Abstract
The consequences that educational underperformance has on both individuals and society as a whole lead policy makers and planners to focus on how to measure it properly. The aim of this paper is to propose an index to measure educational poverty which, taking as a starting point the economic literature on multidimensional poverty measurement, turns out to be appropriate in the educational context. With this purpose, the following two features are demanded: (1) an individual should be identified as poor whenever they do not reach the basic level of knowledge in at least one of the relevant subjects; (2) the degree of poverty of individuals who present the same level of insufficiency in some subjects but have different scores in others should be different. Based on these premises, we introduce a multidimensional adjusted poverty index, called BCa index, which is an extension of Bourguignon and Chakravarty index, and we apply it to measure educational poverty in the OECD countries by using data from PISA 2012 and 2015 reports.
Similar content being viewed by others
Notes
Once we have constructed the deprivation matrix, \( G\left( {{\mathbb{X}},{\text{z}}} \right) \), the proportion of people who are poor and deprived in any set of attributes can be easily obtained (see “Appendix 1”).
References
Alkire, S., Foster, J., Seth, S., Santos, M. E., Roche, J. M., & Ballón, P. (2015). Multidimensional poverty measurement and analysis. Oxford: Oxford University Press.
Bourguignon, F., & Chakravarty, S. R. (2003). The measurement of multidimensional poverty. Journal of Economic Inequality, 1, 25–49.
Chakravarty, S. R. (2009). Inequality, polarization and poverty. Berlin: Springer.
Denny, K. (2002). New methods for comparing literacy across populations: insights from the measurement of poverty. Jouranl of the Royal Statistical Society A, 165(Part 3), 481–493.
Dutta, I., Pattanaik, P. K., & Xu, Y. (2003). On measuring deprivation and the standard of living in a multidimensional framework on the basis of aggregate data. Economica, 70, 197–221.
Foster, J. E., Greer, J., & Thorbecke, E. (1984). A class of decomposable poverty measures. Econometrica, 52, 761–766.
Kelly, D., Nord, C. W., Jenkins, F., Chan, J. Y., & Kastberg, D. (2013). Performance of US 15-year-old students in mathematics, science, and reading literacy in an international context. first look at PISA 2012. NCES 2014-024. National Center for Education Statistics.
Lasso de la Vega, C., Urrutia, A., & Diez, H. (2009). The Bourguignon and Chakravarty multidimensional poverty family: A characterization. Working Papers No 109, ECINEQ, Society of the Study of Economic Inequality (109).
Lohmann, H., & Ferger, F. (2014). Educational poverty in a comparative perspective: Theoretical and empirical implications. SFB 882 Working Paper Series, no 26, DFG Research Center (SFB) from Heterogeneities to Inequalities. http://www.sfb882.uni-bilefeld.de/. Accessed 21 Dec 2018.
Minzyuk, L., & Russo, F. (2016). La misurazione multidimensionale della povertà in istruzione in Italia multidimensional measurement of educational poverty in Italy. Politica economica, 1, 65–122.
OECD. (2014a). PISA 2012 results: What students know and can do—Student performance in mathematics, reading and science (Revised ed., Vol. I). Paris: PISA, OECD Publishing. https://doi.org/10.1787/9789264208780-en.
OECD. (2014b). PISA 2012 technical report. Paris: OECD Publishing.
OECD. (2016a). PISA 2015 results: Excellence and equity in education. Paris: PISA, OECD Publishing. https://doi.org/10.1787/9789264266490-en.
OECD. (2016b). PISA 2015 results in focus. https://www.oecd.org/pisa/pisa-2015-results-in-focus.pdf. Accessed 21 Dec 2018.
Permanyer, I. (2014). Assessing individuals’ deprivation in a multidimensional framework. Journal of Development Economics, 109, 1–16.
Sen, A. K. (1976). Poverty: an ordinal approach to measurement. Econometrica, 44, 219–231.
Thomas, V., Wang, Y., & Fan, X. (2000) Measuring educational inequality: Gini coefficients of education. Working Paper 2525. World Bank, Washington DC: World Bank.
Villar, A. (2016). Educational poverty as a welfare loss: Low performance in the OECD according to PISA 2012. Modern Economy, 7, 441–449. https://doi.org/10.4236/me.2016.74049.
