Environmental Issues > Urban Solutions Main Page > All Urban Solutions Documents

Top of Report

Part 1

Before entering into a detailed discussion of density and its affect on utility service costs, it is useful to discuss the organizational and physical characteristics of water and wastewater utility systems. These two types of systems share some common functions. For instance they must treat water by removing contaminants and they must transmit, distribute or collect water through a system of pipes. And, both systems usually need to dispose of solids, although the volume of solids is ordinarily much higher for wastewater systems. Water systems also engage in activities related to obtaining and/or storing water.

Water distribution systems almost always require active pumping to maintain adequate pressure throughout the system. Wastewater collection systems may also require pumping if slopes are insufficient for gravity systems to maintain flows through the system or if there are geographic obstacles.

For the most part, this study is concerned with what are usually the two largest functions for either a water or wastewater system. These are: 1) treatment (sometimes referred to as "plant" since treatment activities usually occur in a treatment plant); and, 2) conveyance (which includes both water distribution and wastewater collection).


Water and wastewater systems are actually composed of several inter-related sub-systems. These are listed below. The two sub-systems which account for most of the costs related to water and wastewater service are treatment and conveyance. Other sub-systems typically include water supply, sludge disposal for wastewater systems, customer service and billing, and storage. Some systems have all components listed below, while others do not.

Water System Components[2]

Source of Supply (S): This consists of wells, reservoirs, rivers or lakes, and ordinarily includes intakes, pumps, dams, head gates or other similar facilities and equipment.

Source Transmission (S): In the event the source of supply is located at a distance from the treatment plant there will be a system of intakes, pipes, canals or aqueducts and associated pumping facilities to transmit raw water to the treatment plant.

Treatment (T): The treatment plant involves a series of settling chambers, holding tanks, filtration, chlorination or UV equipment which extract impurities from the water, fluoridation equipment, other related equipment.

Transmission and Pumping (C): Major transmission lines (ordinarily those in excess of 6 to 8"), associated pumping stations, force mains, and other equipment serving to transmit treated water and maintain pressure throughout the water service area.

Distribution Storage (C): Storage facilities and towers help to maintain and equalize pressure for certain areas of the system and provide reserves for firefighting.

Distribution Lines (C): Distribution lines branch off the major transmission lines to provide service to subareas.

Branch Lines (C): Branch lines connect to the distribution lines and serve individual streets or areas.

Laterals and Meters (C): Laterals and meters provide the service connection between the customer and the branch or distribution line.

Wastewater System Components[3]

Local Collectors (C): Inclusive of successively larger sewers, including branch sewers, lateral sewers and trunk lines which primarily serve local areas.

Interceptors (C): Larger sewers which receive flows from the local collection system and transmit flows to the treatment plant.

Primary Treatment (T): Consists of settling ponds, screens, pumps, grit chamber and primary tanks.

Secondary Treatment (T): Consists of the secondary treatment plant, final tanks, chlorine contact tank and sewer outfall. Some systems may substitute biological treatment systems (such as constructed wetlands) for secondary facilities or augment them with such systems.

Sludge Handling (T): Consists of sludge thickener, digestors, drying beds, de-watering chambers and residuals disposal.

Tertiary Treatment (T): A third level of treatment facilities that is sometimes required to remove contaminants not removed through primary and secondary treatment. Sometimes spray irrigation facilities or wetlands may be used in lieu of active treatment processes.

Chlorine Contact Tank (T): Facilities for chlorinating treated wastewater before it is returned to the receiving water.

Outfall Line (T): Pipes through which treated wastewater is delivered to the receiving water.

Each of the above components correspond to a functional cost center for which both capital, and operation and maintenance (O&M) costs are incurred. Of these components, land use and service area characteristics have been previously shown to be an influence on capital and operating costs related to the treatment plant and on capital costs of conveyance systems.


The relative importance of plant and conveyance functions for a particular utility system is subject to a number of factors. First, there are positive economies of scale for water and wastewater treatment. A study published in 1969 quantified these economies of scale for three wastewater treatment levels (primary, trickling filter, activated sludge). [4] On a unit basis (per thousand gallons), long-run O&M costs for a 1.0 mgd plant were found to be between 54 to 66 percent less than for a 0.1 mgd[5] plant. And, unit costs for a 10.0 mgd plant were between 37 to 45 percent less than a 1.0 mgd plant.

