Environmental Engineering and Management Journal September 2010, Vol.9, No.

9, 1145-1149
http://omicron.ch.tuiasi.ro/EEMJ/

“Gheorghe Asachi” Technical University of Iasi, Romania

______________________________________________________________________________________________

MEASURING ENVIRONMENTAL EVOLUTION USING SYNTHETIC

INDICATORS

José Mondéjar-Jiménez1*, Manuel Vargas-Vargas1, Juan-Antonio Mondéjar-Jiménez2

1
University of Castilla-La Mancha, Department of Statistics
2
University of Castilla-La Mancha, Department of Business Faculty of Social Sciences, 44 Alfares Avda, 16071 Cuenca, Spain

Abstract

Environmental evolution is of great interest among researchers, at national and international level. Countries need to comply
scrupulously with their legislation on sustainability issues, while also promoting protection of the environment internationally.
This paper examines the evolution of environmental impact in Spain. For this purpose, secondary information available about the
main environmental indicators in Spain has been taken, and we built a synthetic indicator built capable of measuring the monthly
evolution of the environmental impact.

Key words: environmental evolution, measurement, sustainability indicators, synthetic indicators

Received: September, 2010; Accepted: September, 2010∗∗

1. Introduction considered logical by the aforementioned experts

Measurement of environmental impact,
together with the definition of the factors causing it,
are two of the most interesting questions in research
terms in recent years. At present, sustainable
development is closely linked to environmental
analysis (Condrea and Bostan, 2008; Duran et al.,
2009; Ioan et al., 2008; Parris and Kates, 2003 Pîslaru
et al., 2008).
Social indicators and sustainable development
are generally defined as quantitative tools exclusively
intended to initiate policies defined as rationally
organized collections of instrumental measures and
actions intended to solve a well defined social
problem (Boulanger, 2007).
The idea of sustainable development has
become an essential reference point affecting all
public policies and which is the real reason why a
series of international, institutional initiatives has
arisen of groups of experts working to prepare
indicators of sustainable development, following a
rationale of standardization based on indicators
provided on a “rational” basis, the use of which is
(Bossel, 1999; Rey-Valette et al., 2007).
Quantification of these impacts is a wide and
also complex field of study. For this reason,
modelling any environmental phenomenon with a
single indicator faces a host of empirical difficulties,
such as the selection of partial indicators, preliminary
statistical treatment, signal extraction or the
aggregation of partial indicators (Eakin and Luers,
2006; Mondéjar et al., 2008; Mondéjar et al., 2009).
The individual indicators make up a list of
values with clear, easy-to-use numbers. If the end
users accept this list of indicators as valid, they can be
used as a point of reference to monitor a particular
phenomenon (Laloë, 2007). An indicator of
sustainable development implies more than providing
a measurement from the available information. The
use of indicators is a complex matter, because it
depends on the variety of functions carried out and
their involvement in management, for this reason the
existence of contradictory indicators causes non-
uniformity in their use (Rey-Valette et al., 2007).
Some authors suggest the integration of the
different indicators in a single index (Barrera-Roldán
and Saldivar-Valdés, 2002; Nambiar et al., 2001), but

*
Author to whom all correspondence should be addressed: e-mail: Jose.Mondejar@uclm.es; Phone: +34 902204100; Fax: +34 902204130
∗∗
Paper presented within the third International Congress on Tourism and Environment, Florence, Italy, 2010
 Mondéjar-Jiménez et al./Environmental Engineering and Management Journal 9 (2010), 9, 1145-1149

