Theoretical Economics 17 (2022), 1529–1557 1555-7561/20221529

Sufficientarianism
José Carlos R. Alcantud
Multidisciplinary Institute of Enterprise (IME), University of Salamanca

Marco Mariotti
School of Economics and Finance, Queen Mary University of London and Deakin Business School

Roberto Veneziani
School of Economics and Finance, Queen Mary University of London

Sufficientarianism is a prominent approach to distributive justice in political phi-

losophy and in policy analyses. However, it is virtually absent from the formal nor-
mative economics literature. We analyze sufficientarianism axiomatically in the
context of the allocation of 0–1 normalized well-being in society. We present three
characterizations of the core sufficientarian criterion, which counts the number of
agents who attain a “good enough” level of well-being. The main characterization
captures the “hybrid” nature of the criterion, which embodies at the same time
a threshold around which the worst off in society is prioritized, and an indiffer-
ence to equality in other regions. The other two characterizations relate sufficien-
tarianism, respectively, to a liberal principle of noninterference and to a classic
neutrality property.
Keywords. Sufficientarianism, distributive justice.
JEL classification. D63, D70.

1. Introduction
Sufficientarianism is a prominent approach to distributive justice in political philosophy
(Frankfurt (1987, 2000, 2015)). It is “the doctrine advising the ethical observer to ‘maxi-
mize the number of people who have enough’ in any situation” (Roemer (2004, p. 278)).
According to this approach, a concern for equality is philosophically misguided. The
social objective should not be to achieve equality in the relevant space (income, well-
being, opportunities, and so on). Sufficientarianism grants special status to the thresh-
old, which defines what is “enough.” As Casal (2007) has put it, as a theory of distribu-
tive justice, sufficientarianism comprises two separate principles: its “negative thesis”
José Carlos R. Alcantud: [email protected]
Marco Mariotti: [email protected]
Roberto Veneziani: [email protected]
We thank Marc Fleurbaey, Jordi Massó, Hervé Moulin, John Roemer, Philippe van Parijs, and three anony-
mous referees for detailed and insightful comments. We also thank audiences at the Universities of Ed-
inburgh, Catania, Murcia, Glasgow, and Colorado State, and participants in the annual meeting of the
September Group. J. C. R. Alcantud is grateful to the Junta de Castilla y León and the European Regional
Development Fund (Grant CLU-2019-03) for the financial support to the Research Unit of Excellence “Eco-
nomic Management for Sustainability” (GECOS).

© 2022 The Authors. Licensed under the Creative Commons Attribution-NonCommercial License 4.0.
Available at https://econtheory.org. https://doi.org/10.3982/TE4232
1530 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

is that society should not be concerned with distributive issues above the threshold. Its
“positive thesis” is that it is desirable to bring individuals above the threshold.1
From an economic perspective, the relevance of sufficientarianism stems from the
fact that it is applied explicitly or implicitly in policy contexts ranging from public health,
to education, to poverty. The idea that universal access to certain social goods up to a
given level should be guaranteed, but it is less pressing to provide additional amounts
above the threshold, is rather common in political and public discourse, and in a num-
ber of policy arenas. A sufficientarian approach underpins, for example, two-tiered
health care systems such as those of Canada and, to an increasing extent, of the UK:
“Universal access to basic care is part of what one could call the first tier of a health care
system, whereas additional care, provided via a second tier, includes treatments that are
only provided to individuals when they opt in to additional insurance” (Fourie (2016,
p. 194)). An emphasis on “adequacy” characterizes debates on educational policies in
the USA at least since San Antonio Independent School District v. Rodriguez (1973), when
the Supreme Court effectively sanctioned inequalities in educational funding per pupil
by ruling that state-funding formulae for schools based on local taxes were not an un-
constitutional violation of the Equal Protection Clause of the Fourteenth Amendment,
while acknowledging the importance of securing an “adequate” education for students
in all districts (Satz (2007)). In virtually all developed countries, sufficientarianism—
rather than, for example, egalitarianism—is the dominant view defining moral obliga-
tions in the international realm and concerning humanitarian aid (Satz (2010)). Finally,
recent proposals for a universal basic income can also be naturally justified from a suf-
ficientarian perspective (Huseby (2010)).
In spite of its importance and popularity, the theoretical contours of sufficientarian-
ism remain rather vague and undefined. Indeed, one wonders whether the wide appeal
of sufficientarianism among both theorists and practitioners holding rather disparate
views of distributive justice—including both egalitarians and antiegalitarians—may be
partly explained by its embodying seemingly different, if not inconsistent, ethical intu-
itions.
Sufficientarianism is largely unexplored in normative economics and social choice
theory. To the best of our knowledge, it lacks a formal characterization of the type that
can be found for most other major approaches to distributive justice, such as egalitari-
anism and utilitarianism. Benbaji’s claim that “as it stands, it does not have a canonical
interpretation” (Benbaji (2005, p. 310)) remains valid.
In some recent contributions, Bossert, Kamaga, and Cato (2022, 2021) have applied
the axiomatic method to sufficientarian principles. However, their aim is to charac-
terize a class of social welfare orderings within the sufficientarian family and not to
explore the foundations of sufficientarianism as a distributive ethic. Therefore, they
assume the existence of a unique, exogenously given, normatively relevant threshold,
1 “The notion of ‘having enough’ and its ethical significance are by now central to any discussion of the

ethics of distribution” (Benbaji (2006, p. 327)). The literature is too vast for a comprehensive set of ref-
erences. See, for example, Crisp (2003), Temkin (2003), Brown (2005), Casal (2007), Shields (2012), and
Vandamme (2017).
Theoretical Economics 17 (2022) Sufficientarianism 1531

which is part of the analytical framework and enters explicitly the formulation of sev-
eral axioms, which are conceived of as restrictions on sufficientarian orderings. More
generally, their axiomatic framework is rather different from ours. As their analysis
of sufficientarian principles is motivated by an interest in population ethics, they fo-
cus on variable-population properties. Further, they impose fairness and efficiency ax-
ioms that we derive from more basic principles (see, respectively, Propositions 2 and 3
below).
In this paper, we fill this gap by examining the analytical foundations of sufficientar-
ianism. We provide axiomatic characterizations that dissect its ethical building blocks
in a novel way, thus complementing the philosophical analysis, and are a first step in
developing a canonical interpretation of the sufficientarian approach.
The plausibility of the sufficientarian view clearly depends on the appropriate inter-
pretation of the threshold that identifies what is “good enough.” In turn, this raises the
issue of the appropriate variable of normative concern. In his seminal paper, Frankfurt
(1987) focuses on income, but this is disputable (Sen (1985)). Bossert et al. (2022, 2021)
adopt individual utility as the focus of distributive concern. This is much more satisfac-
tory in that it focuses on what ultimately matters to individuals. However, it also raises
complex issues in terms of defining a meaningful, interpersonally comparable utility
threshold. Alternatively, one could focus on opportunities in their “chances of success”
interpretation (Mariotti and Veneziani (2011, 2018)), as in the sufficientarian approaches
proposed by Axelsen and Nielsen (2015) and Nussbaum (1988, 1990). This would lend
an objective nature to the alternatives and establish an absolute scale of measure, but
one may object—following the literature on luck egalitarianism—that lacking any refer-
ence to individual responsibility, chances of success are not an ethically sound way of
measuring opportunities.
We do not enter the debate on the appropriate variable of normative concern and
analyze sufficientarianism in an abstract framework focusing generically on individual
well-being. Given the central role of the threshold in sufficientarian approaches, how-
ever, we assume that well-being can be normalized and measured on a 0–1 scale (e.g., as
in Karni (1998), Dhillon and Mertens (1999), Segal (2000), and Borgers and Choo (2017)
for the case of utilitarianism). We consider criteria that rank profiles of normalized well-
being vectors.
Our main characterization of sufficientarianism isolates four key conceptual con-
stituents:

1. The existence of a distinguished threshold profile around which judgments are

prioritarian-like, in the sense of giving strict precedence to the worst off in society.

2. A principle asserting that raising an individual t to the maximum level of well-being

from a lower level—while leaving the well-being of all other agents unchanged—
is an “absolute individual improvement” and cannot justify a switch in society’s
collective ranking against t.

3. A standard separability principle: the comparison between two profiles uses as in-
put only the well-being of the individuals who stand to gain or lose in moving from
1532 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

one profile to the other; and ignores the precise level of well-being of indifferent
individuals.