Similar economies of scale for treatment were found on the capital side as well. Marginal capital costs for a 1.0 mgd wastewater treatment plant were between 41 percent-63 percent less than a 0.1 mgd plant. And costs for a 10.0 mgd plant were between 41 to 64 percent less than for a 1.0 mgd plant. [6]

To the extent previous studies have examined conveyance costs, they have addressed capital costs only, and have relied on indirect methods. For example, a longstanding study examined the affect of the density of development for a hypothetical 2,480 unit residential subdivision. It found the capital costs for sanitary sewers were 62 percent lower in a high density community (16 dwelling units per acre) than a medium-density community (4 dwelling units per acre). [7] No studies were found which examine O&M costs for conveyance, or which relied on actual O&M cost data for either conveyance or treatment.

The extent of a utility system's service area can also affect its cost structure. For example, smaller systems have a larger portion of their capital assets invested in the treatment plant, and ordinarily a greater share of their system's O&M costs are devoted to the plant. Eventually, as a utility's service area increases, the share of capital assets invested in conveyance increases, as does its conveyance O&M outlays. For larger systems, conveyance O&M costs can exceed those for treatment. The CDOW typifies many larger utility systems in this regard. In 1995, for example, it employed 560 persons in its distribution section, while its treatment plant operations section employed just 260 persons. [8]


It is a common misperception that the costs of treating, producing and delivering water and wastewater are dominated by the capital costs of constructing these facilities. While water and wastewater utilities are among the most capital intensive of all local government services, capital costs are ordinarily only a fraction of total annual utility costs. Although outlays for capital improvements can be quite large, their annual budgetary and rate impacts are usually moderated by borrowing.

Two other factors tend to reduce the importance of capital costs relative to O&M costs. Since the actual useful life of many capital improvements exceeds the term over which they are financed, there may be no debt service[9] payments associated with these facilities in later years. Furthermore, the annual debt service associated with a particular capital improvement remains constant over time, while inflation ordinarily causes its associated O&M costs to increase. Added to this, utilities can take advantage of lower market interest rates to periodically refinance outstanding debt, thereby reducing debt service costs.

An examination of fifteen water and/or wastewater systems illustrates the relative importance of O&M and capital costs. Located in four states, these systems provide utility service to communities with populations as small as 8,000 to more than 1 million. Of the fifteen systems represented in Table 1, debt service as a percent of total operating costs averages 24 percent - ranging from a low of 5.3 percent to a high of 61.8 percent. Average annual O&M costs are about three times greater than annualized capital costs.

Only in Avon, Ohio are annual debt costs higher than operating costs, and this is due primarily to recent major capital improvements to provide additional system capacity for new and future development. The CDOW has the next highest portion of debt service as a share of total operating costs. Its relatively high debt service expense is attributable in large part to an ongoing $250 million capital improvement initiative aimed at providing additional treatment capacity for its western suburbs and replacement of some of the older parts of its distribution system some of which dates back to the earlier part of the 1900's.

Table 1: Water and Wastewater Operating Expenses and Debt Service Costs
Jurisdiction, State, Fiscal Year Total Expenses
($ millions)
Debt Service[10]
($ millions)
Debt Service as % of Total
Naperville, IL, FY 96 $19.260 $5.074 26.3%
Sycamore, IL, FY 96 0.862 0.046 5.3%
Woodstock, IL, FY 96 4.515 0.45 10.0%
Plainfield, IL, FY 96 1.167 0.29 24.9%
Oak Park, IL, FY 91 4.322 0.561 13.0%
Rocky River, OH, FY 96 0.822 0.253 30.8%
Avon, OH, FY 95 1.593 0.984 61.8%
Westlake, OH, FY 96 1.129 0.261 23.1%
Cleveland Division of Water, OH, FY 95 96.719 41.821 43.2%
North Olmstead, OH, FY 95 3.469 0.439 12.7%
City of Bay Village, OH, FY 95 1.03 0.229 22.2%
Fairview Park, OH, FY 95 1.079 0.297 27.5%
Newport News Waterworks, VA, FY 96 47.596 11.947 25.1%
Prince William County Service Authority, VA, FY 93 29.354 7.662 26.1%
King George County Service Authority, VA, FY 96 .729 .135 18.5%
Hendersonville, NC, FY 95 11.457 1.463 12.8%


Surprisingly, the greater role of O&M costs in determining the impact of various land uses on utility service costs has been given little attention. Indeed, most local programs and policies designed to address the fiscal impact of local land use and development policies concern themselves almost exclusively with capital costs and facilities. [11]