Acknowledgements
This work was supported by the Ministerio de Economía y Competitividad (Spain) under Grant Numbers: ECO2013-43119-P; ECO2016-77200-P.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix 1: Calculation of the Proportion of the People Who Are Poor and Deprived in Any Set of Attributes
Let \( k \in N, J = \left\{ {1, \ldots ,k} \right\} \) and \( S \in P\left( J \right){ \setminus }\emptyset , \) where \( P\left( J \right) \) denotes the set of all the subsets of J. We define the exclusive identification function, \( {\text{ID}}^{E} :{\mathbb{R}}_{ + }^{k} \times P\left( J \right){ \setminus }\emptyset \to \left\{ {0,1} \right\} \), such that for any \( x \in {\mathbb{R}}_{ + }^{k} \) and any \( S \in P\left( J \right){ \setminus }\emptyset \),
Therefore, for any \( \left( {{\mathbb{X}},z} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \), \( J = \left\{ {1, \ldots ,k} \right\} \), \( S \in P\left( J \right){ \setminus }\emptyset , \) and by denoting the i-th row of matrix G \( \left( {{\mathbb{X}},z} \right) \) by \( g_{i} \left( {{\mathbb{X}},z} \right) \), then
provides the number of poor who are exclusively deprived in the set of attributes S.
Therefore, for any \( \left( {{\mathbb{X}},z} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \), \( j \in J = \left\{ {1, \ldots ,k} \right\} \),
provides the number of poor people who are exclusively deprived in attribute \( j \). Moreover, \( \sum\nolimits_{i = 1}^{n} {ID^{E} \left( {g_{i} \left( {{\mathbb{X}},z} \right),J} \right)} \) provides the number of people deprived in all the attributes.
We denote, for all \( x \in {\mathbb{R}} \), by
Moreover, for any \( \left( {{\mathbb{X}},z} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \), \( j \in J = \left\{ {1, \ldots ,k} \right\} \), \( i \in N = \left\{ {1, \ldots ,n} \right\}\;{\text{and}}\;S \in P\left( J \right){ \setminus }\emptyset , \)
provides the number of poor in the population;
is the proportion of poor who are exclusively deprived in all the attributes belonging to S;
is the proportion of the population that is exclusively deprived in all the attributes belonging to S.
Appendix 2: Proof of Proposition 1
Proposition 1
Given\( \left( {{\mathbb{X}}, z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \times {\mathbb{R}}_{ + }^{k} \)the Adjustment Factor,\( A_{i} \left( {{\mathbb{X}},z,m} \right) \), satisfies:
-
i.
It is non-negative; if for any\( {\text{i}} \in {\text{N}} \)and\( j, t \in \left\{ {1, \ldots , k} \right\} \), \( j \ne t, r_{ij} \left( {{\mathbb{X}}, z, m} \right) > 0 \)and\( g_{it} \left( {{\mathbb{X}}, z} \right) > 0 \), then\( A_{i} \left( {{\mathbb{X}},z,m} \right) \) > 0; and\( A_{i} \left( {{\mathbb{X}},z,m} \right) = 1 \)when the individual is deprived in all the attributes.
-
ii.
It is increasing with respect to \( \phi_{i} \left( {G\left( {{\mathbb{X}}, z} \right)} \right) \) and decreasing with respect to \( \phi_{i} \left( {R\left( {{\mathbb{X}}, z,m} \right)} \right). \)
-
iii.
Deprivation and non-deprivation levels considered to construct it are defined in a parallel way.
-
iv.
For each\( i \in {\text{N}}, \)the curves defined by the sets of pairs\( \left\{ {(\phi_{i} \left( {G\left( {{\mathbb{X}}, z} \right)} \right) ,\phi_{i} \left( {R\left( {{\mathbb{X}}, z,m} \right)} \right) \in \left[ {0,1\left] \times \right[0,1} \right]\text{ / } BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = C} \right\} \), called\( BC_{i}^{a} \)-isopoverty curves, are increasing and convex.
□
Proof
-
i.
It is obviously satisfied by definition of the Adjustment Factor,\( A_{i} \left( {{\mathbb{X}},z,m} \right) \).
-
ii.
\( A_{i} \left( {{\mathbb{X}},z,m} \right) \) is increasing with respect to \( \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \)
since
$$ \partial A_{i} \left( {{\mathbb{X}},z,m} \right)/\partial \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = \frac{{\phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)}}{{\left[ {\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) + \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)} \right]^{2} }} > 0. $$\( A_{i} \left( {{\mathbb{X}},z,m} \right) \) is decreasing with respect to \( \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) \)) since
$$ \partial A_{i} \left( {{\mathbb{X}},z,m} \right)/\partial \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) = \frac{{ - \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right)}}{{\left[ {\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) + \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)} \right]^{2} }} < 0. $$ -
iii.
