Annals of Tourism Research, Vol. 30, No. 4, pp. 942­956, 2003
 2003 Elsevier Ltd. All rights reserved
Printed in Great Britain
0160-7383/$30.00
www.elsevier.com/locate/atoures
doi:10.1016/S0160-7383(03)00090-2
DECOMPOSING SEASONAL
CONCENTRATION
Antonio Fernández-Morales
Universidad de Málaga, Spain
Abstract: The main purpose of this article is to analyze seasonal concentration in tourism
demand series in three Spanish Mediterranean destinations by means of the Gini index,
decomposing it into inequality between and within seasons. This method is applied to the
monthly series of hotel nights, covering the period 1980­2001. The results suggest that the
"between" seasons component is the most significant one. In the most mature destination
this component is stabilized around 90% of the annual Gini index. In contrast, in the desti-
nation with the smallest Gini index, which offers a more diversified tourism product, the
between component only rises to the 70% and is still decreasing. Keywords: seasonal concen-
tration, Gini index.  2003 Elsevier Ltd. All rights reserved.
Résumé: La décomposition de la concentration saisonnie`re. Cet article a comme but princi-
pal l'analyse de la concentration saisonnie`re de la demande touristique dans trois desti-
nations méditerranéennes espagnoles en utilisant l'indice de Gini et en la décomposant en
parties inégales  entre  et  pendant  les saisons. La méthode est appliquée a` des séries
de nuits d'hébergement pendant la période 1980-2001. Les résultats sugge`rent que le com-
posant le plus significatif est celui qui est  entre  les saisons. Dans la destination la plus
mu^re ce composant s'est stabilisé vers 90% de l'indice annuel de Gini. Dans la destination
ayant l'indice de Gini le plus bas, par contre, du^ a un produit touristique plus diversifié, le
composant  entre  atteint seulement 70% et continue a` décroi^tre. Mots-clés: concentration
saisonnie`re, indice de Gini.  2003 Elsevier Ltd. All rights reserved.
INTRODUCTION
Seasonality is one of the most representative attributes of tourism
and it is often seen as a major problem that this industry has to face,
particularly in peripheral regions of Southern Europe, which base their
tourism product on sun and sea.
There are many definitions of this concept. Following Allcock (1994)
it is the tendency of tourist flows to become concentrated into rela-
tively short periods of the year. This characterization is based on the
annual distribution of the tourist flows, stressing the fact that it can
be found not only in the climatic conditions of the destination, but also
in originating markets. Along the same line, Butler (1994) describes it
as a temporal imbalance, which may be expressed in terms of elements
Antonio Fernández-Morales is Professor of Statistics at the School of Tourism of the Univer-
sity of Málaga (Departamento de Estadística y Econometría, 68, Campus de Teatinos, s.n.,
29071 Málaga, Spain. Email <afdez@uma.es>). His tourism research interests include measur-
ing service quality, economic impact studies, sustainable development and seasonality. He is
co-author of two books which cover statistical techniques for tourism analysis.
942
943ANTONIO FERNÁ NDEZ-MORALES
such as numbers or expenditure of tourists, traffic on highways,
employment, or admissions to attractions. At this point, it is important
to recall one of the distinctive features of seasonality: its regularity. In
this sense, BarOn (1999) considers that it gathers the effects occurring
each year with more or less the same timing and magnitude.
It is widely recognized that the main causes of seasonality are natural
and institutional (BarOn 1976). The former are originated by regular
fluctuations of the weather (temperature, rainfall, sunlight, daylight,
etc.) in a destination. In contrast, the latter are related to cultural,
ethnic, and social factors that affect several human activities like the
vacation. In addition, Butler (1994) cites other causes: social pressure
or fashion, sporting calendars, and inertia or tradition. Butler and Mao
(1997) distinguish between factors in origin and in receiving areas,
which do not act independently, but are interrelated and interact with
each other.
Seasonality is mainly perceived as a problem with serious impli-
cations for all aspects of supply-side behavior (Baum 1999). The prob-
lems more often mentioned are related to investment (difficulties in
securing the return of capital and under-utilization of physical capital)
and employment (seasonal unemployment and difficulties for the
employers to recruit and retain full-time staff). There are also adverse
effects on the environment, caused by the intensity of pressure on this
fragile resource and by its overuse during the peak period.
As a consequence of this problematic perception, much attention
has been paid and much effort made by both public and private sectors
to reduce seasonality through a variety of approaches (Butler 1994).