4. A standard requirement of efficiency (monotonicity).

We show that these four properties fully characterize sufficientarianism (Theorem 1).
The main interest of this way of characterizing sufficientarianism is, in our opinion,
that it brings to light the tension between two different ethical stances embedded in
the criterion: the focus on lifting the well-being of low well-being individuals around
the threshold profile, captured in (1), which tends toward equalizing well-being; and the
simultaneous tolerance of major inequalities: as we shall elaborate, the principle in (2)
can have extremely inegalitarian consequences.
What is more, we show that these properties imply a key impartiality principle
(anonymity), asserting that the identities of the agents do not count in the criterion.
The other two characterizations emphasize in different ways the distinctive suffi-
cientarian focus on the individual (captured by the idea of absolute improvements in
the main result). Our second characterization (Theorem 2) shows how sufficientarian-
ism satisfies a principle of respect for autonomy with a liberal flavor recently proposed
in the literature (Mariotti and Veneziani (2009, 2011, 2013), Lombardi and Veneziani
(2016), and Alcantud (2013)). Finally, our third characterization (Theorem 4) uses the
classical neutrality axiom of social choice. Evaluating to what extent sufficientarianism
incorporates neutrality-like principles will help us understand the informational con-
tent of sufficientarian rankings, namely which order-preserving transformations they
are invariant to.
An important point to note is that our aim is not to defend sufficientarianism as a
comprehensive approach in political philosophy and normative economics. Rather, it
is to provide a full characterization of the core sufficientarian view so as to clarify its
foundations and implications. The use of our axiomatizations can well be negative: if
some of the properties in the characterizations we offer are considered unacceptable,
then sufficientarianism must be rejected.

2. The framework
2.1 Preliminaries
Let T = {1,    , T } denote a society of T individuals with T > 1. Individual well-being is
measured on a [0, 1] scale, where 0 and 1 denote, respectively, the lowest and highest
levels of well-being attainable by an individual, and for any agent t ∈ T , at ∈ B ≡ [0, 1]
denotes t’s level of well-being or welfare.2 We are interested in a criterion that guides the
allocation of well-being among the T individuals.
A well-being profile (or simply a profile) is a point a = (a1 , a2 ,    , aT ) in the box BT ≡
[0, 1]T . The points (0, 0,    , 0) ∈ BT and (1, 1,    , 1) ∈ BT can be thought of as Hell and
Heaven, respectively, and individual t is in Hell (resp., Heaven) at a ∈ BT if at = 0 (resp.,
2 None of our results depends on a specific interpretation of the distribuendum, for example, income,
wealth, utility, chances of success, which will be generically referred to as well-being.
Theoretical Economics 17 (2022) Sufficientarianism 1533

at = 1). For any a ∈ BT , t ∈ T , and at ∈ [0, 1], we denote (at , a−t ) the profile obtained
from a by replacing at with at , that is, (at , a−t ) = (a1 ,    , at−1 , at , at+1 ,    , aT ).
For all a, b ∈ BT , we write a ≥ b to mean at ≥ bt , for all t ∈ T ; a > b to mean a ≥ b and
a = b; and a  b to mean at > bt , for all t ∈ T .
A permutation π is a bijective mapping of T onto itself. For all a ∈ BT and all π,
aπ ≡ π(a) = (aπ(t ) )t∈T is a permutation of a.
Given a binary relation  on a set X and x, y ∈ X, we write x  y if and only if x  y
and y  x, and x ∼ y if and only if x  y and y  x. We aim to specify desirable properties
for a binary relation  on the box BT , interpreted as a social welfare criterion (or simply
a criterion).3

2.2 The key properties

We discuss here four properties of sufficientarianism. These properties are not to be
taken as compelling desiderata for a criterion, but just as ethical features of sufficien-
tarianism that we consider relevant. We will show that they are also exhaustive, that is,
jointly equivalent to sufficientarianism.
The first two properties are completely standard. One imposes a mild requirement
of efficiency.

Monotonicity. For all a, b ∈ BT , a > b ⇒ a  b.

The other standard property introduces a form of separability in the criterion by

requiring some independence across individuals.

Separability. Let a, b, a , b ∈ BT be such that, for some t ∈ T ,

at = bt and at = bt ,

 
a = at , a−t ,
 
b = bt , b−t .

Then a  b whenever a  b.

The logic underlying this axiom is well known and is common to a host of separa-
bility axioms in social choice and decision theory. The only information the criterion
should use to compare two profiles is the well-being of those individuals who stand to
gain or lose by being at one profile rather than the other. The criterion should ignore the
exact level of well-being of the individuals who are indifferent among the two profiles.
So, if individual t has the same well-being at = bt at profiles a and b and society prefers a
3 Recall that a relation  on a set X is said to be: reflexive if, for any x ∈ X, x  x; complete if, for any

x, y ∈ X, x = y implies x  y or y  x; transitive if, for any x, y, z ∈ X, x  y  z implies x  z. An ordering is

a reflexive, complete, and transitive binary relation. A social welfare criterion is not necessarily an ordering,
but completeness, reflexivity, and transitivity are properties that will be imposed in addition to the axioms
below.
1534 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

to b, it should not change its preference if the levels of well-being of that person change
in both profiles to a common new amount at = bt .
We now introduce two new principles that are more distinctive of sufficientarianism.
The first incorporates an explicit emphasis on levels of well-being.

Absolute individual improvement. Let a, b ∈ BT be such that a  b and at < 1, some

t ∈ T . Then (1, a−t )  (bt , b−t ) for all bt ∈ B.

In words, starting from a situation where a is strictly socially preferred to b, this rank-
ing cannot be reversed if the well-being of an individual t is strictly improved and taken
to its maximum level at profile a, no matter how t’s well-being changes at b, provided
that the well-being of all other agents is unchanged at both profiles. In this sense, taking
t to Heaven is an absolute individual improvement that cannot justify a reversal in the
collective ranking.
Absolute individual improvement captures the sufficientarian idea that absolute lev-
els of well-being matter, and that they matter for individuals—not because of distribu-
tive considerations. As Frankfurt (1997, p. 6) puts it, “what is of genuine moral concern
is not formal but substantive. It is whether people have good lives . . . The evil lies simply
in the unmistakable fact that bad lives are bad.” Thus, without any changes in everyone
else’s well-being, lifting an individual to the highest possible level of well-being, which
surely implies a good life no matter how high the threshold that distinguishes a good life,
cannot make a profile strictly worse than another profile, when it was initially strictly
better. From this perspective, by focusing on changes affecting a single agent, absolute
individual improvement captures the fundamentally individualistic nature of the suffi-
cientarian approach. According to Frankfurt (2000, p. 100), for example, “The rights to
which a person is entitled do not depend upon any comparison with the rights others
possess. The entitlements of each person are based simply upon the relevant character-
istics and circumstances of that person. The governing moral requirement is to avoid
being arbitrary in the attribution benefits and disadvantages to each individual.”
The important feature of the principle is that it is assumed to be valid independently
of the possibly harsh distributional consequences. For example, on nonegalitarian views
it is typically conceivable that a sacrifice may justifiably be asked of some individual t 
for the sake of bringing another individual t to a higher level of well-being. But even non-
egalitarians may balk at imposing the same sacrifice on t  just for the sake of bringing t to
the very maximum level of well-being if t’s gain from this achievement was minuscule.
Consider for instance a utilitarian view: suppose that a1 + a2 > b1 + b2 with a2 < b2 , so
that a is utilitarian-better than b because the gain of a1 − b1 for the first individual is
larger than the loss of b2 − a2 for the second individual. But for any γ ∈ (0, 1) for which
γ > 1−(b2 −a2 ), the profile (1, a2 ) is utilitarian-worse than (γ, b2 ), since the gain of 1−γ
of the first individual does not now compensate for the imposition of the loss b2 − a2 on
the second individual. Yet absolute individual improvement (and sufficientarianism)
does justify this loss no matter how high t’s level of well-being γ is and, therefore, no
matter how little the gain for her by reaching the very top.
The next axiom posits the existence of a distinguished profile around which the cri-
terion is prioritarian.
Theoretical Economics 17 (2022) Sufficientarianism 1535

Prioritarian threshold. There exists β ∈ (0, 1) such that (β, β,    , β)  b for all b ∈
BT for which bt < β, some t ∈ T , and bi = bj for all i, j = t.

Prioritarian threshold says that there exists some critical profile in which all agents
have a common well-being level β that strictly dominates any profile in which one agent
has well-being strictly lower than β. Several observations are in order. First, it is intu-
itive (and consistent with monotonicity) to conceive of the common level of well-being
β as being “sufficiently high,” hence the terminology “threshold,” although this feature
is not formally present in the axiom. Second, the “prioritarian” component of the axiom
consists in the fact that for bi = bj > β, it states that it is (strictly) not worth decreasing
somebody’s well-being below β for the sake of increasing everybody else’s above it (while
for bi = bj ≤ β it simply expresses a mild concern for efficiency). Third, the axiom does
not exclude the existence of multiple thresholds (as, e.g., in Roemer (2004)): for an ex-
treme example, consider that, in addition to monotonicity, the leximin or maximin cri-
teria satisfy prioritarian threshold for a continuum of values β—uniqueness will be im-
plied by the conjunction with absolute individual improvement (Lemma 1). Fourth, by
excluding the case β = 1, and thus preventing the threshold profile from being Heaven,
the axiom does not enforce the rather uncontroversial statement that Heaven is better
than anything else. Finally, the common critical level of well-being β is restricted to be
strictly positive in order to avoid making the property trivial: any social ordering vacu-
ously satisfies the axiom when β = 0 since the condition bt < β = 0 cannot be satisfied
for b ∈ BT .