Deprivation and non-deprivation levels considered to construct the Adjustment Factor are defined in a parallel way since \( A_{i} \left( {{\mathbb{X}},z,m} \right) =\upzeta(f_{1} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \), \( f_{2} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) \) where \( f_{1} = f_{2} = \phi_{i} \).
-
iv.
For each \( i \in {\text{N}}, \) consider the \( BC_{i}^{a} \)-isopoverty curve \( \left\{ {(\phi_{i} \left( {G\left( {{\mathbb{X}}, z} \right)} \right) ,\phi_{i} \left( {R\left( {{\mathbb{X}}, z,m} \right)} \right) \in \left[ {0,1\left] \times \right[0,1} \right]\text{ / } BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = C} \right\} \).
Since
$$ BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = \frac{{\left( {\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) } \right)^{2} }}{{\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) + \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)}} = C $$then
$$ \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) = \frac{{\left( {\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) } \right)^{2} }}{C} - \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right). $$Then,
$$ d\phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)/d\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = \left( {2\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) - C} \right)/C > 0 $$since
\( C = BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \times A_{i} \left( {{\mathbb{X}},z,m} \right) \le \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) < 2\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \), that is, the \( BC_{i}^{a} \)-isopoverty curve is increasing.
Moreover,
$$ d^{2} \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right)/d\left( {\phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) } \right)^{2} = 2/C > 0, $$so, the \( BC_{i}^{a} \)-isopoverty curve is convex.
□
Appendix 3: Proof of Remark 1
-
i.
The Permanyer’s index is not consistent in the sense that it measures the deprivation and non-deprivation levels by applying different functions to surplus and poverty gaps.
-
ii.
The construction of Permanyer’s index requires estimating \( \left[ {k\left( {k - 1} \right) + 1} \right] \) parameters more than our index (where k is the number of attributes).
-
iii.
Permanyer’s index is not sensitive to the deprivation levels of the individuals because it is not a function of \( \phi_{i} \left( {G\left( {{\mathbb{X}}, z} \right)} \right) \).
-
iv.
In general, the isopoverty curves defined from Permanyer’s index are not convex.
Proof
-
i.
See the definition of excess gaps and poverty gaps.
-
ii.
By definition, Permanyer’s index (Permanyer 2014) requires estimating \( \gamma \) and the values of \( \lambda_{jl} \) for all \( j,l \in \left\{ {1, \ldots ,k} \right\} \), such that \( l \ne j \), that is, \( \left[ {k\left( {k - 1} \right) + 1} \right] \) parameters more than our proposal, where k is the number of attributes. In Permanyer’s words, his proposal is over-parametrised.
In order to prove (iii) and (iv), we consider the following specification of the general Permanyer’s formulation of a multidimensional poverty index (\( P^{P} ) \).
with \( \varphi_{jl} \left( {r_{il} (\left( {{\mathbb{X}}, z,m} \right)} \right) = 1 + \left( {\lambda_{jl} - 1} \right)r_{il}^{\gamma } \left( {\left( {{\mathbb{X}}, z,m} \right)} \right) \), where \( \psi \left( x \right) = \left( {x/k^{1/\theta } } \right)^{\alpha } \), \( \lambda_{jl} = \lambda \) for all j, l, \( \lambda \in \left( {0,1} \right] \) and \( \gamma > 0 \). Then,
If \( 1 - \lambda = \beta \), \( \beta \in \left[ {0,1} \right) \), the Permanyer’s correction function for any agent i can be written as \( A_{i}^{P} = \left( {1 - \beta r_{i1}^{\gamma } \left( {{\mathbb{X}}, z,m} \right)} \right)\left( {1 - \beta r_{i2}^{\gamma } \left( {{\mathbb{X}}, z,m} \right)} \right) \cdots \left( {1 - \beta r_{ik}^{\gamma } \left( {{\mathbb{X}}, z,m} \right)} \right) = \mathop \prod \limits_{l = 1}^{k} \left( {1 - \beta r_{il}^{\gamma } \left( {{\mathbb{X}}, z,m} \right)} \right). \) Thus, the individual poverty levels of any individual \( i \in \left\{ {1, \ldots ,n} \right\} \) according to this index is:
-
iii.
