16
The Economics and Politics of City Size
By Shirley Svorny
Professor of Economics
California State University, Northridge
Northridge, CA 91330-8374
(818) 677-4504
January 2003
The Economics and Politics of City Size
I. Introduction
The recent, failed attempt to break the San Fernando Valley from Los Angeles, raises the question as to what size is optimal with respect to local governance. Is Los Angeles too large, at nearly 4 million residents? This paper discusses issues related to municipal boundary decisions as they apply to a city as large as Los Angeles. Specifically, it reviews the literature on municipal economies of scale, and raises other related issues, such as the influence of special interests, equity effects, and the implications for regional planning.
II. Economics of City Size
The theory behind identifying the optimal size for the production of city services parallels that for the firm. As with a firm, for each function, it is relevant to consider whether it should be produced within the city or outside, through private means, or for large projects, through regional consortiums. It is thought that per capita costs of production initially fall as the number of residents served grows. Producing services for thousands of households allows a sharing of capital equipment (fire truck, police car) and labor (fire and police personnel) that may reduce the per capita or per household cost of production.
At some point, as the number of residents served increases, it is thought that per capita costs of production start to rise. The economics literature on the firm emphasizes the communication, incentive, and coordination problems inherent in a large service provider. Being too big leads to inefficiencies in service provision and management. In addition, agency costs are thought to increase as entities grow in size. These are the costs associated with getting employees (agents) to act in the interest of a firm’s owners or, for a public entity, voters. The further removed employees are from the oversight of those whose interests are being served, the more it is thought they will be able to shirk, putting their own interests before those who hire them.
The question is, how large must a public provider be before it exhausts economies of scale and runs into diseconomies in the production of services? Initially, researchers tried to determine the efficient size of a city using engineering data or by comparing city per-capita expenditures. Engineering studies estimate costs based on knowledge of the technical aspects of producing a service. They focus on how inputs are combined to produce final products and examine whether, as a city grows, there are some advantages to size. For example, as a city grows, it can buy specialized equipment, like a snowplow. This lowers the cost of clearing the roads. Engineering studies suffer from the fact that they ignore coordination and incentives problems that cause costs to rise as city size increases (Hirsch 1968). Another weakness is that they ignore external conditions that might affect costs (such as population transiency, age, or density) (Duncombe and Yinger 1993).
Fox (1980) pointed out the weaknesses in studies of per-capita expenditures as a means of assessing the optimal city size. Cities that provide a greater array of services will have higher per-capita costs, yet nothing can be concluded about their relative efficiency. If the demand for municipal services is price elastic, total spending may be highest where cities are the most efficient, not the reverse.
Bish and Ostrom (1973) noted that central cities may experience higher per-capita (per resident) spending because many people come into the city to work, but few live there. Also, higher costs of labor in large cities could make large cities appear less efficient in the production of city services, although they are not.
Finally, if cities shift costs to residents, these costs cannot be captured in per-capita spending measures. For example, if residents must drive long distances to obtain construction permits, there is a cost, but it will not show up in estimates of the cost of city permit activity. Local government mandates (for example, requiring sidewalk repairs upon the sale of property) shift public costs to private individuals, again, the cost is hidden.
Concern with these issues led many researchers to focus on the cost of providing specific services – like fire or police, that are more directly comparable than other types of spending, and for which measures of population density and other characteristics can be used to control for variations across cities in the costs of providing services. To assess the impact of population size on city costs, researchers generally estimate equations where average costs are regressed on measures such as inputs (labor, capital), input prices, environmental characteristics (population density, age), and population.
When Hirsch (1968) examined spending on police, fire, library, public housing, welfare, parks and recreation, refuse collection and street maintenance, he found no economies of size beyond 100,000 residents. However, he did find economies in the more capital-intensive services -- water, sewage disposal, transportation, electric power production, hospitals, and some health services -- for cities over 100,000. Fearful that any advantages of size would be outweighed by inefficiencies resulting from “political patronage and administrative top-heaviness” in large cities, he concluded that, “in terms of economies of scale, governments serving from 50,000 to 100,000 urbanites might be most efficient.” (p. 509)
Fox (1980) examined more than 80 empirical studies of the production of police protection, education, fire protection, refuse collection, roads and highways, and water and sewer utilities. Fox outlined problems measuring costs, inputs and input prices, and problems measuring conditions under which the services are produced (demographic, such as population density, or weather) and, most important, measuring output (for example, trying to measure police output using crime and apprehension rates). Reviewing the existing empirical studies, Fox was comfortable drawing conclusions with respect to only two categories of municipal services -- fire protection (economies are exhausted at 10,000 residents) and refuse collection (economies to 20,000 residents).
Fire and Police
Much of the recent cross-sectional research has focused on fire and police protection services, which make up the lion’s share of most city budgets (excluding education). William Duncombe and John Yinger (1990, 1993), like Fox before them, critiqued the empirical literature on economies of scale in government service provision. They emphasized the importance of separating the effects of population size from other characteristics of cities -- such as density or poverty – which might be correlated with size, and affect the cost of providing services. In their own work, Duncombe and Yinger separated the effects of population (as a measure of city size), technical economies of scale in production (the focus of traditional engineering studies), and environmental conditions (conditions under which production takes place; e.g., population density), on the efficiency with which public services are produced.
Looking at data from all 188 full-time fire departments in New York in the mid-1980s, Duncombe and Yinger found costs to be constant or increase with population. The implication is that “consolidating small fire departments will not result in significant cost savings.” (p. 70).
Earlier work by Yinger with Helen Ladd (Ladd and Yinger 1989) examined fire and police services in 86 major central cities in the U.S. For fire protection services, they found that the number of residents served did not affect per-capita costs. For police protection services, they found per-capita costs to be either constant or increasing with the number of residents served.
