Discussion Paper Series, 13(6): 137-164
Past and present patterns of informal housing in
Greece: A spatial analysis
Serafeim Polyzos
University of Thessaly, Department of Planning and Regional
Development, Pedion Areos, Volos, Greece, 38334, tel.:+30-24210-74446.
e-mail: [email protected]
Dionysios Minetos
University of Thessaly, Department of Planning and Regional Development,
Pedion Areos, Volos, Greece, 38334, tel.:+30-24210-74276.
e-mail: [email protected]
Abstract
This paper deals with the primary causes of informal housing in Greece
as well as the observed differentiations in informal housing patterns
across space. The spatial scale of analysis is the prefectural
administrative level (NUTS III) and the methodology used in the
multivariate analysis is multinomial logistic regression. Results indicate
that Greek prefectures differ in the way they experience informal
housing phenomenon. Some justification for the observed differences
may be the separate development paths followed by each spatial unit
and the diverse range of economic activities in each prefecture. The
Greek state has not made provisions for creating the necessary "urban
land stock" in each prefecture, so that everyone interested can find land
parcels in an affordable price. On the contrary, the state follows the
informal housing activity by legalising large areas sporadically and by
introducing new legislative initiatives of limited success in dealing with
the problem. Current policies and institutional arrangements often have
failed to tackle the informal housing processes, in part due to false
assumptions about the causal relations that link the policies to the
phenomenon. The results can be used for improving land use policy
procedures as well as the strategic planning of land use regulatory
interventions.
Key words: informal housing, land uses, multinomial logistic regression,
housing policy, Greece
September 2007
Department of Planning and Regional Development, School of Engineering, University of Thessaly
Pedion Areos, 38334 Volos, Greece, Tel: +302421074462, e-mail: [email protected], http://www.prd.uth.gr
Available online at: http://www.prd.uth.gr/research/DP/2007/uth-prd-dp-2007-06_en.pdf
Past and present patterns of informal housing in Greece: A spatial analysis
139
1. Introduction
The term "informal housing" indicates a group of dwellings that have been created by
their owners without, in some way, following the legal construction procedures defined in
the state regulations. In such areas there is not place a formal urban plan and the
occupied land is either poorly-serviced or there is total lack of vital public infrastructure.
Informal settlements represent a growing part of many metropolitan concentrations in a
great deal of developing countries. Estimates based on current trends from the
developing areas of the world, indicate that such settlements will increase in the near
future providing shelter for as many as 50 per cent of the total urban population over the
next 10–20 years (UN-Habitat 2005). However, in most of the countries which face this
problem there is little strategic thinking about the integration of these housing clusters
into the city as a whole (Abbott and Douglas 2003). Therefore, the problem remains
unsolved and continuous to generate secondary adverse impacts on almost all aspects
of urban life (e.g. transportation, safety, social security etc).
The spatial distribution of informal housing phenomenon is uneven not only between
countries of the world but also amongst regions and provinces within the borders of a
single country. A lot of African, Middle East, Asian, and South and Central America
countries as well as the Caribbean are characterized by intense patterns of informal
housing (Baharoglu and Leitmann 1998; de Souza 2001; Gottdiener 1994; Leontidou
1993; Saff 1998; UN-Habitat 2005). Some Eastern and Southern European countries
are also places of increased informal housing activity. As regards the intra-national
variations in the spatial distribution of informal housing patterns, there are regions with
large rates of illegal construction activity and others that informal housing is not an issue
of concern. Regardless of the site specific circumstances that generate and fuel the
phenomenon, informal housing activity is tightly connected to some serious drawbacks
in the process of allocating the scare land resources in a sustainable way.
Encroachment of public land, unplanned large-scale changes in land uses, urban sprawl
and land segmentation are only some of the considerations that arise when the process
of informal housing advances. Subsequently, additional social considerations such as
poverty, immigration, social justice and equity are becoming even more pressuring.
Finally, the effectiveness of land use planning authorities becomes a major issue of
concern. Bureaucracy, time-consuming legal processes for acquiring developable land
or residence could all contribute to the (Abbott and Douglas 2003; Ferguson 1996;
Huchzermeyer 2003). Different combinations of socio-economic, institutional and
administrative parameters lead, in several countries, to a stage of unplanned urban
development and of randomly arranged informal residential units.
The experience so far has clearly indicated that the so-called “illegal route” in acquiring
developable land or constructing a housing unit makes it difficult to establish legality in
Discussion Paper Series, 2007, 13(6)
140
Serafeim Polyzos, Dionysios Minetos
the construction sector. On the other hand, the construction of informal settlements
causes serious environmental and social problems, not least because of the necessity
for providing the basic social infrastructure in order to relax some of the discomfort in
those areas (Abbott and Douglas 2003).
Informal housing in Greece constitutes a phenomenon with economic, social and
political dimensions. It is tightly connected to the kind of management placed on urban
and non-urban land uses by the state as well as the implemented housing policies. Past
and current housing policies in Greece have introduced specific instruments to address
informal settlements. At the launch of the first specific informal housing policy in 1983
significant mechanisms were set up to deal with the problem. However, the illegalhousing phenomenon proceeds at a high pace, so that about 3.000 unlicensed buildings
each year (almost the size of a small town) are legalised and integrated into the existing
urban system. Moreover, the annual number of illegal buildings that do not get into the
legalising process is believed to be much higher (Minetos et al., 2006; Polyzos and
Minetos 2007a).
This situation tents to become an acute problem with serious economic, social and
environmental implications. Impacts are great and pressuring ranging from aesthetic
deteriorations of landscape qualities, biotic diversity threats, desertification and forest
and open land “squeeze” to increase vulnerability to human settlements, local water
contamination, as well as to cultural degradation issues. In the majority of these
countries, it is recognised that the only way of improving the quality of life of residents in
these areas is through a process of in situ upgrading. However, the focus of this
upgrading tends to be directed at the level of the individual settlement; rarely is the
impact on the wider metropolitan area considered.
Given these considerations, the paper is organized into five main sections. Following the
introduction, section 2 describes the spatial dimension and the characteristics of the
phenomenon in Greece. The study area with regards to the spatial unit of analysis is
also presented. Section 3 provides a thorough description of the explanatory variables
which are used in the empirical model. The fourth section presents and evaluates the
multinomial logistic regression methodology employed in the empirical analysis,
describes the available data as well as the spatial configuration of informal housing
across Greek regions. It also attempts to interpret the parameters estimates yielded by
the analysis. The focus here is on approaching the underlying causes of informal
housing during the study period. The paper ends by commenting on the wide
implications of the phenomenon.
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
141
2. Causes, spatial dimension and characteristics of the
phenomenon in Greece
The issue of informal settlements is of significant importance worldwide and shares
similar implications on natural and urban environments in most of the regions of the
world. However, the type and quality of constructions may vary considerably between
countries (Kombe 2000; Kombe 2005). Attempting a short retrospection in the past, it
can be seen that as long as the Greek case is concerned, the building forms and driving
forces of informal housing have been extraordinarily different depending on the historical
context. In the past, informal housing activity was associated with some significant
urban-rural migration movements as well as the failure of the state to meet the extensive
demand for shelter by the incoming population.
Therefore, urbanization and net increases in the country’s population created pressuring
needs for new housing units. In addition, the income level of the people which where
newly-arrived at most major cities was relatively low and so did their ability of acquiring
a proper shelter. The increase in urban population usually fuels the demand for
available housing units in the real estate market raising urban land and property prices.
