Does User-Charge Financing Reduce Expenditure Levels for the Charge-Financed Services?

Abstract

Spurred by the tax and expenditure limitation movement of the late 1970s, user charges have grown to become a major source of revenue for local governments. Both theory and literature propose that a well-designed user-charge structure could reduce spending on the charge-financed services by strengthening the relationship between service and payment. This study empirically tests such a possibility by examining the level of expenditure for sewer and parks and recreation services. Using a panel of 686 American cities for the sewer services and 715 cities for the parks and recreation services between 1972 and 2004, this study finds evidence that greater reliance on user charges to finance government services leads to a reduction in municipal expenditures.

Keywords

user charge municipal expenditures fiscal illusion sewer service parks and recreation service

American cities have been more reliant on user charges and fees since the tax and expenditure limitation (TEL) movement of the late 1970s. This move by cities away from general tax financing can be seen by the fact that, in 2002, user charges made up 42% of municipal total own-source revenue (excluding liquor stores and social insurance trust revenue) whereas property taxes only generated 23% of the total, compared with 33% and 37% in 1972. In other words, for every US$1 in property taxes collected, municipal governments, on average, received US$1.83 in revenue from user charges in 2002, compared with US$0.89 in 1972. In addition, per capita municipal user charges in real terms have almost doubled over the past three decades (U.S. Census Bureau, 1977, 2002).

User charges are prices charged by governments for voluntarily purchased and publicly provided goods or services that are closely associated with basic government responsibilities (Mikesell, 2007). They are different from general taxes in that user-charge financing restores a direct relationship between the service received and payment rendered. This enables consumers to properly evaluate the service and use the service at an optimum level. From this reasoning, scholars have proposed that greater reliance on user-charge financing leads to a more efficient provision of services than does the tax financing (Bailey, 1994a; Wagner, 1976). With the exception of studies by Bierhanzl and Downing (1998) and Jung, Eom, Kim, and Sun (2009), which found that an increase in user-charge reliance reduces the total expenditure for sewer service (Bierhanzl & Downing, 1998), parks and recreation, and garbage collection services (Jung et al., 2009), no other studies have empirically examined whether the degree of reliance on user charges affects the expenditure level of charge-financed services, specifically at the national level.

To fill this gap, this study investigates the influence of user-charge reliance on the level of municipal expenditures for two specific public programs—the sewer service and the parks and recreation service. Using a panel of 686 American cities for sewer services and 715 cities for the parks and recreation services between 1972 and 2004, this study finds strong evidence that a greater degree of user-charge reliance in the sewer, and parks and recreation services leads to reduced government expenditures for these functions. The results of this study could be used to set up a well-designed user-charge system that could improve public service provision by allocating the available supply to those who value it most and by providing local officials with accurate information on the desirability of expanding a service (Bierhanzl & Downing, 1998). As this study is one of the first analyses examining a longitudinal panel data set for a large number of national sample cities, it is expected to provide additional methodological insights into the research topic.

The remainder of the article is organized as follows. The section below provides a brief overview of the pricing system of user charges. The next section sketches the theoretical framework for understanding the effect of user-charge financing on expenditures, followed by the section that reviews the existing literature on the subject. This is followed by the section that describes the data and outlines the methods employed in the study. The next section reports major findings, and the last section concludes with policy implications, study limitations, and suggestions for future research.

Utilization of User Charges

User charges at the municipal level are commonly used in the areas of transportation, energy, environment, health, and housing services (Netzer, 1992). In most cities, however, charges typically do not cover the full costs (current operating cost plus capital cost) of the service except for a few service areas such as parking services and gas and electricity supplies. The remaining costs are usually covered (or subsidized) with a combination of general funds (mainly taxes), intergovernmental grants, and other funding sources (Bierhanzl & Downing, 2004; Downing, 1992; Moulder, 2002). This variation reflects a combination of economic, political, and practical considerations. In economic theory, user charges are applicable where public goods and services resemble private sector outputs. That is, these goods and services provide significant private benefits to identifiable individuals or business firms; it is feasible to exclude nonbeneficiaries; and the costs of services are measurable (U.S. Advisory Commission on Intergovernmental Relations [U.S. ACIR], 1987). This theory of user charges also depends on political and practical factors, particularly consumer acceptance, equity issues, externalities, and administrative costs (Bierhanzl & Downing, 2004). For example, many cities charge nearly full costs of services such as sewerage, sanitation, and water supply because a large portion of the benefits of these services are private and most citizens readily understand and accept user charges for such services.

In contrast, many municipal governments subsidize their recreation facilities by offering free entry to municipal parks—especially when the administrative cost of a complex fee structure is too high—or charging a reduced price for youth groups or low-income residents, because participation in after-school recreation programs could generate benefits that spill over to the community at large (e.g., reducing juvenile delinquency), and because imposing full charges that cover the total costs of services on the poor might be inequitable (Bierhanzl & Downing, 2004). Other examples include public mass transit, hospitals, and education, which are partly fee-based services but also heavily publicly subsidized. These types of goods are often considered merit goods, which are not suitable for private sector production due to positive externalities and equity concerns. In addition, splitting the financing of these goods between user charges and taxes may also help reduce the tax costs and make the service more politically feasible.

Reflecting the above financing considerations, user-charge reliance—revenues from user charges as a percentage of expenditure for selected services—varies widely among different services. Table 1 illustrates this variation based on aggregate data over all U.S. cities. On average, municipal user charges for highways cover about 6% of total highway expenditure in 2002, 22% for parks and recreation, 58% for solid waste management, 62% for hospitals, 81% for airports, 92% for sewerage, and 121% for parking. Historically, the level of user charges for most services remained relatively stable, with the exception of sewerage, solid waste management, and public transit services. Charges for sewerage and solid waste management showed steady increases over the past 30 years whereas public transit charges experienced a considerable decline. These national average figures provide some insights into the difference between the cost of services and how much citizens have to pay directly as a charge, although variations in service quality may exist across regions and cities.