The main strategies to reach this objective are the variation in the
product mix and the use of price differentials (Allcock 1994). In the
case of Spain, Sutcliffe and Sinclair (1980) proposed deseasonalization
by the development of other cheap off-season holidays for old-age pen-
sioners; the provision of packages based on cultural events, tours of
historic sites, and sports such as skiing, horse riding, or golf; and the
promotion of business tourism by providing adequate facilities for
international conferences. These recommendations can still be seen,
some 20 years later, in many regional and local government policy pro-
grams.
The destinations analyzed in this article belong to the most southern
region of Spain: Andalusia. They are three of the eight provinces of
the region (the ones with the greatest numbers of tourists). They share
some features like the combination of coastal and interior areas and
the predominance of the sun and beach in their product, but have
important differences. The regional government is highly concerned
with the tourism industry as it contributes 13% to the regional GDP.
The most recent plan developed by the regional government to
improve the sector, Plan General de Turismo, mentions the problem of
seasonality and considers its reduction as a challenge.
There is another related area of interest in this field that needs
further investigation: the definition of seasons in particular desti-
nations. Uysal, Fesenmaier and O'Leary (1994) used a standard quar-
terly calendar breakdown to study the concentration of travel time in
944 DECOMPOSING SEASONALITY
the United States. They mentioned that research is needed to examine
the nature of travel during the different seasons, and that it is possible
that specific states may view their monthly configuration for a "season"
to be different. For example, a winter period may actually stretch
across one year into the next. In this sense, Allcock (1994) points that
it is impractical to assume that time can invariably be organized into
spring, summer, fall, and winter. This paper presents seasons as "soci-
ally significant periods of time rather than natural events", because the
evaluation of time is made with respect to tourist flows, which are
human cycles of activities.
There are different forms of seasonality; Butler and Mao (1997) dis-
tinguish among three forms: one-peak, two-peak and non-peak. The
first is typical of "Medsun" countries like Spain, Portugal, Greece, and
Cyprus. It is characterized by a significant peak period, three to four
months, usually in the summer. A very simple method for defining the
peak season is considering the months which systematically are over
the trend-cycle of the series, and hence their seasonal factors are
greater than one (Lim and McAleer 2001). But this approach can be
significantly improved, for example, by including the introduction of
a "shoulder" season. On the other hand, the definition for seasons can
change due to the evolutionary nature of the industry (Kennedy 1999),
and a measure of stability could be very useful.
Butler (1994) mentions that the literature about this topic has been
focused on the analysis of the demand patterns, the description of
particular destinations' seasonality, the negative effects on employment
and investment, and counter-seasonal policies and their implications.
In addition, he recognizes that there is little research on concepts or
theories and only a few specific works examine problems of measure-
ment, citing BarOn (1976) and Sutcliffe and Sinclair (1980).
The main contributions of the present paper are in this field, namely
measuring and decomposing the seasonal concentration of the hotel
demand in three Mediterranean Spanish destinations. To do so, the
evolution over the last 20 years of the monthly series of hotel demand
is analyzed. On one hand, the Gini index is used as a measure of the
annual level of concentration. On the other hand, through the appli-
cation of the additive decomposition of this coefficient proposed by
Dagum (1997), it is used for measuring this concept within and
between seasons, providing a measure of the stability of the seasons'
definition. A cluster analysis is also performed to validate the partition
of the year used in this work. Although the additive decomposition of
the Gini index is well known in the field of economic inequality, it has
not been applied to the tourism industry.
SEASONALITY IN THREE MEDITERRANEAN DESTINATIONS
In general, the Spanish tourism product is based on sun and sea, so
the industry is seriously affected by seasonality. Sutcliffe and Sinclair
(1980) found that the Gini index of arrivals increased by 45% between
1952 and 1975. One of their findings is that this trend was due to a
greater concentration of arrivals in certain months of the year and not
945ANTONIO FERNÁ NDEZ-MORALES
to an increase in fluctuations in the monthly pattern. Ramón and Abel-
lán (1995) in a more recent study which covers 1983 to 1993, state
that the Gini index was almost stabilized during those years. They also
found that the demand is concentrated in the summer months of July,
August, and September (which absorb 40% of the total).
The three destinations analyzed in this article are three contiguous
Spanish provinces facing the Mediterranean Sea: Málaga, Granada,
and Almería. They belong to the southern region of Spain, Andalusia.
Ramón and Abellán, in their study, mention that in Andalusia the dom-
estic hotel nights are concentrated in July and August, and the non-
resident hotel nights expand to the May­October period. The reasons
for this are the good climate, even in winter, the development of a
product based on the history and culture of Andalusian cities, and
improvements in communications and transportation.