In turns, high land-plot prices are a serious obstacle for acquiring a house. The increase
in the available developable land through the expansion of existing urban plans is a
time-consuming process and therefore keeping the urban land prices at an affordable
level is difficult and requires effective monitoring procedures and land use planning
mechanisms. Under the prevailing past conditions of scarce economic resources,
neither the government nor the private sector could provide the urban poor with basic
shelter. Hence, informal housing activity is in part the direct result of the failure of
government approaches to housing and spatial planning policy.
The main reasons for this failure were: (a) the applied regional and economic policy
which fuelled rural-urban migration movements (b) failure to design and implement an
urban land policy that would have provided land plots of an affordable prices for the low
income social group (γ) the inability of the public sector to plan strategically and to
forecast urban land demand patterns.
Nowadays, informal housing in Greece is usually materialized in distant areas far away
from the borders of the formal city and town land-use plans. Up to now the various
legislative initiative launched by the state have not managed to contain the problem. The
sate policy towards informal settlements is still marked by great incoherence. Informal
settlements in Greece form a complex issue. These groups of housing units are
associated with a diverse population of a wide variety of social and economic
backgrounds. Therefore, informal housing units do not have the same characteristics
everywhere. In fact the individual cases have distinctive characteristics and can be
Discussion Paper Series, 2007, 13(6)
142
Serafeim Polyzos, Dionysios Minetos
classified into certain general categories. One such category concerns houses created
by low-income households and they are scattered all over the country both close to
urban centres and in ex-urban areas. A second category is made up by luxury vacation
or secondary houses, placed close to the coastal zone or to remote rural locations.
Finally, a third category encompasses the remaining cases such as illegal business
building of the primary sector (e.g farm buildings), the secondary sector (e.g small family
manufacturing companies) as well as the tertiary sector (tourism-related building all over
the country).
In Greece, the turning point for urban planning legislation and illegal settlement
construction can be traced back to 1983. In 1983, a significant piece of legislation was
introduced for dealing with wider urban land planning and management issues as well
as the phenomenon of informal housing (L.1337/83 1983). The law made provision for
integrating informal settlements into the existing urban system and for lowering the pace
of urban sprawl through the introduction of urban land use zones. As a result, a great
effort was made to survey and organise unregistered urban units that had emerged
since the post-war period (especially after 1955, a point that the Greek state introduced
a certain procedure for constructing a buildings through the requirement for issuing a
building license). By 1995, most of the «first generation of informal settlements» had
been legalised. However, it had already started the process of creating the «second
informal settlement generation».
In addition, remarkable changes have also occurred to the spatial distribution of illegal
housing. Whereas in the past most of the informal settlements were placed in suburban
areas of the great metropolitan concentrations, nowadays the majority of them are
developed in distant areas of great environmental value, close to the coastal zone or on
the islands.
Following, we attempt a spatial, quantitative analysis of the informal housing
phenomenon in Greece, using the existing statistical data acquired by official sources
and relevant studies. The study area covers a geographical area of 131.957,4 km2, has
a population of about 10.9 million people and a population density of 78/km2 (NSSG
2003). The country consists of 13 administrative regions. The regions are divided into 51
prefectures (fig. 1) or districts, which include 900 large municipal districts and 133 small
districts.
The mainland part of the country marks the end of the Balkan Peninsula whereas the
insular part (about 3.000 both habited and inhabited islands situated in the Aegean Sea
as well as the Ionian Sea) borders with Asia and Africa continents. Coastline stretches
for approximately 15,021 kilometers. About 72.8% of the people live in urban areas and
the remaining 27.2% being rural population (NSSG 2003). In mountainous areas live
9.2% of the population, whereas in semi-mountainous and urban areas the figures are
21,8% and 69.0% respectively (NSSG 2003). The country consists of 13 administrative
regions.
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
143
As regards land use distribution, agriculture and pastoral uses cover 49.5% of country’s
surface, forests, shrub and bare land cover 47.2%, inland water 1.3% and urban and
other artificial surfaces cover 2.0% (NSSG 2003). Geomorphologically speaking, most of
the mainland territory consists of mountains. Just a few major agricultural plains exist
the largest of which is placed in central Greece within the administrative boundaries of
the Thessaly region. Agricultural land use in the last decade (1991-2000) showed a
reduction mainly affecting cultivated land (Polyzos et al., in press).
Figure 1: Administrative boundaries of the Greek prefectures
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Agion Oros*
19
Ioannina
37
Lesvos
Achaia
20
Irakleio
38
Magnisia
Aitoloakarnania
21
Karditsa
39
Messinia
Argolida
22
Kastoria
40
Pella
Arkadia
23
Kavala
41
Pieria
Arta
24
Kefallinia
42
Preveza
Attiki
25
Kerkyra
43
Rethimno
Dodekanisos
26
Chalkidiki
44
Rodopi
Drama
27
Chania
45
Samos
Evros
28
Chios
46
Serres
Evrytania
29
Kilkis
47
Thesprotia
Evia
30
Korinthia
48
Thessaloniki
Florina
31
Kozani
49
Trikala
Fokida
32
Kyklades
50
Viotia
Fthiotida
33
Lakonia
51
Xanthi
Grevena
34
Larisa
52
Zakinthos
Ilia
35
Lasithi
Imathia
36
Lefkada
* This area has been excluded from the analysis
Discussion Paper Series, 2007, 13(6)
144
Serafeim Polyzos, Dionysios Minetos
3. Driving factors of informal housing activity
The quantitative analysis that follows in section 4 has been based on a number of
socioeconomic and environmental explanatory variables that are described shortly. This
analysis involves evaluating a stock of informal housing units in Greece that entered into
the legalisation process during the period from 1997 to 2006 against selective
economic, social and environmental factor.
•
Το total population potential of each prefecture (pot). For estimating the influence of
population on informal housing in the coastal zone we use both the "direct" or "self"
population potential (dpot) and the "indirect" population potential (ipot). The total
population potential is a function of a prefecture’s permanent population and the
distances between the prefecture under investigation and the remaining prefectures.
Therefore, it is connected to the interregional distances as well. The total population
potential of a prefecture is provided by the following formula (Polyzos and
Arambatzis 2006):
(pot)r=(dpot)r+(ipot)r=
Mr n Ms
+∑
drr s =1 d rsb
(1)
where:
Mr
=
a measurement of the volume of population “mass” or of economic activities
as expressed through GDP of the prefecture under investigation namely
prefecture r
Ms
=
a measurement of the volume of population “mass” or of economic activities
as expressed through GDP, of prefecture s, s=1,2…n.
drr
=
the mean intra-prefectural distance of the prefecture r (either in time or
length units)
drs
=
the distance between region r and the other regions s
b
=
An exponent which shows the “friction of distance” between precture r and
the other prefectures s
•
Population (pop) changes in each prefecture. Increasing local populations create
new requirements for housing and therefore have an influence on both legal and
informal housing activity in each prefecture. For the purpose of the present research
the statistical data regarding population changes in each prefecture for the study
period derive from the NSSG (NSSG 2004).
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
145
•
The share of tertiary (ter) sector of economy in the total production in each
prefecture. We expect that the prefectures whose economy is based on the tertiary
sectors of production would have higher economic growth rate compared to the
remaining prefectures and therefore they would experience increased demand for
housing units.
•
The urban population variable represents the change in urban population in each
prefecture. It is the ratio of urban population in the year 1991 to the urban population
of 2001. Data come from the NSSG (NSSG 1991, 2001). According to Zelinsky’s
rural - urban migration theoretical suggestion (Zelinsky 1971) there are certain
stages in migration according to the state that a society is. One of his suggested
stages involves the emergence of considerable rural-urban migration mostly in
societies experiencing developing processes. In this stage the flows migrants
increase considerably the demand for urban space in the expense of other uses.
This may was the case in Greece for the period shortly after the World War II which
was characterised by massive rural-urban migration movements (Leontidou 1989).
