You are on page 1of 21

ijcrb.webs.

com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

Multidimensional poverty in Tunisia: a non-monetary approach applied to East-Central Tunisia


Guiga Housseima Doctor in Economic Sciences, Faculty of Economic Sciences and Management of Tunis, Tunis El Manar University-TUNISIA Ben Rejeb Jaleleddine Professer in Quantitatives Method. Institut Suprieur de Gestion de Sousse.

Abstract
This work has the aim of constructing a Composite Poverty Indicator (PCA), on which the identification of the poor and measurements of the povertys indices will be based. Then, to analyze the profile of poverty of under groups of the identified households (poor, average and rich person). The study is based on the data analysis provided by the investigation into consumption, budget and the standard of living of the households in 2005. This survey, conducted by the National Institute of Statistics of Tunisia (INS), relates to sample of 2312 households in East-Central Tunisia. The data used relate to the socioeconomic characteristics of households such as housing, education, access to social services ... We have adopted the Multiple Correspondence Analysis for the calculation of the ICP. Then, we used software DAD (version 4.3) for the estimate of the poverty indices and the decomposition of the latter according to the various household characteristics. The classifications results show that poor households are characterized by the lack of comfort goods and the difficulty of access to social services, while the rich, they are characterized by a well developed pattern of settlement and possession of appliances. Regarding the decomposition of poverty, we note that this one touches more the households directed by a man, who do not have a higher educational level and who are aged from 35 and 54 years. This poverty is more localized in Sfax and which contributes more to average poverty of the sample. The result of estimate of the binary model logit validates the result got by AMC. Indeed, the probability of being poor depends on the educational level, of the CSP of the households head and the access to social services. Classification JEL: C38, I31, I32 Key words: Multidimensional Poverty, Poverty Composite Indicator, Multiple Correspondence Analyses. 1. Introduction In developing country, the improvement of the living conditions of the population constitute the vital objective of any policy centered towards the planning of economic development and social, Tunisia, like all other developing Countries, aims a series of reforms to fight against poverty. Since poverty became a problem in the majority of these countries, this calamity nowadays is as subject of debate relating especially to the policies able to absorb it. So the measurement of poverty is a current subject. It is currently recognized that the income is not the only factor on which depends the individual deprivation and which poverty is a complex and multidimensional phenomenon which is related to the several other factors like housing, instruction, demography, infrastructure, etc. Because poverty is multidimensional, so it is important for researchers to take into account the multiple dimensions preventing an individual from having a decent and quiet life. Thus, we will use a multivariate analysis for better identifying the poor and the strategies allowing fighting it effectively. In a first part, we will present the conceptual framework of poverty, identifying the principal schools of thought in the conceptual approaches of poverty and measurements of multidimensional poverty based on axiomatic and non axiomatic approaches, in a second part. The third part presents variables and adopted methodology. The

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

856

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

fourth part will be devoted to the application. We will measure indices FGT of poverty and their decomposition according to various households characteristics. In the same way, we will identify the characteristics of different subgroups of households following a hierarchical classification based on CIP. The fifth part is interested in the binary logit model to estimate the probability of being poor according to the socio-economic characteristics of households. Finally, we finish this work by a conclusion on the various got results. 2. Conceptual framework for analysis of poverty There exist three principal schools of thought to the measure of poverty: the school welfare, the school of basic needs and the school for the capacities. These three schools seem to agree on the following point (Asselin and Dauphin, 2000) is considered poor any person who does not reach a minimum of reasonable satisfaction of a "thing". What distinguishes them is the nature and level of the minimum. 2.1. Welfare approach The utilitarian approach of neo-classic is based on the concept of economic welfare. It is based on the principle that each individual satisfies his wellbeing according to his preferences. The latter vary according to individual characteristics, needs, capacities and time. According to this approach, an individual is considered poor if he does not reach a level of economic well-being considered as reasonable minimum by the standards of the society where he lives. Indeed, the evaluation of individual wellbeing is based on the order of preferences revealed by his free choices. For example, a person could be considered poor in basing on total consumption, while he could be considered non-poor given his capacity of work. This is explained by the choice of this individual who chooses to work and spend little for, for example, the benefit of more than leisure. Freedom of choice creates heterogeneity of preferences, which are not directly observable, which limits their practical use for measure poverty. Indeed, the use of economic well-being faces two major problems. First is relates to the observation of the revealed preference. In fact, the categorization of individuals as poor and rich person is not linked only to demographic characteristics and living conditions. But it is necessary to refer to the level of utility gotten by an individual, if it is lower or not than a certain level of critical utility according to his perception. For that, it is necessary to have all information about the preferences of each individual, which their collection is difficult to realize. What makes the utilitarian approach impractical The second problem relates to measuring it "physical happiness" or the pleasure coming from the choice of consumer. In fact, the comparison of utility between individuals is difficult considering the preferences are heterogeneous and that personal characteristics, the needs and the capacities are varied. Similarly, households are differed according to the size and the composition and the prices vary in time and space. For that, comparing the well-being has no meaning. Moreover, this approach considers as poor a person who is not satisfied his needs, even if it is materially easy; as it may consider a person as not poor because he is happy and satisfied with its economic level, even if he is private some attributes. In practice, given that utility is a state obtained by a person following the consumption of goods and services, the utilitarian economists take as measures wellbeing the income or the households expenses. In fact, a household that does not reach an acceptable level of income according to the standards of his society is considered as poor. Therefore, low income is a criterion for identification of poverty. Nevertheless, it is to be recalled that poverty based on consumption or income does not take into account the well-being provided by public and non commercial goods, such as safety, freedom, peace and health. The latter are difficult to specify in practice. According to this approach, the policies of poverty reduction are focused on increasing income. Therefore, market forces are supposed to bring the fruits of poor economic growth, while taxation, social services and other state interventions of the State are favorable only to the distribution of the fruits of economic growth. Besides the utilitarian approach, there exists the non-utilitarian school which takes into account other dimensions.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