The influence of land use patterns on conveyance costs, and their effects on equity and efficiency has been recognized previously. Indeed, land use, spatial and geographic characteristics are a primary design consideration for utility conveyance systems, so it is not surprising that they are a major determinant of costs. A Water Pollution Control Federation publicationnotes the importance of spatial features in influencing costs of wastewater services. It identifies total land area and density as being important factors which influence service costs. According to the WPCF, total land area is "a major determinant of the level of capital costs experienced by a wastewater utility and to a lesser extent operating cost." [12]

Likewise, with regard to density, the WPCF states: "When customers are located relatively close to one another...the sewers necessary to serve them also are located closer together, serving more customers per foot of pipe. The density and extent of this grid has a major impact on the level of collection system operating and capital costs." [13]

In addition to being affected by density, WPCF also notes that fixed collection-related costs[14] are also related to length of pipe. Length of pipe is a function not only of area served, but also distance from the treatment plant, interceptor or major transmission line. And, although not mentioned in the literature, total length of pipe can also affect treatment costs since it presents the opportunity for more loss or leakage of potable water, or inflow and infiltration of groundwater into wastewater lines.

Despite these claims, the relationship between spatial features and operating costs has not been previously quantified using empirical data. Both the American Water Works Association (AWWA) and the National Regulatory Research Institute (NRRI) acknowledge the need to determine the influence of size and density on capital and operating costs. [15] If spatial costs differences are significant, they can cause fiscal inequities within systems with widely differing land use patterns. The widespread practice of charging uniform rates, or rates differentiated only by customer class, could cause customers located in higher cost areas to be subsidized by those in lower cost areas.

Indeed, the NRRI notes that: "Uniform rates over geographic space involve cross-subsidization. The rate averaging results in prices exceeding costs for some users and failing to meet costs for others." [16] NRRI also recognizes that spatial pricing can help to reduce or eliminate subsidies where significant geographic cost differentials exist, and recognizes that spatial pricing is most "...appropriate for utilities with core and satellite areas." [17]

Further, while the AWWA manual on water rates does not specifically mention spatial pricing, it also recognizes that rate differentials are justified on the basis of "...special considerations...such as facilities required, extent and nature of service and other special items."

While some utility systems have instituted geographically sensitive development charges[18] to recover capital related costs, only a handful of utility systems in the United States have adopted user rates which reflect spatially sensitive O&M costs attributable to the conveyance system. The Cleveland Division of Water (CDOW) is one of the few systems in the U.S. which recognize geographic-based cost differentials in water user rates. Its pressure zone rate structure causes customers located in its higher elevation areas to pay for the excess pumping and pressure-related costs for these areas.

Customers located in CDOW's three higher pressure zones pay rates that are 1.7, 2.0, and 2.3 times greater than those in its lowest pressure zone. Costs to serve CDOW's higher pressure zones are largely attributable to their greater water pumping and pressurization requirements. While the scale of these extra costs may be greater for CDOW than for many other water systems, the need to incur extra pumping and pressurization costs generally becomes greater as service areas increase in size, and/or as distance from the treatment plant increases.

Given that spatial and land use characteristics can and do significantly affect conveyance costs, why aren't these cost differences more widely reflected in user rates? In general, there are two obstacles towards broader adoption of geographic or spatially sensitive utility rates. These are a lack of functional accounting data and a lack of information on how different land uses affect conveyance costs at the community, subdivision, or sub-area level.

As noted by the WPCF, conveyance costs are most sensitive to spatial variables. To enact geographically sensitive rates, a utility service provider must know what its conveyance costs are (as distinct from other system costs), and how much these costs are affected by different land use patterns and spatial characteristics within its system.

Identification of functional costs can be accomplished through methods such as those recommended by the National Association of Rate Utility Consultants (NARUC) and the AWWA. Appendix A contains an example of the NARUC recommended uniform chart of accounts for a water utility.

Many municipally owned[19] systems have yet to adopt charts of accounts that reflect functional costs. Rather, they rely on aggregated or line item cost-accounting. For systems providing both treatment and conveyance functions, these methods do not adequately differentiate costs for these separate functions. When functional accounting data are lacking, the extent of any excess costs attributable to spatial or land use characteristics is difficult to determine. This explains why so few utility systems have adopted geographically sensitive user rates.

Functional accounting data, when available, must also be related to those portions of a utility's service area with significantly different land uses. For systems such as CDOW, this would not be a difficult task as its capital and O&M costs related to its pumping and pressure-related facilities can be specifically linked with its pressure zones.