As the country get into the developed stage we might have the opposite
phenomenon of rural-rebound where people seek dwelling in peri-urban or even
rural location despite the fact that they work in cities.
•
The sandy beaches length (sbeach) in each prefecture is a measure of the total
length of the coastline and to some extent indicates the existence of suitable areas
in each prefecture for situating vacation and holiday houses. We assume that this
factor depicts better the potential in each prefecture for tourism development. The
existence of extensive scenic coastal locations is a factor of attraction for tourism
investment because of the economic benefits traditionally associated with tourism.
The effect of tourism growth on sandy-shore ecosystems through informal housing
of vacation units is intensified by the tendency of people to move to the coast for
permanent stay. Data for this variable come from the NSSG (2004).
•
The total area (area) added to the existing urban plans (legalised area) in each
prefecture for the period 1985-2003. It is obvious, that if the new areas that come
into the urban plans are equal or exceed the demand for urban space, urban land in
the real estate market will be of an affordable price. This is because there is a direct
relationship between city growth boundaries and affordable housing. Therefore, give
the availability of developable land in reasonable prices there would hardly be scope
for informal settlement construction if land prices is an important determining factor
of informal housing. An additional supportive reason for reduced informal housing
activity when there are available sufficient quantities of developable urban space is
the fact that within city boundaries there are certain advantages for investors such
as provision of public infrastructure and utilities fresh water distribution system,
sewerage, electricity etc. Building outside the boundaries of urban plans requires an
Discussion Paper Series, 2007, 13(6)
146
Serafeim Polyzos, Dionysios Minetos
additional significant cost to the individuals for acquiring such utilities. The statistical
data for this factor are taken from YPEHODE (2006).
•
Legal housing per capita 1990-2000: Legal housing per capita is a measure of the
housing activity in each prefecture. By employing this variable we want to
investigate whether the informal housing activity has an analogous relationship with
the legal (licensed) housing activity or they are uncorrelated. The data for this
variable are taken by the NSSG (2005).
•
Urban sprawl: The data for this variable come from the NSSG (1990; 2000) and
refer to the percentage of urban land outside the existing urban plans in each
prefecture for the year 2000. The indicator is the ratio of the number of building
constructed outside the existing urban plans in a prefecture multiplied with 100 to
the total number of buildings in that prefecture. This is a proxy variable for total
urban sprawl in each prefecture.
•
The nights spend by foreigner tourists in each prefecture (ftour) is a proxy variable
that may capture tourist attractiveness of each prefecture as regards foreign
tourism. In turns, this may affect the total demand for accommodation. An additional
variable is the nights spend by domestic tourists in each prefecture (dtour). This
variable is an indicator of tourist attractiveness of each prefecture as regards
domestic holidaymakers as well as for secondary and vacation housing
attractiveness. Data for this variable come from the NSSG (NSSG, 2002).
4. A quantitative analysis
4.1 Methodology
Much of the past and current research on the process of informal housing in developed
countries reflects a descriptive approach and orientation which results in aggregate
conclusions. This approach has the limitation of lacking firm theoretical grounding and
adequate quantitative justification. Being narrative in character it attempts to provide a
better understanding of informal settlements through painstakingly structured historical
interpretations and synthetic descriptions of events. Establishment of a systematic and
coherent analytical framework for studying informal settlements, based on sound
theoretical perspectives and quantitative modelling tools seems difficult due to lack of
reliable data on such housing activity. The development of robust modelling tools
needing medium quality data for investigation the proximate and underlying causes of
the process could possibly show new direction in informal housing research.
The methodology we choose in order to study the impacts of the aforementioned
variables on informal housing activity is multinomial (or polytomous) logistic regression
which is a variation of ordinary regression. It is used when the dependent variable is
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
147
categorical and the explanatory variables are continuous, or categorical (Lesschen et
al., 2005; Norusis 2004). The logistic regression directly estimates the probability of a
certain prefecture experiencing low, medium or high illegal housing activity subject to a
set of socio-economic and environmental factors. Thus the methodology can be used to
present data of illegal housing activity, to calculate the coefficients of the mathematical
formula and then apply the coefficients to grid cells representing districts, prefectures or
municipalities, yielding illegal housing activity surfaces (Lesschen et al., 2005).
In this study we take legalised housing units per 1000 residents per prefecture for the
period from 1997 to 2006 and use it as a proxy variable for the total illegal housing
activity in the prefectural level. For a number of pragmatic reasons, actual illegal
housing activity is extremely difficult to observe and count precisely. Therefore, we need
an observable variable that can be safely considered as indicative of illegal housing
activity volume. Proxy variables should be variables with good correlations to real
variables and they are used because the real variables are difficult to measure and
unobservable in practice. We set up a regression model in which one of the observed
variables is a proxy for some unobserved "true" variable. Therefore, this paper follows
the alternative approach of modelling illegal housing as an unobservable variable. In
order for the results to be valid, the proxy variable must have a close correlation with the
inferred value. Using early release data from the National Statistical Service of Greece,
we set off to find some evidence that illegal housing process is related to certain
socioeconomic and environmental driving forces. Because the legalised housing unit
indicator is relatively crude, we choose the multinomial logistic model that performs well
when fed with medium or even low quality data.
After transforming the continuous dependent variable into categorical with three classes
we form two logits (one logit less than the classes of the dependent variable) and apply
the model that is based on maximum likelihood estimations and not on the least-square
method as in multiple linear regression. The multinomial logistic model does not assume
that the relationship between the explanatory variables and the dependent variable is a
linear one (Norusis 2004). Furthermore, it does not assume homoscedasticity nor that
the dependent and independent variables or the error terms are distributed normally.
The only assumptions of the model are that the observations are independent and that
the independent variables are linearly related to the logits of the dependent. The benefit
of using a multinomial logistic model is that it models the odds of each subcategory
relative to a baseline category as a function of covariates. It can test the equality of
coefficients even if confounders are different unlike in the case of pair-wise logistics
where testing equality of coefficients requires assumptions about confounder effects.
Several studies in land use change literature adopted this methodology. Morita et al.
(1997) used a multinomial logit model to assess changes in land use by type in Japan.
The dependent variable was a four-dimensional vector of land-use ratios, representing
farmland, forest, built-up areas, and other areas. The study area was divided into cell
units and it was estimated the probability of choice of a particular land use type in each
Discussion Paper Series, 2007, 13(6)
148
Serafeim Polyzos, Dionysios Minetos
of the cells. In doing so, they used a group of explanatory variables concerning
economic, social and environmental factors. In a another study Chowdhury (1998)
applies multinomial logit regression in order to evaluate the socioeconomic and
biophysical forces driving deforestation processes in southern Yucatán peninsular
region, Mexico. In Sonoma County, California, Newburn et al. (2006) in a study
concerning purchasing land and easements for conservation purposes employ a
multinomial logit model to explain land-use transitions as a function of parcel site and
neighbourhood characteristics. Estimated coefficients from the multinomial logistic
regression are employed to predict the relative probability of land-use change, since the
site characteristics are known.
Instead of using the legalized housing units per prefecture data as a continuous variable
we transform it into a categorical variable in order to account for errors such as
undetected informal housing activity in each region and other errors involved in
recording the process (Kaimowitz and Angelsen 1998; Mahapatra and Kant 2005). The
magnitude of legalized housing units per prefecture is use as a dependent variable.
Therefore, in the dependent variable we have the following categories:
•
Low informal settlements prefectures where the per 1000 residents legalized
housing units value is from zero to 2.00 (0≤ legalized housing units per 1000
residents ≤2.00).
