| Front page | | Contents | | Previous | | Next |

Water Prices in CEE and CIS Countries. Volume I: Main Text

Chapter 4
Customer Perceptions, Willingness to Pay and Demand for Water

4.1 Introduction

4.1.1 Purpose

This chapter and the following chapter have a joint purpose namely to provide guidance as to how to assess the demand for water and wastewater services. The demand for services is a function of relative prices, income and preferences of the customers.

The present chapter provides guidance as to methods to carry out an analysis of the households' willingness to pay for water services.

4.1.2 Target group and structure of chapter

The target group of this chapter is consultants who are familiar with market research, but who have not yet used advanced market research methods in the water services sector. While much information in this chapter may be of interest to policy makers, the main part has been written for their consultants / advisors. Policy makers may read Section 4.2 and skim the remaining sections.

Following a discussion in Section 4.2.1 of why to use these tools and when to use them, the text focuses on how to do market surveys and willingness to pay analysis within water and wastewater in the CEE and CIS countries. In Sections 4.5 and 4.6, we assume that the reader is familiar with market research methods¹², including the stated preference methodology suggested for use.

4.1.3 Approaches to willingness to pay surveys

There are a number of alternative approaches to willingness to pay surveys. We recommend a particular approach to assess the willingness of the consumer to pay for water and wastewater services in medium-sized and large municipalities in the CEE and CIS countries¹³. The following sections justify our recommendation and describe the recommended approach. They describe the typical process and discuss the need to interface with experts and recipients. The suggested approach for small towns is given in Chapter 7.

The term "willingness to pay" describes the consumer's preferences in relation to changes in water and wastewater services and prices. The "willingness to pay" is the (expected) payment a user is willing to pay for a given service/product or a given change in service level or product attributes.

When improvements are introduced, willingness to pay reflects the level of increase in payment that leaves the consumer indifferent as to the situation before and after the change. Likewise, the willingness to pay to avoid a deterioration of the situation represents the compensation in payment which will be necessary to leave the consumer indifferent.

Willingness to pay for water and wastewater services is determined by a large number of factors including, but not limited to, the public perception of the quality of water and wastewater services. The main determinants of willingness to pay are illustrated in Figure 4.1

Figure 4.1:
Determinants of willingness to pay for water and wastewater services

Sections 4.2-4.7 of this chapter deal with the methodology for quantifying willingness to pay for water and wastewater services. The methodology is sufficiently flexible to allow for all of the determinants to be included.

Section 4.8 briefly comments on how to deal with the estimation of price and income effects in relation to water and wastewater services.

4.2 Suggested approach, timing and resources

4.2.1 The selected approach - a brief justification

There are basically two types of data that can be collected and applied to assess willingness to pay:

	Revealed preference data
	Stated preference data

Revealed preference (RP) data is often preferred, when available, as it reflects observed behaviour. The traditional approach to demand assessment based on RP data is to assess the sensitivity of demand to changes in relative prices and income (elasticity). However, when faced with major infrastructure and service level decisions in the CEE or CIS water and wastewater services sector, this approach has limited value for a number of reasons.

First, for households the concept is only meaningful if the household is metered, thus giving a direct link between changes in demand and the price that the household pays for the service. Thus, the analysis cannot be made where households are charged according to norms.

Second, the traditional approach implies requirement for data based on observed behaviour. Ideally, panel data should be available and cover a period with major swings in relative prices and changes in income. Alternatively, a long time series for a period with unchanged institutions should be available. Such data is rarely available.

Third, the traditional time-series approach has a forecasting power only if the service will be essentially the same in the future as it has been in the past. However, the basic rationale of major infrastructure investments and/or a PPP arrangement is often to change the future service level compared to the past. Furthermore, customers do not choose between alternative water services as they choose between, say, alternative modes of transport. Also for this reason, the required data are unlikely to be available.

Fourth, the notion of total economic value recognises the fact that individuals may receive benefits from goods and services which may not be marketed. For example, water consumers may benefit from knowing that the wastewater treatment plant contributes to avoiding pollution. Assessment of the value of non-use benefits is often important for water services projects.

These considerations make methods based on hypothetical behaviour more appropriate when dealing with the valuation of water and wastewater services, in particular when future services are likely to differ from present services, or when non-use benefits are important.

There are two types of methods that are based on hypothetical behaviour and have traditionally been widely used:

	Contingent valuation
	Stated preference (or conjoint analysis)

In contingent valuation, consumers are asked to state their willingness to pay for a specific package of improved water services. This methodology can be useful if there is a specific package for the consumer to consider. However, there is a risk of consumers answering strategically, whereby respondents understate or overstate their valuation of the good or service in question.

