JCIT-01751; No of Pages 8 Cities xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Cities journal homepage: www.elsevier.com/locate/cities
Specialized vs. diversified: The role of neighborhood economies in shrinking cities James Murdoch, III 2M Research Services, LLC, 500 E. Border St. Ste. 680, Arlington, TX 76010, United States
1. Introduction A recent strain of research highlights the problems faced by “shrinking cities” that have experienced prolonged population loss and problems associated with the loss of jobs and industry (Oswalt, 2006; Martinez-Fernandez, Audirac, Fol, & Cunningham-Sabot, 2012). Research in this vein challenges assumptions of unfettered growth and argues that decline and shrinkage are inevitable in the global economy (Audirac, 2009; Martinez-Fernandez et al., 2012; Rieniets, 2009). Developed economies in particular have experienced a concentration of economic success in global cities that support financial, telecommunication, and other advanced services, while cities that once prospered from manufacturing and other production-oriented firms have experienced job loss and abandonment (Sassen, 2001). In the U.S., classic examples include Detroit and Flint, MI and Buffalo, NY. In response, many urban planners, public officials, and other stakeholders question neoliberal forms of governance that rely on marketoriented solutions and fierce competition that inevitably leave many cities in a state of decline. Researchers are suggesting planning that encourages ground-up, cooperative, and participatory solutions that are more sustainable in the long-run (Dewar & Thomas, 2013; Ryan, 2012). This form of planning, however, often takes place at the neighborhood and community level, in which shrinking city research has paid little attention. An exception is Murgante and Rotondo (2013), who point out the likelihood of spatial patterns of growth and decline within shrinking cities. Particular neighborhoods may continue to support strong local economies and vibrant communities, while others suffer from problems of continual decline and abandonment. This study draws on the literature of agglomeration economies to determine how economies that specialize in particular industries compare to more diversified economies in terms of population loss within shrinking cities. Results suggest that neighborhoods with diverse economies experience less shrinkage on average than neighborhoods with more economic specialization; however, this difference is due to other neighborhood characteristics that tend to be present in these neighborhoods. Once these are controlled for, the analysis suggests neighborhoods in advanced economies that are more specialized experience slightly less population loss, although other factors in the neighborhood have more explanatory power. These factors include high levels of population shrinkage in surrounding neighborhoods, rental and seasonal
E-mail address:
[email protected].
housing, people that work remotely from home or travel to work in other cities, and large families, which are strongly associated with increased neighborhood shrinkage, as well as higher levels of income, education, single and nonfamily populations, immigrants, and migration, which are associated with less shrinkage. The remaining sections begin with a brief review of the literature that informs the specification of a neighborhood-level spatial lag regression model explaining the variation in population loss within shrinking cities. Next, the data and methods used to perform the analysis are detailed and the results of the analysis are presented and discussed. The paper concludes with a discussion of how the findings can help inform the planning and policy efforts advocated and pursed in shrinking cities. 2. Literature review At the macro-level, one of the major causes of urban shrinkage that relates to the experience of the United States is deindustrialization, or the loss of manufacturing jobs as firms outsource to locales with lower labor costs (Audirac, 2009; Großmann, Bontje, Haase, & Mykhnenko, 2013; Martinez-Fernandez et al., 2012). Central to this argument is the theory that cities in the global economy have undergone an economic restructuring as the logic of industrial location has changed. Advancements in transportation technologies removed the need for manufacturing firms to locate close to raw materials and/or consumer markets, while, at the same time, the high cost of inputs such as fiber optics and the need for an increasingly skilled labor force encouraged the co-location of financial, high-tech, and other firms that offer advanced services (Castells, 1989; Clark, 2011; Sassen, 2001; Stanback, 2002). The result of this process was that cities that were once prosperous manufacturing hubs such as Detroit and Flint, MI, Youngstown, OH, and Buffalo, NY faced significant decline, while cities such as San Francisco, CA and New York City able to support advanced services experienced prolonged growth. The implication for many urban economists is that, in order to grow, cities must attract talented individuals with high levels of human capital to provide the needed labor pool for the advanced services that drive the new economy (Clark, Lloyd, Wong, & Jain, 2002; Florida, 2002; Glaeser, Kolko, & Saiz, 2001; Markusen & Schrock, 2009). Although the focus on human capital and labor rather than physical capital such as natural resources or transportation arterials seems like a major shift, both are factors that have long been considered important in explaining the co-location, or agglomeration, of firms. In the 19th Century, Alfred Marshall famously discussed the importance of skilled
http://dx.doi.org/10.1016/j.cities.2016.12.006 0264-2751/© 2016 Elsevier Ltd. All rights reserved.