In spite of the proximity, the degrees of seasonal concentration of
the three destinations analyzed in this paper are very different. Málaga
coasts are known as the Costa del Sol, a very important sun, sand, and
sea destination since the sixties. Of the hotel nights of the province,
97% correspond to the coastal area, and grew to 15.3 million in 2001.
This is the reason there is a high concentration ratio in Málaga. Recent
local policies that try to diversify the supply with golf and rural tourism
have not yet been fully successful in reducing this level significantly.
Málaga is a mature destination, with some of the negative effects of
mass tourism. The main resorts of the province--Torremolinos, Mar-
bella, and Fuengirola--reached the last stage of Butler's cycle (1980),
and show an almost saturation level in the peak period with its adverse
effects on the environment and on the image of the destination, as
Pollard and Domínguez Rodríguez (1993) noticed for Torremolinos.
The province of Granada shows a more diversified supply. The city
of Granada (at 50 km. from the coast) is a very attractive historic and
cultural destination. There is a ski resort in the Sierra Nevada, the
southernmost ski resort of Western Europe, and the beaches of the
province are the destination of a (mainly national) tourism of sun and
beach. In 2001, the number of total hotel nights in the province grew
to 3.7 million. Due to the diversified character of the supply, the sea-
sonal concentration in Granada is the smallest among the three stud-
ied areas.
The third destination is the province of Almería. This one is even
more specialized in sun, sand, and sea and has shown a spectacular
growth in recent years: from one million overnights in 1980 to 4.2
million in 2001. The seasonal concentration in this province is the
highest among the three. Almería experienced a late tourism develop-
ment, compared to other coastal areas of Spain, mainly due to the lack
of communications and infrastructure. The main resort of Almería,
Roquetas, ranks third in Andalusia in hotel nights and absorbs 60%
of the province total. Since 1990, the supply in this province grew at
higher rates than the average of the region, and the construction of
the Mediterranean and the A-92 highways (connecting Almería with
the rest of the Spanish Mediterranean coast and with the main interior
cities of Andalusia) contributed enormously to the expansion of tour-
946 DECOMPOSING SEASONALITY
ism in this destination. Almería, in spite of its short life is also begin-
ning to experience problems of saturation in summer months, and the
local and regional authorities are looking for solutions.
The three destinations face the problems of seasonal concentration,
with saturation in some of their resorts; and the local and regional
authorities, as well as the private sector agents, are concerned. As a
consequence, in recent years a close coordination has been observed
between municipal authorities and private enterprise in an effort to
implement measures to restructure the tourism product, as Priestley
and Mundet (1998) mention for Catalonia resorts. In this sense, it is
very important to get deeper quantified insight, by extending the meas-
uring of the demand, defining the seasons in every destination, and
investigating the stability and the concentration within and between
seasons.
Study Methods
The most common approach for measuring seasonality in tourism
consists of estimating seasonal factors in time series, using proportional
deviates to moving averages, or through dummy variables in multiple
linear regression. Examples of more advanced econometric models are
those of Sorensen (1999) who uses univariate ARIMA models for study-
ing hotel nights in Denmark; Ashworth and Thomas (1999) who apply
recent econometric tests to examine changes in the seasonality in
employment in the United Kingdom, and González and Mora (1996)
who use structural time series models with stochastic factors to analyze
international arrivals to Spain. These factors are often estimated over
a monthly basis. In addition to their descriptive use, they are often
used to deseasonalize series in order to study the changes in the trend-
cycle component of the series.
Because monthly or quarterly factors do not provide a synthetic mea-
sure of the annual level of seasonality, an annual single measure of
the extent of this phenomenon may permit the identification of those
years in which it has increased or decreased as well as providing infor-
mation about whether counter-seasonal policies had been effective or
not. Many statistical measures of dispersion and concentration can be
used to estimate the degree of seasonality, like the standard deviation,
the Theil and the Gini index. Although there is not enough empirical
evidence, it seems that all three give very similar results when estimat-
ing the changes in the degree of seasonality. Sutcliffe and Sinclair
(1980) applied these coefficients to monthly arrivals in Spain for the
period 1951 to 1976, and found correlations higher than 0.989 among
the three measures. In this study, the linear correlations among the
Gini and Theil indexes and the standard deviation in the three desti-
nations are all higher than 0.99.
The Gini coefficient is possibly the most widely used in this area.
Some authors who apply it are Sutcliffe and Sinclair (1980), Yacoumis
(1980), Lee (1996), Grabler (1997), Wo¨ber (1997), and Lee and Kang
(1998). In addition, this index can be decomposed in a very useful
947ANTONIO FERNÁ NDEZ-MORALES
way for assessing the degree of variation within and between seasons,
as well as providing a measure of transvariation between them.