Table 1.

User Charges as a Percentage of Expenditure for Selected Functions

Functional category	1972	1982	1992	2002
Parking	119.84	92.89	113.29	121.15
Electricity supply	96.22	96.75	100.61	103.36
Gas supply	111.59	105.41	106.72	100.25
Sewerage	46.31	53.10	83.00	92.21
Water supply	97.18	81.59	86.52	85.90
Airports	61.83	74.00	70.98	80.51
Water transport	56.80	81.28	115.87	70.01
Hospitals	40.42	66.58	60.76	62.20
Solid waste management	23.44	32.82	49.79	58.28
Public transit	64.65	48.14	33.62	34.44
Parks and recreation	15.22	17.77	22.21	21.76
Housing and community development	16.90	11.05	14.52	11.96
Highways	3.96	4.85	6.23	5.54
Education	3.70	3.14	2.10	1.57

Source: Authors’ calculations from U.S. Census Bureau, Historical Finances of Federal, State and Local Governments: State Aggregates (Fiscal Years 1972-2004).

The differences in user-charge reliance may also reflect the differences in pricing structures. For example, the sewerage charge formulas vary between municipalities and sanitation districts. Some municipalities simply charge a flat rate per month or per quarter for all service users or all users within a group (e.g., US$10 per month for single family residence, and US$7 per month for each unit in a multifamily dwelling), whereas other jurisdictions adopt more complicated formulas based on the amount of water used, plumbing fixtures (such as garbage grinders in residences, wash-racks in service stations, and washing machines in commercial laundries), water meters and sewer connections, assessed property values, or a combination of two or more of these factors. Sewer strength (or pollution potential) is frequently used as the basis for pricing industrial sewage (Bierhanzl & Downing, 2004; Johnson, 1969).

In the case of parks and recreation services, the International City/County Management Association (ICMA)’s 2001 survey (for 1,563 cities with varying population sizes and for 215 counties with populations of 50,000 and above) provides recent information on the use of charges (Moulder, 2002). According to the survey results, about 89% of responding local governments provides parks and recreation services; if the government does not, another organization or governmental entity provides the service. In regard to the funding sources, overall, general fund and user charges are identified as the most important sources for parks and recreation operating budgets. Specifically, of the 1,334 jurisdictions (75% of the sample) that use the general fund as a funding source, the average percentage from the general fund is approximately 75% of the budget; of the 936 jurisdictions (53% of the sample) that rely on user charges, the average percentage is 28%, with nine jurisdictions funding 100% of their parks and recreation operating budgets through user charges. Other funding sources include intergovernmental aid, private grants, and fund raising by the local government.¹ Recreation facilities that generate user-charge revenues include community centers, playgrounds, tennis courts, swimming pools, and so forth. Although most local governments provide certain services, such as playgrounds and hiking trails, for free, they also impose fixed or variable charges for select services such as swimming pools and public golf courses.

Theoretical Framework

Public finance and public choice theory considers bureaucratic behavior to be the result of “fiscal illusion” (Buchanan, 1967; Goetz, 1977; Puviani, 1903). This theory suggests that when government revenues are unobserved or not fully observed by taxpayers, the cost of government is perceived to be less expensive than it actually is. Thus, from the politician’s perspective, the existence of a fiscal illusion enables a politician to select a revenue mechanism that maximizes the extent to which spending can be increased without attracting public attention, while at the same time, this same revenue mechanism will tend to promote greater-than-optimal consumption on the part of taxpayers by concealing from them the true costs of public services (Wagner, 1976).

Ideally, a properly priced user-charge financing scheme could restore the direct relationship between payment rendered and service received; this has led to much theoretical discussion hypothesizing that user charges could break this fiscal illusion by providing taxpayers with more accurate cost signals concerning their consumption, reducing the ability of a government agency to extract rent from taxpayers (Bailey, 1994a; Bierhanzl & Downing, 1998, 2004). Proponents of this argument reason that user charges act as a market mechanism in the public sector, allowing consumers to pay the full or majority of the cost of the service, and adjust consumption accordingly (Bierhanzl & Downing, 1998). The underlying assumption of this argument is that public demand for charge-financed goods or services is fairly elastic or at least not perfectly inelastic, so that resource allocation and service utilization would be sensitive to the pricing system of user charges.

User charges may also provide public officials with invaluable information about consumer preferences for public goods and services, allowing them to make more effective long-run investment decisions concerning the types and levels of services provided. Such policy-making patterns could prevent local governments from “providing the wrong level of output at too high a cost to the wrong people” (U.S. ACIR, 1987, p. 31). In fact, U.S. ACIR, (1987) suggests that government may expand local services “only if direct users are willing to pay the full costs of the expansion” (p. 3).

These arguments have led scholars to propose that a greater degree of reliance on user-charge financing could result in efficiency gains by reducing the level of service demands as well as cost savings through more efficient service delivery, compared to a traditional tax financing scheme, provided that the administrative costs of collecting charges are relatively low and alternative tax measures related to service usage can be devised (U.S. ACIR, 1987; Bailey, 1994a; Bierhanzl & Downing, 1998, 2004; Wagner, 1976). Although the primary focus of this article is to examine whether a greater degree of reliance on user-charge financing would result in the level of charge-financed services (or allocative efficiency), it should be noted that equity issue associated with the use of user charge is equally important, which needs to be analyzed in future studies.²