3. The core sufficientarian view

In this section, we define and characterize the core sufficientarian criterion. Let α ∈
(0, 1) denote an ethically critical threshold identifying a sufficient, or satisfactory level
of well-being. Then, for all a ∈ BT , let P(a, α) = {t ∈ T : at ≥ α} denote the set of in-
dividuals who have (at least) a sufficient level of well-being at profile a, and let n(a, α)
be their number—formally, the cardinality of P(a, α). Then, for all a, b ∈ BT , define the
sufficientarian criterion sα on BT as follows:

a sα b ⇔ n(a, α) ≥ n(b, α).

Our first result proves that the sufficientarian criterion satisfies the main axioms.

Proposition 1. The sufficientarian social welfare relation sα on BT is an ordering, and

it satisfies absolute individual improvement, prioritarian threshold, monotonicity, and
separability.

Proof. It is immediate to see that sα on BT is an ordering and that it satisfies mono-
tonicity, separability and prioritarian threshold by setting β = α (and only for this choice
of β). In fact, it satisfies the stronger condition that n(a, β) > n(b, β) implies a sα b.
To see that sα on BT satisfies absolute individual improvement, consider a, b ∈ BT
such that a sα b and at < 1 for some t ∈ T . By definition, this implies n(a, α) > n(b, α).
Consider a , b ∈ BT such that a = (at , a−t ), b = (bt , b−t ), and at = 1. Since at ≥ bt ,
n(a , α) ≥ n(b , α) and so a sα b , as sought.
1536 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Note that while the sufficientarian criterion is egalitarian at a distinguished level of

well-being β, it is not difficult to show that it violates important egalitarian principles,
such as the Hammond equity axiom or the Pigou–Dalton condition (Hammond equity
says that if a, b ∈ BT are such that ai < bi < bj < aj for some i, j ∈ T , and ak = bk for all
k ∈ T \{i, j}, then b  a. The Pigou–Dalton condition states that if a, b ∈ BT are such that
bi = ai − δ ≥ aj + δ = bj for some i, j ∈ T and some δ > 0, and ak = bk for all k ∈ T \{i, j},
then b  a. To see that the sufficientarian criterion does not incorporate the intuitions
behind these properties, let α = 1/2 and consider two profiles a, b ∈ BT such that a =
( 58 , 0, 1, 1, 1,    , 1) and b = ( 38 , 14 , 1, 1, 1,    , 1). By definition a sα b, which violates both
axioms).
The main result of the paper is the following.

Theorem 1. A social welfare ordering  on BT satisfies absolute individual improve-

ment, prioritarian threshold, monotonicity, and separability if and only if it is the suffi-
cientarian social welfare ordering sα .

Before presenting the proof of Theorem 1, we need a series of preliminary results.

3.1 The ethical threshold, decency, and penury avoidance

We prove here a fundamental lemma that is useful in the proof of a converse to Propo-
sition 1, but is also interesting in its own right. It establishes the existence of a unique
ethical threshold β such that profiles in which the well-being of all agents is at least β
are strictly better than profiles in which some agents have low well-being.

Lemma 1. Let  be an ordering on BT that satisfies absolute individual improvement,

prioritarian threshold, and monotonicity. Then there exists a unique β ∈ (0, 1) such that
for all a, b ∈ BT , at ≥ β, all t ∈ T , and bt < β, some t ∈ T ⇒ a  b.

Proof. We first prove that there exists a β ∈ (0, 1) such that for all a, b ∈ BT , at ≥ β, all
t ∈ T , and bt < β, some t ∈ T ⇒ a  b. Then we prove that such β is unique.
1. Let β ∈ (0, 1) be a parameter value for which prioritarian threshold is satisfied, and
consider any a, b ∈ BT such that at ≥ β, all t ∈ T , and bj < β, some j ∈ T . By monotonic-
ity, a  (β, β,    , β, β). By prioritarian threshold, (β,    , β)  (1,    , 1, bj , 1,    , 1). By
monotonicity, (1,    , 1, bj , 1,    , 1)  b. The desired result then follows by transitivity.
2. In order to prove uniqueness, suppose by contradiction that there are β, β ∈
(0, 1), β > β , such that for all a, b ∈ BT ,

n(a, β) = T > n(b, β) implies a  b, (1)

   
n a, β = T > n b, β implies a  b. (2)

By (2), (β ,    , β )  (β,    , β, 0). By absolute individual improvement, (β ,    , β , 1) 

(β,    , β), a contradiction with (1).
Theoretical Economics 17 (2022) Sufficientarianism 1537

In view of this result, when an ordering on BT satisfies absolute individual improve-

ment, monotonicity, and prioritarian threshold, we will henceforth assume that the
latter axiom holds for the unique parameter value α that satisfies the consequent of
Lemma 1.
In order to better understand the implications of Lemma 1, it is instructive to con-
sider two properties recently proposed by Roemer (2004, pp. 274 and 277), called univer-
sal decency and avoidance of penury. Let β1 , β2 , β3 ∈ (0, 1) represent three prespecified,
ethically relevant well-being thresholds such that β1 ≤ β2 ≤ β3 . If the inequalities are
strict, they can be interpreted, respectively, as the levels of well-being associated with a
life barely worth living, a mediocre life, and a good or excellent life.
Universal decency states that a profile such that all individuals flourish is preferable
to one in which only some of them enjoy a good or excellent life.

Universal decency. For all a, b ∈ BT , at ≥ β3 all t ∈ T , and bt < β3 , some t ∈ T ⇒

a  b.

Avoidance of penury states that a profile such that all individuals have a decent life
is preferable to one in which some of them have a life not worth living.

Avoidance of penury. For all a, b ∈ BT , at ≥ β2 , all t ∈ T , and bt < β1 , some t ∈ T ⇒

a  b.

The property in the consequent of Lemma 1 is somehow reminiscent of universal

decency and avoidance of penury. However, in our case the property follows from the
other, more basic properties. What is more, prioritarian threshold is a merely existen-
tial property and, as noted before, it does not impose any restriction on the number of
thresholds.
Before proceeding with the general characterization, armed with Lemma 1, it is pos-
sible to provide a simple graphical representation of the key steps of the proof of our
main result in the case with T = 2. This may help gauge the relevance of the various
axioms. The uninterested reader may safely skip to the next section for the full proof.

3.1.1 An illustration with T = 2 Consider the following partition of the box B2 :

 
A = a ∈ B2 : a1 ≥ α, a2 ≥ α ,
 
B1 = a ∈ B2 : a1 < α, a2 ≥ α ,
 
B2 = a ∈ B2 : a1 ≥ α, a2 < α ,
 
C = a ∈ B2 : a1 < α, a2 < α .

The following subsets will also be useful in the proof below:

B1F = {a ∈ B1 : a2 = 1}, B1Int = B1 \ B1F .

1538 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Figure 1. The subsets defined in Section 3.1.1.

The set B1F is the intersection between B1 and the frontier of the box, while B1Int can be
loosely interpreted as the “interior” of B1 in the sense that for all a ∈ B1Int , we have at < 1,
t = 1, 2. Similarly,
B2F = {a ∈ B2 : a1 = 1}, B2Int = B2 \ B2F .
Finally, let

AF1 = {a ∈ A : a1 < 1, a2 = 1},

AF2 = {a ∈ A : a1 = 1, a2 < 1},

  
AInt = A \ AF1 ∪ AF2 ∪ (1, 1) .

The partition can be illustrated in the diagram shown in Figure 1.

Assume that the social welfare ordering satisfies absolute individual improvement,
prioritarian threshold, monotonicity, and separability. We know that this implies that
Lemma 1 holds and, therefore, a ∈ A, b ∈ B2 \ A, implies a  b.
First of all, we prove that B1Int ∪ B2Int is contained in one equivalence class. Suppose,
by way of contradiction, that a ∈ B1Int , b ∈ B2Int but a  b. By completeness, either a  b
or b  a. Suppose first that a  b. Then consider a , b ∈ B2 such that a2 = 1, b2 = α,
a1 = a1 , and b1 = b1 . By absolute individual improvement, a  b . However, noting that
a ∈ B1 and b ∈ A, Lemma 1 implies b  a , yielding the desired contradiction. A similar
argument—perturbing agent 1’s well-being instead—rules out b  a.
We conclude that for all a ∈ B1Int and b ∈ B2Int it must be a ∼ b, and transitivity implies
that for all a, b ∈ BiInt , i = 1, 2 it must be a ∼ b.
Then separability and transitivity immediately imply that any pair of profiles in C,
respectively, AInt , B1F , B2F , AF1 , and AF2 , are indifferent.
Finally, by Lemma 1, (α, α)  (1, b2 ) for any b2 < α. By absolute individual improve-
ment, (α, 1)  (1, 1). By monotonicity, (1, 1)  (α, 1) and, therefore, (1, 1) ∼ (α, 1). Sim-
ilarly, (1, 1) ∼ (1, α). By transitivity, this implies that any profiles in AF1 ∪ AF2 ∪ {(1, 1)}
Theoretical Economics 17 (2022) Sufficientarianism 1539

are indifferent. By separability and transitivity, it follows that A is an equivalence

class.
The rest of the ranking then follows from separability and transitivity.

