Transcript スライド 1 - Hitotsubashi University

Hitotsubashi G-COE Conference Series of Choice,
Games and Welfare:
Equality and Welfare
Securing Basic Well-Being for All
Reiko Gotoh and Naoki Yoshihara
March 16-17, 2012
1
• According to Sen, the concept of “capability” of an individual
is defined as a set of “functionings” (vectors of various
‘doings’ and ‘beings’) which can be realized by transforming
transferable goods and services (not only economic but also
political or social, for example, a knowledge that “everyone
has the right to recognition everywhere as a person before
the law”, Article 6, the Universal Declaration of Human
Rights), through the individual’s non-transferable ability to
utilize them.
• This concept provides a good introduction to understand the
reason of inconveniences an individual faces. For example, if
we know that an individual is too limited in the achievement
of “moving”, we will ask ourselves: does she dislike moving or
is she difficult in using transformations?
2
• And if we wish to promote her achievement of “moving”, we
will consider two kinds of way: making her attitude
(preference, or evaluation of “moving”) change; or expanding
her capability to choose more achievement of “moving”, for
example, through the improvement of public transportations.
• (Note)The following definition of disability is similar to
capability approach: “disability results from the interaction
between persons with impairments and attitudinal and
environmental barriers that hinders their full and effective
participation in society on an equal basis with
others”(preamble, Convention on the Rights of Persons with
Disabilities, 2006).
• Let us illustrate the implication of “capability” concept
contrasting these two ways.
3
accessin
g to IT
y
(x, y) achieved with
evaluation v
(x’ , y’) achieved with
evaluation v’
i’s evaluation , v
i’s capability C
0
45°
moving
x
i’s changed evaluation, v’
Figure: improvement of “moving” by the change of evaluation on functionings from v to v’
4
accessing to
IT
y
(x’ , y’) achieved
under capability C’
(x’, y) achieved
under capability C’
1’s capability C’
i’s evaluation
1’s capability C
0
45°
moving
x
Figure: improvement of the achievement of “moving” through the expansion of capability from C to C’
5
• Here, we especially focus on the concept of “basic capability”
(Sen, 1980, p.367) and investigate the social choice procedure
to secure “basic capability for all”.
• Ｗe define the concept of “basic capability” slightly modifying
the original Sen’s idea in the following way.
• When an individual's capability is below “basic capability”, she
has a legitimate claim to society to provide the goods and
services which are necessary to secure her “basic capability”,
while the freedom to choose the appropriate functionings
from her capability is preserved for her.
• Conversely, we assume that society has an obligation to
secure “basic capability” for all individuals.
6
• But in what way? To say ‘for all’ sounds as much ambiguous as
to say ‘for no one special’.
• To secure basic capability for ‘all’ seriously, it is necessary to
identify ‘the least advantaged’ in terms of capability in each
social policy and choose social policies, under which the
capabilities of the least advantaged are at least as good as (or
not worse than) basic capability.
• In this context, we can refer to “the difference principle”
proposed by John Rawls(1971), which requires maximizing the
benefits of the least advantaged, given economic
circumstances including individuals’ work incentives.
7
• Yet we must note that: Rawls’ difference principle assumes
that the least advantaged can be endogenously or
unanimously defined in a model, for example, as individuals
with the least income in a society in each alternative policy.
• In contrast, when we adopt the concept of capability as index,
it is unrealistic to assume that the least advantaged can be
identified endogenously or unanimously depending on
alternative social policies.
8
• This is first because unlike what is the case of income, the
positions of individuals in terms of capabilities (which are
historically, culturally and personally characterized) cannot be
replaceable through social policies or transferable goods and
services,
• second, we find different types of “the least advantaged” who
are common in lacking “basic capability” but cannot be
compared one another, namely, we cannot say that their
positions in terms of capabilities are ‘indifferent’, needless to
say, one is ‘better’ or ‘worse’ than another.