In a monthly series Y, the Gini index (G) of an annual set of obser-
vations Y1, Y2, ..., Y12, can be expressed in various ways. The original
formula, Gini (1912), is shown in Table 1, where is the Gini mean
difference. This means that the mean absolute difference between all
the 12 × 12 pairs (Yi,Yj), and Y is the arithmetic mean. Continuous
variable G ranges from 0 to 1. But here the data set has a finite size,
12, and the range is (0, 0.9167). G equals 0 when the 12 data are the
same, Y1 = Y2 = ... = Y12, indicating that there an egalitarian distribution
over all months. On the other hand, the maximum value of G is
reached when 11 data are 0 and only one month has a non-null value.
Therefore, the higher the Gini index, the greater the degree of sea-
sonal concentration.
The former consideration reveals that this measure is not useful to
compare series with different periodicity. For example, the index of
Table 1. Additive Decomposition of the Annual Gini Index
Annual Gini index
G =
2Y
=
1
2
12
i=1
12
j=1
|Yi-Yj|
122
Y
Decomposition of the annual index G = Gw + Gnb + Gt
"Within" seasons component
Gw =
k
r=1
Gr,rprsr
"Net between" seasons component
Gnb =
k
r=2
r-1
t=1
Gr,t(prst+ptsr)Dr,t
"Transvariation" component
Gt =
k
r=2
j-1
t=1
Gr,t(prst+ptsr)(1-Dr,t)
Gini ratio between seasons r and t
Gr,t =
r,t
Yr+Yt
=
1
Yr+Yt
nr
i=1
nt
j=1
|Yi-Yj|
nrnt
Distance ratio between seasons r and t
Dr,t = (Yr-Yt)/ r,t,Yr >Yt
948 DECOMPOSING SEASONALITY
quarterly series varies between 0 and 0.75, while for monthly series the
range goes from 0 to 0.9167. This is due to the fact that indexes
obtained from quarterly data are implicitly imposing an egalitarian dis-
tribution within every quarter. Therefore, in practice, every variable
shows smaller Gini indexes for quarterly than for monthly data.
According to the observed monthly patterns, the year can often be
decomposed into well-defined seasons. They can exhibit a homo-
geneous internal distribution or, in the contrary, an important degree
of internal concentration. On the other hand, the differences among
them also affect the level of annual concentration. Hence, it is of great
interest to measure both contributions to the annual coefficient. These
measures can help to explain if a decrease in the value of G is due to an
internal homogenization of a specific season or to an approximation
between seasons.
In order to estimate the effect over the annual value of G of internal
and external differences among seasons, it is necessary to carry out a
mathematical decomposition of the index. Although there are several
ways to decompose it, an important topic in the literature of economic
inequality since the 70s, the one proposed by Dagum (1997) is used
in this work because it has very interesting features. With this decompo-
sition, it is possible to assess the inequality within the seasons, the
inequality among them, and how they contribute to the total (yearly)
concentration ratio, and to obtain a measure of transvariation
(overlapping) within this annual frame.
Following the decomposition of Dagum (1997), the Gini index of
an annual set of the monthly data that is partitioned into k seasons
(i.e., if the year is divided into low, peak, and shoulder, k = 3) of sizes
n1, n2, ..., nk ; i ni = 12 can be decomposed in the following additive
components: Gw, the contribution of the inequality within seasons to
the annual ratio G; Gnb, the net contribution of the inequality between
seasons; and, Gt, the contribution of the intensity of transvariation
between seasons. The formulae of these components are shown in
Table 1. A more detailed explanation can be found in Dagum (1997).
The component Gw indicates the part of G that can be attributed
to the internal concentration within each season. This component is
obtained as a weighted sum of the Gini indexes of each period (in the
formulas of Table 1, Gr,t stands for the Gini index between seasons r
and t, being Gr,r the internal index of the former r; pr for the ratio of
the number of months of season r to 12, and sr for the share of season
r in the annual value of Y). If the variable shows very similar values
for the months of each season (that is, very homogeneous periods),
this component will be small. The extreme case consists of seasons with
identical monthly values (for example, low: Y1 = Y2 = Y3 = Y4 = Y5 =
15, shoulder: Y6 = Y7 = Y8 = 35, and peak: Y9 = Y10 = Y11 = Y12 = 100),
which yields GL,L = 0, GS,S = 0 and GP,P = 0, and thus Gw = 0.