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
2
J. Murdoch, III / Cities xxx (2016) xxx–xxx
labor and the inputs of production in explaining where firms locate. In particular, Marshall suggested that similar firms co-locate so that they can increase productivity by sharing the costs of training the labor force as well as sharing the costs of expensive inputs necessary to the production process (Beaudry & Schiffauerova, 2009; Martin & Sunley, 2003). Marshallian agglomerations are often seen as drivers of innovation, employment growth, and prosperity (Martin & Sunley, 2003; Porter, 2003; Storper & Scott, 2009; van der Panne, 2004). The recent push for cities to support human capital to encourage economic success can be seen as a line of research in this vein. Arguably, a more fundamental shift in economic thinking occurred with the work of Jane Jacobs in the 1960s. She begins similar to Marshall by suggesting that firms co-locate to share access to markets and costly infrastructure; however, she departs from the theory saying that the agglomeration of different types of firms is what increases productivity, not the agglomeration of similar firms that can share costs (Jacobs, 1969). Central to the theory is the idea that the close proximity of diverse people and firms encourages a crosspollination of ideas that can lead to new products and more efficient and advanced methods of production that stimulate economic growth (Lucas, 1988). Moreover, Jacobs and more recent followers suggest that a diversified economy is much more likely to be able to weather severe economic shocks, as losses in employment are made up by gains in other industries less affected by the economic downturn (Dissart, 2003; Malizia & Ke, 1993; Wagner & Deller, 1998; among others). In the context of shrinking cities, both theories have explanatory power. The relatively recent success of Sunbelt cities that attract skilled labor forces and support agglomerations of high-tech, finance, and other advanced services demonstrate the power of Marshallian economies. On the other hand, the decline of cities in the Rustbelt that were predominantly focused in manufacturing and the resilience of diversified economies such as New York City and Boston demonstrate the importance of Jacobian diversity. Moreover, both theories are found to be valid in empirical research (for a review see Beaudry & Schiffauerova, 2009). What remains uninvestigated, however, is how these theories help to explain the variation in growth and decline within shrinking cities. 3. Data and methods The following sections describe the data and methods used to address two research questions: 1. In shrinking cities, do neighborhood-level Marshallian or Jacobian agglomerations help to explain variation in population decline? 2. How do these economic variables compare to other neighborhood variables including urban context, demographics, migration, housing characteristics, and labor force characteristics? 3.1. Census tracts in the US The analysis uses a sample of 5090 census tracts in shrinking cities in the 50 states and Washington D.C in the United States.1 The majority of studies examining the factors associated with economic agglomeration focus on the city, region, or state level of geography (Beaudry & Schiffauerova, 2009). There is, however, some evidence that the impacts of agglomeration are highly localized and fade quickly with distance (Glaeser, Kallal, Scheinkman, & Shleifer, 1992; van Soest, Gerking, & van Oort, 2002). Moreover, the purpose of the study is to determine factors that are associated with decline within Shrinking Cities. The neighborhood is thus a viable unit of analysis, as it facilitates an analysis that can uncover the potentially highly localized effects of economic agglomeration as well as examine variation in population loss within a city. 1 The sample does not include tracts with a total population less than 1000 or tracts with a total employment less than 500.
Although the neighborhood is the unit of analysis, the study does not assume that economic agglomeration will conform to neighborhood boundaries. Rather, as shown in Fig. 2, each neighborhood included in the analysis is classified as being within a specific economy type, which can have a boundary that spreads well beyond the neighborhood boundary. A shrinking city is defined as any city that has continually lost population in each decennial census since 1970, or since the census date closest to incorporation if after 1970. Each of the census tracts in the sample is in one of four categories that include (1) urban, (2) suburban, (3) small town, or (4) rural.2 Table 1 shows the number and percent of census tracts within each of the classifications. It also provides the definitions used for each classification, which are based on the categories of urbanicity defined by the National Center for Education Statistics (NCES). 3.2. Measures of shrinkage The degree of shrinkage, or population loss, from the year 2000 to 2010 in each census tract is calculated with a standard population growth formula: Shrinkage ¼ − ln
pop2010 pop2000
The formula adds a negative sign so that higher levels of population loss are shown with higher levels of the variable Shrinkage and lower levels of population loss are shown with lower levels of the variable. Without the negative sign, this would be reversed. In addition to this variable, the analysis uses a spatial lag term that represents the degree of shrinkage in surrounding neighborhoods.3 Murgante and Rotondo (2013) find that shrinkage and decline are often concentrated in specific areas and it is likely that decline in one locale may influence how another in close proximity develops. 3.3. Measures of economic agglomeration To capture economic agglomeration, 2-digit North American Industry Classification System (NAICS) codes are used to classify employment into 14 industry categories shown in Fig. 1.4 The figure also shows the results of a factor analysis of total employment in the categories that produces two constructs, economy size and economy type, that show how the variables group together.5 After the factor analysis, a score for each construct is produced using the regression scoring method (Thomson, 1951). Economy size has strong positive loadings for virtually all of the employment categories and reflects the overall level of employment within a census tract. High scores for this factor indicate a tract with a high level of total employment and thus a larger economy, while lower scores indicate a smaller economy. Economy type, on the other hand has some