9
• [Example]: the quantitative achievement of ‘moving’ for a blind person,
for example, ‘10 times a week’ must be discounted by the neglects he
suffers (when he asks neighborhoods) during his “moving”, while the one
for an injured person must be discounted by the pains she suffers (when
she cannot find places for laying down) during her moving. How should we
set up their discounting rates?
• Based on this consideration, we adopt a model in which the
concrete content of “basic capability”---the lists and the sizes
of functionings which constitute “basic capability” as well as
the unit of measurement of each functioning---can be
different according to types of disadvantage.
• In other words, we assume a situation that the conception
(operational form) of ‘basic capability’ can take particular
forms subject to the common acceptance of the universal
concept of ‘basic capability’.
10
• Besides, we introduce the concept of “(disadvantaged)
group”, which consists of persons with the same type of
difficulties in utilization ability (impairments) and shares a
common conception (operational form) of “basic capability”
as well as the unit of measurement.
• It is assumed that a group can form a common appraisal on
capabilities of its members and a common evaluation over
alternative social policies to the extent that is needed to
secure the basic capability for all the members, focusing on
the least advantaged within the group.
11

Accessing to IT
capability of
Individual I
capability of
Individual j
Social policy
No support
IT support
moving support
moving & IT support
moving of i  moving of j
accessing to IT of i  Accessing to IT of j
BC: basic capability
Moving x
figure: Group of blind individuals
the least advantaged i and j have different utilization ability to achieve
“moving” and “accessing to IT” and their capabilities change differently
according to the kinds of supports (moving support &IT support).
12
• The agenda of this paper is to examine the possibility of
constructing a social choice rule (called social relation
function: SRF) that associates a social evaluation of
alternative social policies to each individual or/and group
information, in order to choose the class of social policies
securing ``basic capability for all'', given each circumstance.
• The social relation function we investigate here shares an
essential feature with the traditional Arrovian social welfare
function (Arrow, 1951) or the Sennian social welfare
functional (Sen, 1970, p.126, d'Aspremont and Gevers, 2002)
in that the ultimate informational basis to determine “social
values” is nothing more than “individual values” within a
society. Yet, we diverge from both of them in several aspects.
13
• In Arrovian framework, each individual's characteristic within
a profile is her preference which may represent her
``subjective evaluation''; further full intra-personal
comparability (completeness of ordering) as well as full interpersonal non-comparability are assumed.
• The Sennian social welfare functional permits each
individual's utility function to be interpreted as her
``observable situation'' which is inter-personally comparable,
yet it also assumes full intra-personal comparability
(completeness of ordering) and is usually represented by a
numerical function (individual utility function).
14
• In contrast to the Arrovian and the Sennian frameworks, in
this paper, a capability appraisal with respect to individual i,
which can be interpreted as a ranking of her capabilities, does
not necessarily means her ``subjective appraisal''. Rather, we
interpret that it is an appraisal which is publicly scrutinized
and recognized as reasonable in terms of her observable
capabilities.
• Moreover, an individual appraisal on capabilities is allowed to
be incomplete and ordinal binary relations, as well as interpersonally comparable to the extent that the least
advantaged can be identified within a group but does not
guarantee full comparability among the least advantaged
even in a group. For, example,
15
BC
C(x) of the least advantaged i
C’(x) of the least advantaged j
Figure. Example: capability C (x)and C’ (x)cross each other
and C is included in basic capability while
C’ (x) is cross with BC.

16
These characteristics of our framework derive from the
characteristics of the concept of capability or basic capability
itself: observability (see “positional objectivity”, Sen 2002)
and plurality: the plurality of lists, units and weights of
functionings.
Note:
• Within a group, which shares the same list of functionings and
unit of measurement, comparison of capabilities whether
intra-personally or inter-personally may be possible, for
example, by adopting a specific weight of functionings and a
certain “Atkinson Welfare Index”.