3.2 Impartiality
We now proceed to establish an auxiliary result, which shows that the distributions
where everyone is above the ethical threshold are all equivalent. This will allow us to
derive an important property of impartiality.

Lemma 2. Let  be an ordering on BT that satisfies absolute individual improvement,

prioritarian threshold, monotonicity, and separability. Then a ∼ b for all a, b ∈ BT with
n(a, α) = n(b, α) = T .

Proof. It suffices to prove that a ∼ (α,    , α) for all a ∈ BT with n(a, α) = T because
this yields the desired result by the transitivity of ∼.
Fix a ∈ BT with n(a, α) = T . The fact that a  (α,    , α) follows from mono-
tonicity. In order to prove (α,    , α)  a, note, first of all that Lemma 1 implies
(α,    , α)  (1, α,    , α, α/2). Therefore, as α < 1, by absolute individual improvement,
(α,    , α, 1)  (1, α,    , α, 1) and by separability, (α,    , α)  (1, α,    , α). Similarly,
Lemma 1 implies (α, α,    , α)  (α/2, 1,    , 1). Therefore, by absolute individual im-
provement, (1, α,    , α)  (1,    , 1).
Next, by monotonicity, (1,    , 1)  a. The desired result then follows from transitiv-
ity as (α,    , α)  (1, α,    , α)  (1,    , 1)  a.

Lemma 2 formalizes the “negative thesis” of sufficientarianism, according to which

distributive issues cease to be a concern once all agents are above the threshold: “if ev-
eryone had enough, it would be of no moral consequence whether some had more than
others” (Frankfurt (1987, p. 21)). The normative appeal of Lemma 2 may perhaps derive
from its providing a resolution of the tension between equality and freedom. As Arneson
(2000, p. 55) puts it, the “negative thesis” addresses “the worry about illiberal restriction
of freedom by leaving a wide space of above-threshold matters wherein individual free-
dom is not constrained by social justice.”4
Given the previous results, we can now prove that under the conditions of Lemmas
1–2, the social welfare criterion also satisfies a notion of fairness as impartiality.

Anonymity. For all a, b ∈ BT , a = bπ for some permutation π ⇒ a ∼ b.

Anonymity requires the allocation rule to be insensitive to individual identities.

Proposition 2. Let  be an ordering on BT that satisfies absolute individual improve-

ment, prioritarian threshold, monotonicity, and separability. Then  satisfies anonymity.
4 As noted by an anonymous referee, sufficientarianism may be interpreted precisely as asserting that the

role of government is limited, and instead of maximizing social welfare, it should aim to ensure acceptable
well-being for as many individuals as possible.
1540 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Proof. Let a, b ∈ BT , and let π be a permutation such that a = bπ . Because any per-
mutation is a composition of transpositions (i.e., permutations of two elements), and
given the transitivity of , in order to prove a ∼ b we just need to assume that π is a
transposition. For notational convenience, we consider the case a = (x, y, a3 ,    , aT )
and b = (y, x, a3 ,    , aT ). The other transpositions can be dealt with using similar argu-
ments. Without loss of generality, let x > y.
Case 1: x ≥ α. Two subcases arise.
If y ≥ α, then a = (x, y, a3 ,    , aT ) ∼ (y, x, a3 ,    , aT ) = b if and only if (x, y, 1,    ,
1) ∼ (y, x, 1,    , 1) by separability, and the latter equivalence holds true by Lemma 2.
Assume instead y < α.
Suppose, by way of contradiction, that b  a. Then separability yields (y, x, 1,    ,
1)  (x, y, 1,    , 1). If x < 1, then by absolute individual improvement, (y, 1, 1,    , 1) 
(x, 1, 1,    , 1), against Lemma 1. Suppose x = 1, that is, (y, 1, 1,    , 1)  (1, y, 1,    , 1).
By Lemma 2, (1, 1,    , 1) ∼ (1, (1 + α)/2, 1,    , 1), and separability implies (y, 1,    ,
1) ∼ (y, (1 + α)/2, 1,    , 1). By transitivity, (y, (1 + α)/2, 1,    , 1)  (1, y, 1,    , 1) and
absolute individual improvement yields (y, 1,    , 1)  (1,    , 1), against Lemma 1.
A similar argument rules out a  b.
Case 2: x < α.
Let us first prove (0, z, a3 ,    , aT ) ∼ (0, 0, a3 ,    , aT ) when z < α. By contradiction,
suppose (0, z, a3 ,    , aT )  (0, 0, a3 ,    , aT ), noting that  is monotonic. By separa-
bility, (α, z, a3 ,    , aT )  (α, 0, a3 ,    , aT ). We use Case 1 and transitivity to deduce
(α, 0, a3 ,    , aT ) ∼ (0, α, a3 ,    , aT ) and then (α, z, a3 ,    , aT )  (0, α, a3 ,    , aT ). By
absolute individual improvement, (1, z, a3 ,    , aT )  (α, α, a3 ,    , aT ). From separabil-
ity, we get (1, z, 1,    , 1)  (α, α, 1,    , 1), in contradiction with Lemma 1.
Similarly, one can prove (z, 0, a3 ,    , aT ) ∼ (0, 0, a3 ,    , aT ) when z < α.
Hence, by transitivity, and noting that α > x > y, (0, y, a3 ,    , aT ) ∼ (0, x, a3 ,    , aT )
and (x, 0, a3 ,    , aT ) ∼ (y, 0, a3 ,    , aT ). By separability, the former equivalence implies
(x, y, a3 ,    , aT ) ∼ (x, x, a3 ,    , aT ), while the latter equivalence implies (x, x, a3 ,    ,
aT ) ∼ (y, x, a3 ,    , aT ). The result then follows from the transitivity of ∼.

Anonymity is often considered to be a fundamental criterion of fairness in normative

judgements as it rules out the possibility that agents’ identities influence social evalua-
tions. By Proposition 2, we do not need to impose it separately: it follows from our key
axioms.
In order to grasp the intuition of the proof, the reader can look at the graphical illus-
tration of the case T = 2, showing that B1Int ∪ B2Int is contained in an equivalence class.
In fact, the logic of the proof is essentially the same, as B1Int , B2Int are on opposite sides of
the 45-degree line. The key steps of the argument in Proposition 2 are Lemma 1, show-
ing that a unique threshold exists; and Lemma 2, showing that there is “indifference
at the top.” Then our “perturbational” axioms (notably separability and absolute indi-
vidual improvement) force the indifference at the top to carry over to profiles that are
permutations of one another and have at most two elements below the threshold.
Theoretical Economics 17 (2022) Sufficientarianism 1541

3.3 Completing the characterization

We are now ready to establish the main result.

Proof of Theorem 1. Proposition 1 proves necessity.

To prove sufficiency, let  be an ordering on BT that satisfies the axioms. By
Lemma 1, there is a unique α ∈ (0, 1) such that for all a, b ∈ BT , at ≥ α, all t ∈ T , and
bt < α, some t ∈ T implies a  b. We must show that for each a, b ∈ BT , a  b if and only
if n(a, α) ≥ n(b, α).
We show that for all natural numbers h, such that T ≥ h ≥ 0, and for all a, b ∈ BT ,

n(a, α) = T − h > n(b, α) implies a  b, (3)

and
n(a, α) = T − h = n(b, α) implies a ∼ b. (4)
We proceed by induction on h.
(h = 0) Lemma 2 proves (4), while (3) follows from Lemma 1.
(Inductive step) Suppose that (3) and (4) are true for all 0 ≤ h ≤ k − 1 < T , and con-
sider h = k. We prove first that (4) must hold.
Suppose, by way of contradiction, that there exist a, b ∈ BT such that n(a, α) = T −
k = n(b, α) but a  b. By completeness, suppose that a  b without loss of generality.
Suppose T − k > 0. Then, noting that T − k < T , there are t, t  ∈ T such that at ≥
α and bt  < α. By Proposition 2 and transitivity, we assume that t = t  without loss of
generality.
If at < 1, then absolute individual improvement implies a = (1, a−t )  b = (α, b−t ).
This contradicts the induction hypothesis for (3), noting that n(b , α) = n(b, α) + 1 >
n(a, α) = n(a , α).
Therefore, suppose that for all k ∈ T , ak ≥ α implies ak = 1. By Lemma 2 and separa-
bility, it follows that a = (1, a−t ) ∼ (α, a−t ) = a and, therefore, a  b by transitivity. Then
absolute individual improvement implies a = (1, a−t )  b = (α, b−t ). This contradicts
the induction hypothesis for (3), noting that n(b , α) = n(b, α) + 1 > n(a, α).
Suppose T − k = 0.
If there are t, t  ∈ T such that at = bt  , then consider a permutation bπ of b such
  
that bπ t = bt  . By Proposition 2 and transitivity, a  b . Let a , b ∈ B be such that at ≥ α,
π T

bt ≥ α, at = bt , and aj = aj , bj = bπ  

j , for all j = t. By separability, a  b , which contradicts
the induction hypothesis for (4). Therefore, suppose that there are no t, t  ∈ T such that
at = bt  . By monotonicity, a  b implies that there is at least some t ∈ T such that at > bt .
Then consider a ∈ BT such that at > at = bt and aj = aj for all j = t. Noting that n(a, α) =
n(a , α) = n(b, α) = 0, by the previous argument it must be both a ∼ a and a ∼ b, which
yields the desired contradiction by transitivity.
Next, we prove that (3) must also hold.
Suppose, by way of contradiction, that there exist a, b ∈ BT such that n(a, α) = T −
k > n(b, α) but a  b. By completeness, this implies that b  a. Let us fix t ∈ T such
that at < α, bt < α (this is without loss of generality by Proposition 2 and transitivity, and
noting that T − k < T ).
1542 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Suppose b  a. Then by absolute individual improvement, b = (1, b−t )  a =