The other two terms of the decomposition need additional expla-
nations. Dagum (1997) calls Gnb the "net contribution" of the
inequality between groups to G. The magnitude of this component is
a measure of the contribution to the annual of the differences between
seasons. If very pronounced, the component Gnb will be responsible
949ANTONIO FERNÁ NDEZ-MORALES
for an important annual concentration index, but if there are few dif-
ferences, this component will be small.
The effect of overlapping is measured by Gt. Dagum calls this the
contribution of the intensity of transvariation to G (other authors call
it interaction or cross-over term). The concept of transvariation goes
back to Gini (1916). There is no null transvariation between two sea-
sons when at least one of the possible differences between an element
of one and that of the other has the opposite sign than the difference
between the means of their corresponding seasons.
The component, Gt, may be very useful to the analysis of seasonality.
It is an indicator of the stability of the definition of the seasons. A
stable classification of the months should show null transvariations year
after year, while no null transvariations components appear when the
seasonal distribution of the months changes over the observed period.
Although it is possible to obtain the decomposition of Table 1 using
alternative approaches, in this article, the method of Dagum (1997) is
followed in the calculations. The hotel demand is studied by means of
the monthly series of nights spent in every one of the three provinces.
The series of hotel nights are published by the Instituto Nacional de
Estadística (1999), and are an essential part of the survey Movimiento de
Viajeros en Establecimientos Hoteleros, conducted over all the national ter-
ritory.
Study Results
The evolution of the Gini index (G) of the monthly series over the
period since 1980 to 2001 can be seen in Figure 1. It shows that during
the 21 observed years, Granada was the destination with the lowest
index. In addition, this province began at the end of the 80s to reduce
even more its seasonal concentration, and now shows values around
0.10. Almería has also experienced a general decreasing trend over
the years, but remains with the highest values of G. In contrast, Málaga
shows a more erratic pattern. These trends are consistent with the gen-
Figure 1. Decomposition of Gini Indexes
950 DECOMPOSING SEASONALITY
eral one observed in the whole country. With a set of data from 1980
to 2001, it seems that there has not been any important change in the
monthly distribution of hotel nights in Spain, with the exception of a
slight reduction in the peak months. G exhibits a slight decreasing
trend during these years, from 0.24 in 1980 to 0.22 in 2001. In
addition, this coefficient in Andalusia was 0.19 in 1980 and 0.18 in
2001. Between these years, the maximum was reached in 1983 (G =
0.22) and the minimum in 1988 (G = 0.16).
There are important differences among the monthly patterns of the
three destinations. While Málaga and Almería exhibit a unimodal dis-
tribution, Granada has a bimodal one. In Málaga and Almería, the
factors for the winter months (November, December, January, and
February) are below 0.7 and for the summer ones (July, August, and
September) are over 1.3. It is very interesting to notice that summer
and winter in Almería and Málaga have the same pattern, but more
marked in Almería (especially in the summer). In contrast, in Granada
the factors are less dispersed. All of them are within the interval (0.7,
13). In this province, there is a peak in March, April, and May and
another relative peak in July, August, and September.
The partition of the year could be done using the common and
simple criterion of considering two seasons in such a way that the peak
period would include all the months with a factor greater than one
and the low one the rest of the year. In Málaga and Almería, the
national factors can be adopted to define two seasons. But it seems
more adequate to define a third shoulder season containing months
with factors close to one, like April (0.88), May (1.12), or October
(1.07), in spite of joining them with the peak or through months, like
January (0.57) or August (1.76). Hence, in the provinces of Málaga
and Almería, the year can be decomposed in three seasons: low, which
comprehends the winter months (November, December, January and
February), high, including the summer months (July, August, and
September), and finally "shoulder" (March, April, June and October).
This decomposition yields no null transvariations except for one year
in Málaga and four years in Almería.
If the national factors were taken to define two seasons in Granada,
the transvariation would have very high values (in 2001 Gt = 0.0293, a
31.19% of G). It is necessary to use specific seasonal factors in this
province. Even with its own specific factors, the pattern of Granada
does not allow three well-differentiated seasons. For this reason it is
preferable to work with only two seasons in Granada: low (November,
December, January, February, June, and July) and high (March, April,
May, August, September, and October), which contains the two relative
modes of the annual distribution.
A k-means non-hierarchical cluster analysis has been performed for
the series of hotel nights in the three destinations to confirm the estab-
lished year partition. To obtain a three seasons partition, it has been
necessary to take a four-clusters solution, due to the fact that a previous
hierarchical cluster analysis shows August in the three provinces as an
independent cluster. In Málaga and Almería, the low season corre-
sponds to the first cluster (November, December, January, and
951ANTONIO FERNÁ NDEZ-MORALES
February), the peak season to clusters three (July and September) and
four (August), and the months in cluster two (March, April, May, June,
and October) are classified as a shoulder season. In Granada, the peak
season includes clusters three (March, April, May, September, and
October) and four (August), and clusters one (November, December,
January, and February) and two (June and July) are joined as the low
season because the factors of these two groups of months are very simi-
lar in the observed years and show an important degree of trans-
variation between them.
