Descomposing Agricultural Multifunctionality

Journal of Agricultural Economics, Vol. 58, No.
2, 2007, 218–241
Decomposing the Value of Agricultural

Multifunctionality: Combining
Contingent Valuation and the
Analytical Hierarchy Process
Zein Kallas, José A. Gómez-Limón and Jesús Barreiro Hurlé1
(Original submitted April 2006, revision received October 2006, accepted October
2006.)
Abstract
Agricultural multifunctionality is the recognition of the joint exercise of econo-
mic, environmental and social functions by this sector. Nevertheless, not all these
contributions to society are valued in markets, moreover a large share of them
are public goods. For this reason, in order to make this concept of multifunction-
ality operative for the design of public policies, it is necessary to estimate the
social demand of such functions. The objective of this article was to implement
an empirical application along these lines. For this purpose, the agricultural sys-
tem of cereal steppes in Tierra de Campos in Spain is taken as a case study.
The economic valuation technique used relies on a combined implementation of
contingent valuation and the analytical hierarchy process. The results obtained
demonstrate the existence of a significant demand for the different attributes
included in the multifunctionality concept, although this demand is heterogeneous
and is based on the socioeconomic characteristics of individual persons.
Keywords: Agricultural multifunctionality; analytical hierarchy process; contin-

gent valuation; Castilla y León; economic valuation; Spain.
JEL classifications: Q18, Q11, Q25.
1
Jesús Barreiro Hurlé, Área de Economı́a y Sociologı́a Agrarias (AESA), Instituto de Investi-
gación y Formación Agraria y Pesquera (IFAPA), PO Box 2027, 18080 Granada, Spain.
E-mail: [email protected] for correspondence. Zein Kallas is at
CREDA-UPC-IRTA, Universitat Politécnica de Catalunya. José A. Gómez-Limón is at the
Agricultural Economics Department, E. T. S. II. AA. Palencia, Universidad de Valladolid.
This study was partly financed by the Spanish Commission for Science and Technology (CI-
CYT) through the MULTIAGRO project (AGL2003-07446-C03-01), by the Spanish National
Institute for Agricultural Research (INIA) through the DISOPTIPOL project (RTA2005-
000020) and by the Education Department of the Regional Government of Castilla y León
through project VA006A05. The authors wish to thank two anonymous reviewers and JAE’s
Associate Editor and Editor-in-Chief for their constructive comments, which have improved
the quality of the paper. The usual disclaimer applies.
Blackwell Publishing Ltd. 2007, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Steet, Malden, MA 02148, USA.
Decomposing the Value of Agricultural Multifunctionality 219
1. Introduction and Objectives

The past decade has seen an increase in both political and academic interest in
multifunctionality in the agricultural sector. The use of this concept by the Euro-
pean Council of Agricultural Ministries in December 1997 fuelled the expansion of
studies dealing with this new notion of agriculture (see also EC, 1998, 1999, 2000).
The term ‘multifunctionality’ was formally defined for the first time by the Euro-
pean Commission (EC, 1999), as the recognition of the three different roles played
by agriculture in the EU:
(i) producing food and fibre products;
(ii) preserving the rural environment and landscape; and
(iii) contributing to the viability of rural areas and a balanced territorial develop-
ment. From this definition, it can be seen that multifunctional agricultural
production comprises both market and non-market goods (commodities and
non-commodities). The former comprise mainly, although not exclusively, food
and fibre products, while the latter include environmental and social functions,
which in most cases also have public good characteristics.
Following the definition of this concept, broadening and deepening its under-
standing has been the goal of several research agendas. However, these have
focused to date on theoretical and qualitative approaches to multifunctionality
(see, e.g. Cahill, 2001; Peterson et al., 2002; Batie, 2003; Brouwer, 2004; Harvey,
2003; Prety, 2003; Van Huylenbroeck and Durand, 2003). The Organisation for
Economic Co-operation and Development (OECD) has been important in the
development of this agenda and has commissioned several studies on the topic,
including: an in-depth analysis of the economic concepts, summarised in OECD
(2000) and an international review of the existing literature OECD (2001). This
review identified a lack of empirical quantitative applications capable of measur-
ing multifunctionality, a fact that prevents reaching solid conclusions regarding
public intervention for promoting agriculture as a multifunctional activity
(OECD, 2003).
Our study aims to overcome this handicap, focusing on the analysis of social
demand for agricultural multifunctionality. Many valuation studies have looked
into a particular value of agricultural multifunctionality (e.g. landscape, non-point
source pollution, erosion, food safety, etc.) which Lima Santos (2001) reviewed.
However, the valuation of ‘‘complex’’ goods, such as multifunctional agriculture,
should involve the joint valuation of the whole set of their attributes (Hoehn and
Randall, 1987, 1989). Otherwise, we run the real risk of part-whole bias (Mitchell
and Carson, 1989; Bateman et al., 1997; Randall, 2002). Despite this, studies
adopting this methodological approach have been scarce (e.g. Bennett et al.,
2004).
This paper has a double objective. First, we estimate the social demand for
multifunctional outputs of a particular agricultural ecosystem. Second, we analyse
the demand, identifying significant relationships between both aggregated and par-
tial values declared for each attribute and individual socio-demographic charac-
teristics.
This advances the existing literature in two ways. First, we estimate an aggregate
monetary value for the multifunctional goods and services provided by a given agri-
cultural ecosystem (an estimate we have not found in the literature reviewed) and
Blackwell Publishing Ltd. 2007
220 Zein Kallas, Jose´ A. Gómez-Limón and Jesús Barreiro Hurle´
partial values for each of the attributes. Second, we propose the combination of
two valuation techniques: contingent valuation (CV) and the analytical hierarchy
process (AHP), as alternatives to choice experiments (CE) (Bennett et al., 2004;
Yrjölä and Kola, 2004 or Colombo et al., 2005). The case study area is the agricul-
tural ecosystem of Tierra de Campos in Castilla y León (an Autonomous Region in
northern Spain). Although, our results cannot be directly transferred to other agri-
cultural ecosystems, the case study is relevant for two reasons. The selected eco-
system is representative of Mediterranean dry-land farming in marginal areas
(common in inland Spain, Portugal, Italy and Greece), where multifunctionality is a
key policy goal. Second, our results could help future studies develop benefit trans-
fers and also demonstrate alternative valuation techniques for complex multifunc-
tional goods.
The paper is structured as follows. Following this introductory section, section 2
describes the agricultural ecosystem of Tierra de Campos and its multifunctional
aspects. We then present and justify the methodological approach selected to esti-
mate the demand for multifunctionality: a combination of CV and AHP. Section 4
describes the main features of the application, followed by a presentation of the
results obtained. We conclude and offer insights for further research.
2. Case Study
2.1. Agriculture in the Tierra de Campos ecosystem
Pseudo-steppes are ecosystems characterised by scarce vegetation, an almost total
absence of trees, a plain relief and average rainfall below 450 mm (Suárez et al.,
1997). An important part of Castilla y León can be classified as pseudo-steppe, an
ecosystem associated with dryland extensive culture, cereals in particular, and they
are therefore referred to as ‘‘cereal steppes’’.
The focus is on the ‘‘cereal steppe’’ area in Castilla y León known as ‘Tierra de
Campos’; 1 million hectares (comprising two-thirds of the total area of steppe in
the region), located to the northeast of central Castilla y León, and including 267
municipalities in four provinces. The area shares common ecological, political,
social and economic characteristics, and can be classified as predominantly rural.
According to the latest available data (1999 Agricultural Census; INE, 2001), uti-
lised agricultural area accounts for 83.8% of the territory, and is mainly represen-
tative of marginal extensive agriculture, where the social and environmental
functions are often supposed to have greater importance than food and fibre
production.
2.2. Agricultural multifunctionality in Tierra de Campos

2.2.1. Economic functions
In 2003, the agricultural sector in Castilla y León generated a gross added value of
over 2.3 billion euro, representing 7.5% of the regional economy (INE, 2004),
nearly double the national average for Spain (4.1%) and nearly five times the
EU-25 average (1.6%) (DG AGRI, 2006). Although still limited, the impact of agri-
culture in the overall economic activity is relevant due to the dual territorial struc-
ture of the region. While urban areas concentrate on industrial activity and services,
rural areas depend to a great extent on agricultural activity.
2.2.2. Social functions