(α, a−t ), which violates the induction hypothesis for (3).
Suppose b ∼ a. Consider a ∈ BT such that at = bt and aj = aj for all j = t. Because
n(a, α) = T − k = n(a , α), by the previous argument, it follows that a ∼ a , and so by
transitivity b ∼ a . Separability then implies b = (α, b−t ) ∼ a = (α, a−t ), however, not-
ing that n(a , α) = T − (k − 1) > n(b , α) = n(b, α) + 1, a contradiction ensues from the
induction hypothesis for (3).

Remark. Theorem 1 continues to hold even if the restrictions bt < at and bt = 1 are im-
posed in absolute individual improvement.5 This is a logical weakening, which leads to
greater independence between the axioms. However, we have opted for the current for-
mulation of the result because the weaker version of the axiom seems to be less easily in-
terpretable. Furthermore, the current axiom viewed as a necessary condition highlights
more directly a strong implication of sufficientarianism.

The proof that the properties in Theorem 1 are independent can be found in Ap-
pendix B.1.

4. Alternative characterizations
In this section, we extend our axiomatic approach to dissect sufficientarianism from two
alternative perspectives. Specifically, holding fixed separability and prioritarian thresh-
old, we provide characterizations based on key axioms that are different from absolute
individual improvement. The first characterization relates sufficientarianism to some
elements of liberal thought, while the second harks back to the classical social choice
discussion about the informational content of welfare comparisons. In the concluding
part of the section, we also offer some remarks on the property of continuity.

4.1 Liberal roots of sufficientarianism

Next, we consider a liberal principle of noninterference recently proposed by Mariotti
and Veneziani (2013).

Noninterference. Let a, b, a , b ∈ BT be such that, for some t ∈ T ,

  
at − at bt − bt > 0,
   
a = at , a−t , and b = bt , b−t .

Then b  a whenever at > bt and a  b.

Observe that the first inequality condition in the axiom says that in moving from the
unprimed to the primed pair of profiles, individual t either strictly loses at both pro-
files or strictly gains at both profiles. Then, noninterference says the following: suppose
5 We are grateful to an anonymous referee for pointing this out to us.
Theoretical Economics 17 (2022) Sufficientarianism 1543

that society strictly prefers a to b and that an individual’s well-being changes, for the
better or for the worse, while all other agents are unaffected; then society should not
reverse its strict preferences in a way that is adverse to the individual whose well-being
has changed.
This principle captures the liberal idea of the existence of a sphere of individual
autonomy when others are unaffected by a change in someone’s circumstances. In
particular, society should not use as ethical arguments the reasons behind changes in
well-being that concern exclusively one individual, for example, it should not punish
further an individual for misfortunes that, because of his negligence, caused harm to
him, and him alone (while society remains free to compensate this individual). As ar-
gued in Mariotti and Veneziani (2020), the roots of noninterference—and its normative
foundations—can be traced back to John Stuart Mill’s Harm Principle.
Noninterference may be deemed objectionable “as it requires ignoring all informa-
tion concerning the size of the changes in welfare, and their potentially relevant impli-
cations for total utility or for the welfare of the worst off” (Mariotti and Veneziani (2020,
p. 572)). Similarly, the restriction to changes in well-being that affect only one individ-
ual may be considered arbitrary. The key point to note here is that we are not proposing
noninterference as an independently desirable property. Rather, our interest in a lib-
eral axiom falls within our general aim to unpack the normative building blocks of the
sufficientarian approach.
From this perspective, one important characteristic of noninterference is its focus
on changes in the well-being of a single agent. “The individualistic and nonaggrega-
tive nature of noninterference (focusing on changes in the situation of a single agent
while keeping everyone else indifferent) aims to capture widely shared liberal views”
(Mariotti and Veneziani (2020, p. 572)). As already noted, the individualistic and nonag-
gregative outlook of classical liberalism is also a key conceptual feature of sufficientarian
approaches.
Theorem 2 shows that nonnterference can replace absolute individual improvement
and monotonicity in the main characterization.

Theorem 2. A social welfare ordering  on BT satisfies noninterference, prioritarian

threshold, and separability if and only if it is the sufficientarian social welfare order-
ing sα .

Proof. See Appendix A.1.

The proof that the properties in Theorem 2 are independent can be found in Ap-
pendix B.2.
Theorem 2 highlights an aspect of sufficientarianism that is usually ignored in philo-
sophical debates.6 What is more, it sheds further light on the relation between sufficien-
tarianism and standard notions of efficiency. To see this, consider the following standard
properties.
6 A partial exception is Crisp (2003) who provides a justification for a sufficientarian view of justice based

on a specific interpretation of Mill’s concept of the “sentiment of justice.”

1544 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Weak Pareto. For all a, b ∈ BT , a  b ⇒ a  b.

Strong Pareto. For all a, b ∈ BT , a > b ⇒ a  b.

Efficiency as weak or strong Pareto is often considered to be a desirable property of

social orderings, but it is obvious that the core sufficientarian view is not concerned with
full efficiency. For example, its “negative thesis” part asserts precisely the irrelevance of
benefits above the critical threshold α.
Consider next the uncontroversial requirement that the social ranking of alternatives
does not coincide with the ordering of any individual.

Nondictatorship. For all t ∈ T , there are a, b ∈ BT such that at > bt and b  a.

Mariotti and Veneziani (2013) have proved the following result (adapted to the
present context):

Theorem 3. There exists no ordering on BT that satisfies noninterference, weak Pareto,

and nondictatorship.

It follows that, since it satisfies noninterference and is not dictatorial, not only
is sα not weakly Pareto optimal, but especially it cannot be extended—by breaking
indifferences— to any relation that satisfies weak Pareto (or, a fortiori, strong Pareto)
while preserving noninterference and nondictatorship.

4.2 Neutrality
In social choice theory, a standard way of capturing consistency in social evaluations is
by means of “single-profile” neutrality conditions (see Rubinstein (1984)).

Neutrality. Let a, b, a , b ∈ BT , a  b and sign(at − bt ) = sign(at − bt ) for all t ∈ T ⇒

a  b .

Neutrality is often taken to incorporate an emphasis on ordinal judgements with

no weight assigned to interpersonal comparisons (for a comprehensive discussion, see
Fleurbaey and Mongin (2005)). However, because sufficientarianism does allow certain
kinds of welfare and level comparisons while being invariant to some ordinal transfor-
mation, neutrality needs to be suitably modified.
Although neutrality seemingly incorporates some reasonable properties, in its full
force, it has two major shortcomings. First, it is a rather strong requirement, which al-
most immediately leads to impossibility results (Fleurbaey and Mongin (2005, pp. 386–
387)). Second, and related, it has rather unappealing distributive implications. As
Samuelson famously argued, neutrality
“says, ‘If it is ethically better to take something (say 1 chocolate or, alternatively, say 50
chocolates) from Person 1 who had all the chocolates in order to give to Person 2 who had
none, then it must be ethically preferable to give all the chocolates to Person 2’. One need
Theoretical Economics 17 (2022) Sufficientarianism 1545

not be a doctrinaire egalitarian to be speechless at this requirement. Is it ‘reasonable’ to

put on an ethical system such a straightjacket? Few will agree that it is.” (Samuelson (1977,
p. 83))

As Fleurbaey and Mongin (2005, p. 405) noted, “This ingenious parable is virtually all
that is needed to deprive [neutrality] from [its] normative appeal as far as distributive ap-
plications are concerned . . . Incidentally, the parable also illustrates Samuelson’s math-
ematical point that neutrality implies dictatorship ‘transparently’.”
Both problems arise from the fact that neutrality considers perturbations that af-
fect the well-being of more than one agent. As Samuelson’s example shows, in this case
imposing a consistency requirement on order preserving perturbations may lead to eth-
ically unattractive implications. However, the property is immune from this line of criti-
cism when the allowable perturbations are restricted to those that concern only a single
individual.
The following axiom captures the individualistic nature of sufficientarianism while
avoiding Samuelson’s critique.

Individual neutrality. Let a, b, a , b ∈ BT be such that for some t ∈ T , a−t = a−t ,

b−t = b−t and at > bt . Then a  b whenever a  b and at ≥ bt .