The additive decomposition of the Gini index can be a very
important tool to get a deeper insight into the monthly distribution of
hotel demand in the three destinations. The evolution of the internal
indexes of every season and the indexes among them is presented in
Figure 2 for the years 1980 and 2001. In this pattern, although there
are several oscillations, the province of Málaga shows along the
observed period low values of the internal concentration ratios of the
three seasons, GL,L, GS,S, and GP,P are below 0.08. This indicates a rela-
tive internal homogeneity of the three seasons. In contrast, Almería
begins the period with high values of GS,S, and shows high values of
GL,L in the middle of the period. Only at the end of the 90s do the
values of these ratios become smaller. This is due to the fact that this
destination is not yet sufficiently mature and the seasonal patterns are
still changing, as opposed to Málaga.
The province of Granada has experienced an important decline in
the internal concentration ratios of both the peak and the low seasons.
The peak one shows a very low GP,P index (the smallest of all of them),
that corresponds to a peak season with a distribution of hotel nights
very close to the uniform distribution. However, the Gw in Granada
represents a percentage of G higher than the ones of Málaga and
Almería.
Turning attention to the Gini indexes between seasons, the greater
values are found in the province of Almería. Although there is a
Figure 2. Gini Indexes Between and Within Seasons
952 DECOMPOSING SEASONALITY
decline in the evolution of the ratios between low- and shoulder-peak,
GL,P and GS,L they are still high. GL,P was 0.5519 in 1980 and grew to
0.4592 in 2001. Something similar is observed in Málaga. These ratios
are high throughout the whole period, indicating a marked seasonal
pattern with few variations. In contrast, GL,P in Granada has experi-
enced a significant decrease. This fact, and the decrease of the within
component, Gw, are the causes of the very significant decrease of the
total Gini index. In this destination, the monthly pattern of the hotel
nights is evolving to be a more uniform distribution throughout the
year, not only within every season, but also between seasons.
The previous results help in understanding the decomposition of
the annual Gini indexes in the three destinations (Figure 1). In Málaga
and Almería, the most important component is the "between" one,
Gnb. However, Almería in 1980 showed a "within" component which
accounted for the 15% of G and reduced it to a 11% in 2001, while
in Málaga this component was around 10% during all the observed
period. It seems that Almería is evolving into a structure similar to
the one Málaga shows since 1980. These two destinations exhibit a
concentration structure with a very important "between" component
that accounts for the 90% of the annual ratio and the rest is due to
the "within" component. The null transvariarion component is a symp-
tom of a very stable definition of seasons. An immediate consequence
of this is that in these two resorts it will be difficult to change the
monthly distribution of the demand. The recent contra-seasonal
experiences in Málaga do not seem to have been successful yet.
The case of Granada is completely different. This province, with
lower values of G at the beginning of the observed period has been
successful in the further reduction of the concentration ratio. In
addition, in 2001 the "between" component was 70% of G, being this
an indicator of a monthly structure with a lower distance between sea-
sons than in the other two destinations.
CONCLUSION
Many Mediterranean resorts are trying to find solutions to the
important degree of seasonal concentration that they face. The three
Spanish destinations analyzed in this paper are among them. Although
they are in the same region of Spain and are contiguous, they exhibit
some important differences in their monthly patterns and struggle with
seasonality in different ways.
Some of the resorts in these destinations are in the two last stages
of Butler's product cycle: stagnation and post-stagnation. Hence, the
three provinces are beginning to consider some of the solutions men-
tioned by Montanari (1995), such as rediscovering the cultural and
natural riches that they possess, tourism in protected areas, archaeolog-
ical and historic cities, and others like promoting senior or golf mar-
kets, in order to redefine their product, in a way similar to the Agar-
wal's (1994) reorientation phase. The main adopted strategies can be
classified into Agarwall's (2002) product transformation in the coastal
resort restructuring. In addition, it is also important to notice the col-
953ANTONIO FERNÁ NDEZ-MORALES
laboration strategy in the three destinations, by means of local and
regional specific development plans.
