties—impact various health outcomes, including mortality rates, diabetes, obesity, and self-reported health measures.

Our work builds on this earlier research and considers the role of social capital on several health outcomes, focusing on the Appalachian Region.

This research uses a two-period panel and controls for a number of explanatory factors, including health behaviors, access to care, race and ethnicity, and socioeconomic variables. Our preliminary findings suggest that social capital has a greater impact on health outcomes in the Appalachian Region than in the rest of the country.

ABSTRACT. In January 1973 the Supreme Court of the United States ruled that a woman has a right to privacy when it comes to a woman's decision to have an abortion, but that states have compelling interests regarding women's health and prenatal life that allow states to regulate abortion. In other words, women have a legal right of access to an abortion. However, some state legislatures disagree and are trying to restrict access in many ways. These ways include mandatory waiting periods, having to repetitively answer questions about the personhood of the fetus, parental notification laws, and Arkansas’ 2013 ban on abortions after 12 weeks. However, geographic access is affected by regulations restricting the supply of doctors willing to perform the procedure and the supply of clinics, as well as an endogenous demand relationship. In this paper I describe some preliminary analyses measuring accessibility to clinics across the US, and plans for future research.