2 The Missouri Census Data Center's correspondence tool matches census tracts to principal cities, urbanized areas, and urban clusters. The four designations are determined based on proximity to these geographies (see Table 1). 3 The open source software Geoda generated the spatial weights matrix used to calculate the spatial lag by defining neighbors of a given census tract as any tract that shares any part of the boundary of the given tract (queen contiguity). 4 Two digit codes are the most general (the most specific contain six digits) and capture a wide range of industry types within each category. Thus, the measures of specialization in this paper may carry the potential problem of obscuring some of the effect of diversified agglomerations (Beaudry & Schiffauerova, 2009). Two digit NAICS codes fit the purposes of this paper, however, because they allow easy identification of the broad sectors of interest. Future research may wish to examine the effects of more nuanced agglomeration clusters at the neighborhood level. 5 To determine the number of factors to retain in the analysis, I examined a scree plot and found that the first two factors explain virtually all of the variation in variables included in the analysis. Second, the first two factors meet Kaiser's (1960) selection criterion that states factors should have an eigenvalue above 1, while all other factors do not.
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
J. Murdoch, III / Cities xxx (2016) xxx–xxx Table 1 The classification of census tracts. Number (%) of tracts
Category
Definition
Urban
Census tracts within principal citiesa
Suburban
Census tracts outside principal cities, but within urbanized Areasb Census tracts within urban clustersc
Small town Rural Total
Census tracts outside urbanized areas and urban clusters
2600 (51%) 1650 (32%) 587 (12%) 253 (5%) 5090 (100%)
a A city that, according to the US Census Bureau, contains the primary population and economic center of a metropolitan area. A metropolitan area can have more than one principal city. b A cluster of census-defined blocks with a combined population of at least 50,000 people. c A cluster of census-defined blocks with a combined population of at least 2500 people, but less than 50,000 people.
positive and some negative loadings. High scores on the economy type factor reflect tracts with high levels of economic traditionalism, with employment in industries such as construction, manufacturing, agriculture, utilities, retail trade, transportation and warehousing, wholesale trade, accommodation and food services, mining, and administration and support and waste remediation services. Low scores, on the other hand, reflect the advanced industries that Florida (2002), Sassen (2001), and others suggest are the most important in the postindustrial new economy including professional, scientific, and technical services, information, finance and insurance, real estate and rental and leasing, arts, entertainment, and recreation, management of companies and enterprises, educational services, and health care and social assistance.
3
In addition to the measures of economy size and type, economic diversity for each census tract is calculated using an entropy index that takes the form Diversity ¼ ∑pi ln
1 ; pi
Where pi is the proportion of employment in NAICS industry i. After calculating the index, the median is subtracted from each census tract's value so that the index spans both negative and positive values. Negative values indicate economies that have lower than the median level of diversity, while positive values indicate economies that have higher than the median level of diversity. Fig. 2 shows a scatterplot of the economy type factor score and the measure of economic diversity. Since the measure of diversity ranges from perfectly specialized (Marshallian) to perfectly diverse (Jacobian) and the economy type measure ranges from perfectly advanced to perfectly traditional, census tracts that fall in the upper left portion of the graph are classified as “Advanced Jacobian” economies, tracts that fall in the upper right are “Traditional Jacobian” economies, tracts that fall in lower left are “Advanced Marshallian” economies, and tracts that fall in the lower right are “Traditional Marshallian” economies. These classifications are used as the independent variables of interest in the regression analysis described below. 3.4. Measures of neighborhood context Measures of neighborhood context control for other factors that may be associated with both the economic classifications and population decline. At the census tract level the 2000 Census of the Population contains measures of neighborhood context including, but not limited to, the population density per square mile; characteristics of households such as median income, household size, and family type; characteristics of population such as the age distribution, race/ethnicity, educational attainment, place of birth, and migration; characteristics of housing such as median housing age, housing value, and median gross rent,
Fig. 1. Factor loadings of industries for economy type and size.