The previous expression shows that Permanyer’s correction function, \( A_{i}^{P} \left( {{\mathbb{X}}, z,m} \right) \) is not a function depending on \( \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \). That is, it is not sensitive to the level of deprivation of the individual since \( \partial A_{i}^{P} \left( {{\mathbb{X}}, z,m} \right)/\partial \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = 0. \)
-
iv.
Next counter example shows that the isopoverty curves defined from Permanyer’s index, \( \left\{ {(\phi_{i} \left( {G\left( {{\mathbb{X}}, z} \right)} \right) ,\phi_{i} \left( {R\left( {{\mathbb{X}}, z,m} \right)} \right) \in \left[ {0,1\left] \times \right[0,1} \right]\text{ / } P_{i}^{P} \left( {{\mathbb{X}},z,m} \right) = C} \right\}, \) are not convex.
Suppose there are two attributes, \( j = \left\{ {1,2} \right\} \), \( \theta = 2 \), \( \gamma = 1 \), \( \beta = 1/2 \), \( g_{i} = (g_{i1} ,0) \) and \( r_{i} = (0,r_{i2} ) \).
Then, \( \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = g_{i1} /\sqrt 2 \), \( A_{i}^{P} = 1 - (r_{i2} /2) \) and \( P_{i}^{P} = \left( {g_{i1} /\sqrt 2 } \right)\left[ {1 - (r_{i2} /2)} \right] \).
If \( g_{i} = \left( {0.5,0} \right) \) and \( r_{i} = \left( {0,0.71} \right) \), we have that
$$ P_{i}^{P} = \left( {0.5/\sqrt 2 } \right)\left[ {1 - (0.71/2)} \right] = 0.3225/\sqrt 2 . $$Now, we want to know the non-deprivation level, \( r_{i2}^{*} \), which combined with the previous deprivation level increased by 0.05, \( g_{i1} + 0.05, \) provides the same poverty level, \( P_{i}^{P} = 0.3225/\sqrt 2 . \)
Then, solving the equation
$$ 0.3225/\sqrt 2 = \left[ {\left( {0.5/\sqrt 2 } \right) + 0.05} \right]\left[ {1 - \left( {r_{i2}^{*} /2} \right)} \right] $$we obtain \( r_{i2}^{*} = 0.869830152 \), so \( \Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2}^{*} )} \right] = 0.615062799 \). Therefore, the pair \( \left( {\left( {0.5/\sqrt 2 } \right) + 0.05, 0.615062799} \right) \) also belongs to the isopoverty curve of level \( 0.3225/\sqrt 2 \).
Analogously, we want to know the non-deprivation level which combined with the initial deprivation level increased by another 0.1, \( g_{i1} + 0.1 \), provides the same isopoverty level, \( P_{i}^{P} = 0.3225/\sqrt 2 . \)
Then, solving the equation
$$ 0.3225/\sqrt 2 = \left[ {\left( {0.5/\sqrt 2 } \right) + 0.1} \right]\left[ {1 - \left( {r_{i2}^{**} /2} \right)} \right] $$we obtain \( r_{i2}^{**} = 0.994420759 \), so \( \Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2}^{*} )} \right] = 0.703161662 \). Therefore, the pair \( \left( {\left( {0.5/\sqrt 2 } \right) + 0.1, 0.703161662} \right) \) also belongs to the isopoverty curve of level \( 0.3225/\sqrt 2 \).
Then, we have that when \( \Phi _{i}^{G} = 0.5/\sqrt 2 \) and it increases by 0.05, it has to be compensated by an increase of the non-deprived level that amounts to \( \Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2}^{*} )} \right] -\Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2} )} \right] = 0.113016985 \) to maintain the poverty level at \( 0.3225/\sqrt 2 . \) However, when \( \Phi _{i}^{G} = \left( {0.5/\sqrt 2 } \right) + 0.05 \) and it increases by 0.05, it has to be compensated by a smaller increase of the non-deprived level, which amounts to \( \Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2}^{**} )} \right] -\Phi _{i}^{R} \left[ {(r_{i1} ,r_{i2}^{*} )} \right] = 0.088098863 \), to maintain the poverty level at \( 0.3225/\sqrt 2 \), which is completely unnatural in the educational context.
