THE RURALIZATION OF DETROIT: IMPLICATIONS FOR ECONOMIC REDEVELOPMENT POLICY By Tanner Connors A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Agricultural, Food and Resource Economics Master of Science 2018 ABSTRACT THE RURALIZATION OF DETROIT: IMPLICATIONS FOR ECONOMIC REDEVELOPMENT POLICY By Tanner Connors As industrial cities transition into a post - industrial state, their demographics and socioeconomic characteristics transition as well. Their population sizes and densities are still considered urban, but are they truly as urban as a thriving central city? The literature identifies many characteristics beyond population size that could be used as rural indicators. There are clear distinctions between rural and urban economic development, so it is imperative that there is a clear understanding of where a comm unity fits on a rural - urban spectrum, to create effective redevelopment polic y . There is limited research on urban placement along a rural - urban continuum in the absence of spatial association . Us ing place level data, I study the differences between popula tion ranges across select rural indicators and apply the findings to Detroit, Michigan on the C ensus tract level, a city that has shrunk to half its peak size and has faced extreme financial difficulties while transitioning into the post - industrial state. Significant differences were foun d between population sizes based on several rural indicators. opulation increasingly resembles that of a rural community. These findings support the theory that traditionally rural economic development policies may have positive effects in Detroit. iii A CKNOWLEDGEMENTS Thank you to my major advisor, Dr. Mark Skidmore. I am very grateful for your encouragement and guidance over these past two years. Thank you for the many opportunities you gave me to participate in exciting research and for helping me dev elop my skills and abilities. Thank you to my committee members, Dr. Laura Reese and Dr. Steve Miller. I am deeply grateful for your guidance and willingness to help develop my research. Thank you to the faculty, staff, and graduate students within the department. I would like to especially thank my cohort members who not only made this experience easier, but unforgettably enjoyable. I love you all. I would also like to thank Rosa Soliz fo r all your help, kindness, and countless conversations. Thank you to my incredible family and friends who I can always rely on for constant support and encouragement. I love you all. To my siblings, parents, Nana, Kyle Connors, and Dr. Hillary Sackett, I a m especially grateful for all the help you have given me throughout this journey. None of you have ever let me doubt myself, you have always encouraged me to challenge myself, and have been there to give me the extra push when I need it. To my nephew, Mile s, and niece, Nola, thank you for being the perfect little human beings that you are . To Joe Furness, I cannot thank you enough for your un wavering love and support through every struggle and victory alike . I am incredibly grateful for everything that you have done for me, but I am mostly grateful that you brought three cats into our household. Thank you for the love and laughs, Joe, Scout, Akut, and Calibur. iv TABLE OF CONTENTS . . 5 . 5 . 3 3 6 3.3 Outlier Analysis Methods 7 . 9 31 . 7 APPENDI 4 4 4 9 v LIST OF TABLES . 4 Table 2: Summary of Rural - .. 5 .. 6 Table 4: Number of Detroit Census Tracts that Correlate with Population Ranges... ... 8 . 20 .. 21 . 1 Table 8: Per Capita In . 2 . 3 . ..2 4 . 25 Table 12: Average Poverty and .. 27 Table 1 3 : . 2 Table 1 4 : . Table 1 5 . . 3 Table 16: Size Distribution ... Table 17: .. .. Table A1: Summary Statistics of Rural Indicators in . . Table A2: Mean Values of Rural Indicators in .. ..46 . 47 ... 48 vi LIS T OF FIGURES .. 7 Figure 2: Comparing Detroit Census Tracts to Population Ranges throughout Incorporated Places in the U . S . - based on Population .. . .. 9 ... 20 ... 20 .. 1 .. 2 ... . 3 Figure 8: Rural Classificatio .. ..2 4 Figure 9: Rural Classification based on Percent of Population Employed in the 2 5 Figure 10: Detroit Census Tracts above the National Average Poverty Rate ..... .. .... ........2 6 Figure 11: Detroit Census Tracts above the National Average Unemployment Rate... .. ..2 6 1 CHAPTER 1: INTRODUCTION For many decades , economic development efforts focused on urban areas with rural communities being of secondary importance. Despite this, and over time , two general tracts of development policies have evolved one for urban places and another for rural communities. Urban and rural areas are not just distinct from one another i n terms of population size and density, but also i n other social and economic characteristics that could potentially help distinguish a rural area more fully than a definition that relies solely on population. and economic condition . B ecause urban and rural redevelopment plans and policies emphasize and address different issues , determining location within the urban - rural spectrum is important . For example, urban policies are often place - based, focused on managing congestion and growth, wh ereas rural policies are often person - and community - based, focused on improving the disadvantages such as lack of access to employment and resources . Is it enough to look at the size of a population, or population density of an area, to determine which development policies should be used ? I argue there should be a b roader view on the specific needs of each community. D etroit, Michigan , prides itself o n the resi lience of its community members. H owever , a similar resilience has not historically been present in the sustainability of the . Many have referred to this distressed city as dead (Reese and Sands, 2017) , but the people of Detroit have not given up and still strive to revitalize it . Detroit was once the center of the auto industry and home to the Big Three: Ford, General Motors, and Chrysler. People relocating to Detroit to take jobs in the growing 2 manufacturing industry resulted in a steep population increase, peaking in 1950 at almost 1.86 million people. The regional economy was bui lt and heavily dependent on this industry. R acial tensions and union negotiations contributed to the movement of the auto industry from Detroit to the broader m etropolitan area (Reese, Eckert, Sands, and Vojnovic, 2017; Reese, Sands and Skidmore, 2014; Padnani, 2013, Vojnovic and Darden, 2013) . Th e result was a reduction in employment opportunities within the central city. As automation increased, employment opportunities became scarcer , as the automobile companies required less labor. E ventually, auto companies began to shrink and relocate out of the Detroit region altogether to areas with lower labor costs , such as southern states and out of the U . S . completely, following the energy crisis in the 1970s, the recession in the 1980s, and more recently, increase in foreign competition (Padnani, 2013). ongoing failure to diversify put it at higher financial risk than cities in similar positions, culminating in the largest munic ipal bankruptcy in U.S. h istory (Padnani , 2013; Neuman , 2014). The lack of effective leadership is partly to blame for manufacturing industry (Padnani , 2013 ; Reese, et al . , 2017, Vojnovic, 2013 , Reese, Sands , and Skidmore , 2014 ) . Since 1950, h as fallen from 1.8 million to 683,443 people in 2016 . from the many efforts made to reverse the economic downturn. After four decades of financial struggles, the city government appears to have improved control of its finances and i s no l onger under financial oversight (Domonoske , 2018). However, the majority of th ese improvements have been seen in Downtown and Midtown, creating a divided city. The recent vibrancy displayed in these up - and - coming areas ha s not sp illed over into 3 , wh ich have become isolated from the recent growth (Reese et al . , 2017). Despite the investments made in designated regions, the city still faces high unemployment and poverty rates, low educational attainment leve ls , and high rates of residential vacancy. Although Detroit has made progress, it still faces many challenges and must use the most effective redevelopment policies that are designed for the specific needs of the struggling city. Are strictly urban development policies appropriate , or given the significant and ongoing population loss should development policies designed for rural plac es be consid ered? The evolving soci oeconomic conditions suggest that relying solely on traditional urban development policies may not be the most effective approach . This thesis show s a clear difference between rural and urban areas by identify ing significant differences in sociodemographic characteristics and focal points in economic development policies among the different populations . I then evaluate the city of Detroit to determine the degree to which it resembles urban versus rural places. I conclude by offer ing implications for effective economic development policies for Detroit, given where it lies along a n urban - rural continuum. As a prelude to the full evaluation, based on a number of factors , Detroit resembles rural places more so than urban places . Chapter 2 offers a literature review that includes a discussion of rural indicators and key components of redevelopment policies in both rural and urban areas. In Chapter 3 , data from the United States Census Bureau are used to measure the differences in economic and demographic conditions for places of different populations . These findings are used to assess the degree to which Detroit resemble s urban or rural environments . Regres sion and outlier analyses and a cluster analysis provide further 4 support that Detroit in many ways resembles rural places more so than urban places. Chapter 4 offers a set of development policy implications for Detroit and concludes. 