857

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


2.2. Non welfare approaches

JANUARY 2012
VOL 3, NO 9

According to the non-utilitarian approach, the level of welfare is in accordance with the standards and values of each society, regardless of perceptions of each individual. In contrast of the utilitarian approach which uses an aggregate indicator (income or utility) to analyze the standards of living, the non-utilitarian approach uses multiple dimensions of well-being. There are two main types of non-utilitarian approaches: the approach based on capabilities and the approach based on essential needs. 2.2.1. Capacities approach Capabilities represent the various combinations of functioning that an individual or household can achieve. These capacities are defined as being a functional combination of the knowledge-being and know how that each person must reach it and who allows him to have a definite type of life. For this purpose, the individual must be adequately nourished, get an education, be healthy, be adequately housed, take part in the community life, appear in public without shame, etc. This condition is sufficient to not treat a person as poor. Indeed, the missing "thing" is neither the utility nor the satisfaction of basic needs, but of human capacities or abilities. The capacity school considers as poor, a person who lacks capacity to achieve a certain subset of functioning. According to this approach, the achievement of results is not important. Indeed, we do not consider a poor person who does not choose to achieve certain functioning as he can reach them if he wants. Thus, an individual is considered poor if it lacks the capacity (human and physical capital) necessary to achieve a certain subset of functioning considered reasonably minimal. Based on this approach, strategies for poverty reduction seek to promote capacities of the poor 1.2.2. Basic needs approach This approach of Sen (1985) focuses on the need to satisfy certain basic needs which are necessary to achieve a certain quality of life. The principal basic needs taken into account are: education, health, hygiene, cleansing, drinking water, housing, access to basic infrastructure, etc.... The measurement of welfare, in this case consists of an aggregation of basic multidimensional functioning. The determination of basic needs, essential for a decent life, depends on the characteristics of individuals and societies in which they live. Thus, satisfaction of basic needs is an important element in judging a person has achieved functioning. According to P.Streeten (1981), the basic requirements are necessary to prevent a badly health, undernourishment and social insecurity. For example, a non-poor person must have access to social services like education and health, but it is not necessary to be healthy and well educated. However, the basic needs approach faces a problem of determining the list of needs. In fact, this list is not limited to a specific number of areas, and there is no consensus on what should be this list. Another problem relates to the minimum level which should be reached, in each area, to not be considered poor. It results from it that policies to reduce poverty which characterize this approach, based on satisfaction of basic needs such as improving housing, education of children, the eradication of diseases, reduction of malnutrition, etc. 3. Approaches to multidimensional poverty measurement Referring to the economic literature which is interested in multidimensional poverty, it is difficult to achieve an appropriate measure of it and to put a specific approach to operate. Generally, the synthesis of non-monetary indicators of well-being bases on two large multidimensional analysis of poverty, namely, the axiomatic a non-axiomatic approach. 3.1. Sen and the axiomatic approach for the aggregation of poverty We can simply represent the extent of poverty by adopting the notation of Asselin (2002). Let a population of size n, each individual i of the population owns a vector xi of J attributes, which is orthan with non-negative in euclidean space.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

858

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

Let X an nJ matrix, where each element xij in the matrix gives the amount of attribute j possessed by individual j j i. And z j Z is the poverty threshold for each attribute j, with Z R positive orthan of Euclidean space R . Let F (.) an additive function, then:

1 P( X , z) = n

(x , z)
i =1 i

And let (.) an indicator functions, that is to say :

( xi , z ) = 1 si x ij z j ( xi , z ) = 0 si non
The properties of F (.) and

j=1, 2, , K

This gives a multidimensional extension of the incidence of poverty. (.) are functions of the axioms which measurement of poverty should not violate. According to this approach, poverty is as a complex concept and whose normative analysis can be facilitated by the adoption of axioms for poverty measurement. In this context, Sen (1976) has established a number of axioms that a good poverty index should satisfy. These axioms reflect the ethical and moral principles of a given society vis--vis the poor. The first two axioms of Sen explaining the situations of increase in poverty: The monotonicity axiom: all other things being equal, a reduction of income of a person who is below the poverty line must increase the poverty measure. This is obvious, as when a person becomes poorer aggregate poverty will be larger. The transfer axiom: All things being equal, a transfer of income between a person who is below the poverty line and a person who is richer (called regressive transfer) must increase the poverty measurement. So, the poorest people have a more important contribution to the aggregation of poverty. In fact, the transfer of income between the poor does not involve change in values of H (defined as the ratio between the number of poor and total population) and I (intensity of poverty), while aggregate poverty has been increased by the transfer, since the poor become poorer (Sen, 1992). The critics of Sen led economists to propose new measures, such as the FGT indices (Foster, Greer, Thorbecke (1984)) that take into account the three dimensions at once, such as measures of the incidence, intensity of poverty and inequality. 3.2. Non axiomatic approaches The study of multidimensional poverty is based on the use of multiple dimensions of well-being. Thus, the measurement of poverty consists in using several aggregate indices of well-being at the same time. 3.2.1. Fuzzy set approach In 1990, Cerioli and Ziani developed a first method for multidimensional measure of poverty based on the theory of fuzzy sets. This method identifies the dominant dimensions of poverty and provides the necessary elements for the development of socio-economic policies aimed at reducing the state. Dagum and Costa (2004) then introduced the one-dimensional indexes to measure the state of deprivation of each attribute for the entire population, to measure the contribution of each dimension to global poverty. Thus, the poverty of a person is identified by its degree of belonging to fuzzy sets and this respectively to each of the attributes of poverty (Costa, 2002). The degree is determined by the degree of ownership of the attribute (Ayadi, and Ben Hassine Chitoui (2005)). Indeed, the membership function in fuzzy set B of the ith household (i = 1, , N) relative to the jth attribute (j = 1, ...., m) is defined as follows:

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

859

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


x ij = B ( X j ( a i ) )
;

JANUARY 2012
VOL 3, NO 9

0 x ij 1

With xij = 1if the ith household did not have the jth attribute; xij = 0 if the ith household has the jth attribute; The well-being of an individual increases as the index approaches to 1. The membership function of the ith household in a fuzzy subset B, measures the ratio of multidimensional poverty household where wj is the weight attached to ith attribute. In fact, this function can be defined as the average weight of xij.

(ai) =

j=1 m

x ij w w
j

With

attributes.