•
Medium informal settlements prefectures where the per 1000 residents legalized
housing units value is from 2.00 to 5.00 (2.00< legalized housing units per 1000
residents <5.00).
•
High informal settlements prefectures where the legalized housing units/1000
residents value is from 5.00 to the maximum value observed (5.00≤ legalized
housing units per 1000 residents≤14.17).
Therefore, we have three (3) categories with the Low category being the reference
category. The empirical model with ‘j’ categories of dependent variable can be
expressed as:
prob (i − class )
= e β i 0 .e β i 1 X 1 .e β i 2 X 2 ... e β in X n + ε i = e β i 0 + β i 1 X 1 + β i 2 X 2 + ... + β in X n + ε i
prob ( j − class )
⎡ prob(i − class) ⎤
ln ⎢
⎥ = β i 0 + β i1 X 1 + β i 2 X 2 + ... + β in X n + ε i
⎣ prob( j − class ) ⎦
where,
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
(2)
(3)
Past and present patterns of informal housing in Greece: A spatial analysis
149
prob(i-class))
= The likelihood the dependent variable being in the i category
prob(j-class)
= The likelihood the dependent variable being in the j category (the
baseline category)
Xn
= The explanatory variables 1,…n, employed by the empirical model
βi0
= The intercept for logit i
βin
= The regression coefficient of the variable n for logit i
When we have three classes of prefectures as regards illegal housing activity, there are
two non-redundant logits that can be expressed as following:
⎡ prob( ILLEGAL − Medium) ⎤
LogitA = ln ⎢
⎥ = β Med 0 + β Med 1 X 1 + ... + β Med 10 X 10 + ε med
⎣ prob( ILLEGAL − Low) ⎦
⎡ prob( ILLEGAL − High) ⎤
LogitB = ln ⎢
⎥ = β High0 + β High1 X 1 + ... + β High10 X 10 + ε High
⎣ prob( ILLEGAL − Low) ⎦
Therefore we calculate the parameters estimates for the above logits. The first logit
expresses the log of the ratio of the probability of a prefecture being in the “medium
illegal housing” class compared to being in the baseline class (i.e. low illegal housing
activity). Similarly, the second logit expresses the log of the ratio of the probability of
being in the “high illegal housing activity” class compared to being in the baseline class
(i.e. low illegal housing activity).
4.2 Data description
Looking at the histogram of the dependent variable in figure 2 we can see that the
distribution is not symmetric. There are two peaks on the left hand-side and also the
distribution is skewed to the right having a tail towards larger “illegal housing activity”
values. That is why we constructed 3 classes of illegal housing activity prefectures. The
first class represents the prefectures under the first peak that have smaller values of
illegal housing activity. The cut point here is 2.00. The second class represents the
prefectures under peak number two having a medium illegal housing activity. The cutpoint here is 5.00. Finally, the last class represents the remaining prefectures of high
illegal housing activity under the right tail of the distribution.
Discussion Paper Series, 2007, 13(6)
(4)
(5)
150
Serafeim Polyzos, Dionysios Minetos
Figure 2: Histogramic representation of the distribution of the dependent variable
Histogram
14
12
Frequency
10
8
6
4
2
0
0,00
3,00
6,00
9,00
12,00
15,00
Mean = 3,7502
Std. Dev. = 2,98542
N = 51
Legalised houses per 1000 residents per prefecture
1997-2006
Examining the stem-and-leaf plot and the box-plot in figure 3 we can acquire more
information about the right tail of the distribution. In the stem-and-leaf plot we can seen
that there are two prefectures of 7.6 and 8.8 values of illegal housing activity (high
values) and 3 outliers lying above 10.9. As regards the box-plot, the whiskers that
extend from the top and bottom of the box represent the smaller and largest values that
are not extreme values. The prefectures outside the whiskers are the outliers and they
are between 1.5 and 3 box lengths from the edge of the box. These prefectures have
the highest illegal housing activity and they are a) the insular prefecture of Chios (14.17)
in the Aegean Sea, b) the insular prefecture of Evia (11.95) close to the metropolitan
area of Athens and c) the mainland coastal prefecture of Laconia (10.89) in the south
part of Peloponissos.
Figure 3: Stem-and-leaf plot and box-plot of dependent variable
Legalised houses per 1000 residents per
prefecture 1997-2006
Frequency
Stem & Leaf
5,00
0 . 00135
13,00
1 . 0011245889999
8,00
2 . 01445578
3,00
3 . 559
8,00
4 . 22224566
7,00
5 . 0000379
2,00
6 . 27
1,00
7. 6
1,00
8. 8
3,00 Extremes (>=10,9)
Stem width:
Each leaf:
1,00
1 case(s)\
15
12
Chios
Evia
Lakonia
9
6
3
0
Legalised houses per 1000 residents per prefecture 1997-2006
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
151
4.3 Spatial overview of the classes
Figure 4 exhibits the geographic distribution of informal housing activity for the
prefectures classified in the “high informal housing activity” category. It is obvious that
most of the east coast of the country is designated as high informal housing activity
area. There are also two important rings of informal housing activity. The first is situated
around the great metropolitan area of Athens (i.e. around Attiki) consisting of the insular
prefecture of Evia and the mainland coastal prefectures of Argolida and Viotia. The
other ring of informal housing activity is found around the second largest metropolitan
area of Greek, the city of Thessaloniki. This ring is made up of the coastal prefectures of
Pieria, Imathia, Chalkidiki and Kavala as well as the hinterland prefecture of Kilkis.
The hypothesis here is that proximity to large urban concentrations results in high
informal housing activity in the neighbouring prefectures. Being a coastal prefecture is
also an additional significant factor of high informal housing activity.
Figure 4: Prefectures with high informal settlements activity. The legalized housing
units/1000 residents are between 5.00 and 14.17
Discussion Paper Series, 2007, 13(6)
152
Serafeim Polyzos, Dionysios Minetos
Legalised housing units per 1000 residents for the period 1997-2006
Athens Ring (blue line)
Prefecture
Value
Evia
11.95
Argolida
5.75
Viotia
5.09
--------Peloponissos South Coast
Prefecture
Value
Lakonia 10.89
Messinia 5.06
Thrace Coast
Prefecture
Value
Evros 5.09
Rodopi 6.28
Thessaloniki Ring (yellow line)
Prefecture
Value
Pieria
8.83
Imathia
7.63
Chalkidiki
6.72
Kilkis
5.38
Kavala
5.00
Thessaly Coast
Prefecture
Value
Magnisia
5.99
----Aegean Sea
Prefecture
Value
Chios
14.17
-----
Furthermore, figure 5 depicts the geographical distribution of prefectures with medium
illegal housing activity. Given the unobserved character of the mechanisms of informal
housing activity and the lack of reliable past data on this activity, it cannot be easily
determined whether the prefectures on this category are at the risk of climbing to the
high category of informal housing activity or at a state that any adverse effect will be
less likely in the future. Most of the west coast of the country, Peloponnesus and Crete
are in this category.