17
Figure: Atkinson Welfare Index: To what point in the line of
full equality the actual distribution is regarded to be equivalent?
y
achieved point with a
certain weight of x and y
utilitarian index
CES function’s index
Rawlsian
index
45°
0
V=[ x

 (1   ) y

]

1

x
, where -1<ρ<∞,
Rawlsian function: ρ→∞
δ:weight coefficient of x and y, ρ:substitution coefficient of x and y
utilitarian function: ρ→-1
Thus, distribution depends on δ, ρ and x/y
• Yet, the basic concern of this paper is to avoid some preexperiential decision on specific methods of comparison,
rather to hand in the decision to the disadvantaged groups.
Here is the true value of social choice theory, more precisely,
the value of replacing the problem of measurement of
capability with the problem of social choice.
• In this paper, the group evaluations are preferentially
reflected in a “social evaluation” of alternative social policies,
while the domain of profiles of group evaluations and the
procedure of aggregating diverse group evaluations are
publicly regulated by certain conditions.
• Now is the time to introduce our basic model in a formal way
and explain the conditions imposed on the social relation
function and on the group evaluations, which are supposed to
embody the goal of “securing basic capability for all”.
19
Consider a society with the population
N ={1,2,…,i,…n}（where
2  n   ）. Let us denote a social alternative by x、and the set of
all possible social states by X（where 3 # X   ）. For each i∈N,
let Zi be the set of conceivable functioning vectors with a generic
element zi . Let Z = iN Zi . For each i, let i's capability
correspondence be a mapping Ci : X  Zi such that for each x∈
Ｘ ,   Ci (x) ⊆ Zi holds. Let C
i
be the set of such
correspondences. Let C= (Ci )iN be a profile of capability
correspondences.
20
Denote the admissible set of profiles of capability correspondences
by C . Then the universal class of capabilities is defined as
follows.
Z  {C⊆Z |for some i∈N, for some Ci ∈C i, for some x∈X, s.t.
Ci ( x)=C}.
Given society N, there exists a set of disadvantage-types T with
generic element t. For each C  C and each t∈Ｔ, there exists a
unique subset NCt . Note that NCt may be empty for some t∈Ｔor
may be identical to N for some t∈Ｔ. An individual can belong to
no group or every group.
21
Given Z, for each i∈N, a capability appraisal of individual i is
defined as a reflective and transitive, binary relation ≿ i .
A group appraisal of t-type ≿ t  it ≿ i , a binary relation reflects
a shared criterion which compares capabilities of members.
Let us denote a capability appraisal ≿  ((≿ i ) iN , (≿ t ) tT ).
The set of the least advantaged of t-type group under each social
policy is defined as individuals, whose capability dominates no
one’s capability in that group.
22
Next, given the society N with C  C and a profile of capability
appraisal ≿ on Z, let us define, for each t∈Ｔ, a reflexive relation
RCt on X. In addition, PCt , ICt NRCt are respectively defined as strict,
indifferent, non-comparable parts of RCt .
Moreover, denote admissible class of type t’s such relations on X
by DCt ((≿ i ) and DC (≿)  tT DCt ((≿ i ).
Based on these notations, we are ready to define a social
relation function, a scheme of social aggregation procedure as
follows.
23
Definition 1: Given a reflective and transitive relation ≿ on Z, a
social relation function (SRF) is a mapping F  which associates
each C  C and each ( RCt )tT  DC (≿) to a reflexive and transitive
relation RC defined on X.
With this scheme, let us discuss the conditions which we devise for
a SRF so as to embody the social goal, “securing basic capability
for all”. First are the conditions relating to the domain of the SRF.
The possible set of profiles of group evaluations is restricted so as
to satisfy BCC, RM and RC, which are defined as follows.
24
Suppose x, (resp. x') is a social policy under which the capabilities
of the least advantaged are at least as good as basic capability; y
(resp. y') is one under which the capabilities of the least
advantaged are worse than basic capability and z (resp. z') is one
under which the capabilities of the least advantaged cannot be
compared (neither worse, better or indifferent) with basic
capability.