The recommended approach to determine the willingness to pay for changes in the service level is a stated preference survey, also known as conjoint analysis. This is the most reliable method to assess the willingness to pay for future services or non-use benefits (e.g. avoiding pollution).

So far, stated preference methodologies have been used more in other sectors, such as transport infrastructure. This reflects the way in which the methodology was historically developed. In fact, the stated preference methodology is highly applicable for water services infrastructure. With this toolkit, we hope to achieve such wider usage.

We recommend the use of stated preference methodologies for the following reasons:

	Stated preference analysis is useful when a package of services is not pre-determined, and part of the aim of the market research is to investigate which services to provide.
	When using stated preference analysis, the risk of strategic answers is minimised as the consumers are asked to choose the preferred option among two or more alternatives. The suggested stated preference survey technique involves the trade-off between multi-dimensional choice pairs which makes it possible to find the optimal package of services for the consumer, and which makes strategic behaviour difficult to perform by the respondent.
	Stated preference is based on economic utility theory and has proven successful in market research. The case studies confirm its usefulness in the specific context of water and wastewater services in the CEE and the CIS countries.

The issue of assessing the sensitivity of demand to changes in prices (and income) is still a very important one. The financial viability of many water services projects is crucially dependent on sound assessment of how much water will be demanded (and sold) in the future. Traditionally, stated preference analysis does not directly provide a figure for the sensitivity of demand to changes in price and income.

We therefore suggest an enhanced approach to assess the future demand for water and wastewater services. We suggest to analyse affordability and willingness to pay separately and to carry out market surveys to assess the willingness to pay for water and wastewater services. The quantitative survey of willingness to pay may be designed to capture both willingness to pay for alternative services and the sensitivity of demand. An example hereof is given in Section 4.8.2. The consultant may also assess likely changes in water demand in response to specific combinations of future service levels, prices and incomes by combining analyses of affordability, and willingness to pay with a traditional assessment of demand elasticity based on historic data.

4.2.2 The stated preference approach - summary of process

The methods presented are interview-based surveys directed at users of water and wastewater services. The focus is on households, but the methods are also applicable for commercial customers (excluding industry) with some minor adaptations.

Specific data collection and questionnaire formats as well as typical output formats that have been tested and found useful during the case studies are presented in Appendices 4 to 7.

Text Box 2 provides a summary illustration of the process.

Text Box 2:
The recommended steps in undertaking a willingness to pay survey

Note: Generic examples included in the appendices are marked by *

The analysis of price and income effects on water demand in Section 4.8 can be carried out separately if the total cost of water services to the consumer is related to the consumption, now or in the future. If this is not the case, this analysis should be omitted.

The traditional assessment of elasticity requires adequate data to be available and the results to be deemed relevant for assessment of the future demand for the services to be provided. However, even if adequate revealed preference data is not available, the consultant must consider the effects on water demand of price changes in the situation where the total cost of water services to the consumer is related to the consumption. In these cases, the considerations may be based on the stated preference data collected and analysed.

Information to the recipient

During the process of the analysis, we recommend that the consultant keep in close contact with the recipient to make sure that the recipient is fully informed about the progress and, in particular, about the periods in which the recipient's customers are being interviewed. This is needed to avoid misunderstanding if representatives of the public contact the recipient to obtain information on the research. The recipient must ensure that information officers at the recipient's institution are fully informed about the progress and purpose of the study. The consultant can assist by preparing a short and concise briefing note to be used by information officers, cf. Appendix 11.

The recipient should also be asked to help by, for example, checking and commenting on survey questionnaires, introducing the research to the local population (thereby motivating them to participate in the survey) and providing facilities and logistical support for the consultancy team, for example offices and telephone lines.

4.2.3 Timing

The optimal timing of the willingness to pay analysis does not differ depending on whether the project is a PPP arrangement, such as a concession agreement, or a traditional investment project in a public utility.

The analysis is intended to feed into the determination of both future service and project design. Thus, in the traditional investment case, this is one of the steps undertaken during the feasibility stage, while in the PPP arrangement case, this is one of the steps undertaken as part of the preparation of the tender documents following the pre-feasibility stage.

For both a traditional investment project and a PPP arrangement, it is necessary to make a first estimate of the key service characteristics in order to be able to make a qualified analysis of customer perception and willingness to pay. This first estimate should include a first qualified "guesstimate" of the water balance in the city and an estimate of the "without project" scenario. Furthermore, analysed basic unit expenditure needs shall be available at this stage.

For a traditional investment project, it is important that the results are available before deciding on service level, project design and tariff setting.