Figure 5: Prefectures with medium informal settlements activity. The legalized housing
units/1000 residents are between 2.00 and 5.00
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
153
Legalised housing units per 1000 residents for the period 1997-2006
Aegean Sea and Sea of Crete
Prefecture
Value
Kyklades
4.29
Chania
4.24
Irakleio
2.84
Lasithi
2.41
Ionian Islands
Prefecture
Value
Kerkyra
3.50
Kefallinia
2.53
----Sterea Ellada and Thessaly
Prefecture
Value
Trikala
3.57
Evritania
2.56
Karditsa
2.16
West Coast of Peloponnesus
Corinthians Channel
Prefecture
Value
Achaia
4.61
Arkadia
4.59
Korinthia
4.25
Ilia
2.40
West Greece
Prefecture
Value
Thesprotia
4.42
Ioannina
4.23
Aitoloakarnania
3.98
Aegean Sea
Prefecture
Value
Kozani
2.74
Drama
2.04
Finally, figure 6 presents the prefectures that form the “low informal housing activity”
category. The metropolitan areas of Athens and Thessaloniki belong to this group of
regions. In the ‘60s and ’70, these two areas used to be the hot-spots of informal
housing activity in Greece. Historically speaking, shortly after the Second World War
there was a rapid increase in migration of rural population to Greek major cities seeking
employment or due to political reasons. The flows of new residents towards the cities
were of such a magnitude that the state authorities were unable to cope with the
demand for residential land (Leontidou 1989; Leontidou et al., 2002; Maloutas 2000).
The lack of available land-plots in the central places of cities meant that the population
had to be accommodated elsewhere.
Nowadays, it seem that the pace of informal housing activity have been lowered in the
periphery of large urban concentration. However, there might has been a transformation
of the phenomenon itself. The new generations of informal settlements are not the
homes of the poor (Potsiou and Ioannidis 2006) but the result of land speculation activity
by an amalgam of actors such as middle class individual land owners, land investors,
building societies, investors in tourism infrastructure and upper and middle class owners
of luxury vacation houses. Therefore, the informal settlement phenomenon has been
transformed from an “obtaining a shelter” issue to an act of speculation.
In addition, remarkable changes have also occurred to the spatial distribution of illegal
housing. Whereas in the past most of the informal settlements were place in peri-urban
areas close to the major urban centres of the country, nowadays the majority of them
are developed in areas of great environmental value not in the suburbs but in the
greater neighbourhood of cities, close to the coastal zone or on the islands.
Discussion Paper Series, 2007, 13(6)
154
Serafeim Polyzos, Dionysios Minetos
Figure 6: Prefectures with low informal settlements activity. The legalized housing
units/1000 residents are between 0.00 and 2.00
Legalised housing units per 1000 residents for the period 1997-2006
The S-shaped line of Central Greece
Prefecture
Value
Larisa
1.92
Fokida
1.57
Fthiotida
1.44
Kastoria
0.36
Grevena
0.16
Florina
0.09
Metropolitan Areas
Prefecture
Value
Thessaloniki
1.85
Attiki
1.18
Trace Ionian Islands
Prefecture
Value
Lefkada 1.91
Zakinthos 1.11
North and South Aegean Sea and Crete
Prefecture
Value
Lesvos
1.88
Rethymno
1.05
Dodekanisos
0.51
Samos
0.09
--------West Greece
Prefecture
Value
Preveza
1.25
Arta
1.09
North Greece
Prefecture
Value
Serres
1.99
Xanthi
1.90
4.4 Evaluating the multinomial logistic model
The Likelihood-Ratio Test for the overall model can be seen in table 1. This measure
tests the hypothesis that all coefficients in the logistic model are 0. Because the
observed significance level is small (0.035), the null hypothesis that all coefficients for
the independent variables are 0 can be rejected. Therefore, we can conclude that the
final model is significantly better than the intercept-only model. The variation of the
dependent variable that is explained by the independent variables cannot be measured
directly in logistic regression models as it can be in multiple linear regression ones with
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
155
R2. However, the pseudo r-square statistics can provide an indication of explained
variation similar to the R2 of multiple linear regression models. These approximations
are computed and presented in table 1. Larger pseudo r-square statistics indicate that
more of the variation is explained by the model, to a maximum of 1. The Cox and Snell
R2 and the Negelkerke R2 are large enough. The Negelkerke R2 indicates that 54% of
the variation in the illegal housing activity is explained by the model. This percentage is
satisfactory as the values of logistic regression measures are almost always much
smaller than the corresponding ones for a linear model (Norusis 2005).
Table 1: Case Processing Summary, Model Fitting Information and Pseudo R-Square
Illegal
housing
activity
Case Processing Summary
N
Marginal
Percentage
0=Low
18
35.3%
1=Medium
19
37.3%
2=High
Valid
Missing
14
51
0
27.5%
100.0%
Total
51
Model Fitting Information
Model Fitting Criteria
Model
AIC
BIC
Intercept
115.211
Only
Final
126.421
Pseudo R-Square
Cox and Snell
Nagelkerke
McFadden
0.474
0.535
0.295
-2 Log
Likelihood
119.074
111.211
172.784
78.421
Likelihood Ratio Tests
ChiSquare
df
32.790
22
Goodness-of-Fit
Chidf
Square
Pearson
72.769
78
Deviance
78.421
78
The null hypothesis that the model adequately fits the data can be examined by the
Pearson and Deviance tests in the Goodness- of-Fit part of table 1. As it is evidence the
significance level is much greater that 0.05 so the null hypothesis that the model does
not adequately fit the data is rejected.
The Likelihood Ratio Tests presented in table 2 evaluates the contribution of each effect
to the model. It is a test for the individual coefficient and tests the hypothesis that the
coefficients are 0. This measure For each effect, the -2 log-likelihood is computed for
the reduced model; that is, a model without the effect. If the significance of the test is
small (less than 0.05 or 0,10) then the effect contributes to the model. This test can be
used instead of Wald test presented in table 4. This is because the Wald test sometimes
fails to correctly reject the null hypothesis when coefficients are large. The Wald test
performs well with large sample sizes. The significance values of the test for the
variables “Total population potential”, “Indirect population potential”, “Gross domestic
product in the tertiary sector”, “% urban population”, and “Length of sandy beaches” are
lower than 0,05 so we can conclude that they are important factors to the formation of
illegal housing activity. All other variables have large value of significance (more than
0,10), so we can consider them as not being important factors. Finally, the Akaike
information criterion (AIC) and the Bayesian Information criterion (BIC) are measures of
how “good” a model is compared to other model with different number of variables. The
model with the smallest value of Akaike criterion or alternatively Bayesian criterion is
Discussion Paper Series, 2007, 13(6)
Sig.
0.035
Sig.
0.646
0.465
156
Serafeim Polyzos, Dionysios Minetos
better. As it can be seen, the model that includes all variables has the smallest AIC
(123,139) as well as the smallest BIC (165,639).
Table 2: Model Fitting Criteria and Likelihood Ratio Tests for the individual logistic
regression coefficient
Effect
Model Fitting Criteria
BIC of
-2 Log
Reduced
Likelihood of
Model
Reduced
Model
124.662
167.162
80.662
131.725
174.225
87.725
AIC of
Reduce
d Model
Likelihood Ratio Tests
Chi-Square
df
Sig.
Intercept
2.241
2
0.326
Total population
9.304
2
0.010
potential
Indirect population
129.547
172.047
85.547
7.126
2
0.028
potential
Change in
124.740
167.240
80.740
2.319
2
0.314
population 19912001
Gross domestic
128.493
170.993
84.493
6.072
2
0.048
product in the
tertiary sector
% urban population
129.862
172.362
85.862
7.441
2
0.024
Length of sandy
131.280
173.780
87.280
8.859
2
0.012
beaches
Legalized area per
125.160
167.660
81.160
2.739
2
0.254
100 residents
New legal housing
125.308
167.808
81.308
2.887
2
0.236
area per resident
Urban Sprawl
123.730
166.231
79.730
1.310
2
0.520
Foreign tourists
123.350
165.850
79.350
0.929
2
0.628
overnight stays per
resident
Domestic tourists
123.139
165.639
79.139
0.719
2
0.698
overnight stays per
resident
The chi-square statistic is the difference in -2 log-likelihoods between the final model and a reduced model.
The reduced model is formed by omitting an effect from the final model. The null hypothesis is that all
parameters of that effect are 0.