But in practical terms it is crucial to analyze in detail the main fea-
tures of the seasonal concentration in every destination. The Gini
index is a powerful statistical tool for this purpose. This ratio provides
a synthetic measure of the annual level of concentration, which can be
very useful to evaluate the need of contra-seasonal policies in particular
destinations as well as in assessing the effectiveness of this type of pol-
icy. In addition, it also allows a decomposition, when the year is par-
titioned into seasons, that indicates the contribution to the annual
index of concentration of the internal concentration within seasons,
the differences among them and the contribution of the transvariation
(if it exists).
The study carried out in this paper has shown that among the three
examined Spanish destinations, Granada has experienced the most sig-
nificant reduction in the concentration of hotel demand. The combi-
nation of sun and beach, ski, and cultural tourism is a very interesting
example of how to reduce the seasonality. In 1980, the Gini index in
Granada was 0.17 and in 2001 it has been reduced to 0.09. The perfor-
med decomposition indicates that the inequality both between and
within the two seasons has decreased, but the most important
reduction is that of Gw. In other words, the two seasons are approximat-
ing between them and, but more importantly, the homogeneity within
them is increasing.
On the other hand, Málaga and Almería exhibit little variation in
their concentration ratios through the 20 observed years. However, the
evolution of their components has been different in each province.
Málaga is a mature destination and the components of the index are
very stable. In contrast, Almería is a younger destination and the
observed decomposition shows important changes. The inequality
within seasons has decreased notably, while the inequality among them
has increased. Therefore, the pattern in Almería is evolving towards a
sharper scheme, like the one of Málaga, in which the greater part of
the inequality is attributed to the inequality among seasons and only
a minor part is due to the inequality within them.
As a consequence of the difficulties to reduce seasonality in Málaga,
and having in mind that Málaga is a mature destination, the local and
council authorities are trying to renovate and revitalize the attractive-
ness of the destination, especially in other markets that could reduce
the seasonal concentration. The observed evolution in the province of
Málaga indicate that the contra-seasonal policies adopted until now
have had little success. The most important ones have been the pro-
motion of golf and senior markets. Recently, there has been an extra-
ordinary development of golf, that concentrates in the winter months,
and accounts for approximately 4% of the number of tourists. The
other main market often mentioned as contra-seasonal in this desti-
nation is the senior one. In Spain, through programs of social tourism,
there have been approximately 400,000 tourists annually, and the coast
of Málaga is one of the favorite destinations in the winter.
The council of Málaga city, the main city of the province, which has
954 DECOMPOSING SEASONALITY
600,000 inhabitants and is now out of the tourist circuit, is involved in
a redefinition of its role within the flows to the province. In 2003 it has
planned the opening of the Picasso Museum, with over two hundred
paintings, donated by Christine and Bernard Ruiz-Picasso, daughter-
in-law and niece of the artist. The authorities expect that this museum,
together with a new contemporary arts center and the interesting his-
toric part of the city (which includes the remains of a Roman theatre
and an Arab fortress) will include Málaga in the circuits of historic
cities of Andalusia (Seville, Granada, and Cordoba). The Picasso
museum could produce a similar, or even superior, effect to that
observed in Bilbao with the opening of the Guggenheim Bilbao
Museum, because the city has little else to offer (Plaza 2000). At this
point, it is important to notice that cultural and museum markets could
not only increase the number of tourists, but also reduce the seasonal-
ity of this destination. However, the magnitude of this effect will
require more research.
In the case of Almería, it is even more important to fight against
seasonal concentration, due to the water deficit in the province. Even
though all the Andalusian region has a structural water deficit, it is
more worrying in Almería, where the concentration is not only sea-
sonal, but also geographical in a very few resorts. The main actions in
this line are directed towards the solution of two main problems; the
underuse of the natural richness and diversity besides the beaches, and
the lack of complementary supply, aggravated by the predominance
of domestic (70%) over international demand.
Finally, there is still much to do for the reduction of the seasonal
concentration of tourism demand in these three destinations, in order
to alleviate the problems that it causes. To move in this direction, it
is important to implement adequate statistical measures that help in
the understanding of the seasonality phenomenon and its peculiarities
in every destination. A
REFERENCES
Agarwal, S.
1994 The Resort Cycle Revisited: Implications for Resorts. In Progress in Tour-
ism Recreation and Hospitality Management, C. Cooper and A. Lockwood,
eds., pp. 191­208. London: Belhaven.
2002 Restructuring Seaside Tourism: The Resort Lifecycle. Annals of Tourism
Research 29:25­55.
Allcock, J.
1994 Seasonality. In Tourism Marketing and Managenent Handbook, S. Witt
and L. Moutinho, eds., pp. 86­92. New York: Prentice Hall.