(Example)
C(x)
C(x’)
C(z)?, C(z’)?
*red line means BC
C(y)
C(y’)
25
(Example)
C(x)
C(x’)
C(z)?, C(z’)?
*red line means BC
C(y)
C(y’)
Then, the Basic Capability condition (BCC) requires a group to
evaluate that x (x’) is more just than y (y’) and z (z’); and that z
(z’) is more just than y (y’).
The Restricted Monotonicity condition (RM) requires to evaluate y
is at least as good as (resp. more just than; indifferent to) y' if the
capabilities of the least advantaged in y are at least as good as (resp.
better than; indifferent to) those in y'.
The Refrain condition (RC) requires to restrain from making any
pair-wise comparison between x and x' or z and z'.
26
The Basic Capability condition and the Restricted Monotonicity
condition are necessary for a group evaluation to achieve the goal
of “securing the basic capability for all” members in that group.
The Refrain condition is necessary for a group evaluation not to
exceed the goal of securing basic capability. The spirit of the latter
is similar to the ``focus axiom'' proposed by Sen (1981, p.186) in
the context of measuring poverty, which requires the measure to be
independent of information indifferent relative to poverty.
Lemma 1 proves that these three conditions are mutually consistent,
that is, it proves that each group evaluation which satisfies BCC,
RM and RC is reflexive and transitive.
27
Next are the conditions relating to the aggregation procedure
which associates each combination of a capability appraisal and a
profile of group evaluations to a social evaluation. We devise the
following two conditions, NR and CBP defined as follows.
Non-Negative Response（NR）：
For each C  C, each ( RCt )tT  DC (≿), and each x,y∈X, if there
exists t’ ∈Ｔsuch that x RCt ' y (resp. x PCt ' y) and there exists no
t’’ ∈Ｔsuch that y PCt '' x, then x RC y (resp. x PC y)holds, where
RC  F  (C, ( RCt )tT ) .
Note: when the brackets (resp.--) holds, this condition is called
Positive Response (PR).
28
Capability-Weak Pareto（CWP）：
For each C  C, each ( RCt )tT  DC (≿), and each x,y∈X,
if Ci ( x)  Ci ( y) holds for all i∈N, then x PC y holds, where
RC  F  (C, ( RCt )tT ) .
NR requires that a social evaluation should reflect pair-wise
rankings which are accepted by at least one group and not rejected
by any other group. Its normative implication is to give priority to
group evaluations over individual evaluations, while treating each
group fully symmetrical to one another. In contrast, the normative
implication of CWP---which is distinguished from the usual
formulation of the Pareto condition only in that the index of
individual information is not utility but capability appraisal---is to
treat symmetrically all capability appraisals of individuals in
society.
29
[Example]
Suppose there are three groups in a society; t1, t2, t3, and an individual 1 who does
not belong to any group and every group has one least advantaged in order, t1, t2, t3.
Moreover suppose their capability appraisals are ranked from the top, highest to the
1
1
2
t
bottom, lowest. For example, the ranking of t1
3
t
t
is: Ct1(z)  t1BC  t1Ct1(w)  t1Ct1(x)  t1Ct1(y).
C1(x)
By Capability-Wased Pareto, x Py holds. By
C1 (y)
Non-Negative Response, z Px (since z Pt1x by
C1 (w)
C1 (z)
BCC, z Pt2x by BCC and z Pt3x by RM); w Py
Ct2 (w)
Ct1 (z)
BC
t1
Ct2 (z)
BC
t2
Ct1 (w) Ct2 (x)
Ct1 (x)
Ct1 (y)
Ct2 (y)
BC
t3
Ct3 (z)
Ct3 (w)
Ct3 (x)
Ct3 (y)
(since w Pt1y by RM, w Pt2y by BCC, and w
Pt3y by RM); z Pw (since z Pt1w by BCC, z
NRt2w by RC and z Pt3 w by RM); w Px,
(since w Pt1x and w Pt3x by RM, w Pt2x by
BCC, In conclusion, z Pw Px Py holds. Note,
i.e. the ranking of z and w in each group.