0 B ( ai ) 1 and 0 B ( ai ) 1 if ai has m attributes ; B ( a i ) = 1 if ai deprived of m attributes; 0 B ( ai ) 1 if ai is partially or totally deprived of some attributes but not completely stripped of all

j=1

With the weight wj represent the intensity of deprivation linked to the attribute Xj and wj written as follows:
n f ( a i ) Where fi is the frequency of null values for each variable xij. w j = l o g mi = 1 Thus, the fuzzy index of poverty of the population A can be defined as a weighted average of x ij f ( a i ) is written as follows: j=1

B (ai )

and

i=1

B n

(ai) f (ai) f (ai)

i=1

3.2.2. Entropy approach Entropy approach is a parametric approach, derived from the mechanical dynamics. It is much used in the statistical theory of information. Its adaptation to poverty measures has made by Maasoum.E (1986) who left this theory based on a divergence measure between two distributions to provide a composite optimal (C) that minimize a weighted sum of divergence in pairs. It looks like a distance interdistributional basis for deriving the composite indicator of a C optimization criterion, since this approach relies on the distance or divergence between distributions Ik (Between-distance distribution). A functional form of the optimal transformation function is written as (Asselin (2002)):
K

IC Pi ( I ik ) =

k =1

I ik

0, 1

Ik : quantitatif indicators. ik : value of indicator

I k for individual i of the population.

sont des poids dtermins de manire non arbitraire grce la mthode ACP (Analyse en Composantes avec

Principales)

= 1.

We can simply expose the basics principles of this approach (Asselin, 2002).

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

860

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Let Pi = prob (x = xi), i=1,, n. the probability that an experiences result be xi with Thus, the anticiper information of an experience is define as follow:

JANUARY 2012
VOL 3, NO 9

Pi [ 0,1]

Let g (Pi) a probability function. We suppose that g(.) a function that g(1) = 0 et g(0) =.

H ( P ) = Pi g ( Pi ) 0
i =1

P = ( P1 ,....., Pn )
associated with a random variable.

We define the entropy as a measure of uncertainty If g (Pi) =- log P, H (P) is Shannon entropy. In this case we have: For an event, for example P = (0, 1, 0, ... ..., 0), H (P) = 0; For equiprobable events, H (P) = log n.

0 H(P) log n
Based on the Shannon entropy, three major classes of measures (, , ) are defined in terms of certain axioms or desirable properties. These three classes of indices are: -divergence, the -gain information and generalized entropy. The basic principles of this approach can be adapted to multidimensional poverty measurement (Maweki Batana, 2006) as follows: a population of n individuals indexed by i = 1....., n and m attributes of well-being indexed by j = 1...., m and Xij is the value of an attribute j for individual i. In a first step, is to aggregate the attributive vector of the ith individual in a single value xit. The basic principles of this approach are to find a vector that reflects the highest possible welfare provided to each individual by the set of attributes. This is to minimize the generalized entropy function as follows:

G E ( xT , X , ) =

1 ( + 1)
1

j x iT (
j =1

i =1

x iT ) 1 x ij

0, 1

In general, the solution of the minimization is:

xiT = j ( xij ) j =1
m

j =
with

j
m j

j =1

3.2.2.3. The criterion of inertia The approach of the inertia comes from the static mechanical and relies on techniques of data analysis, the main methods are: Principal Component Analysis (PCA), Correspondence Analysis (CCA) Generalized Canonical Analysis (CGA) and the Multiple Correspondence Analysis (MCA) Multivariate analysis techniques, applied to this matrix of N observations and K variables can be projected in a sub-optimal space of dimension lower (p <m) while minimizing the loss of inertia of the cloud source and keeping only the essential information. This leads to finding eigenvectors associated with the first eigenvalue which measure the inertia of the cloud of points projected. In the new optimal space of dimension p, each unit Ui is represented using coordinates or scores. The score in the space or dimension j ( j p ) is a linear combination of K variables Ik. The interpretation of multidimensional data becomes easier by adopting the Principal Components Analysis (PCA). It uses the Euclidean metric with quantitative variables; these variables are normalized so they have different units of measurement. As for the AFC, it aims to analyze the association between the terms of two qualitative variables. So instead of focusing on individuals, we are interested in characteristics that affect them. In the case of more than two variables, we used the MCFA (Multiple Correspondence Factorial Analysis) which is a generalization of the CAF. This technique is used to represent the same diagram on how more than two variables. It is very useful for the analysis of questionnaires in which the questions are multiple responses. For the technique of GCA considers two sets of variables and analyzes the correlations between variables in both groups. It uses the Mahalanobis metric and is looking for a subset of variables to maximum correlation.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

861

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

The interpretation of multidimensional data becomes easier by adopting the Multiple Correspondence Analysis (MCA). This statistical technique used to summarize the set of variables into a small number of variables called "factors". We have codified the variables to give them an ordinal character and singularly have the property FAOC (First Axis Ordering Consistency) bring a level of welfare increasing or decreasing. A low code corresponds to a low level of well-being while a higher code corresponds to a high level of well-being. The MCP method is to determine the weight ( wmk ), nonlinear way and without assuming a functional form in the initial transformation function that generates the Composite Poverty Indicator. For each household i in the sample, the composite indicator value is determined by ACM and given by the equation:
k

ICPi (.) =
Where K is the number of categorical indicators. Mk: number of categories (terms) of the indicator K.

W
k =1 mk =1

Mk

K mk

K Jm k

k wm The weight (standard score on the first axis) of Class mk. k K J mk is a binary variable that takes the value 1 if household i has the modality mk.

This method has an advantage over the CPA, which requires the variables to be quantitative. In ACM, we can include both qualitative variables and as quantitative variables that can be coded in binary form. 4. Sample and instruments 4.1. Sample and source of data The data come from the budget survey, consumption and living standards of households in 2005, conducted by the National Institute of Statistics. This investigation focused on an initial of 13392 households selected by random sampling stratified two-degree in each governorate. Our case study will be interested in the region of East-Central Tunisia, which has four main governorates: Sousse, Monastir, Mahdia and Sfax. The table below shows the number of primary units and the number of households in our sample. The size of the sample drawn in the first degree of the draw is 216 primary units corresponding to 2592 sample households in the second degree. However, it should be noted that the 2592 household sample drawn from a number of 2312 households was actually investigated, let 89.19% of the initial sample of the draw. This difference is due to the refusal by certain categories of households responding to the survey or, in exceptional circumstances in which some households were during the survey period. For the study of multidimensional poverty, we will use variables for the characteristics of household members, such as education, health status and employment. Similarly, we will use the information on housing characteristics such as distance between the various social services (drinking water, school, health service) and possession of households comfort goods. The variables and the different methods are shown in the following table: 4.2. Construction method of the Composite Poverty Indicator As part of this work we have a set of variables describing the procedures that can take the primary indicators of household living conditions, such as the type of cooking, lighting mode, possession of an estate, etc... The problem we want to solve is how to aggregate these qualitative indicators into one single composite indicator which has the property of being a good summary of the information provided by the initial indicators? The most appropriate method is the MCA. This technique is to maximize the inertia of the cloud of points projected on L, which led to the search U eigenvectors (direction vectors of optimal subspace) eigenvalues associated with matrix initial data. The first eigenvector associated with the first eigenvalue (the highest value) is also called the first factorial axis.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

862

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

This theme has a special meaning. Is the axis towards which the spreading of the cloud of points is maximal? The first axis has a coordinate each variable factor, also called score that reflects the importance of the variable on the first factorial axis. The functional form of the indicator is defined as follows:

Ai =


q =1

W
q

q j

j J

q j

In this case, Jq is the number of terms of the indicator q, weights.