The classification table shows that the model does very well in identifying the
prefectures that experience high illegal housing activity. Almost 65% of them are
classified correctly. In addition, the model classifies relatively well the prefectures with
low informal housing activity. %56 of the prefectures are correctly assigned. The model
does poorly in identifying prefectures with medium informal housing activity.
Table 3: Classification table
Observed
1
2
3
Overall Percentage
1
10
7
2
37,3%
2
6
8
3
33,3%
Predicted
3
2
4
9
29,4%
Percent Correct
55,6%
42,1%
64,3%
52,9%
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
157
4.5 Results and discussion
The parameter estimates table summarizes the effect of each predictor variable for the
two logits. The first logit is the ratio of the probability of a prefecture being in the
“medium informal housing activity” category than being in the “low informal housing
activity” category. Similarly, the second logits represents the probability of a prefecture
being in the “high informal housing activity” category than being in the “low informal
housing activity” category. The ratio of the coefficients to its standard error, squared,
equals the Wald statistic which is used for testing the null hypothesis that the individual
coefficient values are 0. If the significance level of the Wald statistic is small (less than
0.05 or 0.1) then the parameter is different from 0. Alternatively, there can be used the
Likelihood-Ratio Tests for the individual logistic regression coefficients (table 3).
Parameters with significant negative coefficients decrease the likelihood of that
response category with respect to the reference category whereas parameters with
positive coefficients increase the likelihood of that response category.
We start the analysis by investigating the type, magnitude and significance of the
relationship between current demographic patterns and development in Greece and
informal housing. Among the explanatory variables of the model, “total population
potential” (TPP) has a negative coefficient for both logits. However, the relationship
between TPP and informal housing activity is statistically significant only in the second
logits. The TPP is an indicator of population agglomerations in each prefecture and of
the total accessibility of each prefecture in relation to the other prefectures. It seems that
population agglomerations do not necessarily increase the probability of a prefecture
experiencing high informal housing activity. On the contrary, population agglomerations
decrease that that probability. Therefore, the most heavily populated prefecture have
less chances of getting informal housing activity. This was not the case in the past when
strong population movements from rural areas to Athens, Thessaloniki and other major
urban centres were associated with informal settlement creation. As it results from the
analysis of the rest of the variables underneath, the prefectures with large permanent
population choose neighbouring prefectures for the construction of vacation or
secondary housing units which in a lot of cases might be illegal construction. We
probably have a kind of rural rebound process where urban populations construct
informal settlements in rural areas.
A second demography-related variable is “indirect population potential” (IPP). This
variable seems to be a major causal factor of illegal housing activity in Greek
prefectures affecting directly the pace of the process. It is an indicator of the total
accessibility of each prefecture in relation to the other prefectures. The coefficient of this
variable has a positive sign and it is also significant in both logits. The effects of the
variable are similar in both categories of prefectures as regards informal housing activity
(namely Medium and High). However, the effects are more acute in the high/low logit.
With 1% rise in IPP, the likelihood of a prefecture being in the category of medium
Discussion Paper Series, 2007, 13(6)
158
Serafeim Polyzos, Dionysios Minetos
informal housing activity increases by a factor of 1.063, whereas being in the category of
high informal housing activity increases by a factor of 1.202. Hence, in both categories
of prefectures improved accessibility increase the likelihood of informal housing activity.
These results strengthen the initial expectation that the informal housing phenomenon
has undergone significant transformations in Greece. The new generations of informal
settlements are not the homes of the poor but the result of land speculation activity by
an amalgam of actors such as middle class individual land owners and upper and
middle class owners of luxury vacation houses. The informal settlement phenomenon
has been transformed from an “obtaining a shelter” issue to an act of speculation.
Prefecture like Evia, Argolida, Kavala, Viotia, Pieria and Chalkidiki (see figure 4) have
become the target areas of informal housing activity. These areas possess great natural
beauty and are highly accessibly by the metropolitan areas of Athens and Thessaloniki.
Figure 4 indicates that two distinctive rings of informal housing activity have been
formulated around these cities whereas figure 6 implies that there is relatively low
activity of illegal housing within the prefectures of the metropolitan areas. In the past,
Athens and Thessaloniki suburbs used to be the hot-spots of informal housing. It seems
that, nowadays, the spatial distribution of the phenomenon is somewhat different.
Perhaps, major improvements in transportation infrastructure (chiefly road infrastructure)
- which were achieved thanks to extensive structural funding from EU and through large
public investments after the early ‘90s - helped the phenomenon of informal housing to
migrate from the overloaded metropolitan suburbs to neighbouring prefectures.
For further understanding of demography-related effects on illegal housing activity we
use the variable “population change” in the period 1991-2001. The coefficient of the
variable has a positive sign for the first logit and a negative one for the second. This
means that an increase in population increases the likelihood for a prefecture being in
the medium informal housing category than in the low one but decreases the likelihood
of a prefecture being in high informal housing activity category. However, the level of
statistical significance is not satisfactory in any of the logits. Therefore, these results do
not allow us to make any firm inferences about the relationship between population
change and informal housing activity.
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
159
Table 4: Parameter Estimates
3Categories Legalised houses per 1000
residents per prefecture 1997-2006
(Banded)(a)
B
Std. Error
Wald
Sig.
Exp(B)
95% Confidence Interval
for Exp(B)
Lower
Bound
2
The
probability of
having
medium
informal
housing
activity to the
probability of
having low
activity
3
The
probability of
having high
informal
housing
activity to the
probability of
having low
activity
Intercept
Total population potential
Indirect population
potential
Change in population
1991-2001
Gross Domestic Product
(GDP) in the tertiary sector
%urban population
Length of sandy beaches
Legalized area per 100
residents
New legal housing area
per resident
Urban Sprawl
Foreign tourists overnight
stays per resident
Domestic tourists
overnight stays per
resident
Intercept
Total population potential
Indirect population
potential
Change in population
1991-2001
Gross Domestic Product
(GDP) in the tertiary sector
%urban population
Length of sandy beaches
Legalized area per 100
residents
New legal housing area
per resident
Urban Sprawl
Foreign tourists overnight
stays per
resident
Domestic tourists
overnight stays per
resident
Upper
Bound
-9.059
-0.025
0.061
9.066
0.020
0.047
0.999
1.655
1.694
0.318
0.198
0.193
0.975
1.063
0.938
0.969
1.013
1.166
7.138
8.149
0.767
0.381
1258.345
0.000
-0.041
0.051
0.650
0.420
0.959
0.868
10869468
617
1.061
0.042
0.482
0.184
0.033
0.363
0.239
1.680
1.765
0.592
0.195
0.184
0.442
1.043
1.620
1.202
0.979
0.795
0.752
1.112
3.299
1.922
0.009
0.079
0.012
0.914
1.009
0.863
1.178
0.253
-0.036
0.230
0.067
1.212
0.283
0.271
0.595
1.288
0.965
0.821
0.847
2.022
1.100
0.039
0.634
0.004
0.952
0.962
0.278
3.334
7.468
-0.117
0.184
15.702
0.055
0.083
0.226
4.539
4.940
0.634
0.033
0.026
0.890
1.202
0.799
1.022
0.991
1.414
-9.852
14.371
.470
0.493
5.27E-005
3.08E-017
89973224
-0.173
0.081
4.571
0.033
0.841
0.718
0.986
0.138
1.157
0.461
0.060
0.474
0.297
5.348
5.964
2.419
0.021
0.015
0.120
1.148
3.180
1.586
1.021
1.257
0.887
1.290
8.045
2.838
0.194
0.138
1.974
0.160
1.215
0.926
1.593
0.268
-0.255
0.320
0.376
0.700
0.461
0.108
0.497
1.307
0.775
0.698
0.371
2.446
1.619
0.726
1.001
0.526
0.112
2.067
0.291
14.700
a The reference category is: 1=low informal housing activity.