For a PPP agreement, the optimal timing of the presentation of the results of this analysis is likely to be simultaneous with the announcement of the selected short list of firms invited to submit a bid for the lease, concession etc. The short-listed firms will find it useful to have an indication of the willingness to pay for various service levels in order to assess whether they will be able to match service level and tariff expectations.

The analysis may provide additional spin-offs for the municipality and/or the water utility. The results may provide useful information for designing a public awareness campaign, mobilising local political support etc.

4.2.4 Resources required

Typically, the willingness to pay and the demand analysis requires 14 to 18 calendar weeks as well as approximately 12 to 16 person weeks of resource input plus input from interviewers, translators, etc. It is therefore important to initiate the analysis as early in the process as possible.

Table 4.1:
Resources required for the customer perception and willingness to pay tasks

4.3 Engagement of a local partner

Engaging a good local partner is essential to achieve efficient management of the survey programme. The local partner, who may be a market research organisation or a department at a local university or institute, will be able to deliver the following services:

	Provide interpreters for interviews, briefings etc. (if the study team does not include a native speaker, it is essential that the local partner includes interpreters who are also fluent in the international working language);
	Provide translators to translate survey material;
	Provide recruiters to recruit the respondents for the qualitative research;
	Provide market research interviewers to conduct field work; and
	Organise facilities such as rooms for in-depth interviews, groups and briefings; organise equipment such as overhead projectors for the briefing and provide other facilities such as maps (to mark/code household locations), photocopying, etc.

In addition, the local partner will, in some cases, be able to provide knowledge of local conditions which are of relevance to the study, such as details from other relevant research, information on appropriate income and age ranges, household types to be used in the questionnaire, information for sampling purposes, etc.

In some cases, the local partner will also be able to help with estimations, interpretation of results, etc., but a team experienced in stated preference analysis should be responsible for carrying out these tasks.

4.4 Issues in the survey

The issues to be treated in the analysis have to be determined. The focus could be on improving the quality of the water services or rather on avoiding a deterioration of present level of services. There is a range of issues that can influence the willingness to pay for the services and thereby the public acceptability of water and wastewater prices. The next section lists a range of issues that can be considered when conducting the analysis. The consultant should choose the most relevant ones for the research.

This is done by:

	Undertaking background research
	Meeting the relevant people

The ways to determine which issues to include in the survey are discussed in section 4.4.2.

4.4.1 The range of issues

The main determinants of willingness to pay are:

	Perceived quality of water/wastewater service
	Household income
	Current price of water/wastewater

The perceived quality may differ from the "objective" quality. Here the emphasis is on perceived quality, and hence, the issues to be determined are, for instance, whether the focus should be on both drinking water and wastewater or rather on improving the healthiness of tap water.

The consumers' willingness to pay influences the public acceptability of water and wastewater prices. However, a range of other issues is also likely to have an impact, in this respect. These issues are illustrated in Figure 4.2.

Figure 4.2:
Non-traditional issues influencing public acceptability of tariff increases

As it can be seen, there are a number of issues that may influence the public acceptability of tariff increases. Therefore, the research method must ensure that these issues are taken into account, where relevant. A number of procedures may be applied to ensure that the relevant issues are included in the research. These are described below. Of course, the qualitative research itself will also elicit valuable information on the relevant issues.

History of price and service levels

The history of price increases and service level may have a significant impact on the acceptability of future price increases. For example: Have previous price increases been justified with reference to service improvements? Has the utility delivered as promised?

If water services have previously been provided as a free (or heavily subsidised) public good then resistance to charges based on full cost recovery can be anticipated.

Has there been a history of successful public protests against price increases?

Affordability of the tariffs

This issue is discussed in depth in Chapter 5.

Effectiveness of public relations (PR)

Raising charges is never likely to be a popular measure and may become a major political issue. To convince the public of the need to increase charges, it is essential to develop a public relations strategy. This could involve the following elements:

	Transparency of price setting. Assuming that the increased charges are justifiable, then openness about the reasons for raising prices and structuring them in a particular way could gain public support. Conversely, a lack of transparency could lead to alienation of interested stakeholders, public suspicion and a lack of support.
	Dissemination of information. Linked to transparency is the need for wide dissemination of information to both the general public and groups involved in the policy process so that they can understand the reasons why the charges have been increased.
	Professional PR activities (advertising, press conferences, etc.).

Public awareness of environment and public health issues

This is a very basic but important point. If they are not major public concerns, higher charges for environmental and public health benefits are unlikely to be acceptable. This is particularly the case for wastewater treatment where the benefits of improved service are less tangible than in the case of water supply.