30
• Given that persons with a particular disadvantage can be considered as
`specialists` of that disadvantage, the NR condition seems reasonable on
the one hand.
• On the other hand, in terms of respecting the plurality of the conceptions
of the good that individuals have, the CWP condition seems reasonable.
• It is not, however, self-evident that these two conditions are compatible
with each other. The agenda of the rest of the paper is to verify the
compatibility of these two conditions. This is done in the four steps.
• First, we examine whether or not SRF which satisfies NR can without fail
bring about social evaluations which satisfy transitivity.
• Theorem 1 provides a negative answer to this question.
• Therefore, second, we look for a way to avoid this negative result and
introduce an additional condition, Full-comparability of Destitution (FCD).
31
Theorem 2 proves that FCD is sufficient to guarantee the existence of SRF
which satisfies NR, while ensuring the transitivity of social evaluations.
• Based on these results, in the third step, we directly examine the
compatibility of NR and CWP.
• Theorem 3 shows that it is impossible to guarantee the existence of SRF
which satisfies both NR and CWP, even if we strengthen the comparability
of capability appraisals by introducing FCD.
• Finally Theorem 4 clarifies what a kind of condition further strengthening
comparability of capability appraisals is sufficient to avoid this negative
result . Let us introduce these theorems formally.
Theorems
Lemma 1: Let the reflective satisfy BCC, RM, and RC. Then, it is transitive.
32
Theorem 1: There exists a non-comparable ≿ under which there is
no SRF F  satisfying PR.
Full-comparability of Destitution (FCD): For each t∈Ｔ, and for
each social policy x,y∈ Ｔ , if the capabilities of the least
advantaged are all at least as bad as BC and some are worse than
BC in both x and y, then for all the least advantaged in x and all the
least advantaged in y, Ci ( x) ≿ i Ci ( y) or Ci ( y) ≿ i Ci ( x) holds.
Theorem 2: Let FCD hold. Then, there exists a SRF F  which
satisfies NR.
33
• Theorem 3: Suppose FCD. Then, there exists a capability
appraisal ≿, under which there is no SRF which satisfies NR
and CWP.
• Given this impossibility theorem, we introduce an additional
condition, Dominance (D), which requires that if one
capability is clearly worse in comparison to BC, while another
is not (i.e. at least as good as BC), we should make a
comparative judgment between the two, even if we cannot
directly compare them.
34
BC
C(x)
C’(x)
Dominance Condition (D): C’(x)  C i (x)
Figure: capability C (x)and C’ (x)cross each other
and C is included in basic capability while C’
 (x) is cross with BC.
35
• With a condition D in addition to FCD, we can guarantee the existence of
SRF which satisfies NR and CWP as follows:
• Theorem 4: Let FCD and D hold. Then, there exists a SRF which
satisfies NR and CWP.
[implications of theorems]
• Let us summary the implications of these theorems and
confirm the contributions of this paper.
• First, note that if we assume completeness for group
evaluations, PR coincides with the Strong Pareto condition.
36
Positive Response (PR):
For each C  C, each ( RCt )tT  DC (≿), and each x,y∈X, if there exists t’
∈Ｔsuch that x PCt ' y and there exists no
t’’ ∈Ｔsuch that y PCt '' x, then x PC y holds, where RC  F  (C, ( RCt )tT ) .
Positive Response (PR) under Completeness:
For each C  C, each ( RCt )tT  DC (≿), and each x,y∈X, if there
exists t’ ∈Ｔsuch that x PCt ' y and for all t’’ ∈Ｔ such that x RCt '' y,
then x PC y holds, where RC  F  (C, ( RCt )tT ) .
37
• Given that the Strong Pareto condition satisfies transitivity, if we can
assume completeness for a group evaluation, it must satisfy transitivity
under PR. While, theorem 2 shows that FDC which guarantees fullcomparability only in the domain below BC (any pair of y and y’) is
sufficient to avoid cyclicity under NR (stronger version of PR). Why is this
condition sufficient? Isn’t the fully full-comparability necessary? The
answer is no.