5 CHAPTER 2: L I TERATURE REVIEW 2.1 Indicators of Urban and Rural Communities There is a n expansive amount of literature on the comparison of urban and rural communities , but most of this literature focuses on poor areas . Some argue that rural cannot simply be defined by population densit y or proximity to metropolitan areas (Hart, Larson, and Lishner, 2005; Modi, 2009; Halfacree, 1993) . Regardless, researchers to categorize ruralness and urbanicity . The Census Bureau defines rural areas as those outside of urban clusters with populations less than 2,500 people and often lack subst antial commuting activity to the urban center (Hart , Larson , and Lishner, 2005). This assumes that rural is merely a residual category representing all communities not classified as urban (Burchinal and Siff, 1964). Many believe that this definition is not always about which there is no universal agreement ommon definitions frequently rely on stereotypes and personal experience such as pastoral landscapes. A number of s tudies consider factors beyond a traditional framework, provid ing a common set of factors that characteri ze rural areas beyond the limited definition offered by the Census Bureau . Duncan and Tickamyer (1988) show similarities between the average rural population and poor urban po pulations, such as the presence of a low - skill labor force, isolation from established social and economic institutions, limited access to formal education, and a higher level of persistent poverty . Although many agree that having a low population density is a n important indicator of rural (Hart, Larson, and Lishner , 2005; Duncan and Tickamyer , 1988; 6 Halfacree , 1993; Castle and Weber , 2011), there are several other characteristi cs that many studies support, such as l imited access to public services. Rural areas often lack a well - developed public sector and have limited access to individual and community resources (Duncan and Tickamyer , 1988). Hart, Larson, and Lishner (2005) and Dillman and Tremblay (1977) emphasize that h ealth care is an important example of this; although rural residents tend to have worse health conditions, they often make significantly fewer visits to health care facilities. R ural areas tend to severely lag urban areas in terms of both access to and quality of health care providers and services. Medical personnel are fewer, especially trained specialists, and services offered are more limited. Health care facilities and practic es used are often not up to of date. The reason for these disparities between rural and urban places is partly due to the lack of collegial support in rural areas. Education is a nother example of where services are lacking . R ural areas have limited access to formal education and on average have lower quality educational resources , resulting in a population with lower levels of edu cation (Hart, Larson, and Lishner , 2005; Duncan and Tickamyer , 1988; Burchinal and Siff , 1964; Dillman and Tremblay , 1977). Teachers in rural areas are often less educated, receive lower salaries, are responsible for teaching more classes (and subjects), and are less likely to be members of professional societies (Burchinal and Siff , 1964). Urban areas also tend to see higher returns to schooling compared to rural areas (Mills and Hazarika , 2003). Rural education systems also tend to lack vocational schooling and post - high school opportunities (Burchinal and Siff , 1964; Dillman and Tremblay , 1977). 7 Defining rural ity use d to be based on whether an area was dedicated to or supported by agriculture, but many modern rural families have no affilia tions to farming, relying on other industries, such as manufacturing, for household income (Perry, 1984). Castle, Wu, and Weber (20 11) note that agriculture is no longer limited to rural ar eas and is becoming more common in urban areas, generally on a smaller, more compact scale, producing more per acre and more diverse, higher - value crops. Urban areas also tend to have recreational f armers that have other incomes as well. Burchinal and Siff (1964) point out that the majority of rural populations no longer solely depend on agriculture . I n fact, blue - collar workers replaced farm workers as the largest occupational group in rural areas in the 1950s. Rural households struggle financially without access to skilled jobs and on average have lower levels of income than urban households . With a lack of diverse economic activities, they can be more vulnerable to economic downturns due to concentrated economic special izations (Duncan and Tickamyer, 1988; Hart , Larson, and Lishner , 2005). Although cost of living is lower in rural areas (Joliffe , 2003), Dillman and Tremblay (1977) say that on average the more rural the lifestyle, the more worse off residents are economically . Castle, Wu, and Weber (2011) show that urban residents have access to more highly paid jobs on average and that rural per capita income is much lower, res ulti ng in rural counties having a higher probability of being a poverty - persistent county. These characteristics are supported in other literature as well; there are lower market wages in rural areas and less access to jobs, skilled or unskilled (Mills and Haz arika , 2003). Unlike urban areas, rural areas tend to lack a middle class (Duncan and Tickamyer , 1988). There are higher rates of unemployment and 8 under employment in rural areas as well as higher rates of poverty . Rural populations also have higher per cent ages of uninsured and under insured individuals (Hart, Larson, and Lishner , 2005) . Household demographics are another trait that many agree defines . average, rural househo lds have higher white populations and less diversity than urban areas (Duncan and Tickamyer , 1998; Hofferth and Iceland , 1998). Rural communities have higher rates of elderly and children and a lack of young, middle - age adults (Hart , Larson, and Lishner , 2005; Duncan and Tickamyer , 1988; Hofferth and Iceland , 1998). The proportion of people age 50 - 90 years is greater in rural areas while the proportion of people age 15 - 40 is greater in urban areas (Joliffe , 2003). This supports Burchinal y that outmigration consists of mainly youth and young adults. There are also fewer female headed household in rural areas compared to urban on es (Duncan and Tickamyer , 1988). Davis, Grobe, and Weber (2010) found that although rural areas are generally worse off economically, rural families demonstrate less use o f childcare subsidies. They indicate that similar findings are shown in other welfare programs . Even given worse economic conditions with higher unemployment rates and lower wages, rura l families participate in social service programs less often and for s horter periods of time. A reason behind this behavior is based on stronger social networks in rural areas. Hofferth relationships and are more likely to be made up of family members than social networks in urban areas. They also show that rural households are more likely to receive f inancial support from family members compared to similar urban households. These households 9 also differ in the way financial help is pr ovided. I n urban areas, older generations are more likely to give to younger households, but in rural areas , younger hous ehold heads are more likely to give to older households. Dillman and Tremblay (1977) discuss other characteristics that are more common in rural than in urban areas. They find those in rural areas generally have worse housing situations; there is more ho me ownership but lower land values. Homes are more crowded and have less adequate plumbing and worse quality of drinking water. There is a shortag e of credit in rural counties. Less time is spent o n recreational activities in rural areas , and unlike urban areas, most recreational time is spent outdoors. There are fewer formal recreational structures such as movie theatres, bowling alleys, and formal sporting facilities. On the upside, t hey also find that there are fewer reported crimes, especially violent and property crimes . When surveyed, rural residents report higher general levels of satisfaction with their lives but more dissatisfaction with specific components. Rural residents tend to have high er satisfaction level s with in tangible things such as environmental quality, a place to raise children, and safety from crime, but they report dissatisfaction with services such as public transportation and roads. These characteristics have the opposite satisfaction patterns i n urban areas. 