Political/media reaction to tariff increases

If the increased prices are turned into a campaigning issue by the media or by political parties, this will clearly have a major bearing on the reaction of the public to the charges. Furthermore, political parties can perform a leadership function, and political activists will often take a stance on political issues following the position of their chosen party. If their party is either for or opposed to the price increase, it is likely that many activists will follow this line.

The relationship between public acceptability and political acceptability is discussed in Sections 6.1 and 6.3.

Public involvement in the decision-making process

The interests of key stakeholders need to be taken into account if they are to be expected to support the price increases. In many cases, their views will visibly have to be taken into account in order to achieve "buy-in" to the proposed price increases (e.g. providing subsidies to socially disadvantaged groups in order to make the proposal politically acceptable).

Perceived utility efficiency

In most circumstances, the charges will be collected by the water utility which, in return, is expected to make the necessary investments in the water system. For this arrangement to be acceptable, there needs to be a degree of public trust in the utility - that the revenues collected will be used effectively and in the public interest. Different factors that could come into play are:

	The ownership of water utility: Public ownership could be perceived as being inefficient compared with private operators' services, and local authority control may be viewed as being corrupt and/or politicised. Private ownership could be viewed as being more efficient. But alternatively, high charges for a basic commodity such as water could be viewed as profiteering from the provision of an essential public service. Acceptability is likely to be linked to effective public regulation of the utility. Negative reactions to private ownership can be magnified if the owner is foreign (the idea of increased payments resulting in increased profits for foreign shareholders).
	If the service provider is perceived as inefficient / wasteful / politically controlled, this is likely to undermine public support to paying increased charges.
	The effectiveness of consumer protection legislation, an independent regulator, and antimonopoly authorities.

Perceived fairness

If charges are not perceived as being fair, they are unlikely to be perceived as acceptable. Different factors may influence the perception of fairness:

	In the CIS countries, it is common for certain users to provide cross-subsidies for other users. Under such circumstances, those providing the subsidies (e.g. private sector industry) may find it unacceptable that other consumers benefit from reduced rates (e.g. public sector institutions).
	If the charges do not take ability to pay into account, they will certainly be unacceptable to those who find the new charges unaffordable. This is also likely to provoke a negative reaction in those sections of society interested in social welfare and social justice.
	If the charges are calculated without taking into account the actual water use (as it is common throughout the CIS countries), it is possible to envisage a situation where a wasteful user is billed the same as a more economical neighbour who only uses a fraction of the amount consumed by his neighbour. Under such circumstances, acceptability could depend on the perceived fairness of the water use norms.
	If meters are being introduced to some consumers, only, a possible scenario would be for those with meters to reduce their water use (and therefore their bill). If the overall water charges are increased to compensate for this lost revenue, this might have a disproportionate effect on those without meters. They are likely to find such a situation unacceptable.
	If tariff levels are significantly higher than those in other municipalities, the reasons for such difference will need to be convincingly explained, or it could become a source of public dissatisfaction.
	Procedures for enforcement need to be considered. If non-payers are cut off from the water supply, it could be viewed as socially unacceptable in relation to such an essential public utility. Alternatively, if enforcement is not perceived as being uniform (some customers are able to get away without paying), the acceptability of paying increased water charges is likely to be undermined.

4.4.2 Determining the issues

If the terms of reference are specific on certain issues, for example, defining precisely different segments to be analysed or defining which sort of changed/new/improved services to investigate, the willingness to pay for this should, of course, guide the consultant in determining the issues. If the TOR are more open-ended, the following approaches are useful to determine the issues.

Undertaking background research

The relevant issues can be determined by undertaking background research in order to understand:

	the realistic technical options/alternatives and, in particular, their impact on realistic service levels (for a discussion hereof, See Chapter 3);
	financial and administrative issues such as: billing, collection of payment, enforcement of collection of payment and other rules, principles and practices of how to measure water use, the financial standing of the utility etc.; and
	the socio-economic context, in particular income levels and distribution and types of households.

A wide range of material will typically be available which may help to determine the issues on which it is relevant for the stated preference analysis to focus. These may include:

	reports, statistics, working papers etc. giving general introduction to the utility;
	previous market research reports on water or other relevant areas (such as other utilities);
	reports, statistics, working papers on the (local) economy in the area of the study showing income, socio-economic data, etc.

Meeting the relevant people

Interfacing with experts within financing and engineering is necessary whether the willingness to pay and affordability analyses are carried out as stand-alone analyses or as part of a feasibility study. The experts can be permanent staff of the utility in question or consultants.

Discussions with engineers and financial experts are needed to gain an adequate understanding of the project context and the technical and financial operations of the water and wastewater utility.

