Economic valuation of externalities linked to transport ... · ECONOMIC EVALUATION OF TRANSPORTATION PROJECTS 1 November 2009 1 Economic valuation of externa lities linked to transport

1 November 2009

Raúl Brey

Economic valuation of

externalities linked to

transport projects:

Foundations and

procedures

Proyecto: EVALUACIÓN SOCIOECONÓMICA Y FINANCIERA DE PROYECTOS DE TRANSPORTE

Con la subvención del Centro de Estudios y Experimentación de Obras Públicas (CEDEX) Ministerio de Fomento

Ref. PT-2007-001-02IAPP

www.evaluaciondeproyectos.es

ECONOMIC VALUATION OF EXTERNALITIES LINKED TO TRANSPORT PROJECTS: FOUNDATIONS AND PROCEDURES

1 November 2009 Raúl Brey

Summary

Revised 24/04/10 The purpose of this paper is to present, in a simplified way and as free as possible of

specialised terminology, the main procedures for the economic valuation of those

costs and benefits of transport projects that are not captured by the market and to

show the theoretical foundations behind these procedures.

Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresTable of Contents

1. INTRODUCTION ................................................................................................................. 1

2. THE CONCEPT OF ECONOMIC VALUE .................................................................................. 1

3. IMPLICATIONS OF THE CONCEPT OF ECONOMIC VALUE ..................................................... 2

4. PHASES PRIOR TO THE PROCESS OF ECONOMIC VALUATION .............................................. 3

4.1. THE IDENTIFICATION PHASE ......................................................................................................... 3

4.2. THE STAGE OF QUANTIFICATION ................................................................................................... 4

5. ECONOMIC VALUATION METHODS .................................................................................... 5

5.1. TECHNIQUES BASED ON RELATED MARKETS .................................................................................... 5

5.1.1. Averting behaviour method .......................................................................................... 6

5.1.2. Travel cost method (TCM) ............................................................................................ 6

5.1.2.1. Estimation of the economic value of recreational services of a site ..................... 7

5.1.2.2. Estimation of the economic value of changes in the characteristics of a site ....... 9

5.1.3. The hedonic price method (HPM) ................................................................................. 9

5.2. TECHNIQUES BASED ON HYPOTHETICAL MARKETS .......................................................................... 11

5.2.1. Contingent valuation method (CVM) .......................................................................... 12

5.2.2. Models based on multi‐attribute choices ................................................................... 14

6. CONCLUDING REMARKS ................................................................................................... 16

REFERENCES ......................................................................................................................... 19

ECONOMIC EVALUATION OF TRANSPORTATION PROJECTS 1 November 2009 1

Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures1. INTRODUCTION

The cost-benefit analysis requires quantification, ideally in monetary terms, of all the costs

and benefits attributable to the project under evaluation. In some cases, these costs and

benefits can be valued using the prices provided by the market system, making appropriate

adjustments in situations where there are market distortions. On the other hand, in cases

where impacts are not directly valued by the market, the monetary valuation is achieved by

procedures grounded in economic theory that aim to provide a monetary estimate of the

impact of these costs and benefits in the society. This chapter discusses a simplified

explanation of the most important procedures.

It should be emphasised, therefore, that the monetary quantification of impacts that, despite

affecting goods and services not traded in the market can be valued because they affect the

costs and production levels of other market goods and services, is outside the scope of this

chapter

The remainder of the chapter is organised into five sections. Section 2 explains the

principles that guide the process of economic valuation. Section 3 describes the

implications of this formulation. Section 4 sets out the steps to be followed in the process of

economic valuation. Section 5 describes the main methods of economic valuation and the

last section contains some final considerations.

2. THE CONCEPT OF ECONOMIC VALUE

By quantifying, in monetary terms, the non-market costs and benefits derived from a

project, the economist seeks to reflect how they will improve or worsen the welfare of the

society. The economic approach to capture these changes in welfare is based on satisfying

the preferences of individuals. It is first assumed that the satisfaction of individual

preferences results in an increase in their welfare. In this way, decisions taken by

individuals can serve as guidelines to define their welfare. If, for example, we observe that

a person trades a certain amount of money for a good X, it can be concluded that the

exchange improves the person’s welfare because their welfare is greater than if they had not

performed the action.

The next question that arises is: how much does the welfare of the individual improve? The

answer to this question leads to the concept of economic value.

Economic value is defined through the concepts of willingness to pay (WTP) and

willingness to accept compensation (WTA). The economic value to an individual of an

improvement caused by a project can be defined as the maximum WTP for getting it

(Measure 1) or the minimum WTA for renouncing it (Measure 2). Similarly, in the case of

2 1 November 2009 ECONOMIC EVALUATION OF TRANSPORTATION PROJECTS

Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures a cost, economic value is defined as the maximum WTP of the individual to avoid it

(Measure 3) or the minimum WTA for allowing the welfare loss caused by the cost

(Measure 4). Thus, after making the payment, on Measures 1 and 4, the individual retains

the welfare level he had before the project, while in Measures 2 and 3, the individual

maintains the level of welfare he would have had if the project was implemented.