2.2 Policy Review It is logical for urban and rural areas to have different economic development priorities. While urban planners need to focus on managing congestion and finding the right balance between the diverse needs of their communities , rural planners have the opposite issues like a lack of revenue - generating sources and a population that is 10 suffering fr om a lack of available resources (Cruickshank, 2018). This is why the literature seems to agree that urban redevelopment policies are often place - based, compared to people or community - based as seen in rural communities (Daft, 1971; Sutton , 2008; Dandekar and Hibbard, 2016). Therefore , urban planners tend to focus more on the overall economy and structure of the city by focusing on elements such as investments in infrastructure and policies that incentivize large, individual fi rms to enter the market (Reese and Ye , 2015; McCarthy , 1998 ; Sutton , 2008 ). Of course, populations can share many of the same needs, but most rural communities focus attention elsewhere, on more basic human needs, before prioritizing the same concerns as urban areas. Rural plann ers tend to use more people - based approaches , develop ing policies that focus on improving human and community capital , governance, and resilience (Dandekar and Hibbard, 2016) . This is why rural development economics often entail investments in improving household utilities and the skill level of the wo rkforce with a focus o n low - income areas (Drabenstott, 1995; Martin, 196 6; Dewitt, 1993; Bahmura, 1961; Hansen, 1969). Dewitt (1993) support s the need for bottom - up policies that prioritize the rural poor. Many r ural planners focus on reducing the high unemployment rates, not j ust by strengthening the work force through better academic and vocational training, but also by creating employment op portunities through incentivizing business development (McArthy, 1998; Dewitt, 1993). Although aiding smaller businesses is a concern that has been expressed in both urban and rural areas, it has not been as much of a priority in urban economic development (Reese and Ye). Using financial incentive programs to stimulate business development has been common in urban areas by creating specialized zones that offer 11 tax incentives to individual firms, such as Enterprise and Empowerment Zones (Kroop ka and Noonan, 2009; McArthy, 1998 ). Leo a nd Brown (2000) stress that urban policies that are effective in rapidly growing cities are not necessarily appropriate in smaller, more slowly growing places. Large, prosperous cities have the resources and capacity for specialization to support a large relocating firm, but smaller cities in economic distress may not have the means to do so. Rural areas tend to lack employment opportunities, so incentivizing businesses to open within the community is crucial. However, the literature suggests that urban large, individual firms to the area would not be beneficial for rural areas with high unemployment rates. Although industrial recruitment is commonly practiced for business development, academics now suggest that communities are wasting their resources wit h this strategy (Loveridge, 1996). Large enterprises entering the market are often branch plants, which leads to the hiring of those who hav e relocated with the busines s, w hile s maller, locally based businesses are more likely to hire local residents (McArthy 1998). Many other academics also agree that offering tax incentives to firms is not an effective means of generating empl oyment and weakens the tax base (Dewitt, 1993 ) . Rural businesses are often at a disadvantage when it comes to acces s to information. Rural areas often need improvements in telecommunications for their people and businesses to succeed and grow. Much of the literature suggests that local governments need to strengthen their relationships with institutions that can provid e better access to info rmation, particularly in rural areas (Drabenstott, 1995). 12 Quality of life has not been enough of a focus in urban economic development (Reese and Ye, 2015 ). Urban redevelopment efforts often include reuse of underutilized property. The A merican Planning Association (APA) (2004) suggests that althou gh not as common, redevelopment of underused property can and should be used in rural areas as well . Above all, many sources, including Dewitt (1993), Drabensttot (1995), and the APA Policy Guide on Public Redevelopment (2004) , stress that due to varying needs, rural redevelopment policies should not be aimed at broad regional areas, like we see in urban developmen t economics, but rather individualized planning done for each unique area. In summary, it appears tha t rural economic development policies are often people and co mmunity - based, focusing on issue s such as improving household utilities, telecommunications, governance, sustainability and resilience, and human and community capital through investments in education and job skills training . Development policies in both urban and rural areas focus on business development . However, rura l polic i es tend to try to target small and midsize local businesses t hrough human capital formation , whereas urban policies focus on attracting large, individual firms through financial incentives like tax abatements and specialized zones. Traditional urban policies also tend to focus on areas such as managing expansion and congestion, fostering competit iveness and innovation, and investments in infrastructure. The next chapter uses multiple techniques to identify indicators of the rural - urban divi de to later infer which category of policies may be of use to Detroit. 13 CHAPTER 3: EVALUATION This chapter is divided into five sections. The first section discusses the methods of identifying statistically significant rural indicators among places in the United States as well as methods of comparing Detroit on the census tract level to rural and urban places using mean values. The second section presents the results of the qualitative analysis with a series of maps and tables and discussion. The third and fourth section s discuss the methods and results, respectively, of an outlier analysis using three r egressions. The fifth and final section presents a cluster analysis. 3.1 Methods In this section , I conduct an evaluation to determine where Detroit would fit on the rural - urban spectrum using a variety of socioeconomic factors . See Table A1, in the appendix, for summary statistics of these factors. Although places are most commonly defined as urban or rural based on population size and population density , the literature identifies many o ther variables that show clear distinctions between rural and urban places . I used several of the most commonly cited variables that were available through the United States Census Bureau : p opulation size, population density, household vacancy rates, incom e levels, unemployment rates, percent of population under the poverty level as designated by the official family thresholds (United States Census Bureau, 2016), educational attainment, and industry composition. To get an overall impression of the degree to which Detroit may exhibit rural characteristics , first maps are presented comparing Detroit Census tracts to incorporated places within the U . S . based on the above - listed variables. 14 Using the United States Community Surve y data from 2014 and 2016, incorporated places, minus towns, were chosen for the data set . Census Designated Places (CDPs), or unincorporated places, and towns were not used to limit the numb er of places with vast amounts of uninhabited lands that would not accurately represent true population densities. Th us, the evaluation is based on all incorporated areas in the categories of villages and cities. After removing CDPs, towns , and any observa tion that did not have complete data across all variables, we were left with a sample of 14,468 places across the conti guous United States. A cate was created to indicate the rural - urban classification of a place based on a population of fewer than 2,500 people. Rural populations were divided into three categories based on population si ze. Table 1: Summary of Rural Indicator Variable Rural Indicator Population Range Number of Places Urban 2,500+ 6,124 Moderately Rural 1,500 - 2,499 1,408 Rural 500 - 1,499 3,253 Extremely Rural 0 - 499 3,683 To identify the difference in average values for each variable among different To test whether there was a statistically significant difference between means across the populatio n ranges, a one - way analysis - of - variance (ANOVA) model with a Bonferroni multiple - comparison test was used for each variable. Some variables were not statistically significant between all four categor ies . So , - was created t o compare mean values between rural and urban areas, as defined by the 15 Census Bureau. See Table A 2 , in the appendix, for a summary of mean values in rural and urban places with a comparison to Detroit. Table 2: Summary of Rural - Urban Indicator Variable Rural_Urban Indicator Population Range Number of Places Urban 2,500+ 6,124 Rural 0 - 2 ,499 8,344 The variables that had statistically significant differences among mean values between Population density Residential vacancy rates Educational attainment levels Diversity of industry composition; in year 2010 The additional variables that were statistically significant between rural and urban areas were as follows: Mean household income Per capita income Employment share in the manufacturing industry Diversity of industry composition; in year 2016 A thir was created to obtain mean population densities in incorporated places among six population ranges. The urban variable includes the rural values 1 - variable, but also divides the urban population into three categories. The Census Bureau defines an Urban Cluster (UC) as having a population of at least 2,500 people but below 50,000, as a place with a population of at least 50,000 people is classified as an Urbanized Area (UA). A third category was created to represent very large cities with populations of at least 500,000 people, like Detroit. We will refer to them as Extreme Urbanized Areas (EUA). 