Abstract
A monetary approach cannot represent the complex and multidimensional phenomena of poverty, as it only takes economic aspects into account. Many attempts have been made to propose multidimensional approaches to poverty using basic needs or capabilities approaches. For our study we adopted a non-monetary approach, using multivariate correspondence analysis to construct a composite welfare indicator as an aggregation index of the various well-being attributes. In order to measure poverty within a composite indicator distribution we first developed new classes of poverty measures. Indeed, we developed classes of ethical generalized Sen-Shorrocks-Thon (SST) poverty measures. These are a generalization of the Shorrocks (Econometrica 63:1225–1230, 1995) poverty measure which itself is a modified version of the Sen (Econometrica 44:219–231, 1976) poverty measure. We analyzed the non-monetary poverty trend in Tunisia between 1994 and 2006 using the composite welfare indicator and the generalized SST poverty measures. In order to achieve this, we constructed confidence intervals and tested hypotheses based on the bootstrap method. We found that poverty decreased between 1994 and 2006, but that there were inequalities within the poorer population. We also found that poverty was essentially rural poverty and that it was concentrated in the North West, Center West, and South East of Tunisia.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
- 2.
There exist different literature reviews of the capabilities approach, see e.g. Robeyns (2005).
- 3.
In an unpublished master's thesis Chtioui (2004) constructed a composite welfare indicator using factor analysis as developed by Stifel et al. (1999) and analyzed poverty in Tunisia. Bibi (2004) only applied multidimensional poverty measures developed in the literature on Tunisia. In relation to Tunisia both applied bidimensional stochastic dominance as developed by Duclos et al. (2006).
- 4.
A measure that satisfies the scale invariance axiom is called a relative poverty measure. This axiom requires that the value of a poverty measure should remain unchanged after a relative change in both real incomes and the real poverty line. A relative change consists for example of doubling real incomes and the real poverty line or dividing them by the poverty line. A measure that satisfies the translation invariance axiom is called an absolute poverty measure. This axiom stipulates that the value of a poverty measure should remain unchanged after an absolute change of both real incomes and the real poverty line. It requires that adding a real value to all real incomes and the real poverty line should leave the poverty value unchanged.
- 5.
The normative approach was proposed by Dalton (1920) for the measurement of income inequality. He recommended that an inequality index should have a normative foundation in terms of social evaluation and should incorporate society's judgments regarding inequality. Two well-known procedures were developed for the construction of a normative inequality measure. The first procedure, the Atkinson-Kolm-Sen (AKS) approach, was proposed separately by Atkinson (1970) and Kolm (1969) and popularized by Sen (1973) for developing relative indices. The second procedure, the Kolm-Pollak (KP) approach, was proposed separately by Kolm (1969) in income inequality literature and by Pollak (1971) in the consumer theory for developing absolute indices.
- 6.
The three 'I's of Poverty Curve were first introduced by Jenkins and Lambert (1997).
- 7.
The functional form of the members of this class was first proposed by Chakravarty (1983) as an example of an ethical poverty measure.
- 8.
Jenkins and Lambert (1997), who introduced this curve in the measurement of poverty, called it a “TIP” curve (TIP or the three “I”s of poverty: Incidence, Intensity and Inequality among poorer people).
- 9.
If each person receives an equally distributed equivalent income then we obtain a distribution which is socially equivalent to the initial distribution, but which is less unequal.
- 10.
From this point onwards and until the end of the paper, we will use the term “income” to signify a welfare indicator, which is not necessarily money income.
- 11.
The function \( \Upxi_{G}^{n} \left( {\underline{x}^{*} } \right) \) is positively linearly homogeneous if and only if W is homothetic and it is unit-translatable if and only if W is translatable.
- 12.
This inequality is obtained from Eq. (6.12) where \( w_{n + 1} \ge w_{n} . \)
- 13.
This expression was first proposed by Chakravarty (1983) as noted earlier.
- 14.
This inequality is obtained from Eq. (6.24) with \( \pi_{1} \ge \pi_{2}. \)
- 15.
Similar results can easily be derived for absolute single-series SST measures and absolute single-parameter measures.
- 16.
This method is more appropriate when using and retaining only categorical variables for the analysis.
- 17.
Ayadi et al. (2007) conducted MCA using Stata. We opted to use the R language of programing because it gives more appropriate weights for the attributes.
- 18.
- 19.
The Osberg and Xu (1999) approach has been adopted by the World Bank in the Handbook of Poverty Analysis.
- 20.
The Osberg and Xu (2000) approach has been adopted by Luxembourg Income Study.
- 21.