W jq

is the weight given to the modality j or the

For each household, the value of CPI is defined in equation calculated from the scores on the first factorial axis. The extent of discrimination of a variable is the variance of factor scores obtained by the procedures associated with this variable. A high value corresponds to a high discrimination power. The rank is used to display indicators that contribute most to the inertia on the axis and hence the Composite Poverty Indicator (CPI). 5. Application of the MCA and interpretation of results The variables most discredited assuming a discrimination rate greater than 0.5 are associated with the particular mode of occupation and possession of such energy that the heating mode and type of bathroom, then, involved the following procedures: education level of households head, the nature of housing, access to secondary school and health, unemployment, possession of air and finally, the type of toilet and kitchen. The variables that have a low rate of discrimination are associated with the possession of some durable goods (vehicles, radio, TV), access to primary school and nature of residence. We note as follows: kitchen without basin, another source of drinking water, another lighting mode, tap water and public toilet outside house are very distant from the origin. These terms are not well represented in the sample which shows that households do not have a very low standard of living. In fact, it is very rare to find households using a traditional method of lighting and a public tap. Thus, poverty does not show a multidimensional situation for poor households because of the availability of a good infrastructure. For this first MCA, total inertia of the cloud of points is equal to 37.1% (0.238+ 0.133). Following a number of criteria, we reduced the number of variables from the first CMA. The main criterion is the ordinal consistency on the first factorial axis (FAOC). This property is for a partial indicator to see its structure ordinal well-being respected by the ordinal structure coordinates of its terms on the first factorial axis. Indeed, the explanatory power of the first factor is increased from 23.8% to 37.1% and that of the second, of 13.3% to 19.7%. Thus, the percentage of inertia extracted by the first two factors we interpret is 56.8%. The plan made up of two axes shows that indicators of well-being are spread out on a form of parabola which corresponds to a unidimensional structure. Thus, we talked about a "Guttman effect" which allows also interpreting the second factor. Indeed, the second factor is a quadratic function of the first factor. The table above shows the discrimination measures and the ranks of basic indicators on the first factorial axis. The extent of discrimination of a variable is the variance of factor scores obtained by the procedures associated with this variable. A high value corresponds to a high discrimination power. The rank is used to display indicators that contribute most to the inertia of axis 1 and hence the composite indicator of poverty (CPI). On the first axis, indicators that provide the greatest contribution to inertia are those related to housing type and mode of heating such as the type of bathroom and heating mode. Then there are the indicators of human capital

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

863

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


room, the nature of housing, unemployment, access to health service, type of toilet and kitchen.

JANUARY 2012
VOL 3, NO 9

as the level of education of household head, access to secondary school, possession of air conditioner and dining Finally, there are indicators relating to the possession of goods such as iron, computer and internet. Following the parable in Figure1, from the negative end to the other, there are rich households in the negative side and poor households in the positive side. This is an axis that can be called an axis "wealth-Poverty." Thus, variables that has property COPA obey the rule that the property be improved from a situation of poverty to wealth status throughout the first axis. This property simply means that the mode describing a situation of poverty is on the side of the poor (positive side) on the first axis and describes a situation that is the wealth of the rich side (negative side). Terms that are positively correlated with the first axis relate to the residential user, the SPC and services more accessible. This is the home user with a free lighting, a traditional mode of heating (using coal or wood), using a public tap, a kitchen without tank and toilet without flushing water, a bathroom equipped and not owning a free cooking oven. These terms correspond to households who are illiterate, unemployed, artisans and workers in agriculture and have access to social services over 30 minutes and an unemployment rate of over 75%. In contrast, terms that are negatively correlated with the first axis are those relating to housing type and ownership of durable goods. This is a villa style accommodation with possession of air conditioner, computer, internet and dining room. These terms correspond to households that include executives or employees who have a higher educational level. On the second axis introduces a differentiation within each class. At upper class, he distinguishes households rich ways. Also at the poor class, he distinguishes poor households overall average household. The scatter plot shows three groups of households: the rich, average and poor. Quadrant (-, -) is the area of the rich. These are households that have a higher educational level and who occupy important positions namely senior managers and employees with access to modern means of heating, electricity and drinking water. These households own villa well equipped with bathroom with hot water, toilet with flushing water and kitchen with basin. Quadrant (+, +) is the area of poor. This poverty is described mainly by habitat characteristics and problems of employment and access to social sectors. These artisans are poor, without a bathroom, a toilet without flushing water and use carbon or wood as heating mode. Over 75% of the assets of these households are unemployed. As for the social sectors, they have access to education and health service more than 30 minutes. Quadrant (+, -), characterizing a situation of average living standards. These households have an average level of education, namely, college or high school, engaged in agriculture and in industry as middle managers. Concerning housing, housing type is Arab house or studio, with shower with hot water, a method of heating with gas bottle. Referring to the diagram which groups the households into three classes and to confirm the results of this classification, we will appeal, in the remainder, a classification of households by the indicator of multidimensional poverty. 5.1. Classification of households by CPI We conducted an automatic classification of households by the CPI. It was made according to the subscripted hierarchical ascending method (CAHI). It is used to gather individuals into homogeneous classes according to their CPI. The results of this classification of households into five groups are given in the following table: As shown in Table 4, the incidence of poverty with a hierarchical classification of households is the weight of the poorer classes is 3.02%. Tables 5, 6, 7, 8 and 9 respectively represent the poor, quasi-poor, average, quasi-rich and rich class. It should be noted that a modality is characteristic in a class when it appears in the class with a percentage, significantly, higher than the average percentage in the population. The modalities characterizing are arranged according to their percentage in the class. Table 5 characterizing the poor, indicates that terms associated with the non possession of comfort goods, such as camera, air-conditioner, dining room, internet and hair dryers characterize well this class. 100% of poor households do not have durable goods, such as computer, air-conditioner and microwave.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