Social functions provided by agriculture are primarily related to employment oppor-
tunities generated in rural areas. Agriculture generates 12,589 AWU2 in this area
(1999 Agricultural Census; INE, 2001), or 28.5% of total employment, more than
three times as high as the regional average (9.0%), over five times the national aver-
age (5.6%) and the EU-25 average (5.0%) (all figures for 2005). This employment
demand, together with the structure of agricultural holdings, which are mostly fam-
ily owned and managed (average area 48.2 ha. and with 0.68 AWU per holding),
highlights the role agriculture plays in maintaining families in small communities,
and the social networks vital to the continued existence of such communities.
One of the main problems facing these areas is that of ‘‘rural exodus’’ due to
diminishing agricultural activity (Franco and Manero, 2002). This has led to a very
low population density in the area (11.2 inhabitants/km2), again lower than regio-
nal, national or European averages. This low population density generates import-
ant diseconomies of scale for the provision of public and private services, leading
to the progressive disappearance of the region’s historical and cultural heritage
(Camarero, 1993).
2.2.3. Environmental functions

Most of Tierra de Campos’ territory (77%) is devoted to dry land extensive farming.
This agricultural ecosystem is characterised by low yields per hectare but high yields
per AWU and in intensive labour use in comparison with the other two production
factors, land and capital (Pérez, 2001). The maintenance of agricultural activity is
vital to the conservation of the pseudo-steppe ecosystem and the floral and faunal
communities associated with it. These communities include 21 species which are
either endangered or threatened with extinction. Birds, mainly steppe birds, are the
most important category of these species (Suárez et al., 1997).
The role of agriculture in biodiversity conservation has been recognised by the
environmental managing authority, which has identified and declared 15 Sites of
Community Interest and Zones of Special Protection of Birds comprising
221,475 ha during the implementation of the Natura 2000 network. The study area
is a good example of marginal areas in which agriculture plays an important role in
the conservation of the natural heritage (EEA, 2004).
However, agriculture in the area also has negative impacts caused mainly by the
expansion of irrigation (8.5% of the total surface is irrigated). This leads to the dis-
appearance of arid habitats (with the loss of refuge areas and food sources for
steppe birds) and an increase in environmental pollutants (fertilizers and pesticides).
3. Methodology
Valuation of non-marketed multifunctional goods and services can be undertaken
using various methodologies among which are:
(i) opinion polls and consumer surveys;
(ii) proxy analysis of public preferences;
2
Annual working unit, equivalent to full-time equivalent employment (corresponding to the
number of full-time equivalent jobs).

(iii) deliberative methods;

(iv) monetary valuation; and
(v) multicriteria analysis techniques.
All these possess advantages and disadvantages (Hall et al., 2004). Here we com-
bine two approaches in a novel manner to obtain both an aggregated value for
multifunctional agriculture as a whole, and partial values for each of the individual
functions identified.3 CV is used to obtain an aggregated value for agricultural
multifunctionality, while the AHP is used as a multicriteria technique to obtain par-
tial values for each of the separate functions.
The theoretical model is as follows. CV assumes that consumers derive utility
from two goods, agricultural multifunctionality (MF) and all other goods treated as
a composite good with price one (Y). Our valuation question elicits maximum WTP
consistent with the following indifference situation:
UðY WTPDMF ; MFimp Þ ¼ UðY; MFref Þ; ð1Þ
where WTPDMF is the willingness to pay for a concrete improvement in the multi-
functional character of the ecosystem, MFimp denotes the supply level of agricul-
tural multifunctionality in the improved situation and MFref is the actual level or
status quo. Rearranging the previous expression, the monetary equivalent of the
multifunctional improvement, can be expressed as:
WTPDMF ¼ UðMFimp Þ UðMFref Þ: ð2Þ
On the other hand, multifunctional utility can be expressed as:
UðMFÞ ¼ UðF1 ; F2 ; F3 ; . . .::; Fn Þ; ð3Þ
where Fi is function i. Considering two functions for simplicity:
WTPDMF ¼ UðF1imp ; F2imp Þ UðF1ref ; F2ref Þ: ð4Þ
However, individual values cannot be assigned to Fi without further assumptions
about the shape of the utility function U. Multicriteria analysis (in particular AHP)
helps. Assuming a linear utility function specification (see below), AHP allows us to
estimate wi for each function:
UðMFÞ ¼ w1 F1 þ w2 F2 : ð5Þ
Combining equations (4) and (5) we obtain:
WTPDMF ¼ ðw1 F1imp þ w2 F2imp Þ ðw1 F1ref þ w2 F2ref Þ

ð6Þ
¼ w1 ðF1imp F1ref Þ þ w2 ðF2imp F2ref Þ ¼ w1 FD1 þ w2 FD2 :
Hence, the WTP for an individual function of multifunctionality is as follows:
WTPDFi ¼ wi WTPDMF : ð7Þ
3
Other authors have used CV and AHP in combination before (Smith and Lantz, 2003;
McVittie et al., 2004; Wattage and Mardle, 2004; Zoppi, 2004). However, our study is an ori-
ginal application to agricultural multifunctionality.

3.1. Contingent valuation

Contingent valuation is one of the most frequently used environmental monetary
valuation techniques, in particular when no relationship can be identified between
a non-market good and a market good, enabling indirect values to be obtained
(Mitchell and Carson, 1989; Arrow et al., 1993 or Carson, 2006).
3.1.1. Contingent market design

Construction of a hypothetical market for agricultural multifunctionality requires a
clear definition of the functions, and a clear and precise presentation to surveyed
interviewees. We relied on prior research in the same agricultural ecosystem to iden-
tify the functions expected of agriculture by local society (Gómez-Limón and
Atance, 2004). This research suggests that ranking of social preferences is: mainten-
ance of rural villages (weighting 20.6% of total utility); favouring environmentally
friendly agricultural practices (18.0%); assuring safe and healthy food (16.1%).4
These functions are taken as the three most important functions of agriculture in
the study area (social, environmental and economic, respectively).
The proposed functions were discussed in three different focus groups: university
lecturers in the field of agricultural economics; managers from the public sector and
potential interviewees.5 All agreed that these functions were the most important,
but suggested the convenience of splitting the first into two, one related to the eco-
nomic viability of rural environments, and the other one linked with the mainten-
ance of populations in villages. These focus groups also identified proxy variables
to be used to measure these functions:6
1 Contribution to economic activity in rural areas (proxy variable: number of jobs
in the agricultural sector).
2 Maintenance of population in villages (proxy variable: percentage of farmers who
have their permanent residence in the same municipality as their holding is
located).
3 Use of environmentally friendly agricultural practices (proxy variable: number of
endangered species of fauna and flora in the area7).
4
As pointed out by Gómez-Limón and Atance (2004), these social preferences ‘would corres-
pond to the objective implicit in the concept of agricultural multifunctionality; that of guar-
anteeing an appropriate level of public goods provided by agriculture, in their productive,
social and environmental dimensions’.
5
This particular focus group was made up of leaders of local society (unions, cultural associ-
ations and neighbourhood communities).
6
During focus group sessions, the most controversial attribute was the contribution of agri-
culture to economic activity in rural areas. The adoption of the proxy variable ‘‘jobs gener-
ated by the agricultural sector’’ is based on its higher acceptance by the public as a clear
economic indicator rather than, say, agricultural total output or relative importance of the
agriculture in the whole economy. This assumption was confirmed by the pilot survey. How-
ever, interpretation of this variable is limited to the economic relevance of agriculture rather
than its social implications.
7
A decrease in the endangered species present in the area implies that the number of individuals
of these species grows to an extent where the species are no longer endangered: an improvement
of the ‘‘environmental factor’’ of agricultural multifunctionality in the area under study.