16 Table 3: Summary of Urban Indicator Variable Rural Indicator Population Range Number of Places EUA 500,000+ 34 UA 50,000 - 499,999 689 UC 2,500 - 49,999 5401 Moderately Rural 1,500 - 2,499 1,408 Rural 500 - 1,499 3,253 Extremely Rural 0 - 499 3,683 Decennial data and the American Community Survey from the Census Bureau were also used to obtain mean values of the same variables in Detroit at the C ensus tract level. Removing census tracts that did not have complete information ac ross all variables left a data set of 291 trac ts. The mean values were then compared to those among incorporated places to determine where Detroit fits on a rural - urban spectrum. 3.2 Results T o understand where Detroit might fit on a rural - urban spectrum, I considered several socioeconomic characteri stics that the literature indicates vary between rural and urban areas. With a total population of 683,4 43 people in 2016, Detroit would not classify as rural based solely on population size . Population density is another common indicator used to define a place as rural or urban. According to the USDA, a place is considered rural if there is a population density lower than 1,000 people per square mile. With an average population density of 4,92 6 people per square mile in 2016, Figure 1 sho ws that Detroit was largely considered urban based on this criterion. However, this poses the ques tions of how urban is Detroit a nd how closely does it resemble other large cities? Later in the paper, Detroit wil l be considered a single entity. But to begin the evaluation, it will be considered on the Census tract level in this portion of the chapter. 17 Figure 1: Rural Classification based on Population Density in Detroit In Figure 2, a verage population densities were taken in incorporated places in six different population ranges (see legend) and then compared to the population densities within Detroit Census Tracts. The darkest shade of red indicates Census tracts that ha ve population densities that are closest to those in places with total populations population densities comparable to places with smaller total populations, with shades of green denoting rural populations. Table 4 indicates that in 2016, about 42% of Detroit Census tracts have population densities close to or below those in les s - populated incorporated places, which is thirteen percentage points, (38 Census tracts) , higher than in 2010. & Grant Program is to have a population of no more than 50,000 people. Figure 2 and opulation densities that correlate with places with populations less than 50,000 people. This 2016 2010 18 indicates that although Detroit does not qualify for such programs as a single entity, some areas may benefit from policies like those designed by the USDA for ru ral areas. F igure 2: Comparing Detroit Census Tracts to Population Ranges throughout Incorporated Places in the U . S . - based on Population Density Table 4 : Number of Detroit Census Tracts that Correlate with Population Ranges Year Less than 500,000 people Less than 50,000 people 2010 85 44 2016 123 56 Rural areas tend to have higher percentages of vacant households than urban areas. In 2016, about 92% of Detroit Census tracts would have been classified as rural with household vacancy rate s of at least 12.98% compared to the national urban aver age household vacancy rate of about 10.54%. rate almost triple d that with a value of about 31%. Figure 3 and Table 5 presents this 2016 2010 19 information . Note that although the city has recently made efforts to destroy vacant structures (Reese et al . , 2017) , the current rate of vacancies remain s higher than 2010 average residential vacancy rate of about 23% . Simi larly, referring to Figure 4 and Table 6, i n 2010, about 85% of Detroit had a renta l vacancy rate of at least 10.6 %, which is more consistent with rural areas . The average among urban places was only about 9 %. Figure 3 : Rural Classification based on Percent of Vacant Households in Detroit Table 5 : Average Vacant Household Rates Year Average Vacant Household Rates in Urban Places Average Vacant Household Rates in Moderately Rural Places Detroit Average Number of Detro it Census Tracts Classified as Rural 2010 9.67 11.4 23.37 267 2016 10.54 12.98 31 268 2010 2010 2016 20 Figure 4: Rural Classification based on Rental Vacancy Rate in Detroit Table 6 : Average Rental Vacancy Rates Year Average Rental Vacancy Rate in Urban Places Average Rental Vacancy Rate in Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 9.11 10.61 17.64 248 Rural places tend to be worse off financially than urban places. In 2016, the average mean household income in rural places was about $58,674 compared to $7 1,962 on average in urban places. about $37,300, whic h is substantially lower than the national average. About 94% o f Detroit Census tracts would have been classified as rural. Figure 5 and Table 7 presents this information. 2010 21 Figure 5 : Rural Classification based on Mean Household Income in Detroit Table 7 : Mean Household Incomes Year Average Mean Household Incomes in Urban Places Average Mean Household Incomes in Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 $66,259.29 $52,790.04 $37,755.17 266 2016 $71,965.61 $58,674.14 $37,298.67 276 Similar results are shown for another measure of income, per capita income. In 2016, the average per capita income in rural places was about $24,342 compared to the average of about $28,016 in urban places. su bstantially lower at about $15, 473. About 90% of Detroit C ensus tracts would have been classified as rural . Figure 6 and Table 8 presents this information. 2010 2016 22 Figure 6: Rural Classification based on Per Capita Income in Detroit Table 8: Per Capita Incomes Year Average Per Capita Incomes in Urban Places Average Per Capita Incomes in Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 $25,920.06 $21,841.65 $15,064.08 265 2016 $28,016.47 $24,341.93 $15,472.99 271 Educational attainment levels tend to be lower in rural areas compared to urban areas. In 2016, about 85% of Detroit Census tracts would have been classified as rural, where, at the most 19.96% of their population had an educational attainment level of at , about 27% of the population obtained at least a Figure 7 and Table 9 presents this information. 2010 2016 2010 23 Fi gure 7: Rural Classification based on Percent Degree or Higher in Detroit Table 9: Educational Attainment Year Average Percent of Population with a Bachelor's Degree or Higher in Urban Places Average Percent of Population with a Bachelor's Degree or Higher in Moderately Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 25.42 18.23 11.7 243 2016 27.15 19.96 12.94 246 Using the Herfindahl - Hirschman Index (HHI) a s a model of industry diversity, a rating of industry composition was calculated for all incorporated places and Detroit census tracts in the data sets. (1) w here is the industry composition ranking in place and ranges from near zero to 10,000. represents the employment share for industry in place , accounting for all 13 industries categorized in the Census data. This model assigns a ranking value that 2016 2010 24 represents the level of diversity among industrie s, where the higher the value, the less diverse the economic activities. Rural places tend to have less diversity among economic activities (Baldwin, Vinodrai, and Brown, 2001) , and thus would be expected to have higher industry composition rankings. This was true for both 2010 and 2014. In 2016, there was a statistically significant difference in industry composition rankings only between rural and urban communities not across the average industry composition ranking of about 1,600 while urban places had an average industry composition ranking of about 1,400. About 6 1 % of the Census tracts would have been classified as rural . Figure 8: Rural Classification based on Industry Composition in Detroit Table 10: Industry Composition Year Average Industry Composition Rating in Urban