□
Appendix 4: Proof of Proposition 2
Proposition 2
The \( BC^{a} \) index, \( BC^{a} :{\mathcal{M}} \times {\mathbb{R}}_{+}^{k} \times {\mathbb{R}}_{+}^{k}\,{\rightarrow}\,\left[{0,1} \right] \) , satisfies SFI, WF, SYM, NOM, MON, NDMON, CONT, SI, PP and SUD.
Proof
-
(SFI) Let us consider \( \left( {{\mathbb{X}}, z,m} \right) \;{\text{and}} \;\left( {{\mathbb{Y}},z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{\text{k}} \times {\mathbb{R}}_{ + }^{\text{k}} \), \( j \in \left\{ {1, \ldots ,k} \right\}, \)\( i \in \left\{ {1, \ldots ,n} \right\} \) if \( {{x}}_{{ij}} \ge {{z}}_{{j}} \), \( y_{ij} = x_{ij} +\updelta, \;{\text{where}} \;\updelta > 0,\; y_{tj} = x_{tj} \;{\text{for}}\; {\text{all}}\; t \ne i \;{\text{and}}\; j \in \left\{ {1, \ldots ,k} \right\}, \) then
$$ BC_{i}^{a} \left({{\mathbb{X}},z,m} \right) > 0\,{\leftrightarrow}\,\phi_{i} \left({G\left({{\mathbb{X}},z} \right)} \right) > 0\,{\leftrightarrow}\,\phi_{i} \left({G\left({{\mathbb{Y}},z} \right)} \right) > 0\,{\leftrightarrow}\,BC_{i}^{a} \left({{\mathbb{Y}},z,m} \right) > 0. $$ -
(WF) If for some \( i, \,x_{ik} { \geqq }z_{k} \forall k, \)\( {\text{then}} \;g_{ik} = 0\,\, \forall k, \;{\text{which}} \;{\text{implies}}\; \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = 0, \;{\text{so}}\; BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = BC_{i}^{a} \left( {{\mathbb{Y}},z,m} \right) = 0. \) Moreover, since \( \forall r \ne i \)\( BC_{\text{r}}^{a} \left( {{\mathbb{X}},z,m} \right) = BC_{\text{r}}^{a} \left( {{\mathbb{Y}},z,m} \right), \;{\text{then}} \;BC^{a} \left( {{\mathbb{Y}},z,m} \right) \) = \( BC^{a} \left( {{\mathbb{X}},z,m} \right). \)
-
(SYM) For any \( \left( {{\mathbb{X}},z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \times {\mathbb{R}}_{ + }^{\text{k}} \), if \( \Pi \) is any permutation of the rows, then
$$ {\Pi {\mathbb{X}}} = ({\text{x}}_{\sigma \left( i \right)j} )_{{\begin{subarray}{*{20}c} {i = 1,2,..,n} \\ {j = 1,2, \ldots ,k} \\ \end{subarray} }}. $$
Since \( g_{ij} \left( {{\mathbb{X}},z} \right) = max\left\{ {0,\frac{{z_{j} - x_{ij} }}{{z_{j} }}} \right\} = max\left\{ {0,\frac{{z_{j} - x_{\sigma \left( i \right)j} }}{{z_{j} }}} \right\} = g_{\sigma \left( i \right)j} \left( {{\varPi {\mathbb{X}}},z} \right) \),
it follows that
and
then,
-
(NOM)\( For\;all\; z \in {\mathbb{R}}_{ + }^{k} , {\mathbb{X}} \in {\mathcal{M}}, if\,x_{ij} { \geqq }z_{j} \forall i \in \left\{ {1, 2, \ldots , n} \right\}, \;{\text{and}}\; \forall j \in \left\{ {1, 2, \ldots , k} \right\}, \) then \( \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) = 0, \) so \( BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = 0 \), therefore \( BC^{a} \left( {{\mathbb{X}},z,m} \right) = 0. \)
-
(MON) Since \( y_{ij} = x_{ij} + \delta \),