4 Health and food safety concerns (proxy variable: type of farm management: con-
ventional, integrated or organic).
The functions of multifunctionality here are: environmental (3), social (2) and
two economic (1 and 4), which are the attributes of our ‘‘agricultural multifunc-
tional package’’. Considering the historical records of the proxies, the same focus
groups met again to identify the most suitable combinations to define the improve-
ments in the multifunctional character of the agricultural ecosystem. Two options
were selected: one labelled ‘‘moderate’’ and another ‘‘significant’’. The characteris-
tics of these two improvements are presented in Table 1, which identifies the levels
attached to each attribute in each of the scenarios.
3.1.2. Contingent market design: question format and payment vehicle

Several formats to elicit WTP have been used in CV and there is no consensus
about superiority (Bateman et al., 2002). We use a ‘‘payment card’’ format, as it
combines both the advantages of open-ended formats (elicitation of point informa-
tion of WTP) and closed formats (ease of cognitive burden on interviewees) whereas
minimising the risk of ‘‘starting price bias’’ is associated with iterative bidding pro-
cesses. The prices were chosen using information from the focus groups and the
Table 1
Contingent valuation questionnaire using the ‘‘payment card’’ question format and
a compulsory increase in taxes as payment vehicle
Functions carried out by agriculture Current situation “Moderate” “Significant”

in Tierra de Campos (status quo) improvement improvement
Contribution to economic activity 12,600 14,000 16,000

through job generation in rural areas. agricultural jobs agricultural jobs agricultural jobs
(JOB0) (JOB1) (JOB2)
Maintenance of population in rural

areas and cultural heritage through 70% farmers 80% farmers 90% farmers
increase of number of farmers with with permanent with permanent with permanent
permanent residence in the residence in residence in residence in
municipalities where their holding is villages villages villages
located. (RES0) (RES1) (RES2)
Use of environmentally friendly

agricultural practices reducing the 21 endangered flora 15 endangered flora 9 endangered flora
number of endangered flora and fauna and fauna species and fauna species and fauna species
species. (END0) (END1) (END2)
Food produced by Food produced by Food produced by

Contribution to production of healthier
conventional integrated organic
and safer food products changing farm
management. farming farming farming
(MAN0) (MAN1) (MAN2)
Achieving these improvements implies an additional cost to society which must be funded by an increase in taxes. The
increase of taxes per capita was to be located among the figures in the following card:
0 5 10 15 20 25 30 35 40 45 50 >50
€/year €/year €/year €/year €/year €/year €/year €/year €/year €/year €/year €/year
My willingness to pay to achieve the “moderate” improvement is a maximum of: €/year
My willingness to pay to achieve the “significant” improvement is a maximum of: €/year

pilot survey (which used an open-ended format) to cover the central 90% of
observed distribution of WTP in the pilot (Cooper, 1993; Kaninen and Kriström,
1993). A compulsory payment vehicle, an increase in taxes,8 was selected to mini-
mise strategic bias associated with voluntary payments to conservation funds. This
format was used for both levels of improvement, and was presented in the question-
naire to interviewees as given in Table 1.
We obtained two values for the aggregate multifunctional bundle, WTPMOD, for
moderate improvement and WTPSIG, for significant improvement.
3.2. Analytical hierarchy process

As separately valuing each multifunction through an individual CV exercise could
lead to seriously biased estimations, due to series of ‘‘instrumental biases’’ (see sec-
tion 3.4), we used the alternative AHP as a complementary technique to CV.
Analytical hierarchy process was developed by Saaty in the late 1970s (Saaty,
1977, 1980) as a technique to support multicriteria decision-making. Here, AHP is
used to decompose the aggregated value of agricultural multifunctionality into its
particular attributes (agricultural functions).9 Following Saaty (2001), we use AHP
to elicit weights for each of the attributes (functions) of agricultural multifunctional-
ity, as indicators of their relative importance to society.
The AHP hierarchical structure used here is shown in Figure 1, where the selected
attributes define the good to be valued.
The relative attribute weights (wi) are obtained from paired comparisons. To
make these comparisons and to determine the intensity of preferences for each
option, Saaty (1980) argues for a 9-point scale, which can be easily understood by
most interviewees. From the answers provided to the question of comparing each
pair of attributes, the following matrix is generated for each individual (k) (Saaty
matrix):
Agricultural multifunctionality
Attribute 1 Attribute 2 Attribute 4

Attribute 3
(JOB) (RES) (MAN)
(END)
Maintenance of Contribution to
Contribution to Use of environ-
population in rural production of
economic activity mentally friendly
areas and cultural healthier and
in rural areas agricultural practices
heritage safer food products
Figure 1. Hierarchical structure used to value agricultural multifunctionality
8
People surveyed were informed that a new levy on income tax would be implemented to col-
lect revenues to be devoted to the improvement of local public goods provided by agriculture.
This levy would be fixed specifically for each territory by local authorities, and would be col-
lected from all adult individuals (above 18 years old) living in the corresponding area.
9
Saaty (1980) or Golden et al. (1989) provide a detailed description of this methodology.

2 3
a11k a12k ... a1nk
6 a21k a22k ... a2nk 7
Ak ¼ 6
4 ...
7; ð8Þ
... aijk ... 5
an1k an2k ... annk
where aijk represents the value obtained from the comparison between attribute i
and attribute j for each individual (k). For consistent preferences, this matrix has
two fundamental properties:
(i) all elements of its main diagonal area take a value of 1 (aiik ¼ 1 "i); and (ii) all
other elements maintain that paired comparisons are reciprocal (if aijk ¼ x then
ajik ¼ 1/x).
In addition, aihk · ahjk ¼ aijk for all i, j and h, because for a perfectly rational
decision-maker aijk ¼ wik/wjk for all i and j. As a result, the Saaty matrix can also
be expressed as follows:
2 w1k w1k w1k 3
w1k w2k . . . wnk
6 w2k w2k w2k 7
6 w1k w2k . . . wnk 7
6 7 ð9Þ
6 . . . . . . wik . . . 7:
4 wjk 5
wnk wnk . . . wnk
w1k w2k wnk
Thus, if perfect consistency holds, n weights (wik) for each attribute can be deter-
mined from the n(n ) 1)/2 values for aijk elicited in the survey. Unfortunately, per-
fect consistency is seldom present in reality, where personal subjectivity plays an
important role. For Saaty matrices (Ak ¼ aijk) in which some degree of inconsis-
tency is present, alternative approaches have been proposed to estimate the weight
vector that best reflects the decision-maker’s preferences. Saaty (1980, 2003) pro-
posed two alternatives: the geometric mean and the main eigenvector. Other authors
have proposed alternatives based on regression analysis (Laininen and Hämäläinen,
2003) or goal programming (Bryson, 1995). No consensus has been reached regard-
ing any alternative that outperforms the other (Fichtner, 1986). As all criteria meet
the requirements, we chose the operationally simplest alternative, the geometric
mean (Aguarón and Moreno-Jiménez, 2000). Particular weights assigned by individ-
uals to each attribute are estimated as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffi
u i¼n
u Y
wik ¼ t aijk 8i; k:
n
ð10Þ
i¼1
The AHP has been extended as a valid technique for the analysis of group decisions
(Easley et al., 2000). The steps needed to obtain an estimate of the population’s
weights are as follows: (i) choice of a representative sample of the society; (ii) eli-
citation of paired comparisons and individual Saaty matrices for the sample (Ak ¼
aijk) and estimation of corresponding weights (wik); and (iii) aggregation and synthe-
sis of weights for society as a whole (wi).

We follow Forman and Peniwati (1998), in aggregating individual weights (wik)

with the geometric mean:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uk¼m
u Y
wi ¼ t
m
wik 8i: ð11Þ
k¼1
Zahedi (1987) has demonstrated that resolving a multicriteria problem through

AHP can be considered to be equivalent to the optimisation of a multiattribute util-
ity function. Traditionally, AHP has been associated with an additive utility func-
tion (Kamenetzky, 1982). Thus, the decision-maker’s utility associated with each
alternative (j) can be calculated using the following expression:
X
i¼n
Uðxj Þ ¼ wi Ui ðxj Þ 8j; ð12Þ
i¼1
where U(xj) is the total utility that alternative j provides to the decision-maker (or
group of decision-makers, as is our case), wi represents the attribute weights and
Ui(xj) the partial utilities provided by attribute i in alternative j.
3.3. Combining CV and AHP data

Assuming an additive utility function, multiplying attribute weights (wJOB, wRES,
wEND and wMAN) by the corresponding aggregated WTP provides estimates of the
WTP for various levels of the attributes. Table 2 illustrates this transformation.
3.4. Additivity assumption underlying utility functions

Our methodological approach is based on the assumption that individuals’ utility
functions (and also the social utility function) are additive. The sum of partial utilit-
ies for each attribute is equal to the total utility of the complex good. This assump-
tion is a critical part of our method, and requires discussion. For a utility function
to be additive the attributes should be mutually utility independent (an attribute ri
is utility independent of the other n ) 1 attributes rj if preferences for lotteries
involving different levels of attribute ri do not depend on the levels of the other
n ) 1 attributes rj) (Fishburn, 1982; Keeney and Raiffa, 1993). This requirement is
somewhat restrictive, and is seldom fully held by real-world individuals.
The effects of non-linear utility functions on the valuation of environmental
goods have been addressed by a number of studies in the 1980s and 1990s (e.g.
Hoehn and Randall, 1989; Hoehn, 1991; Madden, 1991; Hoehn and Loomis, 1993
or Randall and Hoehn, 1996). These studies demonstrate the presence of substitu-
tion and complementarity effects in the demand for environmental complex goods,
which could involve valuation interdependences which rule out additive separateness
and would result in over- or under-estimation of resulting WTPs.
On the other hand, other more recent studies (Adamowicz et al., 1998; Hanley
et al., 1998; Colombo et al., 2006; Jin et al., 2006 or Mogas et al., 2006) support
the idea that the additive form can be regarded as a reliable proxy of real (and
unknown) utility functions for the valuation of environmental goods. These authors
reach this conclusion by comparing the empirical results of the whole value of com-
plex goods, obtained through CV, with the respective sums of values of their partic-