Background information

The type of information that can be collected at this stage includes:

	Details of tariff structure - how are tariffs calculated? Is it a flat rate? How much is for water and how much for wastewater?
	Possible privileges and/or subsidies, etc.? Who qualifies?
	Procedures for collection of water bills. Are they part of a household bill that also includes rent etc.?
	Collection rate.
	Consequences for nonpayers.
	Existence of meters or plans to install meters.
	Date of the last rise in water rates. Size of the last rise. Development in other key prices such as those of utilities, rent, food, transport etc.
	Other relevant information such as investment plans, environmental issues, major incidents, etc.
	The state of the economy - inflation, growth, etc.

4.5 Qualitative research

Following the determination of the issues the qualitative research can be conducted. If it has been concluded that a key issue of interest is to avoid a deterioration of the present water services, the qualitative research uses this as its starting point.

Qualitative research comprises open but structured interviews or discussions with a sample of users with the purpose of:

	Determining the important factors as perceived by the users (healthiness of drinking water, taste, smell etc.), including checking that the factors already identified are relevant and identifying new factors of importance to the customers;
	For the key factors, e.g. healthiness of tap water, determining the likely ranges of willingness to pay to be used later for the quantitative research (see Section 4.6). An important point here is to ask respondents to bring their last water bill, as this will facilitate discussions on willingness to pay;
	Determining the language and vocabulary used and understood by the re-spondents for later formulation of questionnaires; and
	Identifying other issues of relevance not identified during the background research.

Qualitative research is a key part of the willingness to pay methodology. It comprises the first contacts made with the users, and it helps define the issues being analysed through the quantitative research.

An important factor or issue that is not identified in the qualitative research is likely to remain undiscovered and hence constitute a major weakness of the whole method.

4.5.1 Determine the method - in-depth and focus group discussions

There are two main formats for qualitative research: in-depth interviews and focus group discussions.

In-depth interviews are face-to-face and conducted with respondents in a suitable, comfortable environment. One or two interviewers interview each respondent with consecutive translation by a professional interpreter. Each interview lasts between 30 and 60 minutes. A programme of 6-12 interviews is required. The number of interviews required could be higher if there are different groups with different attitudes in the area, or if the level of service provided differs substantially among consumers in the area.

A focus group consists of a group of typically eight respondents. The members of the group should represent water users in general. If more than one focus group is established then there is an opportunity for segmenting the groups (e.g. one focus group of men and one of women) to identify whether there are differences between the segments. Obviously, awareness questions can only be tested in a focus group by including, for example, a small questionnaire to be filled in individually by group members. Focus group sessions typically last approximately 2 hours.

Focus groups have the following advantages compared to in-depth interviews:

	respondents can inspire each other;
	respondents also act as interviewers when asking each other questions;
	respondents might feel less "interrogated"; and
	it takes less time per respondent included in the research.

Some disadvantages are:

	risk of side-tracking the discussions;
	risk of very dominant/reluctant group members;
	difficulty of testing the awareness of group members since they educate others;
	the general demands on the moderator are greater than those on experts undertaking indepth interviews;
	risk of group pressure, hiding the points of view of individual respondents; and
	some personal questions are more difficult to discuss in a focus group e.g. income.

For both in-depth interviews and focus groups, a financial incentive for participation should be provided (equivalent to approximately two hours' wage income or 2-15 EURO, the lower end being relevant in the CIS, the higher end in CEE).

Both the in-depth interviews and the focus groups use a topic guide (see Section 4.5.3 below). This is prepared using relevant background information.

Recommendation: A mix of in-depth interviews and focus group discussions provides the best approach for the qualitative research. For example, the focus group can provide information that cannot be elicited in the in-depth interviews and vice versa.

Text Box 3:
Focus group in Kaliningrad, Russia

   
There are two options in relation to the moderation of the focus groups:

Moderated by a member of the research team with consecutive translation of questions and simultaneous translation of responses

or

Moderated by a local expert with the research team following the discussion through consecutive translation of questions and answers.

Both options have advantages and disadvantages. The former puts the research team more in control and allows issues to be probed further, if necessary, but it seems to inhibit some freeflowing discussion because of the breaks due to translation.

The latter provides for a much more free-flowing discussion, although we would recommend that the group be taped (sound only or video and sound) so that the translation does not take place in the same room. This is because the translation can affect the group dynamics, with some participants talking to the study team and not the other group members. Also, this approach would work much better if the moderator is an experienced group moderator and well versed in the local water issues.

Text Box 4:
Quantitative research in Brno, Poznan and Kaliningrad

  
4.5.2 Design sample and recruit respondents

For qualitative research only a small sample is required. It would typically cover representatives of the major segments of the relevant users.