Places Average Industry Composition Rating in Moderately Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 1370.99 1464.2 1739.9 205 2016 1407.8 1598.7 1642.2 129 2010 2016 25 The manufacturing industry has becom e more common in rural areas. In 2016, rural places had a mean value of about 15% of their population being employed in the manufacturing industry, as opposed to a mean value of about 12.4% in urban places. About 47% of Census tracts in Detroit would have classified as rural. Note tha t there has been an increase in the share of employment in the manufacturing industry since 2010 , as shown in Figure 9 and Table 1 1 . Figure 9: Rural Classification based on Percent o f Population Employed in the Manufacturing Industry Table 11: Manufacturing Industry Year Average Percent of Population Employed in the Manufacturing Industry in Urban Places Average Percent of Population Employed in the Manufacturing Industry Rural Places Detroit Average Number of Detroit Census Tracts Classified as Rural 2010 12.68 15.45 13.75 127 2016 12.41 14.95 14.31 136 The literature states that rural areas tend to have higher poverty and unemployment rates. Although there was a statistically significant difference in the 2010 2016 26 complete data set obtained from the Census Bureau, the sample of incorporated places did not show a statistically significant difference between rural and urban areas for either variable. So rather than displaying maps of rurality, Figures 10 and 11 illustrate how Detroit compare s to national averages. They clearly illustrate that Detroit has above - average unemployment and poverty rates, and as shown in Table 12, both rates in Detroit are abou t three times larger than the average national rates in 2016. Figure 10 : Detroit Census Tracts above the National Average Poverty Rate Figure 1 1 : Detroit Census Tracts above the National Average Unemployment Rate 2016 Below the National Average Above the National Average Below the National Average Above the National Average 2016 27 Table 1 2 : Average Poverty and Une mployment Rates Year National Average Poverty Rate Detroit Average Poverty Rate National Average Unemployment Rate Detroit Average Unemployment Rate 2010 11.08 31.32 4.77 13.97 2016 12.05 36.038 4.21 12.11 3.3 Outlier Analysis Methods Population density is often used to determine whether a place is rural or urban (Castle and Weber, 2011) , and therefore which redeve lopment policies are prescribed to that area. The analysis explained in section 3.2 show s that, based on several non - population rural indicators, Detroit compares heavily with rural communities. Therefor e, it can be inferred that population size and density should not be the only varia bles used to determine where the city lies on a rural - urban spectrum. T o further explore this issue, using the 2016 incorporated places data set, I regressed population dens ity on several other rural indicators and used them to predict values for population density. How these predictions compare to actual population density in Detroit will tell us how well of a fit population density is as a proxy for rurality. The first anal ysis is a simple regression, only regressing the natural logarithm of population density, ln ( popdens ) , on the natural logarithm of total population, ln(total) and a constant. Natural logs and robust standard errors were used to co rrect for heteroskedasticity. (2) 28 where is constant, is the coefficient on the natural logarithm of total population for place , and is the error term. The predicted values from this regress ion reveal expected population density based only on total population. Next, a full regression was computed with several other variables that , based on the literature and above analysis , are rural indicators. (3) where is the coefficient on , percent of vacant households, for place . is the coefficient on , the natural lo garithm of per capita income, for place . is the coefficient on , percent of the population with an educational attainment of 1 , for place . is th e coefficient on , the natural logarithm of the calculated industry composition variable, for place , and is the coefficient on , percent of the population employed int eh manufacturing industry, for place . A differenc e in predications from Equations 2 and 3 would infer that the other included variables are in fact rural indicators. Finally, the full regression, Equation 3, was estimated again but minus total population. (4) 1 It was assumed that there would be a U - degree or higher and population total and/or density, with low educational attainment levels in both rural and highly populated areas. However, no s uch pattern was determined. 29 This model is estimated to compar e predicted densities with and without population size. A significant difference between the predicted values between Equations 3 and 4 suggests that the demographics, or soci oeconomic characteristics, are also significant indicators of the rural - urban divide. I t should be noted that population densities of incorporated places are not normally distributed, but rather exponentially distributed toward low - densi ty places, and therefore has a tendency to favor under - predictions. 3.4 Outlier Analysis Results Equations 2, 3, and 4 were estimated using the incorporated places data set ; see Table A 3 , in the appendix, for the regression results population was 683,443 with a population density of 4,926 pe ople per square mile. Equation 2 predicts a population density of 6,963 people per square mile, a difference of 2,037 people, showing that based on population si ze alone , Detroit is predicted to be much more densely populated than it is. Although the difference between predicted and actual population density is not among the largest in terms of percentage , the substantial difference is among the largest 10% in abs olute terms. This is also quite predict ed population densities of larger metropolitan areas . However, once other factors are included in the regression as reflected by Equation 3 , predicted populatio n density falls to 3,802 people per square mile. This predicted value now underestimates the actual value by about 1,123.3 people. This shows that when taking into account socio economic factors, Detroit is more like a rural community. The large predicted value from Equation 2 infers that population size alone is not enough to accurately predict population density. The lower 30 prediction from Equation 3 indicates that once other indicators are included, Detroit appears to be more similar t o areas with smaller populations. Since population density correlates strongly with total pop ulati on, researchers often use population density as a proxy for rurality. However, the difference in predicted values between Equations 2 and 3 suggests tha t it i s useful to also consider other socio economic factors to determine the level of rurality. Finally, by removing total population as a regressor, Equation 4 further supports this theory. Based solely on the non - tion density is predicted to be 795 people per square mile. This is within the Census population density (979 people per square mile) of communities with a total population o f fewer than 2,500 people. The economic and social characteristics of Detroit are not what would be expected for a city of its size and population density. In fact, soci oeconomic factors alone predict Detroit to be rural. It therefore may be prudent to consider whether rural redeve lopment policies would be more appropriate for Detroit and other declining urban places. The results indicate that using population size or density alone as a proxy for rurality may not offer a complete assessment. The regression based on E quation 2 showed that although there is a strong relationship between the two variables, it is not enough to imply that population density can be estimated solely based on po pulation size . Using population size alone could lead to the conclusion that Detroit is extremely urban. Equations 3 and 4 indicate that including other rural indicators into the model shows that Detroit, in some respects, reflects the characteristics of r ural places. 31 3. 5 Cluster Analysis Several variables correlate with rurality. These variables can also be used to predict membership among calculated clusters with varying levels of rurality. T o further examine just how rural the city is, a k - means partition cluster method was estimated to determine wh ich incorporated places Detroit is most similar to. This technique group s observations based on common traits. Incorporated places were grouped into ten clusters 2 using the following rural indicators as traits: population density, percent of vacant housing, per capita income, percent of population with an educational attainment population employed in the manufacturing industry and average household size. These clusters are evaluate d using data from 2010 (Tables 13 and 14) and 2016 (Tables 15 and 16 ). In 2010, Detroit was placed in cluster 2 , among places with an average populati on of 3,445 people and a population density of 1,034 people per square mile, the lowest mean population density among all clusters. Table 13 presents this information. These figures are very close to the criteria used to indicate a rural place and only one cluster has a smaller average population size. Table 14 shows that highest average percent of the population employed in the manufacturing industry, the highest av erage vacancy rate, and the lowest average per capita income and educational attainment levels. 