228
Table 2
Attribute valuation for ‘‘moderate’’ and ‘‘significant’’ improvement alternatives
Contingent valuation

AHP ‘‘Moderate’’ improvement ‘‘Significant’’ improvement
Functions Weights Attribute WTPMOD Attribute WTPSIG
Contribution to wJOB 14,000 WTPJOB1 ¼ wJOB · WTPMOD 16,000 WTPJOB2 ¼ wJOB · WTPSIG
economic activity jobs jobs
in rural areas
Maintenance of wRES 80% WTPRES1 ¼ wRES · WTPMOD 90% WTPRES2 ¼ wRES · WTPSIG
population in rural Farmers with Farmers with
areas and cultural permanent permanent
heritage residence residence
Use of environmentally wEND 15 WTPEND1 ¼ wEND · WTPMOD 9 WTPEND2 ¼ wEND · WTPSIG
friendly agricultural endangered endangered
practices species species
Contribution to wMAN Integrated WTPMAN1 ¼ wMAN · WTPMOD Organic WTPMAN2 ¼ wMAN · WTPSIG
production of farming farming
healthier and safer
food products
This table shows that the joint implementation of CV and AHP can be either in terms of individual values (k), using individual wik and WTPMOD_K
and WTPSIG_K, or aggregated values, using sample mean values of wi, WTPMOD and WTPSIG. This dichotomy allows us to obtain average values
Zein Kallas, Jose´ A. Gómez-Limón and Jesús Barreiro Hurle´
for society as well as to investigate preference heterogeneity among individuals (see section 5.2).
ular attributes, estimated through choice modelling approaches. The similarity of

total values obtained by both methods demonstrates that the additive function
yields extremely close approximations to the hypothetical functions, even when the
conditions of utility independence are not satisfied (through substitution and com-
plementarity effects).10 On these grounds we suppose that the additive utility func-
tion is an adequate simplification of the true utility functions,11 and thus, the
decomposed agriculture values obtained through the implementation of AHP are
reliable.
Testing this supposition would require comparison of the separate values for each
attribute estimated as above with those obtained by determining separately the
WTP for each attribute in an attempt at ‘‘convergent validation’’ (Mitchell and
Carson, 1989). However, CV valuation of the separate attributes could be subject to
a number of ‘‘instrumental biases’’, including embedding effects,12 ‘‘part-whole’’
bias, sequence or order effects13 and violation of the budgetary constraints of indi-
vidual respondents.
In fact, the only reliable way to test the accuracy of the additivity assumption
would be to elicit or estimate individuals’ implicit utility functions. However, this
question is far beyond the objectives of this paper, remaining as an interesting topic
for future research.
4. Empirical Application
4.1. Target population, sample size and sample selection criteria
Focusing on the local demand for multifunctional agriculture, the target population
is all the adult population of Tierra de Campos, 213,749 people. We restrict atten-
tion to the local population because it is difficult to determine a priori the geogra-
phical limits of the relevant and interested population. Furthermore, selecting
non-residents increases the biases due to the embedding effect and scope sensitivity
(Carson and Mitchell, 1993). As there is likely to be a positive WTP for these goods
among non-residents (e.g. in the nearby cities of Valladolid, León and Salamanca),
our WTP estimates could be regarded as lower bounds.
A quota sampling approach was taken. Sampling quotas were based on: place of
residence, municipality population size, age and gender. A total of 120 categories
were determined (five places of residence · three municipality size · four age ranges
10
This additive approximation is also supported by psychological studies regarding human
behaviour, as pointed out by Dawes and Corrigan (1974), Einhorn and Hogart (1975) and
Dawes (1979).
11
As Hwang and Yoon (1981, p. 103) point out: ‘theory, simulation computations, and
experience all suggest that the additive method yields extremely close approximations to very
much more complicated non-linear forms, while remaining far easier to use and understand’.
12
This bias indicates that the valuation of a particular good can vary depending on whether
it is valued individually or embedded into complex goods (see Kahneman and Knetsch, 1992;
Loomis et al., 1993 or Randall and Hoehn, 1996).
13
This bias, to make a long story short, would imply that the value obtained for each func-
tion could be affected by the place in which the valuation question is presented within a valu-
ation sequence. A comprehensive review of these effects can be found in Bateman et al.
(2004).

· two genders) and sample quotas assigned proportionately to the target population
in each quota. Given these quotas, random routes were established in each district
of residence, determining the municipalities in which the effective sample extraction
should be made as well as the number of interviews to be carried out within each
category. Interviews were carried out face-to-face in the interviewees’ homes during
the period June to July 2005, with a total of 383 valid surveys.
4.2. Questionnaire design

Our questionnaire14 was made up of four sections:
1 An information package regarding the components of multifunctionality in the
agricultural ecosystem of Tierra de Campos, detailing the selected attributes and
their corresponding quantification through proxy variables.
2 A description of the current situation of the attributes and proxy variables pre-
sented in section (1) (status quo), as well as the ‘‘moderate’’ and ‘‘significant’’
improvements to be considered. The WTP for these improvements was then elici-
ted using a CV questionnaire as described in Table 1.
3 Information for the application of AHP to obtain the relative importance of the
different attributes considered.
4 A group of questions about relevant socio-demographic characteristics of the par-
ticipant: gender, age, level of education, income, size of municipality, childhood
place of residence (urban or rural environment), occupation and household size.
The questionnaire was refined and validated through a preliminary pilot survey
of 33 individuals, belonging to all gender, age range and municipality size categor-
ies. Only two of 33 individuals refused to complete the questionnaire, the pilot con-
firmed that the questionnaire adequately explained the issues and tasks (a specific
final question on this aspect was included in this pre-test), and that the time for
each interview was reasonable (about 15 minutes).
5. Results
5.1. Aggregated results for the target population
5.1.1. CV results
Ninety-two people (24%) expressed a zero WTP for both improvement scenarios.
Following Halstead et al. (1992) and Hanley et al. (2002), a screening question was
included to identify ‘‘real’’ zeros (zero marginal utility or corner solution due to
lack of income) and ‘‘protest’’ zeros (individuals who do value the good in question,
but set their WTP equal to zero because they disagree with some of the assumptions
of the valuation scenario). Individuals were classified as protest bidders when the
reason behind their zero WTP was a ‘‘rejection of any further increase in taxes’’,
‘‘consider that they already pay sufficient taxes’’ or ‘‘the government should finance
these improvements’’. On the other hand, those arguing that they ‘‘were not interes-
ted in the subject being valued’’, or ‘‘multifunctionality is nor a priority for me’’ or
‘‘I cannot afford to pay any amount’’, were considered to be ‘‘real’’ zeros. Seventy
14
The survey instruments are available upon request from the authors.

Table 3
Willingness to pay (WTP) for improvements in agricultural multifunctionality
(€/year-inhabitants)
WTP for moderate WTP for significant

improvement (WTPMOD) improvement (WTPSIG)
Mean 27.41 36.12

Trimmed mean* 26.86 37.26
Median 25.00 40.00
Mode 50.00 60.00
Variance 341.67 412.84
Note: *Computed discarding the 25% lowest and highest scores.
of the 92 zero bidders were classified as protest bidders and excluded from the ana-
lysis, the remaining 22 were retained in the sample, leaving 313 valid responses.
The average values of the improvements in multifunctionality (from the answers
provided to the open-ended valuation question) are shown in Table 3.
As these WTP estimates are based on a non-normal distribution,15 a Wilcoxon
rank test is applied. The mean WTP estimates are significant in both cases,16 the
significant improvement being valued more highly than the moderate improve-
ment.17
5.1.2. AHP results

The results of the aggregation of the weights obtained through the AHP method
for the four attributes (wJOB, wRES, wEND and wMAN) are shown in Table 4.
The results show that ‘‘Contribution to production of healthier and safer food
products’’ (MAN) is the most valued attribute for the local population, with an
aggregated weight (wMAN) of 37.6%. Second in importance is ‘‘Maintenance of pop-
ulation in rural areas and cultural heritage’’ (RES) with an aggregated weight
(wRES) of 22.5%. The other two attributes (‘‘Use of environmentally friendly agri-
cultural practices’’ and ‘‘Contribution to economic activity in rural areas’’) appear
to be less valued, with aggregated weights of 20.3% (wEND) and 19.6% (wJOB). All
four attributes appear important contributors to the aggregate value of multifunc-
tionality for this population.
Applying the non-parametric Wilcoxon ranks test,18 the weights are significantly
different from zero. Although the values seem quite balanced, this same test also
15
A Kolmogorv–Smirnov test was undertaken with sample values for WTPMOD and WTPSIG
and the null hypothesis of data following a normal distribution was rejected in both cases.
16
The Z-statistic values for the null hypothesis of mean WTPMOD equal to 0 and mean
WTPSIG equal to 0 are 14.82 and 14.84, respectively, with associated p-values below 0.001 in
both cases. Thus, we reject that WTP for these improvements is equal to zero.
17
The Z-statistic value for the null hypothesis of mean WTPMOD equal to mean WTPSIG is
11.06, with an associated p-value below 0.001. Thus, we confirm that both values are signifi-
cantly different from each other.
18
p-values for Kolmogorov–Smirnov Z-statistic are below 0.01 in all cases.