For water research, "relevant users" include the entire population in the geographical area in question. The sample for qualitative research should be chosen in such a way that it can be expected to reflect all possible service issues. The sample should not reflect the proportions in the population.

It is our experience that in-depth interviews with 6-12 customers and one focus group discussion are sufficient.

Text Box 5:
Characteristics of sample for in-depth interviews and focus groups in Poznan, Poland

  
4.5.3 Topic guide and conducting qualitative interviews

The content of the topic guide will depend on the local circumstances as elicited from the terms of reference, background research and discussion with relevant people as described in Section 4.4. The topic areas are likely to include most or all of the following areas:

	Introduction to the survey
	Respondent characteristics
	(Cold) water supply14
	Assessment of present (cold) water service
	Assessment of present wastewater service
	Political issues
	Attitudes to the water company
	Payment of water bills
	Willingness to pay.

A generic example of a topic guide is included in Appendix 4.

The format of the topic guide should allow plenty of room to write down responses.

The nature of the qualitative research process means that the topic guide is used as a structured guide to the process. It is important that the topic guide is used as a launch pad for enquiry rather than a structured questionnaire. The interviewer should explore further areas which reveal themselves to be of interest.

Not all topics in the guide are necessarily covered, and also, those covered may occur in a different order than that shown in the guide.

Text Box 6:
Evolution of the topic guide in Brno, the Czech Republic

   
4.5.4 Debrief/prepare report

Following the qualitative research, the consultancy team will meet to discuss the findings of the research. This process involves going through the written notes from all the interviews/focus group discussions, topic by topic to assess the importance or weight of opinion for each topic area.

A report is then prepared based on the results of the qualitative research.

4.6 Quantitative research

The quantitative research is designed using the findings of the qualitative research, and will provide quantitative measures of willingness to pay as well as other aspects of interest. The quantitative research includes the following steps:

	Design of the sample
	Design of the questionnaire
	Design of stated preference games
	Translation of survey materials
	Briefing of interviewers
	Pilot phase of the field work
	Pilot phase report
	Main phase of the field work
	Field work quality control
	Coding of questionnaires and data entry into database
	Data analysis
	Reporting

Each of the steps is discussed in detail below, and the text includes examples from the case studies.

4.6.1 Design of the sample

The sampling methodology is critical since the results must be representative of the population in order to be valid.

Ideally, a stratified random sample using electoral roll data would be used. In practice, the budget and/or availability of appropriate records may not allow such an approach. Therefore, a sampling method must be designed which, as far as possible, results in a representative sample.

Text Box 7:
Sampling methodologies. Examples from Brno, Poznan, and Kaliningrad

  
Regardless of the method applied, special care should be taken to ensure that the sample is randomly chosen and that it is appropriate. For example, in a region with a low level of telephone ownership, phoning respondents would not be appropriate.

If the aim is to interview a representative sample of the inhabitants of the city, extensive and recent data on inhabitants has to be available.

In practice, a random sampling approach is most appropriate when data on inhabitants is not available or of poor quality.

Whatever the sampling approach, a call-back regime should be instituted so that the sample does not over-represent those who are more often at home (typically the elderly and the unemployed). Such a call-back regime means that, if there is no one at the sampled location, you call back up to three times at different times and days.

Finally, the approach should ensure that the person responsible for paying the water bill is interviewed.

4.6.2 Design of the questionnaire

The questionnaire can be divided into two parts: a general part on the socio-economic variables, water usage and attitudes towards the present service levels, and another part including the stated preference questions. The two parts of the questionnaire are described below. A generic questionnaire with a wide range of questions is included in Appendix 5.

General part of the questionnaire

The questionnaire is designed on the basis of the information gathered in the background research and the qualitative work.

The questionnaire should include:

	Assessment of existing water and wastewater services
	Attitudes towards water and wastewater services
	Importance of improvements in water and wastewater services
	Payment of water bills
	Respondent characteristics
	Other subjects of interest, e.g. attitudes towards the water company, political attitudes, attitude towards recycling.

The questionnaire may also include questions on expenditure, if income data is deemed unlikely to be of great validity (for instance because public sector employees have not been paid for a considerable amount of time, and/or there is a significant grey economy).

Some questions should be asked with respect to different seasons. For example, bottled water usage in a typical summer month and a typical winter month.

Many subjects can be covered by a questionnaire. However, a selection has to be made in order to keep the length of an interview acceptable to the respondent (20 to 30 minutes). It should be borne in mind that the stated preference part of the questionnaire will take about 10 minutes.

Identification of stated preference variables and levels

A list of possible improvements in the water services will have been identified in the background research and the qualitative research. Probably, it will not be possible to accommodate all options in the stated preference game so the most appropriate ones should be selected.

