Municipalities, especially those that are mid-sized or smaller, often face significant challenges in providing services and amenities to meet the needs of their diverse and changing populations. Solutions are usually context-specific and must factor in larger demographic and economic trends, in order to be effective. And, yet, in spite of contextual differences, cities frequently have meaningful similarities. However, identifying peer cities is often informed more by conversation than by data or evidence.
The Peer Cities Identification Tool (PCIT) developed by the Community Development and Policy Studies division (CDPS) of the Federal Reserve Bank of Chicago is a data comparison and visualization instrument that can help policymakers and practitioners understand a municipality in the context of peer cities. The tool stems from the Industrial Cities Initiative (ICI), a study that originally profiled ten midwestern cities with manufacturing legacies, at least 50,000 population and at least 25 percent employed in manufacturing in 1960, and how they have fared in socioeconomic terms over time.1
The original study generated a great deal of attention among leaders of cities with comparable histories. The PCIT is in part a response to inquiries from these leaders as to how they "compare" to similar cities both within the region and in other regions of the country, as well as in response to a stated need/desire to share and learn from best practices to address entrenched municipal challenges.
The PCIT is different from other "city-data" tools in that it is not a ranking, but a comparison tool that provides the user with a baseline of data from which to ask questions and interpret and apply the answers. This approach is based on a fundamental belief that every city is different, possessing its own assets and liabilities. Usually no one is more aware of the "municipal balance sheet" than the people who live in and lead a city.
The PCIT allows city leaders concerned with community and economic development issues to identify groups of cities experiencing similar trends, challenges, and opportunities along economic, demographic, social, and housing dimensions. Using data on 300 cities from the 2010-2014 American Community Survey, as well as longitudinal historical census data, the PCIT performs a cluster analysis to identify similar cities. The 300 cities located nationwide have a common baseline: a population of at least 50,000 in 1960. Today, the 300 cities have a median population of just over 100,000.
How it works
From the PCIT website, users enter a city and select one of four themes off which to base their analysis: ‘Equity, Affordability, Resilience and Outlook.’ Users first see a map of the United States highlighting the identified peer cities – usually between five and 15 cities. While often peers are geographically proximate (i.e., within the same general region of the country), sometimes a peer search can yield surprising results. The PCIT will also present the user with data from the peer cities and a table of key variables that were used to identify the group. In addition, the tool generates peer median, minimum and maximum for each variable, as well as the ICI 300 median for the selected variable enabling comparison across and within the cities, as well as the (full) dataset. This perspective can provide further context, especially in identifying areas in which the subject city might deviate from its peers, which can serve to highlight particular challenges or opportunities. Users can also select variables to graph or chart, providing a useful visual. All data and images can be exported.
Understanding the themes
Peer cities are grouped along four key themes (others may be added at a later time), which are essentially ‘portals’ to the data. These themes are designed in response to key areas of concern voiced by city leaders following more than 200 interviews across almost a dozen cities as part of the Industrial Cities Initiative and other place-based research.
•Equity addresses questions regarding inclusion, access, and diversity using wage-based Gini coefficient, race and ethnicity-based dissimilarity indices, changes in poverty levels, and educational attainment. City leaders cited challenges of creating and implementing inclusive growth strategies that attract new businesses and jobs to their cities, while creating policies that allow marginalized populations to benefit from these new opportunities. The PCIT uses the wage-based GINI coefficient (as opposed to the income-based coefficient more frequently used) to focus in on wage-earning workers who have been employed for the full year.
•Housing speaks to issues of affordability by incorporating data relating to home ownership (income-to-home value ratio) and renting (rent burden), the quality and competitiveness of housing stock by using the age of housing as a proxy, and monthly living costs. Providing competitive housing affordable and attractive to both renters and buyers was a primary discussion point among leaders.
•Resilience speaks to issues related to economic diversification in terms of changes in manufacturing employment, existing levels of manufacturing employment, labor force participation, and unemployment. Many cities experienced economic shocks during the Great Recession, but had experienced decline along these measures during the preceding decades. Economic diversification and labor force conditions provide broad insights into areas of vulnerability and strength.
•The Outlook theme explores signs of a city's demographic and economic future by incorporating changes in the working age population, family composition, and mobility (over time). Changes in the age distribution of a population, net migration, and household size and composition, can all provide clues about a city's future. Cities experiencing unusual demographic shifts may look to peers undergoing similar shifts, and to (non-demographic) factors such as employment and educational opportunities, that may be drivers.
Methodology
The tool works by performing a hierarchical cluster analysis on all 300 cities, using the variables included in the selected theme. A cluster analysis is a way of grouping data based on the similarity of responses to several variables. A cluster analysis treats the subject city data as a “case” and will find "similar" or "peer" cases based on several variables (such as our themes of socioeconomic status). The clustering method used is Ward's method, which minimizes the variance across all variables in a given group.2If a cluster produces only a small number of results, the program has the option of using the ranked values instead of the normalized values, which tends to produce more evenly distributed groups, but does not allow for easy distinction between extreme outliers and more typical cities. The cluster containing the focus city is squared off for ease of explanation and verification, by looking at the maximum and minimum values for each variable within the cluster and including all cities within the given range for each variable as peer cities for the focus city. Finally, the program produces a table of all the included variables for all peer cities in the cluster.
As mentioned above, the PCIT has several potential uses. For many municipal planners, comparison cities are often, for practical reasons, limited to those that are geographically proximate, subject to similar regional trends, and to the planner’s personal knowledge and familiarity. Sometimes, this is satisfactory, for example when planners may want to understand cities subject to similar statewide policies or conditions. However, at other times this purview is limiting and frustrating to planners and other practitioners who wish to go outside of their ‘familiarity zone’ to interact with other places that may be experiencing similar challenges or changes. In particular, cities that have experienced changes in their economy, with respect to manufacturing employment, for example, may find it useful to learn about cities outside of their specific region.
To this end, the PCIT will return cities that may not initially appear to be peers – the most evident difference is often that the peers are in very different regions of the country – but that upon closer look are experiencing similar conditions, at least along one of the variable clusters. Different variable clusters will return different sets of peers – occasionally there will be common cities across the theme-based peer groups – and additional data exploration can often shed light on similarities and differences. Usually, however, the PCIT peer group will include regionally proximate cities, as the methodology used specifically seeks to minimize variance across clusters.
While the PCIT can be a useful comparison tool from which to initiate planning discussions, it is not a planning tool per se. Users are cautioned against taking high level, longitudinal data as a directive or prescription in any way: each of these cities is unique, with its own distinct characteristics. However, as the case study illustrates, it can be helpful in answering a specific question (about housing, for example). It can be especially useful in informing, without judgment or qualification, broader discussions.
Note: We expect to bring the PCIT online in the first quarter of 2017, and will provide an update in the next edition of ProfitWise News and Views.
1For more information on the Industrial Cities Initiative, please visit https://www.chicagofed.org/region/community-development/community-economicdevelopment/ici/index
2For more information regarding Ward's Method, the original article detailing the method is publicly available at: http://homes.mpimf-heidelberg.mpg.de/~mhelmsta/pdf/1963%20Ward%20JASA.pdf.
