Diapositiva 1 - Vienna University of Technology

download report

Transcript Diapositiva 1 - Vienna University of Technology

Composite Indices to Measure
Poverty and Social Inequality in
Europe
Matteo Mazziotta and Adriano Pareto
Istat - Methodological Office
Valentina Talucci
University of Rome “La Sapienza”
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Introduction
Aim of the work
Individuating
phenomena
individual
indicators
that
represent
the
Comparing different composite indices in order to find a
robust solution
“Designing” social inequality in Europe
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Definitions
Social inequality
“...refers to the ways in which socially-defined categories of persons
are differentially positioned with regard to access to a variety of
social “goods”, such as the labour market and other sources of
income, the education and healthcare system, and form of political
representation and participation.”
Poverty (UN Statement, June 1998)
“…is a denial of choices and opportunities, a violation of human
dignity. It means lack of basic capacity to participate effectively in
society. It means not having enough to feed and cloth a family, not
having a school or clinic to go to, not having the land on which to
grow one’s food or a job to earn one’s living, not having access to
credit. It means insecurity, powerlessness and exclusion of
individuals, household and communities. It means susceptibility to
violence, and it often implies living on marginal or fragile
environments, without to clean water or sanitation”
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Multidimensional phenomena
•
Social inequality and poverty are complex
•
They cannot be reduced to the income dimension
•
Multidimensionality has theoretical advantages but statistical
difficulties
•
Which is the better approach to measurement and evaluation?
Wien, february 24-26 2010
multidimensional
Conference on Indicators
and Survey Methodology
2010
Measuring social inequality and poverty - 1
From one to many dimensions
•
A growing consensus about going beyond income and per capita
product
– UNRISD, ILO, basic needs, PQLI
– Capability approach and Human Development
•
An increasing number of composite indices of development, wellbeing, QoL, …
(OECD Global Project “Measuring Progress of Societies”)
•
…but not of social inequality and poverty
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Measuring social inequality and poverty - 2
Working with many dimensions:
Theoretical and methodological problems
1. Choice of dimensions or of the “informational basis” (Sen,
1999) concept of justice or ethics
–
Choice of indicators
2. Use of the included information
–
Standardization/normalization
–
Space/time comparisons
•
Using “profiles”: pro and cons
•
Using composite indices: pro and cons
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Composite index: general aspects
Main steps
 selecting a group of elementary indicators, usually expressed in
different unit of measurement
 normalizing elementary indicators to make them comparable
 aggregating the normalized indicators by composite indices
(mathematical functions)
Problems
 finding data
 losing information
 researcher arbitrariness for:
 selection of indicators
 normalization of data
 choice of the aggregation function
Advantages
 unidimensional measurement of the phenomenon
 immediate availability
 simplification of the geographical data analysis
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Methods for composite index building
An alternative methodology: MPI
The method proposed by the authors wants to supply a synthetic
measure of a set of “non-substitutable” indicators.
The alternative composite index, called MPI (Mazziotta-Pareto Index),
starts from a linear aggregation and introduces penalties for the
geographical areas with “unbalanced” values of the indicators.
The steps to compute MPI are the following:
(i) normalization of the individual indicators by “standardization”;
(ii) aggregation of the standardized indicators by arithmetic mean with
penalty function based on “horizontal variability” (standardized values
variability for each unit). The penalty is based on the coefficient of
variation.
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Methods for composite index building
An alternative methodology: MPI
The standardization
Transforming each indicators in a standardized variables with mean equal to
100 and standard deviation equal to 10: the obtained values are usually
included in the range 70-130.
This standardization allows to solve the problem of the “ideal country”, because
the corresponding vector is composed by the mean values.
In this way, it is easy to individuate the geographical areas that are over the
mean value (values greater than 100) and the geographical areas that are
under the mean value (values smaller than 100).
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Methods for composite index building
An alternative methodology: MPI
The penalty function
The target is to penalize the geographical areas that present an “unbalanced”
set of indicators (for example, an indicator shows good result and another a
bad one).
The “horizontal variability” can be measured by the coefficient of variation (CV).
In this way, it is possible to penalize the “score” of each area (the mean of the
standardized values) by a directly proportional quantity to the CV.
The “penalty” can be add or subtract depending on the phenomenon nature
(development or poverty, wealth or social inequality).
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Methods for composite index building
Steps for computing MPI: Normalization
Being X={xij } the original data matrix, we denote Mx j and S x j the mean
and the standard deviation of the j-th indicator, where:
n
n
Mx j 
 xij
i 1
n
;
Sx j 
 ( xij  Mx j )2
i 1
.
n
The standardized matrix Z={zij } is computed as follows:
zij  100 
zij  100 
( xij  Mx j )
Sx j
( xij  Mx j )
Sx j
Wien, february 24-26 2010
10
if the j-th indicator is concordant;
10
if the j-th indicator is discordant .
Conference on Indicators
and Survey Methodology
2010
Methods for composite index building
Steps for computing MPI: aggregation
Let cvi be the coefficient of variation for the i-th unit:
cv i 
where
Sz
i
Mz
i
m
m
Mz 
i.
 (zij  Mz )2
 zij
j 1
Sz 
m
i
i
j1
m
The generalized form of MPI is given by:


MPIi /   Mzi 1  cv i  Mzi  Sz i cv i
2
where the sign ± depends on the direction of the indicators.
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Social inequality in Europe
ec.europa.eu/eurostat
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
The dimensions of social inequality
Wien, february 24-26 2010
The dimensions
SUBJECT
Social Inclusion
INDICATOR
At risk of poverty
rate
DEFINITION
Share of persons with an equivalised disposable income below 60% of the
national equivalised median income.
Equivalised median income is defined as the household's total disposable
income divided by its "equivalent size", to take account of the size and
composition of the household, and is attributed to each household
member.
Summary measure of the cumulative share of equivalised income accounted
for by the cumulative percentages of the number of individuals.
Its value ranges from 0% (complete equality) to 100% (complete
inequality).
Income
Gini coefficient
Education and
training
Share of persons aged 18 to 24 who have only lower secondary education
(their highest level of education or training attained is 0, 1 or 2 according to
Early school leavers the 1997 International Standard Classification of Education – ISCED 97)
and have not received education or training in the four weeks preceding
the survey.
Labour market
Long term
(including LFS unemployment rate
Labour Force Survey)
Total long-term unemployed population (?12 months; ILO definition) as a
proportion of total active population aged 15 years or more.
Age-related projections of total public social expenditures (e.g. pensions,
health care, long-term care, education and unemployment transfers),
current level (% of GDP) and projected change in share of GDP (in
percentage points) (2010-20-30-40-50)
Specific assumptions agreed in the AWG/EPC. See "The 2005 EPC
projections of age-related expenditures (2004-2050) for EU-25: underlying
assumptions and projection methodologies"
Social Protection
Projected Total
Public Social
expenditures
Health Care
Total self-reported unmet need for medical care in terms of number of
Self reported unmet
people who reported that at least once in the previous 12 months they felt
need for medical
they needed medical care and did not receive it either because they had to
care
wait, or it was too expensive, or it was too far away.
Conference on Indicators
and Survey Methodology
2010
Country
at Austria
be Belgium
bg Bulgaria
cy Cyprus
cz Czech Republic
de Germ any
dk Denm ark
ee Estonia
es Spain
fi Finland
fr France
gr Greece
hu Hungary
ie Ireland
it Italy
lt Lithuania
lu Luxem bourg
lv Latvia
mt Malta
nl Netherlands
pl Poland
pt Portugal
ro Rom ania
se Sw eden
si Slovenia
sk Slovakia
uk United Kingdom
The matrix of the individual indicators
At risk of
Gini coefficient
poverty rate
13
15
14
16
10
13
12
18
20
13
13
21
16
18
20
20
14
23
14
10
19
18
19
12
12
12
19
Wien, february 24-26 2010
25
28
24
29
25
27
24
33
31
26
27
34
33
32
32
35
28
39
28
26
33
38
33
24
24
28
32
Early school
leavers
Projected Total
Public Social
expenditures
19,7
33,4
74,1
30,9
9,8
16,7
18,3
11,5
51,8
20,4
33,5
41,3
22,0
35,0
48,9
11,6
34,8
15,5
74,1
27,7
14,3
72,9
74,1
16,0
18,6
11,4
27,5
3507
3421
734
1550
1447
3251
3169
846
2260
2523
3306
2283
1440
3126
2496
862
4153
860
1733
3192
843
2029
507
3012
1959
1130
2580
Long term
Self reported
unemployment unmet need for
rate
medical care
1,2
3,8
4,0
0,7
2,8
4,7
0,6
2,3
1,7
1,6
3,3
4,1
3,4
1,4
2,9
1,4
1,3
1,6
2,6
1,3
4,9
3,8
3,2
0,8
2,2
8,3
1,3
1,0
1,8
28,9
6,6
1,4
4,1
0,2
14,4
0,9
4,7
4,3
7,9
3,9
2,7
9,2
13,6
0,8
28,9
3,4
0,9
13,3
9,6
28,9
4,1
0,3
6,4
2,6
Conference on Indicators
and Survey Methodology
2010
Cograduation Spearman’s index
Arithmetic
mean
Arithmetic
mean
1
MPI
Geometric
mean
Wroclaw's
method
First
principal
component
Rizzi's
method
Weighted
mean of
factors
Wien, february 24-26 2010
Geometric Wroclaw's
MPI
mean
method
First
principal
component
Rizzi's
method
Weighted
mean of
factors
0,994
0,998
0,979
0,983
0,915
0,985
1
0,992
0,990
0,968
0,915
0,979
1
0,976
0,983
0,921
0,983
1
0,954
0,910
0,974
1
0,920
0,971
1
0,897
1
Conference on Indicators
and Survey Methodology
2010
Why using MPI?