The possible changes in service levels can then be divided into two or three groups. The most important services are included in a stated preference game as described below.

The other, less important changes in service levels can be included in importance-rated questions in the questionnaire in order to obtain information on the respondents' evaluation of other services. If the changes in service levels from the stated preference are also included in the importance ratings, then a link can be made between the stated preference 'willingness to pay' values and the other values (although this is not a statistically robust approach).

Some possible changes in service levels may be assessed to be of minor relevance/importance and, therefore, not included in the questionnaire at all.

4.6.3 Design of stated preference game

This section describes the design of a stated preference game. We would recommend that just one stated preference game be undertaken. This is to ensure that the questionnaire does not become too long, and that the task is not too complex for the respondent. It is technically possibly to include more than one stated preference game and, with a link variable, to assess the willingness to pay for a large number of variables.

The willingness to pay for water service is examined using stated preference techniques. A stated preference game consists of a number of factors - normally three or four. One of these factors is the tariff, and the rest are service factors of interest, e.g.

	Water quality
	Smell
	Colour/clear water
	Pressure
	Hours of water supply per day
	Standard of wastewater treatment.

The choice of the service factors depends on the current standard of the supply, the feasible improvements and consumer requirements. Each of the factors has two, three or four levels, each of which describes a certain service level.

An example of a choice task in the stated preference game is presented in Text Box 8.¹⁵ From the choices that the customers make, it is possible to infer the willingness to pay for each of the changes included in the choice task.

There is a limit to the number of choices a respondent can undertake before losing interest. As a rule of thumb, we recommend that eight choice tasks be presented to each of the respondents. Eight choice tasks are sufficient to estimate a model, and adding more choice tasks to the questionnaire often does not provide more information as there is a fatigue effect making respondents answer randomly to the last questions. The number of choices should be tested in the pilot interviews.

Text Box 8:
Stated preference game choice task, Kaliningrad, Russia

   
Text Box 9:
Factors and levels in a stated preference game, Kaliningrad

   
The actual choice pairs presented to the respondent should be chosen in such a way that all parameters can be estimated without multicollinearity. This is achieved by applying an orthogonal design. This means that the various combinations of the factors and levels are sufficiently varied so that the willingness to pay can be estimated for all service changes. This is done by using the appropriate software like SAS¹⁶. Other software packages include Speed/Mint from Hague Consulting Group (www.hcg.nl) and Sawtooth Software's SMRT (www.sawtooth.com).

It is recommended not to include trivial choice tasks where one alternative dominates the other, e.g. where the left-hand side describes an improved situation with improved water services at a lower tariff, and the right-hand side describes the present situation.

4.6.4 Translation of survey materials

The questionnaire and briefing notes for interviewers should be translated into the language of the c s that there is more room for misinterpretation than in prose. Also, there may be cultural differences in the use of language.

Text Box 10:
Cultural differences: Quality rating in Brno and Poznan

   
Therefore, the translation of the questionnaire should be discussed with the translators, allowing them an opportunity to raise any relevant issues.

Furthermore, the design and levels of the stated preference games should be thoroughly checked so that design and levels are exactly the same in the English language version and in the local language.

4.6.5 Briefing of interviewers

The interviewers for the pilot and the main phases will need to be briefed face to face. The briefing is required in order to:

	Explain the background of the research;
	Describe the tasks they are to perform in terms of:

	Sampling including how to choose appropriate respondents, what to do in case of non-response, call backs procedures, etc.
	Recruitment questionnaire, if applicable (to register appointments for later interviews)
	How to fill in the questionnaire including explanation of probes (asking if there are any other reasons), jumps ("go to" instructions from one question to another), what to do if responses cannot be coded etc.
	Manner and approach with respondents.
	Handing out incentives (if applicable).
	Return of questionnaires etc.

An example of the pilot analysis undertaken in Kaliningrad is given in the table below. Here, the respondents were given eight choice tasks, numbered 1 - 8. All respondents agree that A is preferred to B in choice task 3. Thus, the information from this choice pair will be poor, as the choice is too obvious.

Chosen alternatives by choice number

This means that e.g. choice task 3 is not giving much extra information. In order to examine the cause, it is necessary to analyse the actual choice pair, cf. Text Box, 12.

The table below gives the number of respondent who chose the 'better' alternative for each factor. Here Cost is a dominating factor as most of the 25 respondents in the pilot choose the 'better' (i.e. lowest cost) alternative in most of the games. Choice no. 1 is the one where most respondents chose the more expensive alternative (cost did not vary in choice no. 7).