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
4
J. Murdoch, III / Cities xxx (2016) xxx–xxx
to 2010. Table 3 shows the number (%) of shrinking vs. growing census tracts in small shrinking cities with populations under 100,000 as well as for each midsized and large shrinking city. Only Gary, Indiana experienced population loss in all of its census tracts. The other cities experienced shrinkage in some neighborhoods and growth in others. In Birmingham, Evansville, Kansas City, Portsmouth, Rochester, Syracuse, Baltimore, and Milwaukee, over a quarter of the neighborhoods in the city experienced population growth from 2000 to 2010. 4.2. Regression results
Fig. 2. Neighborhood economy classifications.
housing tenure, and housing occupancy; and characteristics of the labor force such as the unemployment rate, place of work, and form of transportation to work (Martinez-Fernandez et al., 2012; Großmann et al., 2013; Murgante & Rotondo, 2013). Variables for each of these categories (Table 2) are included as controls in the regression model described below. 3.5. Regression model The following spatial autoregressive model tests the association of Marshallian and Jacobian economies to the degree of population loss at the neighborhood-level: Shrinkage ¼ λWShrinkage þ βX þ ϵ where Shrinkage represents the values each tract has for the measure of shrinkage defined in Section 3.2, X represents the values each census tract has for the independent variables in the model (Table 2), β represents coefficients estimating the effects of marginal changes in the independent variables on the degree of population loss, W is a spatialweighting matrix which is multiplied by Shrinkage to create the spatial lag, λ is an estimate of the coefficient of the spatial lag for shrinkage, and ϵ is a random error term. To account for multicollinearity, all variables with a variance inflation factor (VIF) above 10 are removed from the analysis. The model is estimated using generalized spatial twostage least squares, which is robust to heteroscedasticity (Drukker, Prucha, & Raciborski, 2013). 4. Results 4.1. The geography of shrinkage Fig. 3 shows a map of shrinking cities that have continuously lost population since 1970. The large population category contains 10 cities that are primarily in the Rustbelt of the Midwest and Northeast, including Baltimore, Buffalo, Cincinnati, Cleveland, Detroit, Milwaukee, New Orleans, Pittsburgh, St. Louis, and Toledo. Similarly, the 14 midsize cities of Akron, Birmingham, Dayton, Erie, Evansville, Flint, Gary, Kansas City, Lansing, Livonia, Portsmouth, Rochester, Syracuse, and Warren are mostly in the same geographic area. The 1146 small shrinking cities, on the other hand, are much more numerous and dispersed throughout the country. This is an important finding, as much of the shrinking city literature focuses on the large and midsize cities, ignoring the prevalence of shrinkage in small cities throughout the country. Second, population shrinkage within shrinking cities is varied. Rather than places of comprehensive population loss, many cities contain several neighborhoods that experienced population growth from 2000
Table 4 shows the results of regression models that examine the relationship of the variables shown in Table 2 to neighborhood population shrinkage within shrinking cities. The table begins with the simplest model with only the economy types included and each column successively adds more variables to see how results change. The first model shows that, on average, Advanced Jacobian and Traditional Jacobian economies, in other words more diversified economies, experience less shrinkage than Marshallian economies with economic specialization. Specifically, population shrinkage is about 3.0% and 1.8% less in
Table 2 Independent Variables in Regression Model. Economy types
Migration
Traditional Marshallian (reference category) Advanced Marshallian Traditional Jacobian
% Moved from a foreign country in last 5 years
Advanced Jacobian
% Moved from another state in last 5 years % Moved from another county in the same Sate in the last 5 years % Moved from another house in the same county in the last 5 years
Urban context
Housing characteristics
Urban (reference category) Suburban Small town Rural Economy size factor Spatial lag of shrinkage City population size Census region
Median VALUE Median Rent Median age % Renter occupied % Vacant, for rent % Vacant, for sale % Vacant, seasonal or vacation home % Other vacant
Demographics
Labor force characteristics
Median Income % Families with 5 or less people % Families with 6 or more % Single households % Non-family households with 2 to 3 people % Non-family households with 4 or more % Married with kids % Single female with kids % Under 18 % 18 to 24 % 25 to 34 % 35 to 49 % 50 to 64 % 65 and up % Female % Black % Hispanic % Asian % Two or more % Other Race Racial diversity (entropy of above categories) % No high school % High school % Some college % Foreign born % in poverty
% Work outside State % Work outside county within state % Work outside city within county % Work at home % Take public transit % Bike % Walk % Commute 15 to 30 min % Commute 30 to 60 min % Commute over 60 min % Unemployed
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
J. Murdoch, III / Cities xxx (2016) xxx–xxx
5
Fig. 3. The geography of shrinking cities.