that leads to doubts about the respective contribution Table 1. Selected partial sustainability indicators
of the different components. However, use of (MEMRA, 2010)
synthetic indicators is an internationally tried and
tested method and an increasingly used tool. The Description Units
main reason this use has become more common is Average surface ozone in the EMEP network µg/m3
that these instruments, which seek to find a common Average SO2 in the EMEP network µg/m3
trend among many, separate indicators, are much Average NO2 in the EMEP network µg/m3
more easily interpreted. Synthetic indicators have Average NOx in the EMEP network µg/m3
Electrical discharges nº
shown themselves to be highly useful, especially in
Mean mainland rainfall l/m2
countries with better sources of information or more
Total mainland hydrological reserves %
reliable statistics (Mondéjar and Vargas, 2009).
Forest fires declared nº
For this reason, use of synthetic indicators Total wooded area affected Ha
needs to follow a series of premises (Montmollin and Thermal electrical power production GWh
Scheller, 2007): Nuclear electrical power production GWh
- all aspects related with sustainable development, Electricity generated from renewable energy GWh
including those which do not directly affect policies, sources
should be included, Total consumption of petroleum products kton
- based on existing indicators, Environmental crimes number
- as much information as is available should be
included in their constitution, Final selection of the indicators especially
- a tool of use to the end user who has to take into stresses the environmental impact caused by
consideration and guarantee broad support for economic activities. When the indicators have been
sustainability and awareness-raising. homogenized, the next step is filtration of the
individual indicators. In this stage, the intention is to
2. Material and methods eliminate from each simple indicator the noise
inherent to each series as well as seasonal
The need for information is the greatest components, so isolating the trend-cycle component.
difficulty and an indispensable requirement for the To carry out the filter process, the dynamic
construction of a synthetic indicator. However, its state-space Bayesian method was used (Mondéjar,
interpretation is not always simple, as they can be 2007); the Bayesian system used to filter the
poorly constructed or wrongly interpreted, thereby indicators is closely linked to dynamic state space
sending erroneous information (Mondéjar and models. Using Bayes' theorem, Bayesian learning
Vargas, 2008; Saisana and Tarantola, 2002). combines information from observations, expressed
The construction of synthetic indicators is not according to likelihood, with the state of the analyst's
simple, involving both theoretical and methodological knowledge before the observations are available: this
suppositions that must be analyzed to prevent provides a mechanism for updating knowledge of the
incorrect results. Nardo et al. (2005) lists a series of series which can be explained as follows:
considerations to be taken into account for indices to 1. Given indicator Yt and its trend-cycle signal θt,
be built properly. Specifically, he highlights the need it is assumed that their relationship is disturbed by a
for an explicit conceptual framework for the index, noise, and that the signal has a Markovian evolution,
and the usefulness of multivariate analysis prior to as explained by Eqs. (1):
aggregation of the individual indicators (Mondéjar
and Vargas, 2009). Yt = Fθ t + Vt
Nardo et al. (2005), defines a synthetic (1)
θ t − 1 = Gθ t + Wt
indicator as “… a mathematical combination of the
different dimensions of a concept the description of
2. Call the knowledge existing at instant (t-1), Dt-1.
which is the purpose of the analysis”.
The prior information about the signal at instant t can
The first step in carrying out their construction
is the selection of the partial indicators that are to be be expressed through the distribution θ t D t − 1 .
used in the aggregation. Their classification needs 3. The forecast from the prior information will be
knowledge of the available statistical sources and generated through the observation Eq. (2):
quality analysis of the available information
(Mondéjar, 2009). Using the existing information Yt D t − 1 = (Ft θ t D t − 1) + (Vt D t − 1) = Ft (θ t D t − 1) + (Vt D t − 1) (2)
available and based on the selected frequency
(monthly), a selection of indicators has been carried 4. Dt = Dt − 1 ∪ Yt can be used to obtain the
out for the reference period January 2002 to posterior distribution, which will be proportional to
December 2008 (Table 1). the product of the prior distribution and the likelihood
Selection of the final series of indicators is function (Eq. 3).
based on the principle of representative diversity
(O'Connor and Spangenberg, 2008), with the (θ t Dt ) ∝ (Yt {θ t , Vt})· (θ t D t − 1) (3)
intention of taking in different points of view.