The Hall (1992) method consists of computing the difference between each bootstrap point estimate and the point estimate of the initial data: \( \hat{y}_{b}^{*} = \hat{P}_{b}^{*} - \hat{P} \). Then, it establishes the \( \frac{\alpha }{2} \) and the \( (1 - \frac{\alpha }{2}) \) limits of the new distribution of the \( \hat{y}_{b}^{*} \)s. Finally, the confidence interval limits are obtained as follows
\( \left( {1 - \alpha } \right)100\% CI = \left[ {\hat{P} - \hat{y}_{{\left( {1 - \frac{\alpha }{2}} \right)}}^{*} ,\hat{P} - \hat{y}_{{\left( {\frac{\alpha }{2}} \right)}}^{*} } \right] \); \( \alpha \): significance level.).
- 22.
We developed a program using the R language of programing which contains estimates of different poverty measures, statistics to be computed (mean, standard error), estimates of the different bootstrap confidence intervals and estimates of the different hypothesis tests to be conducted.
- 23.
We conducted a first MCA on the preliminary attributes selected. This technique produced non-intuitive results for the “statute of occupation” variable. They showed a negative weight for its “owner” and “joint owner” modalities and a positive weight for its “Rent the house” modality along with a low contribution to the first factor. It appears evident that these variables should not be included.
- 24.
As noted before, this absolute change has no effect in terms of poverty on the values of absolute poverty measures that we apply in this section.
- 25.
For area comparisons and regional comparisons we chose a poverty line equal to 75 % of the median rather than a poverty line equal to 50 % of the median. This is because for rural areas and some regions, when applying the second threshold we only obtained a small number of poorer people and this prevented us from using the bootstrap technique and computing poverty measures.
- 26.
We limited our regional comparisons. For each of the three cases we chose to test regional disparities between the most deprived regions and the least deprived regions (the Capital Tunis and its surrounding area, as shown by the poverty values in Table C.8).
- 27.
For more details on the economical and social changes in Tunisia since the independence see Ayadi et al. (2007).
References
Asselin, L. M. (2002). Multidimensional poverty: composite indicator of multidimensional poverty. Lévis: Institut de Mathématique Gauss.
Atkinson, A. B. (1970). On the measurement of inequality. Journal of Economic Theory, 2, 244–263.
Ayadi, M., El Lahga, A., Chtioui, N. (2007). Poverty and inequality in Tunisia: a non-monetary approach. PMMA Working Paper No. 2007-05.
Bibi, S. (2004). Comparing multidimensional poverty between Egypt and Tunisia. CIRPEE Working Paper No. 04-16.
Booth, C. (1902). A case of town life. London: Macmillan.
Chakravarty, S. R. (1983). Ethically flexible measures of poverty. The Canadian Journal of Economics, 16, 74–85.
Chakravarty, S. R. (1990). Ethical social index numbers. New York: Springer-Verlag.
Chtioui, N. (2004). Les Mesures de la Pauvreté Multidimensionnelle. Mémoire de Mastère de Modélisation non publiée. Tunis: Institut Supérieur de Gestion.
Dalton, H. (1920). The measurement of the inequality incomes. Economic Journal, 30, 348–362.
Donaldson, D., & Weymark, J. A. (1980). A single-parameter generalization of the Gini indices of inequality. Journal of Economic Theory, 22, 67–86.
Duclos, J.-Y., & Gregoire, P. (2002), Absolute and Relative Deprivation and the Measurement of Poverty, Review of Income and Wealth, series 48(4).
Duclos, J.-Y., Sahn, D., & Younger, S.D. (2006). Robust multidimensional poverty comparisons. Economic Journal, 116(514), 943–968.
Efron, B. (1979). Bootstrap methods: another look at the jackknife. Annals of Statistics, 7, 1–26.
Efron, B. (1981). Nonparametric estimates of standard error: The jackknife, the bootstrap and other methods. Biometrika, 68, 589–599.
Efron, B. (1982). The Jackknife, the Bootstrap, and other resampling plans. Philadelphia: Society for Industrial and Applied Mathematics.
Efron, B., & Tibshirani, R. J. (1993). An introduction to the Bootstrap. New York: Chapman and Hall.
Filmer, D., & Pritchett, L. (1998). Estimating wealth effects without expenditure data or tears: an application of educational enrollment in states of India, Mimeo. Washington: The World Bank.
Foster, J. E. (1984). On economic poverty: a survey a aggregate measures. In R. L. Basmann, G. F. Rhodes (Eds.), Advances in Econometrics, 3.Connecticut (pp. 215–251). Bingley: JAI Press.
Foster, J. E. & Sen, A. (1997). Economic inequality after a quarter century. In On Economic Inequality (expanded edition). Oxford: Clarendon Press.
Hall, P. (1992). The Bootstrap and Edgeworth expansion. New York: Springer.
Hammer, J. (1998). Health outcomes across wealth groups in Brazil and India. Mimeo. DECRG. Washington: The World Bank.