One recent study of police services in the Los Angeles metropolitan area presents evidence of diseconomies of scale in police provision.[1] Miles Finney (1997) used panel data techniques (which control for unobserved differences across cities, such as city specific regulations) to examine police costs in 14 communities in Los Angeles County over a four-year period. Some of these cities contracted with the Los Angeles County Sheriff (large scale service provision), others produced services for themselves (on a smaller scale). Finney found that the police departments produced safety (the inverse of the crime rate) and arrests under decreasing returns to scale. In other words, safety was best produced on a smaller scale. Finney concluded that contracting with the County was not a good deal for the contracting cities. He noted, “Whatever quantity of police services the contracting municipalities are presently purchasing could have been produced by the individual jurisdictions at a lower cost.” (p. 126).
Finney’s findings are consistent with work by Mehay two decades earlier (1979). Mehay examined manpower, property crime and violent crime in the Los Angeles metropolitan area. Controlling for other factors, he found that 25 contract cities had fewer police on patrol and were subject to more crime than residents in 46 independent cities (served by small municipal police departments).
Looking Closer at Police Provision
Elinor Ostrom used a different method to examine police service provision. Ostrom (1976) describes the “most similar systems” research design, in which neighborhoods are paired on the basis of race, density, age and wealth. Neighborhoods served by small police departments were compared to similar neighborhoods served by large police departments (in Indianapolis, Grand Rapids, Nashville-Davidson County, and St. Louis). Researchers examined measures such as police response, criminal investigation, emergency services, and other neighborhood level police services. They found that small to medium sized departments consistently performed more effectively and frequently at less cost than did large police departments (here large is in the range of serving one million residents). Ostrom was one of the first to point to the lack of evidence to support claims of benefits from municipal consolidation.
Furthermore, Ostrom, along with Roger Parks (1973), examined National Opinion Research Center (NORC) survey responses on crime for 109 cities with populations greater than 10,000. Both feelings of safety and ratings of police honesty were found to decrease with city size. Ostrom and Parks checked what these cities were spending on police, and found that the better service could not be explained by greater per-capita expenditures in the smaller cities. When they focused on responses of central city respondents specifically, they found city size and positive feelings about services were positively related -- up to a city size of 100,000 residents. Beyond this, the direction of the relationship reversed; residents of larger cities were less satisfied with police services than their neighbors in smaller cities.
Summarizing the findings on the provision of municipal services, it appears that the smallest efficient cites are surprisingly small. There is no evidence that services can be provided more cheaply by very large municipal governments, and some evidence that large service providers are, indeed, more expensive.
Hirsch (1968) explained the logic behind this. He labeled the majority of services provided by local government as “horizontally integrated.” By this he meant that additional residents are served by adding yet another identical police station or park or fire department. All of the units offer the same service, and there are no advantages of central provision that might result in economies of scale for these services. If city services must be replicated from one neighborhood to the next (parks, police patrols, fire stations), then certainly small cities can be as efficient as larger cities in producing city services.
An important point is that, in urban areas where contracting-out is an option, even the smallest cities can take advantage of any economies of scale that exist in the production of municipal services. By contracting with a neighboring public or private service provider, residents who want very local government (perhaps 5,000 residents) can pay competitive prices for local services.
III. Politics and Participation
Jurisdiction size affects the relative political power of groups that have an interest in city policies and spending. Size affects the costs stakeholders face in influencing public policy. Some groups benefit from size, for others, city size reduces their influence over local politics.
One reason power shifts is that city size may influence the monitoring function of residents (Davis and Hayes 1993). In large cities, residents tend to be removed from governments’ day-to-day actions. The concept of “rational ignorance” was coined to refer to the detachment from government that results when information is costly (Downs 1957). Residents compare the costs of participation to the benefits; when costs of information rise, participation rationally falls.
Nellor (1984) suggests that rational ignorance increases with jurisdiction size, increasing the power of bureaucrats to pursue their own agendas. In support of this premise, Oliver (2000) finds that resident participation increases as municipal jurisdiction size falls. Specifically, he finds that participation in a small city within a dense metropolitan area is similar to that observed in smaller cities outside large metropolitan areas. With respect to voting, researchers at the Public Policy Institute of California found a negative association between voting and city size (Hajnal 2002).
As cities grow in size and it becomes increasingly costly for residents to follow local government actions, residents pay less attention to city government actions and their function as monitors weakens. With less oversight, politicians are more likely to court campaign contributions by supporting public policy initiatives that benefit special interests at the expense of the city as a whole. As a result, public policy shifts marginally to serve the objectives of special interests.
Martin and McKenzie (1975) model the behavior of bureaucrats as part of local government consolidation. They conclude that consolidations simultaneously create monopoly power that can be exploited. Bureaucrats can absorb these gains in the form of higher salaries, nicer working conditions, and by defining easier jobs. Similarly, Persson and Tabellini (1994) argue that centralizing the provision of services creates opportunities for rent-seeking. They predict an increase in the size of government and overprovision of local public goods.
Where there are many cities (jurisdictional fragmentation), it is hypothesized that competition for residents and businesses leads to better public services at lower cost. Research tends to support the hypothesis that competition among governments leads to leaner governments. Eberts and Gronberg (1990) attempted to reconcile earlier contradictory results using state level data.[2] They looked at more than 200 metropolitan areas in the late 1970s, examining suburbs, central cities and other jurisdictions separately. They found that leaner governments where competition among local governments was greatest. Stansel (2002) found an association between jurisdictional fragmentation and economic growth.