The choice between one way or another to measure changes in welfare (WTP or WTA)

basically depends on the perception by the individual of his right to improve or not to bear

the cost. The results provided by both approaches tend to be quite different due to various

theoretical considerations (Hanemann 1991, Freeman 1993) and other considerations (see,

for example, Kahneman and Tversky 1979, and Hanley 1988).

Through these measures, an estimate of individual preferences for the proposed change

would be obtained and, since as stated above it is assumed that what the individual is

seeking when he satisfies his preferences is to maximise his welfare or utility, a monetary

measure of the change in welfare that this service, good or status of the environment poses

to the individual. Adding the welfare measures obtained for all individuals, we get the

monetary quantification of the project's impact on the society as a whole.

3. IMPLICATIONS OF THE CONCEPT OF ECONOMIC VALUE

From the discussion in the previous section we get the idea that the economic concept of

value is an anthropocentric concept, it is based on the individual and it revolves around

him. What are valued are the preferences of the society for the impacts caused by a project.

This concept of value has a number of important implications, some of which are detailed

below:

Goods or services will only have economic value if the individual values them

directly or indirectly. Since the economic value is measured in terms of WTP or

WTA for the goods or services, someone must be willing to pay for them or to

accept a direct or indirect compensation. Thus, if any impact caused by a

transportation project is not perceived by individuals, it will not entail any changes

in their welfare and therefore will have no value from this point of view. For

example, consider a coastal area. A biologist would look into the water of that area

and classify it as more or less degraded compared to other areas depending on the

degree of presence of micro-organisms. However, in economic terms, the

degradation of an area over another only has economic value if it is perceived by

individuals. If no one notices the difference in the quality of water then there is no

difference in value regarding this aspect.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures The approaches based on the costs for the monetary quantification of the impacts

of a project n not captured by the market1 are not related to individuals' preferences

and therefore to the concept of economic value previously explained. As seen

below, the use of economic valuation methods is not an easy and quick task. This

means that studies sometimes resort to simpler and quicker procedures consisting

of monetarily quantifying the impacts of a project through costs (for example,

through the restoration costs, costs of replacement or substitution, or control

costs). The main drawback of these procedures is that they are not based on the

preferences of individuals. Most of these methods assume that if the costs were

actually incurred it is because society values the good or service lost at least that

amount of money. Therefore, these methods must be applied when the cost to

replace, to avoid the loss or restore the non-market good or service have been

completed. This means admitting that decision-maker agents are capable of

making optimal choices when deciding whether to undertake these expenditures. If

the social agents have made an investment, for example, to restore a particular

environmental good it is because they "know" that the gains to society of such

action outweigh the costs and, therefore, these costs may be taken as a lower limit

for the economic value of that property.

The economic value of a good or service depends on both its characteristics and

the preferences of individuals with respect to those characteristics. This causes the

same good to be valued differently by different individuals and that its valuation

cannot be transferred directly to the economic valuation of the impacts of projects

in different studies.

4. PHASES PRIOR TO THE PROCESS OF ECONOMIC VALUATION

Before valuing a cost or benefit of a project, it is necessary to characterise the impact, i.e.

knowing exactly what we want to evaluate. Therefore, as a prelude to the monetary

valuation process, it is necessary to conduct a series of phases that are discussed below.

4.1. The identification phase

The purpose here is to discover and bring out all the possible potential impacts caused by

the development of an action. This phase is particularly important and complex in the case

of environmental impacts

1 As noted in the Introduction, we exclude the impacts on non-tradable goods and services when they act as factors of production,

since in such cases these impacts lead to changes in production costs, so they are directly reflected in the market. In these cases,

the cost-based approaches can be used to obtain the monetary quantification of these impacts (Freeman 1993).


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures In order to undertake this identification stage, the holistic and interconnected nature of

natural systems requires in-depth knowledge of the various relationships. Furthermore, the

ability to set limits on the analysis is required so as to focus on a particular field, otherwise

the analytical spectrum would be unmanageable and too complex (space, time, type of

relationship, etc.). The establishment of these limits will markedly influence the quality of

the analysis, so they must be established rationally and reflected in the statement of results.

For a definition of the "catchment area" of the study there are different approaches. Below

are a list of those suggested by the International Union for the Conservation of Nature and

Natural Resources (IUCN) (1980):

a. Temporal and spatial horizon: It is necessary to define the geographic and time

scope in which we will measure the environmental effects. Obviously, there is no

general rule, but with regard to the natural limits, it is recommended to respect

them whenever possible. In the case of time limits, they should be long enough to

cover the lifetime of the investment project, while being aware that the process of

discounting that is usually done within the cost-benefit analysis may cause the

long-term values to eventually become negligible values, meaning extending the

time limit too far is not worthwhile.

b. Criterion of urgency: Under this approach we try to collect what the sensitivity of

natural systems to the proposed action is and the time we would have to develop

corrective and rehabilitative measures.

c. Degree of irreversible damage: This would involve determining the level of

irreversible effects produced by the plan.