The coefficients of the share of the tertiary sector in the GDP inform us about the
relationship between the logits and the level of specialization in the service sector. For
both logits the coefficients are negative but statistically different from 0 only for the
second logits. These results indicate that the specialization of regional economy in the
tertiary sector appears to be negatively related to high illegal housing activity. Namely,
the prefectures whose economy mainly relies on the service sector do not suffer great
informal housing activity. Bearing in mind that a large part of informal housing activity
consists of vacation housing units, it is logical that these houses are locate in relatively
less developed prefectures with low specialization in the service sector.
Discussion Paper Series, 2007, 13(6)
160
Serafeim Polyzos, Dionysios Minetos
As regards the variable “percentage of urban population” the coefficient β has positive
signs in both logits but it not statistically different from 0 for the first logit whereas it is
statistically different for the second. This indicates that the urbanisation is not
significantly related to the separation of the medium and low informal housing activity
prefectures. However, it appears to be related to high informal housing activity
prefectures. Two possible explanations of this outcome is that (a) increased
urbanisation rates in a prefecture induces informal housing activity regarding manly
secondary housing and (b) the magnitude of urbanisation influence the real estate
market by increasing land prices. In turns, this decrease the likelihood of acquiring land
in low prices driving people to getting involved in informal housing activity.
The length of sandy beaches in each prefecture is an indicator of the total length of the
coastline and to some extent indicates the existence of suitable areas in each prefecture
for situating vacation and holiday houses. The positive and significant parameter
estimate in High/Low logit denotes that there is a positive relationship between on the
one hand, natural amenities (environmental quality, landscape aesthetics, natural
recreational resources, unspoiled environment etc) and on the other hand informal
housing activity. This is logical since as the length of sandy beaches increase so does
the relative attractiveness of an area for locating secondary and vacation housing. As
regards the Medium/Low logit, even though the relationship is positive, it seems that
there is not a strong association in place.
The total land surface added into the existing urban plans in each prefecture for the
period 1985-2003 (legalised area per 100 people) is a measure of the available
developable land. For the first logit the null hypothesis that the coefficient is zero can not
be rejected because of the level of significance. However, in the second logit, the
significance level is below 0.05. This indicates that the probability of a prefecture being
in the high informal housing category increase by a factor of 1.58 with 1% increase in
the legalised area per 100 residents. In other word, the legalisation process of former
illegal housing units and the addition of new developable land into the urban land-stock
do not restrain the phenomenon of informal housing.
From the observed significance level for both logits, the variable of “new legal housing
areas per resident” does not appear to be related to the medium or high level of informal
housing activity in relation to the low level because its coefficients are not significantly
different from zero. In other words the construction of new building units for residential
use does not seem to halt (or fuel) the phenomenon of informal housing activity.
Therefore, the supply of new housing units through the formal construction process can
not meet the particular piece of demand that is met by informal housing.
The variable of “urban sprawl” shows a positive relationship with the dependent variable
in both groups of prefectures. The parameter estimate is not statistically significant in
the first logit and it significant in the second at approximately 10% level of significance.
This result leads us to the conclusion that in prefectures of “high informal housing
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
161
activity” there is an association between the response and the predictor variable. Urban
sprawl brings basic infrastructure (roads, electricity lines etc) to ex-urban areas (i.e.
areas lying outside the formal urban plans). Subsequently, it is likely that the provision of
such infrastructures might attract informal housing as well. Ultimately, we come to the
point that it is not clear which one is the cause and which is the effect. More urban
sprawl means more informal housing and more informal housing means more urban
sprawl.
The last two variables of the empirical model concern tourism activity across the Greek
prefectures. They represent the nights spend by foreign and domestic tourists per
prefecture. Foreign tourism shows a negative association with informal housing in both
logits. However, the null hypothesis that the coefficient of the variable is significantly
different from zero can not be rejected due to the observed significance level. Even
though we cannot be sure whether there is a negative relation in place or no relation at
all, we can say that foreign tourism activity does not seem to produce informal housing
activity. In some respects, this is logical because foreign tourism is usually
accommodated in hotels which are licensed and regularly checked by the Greek
National Organisation of Tourism. Thus, although the scope for getting involved in
informal building activities is not eliminated, at least it is reduced.
On the other hand, the domestic tourism variable has a positive relationship with
informal housing in both logits although statistically significant only in the second logit at
approximately 11% level of significance. We can draw firm conclusion about this
relationship due to the observed significance level. However, it seem that there is a
positive association in place. The positive relationship can be interpreted that unlike
foreign tourism, the domestic tourism activity fuels illegal housing. In some respects this
is logical because regions that are attractive to domestic tourists are also attractive to
illegal housing if this housing mainly concerns vacation and secondary units. This is
further justified when bearing in mind that the prefectures of “high informal housing
activity: (e.g. Evia, Pieria, Chalkidiki) are situate close great metropolitan areas and
therefore they have large indirect population potential. This is in accordance with the
relationship between informal housing and direct population potential.
5. Conclusions
This empirical research has attempted to address the issue of informal housing in
Greece in a quantitative manner by providing some evidences on the likely causes of
the phenomenon. The main outcome of this study is that the processes in which lowincome households used to engage in order to obtain shelter for themselves might not
be longer the case in Greece. The new generations of informal settlements are not the
homes of the poor (Potsiou and Ioannidis 2006) but the result of land speculation activity
Discussion Paper Series, 2007, 13(6)
162
Serafeim Polyzos, Dionysios Minetos
by an amalgam of actors such as middle class individual land owners, land investors,
building societies, investors in tourism infrastructure and upper and middle class owners
of luxury vacation houses. Therefore, the informal settlement phenomenon has been
transformed from an “obtaining a shelter” issue to an act of speculation (Minetos et al.,
2006; Polyzos and Minetos 2007b).
Reflecting on the results of this empirical analysis, several conclusions can be drawn.
The first point is that the characteristics of informal settlements in Greece seem to be
different of those of third world countries as reported by the relevant international
literature. Geographically speaking, the spatial distribution of illegal housing has
changed. Whereas in the past most of the informal settlements were place in peri-urban
areas close to the major urban centres of the country, nowadays the majority of them
are developed in distant areas with large indirect population potential (e.g. Evia,
Chalidiki), great environmental value, close to the coastal zone (e.g. Evia, Argolida,
Pieria, Chalkidiki) or on the islands (e.g. Evia, Chios). This is a major difference
compared to the favelas informal housing in Brasil, gecekondus housing in Turkey,
informal urban settlements in South Africa, squatter problem in Kenya or slum areas in
India.
Secondly, in spite of the numerous legislative reforms the informal housing phenomenon
continuous to grow. For instance, the legalisation initiatives of the ’80 and ’90 added
considerable land into the existing urban plans and solved pressuring social and
environmental problems. However, all these legalisation initiatives as well as the
updating and restructuring of building-related procedures and the reform of land use
planning policy did not manage to tackle the problem. In most of cases the state
intervention regarding the integration of new space into the existing urban plans virtually
followed the informal housing process instead of going before. Hence the increase in
urban space did not precede but followed the demand already met by the process of
illegal housing.
Containing or eliminating informal housing activity presupposes the formulation of a
sustainable, integrated land use policy and the establishment of effective regional and
urban land use planning mechanisms. Major issues that need to be taken care of
include, complex and insufficient legal framework, the inappropriate planning and land
use allocation provisions and procedures, shortage of available well planned and
organised land for vacation housing, insufficient control mechanisms, low political will
and commitment and land speculation.