Making this link may not be as easy for other systems. Other than debt service, most conveyance costs are for labor, equipment, materials and supplies, electrical costs, engineering and overhead. Even when functional accounting data are available, it may still be difficult to attribute these costs to specific areas. The obstacles involved in attributing costs to specific geographic areas or land uses are prohibitive for many utilities, particularly smaller and medium-size systems.

This study provides an alternative method for determining the relationship between wastewater conveyance costs and land uses patterns, determining this relationship "cross-sectionally." This is accomplished by examining a number of conveyance systems exhibiting a range of land use and density characteristics. Conveyance costs for each system are representative of that system's development and service patterns. This approach avoids having to determine and allocate geographically specific conveyance costs within a system. Further, since the cross-sectional approach relies on multiple systems, the results can be more confidently applied to other systems.


To the extent their service costs also vary, different types of land use can raise issues of equity and efficiency. Equity issues are raised when one type of land use or geographic area within a community or system is subsidized by another. This situation is most likely to occur when these areas are served by a single utility with a rate structure that relies on average conveyance costs. For such systems, customers located in those portions of the utility's service area with lower unit service costs will likely subsidize those in the higher cost areas.

Another issue is efficiency. Land use patterns can affect the total public and private costs for utility services. If a community develops with a less costly development pattern from the standpoint of utility service costs, then savings can be achieved system-wide. Such savings, if realized, could be used to increase disposable income, or they could be applied to other public services.


Ten wastewater conveyance systems were studied to determine the relationship between land use characteristics and unit conveyance costs. [20] Each system is a conveyance-only system, or one whose accounting practices isolate conveyance from other system costs. The service area for each conveyance system is co-terminus with the municipal boundaries in which it is located. [21] Environmental and geographic factors which can influence each system's costs are substantially similar for the Cleveland and Chicago areas. For all systems, topography is essentially flat to gently sloping, meaning that elevation-related pumping costs are minimized (see Appendix C).

Conveyance costs for each system were obtained. Total conveyance costs were divided by annual water usage (thousands of gallons) within each system to obtain each system's unit conveyance costs. Unit costs were related to factors previously cited as exerting a potential influence upon them such as the total number of service units, miles of pipe, and square miles of the service area. Linear regression techniques were applied to determine the statistical relationship between these variables and unit conveyance costs.

Each community[22] and its corresponding conveyance system contains primarily residential land uses with associated retail, and other commercial and institutional development. Median household income in these communities ranges between about $29,000 to $60,000. Population densities range between 190 to 11,400 persons per square mile. Although some of the systems included in this study provide service to manufacturing industries, none are known to provide service to a high volume water- or wastewater-intensive industry.

With regard to population, four communities have experienced declines of between 7.5 to 19 percent since 1970. One community had essentially no change in population. Two communities experienced population growth rates that have at least doubled their 1970 populations, while three others experienced modest gains in population. All of the communities in this study have experienced gains in employment since 1970. In terms of their demographic and employment characteristics, these communities are representative of many developing and developed suburban areas throughout the U.S.

Seven of the communities included in this study are responsible for wastewater conveyance only - that is they do not own or operate a wastewater treatment facility. Rather they contract with other nearby service providers for treatment. Three communities (Lakewood and North Olmsted, Ohio and Naperville, Illinois) operate their own treatment facilities. These three communities report treatment and collection costs separately. Four communities (Fairview Park, Rocky River, Bay Village and Westlake, Ohio) operate a joint venture treatment facility. [23] These four communities are examined both individually and collectively, yielding a total of nine individual communities and one additional "community" consisting of four aggregated joint venture communities. [24]

Of the two Chicago-area communities, Oak Park, located in Cook County, is an older inner-suburb which is substantially built-out. Naperville, in DuPage County, has come within commuting range of major Chicago-area employment centers over the last decade, and has itself become a major employment center. This has caused it to be one of the most rapidly growing communities in the U.S. over the last decade.

The eight Cleveland area systems are located in the western-Cleveland suburbs. Seven are in Cuyahoga County and one is in Lorrain County. These communities include three of Cleveland's inner suburbs (Lakewood, Rocky River and Bay Village), which were substantially built-out prior to the 1970's. They also include newly developing areas such as Westlake, which has experienced substantial new development in more recent times, as well as Avon, which is just beginning to experience significant new development.

Although this study focuses on land use patterns, other factors not previously mentioned can also influence unit conveyance O&M costs. These include: overcapacity, maintenance practices, and the physical condition and age of the conveyance system. Overcapacity can be a result of either loss of population and/or employment, or expansion to accommodate expected growth. Lakewood, Fairview Park, Bay Village, Rocky River, and North Olmsted each have some overcapacity caused by loss of population. Naperville and Avon have substantial excess capacity caused by recent expansions designed to accommodate expected new development. Unit conveyance costs (dollars per thousand gallons) in communities with substantial overcapacity are somewhat greater than they would be if they were operating at full capacity.