Along with these criteria we should take into account other aspects of this process of

identification of impacts, including the nature of the effects (classification of different

effects according to various criteria such as: if they affect human health, productive

capacity of the environment, survival of certain species, etc.) or possible synergistic and

cumulative effects between the components of the project and other projects, because even

though individually the effects of various projects or their components can be small,

together they may prove very damaging to the environment.

4.2. The stage of quantification

Once the effects have been identified, the next step is to quantify them. This phase does not

consist of monetary quantification (this will be done later in the economic valuation), but to

express these effects in physical units. Obviously, it is not usually possible to quantify all

the effects. In such cases, these immeasurable effects should be expressed qualitatively.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresThe impacts caused by the project cannot be meaningfully quantified without a basis for

comparison, i.e. the conditions that would exist in the event of no action taking place. What

matters are the impacts directly attributable to the development of the action. Therefore it is

necessary to distinguish between changes attributable to the project from those that result

from the natural evolution. Thus, it is necessary to compare the situation "with" and

"without" the project, so it will be necessary to establish evolutionary trends or scenarios of

the environments affected by it.

Once these impacts have been identified and characterised (quantitatively or qualitatively),

the next step is to undertake their monetary valuation following the criteria explained

previously.

5. ECONOMIC VALUATION METHODS

Economic value has been defined above from the preferences of individuals. When the

costs or benefits of a project involve goods and services traded in the market, the decisions

that individuals take in that market may serve as a means to explore their preferences for

these goods and services and thus quantify the impact monetarily. In the case where the

costs or benefits do not have a market in which they are traded, alternative means must be

sought to reveal the preferences of individuals. These methods are the focus of this section.

These techniques rely on different assumptions and represent different ways of approaching

the preferences of individuals. This means that all the techniques cannot be used for

measuring any impact and it is necessary to consider each situation and clarify exactly what

is to be evaluated when choosing one or another technique. It may sometimes be necessary

to simultaneously use several techniques in order to each one picking a different aspect or

type of monetary value so we can then quantify the full impact.

These techniques are usually classified into two main groups: techniques based on related

markets and techniques based on hypothetical markets.

5.1. Techniques based on related markets

These techniques elicit the preferences of individuals for non-tradable goods and services

from the decisions that they take in the market on other goods and services that have some

kind of relationship (of substitutability or complementarity) with those non-tradable goods

and services.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures From this idea we conclude that this group of techniques can only monetarily quantify

those impacts resulting from a project on the use that individuals make of non-tradable

goods and services.

Within this group we can distinguish three types of methods or procedures: averting

behaviour method, the travel cost method and the hedonic price method.

5.1.1. Averting behaviour method

This method is based on the idea that individuals can compensate changes in the quantity or

quality of non-tradable goods or services through changes in the quantity or quality of

tradable goods and services, thereby maintaining their welfare or utility unchanged.

Following this idea, and assuming perfect substitutability and other additional hypotheses

(Maler 1974), the WTP of the individual for a marginal change in the quantity or quality of

the non-tradable goods or services affected by a project can be expressed as the marginal

change in the spending on the private goods or services.

Examples of this type of behaviour can be found on the expenses that households make to

better insulate their homes against external noise when there are increases in noise levels, or

expenses in buying bottled water, filtering tools and/or constructing private wells to offset

declines in the quality of supplied water.

Despite its conceptual attractiveness, this method presents a number of drawbacks. Firstly,

this procedure does not apply to non-marginal changes. The benefit of a non-marginal

change in the non-tradable good or service q is monetarily quantified according to this

procedure as the reduction in spending in the private good z that makes the individual retain

its level of utility. However, since they have more resources because of the improvement in

q, individuals will increase their consumption of goods and services, including in the good

z. Thus, the reduction in the spending in good z will be less than necessary to keep the

utility level constant and therefore we will underestimate the benefit of the change in q.

Secondly, this method assumes that individuals rapidly adjust to the change in q, when

really some time may be needed. Finally, an averting behaviour may not fully compensate

for a decline in q or lead to other benefits.

5.1.2. Travel cost method (TCM)

This method is based on analysing the relationship of complementarity between a non-

tradable good or service (in this case recreational, cultural, historical or scenic goods that

require a displacement of the individual for their enjoyment) and a private good (the travel).

This relationship of complementarity implies that the use of the non-marketed good (e.g.,


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresvisiting a natural landscape) requires the use of a private good (the trip), thus being able to

indirectly capture the preferences of individuals for the use of that non-tradable good or

service.

The idea behind this method is that the costs that people incur to visit a site can be

interpreted in some cases as a "price" for accessing this site and, therefore, can be used to

estimate the WTP of individuals for services provided by the site.

The TCM allows both the economic value of sites and the economic value of specific

characteristics, or changes in quality, of sites to be estimated.

5.1.2.1. Estimation of the economic value of recreational services of a site

For the first objective, the two most common variants of TCM are the zonal or aggregate

and the individual (Bateman 1993). The zonal TCM is to establish a relationship through a

regression between the rates of visits to a site from different geographical areas of origin

(usually expressed as number of visits per 1,000 inhabitants in the area for a period of time)

and the cost which is incurred by travelling from each zone of origin to the site in question.