(a) if \( y_{ij} < z_{j} , \) then \( g_{ij} \left( {{\mathbb{Y}},{\text{z}}} \right) = g_{ij} \left( {{\mathbb{X}},{\text{z}}} \right) - \frac{\delta }{{z_{j} }} \) and \( r_{ij} \left( {{\mathbb{Y}},{\text{z}},{\text{m}}} \right) = 0, \) therefore \( g_{ij} \left( {{\mathbb{Y}},{\text{z}}} \right) < g_{ij} \left( {{\mathbb{X}},{\text{z}}} \right) \) and \( \phi_{i} \left( {G\left( {{\mathbb{Y}},z} \right)} \right) < \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \). Moreover \( \phi_{i} \left( {R\left( {{\mathbb{Y}},z,m} \right)} \right) = \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) \). By applying (ii) of Proposition 1, \( A_{i} \left( {{\mathbb{Y}},z,m} \right) < A_{i} \left( {{\mathbb{X}},z,m} \right) \), so \( BC_{\text{i}}^{a} \left( {{\mathbb{Y}},z,m} \right) < BC_{\text{i}}^{a} \left( {{\mathbb{X}},z,m} \right), \) then \( BC^{a} \left( {{\mathbb{Y}},z,m} \right) < BC^{a} \left( {{\mathbb{X}},z,m} \right). \)
(b) If \( y_{ij} > \)\( z_{j} , g_{ij} \left( {{\mathbb{Y}},{\text{z}}} \right) = 0, g_{ij} \left( {{\mathbb{X}},{\text{z}}} \right) > 0 \) and \( r_{ij} \left( {{\mathbb{Y}},{\text{z}},{\text{m}}} \right) > 0 \), then \( \phi_{i} \left( {G\left( {{\mathbb{Y}},z} \right)} \right) < \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \) and \( \phi_{i} \left( {R\left( {{\mathbb{Y}},z,m} \right)} \right) > \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right) \). Analogously by applying (ii) in Proposition 1, \( A_{i} \left( {{\mathbb{Y}},z,m} \right) < A_{i} \left( {{\mathbb{X}},z,m} \right) \), so \( BC_{\text{i}}^{a} \left( {{\mathbb{Y}},{\text{z}},{\text{m}}} \right) < BC_{\text{i}}^{a} \left( {{\mathbb{X}},{\text{z}},{\text{m}}} \right), \) then \( BC^{a} \left( {{\mathbb{Y}},z,m} \right) < BC^{a} \left( {{\mathbb{X}},z,m} \right) \).
-
(NDMON) Since \( x_{ij} > z_{j} \), then \( y_{ij} = x_{ij} + \delta > x_{ij} > z_{j} \) implies that \( g_{ij} \left( {{\mathbb{Y}},{\text{z}}} \right) = g_{ij} \left( {{\mathbb{X}},{\text{z}}} \right) = 0\), therefore \( \phi_{i} \left( {G\left( {{\mathbb{Y}},z} \right)} \right) = \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \). Moreover,
$$ r_{ij} \left( {{\mathbb{Y}},z,m} \right) = \frac{{x_{ij} + \delta - z_{j} }}{{m_{j} - z_{j} }} > \frac{{x_{ij} - z_{j} }}{{m_{j} - z_{j} }} = r_{ij} \left( {{\mathbb{X}},z,m} \right),$$then
$$ \phi_{i} \left( {R\left( {{\mathbb{Y}},z,m} \right)} \right) > \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right). $$So, by applying Proposition 1, \( A_{i} \left( {{\mathbb{Y}},z,m} \right) < A_{i} \left( {{\mathbb{X}},z,m} \right) \), and
$$ BC_{i}^{a} \left( {{\mathbb{Y}},z,m} \right) \leq BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right) = \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right) \times A_{i} \left( {{\mathbb{X}},z,m} \right).\quad Therefore\quad BC^{a} \left( {{\mathbb{Y}},z,m} \right) \leq BC^{a} \left( {{\mathbb{X}},z,m} \right). $$ -
\( \left( {CONT} \right)\,BC^{a} \left( { {\mathbb{X}} , z,m} \right) \) is a composition of continuous functions.