An interesting—and perhaps surprising—implication of individual neutrality is that

if it is assumed, together with the separability and prioritarian threshold axioms, then it
is unnecessary to impose monotonicity as a separate requirement.

Proposition 3. Let  be an ordering on BT that satisfies individual neutrality, priori-

tarian threshold, and separability. Then  satisfies monotonicity.

Proof. We first prove the following particular instance: for each t ∈ T and b ∈ BT , at >
bt with at ∈ B implies (at , b−t )  b.
Fix a β ∈ (0, 1) that verifies prioritarian threshold. For notational convenience, we
proceed when t = 1, the other cases being identical.
By prioritarian threshold, (β,    , β)  (0, β,    , β). By individual neutrality,
(a1 , β,    , β)  (b1 , β,    , β). Now a sequential application of separability implies
(a1 , b−1 )  b.
Once this property has been established then by transitivity, a routine application to
the successive components proves that for all a, b ∈ BT , a > b implies a  b because

a  (b1 , a2 ,    , aT )  · · ·  (b1 , b2 ,    , bT −1 , aT )  (b1 , b2 ,    , bT ) = b.

Given Proposition 3, the next result proves that individual neutrality can replace ab-
solute individual improvement and monotonicity in the main characterization.

Theorem 4. A social welfare ordering  on BT satisfies individual neutrality, prioritar-

ian threshold, and separability if and only if it is the sufficientarian social welfare order-
ing sα .
1546 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Proof. Necessity is trivial.

In order to establish sufficiency, we shall prove that if an ordering  on BT satisfies
individual neutrality, prioritarian threshold, and separability, then it satisfies absolute
individual improvement. To see this, let a, b ∈ BT be such that a  b and at < 1, some
t∈T.
If at > bt , individual neutrality implies (1, a−t )  (bt , b−t ) for all bt ∈ B.
If at = bt , separability implies (1, a−t )  (1, b−t ), and then by Proposition 3 and tran-
sitivity, (1, a−t )  (bt , b−t ) for all bt ∈ B.
If bt > at , then (bt , a−t )  a  b  (0, b−t ) by Proposition 3 and transitivity. By indi-
vidual neutrality, (1, a−t )  (bt , b−t ) for all bt ∈ B.
This proves that, under the axioms in the statement, the ordering  on BT satis-
fies absolute individual improvement. Given Proposition 3, the result then follows from
Theorem 1.

The proof that the properties in Theorem 4 are independent can be found in Ap-
pendix B.3.

4.3 Continuity

Our results also highlight another key aspect of sufficientarianism, namely the discon-
tinuity in ethical judgements at the sufficiency threshold. This emphasis may be dis-
putable, and it may be objected that, intuitively, there is no threshold, which marks a
discontinuous, qualitative change in people’s lives. Two points can be made in response
to this objection.
First, it is not obvious that continuity is a desirable property. Although it is of-
ten presented as an innocuous technical condition on social preferences, it rules out
some widely used and normatively appealing approaches, such as the lexical version of
John Rawls’s maximin principle or Amartya Sen’s poverty index. (For a discussion of,
and counterpoint on, the relevance of continuity properties see, for example, Baigent
(2011).) Second, discontinuity may actually express an ethically relevant property: as
Shields (2012, p. 108) nicely puts it, “The sufficiency threshold . . . seems to mark a shift
in the nature of our reasons to benefit people further. This intuitive thought can be for-
mally expressed by what I term the shift thesis.” The ethical discontinuity at the thresh-
old is indeed a defining feature of sufficientarianism in all of its variants. Benbaji (2006,
p. 332) calls it the “sufficientarian discontinuity,” according to which “benefiting a per-
son just below the threshold is much more important than benefiting those who are just
above it.” From this perspective, the analysis in Section 4.1 can be extended to show
precisely the nature of the discontinuity, and loosely speaking, the exact amount of con-
tinuity over the entire box BT that is compatible with sufficientarianism.7

7 In their characterization of a class of sufficientarian orderings, Bossert et al. (2022, 2021) incorporate
the ethical discontinuity inherent in the sufficientarian approach by imposing two axioms that require con-
tinuity to hold above and below the threshold separately, but not over the entire domain.
Theoretical Economics 17 (2022) Sufficientarianism 1547

To be precise, as shown in Appendix A.2, if prioritarian threshold is weakened to

include β = 1, then the three properties in Theorem 2 are not sufficient to obtain the
result. An additional axiom is necessary, which incorporates the usual view of a strict
preference relation: if two profiles are strictly ranked, then the strict ranking is preserved
for profiles that are sufficiently close to the inferior one:

Upper semicontinuity. For all a ∈ BT , the set {b ∈ BT : a  b} is open in BT in the

Euclidean topology.

5. Concluding remarks
The philosophical analysis of sufficientarianism presents some puzzlingly disparate in-
terpretations. For example, its originator Frankfurt explicitly saw sufficientarianism as
an alternative to egalitarianism. Others (e.g., Anderson (1999), Nussbaum (1988, 1990),
Satz (2007)) have interpreted it instead as a special form of egalitarianism. The root
cause for such discrepancies is that sufficientarianism incorporates just some aspects
of several standard principles, without incorporating any of them in its “pure” form. In
this paper, we have offered formal characterizations that should help clarify this hybrid
aspect of sufficientarianism. The main characterization (Theorem 1) in particular em-
phasizes, on the one hand, its limited prioritarian features, and on the other hand the
possible justification it can provide to harsh inequalities.
We conclude with some comments on possible future directions to extend our work.
Two characteristics of the sufficientarian ordering, among others, have attracted
criticism. First, it justifies major losses in well-being for a large number of destitute
individuals, for the sake of a small increase in the well-being of one agent, provided this
allows her to cross the sufficiency threshold (Roemer (2004)). While sufficientarianism
was proposed by Frankfurt precisely in opposition to the idea that equality should al-
ways be pursued, this conclusion may seem unpalatable nevertheless, because of the
extremeness of its antiegalitarianism. Second, the core sufficientarian view yields very
large indifference classes, making it rather insensitive to both efficiency and distributive
concerns. Taken on their own, neither of these objections seems to seriously undermine
sufficientarianism. One may argue, for example, that the sufficientarian view can be in-
terpreted as capturing some widespread intuitions similar to triage: it is not so implau-
sible to opt for a profile, which allows at least one agent to live a decent life, rather than
having everyone lead a life hardly worth living. Similarly, once everyone thrives, most
people would agree that distributive concerns are much less pressing. The problem
is that the core sufficientarian view cannot respond to both criticisms together (Casal
(2007)). A low threshold allows one to respond to the first objection but it makes the
second objection more salient. A high threshold has the opposite effect.
One possible answer explored in the literature to the large indifference classes prob-
lem is to consider sufficientarianism as part of a more complete distributive theory
(Crisp (2003), Benbaji (2005), Brown (2005), Shields (2012), Vandamme (2017)).8 In
8 For a discussion, see Casal (2007). Bossert et al.’s (2022, 2021) critical level sufficientarianism can also
be interpreted as a hybrid approach.
1548 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

our framework, this approach could be formalized by means of a refined sufficientarian

principle that adds a secondary criterion to the core one. That is, a profile a is judged
better than a profile b if the number of people above a normatively relevant threshold is
greater at a than at b, or if an equal number of people are above the threshold at a and
b, but a is better than b according to a secondary criterion (e.g., utilitarian).
Formally, let P denote an ordering on BT . Given a threshold level of well-being α ∈
(0, 1), define a refined sufficientarian criterion s,P
α on B as follows. For all a, b ∈ B ,
T T

a s,P
α b ⇔ either n(a, α) > n(b, α), or n(a, α) = n(b, α) and a P b.

(where recall that n(a, α) is the number of individuals who have a sufficient level of well-
being at profile a). Observe that s,P
α is transitive and complete by the transitivity and
completeness of sα and P . It is also known that

a s,P
α b ⇔ either n(a, α) > n(b, α), or n(a, α) = n(b, α) and a P b.

A different strategy to respond to the criticisms moved against the core sufficientar-
ian view has been to clearly separate the “positive thesis” and the “negative thesis” by
specifying multiple different ethical thresholds (Benbaji (2006), Huseby (2010); see also
Roemer’s (2004) formalization). One can think of the higher threshold as identifying a
level of well-being above which agents flourish, while the lower threshold can be set at a
level below which an agent will be in a miserable condition. Then, if the former thresh-
old is indeed sufficiently high, it is not implausible to argue that once all agents are above
such level distributive concerns are less pressing. Similarly, if the latter threshold is suf-
ficiently low, then one may argue that it is a matter of moral urgency to push as many
agents as possible above such a minimum threshold. Formally, let α, α ∈ B denote two
(ethically determined) distinct thresholds with 1 > α > α > 0, identifying, respectively,
a sufficient, or satisfactory well-being and a minimum acceptable well-being level (e.g.,
the minimum level guaranteeing leading a life worth living). A natural extension of the
core sufficientarian view, the multithreshold sufficientarian criterion, sα,α , is as follows.
For all a, b ∈ BT ,
       
a sα,α b ⇔ either n a, α > n b, α or n a, α = n b, α and n(a, α) > n(b, α),
   
a ∼sα,α b ⇔ n a, α = n b, α and n(a, α) = n(b, α).