Previous Studies

The literature is replete with studies of efficiency gains and cost-reduction effects of user-charge financing applied to health services, natural resources, and transportation.³ Less work has been done considering basic services such as water, sewer, and parking. Examining municipal solid waste management services, Savas, Baumol, and Wells (1977) found that no relationship exists between the type of financing methods (e.g., taxes, flat fees, and variable fees) and the volume of waste generated, the frequency of collection, or the place of collection. Unfortunately, their study suffers from several methodological limitations, notably the lack of statistical correlation measures and regression analysis controlling for factors that may affect the demand and supply of the service. Mercer and Morgan (1983) compared the actual user charges with a projected cost time series using data on 34 cities and 22 counties in California between 1977 and 1982 and found that overall, cities performed better than counties in efficiency measures over time and small counties performed better than did large counties. This study is limited by the fact that their method of cost determination did not involve capital costs, an important part for capital-intensive services such as water and sewer services, biasing their efficiency measures, which are based on the ratio of user charges to such estimated costs. Bierhanzl and Downing (1998) was the first empirical study that directly and systematically investigated whether a higher degree of reliance on user-charge financing results in improved efficiency measured by the expenditure on charge-financed services. Using data from the 1990 Census Survey of Governments, Bierhanzl and Downing evaluated the sewer service of 751 U.S. cities in 1990 and found that the employment of user-charge financing mechanism significantly reduced government spending levels on sewer services.

Bierhanzl and Downing’s seminal work provided a useful analytic approach in connecting the theoretical argument about the efficiency-enhancing effect of user charges to the actual economic effect of user-charge financing. However, their study is limited by several factors. First and foremost, Bierhanzl and Downing’s study did not control for quality and output of the public service, thus their conclusions about efficiency gains through user-charge financing should be interpreted with great caution. Second, their study only analyzed a limited sample for a single cross-sectional year. Analysis of a longitudinal data set is necessary to appropriately account for initial and recurring capital costs. Third, Bierhanzl and Downing’s study examined only the sewer service.⁴ Because of extensive variation in the degree of user-charge reliance and the resulting level of government spending on charge-financed services, analysis of a pooled data set for multiple service types is necessary to reveal more detailed information on the research topic. Fourth, Bierhanzl and Downing’s analysis only controlled for a limited number of factors that were expected to affect government expenditures on sewerage, such as population, growth rates of population, population density, income, and the percentage of owner-occupied housing. However, municipal expenditures often depend on multiple political, demographic, economic, and fiscal factors. A more thorough analysis should take into account these features that vary across cities and over time. Last but not least, in Bierhanzl and Downing’s empirical model the dependent variable—total expenditure—is also a part of the explanatory variables in the statistical equation, which may cause bias through simultaneity problems. Alternative model specifications should be developed to correct this problem.

Jung et al. (2009) empirically tested the effects of the reliance of user charges on the level of parks and recreation and garbage collection services for a panel of 546 Georgia cities from 1990 to 2007. Their study found that a greater degree of user-charge reliance for the two services reduced expenditure levels although the magnitude of the reduction effects of garbage collection service is a bit larger than that of parks and recreation services. Although Jung et al.’s study advanced the topic by employing a longitudinal data set, their study is still limited to a single state, and thus a longitudinal study using a national sample is expected to further strengthen the generalizability of previous findings.

This study conducts a more thorough empirical analysis that builds on Bierhanzl and Downing’s conceptual framework and Jung et al.’s empirical study to address the above limitations with an alternative model specification and a national sample.

Method

Data and Unit of Analysis

This study extends the time frame of Bierhanzl and Downing’s work from a limited cross-sectional study to a cross-sectional time series analysis for sample cities with 1970 populations of 25,000 or above during a 33-year period (1972-2004). Data were collected from multiple sources: The Census Bureau’s Annual Survey of Local Government Finances and Census of Government (1970-2004); the County and City Data Book (1972, 1977, 1983, 1988, 1994, and 2000); and the ICMA’s Municipal Form of Government Survey (1981, 1986, 1991, 1996, and 2001) and Municipal Year Book (1971, 1974, and 1977). Due to missing values (in particular total expenditures and tax revenues) for some cities for at least 1 year during the study period, the data are unbalanced panels. The data for the sewer service (“sewer sample”) consists of 20,402 observations for 686 cities from 1972 to 2004 or 90% of the possible city-years, and the data for the parks and recreation service (“parks sample”) consists of 21,955 observations for 715 cities from 1972 to 2004 or 93% of the possible city-years.

This study analyzes two specific charge-financed municipal services—sewer service and parks and recreation service—rather than a single service or an aggregate of user charges across different services. This allows for comparisons between services and obtains a clearer understanding of the effects of user charges. The primary reason for choosing sewer and parks and recreation services is that the budgetary norm for public sector sewer services has been full-cost recovery, whereas the common practice for parks and recreation services has been to cover only operating costs (Bierhanzl & Downing, 1998; Netzer, 1992). Sanitary sewer systems (i.e., systems for collection and treatment of sewerage from houses or industry) are among the most common charge-financed services, with benefits that are clearly identifiable and easy to administer (i.e., they are commonly added to user’s water bills) and there is little public opposition to charges for such service (Bierhanzl & Downing, 1998).⁵ In contrast, user charges are of limited use in public parks. Equity concerns and the general public benefits from recreation services limit user charges collected in parks to parking fees and franchise fees. Recreation departments are probably under considerable pressure to adopt some form of benefit-based charges. Thus, user charges for parks and recreations service typically recover between one-quarter and two-thirds of operating costs (Bailey, 1994b). This contrasting situation provides a good avenue to conduct a comparative study for the sewer service and parks and recreation service.

Additionally, sewer service and parks and recreation service are two of the most important municipal functions, and the majority of cities with a population exceeding 25,000 tend to provide these services either fully or partially (U.S. Census Bureau, 1962-2004). According to the Census, municipalities spent about US$43 billion on wastewater operations and capital projects in 2007 compared to US$1.4 billion spent by state governments (U.S. GAO, 2010). Furthermore, charge collections and total expenditure for these two services are clearly disaggregated in the Census data. All of these factors make sewer and parks and recreation services good candidates for analysis.

Variables and Model

This study proposes the following hypothesis: A higher degree of reliance on user-charge revenue reduces the level of government expenditure for the charge-financed service.