2 As a robustness check, the cluster analysis was performed repeatedly with numbers of clusters ranging from five to 15 and Detroit was grouped among the same places for all cluster amounts. 32 Table 1 3 : Summary of 2010 Clusters Cluster Sample Size Mean Population (people) Average Population Density (people per square mile) 1 47 2423.3 1110.3 2 1,841 3445.2 1034.7 3 3,994 5489.7 1107.4 4 139 5874.7 1922.3 5 3,985 9734.1 1264.8 6 251 11529.8 2788.9 7 2,312 20709.9 1603.1 8 540 22344.4 2898.4 9 1,138 23965.8 2247.2 10 221 107255.7 10243.6 Table 14 : Means of Select Features in 2010 Clusters Cluster Percent of Population Employed in Manufacturing Industry Per Capita Income Percent of Vacant Households Percent of Population with a minimum of a Bachelor's Degree 1 8.7 $118,783.00 13.8 73.8 2 14.8 $13,301.00 15.6 9 3 16 $17,845.61 13.2 12.6 4 8.3 $83,814.92 14.9 69.3 5 15.2 $21,764.07 11.4 16.4 6 8.7 $58,963.17 11.8 60.8 7 13.2 $26,314.21 10.3 22.3 8 10 $43,101.37 11.9 48.3 9 11.1 $32,883.75 9.8 34.1 10 11.7 $20,191.60 8.6 17.8 Clusters estimated with the 2016 data produced results similar those in 2010 . Detroit was also sorted into cluster 2 with the second smallest mean population size of 4,340 people and the smallest population density of 1,130 people per square mile. Table 15 presents this information. Again, Detroit has the largest population in its cluster 33 with a total population of 683,433 people. Laredo, Texas , has the second largest population size in the cluster with only 251,671 people. Detroit is also much larger than Laredo in terms of population density. Table A 4 , in the appendix, provides a list of the places in cluster 2 with the twenty largest populatio ns and their corresponding population densities . Table 16 indicates that only eighteen places have populations larger tha n 50,000 people and about 68% of the cluster, 1,358 places, are considered for rurality . Also, Detroit is in the 95 th percentile for population density within the cluster . Although it is not the largest outlier, there are only thirty - five places in cluster 2 with larger population densities. Almost 1,200 places have population densities fewer than 1,000 people per square mile, It is clear that Detroit is not representative of the average place in cluster 2 based o n population size and density . Consequently , other Detroit characteristics must be similar to those of cluster 2 . Table 1 5 : Summary of 2016 Clusters Cluster Sample Size Mean Population (people) Average Population Density (people per square mile) 1 38 2,447.8 1,438.7 2 1,985 4,340.3 1,130.0 3 123 6,552.5 2,117.9 4 4,008 7,351.8 1,204.8 5 240 9,792.8 2,680.0 6 3,968 11,142.2 1,318.0 7 2,396 19,670.8 1,678.8 8 1,032 24,333.7 2,265.0 9 514 24,471.3 3,043.6 10 164 113,288.9 11,567.5 34 Table 16: Size Distribution of Places in Cluster 2 for 2016 Population Size (in people) Number of Places Population Density (people per square mile) Number of Places Less than 1,000 968 Less than 1,000 1,194 1,000 2,499 390 1,000 2,999 670 2,500 49,999 609 3,000 4,999 86 50,000+ 18 5,000+ 35 Hence, sociodemographic traits within are very rural. As presented in Table 17, cluster 2 has the lowest per capita incomes and educational attainment levels and the second highest percent of population employed in the manufacturing industry and residential vacancy rate. Table 1 7 : Means of Select Features i n 2016 Clusters Cluster Percent of Population Employed in Manufacturing Industry Per Capita Income Percent of Vacant Households Percent of Population with a minimum of a Bachelor s Degree 1 7 $132,862.70 18.3 74.2 2 14.8 $14,774.38 18.0 10.4 3 7.5 $91,815.67 13.1 72.6 4 15.4 $19,844.70 14.8 14.0 5 9 $67,040.19 13.8 66.5 6 14.6 $24,306.03 12.4 18.1 7 12.8 $29,392.99 11.3 24.9 8 10.7 $37,034.97 10.5 38.0 9 9.4 $48,724.76 11.6 52.6 10 10.9 $20,429.46 9.6 18.5 As a robustness check, to ensure Detroit was sorted into an appropriate cluster with other similar places, Equation 3 was used 35 well as all other variables by rearranging the model five times, using the places data set, regressing each vari able individually on all the others. (5) (6) (7) (8) (9) Then for each variable, a z - score test was used to determine how many standard deviations away was the predicted value from the mean value in the 2 nd cluster. (10) where is the number of standard deviations away from the variable mean, within is the predicted value for Detroit and is the standard deviation of the variable mean within the cluster. 36 Table 1 8 2016 Cluster All the predicted values were within one sta ndard deviation of the respected variable mean within the cluster except per capita income. It should b e noted that based on population density, Detroit was in the 95 th percentile , reinforcing that rep resentative of the average place in the cluster based on population density. This t to be sorted into the cluster and thus , char acteristics do not resemble th o se of an average highly populated urban area. These results are similar to those in Section 3.4, indicating that population density and population size alone are not appropriate proxies for rurality. Variable Cluster Mean Detroit Prediction De troit Actual popdens 1,144.6 795.0 4,925.7 vacanthh 18.0% 8.4% 29.8% pcincome $14,721.04 $19,483.57 $15,562.00 educ 10.4% 10.6% 13.8% indcomp 1,719.0 1,555.3 1,305.3 manuf 14.8% 9.1% 14.5% 37 CHAPTER 4: POLICY IMPLICATIONS AND CONCLUSION As Cowan (2010) notes, among the foc al points in USDA Rural Development programs are adequate housing, generating employment opportunities , sustainable business development, and improv ing human capital and poverty rates. These are all objectives that are of concern for Detroit . W ith the curren t eligibility requirements, it do es not qualify for such programs. To qualify for USDA Rural Development programs , the place must have a population density of fewer than 1,000 people per square mile and a total population of fewer than 50,000 people. Detro it does not meet these criteria , but as this analysis conclude s , the sociodemographic characteristics within the city are actually closer to rural areas than urban areas . Referencing Figure 2, in section 3.2, even the population density in much of the region is similar to that in much smaller cities , including areas with total populations less than 50,000 people. And as the models estimate, based on social and economic traits, Detroit is similar to places much smaller than itself. Equations 3 and 4 use rural indicators to predict a population density for Detroit that is significantly smaller than the actual value and even further away than the prediction estimated from Equation 2 using population only as a fact or. These findings suggest that population density alone is not an appropriate proxy for rurality. This is further supported by Equation 4 , where only non - population sociodemographic factors are considered, predicting a population density of 7 95 people per square mile for Detroit, which is below the requirement of eligibility for USDA Rural Development programs, and the cluster analysis grouping Detroit among rural communities defined by the Census Bureau. 38 D ecades of struggling with racial and class confli cts, suburbanization, the decline of the automobile industry, and weak governance have put the c i ty in a state of economic distress and transformed the population into one that , in some respects, resemble s a rural commun ity . Thus, it may useful to consider applying what is thought of as traditional rural economic development policies in Detroit . As this analysis has shown , unemployment and poverty, low levels o f human capital, lack of employment and educational opportunities, and high residential vacancy rates. All of these traits are common in rural areas and thus often major points of focus in rural redevelopment policies. (D etroit City Council , 2009) is a working document that outlines the major issues local government should be concerned with, goals for improvement, and policy theme suggestions. The report makes clear that these goals do not have timelines and cannot all be addressed at the same time . I t is the responsibility of elected officials to prioritize the concerns and choose policies based o n available resources. This work suggests that local government consider both the traditional urban development policies as wel l as policies typically thought of as most appropriate for rural places. In short, policymakers may want to consider redevelopment through both policy lenses to The Master Plan has done a good job of outlining these specific needs and ha s suggested policy angles that would traditionally resemble those from either end of the rural - urban spectrum. Now it is in the hands of local government to prioritize the most pr essing concerns and choose appropriate policies. 