A poorly maintained system will develop breaks and leaks that require substantial labor and materials to repair. As systems age, the need to make unscheduled repairs is exacerbated. Systems with a regular replacement schedule will ordinarily exhibit less need for unscheduled repairs. Historic maintenance practices are also likely to have a significant influence on current O&M conveyance costs. Unfortunately, the necessary records to compare historic maintenance practices and to determine the average age of each system's pipe was not available.

Although exact records are unavailable, the relative age can be approximated from site visits and anecdotal reports from local officials. Wastewater systems serving Oak Park and Lakewood were developed in the earlier part of this century. These two communities experienced their peak in jobs and population between 1960 and 1970. The Lakewood system may have experienced significant deferred maintenance during this time causing current O&M expenses to be greater. A recent scheduled replacement program has been initiated to address prior deferred maintenance. By comparison, Oak Park's system also dates back to the early 1900's. It is the most affluent community in this study, and has been more economically stable over the years than Lakewood. Its relative stability and affluence have enabled it to regularly maintain its system. Naperville and Westlake have experienced substantial recent growth and new development, and likely have the newest collection systems.

The age of the system can also influence unit costs. Like a new building, newly installed conveyance systems tend to have reduced maintenance requirements for the first few years. This may account for Naperville and Westlake having lower unit costs than might otherwise be expected. If so, the results obtained in this study would tend to understate the influence of density variables on unit conveyance costs.


Conveyance costs are examined in this study on a unit basis. In technical terms, unit conveyance costs are the "dependent variable." They depend on, or are influenced by, other "independent" variables such as density, age, and number of service units. Total conveyance costs are divided by average annual water usage within the system to determine the unit cost per thousand gallons. Annual water usage is used because it serves as the basis for wastewater billing, and because wastewater flows attributable to each community were not consistently available.

There are a number of factors which have been mentioned as having the potential to affect unit conveyance costs. These include:

  • Miles of pipe

  • Size of service area

  • Number of service units

  • Volume

  • Age of system

  • Historic maintenance policies

The last two factors are not quantifiable with available data. In an attempt to test for the influence of maintenance policies, median household income was used as a default for historic maintenance policies, under the assumption that higher income communities would be less likely to defer scheduled maintenance. This leaves age of the system as the only variable noted above which could not be explicitly tested.

Some of these variables were further transformed to relate them to each other. For example, the number of service units is divided by the miles of pipe to obtain service units per mile of pipe which is a measure of service density. Similarly, miles of pipe is divided by square miles of the system's service area to obtain a measure of the density of conveyance grid. Single variate regressions were done to determine the influence of each of the following independent variables on unit service costs:

  • Service Units Per Mile of Pipe (SUPMP, used as a measure of service density)

  • Median Household Income (MHI, used as a proxy for historic maintenance policies)

  • Service Units (SU, used as a measure of economies of scale)

  • Volume (VOL, used as a measure of economies of scale)

  • Pipe Miles per Square Mile (PMPSM, used as a measure of grid density)

  • Service Units per Square Mile (SUPSM, used as a measure of area density)

Based upon the results of the single-variate regressions, multi-variate regressions were done on combinations of independent variables. These combinations included: [25]


  • SUPMP and SU

  • SUPMP and VOL




Prior to performing the regression analysis, some adjustments in budget data were made to compensate for budgeting practices and other locally specific conditions described in Appendix D.[26]

Regressions were run on this adjusted data. Single and multi-variate regression results are discussed below. The "adjusted R-square" value is presented for each of the variables analyzed. This value equates to the percent of the relationship between the independent variable(s) and the dependent variable.27

Single-Variate Results

Service Units Per Mile of Pipe (SUPMP), adjusted R-square = .321

SUPMP is a measure of the service density of the conveyance system. It is the number of miles of pipe divided by the number of service units. SUPMP alone was found to have a significant influence on conveyance O&M costs for the ten systems examined. Its adjusted R-square value of .321 means that it accounts for about 32 percent of the variance in costs between the ten systems examined. The relationship is very strong, as measured by a T value of -2.293, and the confidence level of this relationship is high at about the 95 percent level. The negative T-value indicates that, as service density (measured as SUPMP) decreases, costs increase. A low Durbin-Watson score of 1.355 for this variable suggests some co-linnearity with unit costs.