864

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

As shown in table 5, more than 89% of the poor households have access to social services of more than 30 minutes. 100% of this category does not have access to communication (such as Internet) and 80% among them do not have housing of good quality (toilet without water hunting, kitchen without basin and bathroom not equipped). In this class, 47.4% of the poor are illiterate. The category of the quasi-poor gathers households who do not have comfort goods and communication. They use carbon or woods as means of heating and suffer from problem of access to social services (at secondary school and a health service) which are more than 30 minutes. Modalities characterizing the middle class are related to the drinking water resource (89.6% of these households are connected by the water of SONEDE and 96.5% use electricity with invoice). These households do not have comfort goods; they have an acceptable kind of housing, such as kitchen with basin, toilet with water flushing and showers with water-heater. They use the bottle of gas as means of heating but, they do not have an airconditioner. The quasi-reach households are characterized by acceptable conditions of habitat, such as, kitchen with basin and toilet with water flushing. However, more than 87% of these households do not have durable goods, such as air-conditioner, computer and microwave. Concerning the rich class, that are modalities associated with the type of housing that characterize well this class. In this class, 100% of households have a toilet with flushing water. 99.6% have a kitchen with basin and 98.6% have a bathroom with water heating. The rich class gathers the households which have access to the basic services of less than 30 minutes. More than 80% of the rich households have durable and comfort goods, such as, dining room, iron, dry hair and camera, as they characterized by an unemployment rate less than 25%. The results of classification are in conformity with those of the AMC. In fact, these poor do not have comfort goods and good conditions of housing. It should be noted that among the poor households according to the multidimensional approach, 20% are also touched by monetary poverty and 47.37% are touched by non monetary poverty. To specify the percentage of poor population by the multidimensional approach, it is useful to construct a multidimensional poverty line and estimate the poverty profiles. 5.2. Determination of the poverty line and indices FGT Once the CIP is calculate, it is appropriate to determine a poverty line which will make it possible to build a profile of multidimensional poverty. This poverty line is equal to 50% of the median of indices CPI ordered in ascending order, is equal to 0.085. The poverty rate is the percentage of the population whose CPI is below the multidimensional poverty line. It is equal to q/n, q: the number of poor people and n: the size of the population. For our sample, the incidence of poverty is equal to 0.4844. Thus, 48.44% of the population is poor according to the non monetary approach. The poverty gap is defined as the average distance separating the population from the poverty line when a zero distance allocated to non-poor. It calculates the average percentage of lack of income of the poor from the poverty line. Analytically, the poverty gap is written as follows:

I=

1 q g i 1 q z yi = z n i =1 z n i =1

y q = n 1 z

The value of the poverty gap is equal to I = 0.6825, indicating that the lack of average is 68.25% of income for the poor reach the threshold 0.085. The squared poverty gap represents the severity of poverty and is written using the following equation:

p2 =

yi 1 q 1 n i =1 z

The severity of poverty is equal to 1.1451.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

865

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

According to the multidimensional approach, the poverty rate is 48.44%. This raises the question of which are the most contributive factors with the average poverty of the sample. Hence, we will decompose the indices FGT of poverty according to sub-group which share the same sociodemographic characteristics. The software used for this analysis is DAD version 4.3. 5.3. Decomposition of indices FGT of non monetary poverty The FGT index of poverty of a population composed of K sub-groups can be written as the following form:
K

P ( z; ) =

(k ) P (k ; z; )
k =1

With is the FGT poverty index in subgroup k and the proportion of the population in subgroup k. Thus, the contribution of subgroup k to the poverty index of the total population is equal to ( k ) P ( k ; z; ) .

From Table 10, which can provide the profile of poverty on the entire sample by selected characteristics, drew several comments. The incidence of poverty is not different in a very important way according to the sex of households head. It is equal to 51.58% for households headed by a man and 26.87% for households headed by a woman. In the same way, for the others FGT index of poverty. Concerning the contribution of the sub-group of households to poverty, we note a higher contribution for the male household than for the female households. The contribution of male households to the various incidences of poverty is with the order of 92%, against 7% for female households. Then, it follows from the analysis that non-monetary poverty touches more the oldest households (whose chief has between 45 and 54). The latter have an incidence of poverty equal to 54.93% against a rate of 50.86% for households whose chief has between 16 and 34 years. The results indicate that households, whose chief aged between 35 and 44 years, contribute more to average poverty of the sample. This contribution is most important for three FGT indices of poverty. The contribution to povertys incidence of households, whose chief has a primary education level, is most important, with a larger part of 36%. In the same way for the other sub-groups which have a higher level of instruction. It should be noted that non monetary poverty affects more the graduates who have a higher educational level. The study of the professional category is particularly interesting. We see very high values of poverty indices for different subgroups. This shows that non-monetary poverty is not dependent on the professional category of the household head. However, the contribution of households to the average poverty, whose chief is a worker, is most important. These values are equal, respectively, 55.57% and 13.43% for workers in the industrial sector is most important; it is equal to 62.32%. Moreover, non-monetary poverty touches more the urban households, with a rate equal to 73.85% compared with 59.10% for rural environment. Thus, non-monetary poverty is not a rural phenomenon. The decomposition of the FGT indices of poverty according to the geographical location (Table 9) shows several reflections. Initially, we note a strong homogeneity of poverty between the layers. The incidence of poverty, for a line of poverty equalizes to 0.085, is equal to 56% for Sousse and 51.05% for Sfax, with an average rate on the whole of the zone of 48.44%. We also note that Mahdia and Monastir have less relatively low rates, respectively of 34/36% and 47.95%. It is possible to conclude that these zones which are close geographically tend towards a similar poverty incidence. The data in Table 11 shows that the contribution of Sfax to the average poverty of the sample is most important, for the three measurements of FGT. Instead, Mahdia and Monastir always admit a lower contribution to the poverty of the population. To conclude on the most contributive factors with multidimensional poverty, it is useful to confirm the results got by classification and that of the decomposition of indices FGT. Thus, we will use to estimate a binary logit model, gathering the different contributive factors to poverty and to measure the effect of each factor to the variation from poverty.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