Table 4
Aggregated weights for multifunctionality attributes
wJOB wRES wEND wMAN
Aggregated weight 0.196 0.225 0.203 0.376

(geometric mean)
Arithmetic mean 0.203 0.225 0.205 0.367
Trimmed mean* 0.181 0.215 0.208 0.396
Median 0.165 0.215 0.178 0.352
Variance 0.022 0.019 0.017 0.035
Note: *Computed discarding the 25% lowest and highest scores.
Table 5
Mean willingness to pay (WTP) for the different functions and levels of agricultural
multifunctionality (€/year-inhabitants)
WTP for the ‘‘moderate’’ WTP for the ‘‘significant’’

Function improvement of the attribute improvement of the attribute
Contribution to economic WTPJOB1 ¼ 5.37 WTPJOB2 ¼ 7.08

activity in rural areas (JOB)
Maintenance of population WTPRES1 ¼ 6.17 WTPRES2 ¼ 8.13
in rural areas and cultural
heritage (RES)
Use of environmentally WTPEND1 ¼ 5.56 WTPEND2 ¼ 7.33
friendly agricultural practices
(END)
Contribution to production of WTPMAN1 ¼ 10.31 WTPMANS2 ¼ 13.58
healthier and safer food
products (MAN)
confirms significant differences between the mean values, except in the case of
wEND ) wRES, where differences are not significant.
5.1.3. Joint implementation of CV and AHP results

By combining the mean WTP of the proposed scenarios (WTPMOD and WTPSIG)
with the aggregated weights for multifunctional attributes, we derive the mean val-
ues for each attribute as explained above. The WTPs for each attribute are shown
in Table 5.
These estimates should be interpreted as follows. For example, WTPJOB1 (€5.4
per year and inhabitant) means that on average people would be willing to increase
their annual tax bill by €5.4 to increase the number of agriculture-related jobs in
the area from 12,600 (status quo) to 14,000 (the ‘‘moderate’’ improvement scenario).
WTPJOB2 shows that people would be willing to increase their annual tax bill by
€7.1 per year per inhabitant to increase the number of agriculture-related jobs from
12,600 to 16,000 (the ‘‘significant’’ improvement scenario). The same interpretation
holds for the other three attributes.
Perhaps the most surprising finding of these results is the relatively low valuation
of new jobs in agriculture. This could be explained by two relevant factors: (1) The
rate of agricultural employment in the area is well above European levels (28.5% of
working people) and farming labour has a relative low productivity (lower and
decreasing incomes). Because of this, people might consider that rural development
should encourage diversification, creating jobs in the industrial and services sectors,
rather than increasing the number of agriculture-related jobs; and (2) Farming
activities have a poor social status (hard work and low pay). Considering the low
unemployment rate in this area, this means that new jobs in agriculture are mainly
covered by foreign workers, mostly from Africa and South America, a situation
that generates conflicts in predominantly conservative and elderly rural communi-
ties. In this particular context, more labour in the agricultural sector that would
involve an immigrant labour force is not particularly highly valued.
The estimated WTP for all attributes and levels is significantly different from
zero,19 confirming that there is a demand for all these non-commercial functions of
agriculture. Moreover, partial WTPs for each attribute are also significantly differ-
ent from each other (with the exception of WTPEND1 ) WTPRES1 in the ‘‘moder-
ate’’ improvement scenario and WTPEND2 ) WTPRES2 in the ‘‘significant’’
improvement scenario) showing that not all functions associated with multifunction-
ality are valued equally by society.
5.2. Analysis of heterogeneity in preferences

An additional goal of our research was to examine the heterogeneity of prefer-
ences for agricultural multifunctionality, both at the aggregate level and for each
of the individual components. We regressed WTP values with individual’s socio-
demographic characteristics, using a standard ordinary least squares (OLS)
regression model, to identify significant differences among individual valuations
of multifunctionality.
The dependent variables used to estimate these value functions were the answers to
the two WTP questions. Independent variables were selected from data gathered in the
fourth section of the questionnaire. As most of these data were highly correlated, an
exploratory principal components analysis was carried out to identify the most rele-
vant variables, all of which were coded as level-dependent dummies. Five factors were
detected, explaining 64.7% of total variance, and for each the most representative vari-
able was selected for use in the OLS analysis. The variables used are defined as follows:
AGE1: 1, if interviewee is between 35 and 64 years of age; 0, otherwise.
AGE2: 1, if interviewee is above 64 years of age; 0, otherwise.20
SIZE1: 1, if resident in a municipality between 500 and 2,000 inhabitants; 0, other-
wise.
SIZE2: 1, if resident in a municipality with more that 2,000 inhabitants; 0, other-
wise.21
19
Non-normal distributions are also present for these data and Wilcoxon rank tests have
been carried out rejecting the null hypothesis of WTP equal to zero.
20
AGE, reference level is age below 35 years.
21
SIZE, reference level is municipality of residence with less than 500 inhabitants.

Table 6
OLS value functions for aggregated multifunctionality
WTPMOD WTPSIG
Variables b Standard b t-statistics P-value Variables b Standard b t-statistics P-value
b0 33.744 12.12 0.000 b0 36.135 19.63 0.000

AGE2 )7.259 )0.193 )3.442 0.001 AGE2 )8.200 )0.198 )3.668 0.000
SIZE1 )6.361 )0.135 )1.951 0.052 SIZE1 )5.471 )0.105 )1.997 0.047
SIZE2 )4.502 )0.118 )1.664 0.097 INC1 12.295 0.270 4.948 0.000
INC1 4.981 0.120 2.092 0.037 INC2 8.989 0.138 2.94 0.013
INC2 11.181 0.189 3.264 0.001
GENDER )3.856 )0.104 )1.924 0.055
N 313 R2 0.109 N 313 R2 0.159

F 6.237 p-value 0.000 F 14.531 p-value 0.000
Notes: Only variables with significant coefficients are shown.

OLS, ordinary least squares; WTP, willingness to pay.
EDUC1: 1, if interviewee has secondary education; 0, otherwise.

EDUC2: 1, if interviewee has university education; 0, otherwise.22
INC1: 1, if household income is between 1,500 and 3,000 € per month; 0, other-
wise.
INC2: 1, if household income is above 3,000 € per month; 0, otherwise.23
SEX: 1, if interviewee is female; 0, otherwise.
Table 6 shows the results for the multifunctionality scenarios proposed (WTPMOD
and WTPSIG). Both models are significant from a statistical point of view: individu-
als do not have homogenous valuations of multifunctionality. Multifunctionality
seems to be a normal good, as its valuation increases with income, both in the mod-
erate and significant improvement scenarios, as shown by the positive and signifi-
cant effect of the INC variables.
We regard it as surprising that rural people, especially those with permanent resi-
dence in small villages (below 500 inhabitants) value multifunctionality more than
other segments of the community, as demand for multifunctionality has usually
been assumed to be concentrated in urban areas. Nevertheless, it appears that med-
ium-sized rural areas (SIZE1) show a significantly lower valuation for multifunction-
ality. Our results also indicate that older people and women attach lower values to
multifunctionality.
Although these models are highly significant as a whole and as regards individual
socio-demographic characteristics, R2 values are quite low.24 This reflects the fact
that the reasons for such heterogeneity remain largely unknown. Further fieldwork
22
EDUC, reference level is primary education; this variable was identified alone in the factor
analysis.
23
INCOME, reference level for this variable is household income below 1,500 € per month.
24
It seems that variability in the WTP responses are due to many non-observed factors over-
looked in our survey. However, a review of published studies in which OE WTP OLS models
are used shows that our R2 results are not unusual, and are not among the lowest.