Advanced Jacobian and Traditional Jacobian economies, respectively, than Traditional Marshallian economies in the average neighborhood. Population shrinkage in Advanced Marshallian economies is not statistically different than the level in Traditional Marshallian economies in shrinking cities. That being said, once urban context and population demographics are accounted for these differences are no longer significant
Table 3 Variation in Neighborhood Population Loss in Selected Shrinking Cities. Shrinking cities
Shrinking tracts
Growing tracts
Total
Small (Under 100,000)
2626 (84.74%)
473 (15.26%)
3099 (100%)
Midsize (100,000 to 250,000) Akron, OH 57 (89.06%) Birmingham, AL 64 (70.33%) Dayton, OH 55 (93.22%) Erie, PA 24 (80%) Evansville, IN 24 (63.16%) Flint, MI 33 (89.19%) Gary, IN 23 (100%) Kansas City, MO 34 (66.67%) Lansing, MI 35 (81.4%) Livonia, MI 24 (77.42%) Portsmouth, NH 17 (65.38%) Rochester, NY 50 (71.43%) Syracuse, NY 28 (57.14%) Warren, MI 33 (80.49%)
7 (10.94%) 27 (29.67%) 4 (6.78%) 6 (20%) 14 (36.84%) 4 (10.81%) 0 (0%) 17 (33.33%) 8 (18.6%) 7 (22.58%) 9 (34.62%) 20 (28.57%) 21 (42.86%) 8 (19.51%)
64 (100%) 91 (100%) 59 (100%) 30 (100%) 38 (100%) 37 (100%) 23 (100%) 51 (100%) 43 (100%) 31 (100%) 26 (100%) 70 (100%) 49 (100%) 41 (100%)
Large (Over 250,000) Baltimore, MD Buffalo, NY Cincinnati, OH Cleveland, OH Detroit, MI Milwaukee, WI New Orleans, LA Pittsburgh, PA St. Louis, MO Toledo, OH Total
56 (32.56%) 9 (13.04%) 13 (12.87%) 6 (4.41%) 9 (3.6%) 86 (46.49%) 13 (9.29%) 17 (16.5%) 14 (15.38%) 11 (12.09%) 859 (16.88%)
172 (100%) 69 (100%) 101 (100%) 136 (100%) 250 (100%) 185 (100%) 140 (100%) 103 (100%) 91 (100%) 91 (100%) 5090 (100%)
116 (67.44%) 60 (86.96%) 88 (87.13%) 130 (95.59%) 241 (96.4%) 99 (53.51%) 127 (90.71%) 86 (83.5%) 77 (84.62%) 80 (87.91%) 4231 (83.12%)
or meaningful. On the other hand, Advanced Marshallian economies show slightly less shrinkage (1.4% in model 6) than Traditional Marshallian economies after controls are included in the model. Thus, although the first model lends support to the Jacobian theory supporting economic diversity, the analysis shows that this difference is due to other factors associated with economy type. Once these are accounted for, it appears that neighborhoods in advanced economies are slightly better off with some degree of Marshallian specialization. Table 4 shows that the economy types, while statistically significant in some models, do not explain much of the variance in neighborhood population shrinkage within shrinking cities. Urban context, and especially the degree of population shrinkage in surrounding neighborhoods, is far more important. The spatial lag term, which can range from −1 to 1, is consistently strong and significant, indicating that surrounding shrinkage has a strong association with increased shrinkage within a neighborhood. Urbanicity and economy size are also important. Suburban, small town, and rural area neighborhoods all experience higher levels of shrinkage compared to urban neighborhoods in principal cities, while larger economies experience less shrinkage than economies of smaller sizes. Fig. 4 shows the standardized beta coefficients of the significant variables in model 6 above, which allow each of the variables to be compared in terms of the strength of association with neighborhood shrinkage. The bars in the figure represent the degree of association with population shrinkage. Bars that stretch toward the left represent negative beta coefficients and thus indicate variables that are associated with decreased levels of population shrinkage. Bars that stretch toward to right represent positive coefficients and thus indicate variables that are associated with increased levels of population shrinkage. Neighborhoods with older homes, more rental properties (both vacant and occupied), larger families, high percentages of people working outside of the city, high percentages of people with a 30 to 60 min commute, high percentages of people working remotely from home, and housing that is seasonally vacant are likely to experience higher levels of shrinkage. Within shrinking cities, population loss appears to impact specific neighborhoods, especially those which are close to places that are
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
6
J. Murdoch, III / Cities xxx (2016) xxx–xxx
Table 4 Associations with Population Shrinkage (Regression Results).