An analysis of the approaches just sketched seems a promising route to deepen our
understanding of sufficientarian ideas.
On a different front, in our treatment, we have not fully exploited the structure of the
box of well-being. Given the presence of the extreme profiles Hell and Heaven, “duality”
properties analogous to those used in the theory of rationing (e.g., Moulin (2000)) could
be defined. This could uncover different ethical aspects of sufficientarianism.
Finally, another interesting development would be to extend the study of sufficien-
tarianism to the context of intergenerational justice (analogously to what has been done
Theoretical Economics 17 (2022) Sufficientarianism 1549

for other distributive criteria; for a review, see Asheim (2010)), setting the objective of
leaving each generation with a sufficiently high standard of living.

Appendix A: Proof of results in Section 4

A.1 Proof of Theorem 2
The basic structure of the proof of Theorem 2 is similar to that of our main characteriza-
tion. First, we prove that any social welfare ordering satisfying noninterference, priori-
tarian threshold, and separability satisfies monotonicity.

Proposition 4. Let  be an ordering on BT that satisfies noninterference, prioritarian

threshold, and separability. Then  satisfies monotonicity.

Proof. We first prove the following particular instance: for each t ∈ T and b ∈ BT , at >
bt with at ∈ B implies (b1 ,    , bt−1 , at , bt+1 ,    , bT )  b. Let β be an index that satisfies
prioritarian threshold.
We distinguish four cases. For notational convenience, we proceed when t = 1, the
other cases being identical.
Case 1. If a1 = β, then (a1 , β,    , β)  (b1 , β,    , β) by prioritarian threshold. A se-
quential application of separability yields (a1 , b2 ,    , bT )  (b1 , b2 ,    , bT ) = b.
Case 2. Suppose either β ≤ b1 or 0 < b1 < β < a1 . Then (β,    , β)  (0, β,    , β) by
prioritarian threshold. A sequential application of separability yields (β, b2 ,    , bT ) 
(0, b2 ,    , bT ). Noninterference ensures (a1 , b2 ,    , bT )  (b1 , b2 ,    , bT ) = b.
Case 3. If 0 = b1 < β < a1 , then (β,    , β)  (b1 , β,    , β) by prioritarian thresh-
old. A sequential application of separability yields (β, b2 ,    , bT )  (b1 , b2 ,    , bT ) = b.
Noninterference yields (a1 , b2 ,    , bT )  (β, b2 ,    , bT ) and the desired result follows
from transitivity.
Case 4. If b1 < a1 < β, then consider ε > 0 such that β − ε > b1 . By prioritarian
threshold, (β,    , β)  (β − ε,    , β), and a sequential application of separability yields
(β, b2 ,    , bT )  (β − ε, b2 ,    , bT ). The desired result then follows from noninterfer-
ence.
Once this property has been established, by transitivity a routine application to the
successive components proves that a, b ∈ BT , a > b implies a  b because

a  (b1 , a2 ,    , aT )  · · ·  (b1 , b2 ,    , bT −1 , aT )  (b1 , b2 ,    , bT ) = b.

Given Proposition 4, we prove the existence of a unique ethical threshold β such that
profiles in which the well-being of all agents is at least β are strictly better than profiles
in which some agents have low well-being.

Lemma 3. Let  be an ordering on BT that satisfies noninterference, prioritarian thresh-

old, and separability. Then there is a unique β ∈ (0, 1) such that for all a, b ∈ BT , at ≥ β,
all t ∈ T , and bt < β, some t ∈ T ⇒ a  b.
1550 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Proof. Given Proposition 4, the demonstration is a straightforward modification of the

proof of Lemma 1.

The next result generalizes the consequent of noninterference to any two profiles in
which an agent enjoys the same level of well-being.

Lemma 4. Let  be an ordering on BT that satisfies noninterference, prioritarian thresh-

old, and separability. Then for any a, b ∈ BT , if a  b, then a  b for any a , b ∈ BT such
that at = bt some t ∈ T and a = (at , a−t ), b = (bt , b−t ).

Proof. Let a, b ∈ BT be such that a  b. Consider any t ∈ T . If at = bt , then the result

follows from separability. Suppose at = bt . We argue with t = 1, the other possibilities
being symmetrical.
1. First of all, we show that for all a ∈ BT :
 
if a1 ≥ α, then a1 , a−1 ∼ a for all a1 ∈ (a1 , 1), (5)
 
if a1 < α, then a1 , a−1 ∼ a for all a1 ∈ (a1 , α). (6)

To see that (5) holds, fix an arbitrary a1 such that a1 < a1 < 1. We claim that
(a1 , 1,    , 1) ∼ (a1 , 1,    , 1). By Proposition 4, (a1 , 1,    , 1)  (a1 , 1,    , 1). Suppose,
by way of contradiction, that (a1 , 1,    , 1)  (a1 , 1,    , 1).
Observe that by Lemma 3 we have that (α,    , α)  (1, α,    , α, α/2), and non-
interference yields (α,    , α, (1 + α)/2)  (1, α,    , α). Further, by Lemma 3 ((1 +
α)/2, α,    , α)  (α/2, 1,    , 1), and then because 1 > a1 > α > α/2, noninterference
yields (1, α,    , α)  (a1 , 1,    , 1). By transitivity, it follows that (α,    , α, (1 + α)/2) 
(a1 , 1,    , 1).
But then, transitivity implies (α,    , α, (1 + α)/2)  (a1 , 1,    , 1), in contradiction
with Proposition 4. Therefore, we conclude that (a1 , 1,    , 1) ∼ (a1 , 1,    , 1) and the
desired result follows from separability.
To see that (6) holds, fix an arbitrary a1 such that a1 < a1 < α. Proposition 4 im-
plies (a1 , a−1 )  a. Suppose, by way of contradiction, that (a1 , a−1 )  a. Separabil-
ity yields (a1 , (1 + α)/2,    , (1 + α)/2)  (a1 , (1 + α)/2    , (1 + α)/2). Observe that
(α,    , α, α/2)  (a1 , (1 + α)/2,    , (1 + α)/2). [For if (a1 , (1 + α)/2,    , (1 + α)/2) 
(α,    , α, α/2), then by noninterference, (a1 , (1 + α)/2,    , 1)  (α,    , α, α) in contra-
diction with Lemma 3.] By transitivity, (α,    , α, α/2)  (a1 , (1 + α)/2    , (1 + α)/2. By
noninterference, (1, α,    , α, α/2)  (α, (1 + α)/2    , (1 + α)/2), in contradiction with
Lemma 3.
2. Armed with (5) and (6), we can now prove the result. We distinguish two cases.
Case 1: a1 < 1.
If b1 > a1 , then b  (a1 , b−1 ) by Proposition 4. Therefore, transitivity implies a 
(a1 b−1 ) and the desired result follows from separability.
,
Suppose b1 < a1 . By (5) and (6), there exists ã1 ∈ (a1 , 1) such that a ∼ ( ã1 , a−1 ). Then
we deduce ( ã1 , a−1 )  (a1 , b−1 ) by an application of noninterference to a  b. Transitiv-
ity implies a  (a1 , b−1 ) and separability yields the desired conclusion.
Theoretical Economics 17 (2022) Sufficientarianism 1551

Case 2: a1 = 1.
We proceed in two steps. First, we prove that a ∼ ((1 + α)/2, a−1 ). By Proposition 4,
in order to establish the claim it is sufficient to prove that ((1 + α)/2, a−1 )  a.
Let c = ((1 + α)/2, (1 + α)/2,    , (1 + α)/2). The claim ((1 + α)/2, a−1 )  (1, a−1 ) = a
is equivalent to ((1 + α)/2, c−1 )  (1, c−1 ) by separability. Fix an arbitrary x ∈ ((1 +
α)/2, 1). By Lemma 3, (x , α, (1 + α)/2,    , (1 + α)/2)  (1, 0, (1 + α)/2,    , (1 + α)/2).
By noninterference, (x , x , (1 + α)/2,    , (1 + α)/2)  (1, c−1 ). The claim then fol-
lows from transitivity noting that because x ∈ ((1 + α)/2, 1), (5) implies c ∼ (x , c−1 )
and (x , c−1 ) ∼ (x , x , (1 + α)/2,    , (1 + α)/2) and the transitivity of ∼ ensures c ∼
(x , x , (1 + α)/2,    , (1 + α)/2).
The previous argument proves that a ∼ ((1 + α)/2, a−1 ). Then Proposition 4 and
transitivity imply ( ã1 , a−1 )  a for all 1 > ã1 > max(b1 , (1 + α)/2). Then, by transitivity,
it follows that ( ã1 , a−1 )  b for all 1 > ã1 > max(b1 , (1 + α)/2). Then by noninterfer-
ence, a  ( ã1 , b−1 ). Transitivity ensures ( ã1 , a−1 )  ( ã1 , b−1 ) and the desired conclusion
follows from separability.