Median Household Income (MHI), adjusted R-square = -.081 (not meaningful)

MHI is used as a proxy for maintenance practices under the hypothesis that higher income communities would be less likely to defer system maintenance. MHI alone was found to have no influence on conveyance O&M costs for the ten systems examined. This could mean either that current MHI is not a valid proxy for historic maintenance, or that if valid, historic maintenance has no influence on the ten system's O&M costs. The only way to further test if MHI is a valid proxy would be to apply a time series analysis which was not attempted in this study. Accordingly, this study cannot explicitly determine the influence of historic maintenance policies on conveyance O&M costs.

Pipe Miles per Square Mile (PMPSM), adjusted R-square = -.034 (not meaningful)

PMPSM is a measure of grid density, or the amount of pipe relative to the utility service area. It is the miles of pipe divided by the number of square miles within the municipal boundaries. PMPSM alone was found to have no significant influence on conveyance O&M costs for the systems in this study.

Service Units per Square Mile (SUPSM), adjusted R-square = -.039 (not meaningful)

SUPSM is a measure of the density of the service area. It is the square miles of the service area divided by the number of service units. SUPSM alone was found to have no influence on observed unit conveyance O&M costs.

Service Units (SU), adjusted R-square = .183

SU serves to measure for the presence of economies of scale on conveyance costs. SU is the total number of jobs and population served by the system. SU alone was found to explain about 18 percent of the variation in observed unit conveyance O&M costs. The strength of this relationship is moderate as measured by a T-level of -1.738, and the confidence level of this relationship is also moderate at the 88 percent level. This suggests that as a single variable, economies of scale may exert some influence on observed unit costs.

Volume (VOL), adjusted R-square = .226

VOL serves as another measure for the presence of economies of scale. VOL is the total volume of water used by wastewater customers. VOL alone was found to explain about 23 percent of the variation in observed unit conveyance O&M costs. The strength of this relationship is acceptable as measured by a T-level of -1.906, and the confidence level of this relationship is also acceptable at the 91 percent level. This supports a finding that economies of scale are likely to exert some influence on observed unit costs.

Multi-Variate Results

Based upon the above results, multi-variate regressions for selected combinations of variables were run to determine if the inter-relationship of two or more variables provided additional explanatory value for variations in observed unit conveyance costs. The results are presented below:

Service Units Per Mile of Pipe (SUPMP) and Pipe Miles per Square Mile (PMPSM); adjusted R-Square = .538

Together, SUPMP and PMPSM explain about 54 percent of the variation in observed unit costs. SUPMP has an exceptionally high T-value of -3.305, significantly stronger than its value as a single variable. Again, the negative T-value indicates that as service density (SUPMP) decreases, unit costs increase. PMPSM significantly improves the relationship between unit costs and SUPMP. The confidence that SUPMP is related to unit costs is high at the 99 percent level. The strength of the relationship between PMPSM and unit costs is acceptable with a T-value of 2.183. [28] The confidence in this relationship is good at the 93 percent level. PMPSM serves to strengthen the explanatory value of the SUPSM variable. Together, they offer the highest explanatory variable of any single or multiple variables examined. A good Durbin-Watson test score of 2.226 indicates co-linnearity is not a significant factor in the strong relationship among these variables.

Service Units per Mile of Pipe (SUPMP) and Service Units (SU); adjusted R-Square = .318

The adjusted R-square value for this combination of variables would suggest that they explain about 32 percent of the variation in unit costs, with SUPMP being the more dominant of the two variables. However, this relationship does not improve on SUPMP as a single variable, and is not reliable. The T-value for SU of only -.981 indicates a weak relationship, and the confidence in this relationship is also weak at only the 64 percent level. In this combination, SU actually degrades the T-value and confidence of the SUPMP variable, decreasing them to -1.606, and 85 percent, respectively. These results would rule out economies of scale as having a significant influence on observed unit conveyance O&M costs in combination with SUPMP.

Service Units per Mile of Pipe (SUPMP) and Volume (VOL); adjusted R-Square = .313

The adjusted R-square value for this combination of variables would suggest that they explain about 31 percent of the variation in unit costs, with SUPMP being the more dominant of the two variables. However, this relationship does not improve on SUPMP as a single variable, and is not reliable. The T-value for VOL of only -.951 indicates a weak relationship, and the confidence in this relationship is also weak at only the 62 percent level. Similar to SU, VOL actually degrades the T-value and confidence of the SUPMP variable, decreasing them to -1.420, and 80 percent, respectively. These results support ruling out economies of scale as having a significant influence on observed unit conveyance O&M costs in combination with SUPMP.