Two dependent variables in this study are total sewerage expenditures and total parks and recreation expenditures. The Census data for both expenditure types include current operating cost plus capital cost of the service delivery during the 33-year study period from 1972 to 2004. The independent variables of primary interest are user-charge reliance (UCR) for sewer services and parks and recreation services, defined as user-charge revenue for the service divided by expenditures for that service. UCRSEW denotes user-charge reliance of sewer service (UCRSEW = sewerage charges/total sewerage expenditure), and UCRPARK represents user-charge reliance of parks and recreation service (UCRPARK = parks and recreation charges/total parks and recreation expenditure). As such, user charges are measured in a relative term, as opposed to an absolute dollar amount. This helps determine how closely the charge revenue approaches the full budget cost of the charge-financed service, and the measure can also be easily applied to jurisdictions of varying size. To alleviate a potential simultaneity problem caused by the presence of expenditure in both the dependent and primary independent variables, a 3-year moving average of UCR is calculated and enters the statistical model instead of the annual UCR. Although this measurement cannot completely eliminate the simultaneity problem, it helps smooth out the trends in government expenditures and revenues and reduces the estimation bias. Caution should still be exercised, as the results might be biased upward to a certain extent due to the potential simultaneity issue.

As expenditure reflects demand for the level of output, a method for estimating demand functions of individuals for municipal public services is required (Deacon, 1979; Deno & Mehay, 1987). The median-voter hypothesis states that a local government elected by the majority provides the level of service that is most preferred by the median voter (Bergstrom & Goodman, 1973; Bierhanzl & Downing, 1998; Deacon, 1979; Deno & Mehay, 1987).⁶ Basic median voter’s demand function for governmental services can be expressed as:

In G = \ln A + b_{1} \ln T_{j} + b_{2} \ln N + b_{3} \ln Y_{j}

where G is the total amount of public goods supplied; A is any underlying differences in functional responsibility for services across cities and other differences in cities themselves;⁷ T_j is the median voter (j)’s tax share of the local public good, which measures the increment in the household’s tax burden caused by an extra dollar of public expenditure; N is the number of people sharing the local public good (i.e., city population); and Y_j is median voter (j)’s income.

Given the key role of the property tax in local finance, the median voter’s tax share is usually defined as the ratio of median home value to the gross assessed property value in a jurisdiction, which is citizens’ share of local property tax bill. Unfortunately, for this study, although the data set for median home value is available for 1970, 1980, 1990, and 2000, the gross assessed property values for all sample cities during the study period are not available. Therefore, the median voter’s tax share is not included in the regression model.

The general equation for the expenditure model of charge-financed services, thus, can be written as:

E X P_{i t} = α_{1} U C R_{i t} + F_{i t} α_{2} + X_{i t} α_{3} + δ_{i} + ε_{i t}

where EXP_it is total expenditure for sewer service or for parks and recreation service for municipality i in year t; UCR_it is a 3-year moving average of user-charge reliance defined as the average of user-charge revenue for the charge-financed service for municipality i during the previous 3 years (i.e., years t, t-1, and t-2) divided by the 3-year average of expenditure for that service; F_it is a vector of fiscal variables for municipality i in year t, including total intergovernmental revenue (IGR) and the number of special districts (i.e., “sewer districts” for sewer service, and “park districts” for parks and recreation service); X_it is a vector of socioeconomic variables for municipality i in year t that are expected to affect the demand of the services, including population, land area, median family income, percentage of owner-occupied housing, percentage of population age 65 and above, and percentage of non-White population; δ_i is a city-specific intercept; and ϵ_it is the error term.

All variables except dummy variables are in log form. In addition, all monetary terms are adjusted for inflation using the Consumer Price Index (CPI) with 2004 as the base year.

Population, land area, and median family income are included to account for the demand for and cost of public services. Intergovernmental revenue from federal, state, and other local governments, in particular the unrestricted block grants, can be counted as additional income for city residents if the municipal government substitutes the grants for locally raised revenue. In the case of matching grants, there may be a price effect as well as an income effect on the demand for local public services in that the matching grant may reduce the tax price to city residents by the fraction of the unit cost of the local public good financed by the grantor (Oates, 1979). Under fiscal illusion, voters may fail to observe the lump-sum grant or misperceive its impact as an average price effect and thus are willing to support a higher level of government spending. Nevertheless, if the municipal government—the grantee—acts as a budget-maximizer and chooses to spend more than the amount of the intergovernmental grants (i.e., the “flypaper effect”), there is no substitution effect (Courant, Gramlich, & Rubinfeld, 1979; Holsey, 1993; Romer & Rosenthal, 1982; Turnbull & Mitias, 1999).⁸ In either case, intergovernmental revenue is expected to have a positive effect on government expenditures.

Because sewer service and parks and recreation service in some regions are provided through special districts rather than municipal governments, the number of special districts within the county where the municipality is located is included as a control variable.⁹ The Census Bureau classifies 33 types of special districts, including single-function special districts and multifunction districts, according to their designated functions. We count the number of sewer districts including single-function sewerage districts and multifunction sewerage-and-water supply districts, and the number of park districts including single-function parks and recreation districts only. A larger number of special districts are expected to result in a lower level of municipal spending on the services that are provided fully or partially from the special districts.

To capture voter preferences for public services, percentage of population age 65 and above, percentage of population that is non-White, and percentage of owner-occupied housing units are employed in the study. According to the renter’s illusion hypothesis, as renters do not directly pay property taxes and thus are less aware of tax increases than homeowners, renters are more likely to favor expansions in the local budget than homeowners. If this is true, the percentage of owner-occupied housing would be negatively associated with the government spending level (Bergstrom & Goodman, 1973; Martinez-Vazquez, 1983).

