Understanding the costs of urban sanitation: towards a standard costing model


 There is a dearth of reliable cost data for urban sanitation. In the absence of high-quality global data, the full cost of sustainable implementation of urban sanitation remains uncertain. This paper proposes an approach for developing bespoke parametric cost estimation models for easy and reliable estimation of the costs of alternative sanitation technologies in a range of geographical contexts. A key requirement for the development of these models is the establishment of a large database of empirical information on the current costs of sanitation systems. Such a database does not currently exist. Two foundational tools are proposed. Firstly, a standard metric for reporting the costs of urban sanitation systems, total annualised cost per household. Secondly, a standardised approach to the collection of empirical cost data, the Novel Ball-Park Reporting Approach (NBPRA). Data from the NBPRA are presented for 87 individual sanitation components from 25 cities in 10 countries. Broad cost ranges for different archetypal systems have been estimated; these currently have high levels of uncertainty. Further work is proposed to collect additional data, build up the global database, and develop parametric cost estimation models with higher reliability.


INTRODUCTION
More than 1 million people per day must gain and maintain access to safely managed sanitation to meet the sanitation target of the sustainable development goals (Mara & Evans ). A significant majority of this 'new' sanitation investment will be made in urban or urbanising areas.
While standalone household and community sanitation services will continue to be important in rural areas (in both rich and poor countries) in denser urban areas, toilets that This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited provide sanitation at home must be planned and operated as part of a professionalised servicewhich transports excreta either through pipes or road-based networks for management away from the home.
A key challenge for SDG 6.2 arises from the dearth of reliable and comparable benchmark estimates of the unit costs of these 'networked' services (Daudey ). One approach to estimating the cost of services is to use the price as a proxy. Unfortunately, price is a poor proxy for the cost of a sustainable sanitation service for two reasons.
Firstly, many sanitation systems are incomplete or badly operated often due to financial, institutional, and societal failures (Evans ; Ika et al. ; International Water Association ). Secondly, since sanitation is a public good willingness to pay is often lower than the economic value of full-service delivery (Ruiters & Matji )

THE STANDARD COST METRIC FOR URBAN SANITATION; TACH AND TACC
Responding to the call for greater consistency and for generalisable information about the cost of sanitation, we propose a plausible metric which is comparable across geographies and technologies and also understandable both to professionals in the WASH sector and to municipal managers who make decisions regarding sanitation investments. A useful corollary is the concept of costs per kWh of energy which is comprehensible both to governments, who may have to finance the initial capital investment, and households, who ultimately pay the bills.
For urban sanitation, we propose two cost indicators to express the cost of any sanitation system: TACH and total annualised cost per capita (TACC). TACH/TACC takes into account full lifecycle costs which are annualised and expressed on a per-household or per-user basis.
The key question in any infrastructure investment is the total level of financial liability that the planned investment is likely to generate compared to total sources of income To enable comparison across various scales of systems, the total number of people directly using the service is used as the denominator. The choice of household or capita for the denominator is challengingeach has merit.
Per capita costs are perhaps most useful when looking at national level macro-economic performance, or when comparing expenditure on sanitation against that on say the police force, but some local governments rely on property tax for their revenue, and water and sewerage utilities bill on a household basis. For many decision-makers therefore using the household as a denominator makes the most sense.

A PROPOSED APPROACH TO COST ESTIMATION
In selecting an approach to use in estimating the TACH or TACC for any given system in a given place, there is a trade-off between precision, reliability, and level of detail on the one hand, and availability of data and resources on the other. The focus here is on supporting local decisionmakers to make credible plans quickly and cheaply, selecting the best technical options in a given context (Mitlin ). This requires costing data that are reliable, comparable, and easy to make locally relevant (Kalbermatten et al. ). We rejected analytic estimation on the basis that it has very high information requirement, and consequently high cost, yet remains prone to bias (Flyvbjerg Having identified a set of candidate parameters for cost estimation purposes, the next task is to assemble a reliable database which could be used to develop the heuristics needed for the estimation approach to work. As already mentioned, data are scarce. Consequently, an important element of our work is the collection of new data on which to base our estimations.
Some specific challenges arise, however, in the collection and organisation of urban sanitation cost data. Urban Cost data reported by operators are often therefore only a partial representation of the real costs of delivery of safely managed sanitation. To address these challenges, we set out to build a database of cost information that could be used to estimate the full costs of service delivery for the entire value chain. To achieve this, we developed an NBPRA for sanitation.

NBPRA FOR SANITATION
The NBPRA facilitates the collection and collation of data in a consistent format from operators who are delivering ongoing sanitation services. Cases are selected from a broad range of contexts (covering a wide range of parameters) and a range of different sanitation technologies and approaches.
The approach is standardised and based on four main pillars, namely (1) technological homogeneity, (2) acceptable service, (3) basic costing assumption, and (4) the reference business model. Cost data are then normalised to comparable currency equivalent values, annualised, and divided by the household/ people served, so they can be reported as TACH/TACC.

Technological homogeneity
The NBPRA is exclusively concerned with a subset of 'safely Combining these options leaves a set of eight component types for emptying and transport in wastewater based systems (see Figure 1).
For systems which rely on pits or tanks (often referred to as 'onsite' or FSM systems) and container-based systems, we differentiate four main types of component at the containment (household) end of the sanitation value chain (sealed tank with an infiltration structure, sealed tank without an infiltration structure, infiltrating pit, and container for container-based systems). Moving excreta from these systems can either take place in two steps (emptying plus transport) or in a single operation (typically where a single truck is used to empty the tank and transport the contents away to treatment). In either case, we make a distinction between manual powered systems and externally powered systems These aspects of the sanitation system performance are important. However, rather than internalising them resulting in greater complexity and less transparency, the NBPRA explicitly excludes them. The aim is to generate reliable comprehensive cost estimates for systems that deliver 'safely managed sanitation' to which decision-makers may choose to add other performance criteria when making real-world technology selections.