Service Units per Mile of Pipe (SUPMP) and Service Units per Square Mile (SUPSM); adjusted R-Square = .493

Together, SUPMP and SUPSM explain about 49 percent of the variation in observed unit costs. The relationship for SUPMP is very strong with a T-level of -3.070. The confidence in this relationship is high at the 98 percent level. The relationship between SUPSM and unit costs is acceptable as measured by a T-level of 1.930. The confidence in this relationship is also acceptable the 91 percent level. SUPSM serves to strengthen the explanatory value of the SUPMP variable, and together, they offer the second highest explanatory variable of any single or multiple variables examined. A Durbin-Watson test score of 1.855 indicates limited co-linnearity between these variables.

Conclusions of Regression Analyses

Among the single variables examined, the number of service units per mile of pipe (SUPMP) explains more of the difference in observed unit conveyance O&M costs. SUPMP is an expression of service density of the conveyance system. As a single variable, it explains about 32 percent of the variation in unit costs. As a single variable, volume (VOL), which is an expression of economies of scale, explains about 23 percent of the variation in observed unit conveyance costs ranking second to SUPMP.

The most powerful explanation for the variation in observed unit costs, however, is the pairing of SUPMP with PMPSM, which is a measure of the density of the conveyance grid. Together, these two variables explain about 54 percent of the variation in observed unit costs. The second-most powerful combination is SUPMP and SUPSM, which is a measure of the density of service population and employment. These two variables explain about 50 percent of the observed variation in unit conveyance costs. These results affirm that density expressions dominate in explaining the variation in observed unit conveyance O&M costs. The explanation for the balance not explained by density variables is likely to be a combination of factors, including age of pipe, material of pipe, and historic maintenance practices.

Observed unit conveyance O&M costs and service units per mile of pipe are presented in Figure 1 ranked according to the number of service units per mile. Per unit costs associated with the lowest service densities are significantly higher - in some cases more than twice as high - as those associated with the highest service densities. The relationship between unit costs and SUPMP does not fully correspond to the variation in observed unit costs. Given that as a single variable it explains about 32 percent of the variation, one would not expect it to. In general though, the cost per unit increases as the service density per mile of pipe decreases. For example, Oak Park has one of the lowest per unit costs and the highest density. Conversely, North Olmstead has the lowest service density and one of the highest unit costs.

In combination with previous studies, these findings indicate that for water and wastewater systems, the least costly land use patterns are those that optimize each of the following characteristics:

  • The treatment plant serves a large number of residences, businesses and other users, thereby enhancing economies of scale.

  • New development utilizes existing treatment facilities, thereby both benefiting from existing economies of scale and enhancing them.

  • The distance over which lines must run from developed areas to the treatment plant is minimized.

  • Developed areas are designed to maximize the number of households and jobs served per mile of pipe.


Although the regression analysis in this study was for wastewater conveyance, its findings are generally applicable to water conveyance. Others have noted the influence of spatial characteristics on water conveyance costs. [29] That they would be similarly influenced by density and land use characteristics is not surprising since water and wastewater conveyance systems share much in common. Their service areas often overlap or are identical. Both are fluid systems and are designed to reflect geographic, physical and spatial features. Probably the most significant difference is that water conveyance requires pressurization throughout the system, as well as storage facilities. Thus, water systems ordinarily require greater outlays for pumping, whereas wastewater conveyance relies in whole, or in part, on gravity to maintain flows through the system.

More intensive pumping and pressurization requirements would indicate that service and grid density would be at least as important an influence on conveyance costs for water as they are for wastewater, if not more so. Indeed, it is the need for extra pumping, pressurization and related facilities for the more distant and higher elevation portions of its service area which is the basis for CDOW's geographically-based water rates as discussed in Part 2.

Back to contents page | Next page


2. Note: S=Source, T=Treatment, C=Conveyance

3. Note: T=Treatment, C=Conveyance

4. Downing, Paul, The Economics of Urban Sewage Disposal, Prager Special Studies in U.S. Economic and Social Development, 1969.

5. Millions of gallons per day

6. Although this study did not address the costs of collection and transmission, it noted that the length of sewer pipe is a function of area, while the size of the sewer pipe is a function of density.

7. Isard, W. and Coughlin, R., Municipal Costs and Revenues Resulting from Community Growth, Chandler-Davis Publishing Company, 1957.