866

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


6. Analysis of the determinants of non-monetary poverty

JANUARY 2012
VOL 3, NO 9

It is a question of modeling occurred and not occurred of an event. Here we will model the probability of being poor and dependent variable (poverty) will take the value 1 if the household is poor and 0 otherwise. The grouping of households in poor and rich is through hierarchical classification. Thus, the objective is to model occurred of poverty according to certain socio-economic characteristics like the sex of the households head, age, socioprofessional category, type of housing, access to public services,.... etc. The instruments used for this analysis are the marginal effects and the signs of coefficients. The coefficients as presented below are not marginal impacts, only their signs are interpreted. Table in appendix 1 of the estimated logit model shows the significance (at a threshold of 1%) of several variables of interest. The age of the households head effect is positive. In fact, more a head of household is less old, more the probability of being poor increases, considering the least old do not have yet a post at the labor market. Concerning the educational level, more the educations level reduces compared to the higher level, more the probability of being poor increases. In the same way, the occupational category of households head has a positive effect on the probability of being poor. More the households head has a less important station, compared to the senior officer, more the probability that this household belonging to the poor increases. In addition, the effect of unemployment rate on the probability of being poor is positive. As for the access to the social sectors of more than 30 minutes, they admit a positive effect on the probability of being poor. We find that more the access to the secondary school and a health service is more being poor increases. This confirms the result got by AMC and according to which the poor have a problem of access to the social services. The calculation of marginal effects will allow us to appreciate the sensitivity of the probability of being poor compared to the methods of the explanatory variables. They will be obtained by the derivative of the estimated probability compared to the explanatory variables. The last column of the table above shows the average values of explanatory variables, which are values calculated marginal impacts. The results show that the increase in the age of 1% for the category between 35 and 44 years makes increases the probability of being poor by 16.58%. We note that more the households head is illiterate more the probability of being poor increases by 48.67%. The probability of being poor is weakening as the educational level increases. It goes from 48.67% for an individual illiterate to 15.48% for an individual with a higher level of education. More the unemployment rate increases by 1%, more the probability of being poor increases by 47.81%. Thus, more the members of a household have a job, more the probability of being poor increases. For access to social services, more the access for less than 30 minutes increased by 1%, more the probability of being poor decreases by 20.30% and 20.37%, respectively, for the school and the health service. Finally, we note that a 1% increase in household size over 5 members increases the probability of being poor by 5.81%. It should be noted that the effect on the probability of being poor is decreasing as the size decreases. 7. Conclusion Applied to the case of Tunisia, this multidimensional approach of the poverty based on the calculation of a composite indicator, resulting from the Multiple Correspondence Analysis and taking into account the educational, health, drinking water, nutrition, housing, energy, communication, capital equipments of comfort and other assets, allowed to draw important conclusions on poverty. The research conducted allowed to know the profile of poverty in east-central Tunisia. Non monetary poverty affects 3.02% of the population, by adopting the hierarchical classification analysis by CPI. Whereas, the estimate of an absolute threshold of non-monetary poverty gives an incidence of poverty equal to 48.44%. The class of poor is described mainly by the non possession of goods of comfort. These poor are, in particular, unemployed or retired and having houses without a bathroom, a toilet without flushing water and use carbon or

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

867

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

wood as heating mode. For the social sectors, they have access to education and health service more than 30 minutes. Referring to the characteristics of the household head, households directed by a man are more affected by poverty than female households. Similarly, non-monetary poverty is more prevalent among households with heads aged between 35 and 44 years and does not reach a higher level of education. Compared to the socioprofessional category, the industrial and agricultural workers contribute more to poverty in the sample. Geographically, the incidence of poverty is relatively higher in Sousse and Sfax compared to Mahdia and Monastir. The contribution of Sfax to the incidence of average poverty is highest for three indices FGT of poverty. The estimated binary logit model confirms the importance of access to social services in determining the level of welfare than monetary, especially indicators of access to secondary school and access to health services. It also shows the importance of demographic factors on poverty. Although poverty in infrastructure point of view does not exist, our study shows that households in East-Central Tunisia are affected by other types of poverty. It is found that there is a poverty point of view of comfort household poverty resulting in poor quality housing and a poverty point of view of access to public services. Given the complexity of the phenomenon of poverty, effective policies to fight against poverty must be based on multiple and varied instruments. Thus, facilitation of access to physical and human capital, raising living standards of the poor class deserve the attention of authorities. Given the importance of non-monetary indicators to determine the multidimensional poverty, it is interesting to consider other attributes in the construction of monetary indicator of well-being such as food consumption, total... etc. These indicators can provide more information on the needs of poor people and their lifestyles and also to make a comparison between the patterns of income poverty and non-monetary.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

868

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


References

JANUARY 2012
VOL 3, NO 9

Ambapour, S., (2006), Pauvret multidimensionnelle au Congo : une approche non montaire. Bureau dApplication des Mthodes Statistiques et Informatiques. Asselin, L-M., (2002), Multidimensional poverty: Composite indicator of multidimensional poverty Institut de Mathmatique Gauss: Lvis, Qubec. Asselin, L-M., Dauphin, A., (2000), Mesure de la pauvret : un cadre conceptuel . Centre Canadien dEtude et de Coopration international. Ayadi, M., Naouel C., EL Lahga, A., (2005), Analyse multidimensionnelle de la pauvret en Tunisie entre 1988 et 2001 par une approche non montaire, PEP, a paper presented during the 4th PEP Research Network General Meeting, june 13-17, 2005, Colombo, Sri Lanka. Belhadj, B., (2005), Pauvrets persistante, Chronique et transitoire : construction des indices flous, 3rd International Conference : Sciences of Electronic, Technologies of Information and Telecommunications, 2005, Tunisia. Belhadj, B., Matoussi, M.S (2006), Mesure floue de la pauvret, Institut Suprieur de Gestion de Tunis. Bibi, S. (2003), Comparing Multidimensional Poverty between Egypt and Tunisia, 10th ERFs Annual Conference Marekch Maroc. Bourguignon, F et Chakravarty, S., (2003), The measurement of multidimensional poverty, Journal of Economic Inequality. Cerioli A., Zani S., (1990), A Fuzzy Approach to the Measurement of poverty. Costa, M (2002), A Multidimensional Approach to The Measurement of Poverty, IRISS WP 2002-2005. Dagum, C., Costa, M., (2004), Analysis and Measurement of Poverty, Univariate and Multivariate Approaches and their Policy Implications. A case of Study: Italy, In Dagum C. and Ferrari G. (eds.); Household Behaviour, Equivalence Scales, Welfare and Poverty, Springer Verlag, Germany. Foster, J.E., J. Greer et E. Thorbecke (1984), A class of Decomposable Poverty Measures, Econometrica, pp. 761-776. Institut National des Statistiques Enqute sur la consommation et le budget des mnages plusieurs volumes 1990, 1995, 2000 et 2005. Lebart, L., Morineau, A., Piron, M., (1995), Statistique exploratoire multidimensionnelle, Dunod, Paris. Maasoumi, E., (1986), The measurement and decomposition of multidimensional inequality, Econometrica, pp. 771-779. Maweki Batana, Y., (2006), Tests de comparaisons de la pauvret multidimensionnelle bass sur le ratio de vraisemblance, CIRPEE, Universit de Laval. Sahn, D.E. and D.C. Stifel (2000), Poverty Comparisons Over Time and Across Countries in Africa, World Development. Sen, A.K., (1976), Poverty: an ordinal approach to measurement, Econometrica, pp. 219-231. Sen, A.K., (1985), commodities and capabilities, North-Holland, Amsterdam. Sen, A.K., (1992), Inequality Reexamined, Harvard University Press, Cambridge, MA. Streeten, P., (1981), First Things First: Meeting Basic Human Needs in Developing Countries, Oxford University Press, New York.