The estimation of this function involves the implicit assumption that individuals living in

different areas have identical preferences regarding the site, so if they face the same

transport costs they would make the same number of visits. The individual TCM is

conceptually similar to the zonal one; but the relationship between number of visits (per

unit of time) and the cost of travel is set at the individual level. When calculating these

expressions, we may also include other explanatory variables (at the individual level in the

individual TCM and at the aggregate level in the zonal TCM) that could influence the

number of visits.

From these models of trip generation it is possible to get an approximation2 of the net

welfare derived from individuals visiting the site. To do so, we simply need to calculate the

areas below these functions.

In the case of the zonal variant, the area under the function between the cost of visits for a

zone and the cost for which the visitation rate is zero provides an approximation to the net

welfare for the individual of that zone derived from all his visits made to the site in the

period considered. For example, assuming a linear function relating to visitation rates (V/N)

with travel costs (C), this area would be represented for the zone 3 by the triangle C*AB in

Figure 1. This area reflects the difference between what the visits cost to the individual and

what they would be willing to pay as a maximum for them, making it a benefit for the

2 The values obtained are an approximation of the concept of economic value contained in Section 2 (Creel and Loomis 1991).


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures consumer. It would be the welfare that the individual would lose in case he cannot visit the

site. This amount is then divided by the average visits of that zone 3 and multiplied by the

total number of visits from that area to approximate the net welfare of all visits from zone

3. Proceeding similarly for all areas and summing the values obtained gives an

approximation of the net welfare derived from all visits in the period under examination.

Figure 1: Zonal MCV

Similarly, in the case of individual TCM we would calculate the area under the function

and the travel costs incurred by the individual to obtain a monetary approximation of the

net welfare that the individual gains from the trips made to the site. Several authors have

derived expressions that allow calculating the net welfare for different specifications of the

functional form (see, for example, Bockstael and Strand 1987). For example, in the case of

a linear function V = a + bC, the estimated value of the net welfare of the individual would

be 2

0( ( ))

2

V C

b , where C0 is the cost of travel for the individual. To obtain the net welfare

derived from all site visits over a period of time we should add for all visitors.

So far, monetary estimates of the welfare obtained with the individual and zonal TCM

correspond to the complete recreational experience. That is, the individual welfare derived

from travelling and the time on the site. To obtain the net welfare derived exclusively from

the stay at the site we need to calculate what would happen to the rate of visits or the

number of visits to the site with the establishment of different prices for entry to the site.

For this purpose we use the assumption that visitors react to increases in prices in the same

way they react to increases in travel costs according to the travel function previously

estimated. The result is a demand curve for the site which provides site visits according to

C*

A

C

V/N

Zone 1

Zone 5

Zone 3

Zone 4

Zone 2

B


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresdifferent entry prices. The area under this curve provides an approximation to net consumer

welfare derived from the experience in the site.

However, there are authors who have chosen different procedures to distinguish how much

of the welfare derived from the trip is attributable to the stay on the site. For example, one

option is to ask individuals directly how much of their enjoyment is assigned to the stay on

the site and to travel. This information can be used to adjust the travel costs or be

introduced directly as a variable in the function of travel.

From the above it follows that the application of individual and zonal TCM is not without

difficulties. These include, for example: calculating travel costs (and especially the

monetary quantification of the cost of time spent moving), the inclusion in the analysis of

alternative sites, the distribution of costs of travel when the individual visits more than one

site, or how to analyse together individuals performing visits to the site of varying

durations.

5.1.2.2. Estimation of the economic value of changes in the characteristics of a site

When the main objective is the economic valuation of changes in the characteristics of a

site, rather than leisure services provided by the site as a whole, the most widely used

approach of the TCM is based on random utility models (RUM). This variant (Haab and

McConnell 2002, Bockstael and McConnell 2007) is also most appropriate when there are

sites that may be considered as substitutes for the site that is intended to value.

This approach analyses the individual's choice of whether or not to make a trip and, if the

individual decides to make it, his decision on which site to visit. It is a probabilistic

approach that considers the probability that an individual visits a site depending on the

utility (welfare) he gets from visiting that site, which is also dependent on the costs of

travel and the characteristics of the site. The analysis of the RUM on which this approach is

based will be discussed in more detail in Section 5.2.2.

5.1.3. The hedonic price method (HPM)

The hedonic price theory was initially formulated by Rosen (1974), based on an alternative

to the neoclassical theory of consumer raised by Lancaster (1966), whereby a class of

differentiated products can be fully described on the basis of a series of objectively

measurable characteristics. Thus, goods and services are defined by a number of attributes

and characteristics, so that their prices reflect these differences.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures The hedonic model is frequently applied to various markets such as real estate, but it is also

possible to extend its use to others such as the labour or car markets, where the good has

significant features that influence the market price.

In the case of, say, a car, there is a basic model to which a number of features can be added.