Table 7
Socio-demographic characteristics affecting individual attribute valuation
SEX AGE1 AGE2 SIZE1 SIZE2 INC1 INC2 EDUC1 ECUC2
WTPJOB1 – – – – )1.23* – – 1.96** 3.19***

WTPJOB2 – – – – – – – 2.47** 3.58***
WTPRES1 – – – – – 1.40* – 1.39*
WTPRES2 – 1.46* – – – 3.30*** – – 1.83**
WTPEND1 – – )2.55*** – – – 3.69*** 1.32* –
WTPEND2 – – )3.45*** )1.73* – 2.46*** 3.94*** – –
WTPMAN1 )1.73* – )2.56** – – – 2.8* – –
WTPMAN2 – – )2.55** )3.18** – 2.38* – – –
Notes: Significance levels: ***p < 0.01; **p < 0.05; *p < 0.10.
needs to be performed to identify these other factors. Following the earlier literature
(e.g. Van Liere and Dunlap, 1980; Scott and Willits, 1991; Burbank, 1995 or Cott-
rell, 2003), it would be worth analysing attitude- and/or belief-related factors as
demand determinants for agricultural multifunctionality.
Table 7 shows the coefficients and significance levels of the variables that affect
individuals’ valuation of the specific multifunctionality. Again, although all models
are significant (aggregated significance according to the F-statistic with p-values
below 0.01), R2 values are low (in only two cases above 0.10). As can be seen, edu-
cation, age and income are the variables which affect most attributes.
These results inform the findings shown in Table 6. The lower valuation attached
by women to the moderate improvement in multifunctionality seems to be associ-
ated with a significantly lower weight attached to the indicator of healthier and
safer food production. Older people (AGE2) seem to give lower salience to both
environmental protection and integrated or ecological production methods, which
they might consider as handicaps to ‘‘traditional’’ rural activities. The same holds
for inhabitants of medium-sized rural areas (SIZE1), who might, perhaps, view envi-
ronmental protection and food safety and quality as extra limits on their entrepre-
neurship, competitiveness and chances of development. The only attribute which
urban people (SIZE2) value less than rural people is agriculturally related employ-
ment in the moderate improvement scenario, possibly because the moderate increase
in employment is seen as insignificant for the viability of rural areas.
From our analysis, we can confirm that individuals value multifunctionality dif-
ferently although these differences cannot be explained in full by socio-demographic
variables. Higher education and income have been found to be positively correlated
with WTP for multifunctionality, while older people and residents of medium-sized
rural areas seem to be less willing to support those aspects of multifunctionality that
limit economic activities in rural areas.
6. Conclusions
Our results confirm the existence of a real demand for agricultural multifunction-
ality. This demand is not homogeneous, either among different functions considered
as part of the multifunctional bundle or among individuals. Functions related to a
more ‘‘private good’’ facet of multifunctionality (such as health and food safety)

seem to be more valued than those related to ‘‘public goods and services’’. This is
consistent with research on the motives for social valuation of organic farming,
which might be considered as another form of providing more multifunctionality
(Midmore et al., 2005).
Socio-demographic characteristics only partly explain the heterogeneity found in
the demand for multifunctionality. Income and education have positive effects on
the WTP for aggregate improvements in multifunctionality as well as for most of its
individual functions. Older people and residents in medium-sized rural villages are
less willing to value multifunctional functions, possibly related to limitations on
activity in rural areas (environmental protection and food safety and quality), while
residents of small villages place higher values on multifunctionality, probably believ-
ing that this concept is the only way to assure viable rural areas. Nevertheless, most
of the heterogeneity is not captured by these characteristics.
Our research provides support for the combination of valuation techniques
employed to obtain reliable estimates of attribute values for complex goods. Combi-
ning AHP and CV seems to be a promising alternative to CV alone and to CE.
Although both valuation techniques allow estimates of aggregated and attribute-spe-
cific values to be made, major methodological differences remain. First, from a psy-
chological point of view, the process of choosing alternative scenarios required by
the CE is quite different from that associated with revealing WTPs for the implemen-
tation of CV. This arises from the fact that choices required for CE implementation
are driven from reason and arguments to a greater extent than are pricing responses,
unlike the CV method. Thus, as pointed out by Irwin et al. (1993) and McKenzie
(1993), respondents may react differently when they answer similar questions in each
case. In addition, CV has been found to be prone to embedding and order effects
biases (Brown and Shogren, 1998; Bateman et al., 2004; Barriero et al., 2005),
although this problem has been partly tackled by CE (Bennett and Blamey, 2001;
Hensher et al., 2005). Furthermore, CV surveys usually have higher non-response
and protest behaviour rates than CE studies (Stevens et al., 2000). This could also
affect the estimates of value obtained by both techniques. No evidence has yet been
found to determine which method is more suitable to reveal the preferences of real
individuals. Future research should therefore aim to calibrate our results with CE
and other possible valuation techniques to further validate the methodological
approach. The application of different approaches to multifunctionality valuation in
the same area and to the same population seems to be a promising path.
Finally, our results should be treated with caution. They are only related to a
specific case study area and population, at a specific point in time. The area can be
regarded as a good example of marginal agriculture, which generates important pos-
itive externalities both in the social and environmental fields. Similar studies need to
be undertaken in other areas with different characteristics (type of externalities gen-
erated, degree of productivity and reference population considered) so as to obtain
a wider picture of social demand for multifunctionality.
References
Adamowicz, W., Boxall, P., Williams, M. and Louviere, J. ‘Stated preference approaches for
measuring passive use values: Choice experiments and contingent valuation’, American
Journal of Agricultural Economics, Vol. 80, (1998) pp. 65–75.

Aguarón, J. and Moreno-Jiménez, J. M. ‘Stability intervals in the analytic hierarchy process’,

European Journal of Operational Research, Vol. 125, (2000) pp. 114–133.
Arrow, K., Solow, R., Portney, P., Leaemer, E., Radner, R. and Schuman, H. ‘Report of
the NOAA panel on contingent valuation’, Federal Register, Vol. 58, (1993) pp. 4016–
4064.
Barreiro, J., Casado, J. M. and Pérez y Pérez, L. ‘Incorporating uncertainty and zero values
into the valuation of protected areas and species’, in XI Congress of the European Associ-
ation of Agricultural Economists (EAAE, Copenhagen, 2005).
Bateman, I., Carson, R., Day, B., Hanemann, M., Hanley, N., Hett, T., Lee, M., Loomes,
G., Mourato, S., Ozdemiroglu, E., Pearce, D., Sugden, R. and Sawanson, J. Economic
Valuation with Stated Preference Techniques: A Manual (Cheltenham, UK: Edward Elgar
Publishing Ltd., 2002).
Bateman, I., Cole, M., Cooper, P., Georgiou, S., Hadley, D. and Poe, G. ‘On visible choice
sets and scope sensitivity’, Journal of Environmental Economics and Management, Vol. 47,
(2004) pp. 71–93.
Bateman, I., Munro, A., Rhodes, B., Starmer, C. and Sugden, R. ‘Does part-whole
bias exist? An experimental investigation’, Economic Journal, Vol. 107, (1997) pp. 322–
332.
Batie, S. ‘The multifunctional attributes of Northeastern agriculture: A research agenda’,
Agricultural and Resource Economics Review, Vol. 32, (2003) pp. 1–8.
Bennett, J. and Blamey, R. The Choice Modelling Approach to Environmental Valuation
(Cheltenham: Edward Elgar Publishing Ltd, 2001).
Bennett, J., Van Bueren, M. and Whitten, S. ‘Estimating society’s willingness to pay to
maintain viable rural communities’, The Australian Journal of Agricultural and Resource
Economics, Vol. 48, (2004) pp. 487–512.
Brouwer, F. Sustaining Agriculture and the Rural Environment, Governance, Policy and Multi-
functionality (Cheltenham, UK: Edward Elgar Publishing, 2004).
Brown, G. and Shogren, J. ‘Economics of the Endangered Species Act’, Journal of Economic
Perspectives, Vol. 12, (1998) pp. 3–20.
Bryson, N. ‘A goal programming method for generating priority vectors’, Journal of the
Operational Research Society, Vol. 46, (1995) pp. 641–648.
Burbank, M. ‘How do contextual effects work? Developing a theoretical model’, in M. Eagles
(ed.), Spatial and Contextual Models in Political Research (London: Taylor and Francis,
1995, pp. 165–178).
Cahill, C. ‘The multifunctionality of agriculture: What does it mean?’ EuroChoices, Vol. 1,
(2001) pp. 36–41.
Camarero, L. Del Éxodo Rural y del Éxodo Urbano: Ocaso y Renacimiento de los Asentamien-
tos Rurales en España (Madrid: Ministerio de Agricultura, Pesca y Alimentación, 1993).
Carson, R. Contingent Valuation: A. Comprehensive Bibliography and History (Cheltenham:
Edward Elgar Publishing Ltd, 2006).
Carson, R. and Mitchell, R. ‘The issue of scope in contingent valuation studies’, American
Journal of Agricultural Economics, Vol. 75, (1993) pp. 1263–1267.
Colombo, S., Calatrava-Requena, J. and Hanley, N. ‘Analysing the social benefits of soil
conservation measures using stated preference methods’, Ecological Economics, Vol. 58,
(2006) pp. 850–861.
Colombo, S., Hanley, N. and Calatrava, J. ‘Designing policy for reducing the off-farm effect
of soil erosion using choice experiments’, Journal of Agricultural Economics, Vol. 56, (2005)
pp. 81–95.
Cooper, J. ‘Optimal bid selection for dichotomous choice contingent valuation surveys’,
Journal of Environmental Economics and Management, Vol. 24, (1993) pp. 25–40.
Cottrell, S. P. ‘Influence of sociodemographics and environmental attitudes on general res-
ponsible environmental behaviour among recreational boaters’, Environment and Behaviour,
Vol. 35, (2003) pp. 347–375.