Basic costing assumption
The NBPRA is a full-costing approach (Arnaboldi et al. The costs associated with any component are categorised using a standard proforma to facilitate both data collection (acting as a checklist for enumerators), data verification, and modelling of incomplete data (for example, where an operator cannot or will not report the cost of land, or where taxes and fees are not paid). Cost categories used within the NBPRA are shown in Table 1.  Annualisation is applied in line with Stewart et al.  Cost (EAC) of owning, operating, and maintaining the sanitation system (or sub-system) over its entire lifecycle. In other words, it is the annual expenditure needed to cover the servicing of capital debt required to construct and maintain the system plus the annual operational budget.

(). The cost is expressed as the Equivalent Annual
The EAC is calculated, as shown in Equation (1). where: • COST t ¼ are the costs incurred during the lifecycle (i.e. T ) associated with the data point considered.
• T ¼ represents the longest lifecycle associated with the data point considered, and it is calculated in years. T corresponds to the longest lifetime of the infrastructures associated with the data point; • kr ¼ real interest rate, which is calculated using the following formula (Equation (2)) where • kn ¼ Nominal interest rate, assumed 5% as the social discount rate.
• s ¼ annual inflation in the country considered.

RESULTS
The NBPRA has been tested in 25 cities in 10 countries, collecting data on 87 individual sanitation components. A summary of the data collected and processed to date is shown in Table 2.
What the preliminary results mean  Table 3. Median data are preferred to means due to the heavy skewing that can result from a single outlying data point. Further break downs of the data are possiblefor example, in Table 3, we also show selected data for onsite systems in Africa. Estimates of TACH for complete systems have high levels of uncertainty at this stage in the process due to the small sample sizes and clustering of case studies.
The data come from 25 cities, representing only a small sample of the conditions under which urban sanitation systems are implemented.
As might be expected, TACH for the sewerage system shown in Table 3 is significantly higher than TACH for container-based or onsite sanitation systems. However, in Figure 2, we present the data sorted by country, and showing both total costs and CAPEX/OPEX cost breakdown, all annualised on a per household basis. While sewered sanitation systems are often said to be 'more expensive' than FSM-based systems, a closer inspection of the data suggest that the situation may be more complex. For example, the operational liabilities of sewers may sometimes be lower than those for road-based transportation of faecal sludge under some conditions (see Figure 2(b)). Returning to Table 2, it is also possible to see that TACH for containerbased systems is dominated by operational costs whereas onsite systems and sewers have a much higher CAPEX dependency. In all these cases, care is necessary because of the aforementioned clustering of data points.
Candidate cost drivers for parametric estimation Figure 2 shows that for some systems, TACH clusters by country (see, for example, that the data for mechanised aerobic treatment of wastewater are higher in the China cases than in the other three countries for which data points are reported). A much larger dataset will be required to properly understand the combination of factors which drive differing cost performance in each case. While the geographical location is likely to be one driver (as it will determine for example the relative costs of materials, fuel, and labour) other factors, including the scale of the system, population density, and topography are also highly likely to drive costs variation. A significant increase in the data held in the CACTUS database is needed in order to fully interrogate the cost drivers.
Reflections on data collection Figure 2 shows wide cost ranges for many of the components for which data have been collected. The NBPRA approach has proved robust at driving the collection of reasonably complete empirical data on costs, although many operators are unable to fully report on their cost liabilities. For older systems, there is often a lack of data on capital costs, and for newer systems, a lack of data on operational costs. In addition, many operators are not aware of certain implicit subsidies (for example, non-payment of electricity bills issued by national energy-generating organisations). However, the standardised data collection approach has shown promise in helping to drive up the quality of data that is collected. In addition, as the dataset grows, it would become possible to correct for omitted data (for example, by understanding typical cost distributions for particular systems under particular conditions). This type of correction may result in more accurate, clustered estimates of TACH for particular contexts.

CONCLUSION
Compared to other infrastructure sectors, there is a dearth of reliable, internationally comparable cost data for urban sanitation. Sanitation scholars have used both ball-park and analytical cost estimation approaches, and there have been some localised or specific efforts to generate models for parametric estimation. In the absence of reliable global data, the full costs of sustainable implementation or urban sanitation are being systematically underestimated (Flyvbjerg ).
This research considers the lessons learned in other sectors in addressing similar costing challenges and proposes a strategy for developing bespoke parametric cost estimation models to favour easy and reliable cost estimation, for alternative sanitation technologies, in a range of contexts.
Two key requirements of such an approach are the development of standard costing metrics and the development of a large and coherent empirical dataset of sanitation technology cost estimates selected from a range of geographical context and sanitation technologies.
The main contribution to knowledge to date comprises the proposition of the TACH -TACHand per capita -TACC, costing metrics, which are foundational to CACTUS.
The preliminary data collection, based on a standardised approach known as NBPRA, has generated an empirical data set which is larger than any that we have so far been able to find but not yet large enough to form the basis of a reliable global parametric approach to cost estimation.
Further work is proposed to increase the data contained in the CACTUS database and start to develop heuristics to understand how key cost drivers interact to determine the relative costs of different sanitation systems in a range of contexts. The CACTUS database is intended to become a public repository for empirical sanitation cost data which will facilitate future planning.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.