Each option of features will involve an additional different payment, so it is easy to discern

what the price paid for each attribute is. However, when goods and services have a

dimension not captured by the market, it is difficult to determine the non-tradable attribute

price itself, as it is inserted into the total price. In this case, the observed prices, together

with the levels of various attributes, tradable and non-tradable, contained in each good or

service can help to obtain a measure of implicit value that consumers place on each

attribute that forms the good or service, including non-market attributes.

This approach is similar to the TCM explained above, since both are based on a

complementary relationship between a tradable good or service and another non-tradable.

However, there are differences in that the HPM operates through changes in prices of

private goods rather than through changes in their quantities (number of trips) as the TCM

does. In the HPM the private good or service is not acquired to enjoy the non-tradable one,

but this is a feature of the former.

The equilibrium relationship between the price of the good P and its characteristics vector Z

is called hedonic price function.3

( )P h Z

The partial derivative of this function with respect to any good (e.g. Zi,) gives its marginal

implicit price i

hZ , that is, the additional expenditure required to achieve a marginal change

in the feature. In a competitive market, this marginal implicit price will equal the

individual’s WTP for the marginal change in that feature (Freeman 1993).

Since not all individuals have the same preferences for Zi, the next step is to estimate a

function that explains how the marginal WTP varies with the level of the feature Zi, the

socioeconomic characteristics of individuals, and any other variables that might affect

preferences. From this function it is possible, under certain assumptions, to approximate the

economic value for an individual of non-marginal changes in the level of the feature Zi, by

simply calculating the area under this function, the horizontal axis and the two vertical lines

passing through the initial and final values of Zi. However, it should be noted that, given

3 To estimate the hedonic price function we can adopt various functional forms (Taylor 2003).


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresthe existent difficulties, most empirical studies do not perform this second stage (Pearce

and Turner 1990).

The HPM has several problems. On the one hand, from a theoretical point of view, its

validity rests largely on the assumptions of perfect information, perfect mobility of

consumers, and existence of equilibrium in the considered market. These assumptions are

rarely true in reality. On the other hand, from a practical perspective, the problems that may

arise are also numerous. For example, the omission in the hedonic price function of some

relevant variables may lead to biased estimates of the coefficients of the other variables

included, or the variables included in the analysis may present multi-collinearity problems

(Hanley and Splash 1993).

5.2. Techniques based on hypothetical markets

This Section deals with the group of techniques that elicit consumer preferences, and thus

their WTP or WTA, through hypothetical markets created by surveys. Thus, unlike

techniques based on related markets, the estimates are not derived from the observed

behaviour of individuals, but they are inferred from what would be their behaviour

according to their answers in a survey.

Since they are not based on an observed behaviour, and therefore linked to the use of the

good or service you want to value, these are the only techniques capable of capturing, in

addition to use values, the non-use values, i.e., the values that individuals assign to a good

or service even if they are not users. For example, an individual may express WTP for the

preservation of an animal species threatened with extinction despite not having any

intention of going to see it in its habitat.

Initially, these methods seem extremely simple; just to ask individuals questions that serve

to establish their WTP or WTA by a hypothetical change, and to assume that they will

answer exactly what they were asked. The problem arises precisely from this last

assumption, arguing that hypothetical questions tend to produce hypothetical answers.

Therefore, in this group of techniques the design of the survey is of great importance. Since

the questionnaire is the support that allows individuals' preferences about the evaluated

change to be elicited, we must design it so that people perceive the questions as real issues,

minimising as much as possible the existence of bias and strategic behaviour on the part of

respondents (Mitchell and Carson 1989). Usually the design of a good questionnaire

requires much time, being necessary to perform tests on small groups and small samples of

the general population before achieving a final version.

The questionnaire is usually structured in three parts:


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures a. The first part aims to introduce to the surveyed individual the good or service that

is intended to be assessed, and make them think about their preferences.

b. The central part is devoted to the valuation. In this part we must present to the

respondent in a clear and concise manner the choice he is going to make,

describing all those aspects that may affect his decision. Finally, the valuation

questions aimed to get his WTP or WTA by the proposed change are included.

c. Finally, the third part includes questions designed to obtain information about the

characteristics or attitudes of the respondents. The values obtained in the valuation

questions may be put in relation to these data to check its consistency. Moreover,

these questions can also be used to extrapolate the valuation results obtained from

the survey respondents to the population.

Methods based on hypothetical markets can be classified into two groups: the contingent

valuation method (CVM), with its various forms, and multi-attribute methods (Hanley et al.

2001). Both represent two different approaches of the valuation process. The first focuses

on the holistic nature of the goods or services to be assessed, while the latter pays more

attention to the attributes that define them. This different conception of the valuation

process is manifested mainly in the format of the questions of valuation and the type of data

collected.

5.2.1. Contingent valuation method (CVM)

In contingent valuation exercises individuals are offered a given change in exchange for a

certain amount of money, and the process of exchange that the individual makes between

the two is analysed.

