ESLARP East St. Louis Action Research Project
University of Illinois at Urbana-Champaign


Infrastructure Cost Modeling for East St. Louis

Sutanu Bhattacharyya

Arun Pant

August 1996


In order to sustain a basic level of services in any municipal area, a complex fabric of infrastructure needs to be maintained by the municipality or the city, that serves all the units lying within its jurisdiction. While a part of the cost of the services are borne by the property owners in the form of municipal charges such as those for sewer and water, a significant proportion of the cost is shared by the city, out of money received through taxes. This cost is primarily incurred in the process of maintenance of existing infrastructure lines within the city area. The primary services that constitute the basic level of infrastructure for regular functioning of the city are water, sewer, police, fire and health facilities. While thumb rules exist for estimating the cost of establishing new infrastructure facilities, those for estimating the cost of maintaining existing facilities are not easily available.

The question of cost assessment of maintaining existing infrastructure becomes even more significant in the case of communities such as East St. Louis that are either on the decline or in need of economic support. For these communities, the cost of provision of necessary services may constitute a considerable financial burden, both on the residents as well as the tax base of the city. A common feature of such communities is the increase in the decline of housing stock, leading to growing pockets of vacant land between occupied dwellings or businesses. As a result, the infrastructure lines servicing occupied parcels are underutilized when they pass across these vacant lots. This might lead to either of two possibilities. The first could be that in order to maintain the same level of service provision within the community, the city would now levy increased taxes on the remaining occupied units. The second possibility might be a decline in the level and quality of services provided by the city. Either alternative might lead to a further decrease in occupancy rates within the city.

This research tries to analyze the issue of infrastructure costs for a city, with a view to understanding the overall economic impact it may have on the tax base. At this point there can be two alternative means of assessing the actual and perceived impact. The first would be to obtain data from the city to calculate the increase in both taxes and city expenditures on infrastructure due to the spread of vacant land. The second would be to simulate different scenarios with varying arrangement of the existing parcel layout leading to curtailment of underutilized infrastructure lines. The second alternative would help us know if there is a high positive correlation between infrastructure cost and the total span of the service area and consequently the significance of the potential savings that may be achieved through such rearrangements.

In order to arrive at some standard level of assessment of the cost of city services, attempts have been made here to model the maintenance cost of infrastructure as a function of distance between parcel units and service centers of the city, at various levels of aggregation. This model would help us evaluate the impact of the existing vacant land towards increasing the infrastructure cost for the city, and analyze the importance of this impact. In economic terms we would be able to assess the approximate annual burden to both the city and its residents in terms of increased tax, which would allow us to gauge the financial significance or lack thereof of the problem of vacant land for the city.

Costs of municipal services with relation to parcel location.

Municipalities incur certain costs for providing basic services and utilities to the parcels within the city. Though the cost per unit is usually broken down to some fixed rate for all the parcels, it can be allowed that each parcel incurs a differential cost per unit due to location factors. These factors include the distance variable - i.e. the differential cost due to the distance from service centers and can be expressed as function:

DC1 = f (distance) .............(1)

where the differential cost (DC1 ) is some function of distance.

The above function however takes into account the linear distance variations and gives no indication of the spatial arrangement of a set of parcels being serviced. However, if the network distance is used, the effects of this spatial arrangement would be accounted for in the process. Therefore, the two main assumptions relating costs to spatial location of the parcels are:

  1. The distance from parcels to the service centers is related to the costs of providing the service.

  1. The spatial arrangement of the parcels (different permutations are possible with the same set of linear distances) cause the costs to vary.

This methodology is followed for all the services taken into account. A service specific distance function is to be formulated for each service based on the way that the service is provided. The various services are:

  1. Sewers and drains
  2. Water supply
  3. Electricity
  4. Gas lines
  5. Road construction and maintenance
  6. Police
  7. Fire
  8. Ambulance

It can be seen from the above list that the distance functions will be of two general nature. For items 1 to 5, the parcels are served through linear network distances from various service centers. For items 6 through 8, the radial distances from the service centers would be used to limit the extent of the various service areas. Some ideas about existing costs could be obtained from city records or through contractor estimates especially for costs relating to repair and construction of roads, maintenance of sewer lines and installation of pump stations and water supply mains.