Research Findings

This study employs a fixed effects regression model with cluster-robust standard errors. The fixed effects model helps to control for any omitted unobserved factors that differ across cities but are constant over time. In addition, the standard errors are clustered on the variable identifying each municipality to account for intragroup variations, which helps correct potential heteroscedasticity and autocorrelation problems.

Tables 2 and 3 present descriptive statistics of the variables for the sewer sample and for the parks sample, respectively. As the tables show, there are large variations in government spending levels, revenue structures, and socioeconomic characteristics among the sample cities, which justifies the use of regression models to capture the effects of user-charge financing on municipal expenditures.

Table 2.

Descriptive Statistics for the Sewer Sample

Variables	M	SD	Min.	Max.
Dependent variable
Sewerage expenditure (million US$)	15.39	53.78	0.002	1,823.47
Independent variables
UCR sewerage (3-year moving average)	0.90	0.86	0	23.51
Sewerage charges (million US$)	11.30	36.16	0	1,068.05
Sewer districts	4	16	0	280
Intergovernmental revenue (million US$)	85.56	786.81	0	23,900
Population	134,444	379,730	25,019	8,084,316
Land area (square miles)	42.44	71.62	1.00	765.70
Median family income	44,432	13,901	14,409	141,964
Owner-occupied house (%)	57.34	11.96	13.40	93.90
Age above 65 (%)	12.04	4.05	1.50	51.80
Non-White population (%)	19.47	17.75	0	98.80

Note: 20,402 observations.

Table 3.

Descriptive Statistics for the Parks Sample

Variables	M	SD	Min.	Max.
Dependent variable
Parks and recreation expenditure (million US$)	11.05	30.53	0.002	719.43
Independent variables
UCR park (3-year moving average)	0.20	0.18	0	2.17
Parks and recreation charges (mil US$)	2.23	5.56	0	157.64
Park districts	2	9	0	94
Intergovernmental revenue (million US$)	81.71	758.89	0	23,900
Population	130,948	366,785	25,019	8,084,316
Land area (square miles)	43.36	91.66	1.00	1,732.00
Median family income	44,551	14,021	14,409	141,964
Owner-occupied house (%)	57.23	12.04	13.40	93.90
Age above 65 (%)	12.02	4.14	1.50	51.80
Non-White population (%)	20.21	17.99	0	98.40

Note: 21,955 observations.

Tables 4 and 5 summarize the distribution of the 3-year moving averages of UCRSEW and UCRPARK, respectively. For analytical and regression purposes, UCRSEW and UCRPARK are grouped along with five brackets at an interval of 0.5 for UCRSEW and 0.1 for UCRPARK. For example, UCRSEW 0-0.5 indicates a UCRSEW score between 0 and 0.5 and UCRPARK 0-0.1 indicates a UCRPARK score between 0 and 0.1; other brackets follow this pattern.

Table 4.

Distribution of UCR Sewerage (UCRSEW)

Variables	M	SD	Min.	Max.	No. of observations
UCRSEW	0.90	0.86	0	23.51	20,402	(100%)
UCRSEW 0-0.5					5,256	(25.8%)
UCRSEW 0.5-1.0					7,353	(36.0%)
UCRSEW 1.0-1.5					5,680	(27.8%)
UCRSEW 1.5-2.0					1,369	(6.7%)
UCRSEW 2.0 or higher					744	(3.7%)

Note: UCRSEW is a 3-year moving average derived from 686 samples cities for 33 years (1972-2004), which generate 20,402 total observations (excluding missing values).

Table 5.

Distribution of UCR Parks and Recreation (UCRPARK)

Variables	M	SD	Min.	Max.	No. of observations
UCRPARK	0.20	0.18	0	2.17	21,955	(100%)
UCRPARK 0-0.1					7,367	(33.5%)
UCRPARK 0.1-0.2					5,655	(25.8%)
UCRPARK 0.2-0.3					4,060	(18.5%)
UCRPARK 0.3-0.4					2,383	(10.9%)
UCRPARK 0.4 or higher					2,490	(11.3%)

Note: UCRPARK is a 3-year moving average derived from 715 samples cities for 33 years (1972-2004), which generate 21,955 total observations (excluding missing values).

At first glance, 38.2% of sample cities (with UCRSEW ≥ 1.0) rely on user charges to cover the full budget cost of sewer service (see Table 4). However, Bierhanzl and Downing (1998) suggests that a typical municipal budget does not include depreciation or rent of previous capital investments and land expenditure, thus it is not clearly known how much the budget cost understates the true full cost of service delivery. As sewer service is capital intensive, Bierhanzl and Downing (1998) proposes that a UCRSEW of 1.5 or 2.0 could be more representative of a charge structure that covers the full cost. If this is the case, only about 10% of total observations cover the full cost of the service provision and the rest of them (90%) set charges below the full cost. This suggests that sewer service is heavily subsidized by general taxes or other revenue sources.

In the case of the parks and recreation service, it is less well-known what level of the UCRPARK measure would indicate full coverage of the cost of service delivery. However, parks and recreation services tend to be more heavily subsidized by general taxes or other revenue sources than sewerage due to the service’s generation of positive externalities and equity consideration for those who cannot afford the charges for entry into the facilities (Downing, 1992). Nonetheless, it should be noted that only 11.3% of sample cities have UCRPARK scores 0.4 or higher as Table 5 shows. Given that the ICMA’s 2001 survey reports that, on average, user charges account for 99.3% (UCRSEW 0.993) of local government annual sewerage expenditures and 21.6% (UCRPARK 0.216) of annual parks and recreation expenditures, respectively, the sample means (UCRSEW 0.90 and UCRPARK 0.20) in this study are very close to the national averages (Bierhanzl & Downing, 2004). It should be acknowledged that the sample means in this study cover a 33-year period (1972-2004) for municipal governments only whereas the figures of ICMA apply to a single year (2001) and include both municipalities and counties.