1146
 Measuring environmental evolution using synthetic indicators

5. Finally, the transition equation is used to obtain contained in the presumably anomalous datum. The
the prior distribution for the signal at instant (t+1) fact that k periods are taken allows only anomalies
(Eq. 4): which have a relatively important influence to be
detected.
(θ t − 1 D t ) = G t (θ t D t ) + (Wt D t ) (4) If an interest exists in a broader prediction
horizon, k ≥ 1 , the predictive distributions provided
closing the “prior – likelihood - posterior - prior” by the dynamic linear model would be given by Eqs.
cycle, bases of Bayesian learning. This formulation (12).
allows great flexibility in both analysis and
intervention, a desirable feature in a dynamic context (θ t + k Dt ) → tδtnt[at (k ), Rt (k )] (12)
such as environmental synthetic indicators. (Yt + k Dt ) → tδtnt[ ft (k ), Qt (k )]
Assuming a normal prior distribution, it can be
used the algorithm described to update the knowledge where future update parameters are carried out
of the trend-cycle series, and find the posterior according to Eqs. (13).
distributions (Eqs. 5, 6):
at (k ) = Gt + k at (k − 1) at (0) = mt
(13)
Rt (k ) = Gt + k Rt (k − 1)G ′ t + k + Wt + k Rt (0 ) = Ct
(θ t D t ) → tn t [mt , Ct ] (5)
ft (k ) = F ′t + k at (k ) Qt (k ) = F ′t + k Rt (k )Ft + k + St

(ϕ t ⎡ nt dt ⎤
Dt) →G⎢ , ⎥
(6) Finally, the smoothing trend-cycle distribution
⎣2 2⎦ can be obtained by simply carrying back the
information since the last available period using these
where the update parameters are given by Eqs. (7–9). distributions (Eq. 14):
⎛ St ⎞ (7)
mt = at + At et Ct = ⎜ ⎟[Rt − AtA′tQt ] (θ t − k D t ) → t n t (− k )[a t (− k ), R t (− k )] (14)
⎝ St − 1 ⎠ ⎡ n t (− k ) d t (− k ) ⎤
(ϕ t − k D t) → G ⎢ ,
⎣ 2 2 ⎥⎦
RtFt (8)
et = Yt − ft At =
Qt considering that (Eqs. 15).

St − 1et2 (9) at (− k ) = mt − k −Bt − k[at − k + 1(− k ) − at(− k + 1)]

nt = δ t nt − 1 + 1 dt = δ t dt +
Qt Rt(− k ) = Ct − k − Bt − k[Rt − k + 1 − Rt(− k + 1)]B′t − k
(15)
Bt = CtG′t + 1 + Rt−+11
so completing the data learning process. nt (− k ) = nt − k + δ [nt(− k +1) − δ t − k + 1 nt − k]
t − k +1
All the information available was introduced
in the series in our model in a consistent, reasonably S t
−1
(− k ) = S + δ t − k + 1(St−1 (− k +1) − Rt−−1k )
−1
t +1

versatile way. Specifically, the formulation of a dt(− k ) = nt(− k )St (− k )

dynamic linear model allows information from
outside the sample to be entered at any moment of the A smoothed estimate of the trend-cycle given
period, simply altering the prior distribution in that the available set of information will result. The third
period. Another interesting advantage is the detection stage will be the aggregation of the relevant signal of
and processing of outliers. A model is proposed by the partial indicators, which will both be explained in
the foregoing algorithm M and an alternative model the second point. A review of the main methods of
MA, as well. aggregation can be found in the specialized literature
The so called Bayes factor was defined (Eq. (Mondéjar, 2007; Mondéjar and Vargas, 2008), and
10), which may quantify the evidence for and against without wishing to describe them thoroughly it might
the proposed model compared with the alternative. be a good idea to point out that the method finally
chosen was factorial analysis, as recommended in
p (Yt D t − 1 , M ) (10) several specialist studies (Hoskins et al., 2006;
Ht =
p (Yt D t − 1 , MA ) Hoskins and Mascherini, 2009; Mondéjar, 2007;
Mondéjar, 2009; Nardo et al., 2005). Factorial
Using this mechanism, Jeffreys (1961) analysis, a statistical technique which identifies the
suggests that there is evidence in favour of the factors underlying the correlations between
alternative model (Eq. 11). indicators, was used in order to find out whether the
theoretical composition of the dimensions and sub-
p (Yt , Yt 1, ..., Yt k +1 Dt k ,M) (11) dimensions was supported by the data. Factorial
Ht L (k ) = < 0.1
p (Yt , Yt 1, ..., Yt k +1 Dt k , MA) analysis is.

