The Role of Municipal-Level Factors in Neighborhood Economic Change

ABSTRACT:

Despite extensive studies on neighborhood change, the role of municipal-level factors in neighborhood economic change has been underexplored. This article reviews diverse social science literature and makes theoretical connections between city size and homogeneity of city population and municipal performance, which is accordingly associated with neighborhood economic health. Building on the insights from the literature, this study hypothesizes that neighborhoods stay economically healthier in smaller cities and more homogeneous cities. This study presents a multilevel analysis of neighborhood economic change in 35 U.S. metropolitan areas between 1990 and 2000 and finds empirical evidence to support the proposed hypothesis.

Notes

I use the terms of municipality and city interchangeably in this study.

One might say that there is not much variation in size among suburban municipalities. However, suburban municipalities are not necessarily small. Westminster in Colorado and Naperville in Illinois include over 100,000 people each while some suburban municipalities hold only a few hundred people.

One might argue that households in suburban municipalities have similar interests because they can “vote with their feet” in balancing between taxes and public services offered as argued by Tiebout (). However, residential location decisions are constrained by many other factors, such as proximity to employment or family, limitations to mobility, and the multifaceted nature of public goods (Oliver, ; Rubinfeld, Shapiro, & Roberts, ). Additionally, citizens are not fully informed about public services offered by each municipality (Schneider, ).

Alesina et al. () also describe a nice example of the negative relationship between diversity and public spending. The Oakland School Board in Oakland, California, proposed black English (called Ebonics) to be recognized as a separate official language. Black parents believed that this program was good for their children’s needs. However, Hispanic parents were not happy about public spending for Ebonics due to the lack of public resources for their children to have English as a Second Language (ESL) classes or bilingual education. In the meantime, Asian parents complained about the fact that more bilingual resources were used for Hispanic children. Finally, white parents objected to using public resources for any nonstandard English instruction.

Even if a municipality elects some or all council members from districts not at large, neighborhood interests are still better represented in homogeneous cities, because a district often includes several neighborhoods.

A census tract includes 4,000 persons on average (Geolytics, ).

Place in the census includes city, town, borough, and village, which are incorporated, and census designated place, which is unincorporated but delineated for statistical purposes.

My methodology differs from Zielenbach’s () in that I did not include the number of residential loans per housing unit in creating the index. Zielenbach () views a larger number of residential loans per housing unit as an indicator of an economically healthy neighborhood. However, a lower number of residential loans per housing unit in a neighborhood does not necessarily imply that a neighborhood is economically unhealthy when there is a large share of elderly people that have paid off mortgage loans.

Because using an index consisting of more than one economic indicator has not been common in neighborhood change studies, equally weighting the two indicators may be the best place to start. I also ran two additional models using each of the indicators, but did not find practical differences between the two models and the original model.

Average housing value consists of both owner-occupied housing value and capitalized rent. Capitalized rent was added to take into account those neighborhoods with a large share of rental units. It was computed by dividing yearly rent by an interest rate of 0.1, which is commonly used to calculate capitalized rent in real estate (Brueggeman & Fisher, ; Milies, Berens, & Weiss, ). The fact that there is a positive relationship (0.13) between change of the relative ratios of owner-occupied housing value and capitalized rent between 1990 and 2000 indicates that the two values moved in the same direction.

Compared to using median household or family income, using per capita income promotes a more comprehensive analysis because per capita income considers every individual in a neighborhood. To be consistent with per capita income, which is an average individual income, this study used average housing value, although using average housing value instead of using median housing value may shift the distribution upward.

Theories suggest that neighborhood change results from interactions of factors at the neighborhood, municipality, and metropolitan levels. Thus, I could have run a level-3 multilevel model where the level-3 (metropolitan) equation is nested in the intercept of the level-2 (municipality) equation and the level-2 equation is nested in the level-1 (neighborhood) intercept. However, in this study, I ran a level-2 multilevel model by using the change of neighborhood index scores between 1990 and 2000, which combine the relative ratios of per capita income and average housing value to each neighborhood’s respective metropolitan average values for the dependent variable. Using such measures can understate neighborhood economic gain or decline in economically growing or declining metropolitan areas, respectively. However, in this way, this study controls for regional and metropolitan variations and can focus on the link between municipal-level factors and neighborhood economic change.

I computed unique GINI coefficients for each municipality, using the prln04.exe program provided by Francois Nielsen at the following web-site: http://www.unc.edu/˜nielsen/data/data.htm

As suggested by Snijders and Bosker (, ), I calculated proportional reductions of prediction error at each level, which are analogous to R², rather than calculating commonly used percentage of variance explained.

Because the dependent variable is the log ratio of the neighborhood index score in 2000 to the neighborhood index score in 1990, I translated the log values to the percentage terms.

Additional information

Notes on contributors

Hee-Jung Jun

Hee-Jung Jun is an assistant professor in the Department of Political Science and Planning and the Director of the Master of Urban and Regional Planning Program at University of West Georgia. Her research interests include sustainable community development, neighborhood dynamics, residential mobility and social capital.