Jenkins, S. P., & Lambert, P. J. (1997). Three is of poverty curves, with an analysis of UK poverty trends. Oxford Economic Papers, 49, 317–327
Kolm, S. C. (1969). The optimal production of social justice. In J. Magolis & H. Guittor (Eds.), Public economics (pp. 145–200). London: Macmillan.
Moran, T. (2005). Bootstrapping the LIS: statistical inference and patterns of inequality in the Global North. LIS Working Paper No. 378.
Osberg, L. (2000). Poverty in Canada and the United States: measurement, trends, and implications. The Canadian Journal of Economics/Revue Canadienne d’Economique, 33(4), 847–877.
Osberg, L. & Xu, K. (1999). Poverty intensity-How well do Canadian provinces compare? Canadian Public Policy, 25(2), 179–195.
Osberg, L. & Xu, K. (2000). International comparison of poverty intensity: Index decomposition and bootstrap inference. Journal of Human Resources, 35(1), 51–81.
Pollak, R. A. (1971). Additive utility functions and linear Engel curves. Review of Economic Studies, 38, 401–414.
Ravallion, M. (1992). On hunger and public action: a review article on the book by Jean Dreze and Amartya Sen. The World Bank Research Observer, 7, 1–16.
R Development Core Team (2008). R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. ISBN 3-900051-07-0, URL http://www.R-project.org
Robeyns, I. (2005). The capability approach: a theoretical survey. Journal of Human Development, 6(1), 93–117.
Rowntree, B. S. (1901). Poverty: a study of town life. London: Macmillan.
Seidl, C. (1988). Poverty measurement: a survey. In D. Bös, M. Rose & C. Seidl (Eds.), Welfare and efficiency in public economics (pp. 219–231). Heidelberg: Springer-Verlag.
Sen, A. (1973). On Economic Inequality. Oxford: Clarendon Press.
Sen, A. (1976). Poverty: an ordinal approach to measurement. Econometrica, 44, 219–231.
Sen, A. (1985). Commodities and capabilities (p. 89). New Delhi: Oxford India paper backs.
Shorrocks, A. F. (1995). Revisiting the Sen Poverty index. Econometrica, 63, 1225–1230.
Stifel, D., Sahn, D. & Younger, S. (1999). Inter-temporal changes in welfare: preliminary results from nine African countries. CFNPP WP, 6–94.
Streeten, P. (1981). First things first: meeting basic human needs in developing countries. New York: Oxford University Press.
Thon, D. (1979). On measuring poverty. Review of Income and Wealth, 25, 429–439.
Thon, D. (1983). A note on a troublesome axiom for poverty indices. Economic Journal, 93, 199–200.
Weymark, J. A. (1981). Generalized Gini inequality indices. Mathematical Social Sciences, 1, 409–430.
Xu, K. (1998). The statistical inference for the Sen-Shorrocks-Thon index of poverty intensity. Journal of Income Distribution, 8(1), 143–152.
Xu, K. (2000). Inference for generalized Gini indices using the iterated bootstrap method. Journal of Business and Economics Statistics, 18(2), 223–227.
Xu, K., & Osberg, L. (2002). On Sen’s approach to poverty measures and recent developments, paper presented at the Sixth International Meeting of the Society for Social Choice and Welfare (2002). Printed in Chinese in China Economic Quarterly, 1, 151–170.
Zhen G, B. (1993). Poverty measurement, statistical inference, and an application to the United States. Ph.D Dissertation, West Virginia University, Virginia.
Zheng, B. (1997). Aggregate poverty measures. Journal of Economic Surveys, 11, 123–163.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
Appendix A: Proofs
1.1 Proof of Proposition 3
Proof 1:
\( \phi \left( m \right) = \sum\nolimits_{j = 1}^{m} {\pi_{j} } ,m \in I. \)
Proof 2:
\( \pi_{j} = \varphi \left( i \right) - \varphi \left( {i - 1} \right) \).
For i = m, j = 1; prove that \( \pi_{1} = \varphi \left( m \right) - \varphi \left( {m - 1} \right). \).
By proof 1, we have \( \varphi \left( m \right) = \sum\nolimits_{j = 1}^{m} {\pi_{j} } = \sum\nolimits_{j = 1}^{m} {w_{m - j + 1} } = \sum\nolimits_{i = 1}^{m} {w_{i} } . \)
We have j = m−i + 1, if \( i = 1 \to j = m, \)if \( i = m \to j = 1 \)and if \( i = m - 1 \to j = 2 \).
Then, by change of variable in the precedent equation, we obtain
Verification
.
Then,
.
Gradually, we obtain \( \pi_{j} = \varphi \left( i \right) - \varphi \left( {i - 1} \right). \)
Proof 3:
\( \pi_{j} = \varphi \left( {m - j + 1} \right) - \varphi \left( {m - j} \right). \)
We have, j = m−i + 1, then by change of variable, we obtain:
\(\pi_j = \phi{i}-\phi{i-1} = \phi{m-j+1}-\phi{m-j}.\)
1.2 Proof of Proposition 4
Similar to the proof of Proposition 1, we just have to replace i by n−j + 1.