In this way, there can be detected outliers 3. Results and discussion

when the Bayes factor shows evidence against the
proposed model, the alternative model being that From the available information, a synthetic
resulting from elimination of the information indicator has been constructed, capable of

1147
 Mondéjar-Jiménez et al./Environmental Engineering and Management Journal 9 (2010), 9, 1145-1149
characterizing the initial sustainability situation and
showing its evolution over recent years. To interpret
the indicators correctly, bear in mind that values
greater than 0 indicate that the country's
environmental impact is growing, while values below
0 show that the impact is falling. For this reason, as
well as the values of the indicator, the figure shows
the trend graph which allows the evolution over time
in the last six years to be seen. From Fig. 1, it can be
seen that during 2004, the environmental impact fell
in Spain.
Fig. 1. Synthetic indicator of environmental evolution
From that date on, there began a stage of
sustained growth in the environmental impact in this
country, and in 2006 the initial situation was
exceeded. However, in the second half of 2007, a
slight decrease occurred, followed by the highest
increase in recent years in the second half of 2008. In
addition, it can be seen that the evolution of
environmental impact is almost always increasing.
4. Conclusions
Everyone is aware of the desire of the various
agents involved, especially public institutions, to
anticipate the trend found in the matter of
environmental impact fluctuations, because this
conditions their decision-making.
It must be remembered that the information
conditions the whole indicator preparation process
from the outset. For this reason, authors like Pannell
and Glenn (2000) judge the value of an indicator by
the information it provides to improve decision-
making and the construction of a synthetic indicator
provides such information, which is highly valuable
because it provides real time information of
environmental impact generated.
Therefore, it must be kept in mind that the
periodicity and availability of indicators determine
the whole process of updating them.
Data filtering gives the researcher series which
are more suitable for analysis and study of the
environment, because of its highly seasonal nature.
Use of a particular filter model determines the results
1148
obtained, making Bayesian models more
recommendable for study of evolution in the long
term, which was the objective of this study.
Among all the possible solutions, the use of
multivariate methods produces very reliable,
advanced synthetic indicators, which makes them one
of the most used options. In this way, an indicator is
obtained which is perfectly synchronized with
environmental impact. Synthetic, multivariate
indicators record and reflect any change occurring in
nature immediately.
From the results obtained, it can be confirmed
that Spain is steadily increasing its environmental
impact. In this way, the model used confirms this
increase, although not alarmingly. The usefulness of
the principle used in this study lies in its implications
for management, in that it allows exhaustive control
of data to be inferred and enables decisions to be
made on the matter of corrective measures to prevent
such environmental deterioration. This fact should be
taken into account by public decision-makers, to
avoid mortgaging the future of coming generations.
References
Barrera-Roldán A., Saldivar-Valdés A., (2002), Proposal
and application of a sustainable development index,
Ecological Indicators, 2, 251-256.
Bossel H., (1999), Indicators for sustainable development:
theory, method, application, Report to the Balaton
Group, IISD, Winnipeg, Canada.
Boulanger P.M., (2007), Political uses of social indicators:
overview and application to sustainable development
indicators, International Journal of Sustainable
Development, 10, 14-32.
Condrea P., Bostan I., (2008), Environmental issues from
an economic perspective, Environmental Engineering
and Management Journal, 9, 843-849.
Duran D., Duran I., Diaconu A., (2009), Issues on costs and