Tables 6 and 7 report the regression results for the impact of user charges on total service spending under various model specifications: Model I is run with UCR as the primary independent variable; Model II employs four brackets of UCR dummy variables to understand the effect of different magnitudes of the UCR on expenditure reduction; and Model III uses total sewerage charges or parks and recreation charges in an effort to avoid the simultaneity problem that may exist in Model I.¹⁰

Table 6.

Effects of Sewerage Charge Reliance on Sewerage Spending (ln)

	Model I	Model II	Model III
UCRSEW (ln)	−0.217 (0.031)***
UCRSEW 0.5-1.0		−0.121 (0.050)**
UCRSEW 1.0-1.5		−0.313 (0.059)***
UCRSEW 1.5-2.0		−0.513 (0.078)***
UCRSEW 2.0 and higher		−0.855 (0.103)***
Sewerage charges (ln)			0.252 (0.024)***
Sewer districts (ln)	0.038 (0.028)	0.053 (0.029)*	0.031 (0.025)
IGR (ln)	0.123 (0.028)***	0.152 (0.030)***	0.160 (0.026)***
Population (ln)	0.786 (0.115)***	0.797 (0.130)***	0.597 (0.109)***
Land area (ln)	0.303 (0.126)**	0.304 (0.132)**	0.070 (0.111)
Median family income (ln)	−0.289 (0.055)***	−0.184 (0.055)***	−0.164 (0.051)***
% Owner-occupied housing (ln)	−0.244 (0.358)	−0.158 (0.348)	−0.417 (0.295)
% Age 65 (ln)	0.125 (0.094)	0.057 (0.101)	0.051 (0.090)
% Non-White (ln)	0.293 (0.035)***	0.253 (0.036)***	0.087 (0.025)***
Intercept	0.475 (2.095)	−1.000 (2.152)	1.000 (1.761)
R ²	.439	.415	.618

Note: Total 20,402 observations. Standard errors in parentheses are adjusted for clustering within municipalities. All specifications include city-fixed effects.

significant at the 10% level. **significant at the 5% level. ***significant at the 1% level.

Table 7.

Effects of Parks and Recreation Charge Reliance on Parks and Recreation Spending (ln)

	Model I	Model II	Model III
UCRPARK (ln)	−0.074 (0.026)***
UCRPARK 0.1-0.2		−0.027 (0.039)
UCRPARK 0.2-0.3		−0.073 (0.049)
UCRPARK 0.3-0.4		−0.076 (0.063)
UCRPARK 0.4 and higher		−0.085 (0.083)
Parks and recreation charges (ln)			0.212 (0.020)***
Park districts (ln)	0.066 (0.032)**	0.059 (0.033)*	0.011 (0.026)
IGR (ln)	0.143 (0.038)***	0.178 (0.041)***	0.129 (0.032)***
Population (ln)	0.785 (0.132)***	0.816 (0.129)***	0.458 (0.111)***
Land area (ln)	0.426 (0.129)***	0.439 (0.126)***	0.292 (0.097)***
Median family income (ln)	0.087 (0.043)**	0.044 (0.050)	0.094 (0.038)**
% Owner-occupied housing (ln)	0.500 (0.344)	0.335 (0.354)	0.210 (0.296)
% Age 65 (ln)	0.198 (0.079)**	0.226 (0.084)***	−0.035 (0.072)
% Non-white (ln)	0.065 (0.025)**	0.063 (0.027)**	0.026 (0.022)
Intercept	−6.854 (2.311)***	−6.489 (2.297)***	−2.089 (1.907)
R ²	.557	.473	.721

Note: Total 21,955 observations. Standard errors in parentheses are adjusted for clustering within municipalities. All specifications include city-fixed effects.

significant at the 10% level. **significant at the 5% level. ***significant at the 1% level.

As Table 6 shows, the statistically significant and negative coefficient on UCRSEW (−0.217) in Model I implies that a 10% increase in UCR sewerage leads to a 2.2% reduction in total sewerage spending. This finding suggests that as the percentage (or share) of sewerage charge increases, total sewerage expenditure decreases. For example, suppose a city currently finances 60% of its sewerage expenditure from user charges (UCRSEW 0.6). If the city increases the degree of its user-charge reliance and finances 66% of its sewerage expenditure with user charges, its sewerage expenditure will decrease by 2.2%. Considering that the average spending on sewerage for the sample cities is about US$15.39 million annually, the 2.2% reduction would imply a savings of US$338,580 for the municipal government. As Bierhanzl and Downing (1998) points out, the reduction in spending could be attributed to the presence of a consumption-payment link and the breaking of a fiscal illusion in the service provision and consumption of sewer services.

Furthermore, the negative and descending coefficients on the four different brackets of UCRSEW dummy variables in Model II indicate that the higher the UCR sewerage level (bracket), the more sewerage expenditure is reduced.¹¹ In general, the finding suggests that the magnitude of spending reduction becomes larger as the proportion of charge financing increases.

In comparison, Model III shows that a 10% increase in the total amount of sewerage charges is associated with a 2.5% increase in sewerage spending. This positive relationship could occur in a situation where bureaucrats might have incentives to provide the charge-financed services at a more-than-needed level when a significant surplus of the user-charge revenue is available to them. Thus, this finding may serve as evidence of budget maximizing behavior on the part of bureaucrats. Alternatively, this finding could also be indicative of a flypaper effect. For example, suppose that a city uses both general revenue and user charges to fund its sewer service. If the city raises more funds for sewer service from user charges in a given year, it might spend even more on the service instead of diverting general revenues to other purposes, resulting in a positive relationship between user-charge reliance and government expenditures. However, as the direction of the causality could run two ways, another possible and simple explanation could be that an increase in spending requires an increase in user-charge revenue.