The format of the valuation questions is very varied (Mitchell and Carson 1989). Perhaps

the most direct one, called open format, is to ask individuals directly their WTP or WTA

for the analysed change. These responses can be analysed by calculating the arithmetic

mean or simply estimating regression models that explain the stated WTP or WTA as a

function of other variables (e.g. socioeconomic). The main disadvantage of this approach is

that it places individuals in an unusual position. In reality, individuals have to decide

between a set of goods and services with well-defined prices. They rarely face situations

where they are asked to make an offer that can be accepted or rejected by the seller. As a

result, surveys using this format provide higher rates of non-response and high proportions

of implausible extreme values (high or low), and thus high variance in the valuations.

In the single-bounded referendum format, a subsample of individuals is asked if they would

make a payment for the provision of a public good, varying the amount of such payment

among the various subsamples. Thus, the possible answers are normally restricted to closed


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

urescategories “yes”, “no”, or “do not know/no answer”. From these responses it can be

obtained either by parametric techniques (typically logit or probit), semi-parametric or non-

parametric, an estimate of the probability function of the responses of individuals, from

which is possible to calculate the mean or median of the variable maximum WTP of

individuals as a measure of change in welfare. 4

The single-bounded referendum format has as advantages its simplicity and that it

resembles the kind of choices faced by individual in actual markets: the choice to accept or

not accept the provision of a good at a given price. Moreover, it is often considered that

with this format we obtain fewer “protest” and “don’t know” responses, and less dispersed

values than with the open format (Mitchell and Carson 1989). As Carson and Grove (2007)

point out, the single-bounded referendum format can be interpreted by individuals as a

referendum and therefore can be considered compatible with the incentive. This concept

refers to the theoretical property of this preference elicitation format that no individual can

improve, in any combination, acting strategically and reporting a value different from the

real value. All other elicitation formats do not possess this property, although this does not

necessarily mean that individuals realise they are responding to a format that is not

compatible with the incentive, and take advantage of it intentionally by providing answers

that are not those that correspond to their real preferences.

From these two basic formats (open and single-bounded referendum) we can derive other

formats that are just variations or combinations of both, seeking to extract more information

from the individual.

Thus, in the double-bounded referendum format the respondent is asked whether he would

pay a certain amount of money to acquire the good. If his answer is yes, then he is

prompted for an amount higher (lower, if answered no). Thus, regardless of the non-

answers, four types of answers can be obtained: yes–yes, no–no, yes–no, no–yes. The

problem with this approach is that the responses to the second payment may be influenced

by the proposed payment in the first question (called starting point bias). Moreover, as

DeShazo (2000) points out, the responses received to the first payment can sometimes be

inconsistent with the responses to the second.

The bidding game format continues this process up to a change from yes to no (or no to

yes). In this format the individual is asked if they would pay a certain amount for the good

or service in question. If the answer is affirmative, the question is repeated using a higher

price until the answer becomes negative. The highest price with an affirmative response is

interpreted as the maximum WTP. To obtain the minimum WTP, the same iterative process

would occur in reverse. As in the previous case, an objection frequently made to this type

4 The analysis of this type of answers is based on the RUM (see Hanemann and Kanninen 1999).


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures of format is the influence that the initial bid could have on the individuals’ valuations, as

this initial bid can be taken by respondents as a "shortcut" to get their decision and not to

reveal the true value. Another disadvantage is that, although the open questions can be

made by mail or in person, this type of question can only be made face-to-face or by

computer. However, the responses provided by this format often have a low deviation

around the mean in relation to the open format, as well as a smaller number of non-

responses.

Another format, sometimes used, is the mixed-format, which consists of making two

consecutive questions: the first closed and the second open. That is, there is a starting "hint"

that is responded with an acceptance or rejection of the suggested bid, and then we ask for

the maximum WTP. This format shares some problems with the bidding game format

because the bid is given as a reference which can influence the responses, and it also shares

some advantages of the referendum format over the open format.

5.2.2. Models based on multi‐attribute choices

In this group of techniques, individuals have to express their preferences on sets of

alternatives defined by attributes that vary at different levels.

The attributes that define the alternatives in this group of techniques are motivated by the

aim of the study. The researcher has to include as attributes the most important elements

that may be considered by individuals as they make their decisions in the area studied

(Alpert 1971). It is precisely the changes in the values of these attributes which are intended

to be valued by these techniques. It is therefore necessary that one of these attributes

represents a monetary payment. The number of attributes and levels cannot be too high to

ensure the respondents to not have too much difficulty in assimilating the information

provided. Successive pre-tests and pilots are required to design the survey to verify that the

population considers that included attributes and their levels are relevant and easily

understood.

Once the attributes and the levels are selected, they are combined to obtain the different

alternatives. The simplest method is called full factorial design, which is to generate all

possible alternatives from the levels of the attributes considered. Subsequently, these

obtained alternatives are randomly grouped into choice sets, verifying that in the choice sets

that do not coincide alternatives that may be better or worse in all attributes compared to

the other alternatives of the choice set. This type of design allows estimations of both the

main effects and the interactions. A main effect is the direct effect of an attribute

considered individually. In a main effects model, the effect of an attribute would be the

same regardless of the other attributes. The interactions involve two or more attributes. In a

model with interactions, the effect of an attribute is different for different levels of another


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresattribute. With a full factorial design, all main effects, all interactions of second order and

higher order can be estimated and are uncorrelated.