Pilot:
Number of respondents by choice task no. and chosen factors: Cost, water quality, smell and supply and pressure

  
Because these results indicated that cost is very important and, in fact, dominating in the choice pairs, the increases in cost were reduced. In the main round, this lead to responses where the improved and expensive alternative was preferred more often, cf. task 4, 5, 7 in the table below

Final:
Number of respondents by choice task no. and chosen factors: Cost, water quality, smell and supply and pressure

For the Kaliningrad case study, the estimated utility function was:

where the P_i(A) is the probability of respondent i choosing A instead of B Ù is the cumulative logistic function, and the ß's are the parameters to be estimated. Finally, the variables in the formula represent the difference between the values of alternative A and alternative B, e.g. Tariff_i = Tariff_i ^A - Tariff_i^B.

For the Kaliningrad case study, the importance of the price differs between consumers with high income and consumers with low income. Therefore, two different parameters where estimated. Naturally, consumers are expected to prefer lower tariffs to higher tariffs so the signs of both ß_{Tariff (low income)} and ß_{Tariff (high income)} are expected to be negative. Furthermore, it is expected that the consumers with a low income dislike higher tariffs more than consumers in the high income group, and hence it is expected that ß_{Tariff (low income)}<ß_{Tariff (high income)}.

Consumers in different age groups have different evaluations of the importance of the water being always safe to drink. Therefore, different parameters for three age groups have been included in the model. There is no a priori expectation as to which age group has the highest willingness to pay for safe drinking water.

The three remaining parameters estimated in the model do not vary for different socioeconomic groups. However, 24-hour supply is only a relevant improvement for consumers without all day supply.

Tariff

Water safe to drink

No smell

24 hours supply

Always good presure

low income

high income

Age 18-44

Age 44-54

Age 55+

Estimate

-8.825

-5.941

1.636

1.273

0.829

0.191

0.507

0.388

t-ratio

-9.2

-9.0

8.9

6.0

5.0

1.6

3.7

2.5

In the case study in Kaliningrad, the willingness to pay extra for water always being safe to drink among low-income consumers between 18 and 44 years of age was:

The tariff is measured in percentage changes in the current tariff, so the result is that this group is willing to accept a tariff increase of 19% to have tap water that is always safe to drink.

In Poznan, a few selected key policy guidelines were identified, based on the willingness to pay research:

	There was a significant WTP to avoid decreasing quality;
	The WTP is not fully exploited;
	People will pay for the removal of odour from water ;
	People will pay for clear water ;
	The environment can be used as an argument for price increases (for wastewater services);
	The present level of information is considered poor;
	Different information approaches are required for different age groups and gender; and
	Metering will not increase the WTP

If there are consumers who use large amounts of water and/or contribute a large share of the water utility's revenue, interviews should be conducted with these consumers in order to investigate their preferences and likely reactions to an increase of the tariff.

Such interviews should, at least, include the following questions:

	Do the enterprises and the public institutions pay for the water services they receive?
	Is there a cross-subsidisation between these consumers and household consumers? If enterprises cross-subsidise the household consumers, will they also accept this situation in the future?
	How will the large consumers react to an increase in tariffs?
	Are alternatives to public water and wastewater provision available to these largescale consumers? For example: Can they introduce cleaner technologies? Can they establish their own source of water supply? Can they establish their own wastewater treatment plant? If yes, to any of these questions - what is the approximate cost to the large-scale consumer of making such a shift to lower demand / alternative service provision?

No registered data was available on water demand and income in Brno. Therefore, the respondents were asked to bring their water bills for the last three years to the interview. These water bills were used to obtain data on the amount spent on water by each of the households. In cases where the respondents did not have the water bills, the amount was estimated. Furthermore, the respondents were asked to state the changes in their income over the past three years.

Two specific changes in the water tariff were presented to the respondents as SP questions. The respondents were asked to state the expected consumption for each of the tariffs. This was done in order to obtain more variation in the value of the tariff variable in the model.

To compute the price and income elasticity, it is necessary to know how the respondents will react to an increase in price, other things being equal. Therefore, a demand model including all data collected is estimated.

The prefix ln indicates that a logarithmic transformation has been made, ß is the price elasticity and g is the income elasticity. e_it is the unexplained part. The index i is the household, while index t indicates that there are more observations for each household.

The price elasticity is estimated at -0.18, which means that a price increase of 10 per cent results in a 1.8 per cent decrease in water consumption. The income elasticity is estimated at +0.33, which means that an increase of 10 per cent in income results in a 3.3 per cent increase in water consumption.

Hence, the water consumption of the respondents is, as expected, quite insensitive to price, but the income elasticity is relatively high. The results show that the households are not likely to reduce their water consumption very much further in response to an increase in price, but that they will use more water if their income increases.