Table 7 shows the regression results for the effects of UCR parks and recreation on the level of parks and recreation spending. A pattern similar to sewerage spending is seen here. The results of Model I indicate that a 10% increase in UCRPARK leads to a 0.7% reduction in parks and recreation expenditures. The magnitude of the reduction effect is smaller than that of sewerage spending. This blunted effect may be due to the heavy subsidization by general revenue funding for the parks and recreation services. Also, as with sewer spending, Model III in Table 7 shows a positive impact of user charges on expenditures. That is, a 10% increase in the total amount of parks and recreation charges leads to a 2.1% increase in total parks and recreation spending. Unlike the results for sewerage spending, the negative and descending coefficients on the four different brackets of UCRPARK dummy variables in Model II fail to reach statistical significance at any conventional level, suggesting that the magnitude of spending reduction does not vary with the proportion of charge financing in the case of parks and recreation services.

With respect to the control variables, intergovernmental revenue, population, land area, and percentage of non-White population are positively associated with both sewerage expenditure and parks and recreation expenditure. The percentage of population age 65 and above is positively associated with parks and recreation spending but has no statistically significant impact on sewerage spending. Interestingly, the percentage of median family income has a positive impact on parks and recreation spending but a negative effect on sewerage spending. Another unexpected result is the effect of the number of special districts on government spending. Both the number of sewer districts and the number of parks and recreation districts show some evidence of positive impacts on expenditures, although the impacts are not consistent across different model specifications. Instead of employing the number of sewer and park districts, it would be ideal if we could employ the actual amounts of charges and expenditures for sewer districts and parks and recreation districts to precisely estimate their effects. However, as mentioned above, because most districts cover more than a city or county, it is difficult to disaggregate the portion of revenues and expenditure for an individual city, making testing difficult. The coefficients of the percentage of owner-occupied housing fail to reach statistical significance for either sewerage or parks and recreation spending. This is an interesting finding, suggesting that homeowners are not necessarily more sensitive to the cost of charge-financed services than renters.

Conclusion

This study examines the impact of the degree of reliance on user-charge revenue on the level of expenditure for sewer and parks and recreation services for a panel of American cities between 1972 and 2004 by employing fixed effects model of the pooled regression analysis.

Empirical findings in this study demonstrate that a greater degree of user-charge reliance on the sewer and parks and recreation services results in a lower level of expenditure for these charge-financed services. This study also finds that the magnitude of spending reduction of sewer services becomes larger when the proportion of charge financing increases as reflected in descending coefficients of four different brackets of UCRSEW dummy variables employed in the analysis, although this descending effect is not statistically significant for parks and recreation services. The findings in this study largely confirm previous literature that greater reliance on charge revenue for the charge-financed services results in a reduction of total expenditure (Bierhanzl & Downing, 1998; Jung et al., 2009). As explained in Bierhanzl and Downing (1998), this result could be linked to the fact that user-charge financing reduces the demand for government services “through the unbundling of those services, through a direct relationship between consumption and cost, and through the breaking of fiscal illusion” (p. 187).

By employing a more advanced methodology with a longitudinal national sample, this study improved our understanding on the topic. However, the conclusions made in this study are still tentative and several issues remain open to further research. First of all, having expenditures on both sides of the regression equation may bias the estimate results slightly upward, although this study tried to alleviate this problem by employing a moving average method. Second, as numerous factors that vary widely across cities (e.g., the possibility of alternative service delivery arrangements, charge cost practice,¹² the size of tax base, service preference of residents, and the quality of service¹³) could affect the size of government spending, caution should be exercised in interpreting the results of this study to conclude that higher reliance on charge revenue results in a reduction of total expenditure for the charge-financed service than financing primarily by tax revenues (Boyne, 1998a; Boyne, 1998b; Kodrzycki, 1998). Challenging as they are, future studies could consider controlling these factors influencing government spending appropriately.

Finally, future studies should expand the number of service types in the analysis to provide a more accurate picture of the relationship between the degree of user-charge reliance and the expenditure level. This will enable us to generalize the findings to other charge-financed services such as mass transportation, public hospitals, parking service, highways, and so forth.

Footnotes

The author(s) declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

The author(s) disclosed that they received the following support for their research and/or authorship of this article: This work was partially supported by the Lincoln Institute of Land Policy Dissertation Fellowship and the Inha University research grant.

Notes

References

Bailey

S. J.

(1994a). Charging for local government services: A coherent philosophy. Public Administration, 72, 365-384.

Bailey

S. J.

(1994b). User-charges for urban services. Urban Studies, 31, 745.

Bergstrom

T. C.

Goodman

R. P.

(1973). Private demands for public goods. American Economic Review, 63, 280-296.

Bierhanzl

E. J.

Downing

P. B.

(1998). User charges and bureaucratic inefficiency. Atlantic Economic Journal, 26, 175.

Bierhanzl

E. J.

Downing

P. B.

(2004). User charges and special districts. In Aronson

J. R.

Schwartz

(Eds.), Management policies in local government finance (5th ed., pp. 315-351). Washington, DC: International City/County Management Association.

Bowen

H. R.

(1943). The interpretation of voting in the allocation of economic resources. Quarterly Journal of Economics, 58(1), 27-48.

Boyne

G. A.

(1998a). Bureaucratic theory meets reality: Public choice and service contracting in U.S. local government. Public Administration Review, 58, 474.

Boyne

G. A.

(1998b). Competitive tendering in local government: A review of theory and evidence. Public Administration, 76, 695-712.

Buchanan

J. M.

(1967). Public finance in democratic process: Fiscal institutions and individual choice. Chapel Hill: University of North Carolina Press.

10.

Courant

P. N.

Gramlich

E. M.

Rubinfeld

D. L.

(1979). The stimulative effects of intergovernmental grants: Or why money sticks where it hits. In Mieszkowski

P. M.

Oakland

W. H.

(Eds.), Fiscal federalism and grants-in-aid. Washington, DC: Urban Institute.

11.

Creese

A. L.

(1991). User charges for health care: A review of recent experience. Health Policy and Planning, 6, 309-319.