The disadvantage of a full factorial design is that such designs are only applicable to small

designs that use either a low number of attributes, or levels, or both, which is not the case

for most applications. The reason is that, except in those cases, the number of alternatives

generated by the full factorial design is very high and, therefore, the monetary cost and the

risk of fatigue of individuals to respond to the survey is also high. For example, if we

consider a design with five four-level attributes, we will obtain (45) 1,024 different

alternatives.

This problem is solved using fractional factorial designs. These designs imply the selection

through a specific procedure of a particular sample or set of the full factorial design

alternatives, so that the intended effects can be estimated by the most efficient way

possible. The downside is that all these designs generally involve some loss of information,

assuming that the effects of interactions between two or more attributes are zero or

insignificant.

Once the choice sets have been developed, one or more of these sets of alternatives are

presented to individuals to express their preferences. The kind of choice task individuals

have to perform on the set of alternatives presented depends on the specific elicitation

method employed. The most common are the contingent choice (Louviere et al. 2000) and

the contingent ranking (Chapman and Staelin 1982, Hausman and Ruud 1987, Ben-Akiva

et al. 1991, Foster and Mourato 2002).

In a choice experiment, the individual must choose his most preferred alternative from a

choice set, which must include an alternative representing the current situation or status

quo. This format of choice is easy to answer for individuals since it reminds them, to some

extent, of the kind of tasks they must perform in real markets. However, from a researcher’s

point of view, this variant is the one that provides less information for each individual and

choice set. In an exercise of contingent ranking, individuals have to rank all alternatives

included in the choice set according to their preferences. This format provides more

information than choice experiments, but it raises doubts about the ability of individuals to

provide reliable answers when the number of alternatives included in the choice set is high,

when individuals have similar preferences on several alternatives, or when we include in

the choice set alternatives that individuals would never choose or do not know properly

(Louviere et al. 2000). For these reasons, some authors do not recommend the use of all the

information derived from a complete ranking of the individual of all alternatives included in

the choice set (Chapman and Staelin 1982, Ben-Akiva et al. 1991, Louviere et al. 2000).


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures The analysis of the responses provided by individuals in both cases is based on the RUM. It

is assumed that individuals receive a utility from each alternative and they select those

alternatives that provide greater utility. The utility Uni derived by an individual n from an

alternative i is divided into a deterministic component Vni, which captures the influence of

the levels of the attributes present in that alternative, and a random component ni . This

stochastic component contains all those factors that are unobservable or unknown to the

researcher. For example, assuming that the deterministic component takes a linear form, we

obtain:

'nj n nj njU x ,

where njx is a vector containing the levels of attributes for alternative j presented to

individual n, and βi is a vector of parameters that represent the contribution of each

attribute to the utility perceived by the individual. The objective is to estimate these

parameters of the utility function.

According to the assumptions to be made on the distributions of these stochastic

components and their relationship to the random components of the other alternatives

included in the choice set, the applied model for estimating the parameters of the utility

function will change (Louviere et al. 2000). For example, if one assumes that the random

components are independent and identically distributed according to an extreme value type

I distribution, it is obtained in the choice experiment a multinomial logit model and, in the

contingent ranking, an ordered logit model.

Once the parameters have been estimated, it is also possible to estimate the WTP of

individuals to marginal changes in the levels of attributes. We only have to estimate the

trade-off in the utility function between the attribute of interest and the payment attribute.

For example, in the case discussed above of a linear utility function, the WTP for a

marginal change of an attribute is the negative of the ratio between the coefficient for that

attribute in the utility function and the coefficient for the payment attribute.

In the event that we want to value non-marginal changes of one or more attributes

simultaneously, a simplified approach is to assume that the estimated marginal values are

constant for all units and attributes, and aggregate those values in a linear fashion (Hanley

et al. 1998). More correct and formal procedures can be found in Hanemann (1982).

6. CONCLUDING REMARKS

The economic valuation methods play a key role in the cost-benefit analysis since they

allow the direct introduction into the analysis of those impacts caused by a project and not


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

urescaptured by the market. However, the application of these techniques makes it necessary to

resort to various assumptions and simplifications. Therefore, the resulting monetary

estimates of the impacts must not be taken as an exact value, and we have to be cautious at

the time of using these estimates in the study.

This greater caution may be accomplished, for example, by providing confidence intervals

or probability distributions of the estimates obtained. Thus, it would be possible to perform

a sensitivity analysis of the influence of these estimates in the decision on the suitability of

undertaking a project. Another option would be to calculate what should be the economic

value of these impacts so that the net present value or another indicator of the suitability of

a project was null, and then compare this value with the estimates of these impacts obtained

by techniques of economic valuation.

In (Brey 2010) there are collected quantifications in monetary terms, obtained in various

studies, of non-market impacts associated with transportation projects. As discussed above,

these estimates should not be understood as unitary monetary values that can be directly

translated to any study, but merely as an illustration of the importance that such impacts can

achieve.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresREFERENCES

Alpert, M. (1971). Identification of determinant attributes: a comparison of methods,

Journal of Marketing Research, VIII, 184-91.