λ (Spatial lag) Trad. Marshallian (Reference) Advanced Jacobian Traditional Jacobian Advanced Marshallian Urban (Reference) Suburban Small town Rural area Small, under 100,000 (Reference) Midsize (100,000 to 250,000) Large (Over 250,000) Northeast (Reference) Midwest South West Economy size Med. household income, logged % Families with 6 or more % Single households % 2 to 3 Person non-family % 4 or more person non-family % Single mother families % 25 to 34 % 35 to 49 % 50 to 64 % 65 and older % female % Hispanic % Two or more races % Black % Asian % Other race Racial diversity % No high school % High school % Some college % Foreign % Moved from other country % Moved from other state % Moved from other county % Moved from other house within county Median housing value, logged Median rent, logged Median housing age % Renter-occupied % Vacant rental homes % Vacant for sale % vacant off market % Vacant seasonal % Vacant migrant housing % Other vacant % Work outside of state % Work outside of county % Work outside of city/place % Work at home % Take public Transport to work % Bicycle to work % Walk to work % Commute 15 to 30 min % Commute 30 to 60 min % Unemployed Intercept N
(1)
(2)
(3)
(4)
(5)
(6)
Economy types 0.629⁎⁎⁎
Urban context 0.439⁎⁎⁎
Demographics 0.374⁎⁎⁎
Migration 0.423⁎⁎⁎
Housing characteristics 0.421⁎⁎⁎
Labor force characteristics 0.429⁎⁎⁎
−0.030⁎⁎⁎ −0.018⁎⁎⁎ −0.00255
−0.032⁎⁎⁎ −0.00540 −0.020⁎⁎⁎
−0.00137 0.00354 −0.0138⁎⁎
0.000519 0.00319 −0.0131⁎⁎
−0.000432 0.00239 −0.0126⁎⁎
−0.00123 0.00422 −0.0136⁎⁎
0.0168⁎⁎⁎ 0.00171 0.00873
0.0269⁎⁎⁎ 0.0138⁎⁎ 0.0309⁎⁎⁎
0.0274⁎⁎⁎ 0.0179⁎⁎ 0.0379⁎⁎⁎
0.0267⁎⁎⁎ 0.0158⁎⁎ 0.0296⁎⁎⁎
0.0310⁎⁎⁎ 0.0111⁎ 0.0213⁎
0.00466 0.0513⁎⁎⁎
−0.00985 0.0230⁎⁎
−0.00774 0.0190⁎⁎
−0.00152 0.0240⁎⁎⁎
−0.00399 0.0110
0.0112⁎⁎ 0.0181⁎⁎ 0.00737 −0.028⁎⁎⁎
−0.00161 0.000118 0.0316⁎ −0.00829⁎⁎⁎ −0.0699 0.0173⁎⁎ −0.00216 −0.00838⁎⁎⁎
0.000761 0.00528 0.0362⁎⁎ −0.00657⁎⁎⁎ −0.0623 0.0172⁎⁎ −0.00182 −0.00590⁎
0.00700 0.000649 0.0226 −0.00730⁎⁎⁎ −0.114⁎⁎⁎ 0.0177⁎⁎ −0.00387⁎⁎⁎ −0.00855⁎⁎
0.0106⁎ 0.00222 0.0142 −0.00666⁎⁎⁎ −0.115⁎⁎⁎ 0.0156⁎⁎ −0.00424⁎⁎⁎ −0.00928⁎⁎⁎
−0.0224 0.00117 0.000961 −0.00392 −0.00523⁎⁎⁎ 0.000448 −0.00586⁎⁎ −0.00125⁎⁎⁎
−0.0195 0.00185 0.000294 −0.00389 −0.00555⁎⁎⁎ 0.000329 −0.00575⁎⁎ −0.00128⁎⁎⁎
−0.0245 −0.00158 0.000441 −0.00347 −0.00661⁎⁎⁎ 0.000313 −0.00569⁎⁎ −0.00161⁎⁎⁎
−0.0240 −0.00228 0.00179 −0.00340 −0.00607⁎⁎⁎ 0.000370 −0.00509⁎⁎ −0.00160⁎⁎⁎
0.00364 0.000679⁎⁎⁎ −0.000827 −0.00244⁎⁎⁎ −0.0569⁎⁎⁎
0.00309 0.000499⁎⁎⁎ −0.000859 −0.00253⁎⁎⁎ −0.0513⁎⁎⁎
−0.000580 −0.00192⁎⁎⁎
−0.000419 −0.00182⁎⁎⁎
0.00263 0.000391⁎⁎ −0.00108 −0.00325⁎⁎⁎ −0.0336⁎⁎⁎ −0.00133⁎⁎⁎ −0.00230⁎⁎⁎
0.00253 0.000296 −0.00104 −0.00324⁎⁎⁎ −0.0275⁎⁎ −0.00113⁎⁎ −0.00202⁎⁎⁎
−0.000617 −0.00172⁎⁎⁎
−0.000415 −0.00278⁎⁎ 0.00368 −0.000200 −0.00182⁎⁎
−0.000648 −0.00270⁎⁎ 0.00126 −0.00164 −0.00243⁎⁎⁎ −0.00150⁎⁎
−0.000388 −0.00295⁎⁎⁎ 0.00124 −0.00170⁎ −0.00231⁎⁎⁎ −0.00162⁎⁎
0.00795 0.00867 0.000509⁎⁎⁎ 0.00101⁎ 0.00433⁎⁎⁎ −0.00998⁎⁎⁎
0.00736 0.00773 0.000555⁎⁎⁎ 0.000935⁎⁎ 0.00424⁎⁎⁎ −0.0102⁎⁎⁎
0.00534 0.00230⁎⁎⁎ −0.00361 0.00178
0.00453 0.00216⁎⁎ −0.00371 0.00110 −0.000517⁎⁎ −0.000129 0.000445⁎⁎ 0.00349⁎⁎
−0.000701
0.0515⁎⁎⁎ 5090
0.0386⁎⁎⁎ 5090
0.238⁎⁎⁎ 5090
0.230⁎⁎⁎ 5090
0.261⁎⁎⁎ 5090
0.000653 0.00163 0.0000146 −0.000138 0.00113⁎⁎ 0.00128 0.258⁎⁎⁎ 5090
⁎ p b 0.1. ⁎⁎ p b 0.05. ⁎⁎⁎ p b 0.01.
already shrinking and those with populations that may have less longterm attachment to the community as they are renters, seasonal residents, or workers with jobs that are far away. On the other hand, neighborhoods with higher median household incomes and larger percentages of single (non-married) households,
people aged 50 to 64, people graduating from high school, foreignborn or Hispanic populations, 2- to 3-person nonfamily households, females, higher levels of migration, racial diversity, houses for sale, and large economy sizes tend to experience the least amount of shrinkage. These neighborhoods, with economic strength, in terms of income and