39 T he Master Plan presents policies that suggest a continued focus on the downtown area while others suggest focusing on underserved neighborhoods. Downtown and Midtown Detroit have seen the majority of recen t development, whereas the rest of the neighborhoods are being left behind (Reese et al . , 2017 ). These underserved neighborhoods may be e xhibiting rural characteristics because of their lack of access to the growing resources and amenities of Downtown and Midtown. Increasing employment or educational opportunities will have minimal effect if the residents who need them most do not have access to the m. Reese (20 14 ) found that cities still tend to employ traditional economic development policies by investing in basic infrastructure, offering tax incentives and implementing development zones. Detroit has already attempted conventional urban business development plan s, with limited success (Reese, 2014; Reese, Eckcert, Sands, and Vojnovic, 2017 ) . As discussed earlier, although tax incentives and development zones are common urban development strateg ies for helping to generate employment opportunities, such approach es may not be the most effective for Detroit . As previously noted, rural development ap proaches are people - and community - based , placing the needs of the residents as a first priority . challenges close to those of rural residents, such as high unemployment rates partly due to a lack of employment opportunities. that focus on rebuilding the automobile industry as well as diversifying the economy. As previously d iscussed , overreliance on the automobile industry played a large role in et al . , 2017; Reese, Sands, and Skidmore, 2014). D iversifying 40 into other industries is very important for sustainable growth. Pittelko, Bommersbach, and Erickcek (2016) conducted a review of the New Economy Initiative (NEI), a collaboration of 10 foundations pledging $100 million in 2007 to aid economic growth through entrepreneurship and small business development in Southeast Michigan. An estimated total of 17,4 90 direct and indirect jobs were created. Out of the 7,468 direct jobs generated from the initiative, there was an average annual salary of about $44,000 . A bout 51% of these were in professional, scientific, and technical services. The effectiveness of the se efforts support s the notion that public development policies that focus on small and midsize business may generate more employment opportunities for residents, contributing to a health ier and more sustainable economy. However, the Master Plan also suggests offering tax incentives. A wide variety of literature urges cities to take precaution or completely steer away from this strategy (Loveridge, 1996) . Although tax abatements are among the oldest and most commonly used incentive programs nationally (Reese and Sands, 2013), many academics have found them, and other specialized development zones with tax incentives , to be ineffective and expensive (Peters and Fisher, 2004) . Any improvements from past programs were found to be marginal and did not generate long - term effects (McArthy 1998). Studies found that tax abatements were being used by existing facilities and did not successfully increase employment , and any positive effects that occurred in Renaissance z ones, a program used in Detroit , did not spill over into surrounding areas (Reese, 2014) not see any spillover effects from the investments in Downtown and Midtown; employment did not rise and vacancy did not decrease (Reese, et al . , 2017). 41 Detroit needs policies that w ill improve the well - being of the residents in these underserved neighborhoods, and many studies suggest that large investments in specialized zones will not result in positive effects for the overall population. Kang, Skidmore, and Reese (2013) also found that tax abatements were not promising for spillover effects and that such programs have a minimal positive effect in relation to the high costs. In 2013, Reese conducted research among Michigan cities that determined there w as not a statistically significant relationship between residential economic health and tax abatements. This conclusion supports a large array of other studies (Fisher & Peters, 1998; Peters & Fisher, 2004; Sands & Reese, 2012; Wassmer & Anderson, 2001). McArthy (1998) emphasized that tax incentives more often than not attract large, individual firms that are generally branch plant relocations. These firms tend to employ those who have relocated with the business and have little effect on local unemployment rates. Smaller, locally owned businesses are much more likely to employ residents and generate more revenue for the local economy. Bartik (2018) also agrees that tax incentives are expensive and focus should be placed on small and medium - sized manufacturers. Michigan cities have a persistent emphasis on tax abatements and enterprise zones , but they will not aid Detroit in developing the diverse economy they need , and there is little evidence to support the assumption t hat they will attract business development (Reese and Sands, 2007 ). High unemployment rates and the are indicators that employment generation and conserving an inflow of taxes could be very beneficial to the city . It may be useful for policymakers to consider this and focus on suggestions that support local businesses. 42 opportunities, but also because of the low - skilled labor force . Reside nts have limited access to job skills training and quality education (Reese and Sands, 2007). Tens of thousands of students have left the Detroit p ublic s chool system, which is surprising in li ght of the very low standardized test scores (Reese et al, 2017). Local government needs to focus on improving educational attainment and skill level to enhance the quality of the labor force. This is a concern more commonly addressed in rural development policies. The Master Plan has several policy suggestions t hat appear to be influenced by rural economic development, such as increasing opportunities and quality of business education and training, supporting start - ups, providing additional support for working families in the form of child care, transportation, a nd access to support goods. They also sugges t encouraging local institutions to offer training courses for loca l residents . On the human resource side , the Master Plan suggests increasing additional educational opportunities for several specialized populat ions such as adults, immigrants, and at - risk youth, a long with early childhood development. Educational i nstitutions can be great resources to stimulate the whole community. An economic impact study by Erickcek and Pittelko (2015) revealed that North Central Michigan College generated employment opportunities and training programs, access to higher education, and increased revenue for the region. The large percentages of vacant households correlate strongly with rural area s. This trait is interesting because although it is a rural characteristic, traditional urban development policies may be more effective for this concern. The APA Policy Guide on Public Redevelopment explains that reusing and redeveloping underused sites i s a 43 practice more commonly found in urban planning but could be very beneficial to rural areas as well. The Master Plan offers several potentially effective ways of turning the blight into functional spaces for the community, such as attracting entrepreneu rship. The vast amount of vacant land in Detroit leaves a lot of potential for business development. Improving these spaces in distressed neighborhoods will make them more developable, increasing the opportunity for business investment, and could bring emp loyment opportunities to the residents (Bartik, 2018). These spaces could also be offered to public schools or other community organizations. Officials noted increasing community programs in the Master Plan. If used as community gardens or public green spa ce, neighborhood s would benefit from improved air quality and quality of life from increased natural habitat. This excess of vacant land is a great example of how traditional urban and rural development policies would need to be combined and adapted to be st suit the city. Rural development economics stress the notion that every place should be observed uniquel y and should have redevelopment policies specially designed for them. This is true for Detroit as well. There are many factors that set the city apart from other urban areas, including the many rural indicators discussed in this thesis . Detroit may not fit into the definition of a rural place defined by the Census Bureau, but this study infers that the economy and social demographics of th is distr essed city relate more to a rural place than to an urban one. Thus, officials should be looking toward rural development approaches as an example for economic redevelopment policies . 