Bateman, I. (1993). Valuation of the environment, methods and techniques: revealed

preference methods, in R. K. Turner (ed.), Sustainable environmental economics and

management, Belhaven Press, London.

Ben-Akiva, M., T. Morikawa, and F. Shiroishi (1991). Analysis of the reliability of

preference ranking data, Journal of Business Research, 24, 149-164.

Bockstael N. E. and I.E. Strand (1987). The effect of common sources of regression error

on benefit estimates, Land Economics, 63 (1), 11-20.

Bockstael N.E. and K. E. McConnell (2007). Environmental and resource valuation with

revealed preferences. A theoretical guide to empirical models, Springer, Dordrecht.

Brey, R. (2010): Economic valuation of externalities linked to transport projects. Spain. Available at: http://evaluaciondeproyectos.es/SpanishWeb/Resultados/DocTrab.html

Carson, R. T., and T. Grove (2007). Incentive and informational properties of preference

questions, Environmental and Resource Economics, 37, 181-210.

Chapman, R. G., and R. Staelin (1982). Exploiting rank ordered choice set data within the

stochastic utility model, Journal of Marketing Research, 19, 288-301.

Creel M. D. and J. B. Loomis (1991). Confidence intervals for welfare measures with

application to a problem of truncated counts. The Review of Economics and Statistics,

73 (2), 370-373.

DeShazo, J. R. (2000). Designing transactions without framing effects in iterative question

formats, Mimeograph, School of Public Policy and Social Research, University of

California, Los Angeles.

Foster, V., and S. Mourato (2002). Testing for consistency in contingent ranking

experiments, Journal of Environmental Economics and Management, 44, 309-328.

Freeman, A. (1993). The measurement of environmental and resource values: theory and

methods. Resources for the future, Washington D. C.

Haab T. C. and K. E. McConnell (2002). Valuing environmental and natural resources: the

econometrics of nonmarket valuation, Edward Elgar, Northampton, MA.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

ures Hanemann, W. M. (1982). Applied welfare analysis with qualitative response models,

CUDARE Working Papers, nº 241, University of California, Berkeley.

Hanemann, W. M. (1991). Willingness to pay and willingness to accept: how much can

they differ? American Economic Review, 81 (3), 635-647.

Hanemann, W. M. and B. Kanninen (1999). The statistical analysis of discrete-response CV

data, in I. Bateman and K. Willis (eds.), Valuing environmental preferences: theory and

practice of the contingent valuation method in the US, EC, and developing countries,

Oxford University Press, Oxford.

Hanley, N. (1988). Using contingent valuation to value environmental improvements,

Applied Economics, 20, 541-549.

Hanley, N., R. E. Wright and V. Adamowicz (1998). Using choice experiments to value the

environment. Design issues, current experience and future prospects, Environmental

and Resource Economics 11(3-4), 413-428.

Hanley, N. and C. Splash (1993). Cost-benefit analysis and the environment, Edward Elgar

Publishing Limited, Aldershot.

Hanley, N., S. Mourato, and R. E. Wright (2001). Choice modelling approaches: a superior

alternative for environmental valuation? Journal of Economic Surveys, 15(3), 435-462.

Hausman, J. A., and P.A. Ruud (1987). Specifying and testing econometrics models for

rank-ordered data, Journal of Econometrics, 34, 83-104.

IUCN, International Union for the Conservation of Nature and Natural Resources (1980).

World conservation strategy, IUCN, Ginebra.

Kahneman D. and A. Tversky (1979). Prospect theory: an analysis of decisions under risk.

Econometrica, 47 (2), 263-291.

Lancaster, K. J. (1966). A new approach to consumer theory, Journal of Political Economy,

74, 132-57.

Louviere, J. J., D. A. Hensher, and J. D. Swait (2000). Stated choice methods: analysis and

applications, Cambridge University Press, Cambridge.

Mäler, K-G. (1974). Environmental economics: a theoretical enquiry, The Johns Hopkins

University Press, Baltimore.


Econ

omic

val

uati

on o

f ex

tern

alit

ies

linke

d to

tra

nspo

rt p

roje

cts:

Fou

nda

tion

s an

d pr

oced

uresMitchell, R. C., and R. T. Carson (1989). Using surveys to value public goods: the

contingent valuation method, Resources for the Future, Washington D. C.

Pearce, D. and R. Turner, R. (1990). Economics of natural resources and the environment,

Harvester Wheatsheaf Ed Hertfordshire.

Rosen, S. (1974). Hedonic prices and implicit markets: product differentiation in pure

competition, Journal of Political Economy, 82, 34-55.

Taylor L. (2003). The hedonic method, in K.J. Boyle and T.C. Brown (eds.), A primer on

nonmarket valuation, Kluwer Academic Publishing, London.

Documents

Economic valuation of externalities linked to transport ... · ECONOMIC EVALUATION OF TRANSPORTATION PROJECTS 1 November 2009 1 Economic valuation of externa lities linked to transport