Dawes, R. M. ‘The robust beauty of improper linear models in decision making’, American
Psychology, Vol. 34, (1979) pp. 571–582.
Dawes, R. M. and Corrigan, B. ‘Linear models in decision making’, Psychological Bulletin,
Vol. 81, (1974) pp. 95–100.
DG AGRI (2006). Agriculture in the European Union. Statistical and Economic Information.
Available on-line at http://www.europa.eu.int/comm/agriculture/agrista/2005/table_en/2005
enfinal.pdf (Last accessed: 3/2006).
Easley, R., Valacich, J. and Venkataramanan, M. ‘Capturing group preferences in a
multicriteria decision’, European Journal of Operational Research, Vol. 125, (2000) pp. 73–
83.
EC. Contribution of the European Community on the Multifunctional Character of Agriculture
(Brussels: European Commission–DG Agriculture, September, 1998).
EC. Safeguarding the Multifunctional Role of Agriculture: Which Instruments? (Brussels: Euro-
pean Commission–DG Agriculture, September, 1999).
EC. Agriculture’s Contribution to Environmentally and Culturally Related Non-trade Concerns
(Brussels: European Commission–DG Agriculture, July, 2000).
EEA (European Environment Agency). High Nature Value Farmland. Characteristics, Chan-
ges and Policy Challenges (Luxembourg: Office for Official Publications of the European
Community, 2004).
Einhorn, H. J. and Hogart, R. M. ‘Unit weighting schemes of decision making’, Organiza-
tional Behavior and Human Performance, Vol. 13, (1975) pp. 171–192.
Fichtner, J. ‘On deriving priority vectors from matrices of pairwise comparisons’, Socio-
Economic Planning Science, Vol. 20, (1986) pp. 341–345.
Fishburn, P. C. The Foundations of Expected Utility (Dordrecht: Reidel Publishing Company,
1982).
Forman, E. and Peniwati, K. ‘Aggregating individual judgments and priorities with the
Analytic Hierarchy Process’, European Journal of Operational Research, Vol. 108, (1998)
pp. 165–169.
Franco, F. and Manero, F. ‘Valoración global y perspectivas de futuro’, in A. Blanco (ed.),
Envejecimiento y Mundo Rural en Castilla y León (Madrid: Fundación Encuentro, 2002,
pp. 68–79).
Golden, B., Wasil, E. and Harker, P. The Analytic Hierarchy Process: Applications and Studies
(Berlin: Springer-Verlag, 1989).
Goméz-Limón, J. A. and Atance, I. ‘Identification of public objectives related to agricultural
sector support’, Journal of Policy Modelling, Vol. 27, (2004) pp. 1045–1071.
Hall, C., McVittie, A. and Moran, D. ‘What does the public want from agriculture and the
countryside? A review of evidence and methods’, Journal of Rural Studies, Vol. 20, (2004)
pp. 211–225.
Halstead, J., Lulof, A. and Stevens, T. ‘Protest bidders in contingent valuation’, Northeastern
Journal of Agricultural and Resources Economics, Vol. 21, (1992) pp. 160–183.
Hanley, N., Ryan, M. and Wright, R. ‘Estimating the monetary value of health care:
Lessons from environmental economics’, Health Economics Review, Vol. 12, (2002)
pp. 3–16.
Hanley, N., Wright, R. and Adamowicz, V. ‘Using choice experiments to value the environ-
ment: Design issues, current experience and future prospects’, Environmental and Resource
Economics, Vol. 11, (1998) pp. 413–428.
Harvey, D. ‘Agri-environmental relationships and multi-functionality: Further considera-
tions’, The World Economy, Vol. 26, (2003) pp. 705–725.
Hensher, D. A., Rose, J. M. and Greene, W. H. Applied Choice Analysis: A Primer (Cam-
bridge: Cambridge University Press, 2005).
Hoehn, J. P. ‘Valuing the multidimensional impact of environmental policy: Theory and
methods’, American Journal of Agricultural Economics, Vol. 73, (1991) pp. 289–299.

Hoehn, J. P. and Loomis, J. ‘Substitution effects in the contingent valuation of multiple envi-
ronmental programs’, Journal of Environmental Economics and Management, Vol. 25,
(1993) pp. 56–75.
Hoehn, J. and Randall, A. ‘A satisfactory benefit cost indicator from contingent
valuation’, Journal of Environmental Economics and Management, Vol. 14, (1987) pp. 226–
247.
Hoehn, J. and Randall, A. ‘Too many proposals pass the benefit cost test’, American Econo-
mic Review, Vol. 79, (1989) pp. 544–551.
Hwang, C. L. and Yoon, K. Multi Attribute Decision Making (New York: Springer-Verlag,
1981).
INE. Censo Agrario 1999. Encuestas Sobre la Estructura de las Explotaciones Agrıćolas
(Madrid: Instituto Nacional de Estadı́stica, 2001).
INE. Contabilidad Regional de España 2003 (Madrid: Instituto Nacional de Estadı́stica,
2004).
Irwin, R., Slovic, P., Licktenstein, S. and McCelland, G. ‘Preference reversals and
the measurement of environmental values’, Journal Risk Uncertain, Vol. 6, (1993)
pp. 5–18.
Jin, J., Wang, Z. and Ran, S. ‘Comparison of contingent valuation and choice experiment
in solid waste management programs in Macao’, Ecological Economics, Vol. 57, (2006)
pp. 430–441.
Kahneman, D. and Knetsch, J. ‘Valuing public goods: The purchase of moral satisfaction’,
Journal of Environmental Economics and Management, Vol. 22, (1992) pp. 57–70.
Kamenetzky, R. ‘The relationship between the analytic hierarchy process and the additive
value function’, Decision Science, Vol. 13, (1982) pp. 702–713.
Kaninen, B. and Kriström, B. ‘Sensitivity of willingness to pay estimates to bid design in
dichotomous choice valuation models: Comment’, Land Economics, Vol. 69, (1993)
pp. 199–202.
Keeney, R. L. and Raiffa, H. Decisions with Multiple Objectives: Preferences and Value
Trade-offs (Cambridge: Cambridge University Press, 1993).
Laininen, P. and Hämäläinen, R. ‘Analysing AHP-matrices by regression’, European Journal
of Operational Research, Vol. 148, (2003) pp. 514–524.
Lima Santos, J. M. (2001). A Synthesis of Country Reports on Demand Measurement of Non-
commodity Outputs in OCDE Agriculture. Workshop on Multifunctionality, Directorate for
Food, Agriculture and Fisheries, OECD, Paris. Available on: http://www1.oecd.
org/agr/mf/doc/Santos.pdf (Last accessed 3/2006).
Loomis, J., Lockwood, M. and Delacy, T. ‘Some empirical evidence on embedding effects in
contingent valuation of forest protection’, Journal of Environment Economics and Manage-
ment, Vol. 25, (1993) pp. 45–55.
Madden, P. ‘A generzalization of Hicksian Q substitutes and complements with application
to demand rationing’, Econometrica, Vol. 59, (1991) pp. 1497–1508.
McKenzie, J. ‘A comparison of contingent preference models’, American Journal of Agricul-
tural Economics, Vol. 75, (1993) pp. 593–603.
McVittie, A., Moran, D., Allcroft, D. and Elston, D. ‘Beauty, beast and biodiversity: What
does the public want from agriculture?’ 78th Annual Conference of the Agricultural
Economics Society, AES, SouthKensington. Available on http://www.scotland.gov.uk/
Publications/2004/09/19892/42598 (Last accessed 3/2006).
Midmore, P., Naspetti, S., Sherwood, A., Vairo, D., Weir, M. and Zanoli, R. Consumer Atti-
tudes to Quality and Safety of Organic and Low Input Foods: A Review. Deliverable 1.2,
Subproject 1, Quality Low Input Food Project. Available at http://www.qlif.org/research/
sub1/QLIF_Review_Reanalysis_%200509.pdf (Last accessed 3/2006).
Mitchell, R. and Carson, R. Using Surveys to Value Public Goods (Washington, DC:
Resources for the Future, 1989).