•
(example)
C(x)
C(x’)
C(z)?, C(z’)? *red line means BC
•
•
C(y)
C(y’)
• The key to avoid cyclicity in other domains is nothing other than the works
of the Refrain condition (RC) and the Basic Capability condition (BCC). Let
us confirm it.
• First, conceivable cases other than any pair of y and y’ are: any pair of x
and x'; any pair of z and z'; any pair of x and y; any pair of x and z; and any
pair of y and z.
38
• RC prohibits each group to make a comparison between any pair of x and
x' or any pair of z and z', thus NR keeps silent.
• BCC requires full comparison between any pair of x and y; between any
pair of x and z; and between any pair of y and z for every group evaluation.
• In conclusion, cyclicity is avoided in all conceivable cases.
• This result indicates that the conditions devised in this paper are not only
plausible in terms of normative considerations in securing ``basic
capability for all'', but also effective in guaranteeing the transitivity of the
social relation function.
• Second contribution is more philosophical. Recently Sen proposed a
``comparative approach'' to justice in place of a ``transcendental
approach'' to justice (Sen 2009a, pp.15-18, Sen, 2009b, pp.46f).
39
• According to Sen, the latter is a traditional approach in ethics which
focuses on the description of an ideal social state, while the former is in
ethics a new approach, but familiar to economics, which ranks alternative
social states. This paper is an attempt to formulate a ``comparative
approach'' to justice in the following sense.
• First, we can use ``basic capability'' as a criterion to identify whether a
policy is ``just'' or ``unjust''. BCC works by rejecting ``unjust'' policies when
there exists a ``non-unjust'' alternative (non-unjust policies include not
only just policies but also those which are neither just nor unjust policies).
• However, a ``comparative approach'' to justice requires even more. It
requires comparing pairs of ``unjust'' policies to one another and to make
it possible to choose ``less unjust'' policies from the set of feasible
alternatives. In this paper, it is the RM condition with the support of FCD
which does this work.
40
• Instead of summarizing let us clarify the basic conceptions underlying this
paper.
• In this paper, ``(disadvantaged) group'' is defined as the maximum unit
within which inter-personal comparison of capabilities is possible to the
extent that the least advantaged ---individuals whose capabilities never
dominate the capabilities of others--- can be identified in each social
policy.
• Yet, since the relative weights of various functionings cannot be uniquely
determined, there remains a form of incomparability between the least
advantaged within a group, i.e. some of which may be at least as good as
``basic capability'', while some are worse than ``basic capability''.
• In this sense, the process to form a group appraisal is itself a collective
choice which needs to be regulated by certain conditions.
41
• However, the meaning of this form of incomparability should be kept
distinct from that of the incomparability (also called
“incommensurability”) between groups.
• We can treat the former as a technical or political problem and certain
conditions of compromise can be introduced to deal with it as we do in
this paper (the Dominance condition), while the latter is one about which
there should be no compromise as long as we take seriously the plurality
of disadvantages.
• This distinction between two forms of incomparability corresponds to the
distinction introduced by Sen (Sen, 2002, 182) between: ``tentative
incompleteness'' derived from some pairs of alternatives which are not yet
ranked but may all get ranked with more deliberation or information; and
``assertive incompleteness'' derived from some pairs of alternatives which
are asserted to be ``non-rankable''.
42
• The second concerns two different kinds of conflicts which may occur
between different groups.
• One kind of conflicts are ones that arise from each group's need to
achieve ``basic capability''; the other kind of conflicts are ones that derive
from each group's desire to enjoy better capabilities beyond ``basic
capability''.
• The former kind of conflicts is avoidable, if there are resources enough to
secure basic capability for each group, while the latter is not if the desire
towards better capability is without limit. The former deserves
considerations in terms of justice, while the latter does not.