1.3 Proof of Proposition 5
We have \( \pi_{j} = w_{n - j + 1} \).
Then \( w_{1} \le w_{2} \le {\ldots} \le w_{n} \Leftrightarrow \pi_{1} \ge \pi_{2} \ge {\ldots} \ge \pi_{n} \).
So the function \( \varphi \) with \( \pi_{1} \ge \pi_{2} \ge {\ldots} \ge \pi_{n} \).
satisfies the equation \( \varphi (kn) = \varphi (k)\varphi (n), \)
if and only if \( \varphi (n) = n^{\nu } \).
1.4 Proof of lemma 1
I/\( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} = \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} = n^{\nu }. \)
Let, \( k = i - 1 \Rightarrow i = 1 \to k = 0 \) and \( i = n \to k = n - 1 \).
Proof 1:
\( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} = n^{\nu } . \)
Proof 2:
\( \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} = n^{\nu } . \)
We have j = n−i + 1, then \( \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} = \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} . \)
By proof 1, we have \( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} = n^{\nu }. \)
Then \( \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} = n^{\nu }.\)
II/\( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} \tilde{x}_{i}^{*} = \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]\hat{x}_{j}^{*} }. \)
We have j = n−i + 1, then \( i^{\nu } - \left( {i - 1} \right)^{\nu } = \left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } .\)
And we have \( \sum\nolimits_{i = 1}^{n} {\tilde{x}_{i}^{*} } = \sum\nolimits_{j = 1}^{n} {\hat{x}_{j}^{*} } . \) \( \Rightarrow \) \( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} \tilde{x}_{i}^{*} = \sum\nolimits_{j = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} \hat{x}_{j}^{*} . \) \( \Rightarrow \) \( \sum\nolimits_{i = 1}^{n} {\left[ {i^{\nu } - \left( {i - 1} \right)^{\nu } } \right]} \tilde{x}_{i}^{*} = \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} \hat{x}_{j}^{*} . \)
Dividing by \( n^{\nu } \), we obtain \( \Upxi_{S - SST}^{n} \left( {\tilde{x}^{*} } \right) = \hat{\Upxi }_{S - SST}^{n} \left( {\hat{x}^{*} } \right). \)
1.5 Proof of Proposition 6
Proof 1:
\( S - SST^{R} \left( {\underline{x}^{*} ;z} \right) = \frac{{\left[ {z - \hat{\Upxi }_{{G^{\nu } }} \left( {\underline{{\hat{x}}}^{*} } \right)} \right]}}{z}. \)
By Eq. (6.41), we have
.
Proof 2:
\( S - SST^{R} \left( {\underline{x}^{*} ;z} \right) = \hat{\Upxi }_{{G^{\nu } }} \left( {\underline{y}^{*} } \right). \)
We have \( \sum\nolimits_{j = 1}^{n} {\left[ {\left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } } \right]} = n^{\nu } . \).
Hence,
where \( y_{j}^{*} = \frac{{z - \hat{x}_{j}^{*} }}{z} \) so \( \hat{x}_{j}^{*} \uparrow \Rightarrow y_{j}^{*} \downarrow \).
1.6 Proof of Lemma 2
Similar to the proof of Lemma 1, we just have to replace \( i^{\nu } - (i - 1)^{\nu } \) by \( w_{i} \) and \( \left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } \) by \( \pi_{j} \).
1.7 Proof of Proposition 7
Similar to the proof of Proposition 6, we just have to replace \( i^{\nu } - \left( {i - 1} \right)^{\nu } \)by \( w_{i} \) and \( \left( {n - j + 1} \right)^{\nu } - \left( {n - j} \right)^{\nu } \) by \( \pi_{j} \).
Appendix B: Multivariate Correspondence Analysis
Appendix C: Analysis of Poverty in Tunisia
Rights and permissions
Copyright information
© 2013 Springer Science+Business Media New York
About this chapter
Cite this chapter
Chtioui, N., Ayadi, M. (2013). A Non-Monetary Multidimensional Poverty Analysis of Tunisia Using Generalized Sen-Shorrocks-Thon Measures. In: Berenger, V., Bresson, F. (eds) Poverty and Social Exclusion around the Mediterranean Sea. Economic Studies in Inequality, Social Exclusion and Well-Being, vol 9. Springer, Boston, MA. https://doi.org/10.1007/978-1-4614-5263-8_6
Download citation
DOI: https://doi.org/10.1007/978-1-4614-5263-8_6
Published:
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4614-5262-1
Online ISBN: 978-1-4614-5263-8
eBook Packages: Business and EconomicsEconomics and Finance (R0)