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

869

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Figures: Second factorial plan
TY P TOIL

JANUARY 2012
VOL 3, NO 9

Quantifications
2 ouche avec c collge 1 studio entre 25 et primaire retrait ouvindustr moins de 30 carbon ou bo moins de 30 m oins de 30 chambre toil avec ch cuisine avec moins de 25% 2,00 bouteille de secondaire analphabte 1,00 maison arabe salbain avec employ salle 1,00 mang plus de 30 m ordinateur ouvagricol 1,00 internet climatiseur villa cadre moyen chomeur pas de cuisi plus de 75% suprieur 2,00 plus de 30 m artisan gaz de ville toil ss chas lectricit toilette hor cadre supri salbain non plus de 30 m pas de salle

SA LLBA IN M OY CHFFG INTE RNT ORDINA TR TXCHOMG A CCE SA NT A CCE SCD

-1

D i m e n s i o n2

-2

A CCE P OT CSP

-3

cuisine ss b E DUCA TIO

-4 -2 -1 0 1 2 3

LOGE M E NT

Dimension 1

Tables Table 1: Size of sample Nombre dU.P chantillon Nombre de mnages Sousse 60 720 Monastir 48 576 Mahdia 36 432 Sfax 72 864 East-Central 216 2592 Table 2: Variables and modalities Les indicateurs/ modalities Housing Type of housing 1 : bedroom 2 : studio 3 : arabic house 4 : villa Nature of residence 1 : owner 2 : loyer 3 : free Type of kitchen 1 : no kitchen 2 : kitchen without basin 3 : kitchen with basin Type of toilet 1 : no toilet 2 : toilet without flushing water 3 : toilet with flushing water Type of bath room 1 : without bath room 2 : bathroom not equipped 3 : shower with hot water 4 : bathroom with hot water Mode of lighting 1 : other source 2 : free 3 : electricity

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

870

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Mode of heating 1 : coal or wood 2 : gas bottle 3 : gaz de ville 4 : electricity 1 : yes 0 : no 1 : others 2 : eau de robinet gratuit 3 : eau de robinet avec facture 1 : less than 30 minutes 0 : more than 30 minutes 1 : illiterate 2 : primary 3 : college 4 : secondary 5 : high 1 : less than 30 mn 0 : more than 30 mn 1 : less than 30 mn 0 : more than 30 mn 1 : less than 30 mn 0 : more than 30 mn

JANUARY 2012
VOL 3, NO 9

Possession of vehicle Drinking water Source of drinking water Access to water

Education Level of education of househol head

Access to primary school Access to secondary school Health Access to health service

Employment Professional category of household head 1 : Unemployment 2 : retired 3 : agricol worker 4 : industriel worker 5 : employee 6 : artisan 7 : middle management 8 : senior Unemployment rate 1 : > = 75% 2 : = 25-75% 3 : < = 25% Communication TV 1 : yes 0 : no Radio 1 : yes 0 : no Internet 1 : yes 0 : no Possession of durable goods Air conditioner 1 : yes 0 : no Dining room 1 : yes 0 : no Microwave 1 : yes 0 : no Cooking with oven 1 : yes 0 : no

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

871

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Washing machine 1 : yes 0 : no Computer 1 : yes 0 : no Hair dryer 1 : yes 0 : no Camera 1 : yes 0 : no Iron 1 : yes 0 : no Table 3: Measure of discrimination and range Variables Rang Discrimination 0.851 1 Type of bathroom 0.647 2 Mode of heating 0.399 3 Education 0.375 4 Access to secondary school 0.368 5 Air conditioner 0.350 6 Dining room 0.339 7 Type of housing 0.335 8 Rate of unemployment 0.297 9 Access to health service 0.289 10 Type of toilet 0.289 11 Sche cheveu 0.267 12 Type of kitchen 0.247 13 CSP 0.194 14 Access to drinking water 0.182 15 Resource of drinking water 0.143 16 Access to primary school 0.135 17 Internet 0.129 18 Irone 0.093 19 Computer Table 4: Proportion of classes Minimum Maximum 4.94 8.65 2.1 4.75 0.27 2.08 -2.48 -0.82 -0.79 0.25 -0.79 8.65

JANUARY 2012
VOL 3, NO 9

Poor Quasi-poor Average Quasi-rich Rich Total

% 3.02% 15.05% 30.88% 39.01% 12.07% 100%

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

872

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Tableau 5: Characteristics of the poor class Variables Modalits % de la modalit au sein de la classe 100 100 100 100 100 100 100 96.2 96.2 89.7 87.2 83.3 Frquence globale de la modalit 25.7 20.8 41.1 17.3 43.8 22.3 31.1 79.1 40.3 26.8 3.6 16.2

JANUARY 2012
VOL 3, NO 9

Air-conditioner Computer Dining room Internet Dry hair Microwave Means of heating Type of toilet Access to the secondary school Access to the health service Type of kitchen Type of bathroom

No No No No No No Carbon or wood Without water huntig More than 30 minutes More than 30 minutes Cook without basin Bathroom not equiped

Tableau 6: Characteristics of the quasi-poor class Variables Air-conditioner Camera Microwave Dining room Dry hair Internet Means of heating Access to the secondary school Access to the health service Modalits No No No No No No Carbon or wood More than 30 minutes More than 30 minutes % de la modalit au sein de la classe 100 100 99.2 98 97.5 97 94.7 80.6 65.2 Frquence globale de la modalit 95.3 96.5 92.9 41.1 82.5 82.5 51.6 40.3 26.8

Variables

Table7: Characteristics of average class Modalits % de la modalit au sein de la classe Tap with invoice With basin No No No No No Bottle of gas Less than 30 minutes Shower with water heating 98.7 99.6 99.3 98.2 95.8 92.2 86.9 86.9 76.2 67.1