12.

Deacon

R. T.

(1979). The expenditure effects of alternative public supply institutions. Public Choice 34, 381.

13.

Deno

K. T.

Mehay

S. L.

(1987). Municipal management structure and fiscal performance: Do city managers make a difference? Southern Economic Journal, 53, 627.

14.

Downing

P. B.

(1992). The revenue potential of user charges in municipal finance. Public Finance Review, 20, 512-527.

15.

Downs

(1960). Why the government budget is too small in a democracy. World Politics, 12, 541-563.

16.

Goetz

C. J.

(1977). Fiscal illusion in state and local finance. In Borcherding

T. E.

(Ed.), Budgets and bureaucrats: The sources of government growth. Durham, NC: Duke University Press.

17.

Griffin

D. C.

(1992). Welfare gains from user charges for government health services. Health Policy and Planning, 7, 177-180.

18.

Hines

J. R.

Jr. Thaler

R. H.

(1995). Anomalies: The flypaper effect. The Journal of Economic Perspectives, 9, 217-226.

19.

Holsey

C. M.

(1993). Price and income distortions under separate spending and taxing decisions. Journal of Public Economics, 50, 93-114.

20.

Johnson

J. A.

(1969). The distribution of the burden of sewer user changes under various charge formulas. National Tax Journal, 22, 472-485.

21.

Jung

Eom

T. H.

Kim

Sun

(2009). Longitudianl effects of user charges on the expenditure level of charge-financed services: The Georgia case (1985-2007). Municipal Finance Journal, 30(3), 1-28.

22.

Kennedy

P. E.

(1981). Estimation with correctly interpreted dummy variables in semilogarithmic equations. American Economic Review, 71, 801-801.

23.

Knight

B. G.

(2000). The flypaper effect unstuck: Evidence on endogenous grants from the Federal Highway Aid Program. Washington, DC: Board of Governors of the Federal Reserve System (U.S.).

24.

Kodrzycki

Y. K.

(1998, January/February). Fiscal pressures and the privatization of local services. New England Economic Review, 39.

25.

Martinez-Vazquez

(1983). Renters’ illusion or savvy? Public Finance Quarterly, 11, 237-247.

26.

Mercer

L. J.

Morgan

W. D.

(1983). The relative efficiency and revenue potential of local user charges: The California case. National Tax Journal, 36, 203-212.

27.

Mikesell

J. L.

(2007). Fiscal administration: Analysis and applications in the public sector (7th ed.). Belmont, CA: Thomson/Wadsworth.

28.

Moulder

E. R.

(2002). Financing parks and recreation: User fees and fund-raising as revenue sources. Washington, DC: International City/County Management Association.

29.

Netzer

(1992). Differences in reliance on user charges by American state and local governments. Public Finance Review, 20, 499-511.

30.

Niskanen

W. A.

(1971). Bureaucracy and representative government. Chicago, IL: Aldine, Atherton.

31.

Niskanen

W. A.

(1975). Bureaucrats and politicians. Journal of Law and Economics, 18, 617-643.

32.

Oates

W. E.

(1979). Lump-sum intergovernmental grants have price effects. In Mieszkowski

P. M.

Oakland

W. H.

(Eds.), Fiscal federalism and grants-in-aid. Washington, DC: Urban Institute.

33.

Pascal

A. H.

(1984). EBBF: A guide to installing equitable beneficiary-based finance in local government. Santa Monica, CA: RAND.

34.

Puviani

(1903). Teoria della illusione finanziaria. Milano, Italy: Remo Sandron.

35.

Rogers

Silva

R. D.

Bhatia

(2002). Water is an economic good: How to use prices to promote equity, efficiency, and sustainability. Water Policy, 4(1), 1-17.

36.

Romer

Rosenthal

(1982). Median voters or budget maximizers: Evidence from school expenditure referenda. Economic Inquiry, 20, 556-578.

37.

Savas

E. S.

Baumol

Wells

W. A.

(1977). Financing solid waste collection. In Savas

E. S.

Baumol

(Eds.), The organization and efficiency of solid waste collection. Lexington, MA: Lexington Books.

38.

Singhal

(2008). Special interest groups and the allocation of public funds. Journal of Public Economics, 92, 548-564.

39.

Thompson

(2007). The political economy of public administration. In Rabin

Hildreth

W. B.

Miller

(Eds.), Handbook of public administration (3rd ed.). Boca Raton: CRC/Taylor & Francis.

40.

Turnbull

G. K.

Mitias

P. M.

(1999). The median voter model across levels of government. Public Choice, 99, 119.

41.

U.S. Advisory Commission on Intergovernmental Relations. (1987). Local revenue diversification: User charges (No. SR-6). Washington, DC: Author.

42.

U.S. Census Bureau. (1962-2004). Annual survey of local government finances and census of governments.

43.

U.S. Census Bureau. (1977). 1977 Census of governments, volumn 4, number 5, compendium of government finances: 1977: Washington, DC: U.S. Government Printing Office.

44.

U.S. Census Bureau. (2002). 2002 Census of governments, volumn 4, number 5, compendium of government finances: 2002: Washington, DC: U.S. Government Printing Office.

45.

U.S. Government Accountability Office. (2010). Wasterwater infrastructure financing: Stakeholder views on a national infrastructure bank and public-private partnerships. Retrieved from http://www.gao.gov/new.items/d10728.pdf

46.

Wagner

R. E.

(1976). Revenue structure, fiscal illusion, and budgetary choice. Public Choice, 25(1), 45-61.

47.

Winston

(1991). Efficient transportation infrastructure policy. Journal of Economic Perspectives, 5, 113-127.

48.

Zorn

C. K.

(1991). User charges and fees. In Petersen

J. E.

Strachota

D. R.

(Eds.), Local government finance: Concepts and practices (pp. 135-151). Chicago, IL: Government Finance Officers Association.