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
J. Murdoch, III / Cities xxx (2016) xxx–xxx
7
Fig. 4. The Relative Importance of the Independent Variables (Beta Coefficients from Model 6).
economy size, a diversity of people, in-migration, and various types of non-married households, may be able to whether resident turnover as residents that move out are quickly replaced with residents that move in. Moreover, these places that house diverse populations may support economic and cultural scenes that keep the neighborhood desirable, despite an overall trend of population decline. In sum, Table 4 and Fig. 4 together show that while economy type may explain some of the variation of neighborhood population shrinkage within shrinking cities, other factors are more important. Surrounding shrinkage, as well as demographic characteristics, housing characteristics, and in-migration have strong and significant associations with population change.
a result, economic development policy should support economic agglomeration as it exists and, rather than attempting to brand a city with a simple narrative of industry specialization or diversification, adopt a more localized and nuanced approach. Finally, in the most severely declining neighborhoods with little remaining economic activity, non-economic approaches may be better served. Recent work that identifies recommendations for planning and policy action in shrinking cities (e.g. Dewar & Thomas, 2013), advocates neighborhood grassroots and locally-informed responses to shrinkage, which the analysis in this paper supports. References
5. Conclusion To promote success and competitiveness in the global economy, shrinking city governments search for solutions to continual decline and abandonment. This paper examines two potential solutions, a strategy of economic diversification promoted by Jacobs (1969) and followers compared to a strategy of economic specialization in specific industries and clusters promoted by urban economists such as Porter (2003) and Florida (2002). Results in this paper suggest that within shrinking cities, regardless of economy type, these types of agglomeration have little association with reduced shrinkage. Other factors, such as surrounding shrinkage, neighborhood demographics, housing characteristics, migration, and labor force characteristics have greater explanatory power. There is some evidence, however, that once these neighborhood factors are accounted for, economic specialization in advanced services has a small, but significant, association with reduced shrinkage. These results have several implications. First, economic development policy in shrinking cities may be most effective with a strategy of targeting specific neighborhoods with preexisting economic agglomeration. It is an efficient allocation of funds and avoids expensive strategies to create new activity in severely declining neighborhoods that, as the analysis shows, are not likely have much positive impact. Importantly, these targeted strategies should work to include local residents in the planning and implementation process so that the economic assets that already exist can be identified and most effectively supported. Second, the analysis shows that multiple types of industry, both production-oriented and advanced services, can reduce shrinkage. As
