Abstract
We solve the problem of a social planner who seeks to minimize inequality via transfers with a fixed public budget in a distribution of exogenously given incomes. The appropriate solution method depends on the objective function: If it is convex, it can be solved by an interiorpoint algorithm. If it is quasiconvex, the bisection method can be used. Using artificial and realworld data, we implement the procedures and show that the optimal transfer scheme need not comply with a transfer scheme that perfectly equalizes incomes at the bottom of the distribution.
This is a preview of subscription content, access via your institution.
Notes
If we optimize with respect to t_{i}, then \(\frac {y_{i}+t_{i}}{ES_{i}}\) is just an affine transformation of the t_{i} and therefore preserves concavity or convexity. Changing the sums to be weighted also preserves concavity or convexity. See Boyd and Vandenberghe (2004).
See Boyd and Vandenberghe (2004).
As shown in Yitzhaki and Schechtman (2012) there are more than a dozen alternative ways to define the Gini index.
Yitzhaki and Lambert (2013) investigate the relationships between Gini’s mean difference (GMD), the mean absolute deviation, the least absolute deviation, and the absolute deviation from a quantile.
See Boyd and Vandenberghe (2004).
Lambert and Yitzhaki (2013) show that the absolute mean deviation is a special case of the betweengroup Gini mean difference (BGMD). In contrast to the Gini index, the BGMD is not normalized by the mean and is convex.
The relative mean deviation is convex as, in contrast to the absolute mean deviation, the relative mean deviation is divided by the mean.
See Boyd and Vandenberghe (2004, p. 87).
See Boyd and Vandenberghe (2004, p. 294).
References
Arrow, K.J., Enthoven, A.C.: Quasiconcave programming. Econometrica 29(4), 779–800 (1961)
Atkinson, A.B.: On the measurement of inequality. J. Econ. Theory 2(3), 244–263 (1970)
Blackorby, C., Donaldson, D.: Measures of relative equality and their meaning in terms of social welfare. J. Econ. Theory 18(1), 59–80 (1978)
Boyd, S., Vandenberghe, L.: Convex optimization. Cambridge University Press, Cambridge (2004)
Charnes, A., Cooper, W.W. : Programming with linear fractional functionals. Naval Research Logistics Quarterly 9(3–4), 181–186 (1962). ISSN 19319193
Corneo, G., König, J., Schröder, C.: Distributional effects of subsidizing retirement savings accounts. Evidence from germany. Finanzarchiv / Public Finance Analysis (forthcoming) (2018)
David, G.: Luenberger. Quasiconvex programming. SIAM J. Appl. Math. 16(5), 1090–1095 (1968)
Ebert, U.: Using equivalent income of equivalent adults to rank income distributions. Soc. Choice Welf. 16(2), 233–258 (1999)
Ebert, U., Moyes, P.: Equivalence scales reconsidered. Econometrica 71(1), 319–343 (2003)
Glewwe, P.: Household equivalence scales and the measurement of inequality: transfers from the poor to the rich could decrease inequality. J. Public Econ. 44(2), 211–216 (1991)
Joumard, I., Pisu, M., Bloch, D.: Tackling income inequality: The role of taxes and transfers. OECD Journal Economic Studies 2012(1), 37–70 (2012)
Judd, K.L.: Numerical methods in economics. MIT Press, Cambridge (1998)
Prete, V., Sommacal, A., Zoli, C.: Optimal nonwelfarist income taxation for inequality and polarization reduction. University of Verona, Department of Economics Working Papers, 23 (2016)
Shalit, H., Yitzhaki, S.: The meangini efficient portfolio frontier. J. Financ. Res. 28(1), 59–75 (2005)
Shorrocks, A.: Inequality and welfare evaluation of heterogeneous income distributions. J. Econ. Inequal. 2(3), 193–218 (2004)
Wodon, Q.T., Yitzhaki, S.: Inequality and social welfare when using equivalence scales. SSRN Working Paper (2005)
Yitzhaki, S.: A tax programming model. J. Public Econ. 19(1), 107–120 (1982)
Yitzhaki, S.: On an extension of the Gini inequality index. Int. Econ. Rev. 24 (3), 617–628 (1983)
Yitzhaki, S., Schechtman, E.: The Gini methodology: a primer on a statistical methodology. Berlin/Heidelberg (2012)
Yitzhaki, S., Lambert, P.J.: The relationship between the absolute deviation from a quantile and Gini’s mean difference. Metron 71(2), 97–104 (2013)
Author information
Authors and Affiliations
Corresponding author
Additional information
We thank Giacomo Corneo, Frank Cowell, Shlomo Yitzhaki, and participants of the ECINEQ conference 2017 in New York for valuable discussions. We also thank two anonymous referees as well as the handling editor, Valentino Dardanoni. We are also grateful for comments on an earlier version of the paper from Deborah Anne Bowen, Robin Jessen, Adam Lederer, Holger Lüthen, Christoph Halbmeier, Maximilian Stockhausen, and Cortnie Shupe.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendices
Appendix A: Proofs of convexity and quasiconvexity
The following proofs show the property of convexity or quasiconvexity of a variety of inequality indices. With y_{i} we denote the income of an individual or a household and abstain from considering weighting factors as well as equivalence scales and transfers in order to ease the proofs. However, the proofs are without loss of generality, as one could replace y_{i} with \(\frac {y_{i}+t_{i}}{ES_{i}}\) and scale the sums with weights w_{i}.^{Footnote 1}