UNIVERSITY OF THESSALY, Department of Planning and Regional Development
Past and present patterns of informal housing in Greece: A spatial analysis
163
Bibliography
ABBOTT, J. and DOUGLAS, D. (2003), The use of longitudinal spatial analyses of informal
settlements in urban development planning, Development Southern Africa, Vol. 20, No. 1, pp. 319.
BAHAROGLU, D. and LEITMANN, J. (1998), Coping strategies for infrastructure: How Turkey's
spontaneous settlements operate in the absence of formal rules, Habitat International, Vol. 22, No.
2, pp. 115-135.
DE SOUZA, F.A.M. (2001), The future of informal settlements: lessons in the legalization of
disputed urban land in Recife, Brazil, Geoforum, Vol. 32, No. 4, pp. 483-492.
FERGUSON, B. (1996), The environmental impacts and public costs of unguided informal
settlement; the case of Montego Bay, Environment and Urbanization, Vol. 8, No. 2, pp. 171-194.
GEOGHEGAN, J., PRITCHARD, J., OGNEVA-HIMMELBERGER, Y., CHOWDHURY, R.R.,
SANDERSON, S. and TURNER II, B.L. (1998), Socializing the Pixel and Pixelizing the Social in
Land-Use and Land-Cover Change, in LIVERMAN, D., et al. (Eds.), People and Pixels. Linking
remote sensing and social science, National Academy Press, Washington D.C, pp. 51-69.
GOTTDIENER, M. (1994), The Social Production of Urban Space, University of Texas Press,
Texas
HUCHZERMEYER, M. (2003), A legacy of control? The capital subsidy for housing, and informal
settlement intervention in South Africa, International Journal of Urban and Regional Research,
Vol. 27, No. 3, pp. 591-612.
KAIMOWITZ, D. and ANGELSEN, A. (1998), Economic Models of Tropical Deforestation: A
Review, Center for International Forestry Research, Jakarta, Indonesia.
KOMBE, W.J. (2000), Regularizing housing land development during the transition to market-led
supply in Tanzania, Habitat International, Vol. 24, No. 2, pp. 167-184.
KOMBE, W.J. (2005), Land use dynamics in peri-urban areas and their implications on the urban
growth and form: the case of Dar es Salaam, Tanzania, Habitat International, Vol. 29, No. 1, pp.
113-135.
L.1337/83 (1983), Urban plans extension, urban development and relevant arrangements,
Journal of the Greek Parliament, ΦΕΚ-33/Α/83.
LEONTIDOU, L. (1989), Cities of silence: Working class colonisation of urban space, Athens and
Pireaus 1909–1940, Cultural and Technological Institution of the National Bank of Industrial
Development of Greece (ETBA), Athens.
LEONTIDOU, L. (1993), Postmodernism and the City: Mediterranean Versions, Urban Studies,
Vol. 30, No. 6, pp. 949-965.
LEONTIDOU, L., AFOUXENIDIS, A., KOURLIOUROS, E., RONDOS, K. and HATZICHRISTOS,
T. (2002), First Greek Annual Report: Causes of Urban Sprawl in Athens & East Attica, 19812001, HELLENIC OPEN UNIVERSITY.
LESSCHEN, J.P., VERBURG, P.H. and STAAL, S.J. (2005), Statistical methods for analysing the
spatial dimension of changes in land use and farming systems, LUCC Report Series 7, The
International Livestock Research Institute, Nairobi, Kenya and LUCC Focus 3 Office, Wageningen
University, the Netherlands.
MAHAPATRA, K. and KANT, S. (2005), Tropical deforestation: A multinomial logistic model and
some country-specific policy prescriptions, Forest Policy and Economics, Vol. 7, No. 1, pp. 1-24.
MALOUTAS, T. (2000), Social and Economic Atlas of Greece, EKKE and University of Thessaly
Press (in Greek), Volos.
MINETOS, D., POLYZOS, S. and SDROLIAS, L. (2006), Social and Public Responsibility and
Illegal Urban Land Uses in Greece: An Empirical Investigation, paper given at MIBLEManagement of International Business & Economic Systems Conference, Larissa-Greece, 4-5
Nov. 2006.
Discussion Paper Series, 2007, 13(6)
164
Serafeim Polyzos, Dionysios Minetos
MORITA, H., HOSHINO, S., KAGATSUME, M. and MIZUNO, K. (1997), An Application of the
Land-Use Change Model for the Japan Case Study Area; Laxenburg, Austria: International
Institute for Applied Systems Analysis, iii, 27.
NEWBURN, D.A., BERCK, P. and MERENLENDER, A.M. (2006), Habitat and Open Space at
Risk of Land-Use Conversion: Targeting Strategies for Land Conservation, American Journal of
Agricultural Economics, Vol. 88, No. 1, pp. 28-42.
NORUSIS, M. (2004), SPSS 13, Advanced statistical procedures companion, Prentice Hall Inc,
New Jersey.
NORUSIS, M. (2005), SPSS 14.0 Statistical Procedures Companion, Prentice Hall, New Jersey,
USA.
NSSG (1990), Census of buildings, Hellenic Republic, Athens.
NSSG (1991), Population census, Hellenic Republic, Athens.
NSSG (2000), Census of buildings, Hellenic Republic, Athens.
NSSG (2001), Population census, Hellenic Republic, Athens.
NSSG (2003), Statistical Yearbook of Greece, Athens.
NSSG (2004), Statistical Yearbook of Greece, Hellenic Republic, Athens.
NSSG (2005), Yearly Data on Construction Activity in Greece, Hellenic Republic, Pireas.
POLYZOS, S. and ARAMBATZIS, G. (2006), Labour productivity of agricultural sector in Greece:
determinant factors and interregional differences analysis, Mediterranean Journal of Economics,
Agriculture and Environment, Vol. 1, No., pp. 58-64.
POLYZOS, S. and MINETOS, D. (2007a), Causes and management implications in the Greek
coastal zone from Informal settlements development, Proceedings of the International Conference
on Environmental Management, Engineering, Planning, and Economics, vol. I, eds KUNGOLOS,
A., et al., Skiathos, 24-28 June, Greece, pp. 727-732.
POLYZOS, S. and MINETOS, D. (2007b), Causes and management implications in the Greek
coastal zone from Informal settlements development, CEMEPE 1st International Conference on
Environmental Management, Engineering, Planning, and Economics, ed. KUNGOLOS, A.,
Skiathos, 24-28June, Greece.
POLYZOS, S., CHRISTOPOULOU, O., MINETOS, D. and LEAL FILHO, W. (in press), An
overview of urban-rural land use interactions in Greece, Int. J. Agricultural Resources,
Governance and Ecology,, Vol., No., pp.
POTSIOU, C. and IOANNIDIS, C. (2006), Informal Settlements in Greece: The Mystery of Missing
Information and the Difficulty of Their Integration into a Legal Framework, paper given at 5th FIG
Regional Conference: Promoting Land Administration and Good Governance, Accra, Ghana.
SAFF, G. (1998), The Effects of Informal Settlement on Suburban Property Values in Cape Town,
South Africa, The Professional Geographer, Vol. 50, No. 4, pp. 449-464.
UN-HABITAT (2005), Addressing the challenge of slums, land, shelter delivery and the provision
of and access to basic services for all: Overview, UN-Habitat Regional Office for Africa and Arab
States.
YPEHODE (2006), Unpublished data on the expansion of urban plans.
ZELINSKY, W. (1971), The Hypothesis of the Mobility Transition, Geographical Review, Vol. 61,
No. 2, pp. 219-249.
UNIVERSITY OF THESSALY, Department of Planning and Regional Development