Audirac, I. (2009). Urban shrinkage amid fast metropolitan growth (two faces of contemporary urbanism). In K. Pallagst (Eds.), The future of shrinking cities: Problems, patterns and strategies of urban transformation in a global context (pp. 69–80). Berkeley: Center for Global Metropolitan Studies. Beaudry, C., & Schiffauerova, A. (2009). Who's right, Marshall or Jacobs? The localization versus urbanization debate. Research Policy, 38(2), 318–337. Castells, M. (1989). The informational city: Economic restructuring and urban development. Cambridge, MA: Blackwell Publishers. Clark, T. N. (Ed.). (2011). The city as an entertainment machine. (Rev. Ed.). Lanham, MD: Lexington Books. Clark, T. N., Lloyd, R., Wong, K. K., & Jain, P. (2002). Amenities drive urban growth. Journal of Urban Affairs, 24(5), 493–515. Dewar, M., & Thomas, J. M. (2013). The city after abandonment. Philadelphia: University of Pennsylvania Press. Dissart, J. C. (2003). Regional economic diversity and regional economic stability: research results and agenda. International Regional Science Review, 26(4), 423–446. Drukker, D. M., Prucha, I. R., & Raciborski, R. (2013). Maximum likelihood and generalized spatial two-stage least-squares estimators for a spatial-autoregressive model with spatial-autoregressive disturbances. The Stata Journal, 13(2), 221–241. Florida, R. (2002). The rise of the creative class. New York: Basic Books. Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100(6), 1126–1152. Glaeser, E. L., Kolko, J., & Saiz, A. (2001). Consumer city. Journal of Economic Geography, 1(1), 27–50. Großmann, K., Bontje, M., Haase, A., & Mykhnenko, V. (2013). Shrinking cities: Notes for the further research agenda. Cities, 35, 221–225. Jacobs, J. (1969). The economy of cities. New York: Random House. Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement, 20(1), 141–151. Lucas, R. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3–42. Malizia, E. E., & Ke, S. (1993). The influence of economic diversity on unemployment and stability. Journal of Regional Science, 33(2), 221–235. Markusen, A., & Schrock, S. (2009). Consumption-driven urban development. Urban Geography, 30(4), 344–367. Martin, R., & Sunley, P. (2003). Deconstructing clusters: chaotic concept or policy panacea? Journal of Economic Geography, 3(1), 5–35.
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006
8
J. Murdoch, III / Cities xxx (2016) xxx–xxx
Martinez-Fernandez, C., Audirac, I., Fol, S., & Cunningham-Sabot, E. (2012). Shrinking cities: Urban challenges of globalization. International Journal of Urban and Regional Research, 36(2), 213–225. Murgante, B., & Rotondo, F. (2013). A geostatistical approach to measure shrinking cities: The case of Taranto. In S. Montrone, & P. Perchinunno (Eds.), Statistical methods for spatial planning and monitoring (pp. 119–142). Milan: Springer. Shrinking cities, volume 1. Oswalt, P. (Ed.). (2006). International research. Ostfildern-Ruit, Germany: Hatje Cantz Verlag. Porter, M. (2003). The economic performance of regions. Regional Studies, 37(6–7), 549–578. Rieniets, T. (2009). Shrinking cities: causes and effects of urban population losses in the twentieth century. Nature and Culture, 4(3), 231–254. Ryan, B. D. (2012). Design after decline: How America rebuilds shrinking cities. Philadelphia: University of Pennsylvania Press. Sassen, S. (2001). The global city: New York, London, Tokyo. Princeton, NJ: Princeton University Press.
Stanback, T. M., Jr. (2002). The transforming metropolitan economy. New Brunswick, NJ: Center for Urban Policy Research, Rutgers University. Storper, M., & Scott, A. J. (2009). Rethinking human capital, creativity, and urban growth. Journal of Economic Geography, 9(2), 147–167. Thomson, G. H. (1951). The factorial analysis of human ability. London: University of London Press. Van der Panne, G. (2004). Agglomeration externalities: Marshall versus Jacobs. Journal of Evolutionary Economics, 14(5), 593–604. van Soest, D. P., Gerking, S. D., & van Oort, F. G. (2002). Knowledge externalities, agglomeration economies, and employment growth in Dutch cities. Orlando, FL: University of Central Florida. Wagner, J. E., & Deller, S. C. (1998). Measuring the effects of economic diversity on growth and stability. Land Economics, 541–556.
Please cite this article as: Murdoch, III, J., Specialized vs. diversified: The role of neighborhood economies in shrinking cities, Cities (2016), http:// dx.doi.org/10.1016/j.cities.2016.12.006