Resource of drinking water Type of kitchen Air-conditioner Camra Dining room Computer Microwave Means of heating Access to the health service Type of bathroom

Frquence globale de la modalit 96.5 95.3 74.3 61.6 58.9 79.1 77.7 31.1 73.2 33.8

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

873

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS


Table 8: Characteristics of quasi-rich class Variables Type of kitchen Resource of drinking water Type of toilet Unemployment rate Access to health service Access to primary school Internet Air-conditioner Computer Microwave Modalits With basin Tap with invoice With water heating Less than 25% Less than 30 minutes Less than 30 minutes No No No No % de la modalit au sein de la classe 99.6 98.3 95.9 92.6 79.5 65.5 90.9 89.7 84.9 78.3

JANUARY 2012
VOL 3, NO 9

Frquence globale de la modalit 95.3 96.5 82.6 82.3 73.2 59 17.3 74.3 79.1 22.3

Table 9: Characteristics of rich class Variables Type of toilet Type of kitchen Source of drinking water Type of bath-room Unemployment rate Access to drinking water Hair dryer Dining room Access to health service Iron Modalites Toilette avec chasse deau Cuisine avec bassin Eau de robinet avec facture Bath-room with hot water Less than 25% Less than 25% Yes Yes Less than 30 minutes Yes % of modality in the class 100 99.6 98.9 98.6 97.3 95.9 93.86 93.5 91.5 89.8 Global frequency of the modality 82.5 95.3 96.5 48.8 81.9 84.8 40.96 58.8 73.2 50.7

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

874

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

Table 10: Decomposition of FGTs non monetary poverty of households sociodemographic characteristics Caractristiques du mnage Sex of household head Men Women Age of household head between 16 et 34 years between 35 et 44 years between 45 et 54 years more than 55 years level of education of household head Illiterate Primary College Secondary High SPC household head Unemployment Retrait Agricol worker Industriel worker Employee Artisan Average manager High manager Area of residence Urban Rural =0 0.5158 (0.0111) 0.2687 (0.0258) 0.5086 (0.0328) 0.4921 (0.0181) 0.5493 (0.0215) 0.4253 (0.0176) 0.1738 (0.0163) 0.4370 (0.0162) 0.2500 (0.1083) 0.6701 (0.0214) 0.8399 (0.0196) 0.1659 (0.0236) 0.2024 (0.0314) 0.2931 (0.0422) 0.5431 (0.0138) 0.6551 (0.0294) 0.4181 (0.0665) 0.5486 (0.0414) 1.000 (0.000) 0.7385 (0.0221) 0.5910 (0.0267) contribution 0.9295 0.0705 0.1053 0.3339 0.2634 0.2973 =1 0.7263 (0.0185) 0.3824 (0.0412) 0.7149 (0.0533) 0.6883 (0.0298) 0.7193 (0.0336) 0.6422 (0.0306) 0.1502 (0.0163) 0.5156 (0.0219) 0.0212 (0.0092) 1.0566 (0.0391) 1.4544 (0.0464) 0.1857 (0.0294) 0.2211 (0.0402) 0.2759 (0.0453) 0.7453 (0.0223) 1.0157 (0.0576) 0.7011 (0.1231) 0.8797 (0.0747) 1.8555 (0.1253) 0.7385 (0.0221) 0.5910 (0.0268) contribution 0.9297 0.0712 0.1051 0.3315 0.2448 0.3186 =2 contribution 0.9282 0.0717 0.1025 0.3301 0.2282 0.3390

1.2178 (0.0369) 0.6462 (0.0807) 1.1696 (0.0992) 1.1499 (0.0582) 1.1254 (0.0645) 1.1465 (0.0640) 0.1644 (0.0206) 0.7117 (0.0355) 0.0018 (0.0007) 1.8545 (0.0807) 2.8695 (0.1100)

0.0830 0.3625 0.0036 0.2884 0.2625 0.0366 0.0294 0.0303 0.6232 0.1526 0.0205 0.0705 0.0360 0.6715 0.3284

0.0509 0.3035 0.0002 0.3227 0.3225 0.0290 0.0228 0.0203 0.6068 0.1679 0.0244 0.0802 0.0482 0.6715 0.3284

0.0332 0.2497 0.0000 0.3376 0.3796

0.2487 0.0232 (0.2487) 0.3117 0.0192 (0.0673) 0.3145 0.01378 (0.0621) 1.1949 0.5799 (0.0422) 1.8983 0.1871 (0.1297) 1.3249 0.0275 (0.2423) 1.5789 0.0858 (0.1548) 4.0862 0.0632 (0.3544) 1.2468 (0.0440) 0.9786 (0.0531) 0.6758 0.3242

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

875

ijcrb.webs.com

INTERDISCIPLINARY JOURNAL OF CONTEMPORARY RESEARCH IN BUSINESS

JANUARY 2012
VOL 3, NO 9

Table 11: Decomposition of FGT poverty indices according to the geographical location Z= 0.5*Me =0 =1 =2 FGT contribution FGT contribution FGT contribution Sousse 0.5600 0.2705 0.8618 0.2954 1.4932 0.3051 Mahdia 0.3436 0.1294 0.4537 0.1213 0.7065 0.1126 Sfax 0.5105 0.3696 0.7081 0.3639 1.2056 0.3693 Monastir 0.4795 0.2303 0.6433 0.2193 1.0479 0.2129 Ensemble 0.4844 0.6825 1.1451 Appendix.1Marginal effects dy/dx P > |Z| -0.1658 -0.1515 0.4867 0.3398 0.1548 0.2956 0.345 0.2477 0.2158 0.0177 0.1623 0.2135 0.4781 0.1672 0.0581 -0.0868 -0.2030 -0.2037 0.000 0.002 0.000 0.000 0.000 0.000 0.000 0.002 0.048 0.885 0.135 0.016 0.000 0.000 0.000 0.000 0.000 0.000

Pauvret Age [35 44] [45 54] Level instruction Illiterate Primaire Secondaire SPC Unemployment Retrait Agricol worker Industriel worker Employ Artisan Average manager Unemployment rate More than 75% Between 25% et 75% Size < 9 members > 9 members Access to secondary school Access to health service

X 0.2335 0.3407 0.2331 0.4043 0.2100 0.1071 0.0710 0.0505 0.5538 0.1137 0.0230 0.0627 0.0840 0.0976 0.4222 0.0296 0.5978 0.7337

COPY RIGHT 2012 Institute of Interdisciplinary Business Research

876

You might also like