Mogas, J., Riera, P. and Bennett, J. ‘A comparison of contingent valuation and choice
modelling with second-order interactions’, Journal of Forest Economics, Vol. 12, (2006)
pp. 5–30.
OECD. Multifunctionality: Towards an Analytical Framework (Paris: OECD, 2000).
OECD. Multifunctionality: Applying the OECD Analytical Framework. Guiding Policy Design
(Paris: OECD, 2001).
OECD. Multifunctionality: The Policy Implications (Paris: OECD, 2003).
Pérez, M. Agricultura y Medio Ambiente: Necesidades de Formación Medioambientales en el
Sector Agrıćola (Valladolid: Instituto de Formación y Estudios Sociales de Castilla y León,
2001).
Peterson, J., Boisvert, R. and De Gorter, H. ‘Environmental policies for a multifunctional
agricultural sector in open economies’, European Review of Agricultural Economics, Vol. 29,
(2002) pp. 423–443.
Prety, J. ‘The externalities and multifunctionality of agriculture’, EuroChoices, Vol. 2, (2003)
pp. 40–44.
Randall, A. ‘Valuing the outputs of multifunctional agriculture’, European Review of Agricul-
tural Economics, Vol. 29, (2002) pp. 289–307.
Randall, A. and Hoehn, J. ‘Embedding in market demand systems’, Journal of Environmental
Economics and Management, Vol. 30, (1996) pp. 369–380.
Saaty, T. ‘A scaling method for priorities in hierarchical structures’, Journal of Mathematical
Psychology, Vol. 15, (1977) pp. 234–281.
Saaty, T. The Analytic Hierarchy Process (New York: McGraw HillInc., 1980).
Saaty, T. ‘The seven pillars of the analytic hierarchy process’, in M. Köksalan and S. Zionts
(eds), Multiple Criteria Decision Making in the New Millennium (Berlin, Springer-Verlag,
2001).
Saaty, T. ‘Decision-making with the AHP: Why is the principal eigenvector necessary?’ Euro-
pean Journal of Operational Research, Vol. 145, (2003) pp. 85–91.
Scott, D. and Willits, F. K. Environmental Concern of Pennsylvania Citizens: Data from a
Statewide Survey (Technical Report, A.E. and R.S. 219) (University Park, PA: Depart-
ment of Agricultural Economics and Rural Sociology, Pennsylvania State University,
1991).
Smith, H. and Lantz, V. A Methodology for Evaluating Public Values of New Brunswick
Forests (New Brunswick, UK: The Fundy Model Forest, 2003) Available at http://
www.fundymodelforest.net/pdf/socioeconomics/2003_public_values.pdf (Last accessed 12/
2005).
Stevens, T., Belkner, R., Dennis, D., Kittredge, D. and Willis, C. ‘Comparison of contingent
valuation and conjoint analysis in ecosystem management’, Ecological Economics, Vol. 32,
(2000) pp. 63–74.
Suárez, F., Naveso, M. A. and de Juana, E. ‘Farming in the drylands of Spain: Birds of
pseudosteppes’, in D. Pain and M. Pienkowski (eds), Farming and Birds in Europe: The
Common Agricultural Policy and Its Implications for Bird Conservation (London: Academic
Press, 1997, pp. 297–330).
Van Huylenbroeck, G. and Durand, G. Multifunctional Agriculture, A New Paradigm for
European Agriculture and Rural Development (London: Ashgate, 2003).
Van Liere, K. D. and Dunlap, R. E. ‘The social bases of environmental concern: A review of
hypotheses, explanations and empirical evidence’, The Public Opinion Quarterly, Vol. 44,
(1980) pp. 181–197.
Wattage, P. and Mardle, S. ‘Total economic value of fishery dependent area conservation
in Sri Lankan wetlands: Application of contingent valuation method (CVM) and
analytic hierarchy process (AHP) to identify use and non-use values’, in The XVIth
Annual Conference of the European Association of Fisheries Economists (Rome: UN-FAO,
2004).

Yrjölä, T. and Kola, J. ‘Consumer preferences regarding multifunctional agriculture’, Interna-

tional Food and Agribusiness Management Review, Vol. 1, (2004) pp. 78–90.
Zahedi, F. ‘A utility approach to the analytic hierarchy process’, Mathematical Modelling,
Vol. 9, (1987) pp. 387–395.
Zoppi, C. ‘A contingent valuation-multicriteria analysis case study on the taxonomy of three
planning scenarios for a coastal zone of Sardinia (Italy)’, in The 44th European Congress of
the European Regional Science Association (Porto, Portugal: ERSA, 2004).

Descomposing Agricultural Multifunctionality

Uploaded by

Copyright:

Available Formats

Descomposing Agricultural Multifunctionality

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Descomposing Agricultural Multifunctionality

Uploaded by

Copyright:

Available Formats

Journal of Agricultural Economics, Vol. 58, No.

Decomposing the Value of Agricultural

Keywords: Agricultural multifunctionality; analytical hierarchy process; contin-

1. Introduction and Objectives

2.2. Agricultural multifunctionality in Tierra de Campos

2.2.2. Social functions

2.2.3. Environmental functions

Blackwell Publishing Ltd. 2007

(iii) deliberative methods;

Combining equations (4) and (5) we obtain:

WTPDMF ¼ ðw1 F1imp þ w2 F2imp Þ ðw1 F1ref þ w2 F2ref Þ

Hence, the WTP for an individual function of multifunctionality is as follows:

WTPDFi ¼ wi WTPDMF : ð7Þ

Blackwell Publishing Ltd. 2007

3.1. Contingent valuation

3.1.1. Contingent market design

Blackwell Publishing Ltd. 2007

3.1.2. Contingent market design: question format and payment vehicle

Functions carried out by agriculture Current situation “Moderate” “Significant”

Contribution to economic activity 12,600 14,000 16,000

Maintenance of population in rural

Use of environmentally friendly

Food produced by Food produced by Food produced by

My willingness to pay to achieve the “significant” improvement is a maximum of: €/year

Blackwell Publishing Ltd. 2007

3.2. Analytical hierarchy process

Attribute 1 Attribute 2 Attribute 4

Figure 1. Hierarchical structure used to value agricultural multifunctionality

Blackwell Publishing Ltd. 2007

Blackwell Publishing Ltd. 2007

We follow Forman and Peniwati (1998), in aggregating individual weights (wik)

Zahedi (1987) has demonstrated that resolving a multicriteria problem through

3.3. Combining CV and AHP data

3.4. Additivity assumption underlying utility functions

Blackwell Publishing Ltd. 2007

Blackwell Publishing Ltd. 2007

Functions Weights Attribute WTPMOD Attribute WTPSIG

ular attributes, estimated through choice modelling approaches. The similarity of

Blackwell Publishing Ltd. 2007

4.2. Questionnaire design

Blackwell Publishing Ltd. 2007

WTP for moderate WTP for signiﬁcant

Mean 27.41 36.12

Note: *Computed discarding the 25% lowest and highest scores.

5.1.2. AHP results

Blackwell Publishing Ltd. 2007

wJOB wRES wEND wMAN

Aggregated weight 0.196 0.225 0.203 0.376

Note: *Computed discarding the 25% lowest and highest scores.

WTP for the ‘‘moderate’’ WTP for the ‘‘signiﬁcant’’

Contribution to economic WTPJOB1 ¼ 5.37 WTPJOB2 ¼ 7.08

5.1.3. Joint implementation of CV and AHP results

5.2. Analysis of heterogeneity in preferences

Blackwell Publishing Ltd. 2007

Variables b Standard b t-statistics P-value Variables b Standard b t-statistics P-value

b0 33.744 12.12 0.000 b0 36.135 19.63 0.000

N 313 R2 0.109 N 313 R2 0.159

Notes: Only variables with signiﬁcant coefﬁcients are shown.

EDUC1: 1, if interviewee has secondary education; 0, otherwise.

WTPJOB1 – – – – )1.23* – – 1.96 3.19*