A.1 The variance is convex
The variance is defined as,
The functions \(y_{i}\frac {1}{N}{\sum }_{i = 1}^{N} y_{i}\) are affine and, therefore, convex for all i. Squaring these functions and then summing preserves convexity. Therefore, the variance is convex.
A.2 The absolute mean deviation is convex
The absolute mean deviation is defined as,
The functions \(y_{i}\frac {1}{N}{\sum }_{i = 1}^{N} y_{i}\) are composed with the absolute value, which is a norm. Norms are convex and, therefore, the convexity of the expression is preserved.^{Footnote 2} As before with the variance, this implies that the absolute mean deviation is convex.
A.3 The Gini Index is quasiconvex
The Gini index can be written as,^{Footnote 3}
To establish quasiconvexity of G(⋅), we need to establish that G(⋅) is quasiconvex in y_{i}.^{Footnote 4} Next we introduce the Gini’s sublevel sets, \(L_{k}^{}(G)\),
with k denoting the upper bound of the sublevel set. If the elements of \(L_{k}^{}(G)\) are convex for every k, then G(.) is quasiconvex. The sublevel set for an arbitrary k may be denoted by,
This inequality condition holds because of the nonnegativity of the mean and can be shown to describe a convex set in y_{i} by establishing that the lefthand side is a convex function in y_{i} for all k. This is sufficient, since any sublevel set of a convex function is a convex set and here we are studying the sublevel set of the function with levelvalue zero.
Next, we rewrite the lefthand side as a function that has known convexity properties. We note that the lefthand side can be expressed as the pointwise maximum of 2^{N− 1} linear expressions in y_{i} with another linear function in y_{i} subtracted. For example if N = 2 :
The pointwise maximum of linear expressions is convex, so the maximum term is convex.^{Footnote 5} The second term is linear in the y_{i} and, thus, also convex. So the whole lefthand side is convex for any k. Ergo, the Gini index is quasiconvex in the y_{i}.^{Footnote 6}
A.4 The relative mean deviation is quasiconvex
The following proof builds on the convexity of the absolute mean deviation (see proof above). Since \(RMD=\frac {AMD\left (\left \{y_{i}\right \}_{i = 1}^{N}\right )}{\frac {1}{N}{\sum }_{i = 1}^{N} y_{i}}\), we can form the sublevel sets,
The lefthand side contains only convex terms. Hence, the sublevel sets of the RMD are convex. Therefore, the RMD is quasiconvex.^{Footnote 7}
A.5 The Atkinson Index is quasiconvex
The AtkinsonIndex is defined as,
First, consider only the second term of A_{𝜖} and substitute p = 1 − 𝜖. Then,
This function is concave for (p − 1) < 0 or equivalently 𝜖 ≥ 0.^{Footnote 8} To show quasiconvexity, we need to establish that the sublevel sets of A_{𝜖} are convex. This is sufficiently shown by verifying that the negative term of A_{𝜖} has convex sublevel sets, as the rest is just an affine transformation.
The sublevel sets are given by,
Next, we assess if the functions on the lefthand side are convex. If they generate sets that are convex given any k, quasiconvexity is implied. Since the first function is convex – the negative of \(s\left (\left \{y_{i}\right \}_{i = 1}^{N}\right )\) is convex – and the second function is affine, this is the case.
A.6 The Theil Index is quasiconvex
The definition of the Theil Index is,
For quasiconvexity the sublevel sets of the Theil Index need to be convex. Accordingly,
The functions on the lefthand side induce convex sets if they are convex. The second term is affine and, thus, convex. The first term is convex if its Hessian is positive semidefinite. The second partial derivatives of \(f(\{ y_{i}\}^{N}_{i = 1})= {{\sum }_{i}^{N}} y_{i}\text {Log}\left [\frac {N y_{i}}{{{\sum }_{i}^{N}}y_{i}}\right ]\) are,
Then the Hessian of \(f(\{ y_{i}\}^{N}_{i = 1})\) is,
We can reshape the matrix before we test for positive semidefiniteness as
The Hessian is positive semidefinite iff for any vector υ,
To show that this is the case, we rely on the CauchySchwarzInequality. It states that for any two vectors a and b,
State the dotproduct of the Hessian with υ as summations,
To complete the proof, pick \(a^{\prime }=(\sqrt {y_{1}},\sqrt {y_{2}},\ldots )\) and \(b^{\prime }=(\frac {\upsilon _{1}}{\sqrt {y_{1}}},\frac {\upsilon _{2}}{\sqrt {y_{2}}},\ldots )\), which establishes that the above sums are greater or equal to zero.
Since both functions determining the sublevel sets are convex for any k, the Theil is quasiconvex.
Appendix B: Implementation for the Gini index
The optimization problem with the classic formulation of the Gini index is,
Because of the absolute value function in the classic, discrete formulation of the Gini index, it is not differentiable at zero. To derive a differentiable reformulation, we introduce the variables Δ_{ij}, which replace the absolute differences in the objective function, and impose linear constraints that require the Δ_{ij} to be nonnegative: \({\Delta }_{ij} + \left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\leq 0\) and \({\Delta }_{ij}  \left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\leq 0\).^{Footnote 9} To see that this is the case, pick an income difference between any i and j and consider the following scenarios for Δ_{ij}:

1.
Let \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\) be nonnegative. Then Δ_{ij} has to be greater than or equal to \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\) and, thus, will always be greater than or equal to \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\). Accordingly, the Δ_{ij} will be nonnegative.

2.
Let \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\) be negative. Then Δ_{ij} has to be greater than or equal to \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\) and, thus, will always be greater than \(\left (\frac {y_{i}+t_{i}}{ES_{i}}\frac {y_{j}+t_{j}}{ES_{j}}\right )\). So, the Δ_{ij} will again be nonnegative.
Hence, we can convert (37) into an equivalent differentiable optimization problem,
A LinearFractional Problem The objective function in the modified problem (38) has a specific form: an affine function in the numerator and an affine function in the denominator. Problems of this type are called linearfractional (see Boyd and Vandenberghe (2004, p. 151) or originally Charnes and Cooper (1962)). The equivalent problem is,
where we obtain the desired transfer schedule \(t_{i}=\frac {1}{z}\tilde {t}_{i} ~ \forall i\). The major advantage of solving this problem, instead of performing bisection on (38), is the immense saving in computational effort: we need only one run of the interiorpoint algorithm to solve (39) instead of several, as in the case of bisection.
Size of the Problem and Improving Performance Further, there are two possibilities to reduce the number of variables and constraints for a given dataset: First, we can reduce the size of the dataset if two or more households are of the same type – in terms of their equivalence scale – and have the same income. Then we may simply add up their population weights and optimize the collapsed dataset. Second, we may restrict the number of households that may be recipients of a transfer in the optimization by performing the following procedure: 1. Perform a bottom fillup procedure for every equivalencescaletype, where the entire budget at disposal is distributed only among households of this type. 2. Mark those households that are recipients of a positive transfer. 3. Perform the optimization of (39) with free transfer variables for the marked households only.
The justification is that, even in the most extreme case, where just one type of household experiences a bottom fillup, only the marked households can be transfer recipients. Other households of the same type have a weaker effect on the Gini index than the marked households.
Further, to save on memory and reduce computational effort, we provide the following guidelines to enhance the performance of the solver fmincon in Matlab:

1.
The gradient of the objective function, the gradient of the constraints and the Hessian should be generated as sparse matrices to save memory.

2.
The gradient of the objective function and the Hessian are zero everywhere and should be supplied directly by the user.

3.
The gradient of the constraints is constant and can be generated before the execution of interiorpoint algorithm.

4.
Parallel computations should be implemented in order to calculate the gradient of the constraints wherever possible.
Rights and permissions
About this article
Cite this article
König, J., Schröder, C. Inequalityminimization with a given public budget. J Econ Inequal 16, 607–629 (2018). https://doi.org/10.1007/s1088801893803
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s1088801893803
Keywords
 Redistribution
 Public transfers
 Inequality
 Optimization methods
 Interiorpoint algorithm
 Bisection method