-
(SI) Since \( x_{ij}^{'} = \lambda_{j} x_{ij},\,z_{j}^{'} = \lambda_{j} z_{j} , m_{j}^{'} = \lambda_{j} m_{j} \), it is obtained that
$$ g_{ij} \left( {{{\mathbb{X}}^{\prime}},z'} \right) = max\left\{ {0,\frac{{z_{j}^{'} - x_{ij}^{'} }}{{z_{j}^{'} }}} \right\} = max\left\{ {0,\frac{{\lambda_{j} z_{j} - \lambda_{j} x_{ij} }}{{\lambda_{j} z_{j} }}} \right\} = g_{ij} \left( {{\mathbb{X}},{\text{z}}} \right) $$and
$$ r_{ij} \left( {{{\mathbb{X}}^{\prime}},z',m'} \right) = max\left\{ {0,\frac{{x_{ij}^{'} - z_{j}^{'} }}{{m_{j}^{'} - z_{j}^{'} }}} \right\} = r_{ij} \left( {\mathbb{X}}, z, m \right) $$Then,
\( \phi_{i} \left( {G\left( {{{\mathbb{X}}^{\prime}},z'} \right)} \right) = \phi_{i} \left( {G\left( {{\mathbb{X}},z} \right)} \right)\, {\text{and}}\, \phi_{i} \left( {R\left( {{{\mathbb{X}}^{\prime}},z',m'} \right)} \right) = \phi_{i} \left( {R\left( {{\mathbb{X}},z,m} \right)} \right), \) therefore
$$ BC^{a} \left( {{\mathbb{X}},z,m} \right) = BC^{a} \left( {{{\mathbb{X}}^{\prime}},z^{\prime},m'} \right). $$ -
(PP) For any \( \left( {{\mathbb{X}},z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \times {\mathbb{R}}_{ + }^{k} , r \in {\mathbb{N}} \), finite, if \( {\mathbb{X}}^{r} \) is the r-fold of \( {\mathbb{X}} \),
$$ \begin{aligned} BC^{a} \left( {{\mathbb{X}}^{r} ,z,m} \right) & = \frac{1}{nr} \mathop \sum \limits_{i = 1}^{nr} \left( {BC_{i}^{a} \left( {{\mathbb{X}}^{r} ,z,m} \right)} \right)^{\alpha } = \frac{1}{nr}\mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{h = 1}^{r} \left( {BC_{{i_{h} }}^{a} \left( {{\mathbb{X}}^{r} ,z,m} \right)} \right)^{\alpha } \\ & = \frac{1}{nr}\mathop \sum \limits_{i = 1}^{n} r\left( {BC_{i}^{a} \left( {{\mathbb{X}}^{r} ,z,m} \right)} \right)^{\alpha } = BC^{a} \left( {{\mathbb{X}},z,m} \right) \\ \end{aligned} $$ -
(SUD) For any \( \left( {{\mathbb{X}},z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{\text{k}} \times {\mathbb{R}}_{ + }^{\text{k}} \) and any partition of the population into s subgroups, s≥ 2:
$$ BC^{a} \left( {{\mathbb{X}},z,m} \right) = \frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left( {BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right)} \right)^{\alpha } = \frac{{n_{l} }}{n}\mathop \sum \limits_{l = 1}^{S} \frac{1}{{n_{l} }}\mathop \sum \limits_{i = 1}^{{n_{l} }} \left( {BC_{i}^{a} \left( {{\mathbb{X}},z,m} \right)} \right)^{\alpha } = \frac{{n_{l} }}{n}\mathop \sum \limits_{l = 1}^{S} BC^{a} \left( {{\mathbb{X}}_{l} ,z,m} \right) $$
□
Appendix 5: Proof of Proposition 3
Proposition 3
The \( BC^{a} \) index, \( BC^{a} :{\mathcal{M}} \times {\mathbb{R}}_{+}^{k} \times {\mathbb{R}}_{+}^{k}\,{\rightarrow}\,\left[{0,1} \right] \), satisfies FD if \( \alpha = \theta \).
Proof
For any \( \left( {{\mathbb{X}},z,m} \right) \in {\mathcal{M}} \times {\mathbb{R}}_{ + }^{k} \times {\mathbb{R}}_{ + }^{k} \), let \( x_{j} = \left( {x_{ij} } \right)_{{1{ \leqq }i{ \leqq }n}} \) denote the j-th column of \( {\mathbb{X}} \), which represents the achievement vector in attribute j of all the individuals \( i = 1, ..,n \). Then, the poverty level due to attribute or dimension j, according to our poverty index \( BC^{a} \), denoted by \( BC_{dj}^{a} \), is defined for \( \alpha = \theta \), as:
Then,
□
Rights and permissions
About this article
Cite this article
Sánchez-García, JF., Sánchez-Antón, MdC., Badillo-Amador, R. et al. A New Extension of Bourguignon and Chakravarty Index to Measure Educational Poverty and Its Application to the OECD Countries. Soc Indic Res 145, 479–501 (2019). https://doi.org/10.1007/s11205-019-02115-x
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11205-019-02115-x