Ashworth, J., and B. Thomas
1999 Patterns of Seasonality in Employment in Tourism in the UK. Applied
Economics Letters 6:735­739.
Baum, T.
1999 Seasonality in Tourism: Understanding the Challenges. Tourism Eco-
nomics 5:5­8.
Bar-On, R.
1976 Seasonality in Tourism. London: Economics Intelligence Unit.
1999 The Measurement of Seasonality and its Economic Impacts. Tourism Eco-
nomics 5:437­458.
955ANTONIO FERNÁ NDEZ-MORALES
Butler, R.
1994 Seasonality in Tourism: Issues and Problems. In Tourism: The State of
the Art, A. Seaton, ed., pp. 332­340. Chichester: Wiley.
Butler, R., and B. Mao
1997 Seasonality in Tourism: Problems and Measurement. In Quality Manage-
ment in Urban Tourism, P. Murphy, ed., pp. 9­23. Chichester: Wiley.
Dagum, C.
1997 A New Approach to the Decomposition of the Gini Income Inequality
Ratio. Empirical Economics 22:515­531.
Gini, C.
1912 Variabilita` e Mutabilita`. Memorie di Metodología Statistica, Vol. I: Variab-
ilita` e Concentrazione. Roma: Librería Eredid Virgilio Veschi.
1916 Il Concetto di Transvariazione e le Sue Prime Applicazioni. In Memorie
di Metodologia statistica. Vol. II: Transvariazione, C. Gini, ed., pp. 1­55.
Roma: Librería Goliardica.
González, P., and P. Mora
1996 Analysis of Tourism Trends in Spain. Annals of Tourism Research
23:739­754.
Grabler, K.
1997 Cities and the Destination Life Cycle. In International City Tourism, K.
Grabler, G. Maier, J. Mazanec and K. Wo¨ber, eds., pp. 54­71. London: Pinter.
Instituto Nacional de Estadística
1999 Movimiento de Viajeros en Establecimientos Hoteleros. Madrid: Instituto
Nacional de Estadística.
Kennedy, E.
1999 Seasonality in Irish Tourism: 1973­1995. Tourism Economics 5:25­48.
Lee, C.
1996 Input-output Analysis and Income Distribution Patterns in the Tourism
Industry in South Korea. Asia Pacific Journal of Tourism Research 1:35­49.
Lee, C., and S. Kang
1998 Measuring Earnings Inequality and Median Earnings in the Tourism
Industry. Tourism Management 19:341­348.
Lim, C., and M. McAleer
2001 Monthly Seasonal Variations: Asian Tourism to Australia. Annals of Tour-
ism Research 28:68­82.
Montanari, A.
1995 The Mediterranean Region: Europe's Summer Leisure Space. In Euro-
pean Tourism, A. Montanari and A. Williams, eds., pp. 41­65. Chichester:
Wiley.
Plaza, B.
2000 Guggenheim Museum's Effectiveness to Attract Tourism. Annals of Tour-
ism Research 27:1055­1058.
Pollard, J., and R. Domínguez Rodríguez
1993 Tourism and Torremolinos: Recession or Reaction to Environment?
Tourism Management 14:247­258.
Priestley, G., and L. Mundet
1998 The Post-stagnation Phase of the Resort Cycle. Annals of Tourism
Research 25:85­111.
Ramón, A., and M. Abellán
1995 Estacionalidad de la Demanda Turística en Espan~a. Papers de Turisme
17:47­72.
Sorensen, N.
1999 Modelling the Seasonality of the Hotel Nights in Denmark by Country
and Nationality. Tourism Economics 5:9­23.
Sutcliffe, C., and M. Sinclair
1980 The Measurement of Seasonality within the Tourist Industry: An Appli-
cation to Tourist Arrivals in Spain. Applied Economics 12:429­441.
Uysal, M., D. Fesenmaier, and J. O'Leary
1994 Geographic and Seasonal Variation in the Concentration of Travel in the
United States. Journal of Travel Research 32(3):61­64.
956 DECOMPOSING SEASONALITY
Wo¨ber, K.
1997 International City Tourism Flows. In International City Tourism, K.
Grabler, G. Maier, J. Mazanec and K. Wo¨ber, eds., pp. 39­53. London: Pinter.
Yacoumis, J.
1980 Tackling Seasonality: The Case of Sri-Lanka. Journal of Tourism Manage-
ment 1:84-98.
Submitted 27 July 2000. Resubmitted 22 December 2001. Resubmitted 25 December 2002.
Resubmitted 13 March 2003. Accepted 30 April 2003. Refereed anonymously. Coordinating
Editor: René Baretje