44 APPENDI X 45 APPENDI X Table A1: Su mmary Statistics of Rural Indicators in Census Places, FY 2016 Mean Standard Deviation Min Max Total Population (people) 13,059 96,304.80 9 8,461,961 Population Density (people per square mile) 1,595.36 1,949.78 1.78 53,766.98 Unemployment Rate 4.20% 2.74 0 34.40% Mean Household Income $64,319.03 34,129.03 $18,030 $727,189 Per Capita Income $25,904.30 12,621.45 $3,284 $261,848 Percent of Population with a Minimum of a Bachelor's Degree 21.11% 15.19 0 92.40% Residential Vacancy Rate 13.51% 10.57 0 98.40% Residential Rental Vacancy Rate 6.44% 8.71 0 100% Industry Composition Rating 1517.45 460.51 906.91 10,000 Employment Share in Manufacturing Industry 13.88% 8.84 0 100% Poverty Rate among Families 12.05% 9.13 0 100% Number of Observations: 14,468 places 46 Table A2: Mean Values of Rural Indicators in Census Places Urban Rural Detroit Population Density 2,435.7 978.6 4,925.7 (people per square mile) (2,387.4) (963.6) (5,144.3) Residential Vacancy Rate 10.5 15.7 31.0 (9.7) (13.9) (23.4) Residential Rental Vacancy Rate - - - (9.1) (10.6) (17.6) Percent of Population with a 27.2 16.7 11.7 (25.4) (15.2) (12.9) Mean Household Income (dollars) $71,965.61 $58,674.14 $37,298.67 ($66,259.29) ($52,790.04) ($37,755.17) Per Capita Income (dollars) $28,016.47 $24,341.93 $15,472.99 ($25,920.06) ($21,841.65) ($15,064.08) Industry Composition Rating 1,408 1,598 1,642 (1,371) (1,658) (1,740) Employment Share in 12.4 15.0 14.3 Manufacturing Industry (12.7) (15.5) (13.8) Number of Observations: 14,468 places Values for FY 2010 in parenthesis 47 Table A 3 : Regression Results Equation 2 Equation 3 Equation 4 ln(total) 0.330*** 0.271*** - ( - 0.0036) ( - 0.00481) ln(pcincome) - - 0.204*** - 0.770*** ( - 0.0338) ( - 0.0326) ln(Indcomp) - - 0.0932** - 0.826*** ( - 0.0377) ( - 0.0343) manuf - - 0.00246*** - 0.00234*** ( - 0.000842) ( - 0.000908) vacanthh - - 0.0178*** - 0.0270*** ( - 0.000846) ( - 0.000739) percbach - 0.00898*** 0.0306*** ( - 0.000876) ( - 0.000802) Constant 4.414*** 7.680*** 20.46*** ( - 0.028) ( - 0.491) ( - 0.435) R - squared 0.368 0.406 0.247 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 48 Table A 4 : Places in Cluster 2 with the Twenty Largest Populations for 2016 Place Total Population Population Density Detroit city, Michigan 683,443 4,926 Laredo city, Texas 251,671 2,831 San Bernardino city, California 214,581 3,625 Brownsville city, Texas 182,110 1,376 Victorville city, California 121,320 1,658 Rialto city, California 102,418 4,582 Flint city, Michigan 98,918 2,960 Hesperia city, California 92,664 1,268 Nampa city, Idaho 87,896 2,818 Gary city, Indiana 77,858 1,561 Pharr city, Texas 75,172 3,210 Perris city, California 73,718 2,348 Madera city, California 63,398 4,015 Pontiac city, Michigan 59,920 3,000 Porterville city, California 58,472 3,321 Delano city, California 52,538 3,673 Elkhart city, Indiana 52,378 2,233 Caldwell city, Idaho 50,288 2,279 Saginaw city, Michigan 49,892 2,878 Pine Bluff city, Arkansas 45,404 1,019 49 REFERENCES 50 R E FERENCES American Planning Bachmura, F. & Glasgow, R. 1 961. Rural Area Development in a Growing Economy. Journal of Farm Economics : 43 (5), 1532 - 1537. Baldwin, J. R., Vinod rai , T., and Brown, W. M. 2001. Dynamics of the Canadian manufacturing sector in metropolitan and rural regions. Statistics Canada, Analytical Studies Branch . Bartik, T. J. 201 8. Helping Manufacturing - Intensive Communities: What Works? W.E. Upjohn Institute for Employment Research: Prepared for Center on Budget and Policy Priorities (CBPP). Burchinal, L., & Siff, H. 1 964. Rural Poverty. Journal of Marriage and Family, 26 (4), 399 - 405. Castle, E., Wu, J., & Weber, B. 2011. Place Orientation and Rural - Urban Interdependence. Applied Economic Perspectives and Policy : 33 (2), 179 - 204. Cowan, T. 2010. An overview of USDA Rural Development programs. Washington, DC . Cruickshank, J. 2018. Is culture - led redevelopment relevant for rural planners? The risk of adopting urban theories in rural settings. International Journal of Cultural Policy : 24:3, 331 - 349 . Davis, E., Grobe, D., & Weber, R. 2010. Rural - Urban Differences in Childcare Subsidy Use and Employment Stability. Applied Economic Perspectives and Policy : 32 (1), 135 - 153. Daft, L. 1971. Public Policies for Rural America: Legacies or Leading Edges? American Journal of Agricultural Economics : 53 (2), 250 - 256. Dewitt, J. 1993. Where Is Rural Policy Headed? The Annals of the American Academy of Political and Social Science : 529 , 22 - 33. Dillman, D., & Tremblay, K. 1977. The Quality of Life in Rural America. The Annals of the American Academy of Political and Social Science : 429 , 115 - 129. Domonoske, Camila. 2018. Detroit Released From Financial Oversight, 3 Years After Emerging From Bankruptcy. NPR; the two - way. Drabenstott, M. 1 995. Rural Development: Implications of Structural Change for Policies and Institutions: Discussion. American Journal of Agricultural Economics : 77 (5), 1271 - 1273. 51 Duncan, C., & Tickamyer, A. 1988. Poverty Research and Policy for Rural America. The American Sociologist : 19 (3), 243 - 259. Erickcek, G. and Pittelkfo, B. 2015. Economic Impact of North Central Michigan College. W.E. Upjohn Institute for Employment Research: Report prepared for North Central Michigan College. Fisher, P. S., & Peters, A. H. 1998. Industrial incentives: Competition among American states and cities. Halfacree, K. H. 1993. Locality and social representation: space, discourse and alternative defin itions of the rural. Journal of rural studies : 9 (1), 23 - 37. Hansen, N. 1969. Regional Development and the Rural Poor. The Journal of Human Resources : 4 (2), 205 - 214. Hart, L. G., Larson, E. H., & Lishner, D. M. (2005). Rural definitions for health policy and research. American Journal of Public Health : 95 (7), 1149 - 1155. Hofferth, S. L., & Iceland, J. 1998. Social capital in rural and urban communities. Rural sociology : 63 (4), 574 - 598. Jolliffe, D. 2003. Comparisons of metropolitan - nonmetropolitan poverty during the 1990s (No. 96). Washington, DC: US Department of Agriculture, Economic Research Service. Kang, S., Skidmore, M., an d Reese, L.A. 2013. How Effective are Tax Abatements?: The Case of Michigan. Journal of Poli cy Analysis and Management: 35(2): 388 - 414. Leo, C. and Brown, W. 2000. Slow Growth and Urban Development Policy. Journal of Urban Affairs. 22(2): 193 - 213. Loveridge, S. 1996. On the Continuing Popularity of Industrial Recruitment. Economic Development Quarterly : 10(2): 151 - 158 . Martin, L. 196 6 . Effects of Alternative Federal Policies on Welfare of Rural People. Journal of Farm Economics : 48 (5), 1267 - 1276. McCarthy, J. 1998. US urban empowerment zones. Land Use Policy, 15(4): 319 - 330. Mills, B., & Hazarika, G. 2003. Do Single Mothers Face Greater Constraints to Workforce Participation in Non - Metropolitan Areas? American Journal of Agricultural Eco nomics : 85 (1), 143 - 161. Modi, P. 2009. Rural marketing: Its definition and development perspective. International Journal of Rural Management : 5 (1), 91 - 104. Neuman, Scott. 2014. NPR; the two - way. Padnani, A. 2013. The New York Times 52 Perry, C. 1984. Economic Activity and Social Indicators: A Rural - Urban Discontinuum? The American Journal of Economics and Sociology : 43 (1) : 61 - 74 . Peters, A., & Fisher, P. 2004. The failures of economic development incentives. Journal of the American Planning Association : 70 (1) : 27 - 37. Pittelko, B., Bommersbach, and Erickcek, G. 2016. The Employment Impact of the New Economy Initiative (NEI) on the D etroit Region and the State of Michigan. W.E. Upjohn Institute for Employment Research: Prepared for the New Economy Initiative, A special project of the Community Foundation for Southeast Michigan Reese, L. A. 2013. If All You Have Is A Hammer: Finding Economic Development Policies That Matter. American Society for Public Administration: 44(6) . Reese, L. 2014. The Alchemy of Local Economic Development. Economic Development Quarterly : 28 ( 3 ): 206 - 219 . Reese, L., Eckert, J., Cities : 60(2017): 367 - 377. Reese, L. and Sands, G. 2007. Making the least of our differences? Trends in local economic devel opment on Ontario and Michigan, 1990 - 2005. Canadian Public Administration : 50(1): 79 . Reese, L.A., Sands, G., 2012. Money for nothing : industrial tax abatements and economic development. Lexington Books, Lanham, Md. Reese, L. and Sands, G. 2017. Is Detroit Really Experiencing a Full Comeback? Citylab. Reese, L., Sands, G. and Skidmore , M. 2014. Memo to Motown: is austerity here to stay? Cambridge Journal of Regions, Economy, and Society: 2014(7): 99 - 118. Reese, L. and Ye, M. 2015. Trends in Local Development Policy . ICMA Municipal Yearbook : (2015): 15 - 26 . Rural - Urban Contrasts. 1981. Economic and Political Weekly : 16 (22), 967 - 968. Sutton, A. 2008. Urban Revitalization in the United States: Policies and Practices; Final Report. United States Urban Revitalization Project (USURRP). Vojnovic, I. and Darden, J.T. 2013. Class/racial conflict , intolerance, and distortions in urban form: Lessons for sustainability from the Detroit region. Ecological Economics: 96(2013): 88 - 98. Wassmer, R.W., Anderson, J.E., 2001. Bidding for Business: New Evidence on the Effect of Locally Offered Economic Deve lopment Incentives in a Metropolitan Area. Economic Development Quarterly : 15 : 132 148.