Independence from the variability and measurement unit of the indicators

Independence from the “ideal unit”, since it is subjective, it is not univocal
and it can vary during the time

Non substitutability of the indicators

This methodology is not conditioned by the “versus” and by the “range” of
the elementary indicators

Easy computation

Easy interpretation (it is easy to individuate the geographical areas that
are over the mean value (values greater than 100) and the geographical
areas that are under the mean value (values smaller than 100)
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Level of social inequality
COUNTRY
Denmark
Sweden
Austria
Netherlands
Luxembourg
Slovenia
Finland
Czech Republic
France
Germany
Belgium
Ireland
Cyprus
United Kingdom
Hungary
Spain
Malta
Estonia
Italy
Lithuania
Slovakia
Greece
Poland
Bulgaria
Portugal
Latvia
Romania
Wien, february 24-26 2010
MPI
GROUP
89,90
90,97
91,09
91,44
93,35
93,73
94,24
94,50
96,28
96,69
97,52
98,90
98,95
99,52
102,14
102,29
102,62
104,43
104,59
105,39
105,40
106,58
107,48
109,95
109,98
112,59
114,19
1
1
1
1
1
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
4
4
4
4
LEVEL OF
SOCIAL
INEQUALITY
LOW
MEDIUM-LOW
MEDIUM-HIGH
HIGH
Conference on Indicators
and Survey Methodology
2010
The map of social inequality
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
Concluding remarks
The MPI is an alternative composite index based on the property of
“non-substitutability” of indicators that wants, in the scientific
outline, both to respect the desirable characteristics of a composite
index and to be validly applied to different scientific contexts
In fact, this methodology is not conditioned by the “versus” and by
the “range” of the elementary indicators
Therefore, the MPI can be a useful “tool” to synthesize
multidimensional phenomena (positive like development and
negative like social inequality)
The combination of the 6 individual indicators and the MPI
represents a new tool called SINCI (Social Inequality Composite
index)
SINCI is a robust measure of social inequality in Europe
Wien, february 24-26 2010
Conference on Indicators
and Survey Methodology
2010
References
Mazziotta M., Pareto A. (2007) “Un indicatore sintetico di dotazione
infrastrutturale: il metodo delle penalità per coefficiente di
variazione”, in: Atti della XXVIII Conferenza Italiana di Scienze
Regionali, AISRe, Bolzano.
De Muro P., Mazziotta M., Pareto A. (2009), “Composite Indices for
Multidimensional Development and Poverty: An Application to MDG
Indicators”,
Wye
City
Group,
FAO,
Roma.
http://www.fao.org/es/ess/rural/wye_city_group/2009/
OECD (2008) Handbook on Constructing Composite Indicators.
Methodology and user guide, OECD Publications, Paris.
Sen A., 1999, Development as Freedom, Oxford University Press,
Oxford
[email protected] [email protected]
[email protected]
Wien, february 24-26 2010