investment for environmental protection,
Environmental Engineering and Management Journal,
8, 705-708.
Eakin H., Luers A.L., (2006), Assessing the vulnerability of
social-environmental systems, Annual Review of
Environment and Resources, 31, 365–394.
Hoskins B., Mascherini M., (2009), Measuring active
citizenship through the development of a composite
indicator, Social Indicator Research, 90, 459–488.
Hoskins B., Jesinghaus J., Mascherini M., Munda G., Nardo
M., Saisana M., Van Nijlen D., Vidoni D., Villalba E.,
(2006), Measuring Active Citizenship in Europe. Ispra:
European Commission Institute for the Protection and
Security of the Citizen EUR 22530 EN.
Ioan C.C., Macoveanu M., Robu B., (2008), The Gorgon
method as a tool in sustainable behaviour and
development, Environmental Engineering and
Management Journal, 7, 781-784.
Jeffreys H., (1961), Theory of Probability, Oxford
University Press, London.
MEMRA, (2010), Statistical information, Ministry of the
Environment, and Marine and Rural Affairs, (in
Spanish), on line at: http://www.marm.es.
Mondéjar J., (2007), Quantitative analysis of the economic
situation: An application of the state-space
representation of multiple time series, Doctoral Thesis
 Measuring environmental evolution using synthetic indicators
(in Spanish), Ediciones de la Universidad de Castilla-
La Mancha.
Mondéjar J., (2009), Monitoring regional economies with
synthetic indicators, Journal of Business & Economics
Research, 7, 137-142.
Mondéjar J., Meseguer M.L.,Vargas M., (2008),
Measurement of regional economic development
through synthetic indicators activity, original title:
Medición del desarrollo económico regional a través de
indicadores sintéticos de actividad, Revista Venezolana
de Análisis de Coyuntura, XVI, 265-281.
Mondéjar J., Mondéjar J.A., Vargas M., (2009), Economic
evolution measurement of the agricultural sector in
Castilla-La Mancha using synthetic indicators,
Agrociencia, 43, 309-318.
Mondéjar J., Vargas M., (2008), Synthetic indicators: a
revision of aggregation methods, Economía, sociedad y
territorio, VIII, 565-585.
Mondéjar J., Vargas M., (2009), Trends in economic
activity through synthetic activity indicators, African
Journal of Business Management, 3, 736-740.
Montmollin A., Scheller A., (2007), MONET indicator
system: the Swiss road to measuring sustainable
development, International Journal of Sustainable
Development, 10, 61-72.
Nambiar K.K.M., Gupta A.P., Qinglin F., Li S., (2001),
Biophysical, chemical and socio-economic indicators
for assessing agricultural sustainability in the Chinese
coastal zone, Agriculture Ecosystems & Environment,
87, 209-214.
Nardo M., Saisana M., Saltelli A., Tarantola S., Hoffman
A., Giovannini E., (2005), Handbook on constructing
composite indicators: Methodology and user guide,
OECD Statistics Working Papers 2005/3, OECD
Statistics Directorate.
Pannell D.J., Glenn N.A., (2000), A framework for the
economic evaluation and selection of sustainability
indicators in agriculture, Ecological Economics, 33,
135-149.
Saisana M., Tarantola S., (2002), State-of-the-art report on
current methodologies and practices for composite
indicator development, EUR 20408 EN, European
Commission-JRC, Italy.
Laloë F., (2007), Information at the interface between
supply and demand for indicators: the use of a model
with estimated parameter values as a base for
indication in multicriteria analyses, International
Journal of Sustainable Development, 10, 33-45.
O’Connor M., Spangenberg, J., (2008), A methodology for
CSR reporting: assuring a representative diversity of
indicators across stakeholders, scales, sites and
performance issues, Journal of Cleaner Production,
16, 1399-1415.
Parris T.N., Kates R.W., (2003), Characterizing and
measuring sustainable development, Review of
Environment and Resources, 28, 559-586.
Pislaru M., Avasilcai S., Trandabat A., (2008),
Environmental sustainability based on fuzzy models,
Environmental Engineering and Management Journal,
7, 25-29.
Rey-Valette H., Laloë F., Le Fur J., (2007), Introduction to
the key issue concerning the use of sustainable
development indicators, International Journal of
Sustainable Development, 10, 4-13.
1149