• This is the reason why in this paper the work of Monotonicity condition is
restricted to the domain below ``basic capability'', in the domain above
``basic capability'' conflicts are avoided through the work of the Refrain
condition, which prohibits each group to make any kind of ranking.
43
• The third point concerns to two kinds of comparative adjectives, ``better
or worse'' and ``less or more unjust''.
• In this paper, ``better or worse'' is used in the comparison of individual
capabilities while ``less or more unjust'' is used in the comparison of social
policies.
• This usage is derived from the distinction of conceptions of ``good'' and of
``justice'' according to Rawls (Rawls, 1971).
• In the political liberalism of Rawls conceptions of ``justice'' are expected to
realize an overlapping consensus of plural conceptions of the good.
• We suppose that the conceptions of “capability” and of ``basic capability''
as well as of well-being correspond to conceptions of the good, the
plurality of which should be respected, while the goal of securing ``basic
capability for all'' corresponds to the conception of justice, which aims at
an overlapping consensus.
44
• Final point is the distinction between income and capability. It is logically
possible to reduce the capability index to the income index and the goal of
``basic capability for all'' can be reduced to ``basic income for all''.
• If all necessities are satisfied through the market and depend on individual
income, and if we can take the extra demand of income which particular
groups need in order to achieve the same capability as standard
consumers as their legitimate claim this reduction seem reasonable.
• Yet it is usually difficult to do this, since what kinds of functionings are
basic and how much extra income is needed to secure a person’s basic
capability might be different depending on individual difficulties.
•
Moreover, necessities such as right to freedom, dignity as human beings
and the social basis of self-respect cannot simply be satisfied through the
market with more income.
• What then is needed to secure basic capability for all? Such is the central
question which capability approach pursues.
45
•
•
•
•
•
•
•
•
•
•
•
References
Arrow, K. J. (1951/1963): Social Choice and Individual Values, 2nd ed., New York:
Wiley.
d’Aspremont, C. & L.Gevers (2002) “Social Welfare Functionals and Interpersonal
Comparability,” Arrow. K., A. K. Sen and K. Suzumura (eds.) Handbook of Social
Choice and Welfare, vol.1, Amsterdam: North-Holland, 459-541.
Foster, J (1984): “On Economic Poverty: A Survey of Aggregate Measures,”
Advances in Econometrics, 3, pp. 215-251.
Pattanaik, P. K., and Youngsheng Xu (2007): “Minimal relativism, dominance, and
the standard of living comparison based on functionings,” Oxford Economic Papers,
59, 354-374.
Rawls, J. (1971): A Theory of Justice, Cambridge, Mass.: Harvard University Press.
Sen, K. A.(1970): Collective Choice and Social Welfare, San Francisco: Holden-Day.
Sen, A. K. (1981): Poverty and Famines: An Essay on Entitlement and Deprivation,
Oxford.
Sen, A. K. (1985): Commodities and Capabilities, Amsterdam: North-Holland.
Sen, A. K. (1987): The Standard of Living, Cambridge University Press: Cambridge.
Sen, A. K.(1997): On Economic Inequality, expanded edition with a substantial
annex by James E. Foster and Amartya K Sen, Oxford: Clarendon Press.
46
•
•
•
•
•
•
Sen, A. K. (1999a): Reason before Identity, The Romanes Lecture for 1998, Oxford,
Oxford University Press.
Sen, A. K. (1999b): Development as Freedom, New York: Alfred A. Knopf.
Sen, A. K.(2002): Rationality and Freedom, Cambridge: Harvard University Press.
Sen, A. K. (2006): Identity and Violence-The Illusion of Destiny, New York: W.W.
Norton & Company.
Sen, A. K. (2009a) The Idea of Justice, Allein Lane, Penguin Books.
Sen, A. K. (2009b): “Economics, Ethics and Law,” Gotoh R. and P. Dumouchel (eds.)
Against Injustice: The New Economics of Amartya Sen, Cambridge University Press,
39-54.
47