Various approaches in formulating a viable modeling procedure.

A number of different approaches were considered for the purpose of formulating the model due to the lack of any existing standard procedure. These approaches had their own set of advantages and disadvantages. However, the crucial factor in all cases would be the availability of data for empirical analyses. It was useful to consider the different alternatives due to the fact that the exact extent of available data was not known. In addition, the possibility of an improvement of the data base for the area could lead to the use of some of the alternative models in near future. Two of the main approaches considered are given below:

1) Regression approach

If data on the cost incurred by individual parcels for certain services are available, then regression analysis could be used for a sample set of parcels to determine the distance coefficients with the dependent variable being the cost incurred by the parcel. The independent variables would be the distances to the various service centers as well as other variables that characterize the parcels. The regression equation would build upon equation 1 and would be in the following format:

C = f (distance to various service centers, physical and structural parcel attributes)…….. (2)

where C is the cost incurred by the parcel.

The total costs would be the sum of all costs incurred by all the parcels. The advantage of this method would be in the fact that other factors such as parcel size, household incomes, plan designations, etc. could be used as parameters for cost estimation. However, the major drawback is the lack of parcel based cost data. Municipal service costs are measured by the type of service provided in a locality and not by the cost per parcel. This model then can only be used if a methodology is devised to transfer these costs to the parcels.

One way of transferring the costs so as to estimate the coefficients of the independent variables would be to identify the different localities to which a particular service is provided. If the localities are made up of a fairly homogeneous parcel mix (i.e. a residential subdivision) then a sample parcel with the characteristics that is representative of the rest of the parcels would be used as one observation for the regression. It would be necessary to obtain a number of localities that would constitute an adequate sample size. In addition, these localities would have to be representative of the various areas within the city. The size of these localities would depend on the availability of data - the more disaggregated the localities, the greater the precision of which the distance effects are measured. The availability of data from geographical information systems would enable the generation of a number of variables that may affect the cost of services. For example, multi-family zoning would probably mean that there is greater road maintenance activity in the area than in places zoned single family residential. The proposed methodology is summarized below:

It must be noted that the above approach would not be able to differentiate the costs incurred by parcels within a given locality. The size of the locality therefore plays an important role and is determined by the available cost data. In a city wide scenario, this model would be an useful indicator of the service costs and could be used in considering various alternatives for development. However, within a particular locality, the model would not be able to differentiate between the characteristics of the various adjoining parcels.

2) Weighted distance approach

This approach assumes that the variation of the cost between different parcels depends on the distances from the services to the parcels. The distance functions could vary depending upon the type of service. One way of estimating the effect of the distances is by assuming some functional relationship (inverse, inverse square, decay etc.) between the cost and distances. Another way of establishing this relationship is by assigning a set of 'weights' to the distances and using these weighted distances to estimate the cost. An exercise was carried out using a sample parcel arrangement and the location of one particular service center within this arrangement. A number of different weights were applied to the distance measures and used in calculating the cost per distance based on a given total cost figure. These were then used to estimate the change in total cost and these estimates depended on the set of weights utilized. This exercise could be replicated for larger parcel arrangements within the city. One limitation of this method is the current lack of data that would make it difficult to calibrate and check this model at the city wide level.


The usefulness of the model derived from the above methodology would largely depend on the availability of the data. However, with the current increase in the use of digital information systems, it can be expected that it would be less of a problem in future. Therefore, the current emphasis should be in formulating a methodology that would allow the relationship between parcel location and infrastructure costs to be modeled explicitly. This would enable construction of alternatives in determining development and evaluating them by their relative efficiencies regarding the provision of infrastructure.

Document author(s) : Sutanu Bhattacharyya
HTML by : Abhijeet Chavan
Last modified: August 16, 1996


East St. Louis Action Research Project
Refer to the ESLARP mailing address on the ESLARP home page