8. Some CDOW treatment employees are engaged in operation and maintenance of in-plant pumping facilities that are conveyance-related. Employment is a generally reliable indicator of relative costs since wages and benefits constitutes a large percentage of total O&M costs.

9. Debt service, inclusive of principal and interest, is the annualized payment on funds borrowed to construct capital facilities.

10. For some systems depreciation is used where annual debt service is unavailable.

11. Examples of such impact mitigation policies include system development charges, impact fees, and adequate public facilities ordinances.

12. ______, Financing and Charges for Wastewater Systems, Water Pollution Control Federation, in cooperation with the American Public Works Association, and the American Society of Civil Engineers, 1984, p. 44.

13. Ibid, p. 44.

14. Ibid, p. 92.

15. Beecher and Mann, Cost Allocation and Rate Design for Water Utilities, National Regulatory Research Institute and the American Water Works Association Research Foundation, March, 1991, p. 27.

16. Ibid, p. 126.

17. Ibid, p. 125.

18. A development charge is a one-time fee charged to a developer or property owner to recover the capital cost of major system facilities.

19. "Municipally owned" generally refers to any utility system chartered, owned or sponsored by a governmental entity.

20. Nine individual systems were studied, i.e., those for Lakewood, Rocky River, Bay Village, Avon, Westlake, Fairview Park and North Olmstead in Ohio, and Oak Park and Naperville in Illinois. In addition, one combined system consisting of the systems of four individual joint venture communities was included.

21. Naperville and North Olmsted contractually treat wastewater collected in other neighboring communities. These flows pass through portions of their conveyance system. These flows are a small portion of their own wastewater flows, and are not likely to significantly affect their total conveyance costs.

22. The combined joint venture communities are treated herein as a separate community.

23. This plant is a separate legal, budgetary and financial entity from the individual communities and their collection systems.

24. Throughout this study, when ten communities or systems are referred to, this includes each of the individual nine communities in addition to the four aggregated communities treated as a single system.

25. Other multi-variate combinations of single variables were also run which did not produce significant results. The combinations discussed are representative of the range of results obtained.

26. The major adjustments included removal of extraordinary one-time expenses and capital outlays found in the budget data for some systems. Lakewood was found to report its data in a fashion that more fully captured indirect costs, requiring removal of these amounts from its budget data. Naperville recently added significant new miles of pipe to its system, however its budget data did not reflect a full year's operating costs for the expanded system. Data for Oak Park was obtained from its 1995 budget and required adjustment to reflect one year's inflation to put it on par with 1996 data used for all other systems in the study. These and other adjustments are more fully described in Appendix D.

27. An adjusted R-square value of .603, for example would mean that the independent variable(s) explain 60.3 percent of the relationship between it/them and the dependent variable, which in this case is unit conveyance costs. Two other values also help to interpret the R-square value as indication of the strength and direction of the relationship between the independent variable(s) and dependent variable. The larger the T-value, either positive or negative, the stronger the relationship. The significant T-value is an indicator of the reliability or "confidence" of the relationship on a scale of 0-1. When read as a reciprocal (1 less the significant T amount), it indicates the likelihood of the relationship to be accurate. Finally, a "Durbin-Watson" value is a test for co-linnearity between variables, which, if significant, would mean the relationship between two or more variables is due to the fact that they measure or are derived from the same thing. Significant co-linnearity means there is not an independent relationship between two or more variables, and that they cannot reliably be correlated.

28. Within this multiple relationship, PMPSM is positively correlated with unit costs, however it adds additional explanatory value, and strengthens the SUPSM relationship. This does not mean, however, that PMPSM is positively correlated with unit costs, since as a single variable it showed no significant relationship. Rather, it means that in context with SUPSM, it provides a moderating influence.

29. op.cit., NRRI, AWWA.

Get Updates and Alerts

See the latest issue >

Kaid Benfield's Blog

Kaid Benfield writes about development, community and the environment on Switchboard.

Kaid's Recent Posts

New guide from ASLA highlights urban green, literally, in Portland
posted by Kaid Benfield, 10/27/14
Compact and connected communities improve public health, says new research
posted by Kaid Benfield, 10/20/14
For smart growth, not all urban density is created equal
posted by Kaid Benfield, 10/6/14

NRDC Gets Top Ratings from the Charity Watchdogs

Charity Navigator awards NRDC its 4-star top rating.
Worth magazine named NRDC one of America's 100 best charities.
NRDC meets the highest standards of the Wise Giving Alliance of the Better Business Bureau.

Donate now >

Share | |