Next, we show that the profiles where everyone is above the ethical threshold are all
equivalent.

Lemma 5. Let  be an ordering on BT that satisfies noninterference, prioritarian thresh-

old, and separability. Then a ∼ b for all a, b ∈ BT with n(a, α) = n(b, α) = T .

Proof. Given Proposition 4 and Lemmas 3–4, the demonstration is a straightforward

modification of the proof of Lemma 2.

Finally, we prove that under the conditions of Lemmas 3–5, the social welfare order-
ing satisfies anonymity.

Proposition 5. Let  be an ordering on BT that satisfies noninterference, prioritarian

threshold, and separability. Then  is anonymous.

Proof. Given Proposition 4 and Lemmas 3–5, the demonstration is a straightforward

modification of the proof of Proposition 2.

We now proceed to prove Theorem 2.

Proof of Theorem 2. (Necessity) To see that sα on BT satisfies noninterference con-

sider a, b ∈ BT such that a sα b. By definition, this implies n(a, α) > n(b, α). Then con-
sider a , b ∈ BT such that for some t ∈ T , (at − at )(bt − bt ) > 0, and aj = aj , and bj = bj
for all j = t. If at > bt , then it immediately follows that n(a , α) ≥ n(b , α) and so b sα a ,
as sought.
(Sufficiency) Given Propositions 4–5, and Lemmas 3–5, the rest of the proof is a
straightforward modification of the proof of Theorem 1.
1552 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

A.2 Upper semicontinuity

In this section, we provide a formal proof of the claims made in Section 4.3.
First of all, we prove that sα on BT satisfies upper semicontinuity. To see this, take
any a ∈ BT and consider the set L(a) = {b ∈ BT : a sα b}. Consider any b ∈ L(a). By
definition n(a, α) > n(b, α). Let = mint∈T \P(b,α) (α − bt ): is well-defined and strictly
positive. For any a, a ∈ BT , let d(a, a ) be the Euclidean distance between a and a . Then
for any b ∈ BT such that d(b, b ) < , we observe that when t ∈ T \ P(b, α), α − bt = α −
bt + bt − bt ≥ + bt − bt ≥ − d(b, b ) > 0. Therefore, we have n(a, α) > n(b, α) ≥ n(b , α),
as sought.
Next, let prioritarian threshold* denote the weak version of the axiom holding for β ∈
(0, 1]. Example 1 shows that separability, noninterference, and prioritarian threshold*
do not jointly characterize the sufficientarian ordering.

Example 1. Let ∗ be the ordering on B2 defined as follows: a ∗ b if and only if a ∈ Ai ,

b ∈ Aj , and i ≤ j, where A1 = {(1, 1)}, A2 = {(x, 1) : x ∈ (0, 1)} ∪ {(1, y ) : y ∈ [0, 1)}, A3 =
{(0, 1)} ∪ {(x, y ) : x ∈ (0, 1), y ∈ [0, 1)}, and A4 = {(0, y ) : y ∈ [0, 1)}.
The ordering ∗ satisfies separability, noninterference, and prioritarian threshold*
(with respect to β = 1). It does not satisfy upper semicontinuity because ( n1 , 0) ∗ ( 12 , 0),
the sequence {( n1 , 0)}n converges to (0, 0), but ( 12 , 0) ∗ (0, 0). ♦

It is possible to prove, however, that upper semicontinuity is precisely the missing

condition. To see this, note first that Proposition 4 continues to hold (only cases 1 and 4
in the proof are relevant). Then we prove the equivalent of Lemma 3.

Lemma 6. Let  be an ordering on BT that satisfies separability, noninterference, priori-

tarian threshold*, and upper semicontinuity. Then there is a unique β ∈ (0, 1] such that
for all a, b ∈ BT , at ≥ β, all t ∈ T , and bt < β, some t ∈ T ⇒ a  b.

Proof. The proof of existence is identical to Lemma 3.

In order to prove uniqueness, we proceed ad absurdum. Assume that there are
β, β ∈ B, β > β > 0, such that for all a, b ∈ BT ,

n(a, β) = T > n(b, β) implies a  b, (7)

   
n a, β = T > n b, β implies a  b. (8)

By (7), (β,    , β)  (β, β ,    , β ). Separability assures (β , β,    , β)  (β ,    , β ). Let

us prove (β ,    , β )  (β , β,    , β), which ensures the contradiction.
Using (8), (β ,    , β )  (0, β,    , β). By noninterference for each ε > 0 such that
β + ε < 1, it must be the case that (β + ε, β ,    , β )  (β , β,    , β). Upper semiconti-


nuity then yields the desired (β ,    , β )  (β , β,    , β).

The next result generalizes the consequent of noninterference to any two profiles in
which an agent enjoys the same level of well-being.
Theoretical Economics 17 (2022) Sufficientarianism 1553

Lemma 7. Let  be an ordering on BT that satisfies separability, noninterference, and

upper semicontinuity. Then for any a, b ∈ BT , if a  b, then a  b for any a , b ∈ BT such
that at = bt some t ∈ T and a = (at , a−t ), b = (bt , b−t ).

Proof. Consider any a, b ∈ BT such that a  b. Consider any t ∈ T . If at = bt , then the

result follows immediately from separability. Therefore, suppose at = bt . We need to
consider three cases.
Case 1. Suppose at > 0 and bt > 0. Consider any aε , b ∈ BT such that at > aεt , bt >
bt and aεt = bt + ε, some ε > 0, and aεj = aj , bj = bj for all j = t. By noninterference,


it follows that aε  b for any such ε > 0. Therefore, upper semicontinuity yields a =
(bt , a−t )  b . The desired result then follows from separability.
Case 2. Suppose either bt = 0 and 1 > at , or at = 0 and 1 > bt . Then a similar ar-
gument as in Case 1 can be applied to any aε , b ∈ BT such that at < aεt , bt < bt and
aεt = bt + ε, some ε > 0, and aεj = aj , bj = bj for all j = t.
Case 3. Suppose either bt = 0 and at = 1, or at = 0 and bt = 1. In the former
case, by upper semicontinuity, there exists ε ∈ (0, 1), such that a  b∗ = (b1 ,    , bt−1 , ε,
bt+1 ,    , bT ). Then the argument in Case 1 can be applied to a, b∗ . In the latter case
by upper semicontinuity, there exists ε ∈ (0, 1), such that a  b∗ = (b1 ,    , bt−1 , 1 −
ε, bt+1 ,    , bT ). Then the argument in Case 2 can be applied to a, b∗ .

Given Proposition 4 and Lemmas 6–7, the demonstration that the sufficientarian
criterion is the only ordering that satisfies separability, noninterference, prioritarian
threshold*, and upper semicontinuity is a straightforward modification of the proof of
Theorem 2.

Appendix B: Independence of the axioms

B.1 Independence of axioms used in Theorem 1
The following examples prove that the properties in Theorem 1 are independent.

Example 2. Let I be the trivial ordering given by a I b for each a, b ∈ BT .

The ordering I satisfies monotonicity, absolute individual improvement, and sep-
arability but not prioritarian threshold (with respect to any β). ♦

Example 3. Let α ∈ (0, 1) and 1α be the ordering on BT defined by: for each a, b ∈ BT ,
a 1α b if and only if it is false that n(b, α) = T > n(a, α).
The ordering 1α satisfies monotonicity, absolute individual improvement, and pri-
oritarian threshold (with respect to α), but it violates separability: when T = 2, (1, 1) 1α
(1, 0) but it is false that (0, 1) 1α (0, 0). ♦

Example 4. Let 1 > α > β > 0 and α,β be the ordering on BT defined by: for each
a, b ∈ BT , a α,β b if and only if either n(a, α) > n(b, α) or (n(a, α) = n(b, α) and n(a, β) ≥
n(b, β)).
The ordering α,β satisfies monotonicity, separability, and prioritarian threshold
(with respect to α), but it violates absolute individual improvement: when T = 2, α = 0.5,
β = 0.3 one has (0.6, 0) α,β (0.4, 0.4) but it is false that (1, 0) α,β (0.6, 0.4). ♦
1554 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Example 5. Let α ∈ (0, 1) and 2α be the ordering on B2 defined as follows: a 2α b if and
only if a ∈ Ai , b ∈ Aj , and i ≤ j, where A1 = {(1, 1), (α, α), (α, 1), (1, α)}, A2 = {(x, 1) :
x ∈ [0, α) ∪ (α, 1)} ∪ {(1, y ) : y ∈ [0, α) ∪ (α, 1)} ∪ {(x, α) : x ∈ [0, α) ∪ (α, 1)} ∪ {(α, y ) : y ∈
[0, α) ∪ (α, 1)}, A3 = B2 \ {A1 ∪ A2 }.
The ordering 2α satisfies absolute individual improvement, separability and priori-
tarian threshold (with respect to α), but it violates monotonicity. ♦

B.2 Independence of axioms used in Theorem 2

The following examples prove that the properties in Theorem 2 are independent.

Example 6. The ordering I defined in Example 2 satisfies noninterference, upper

semicontinuity, and separability but not prioritarian threshold (with respect to any β).
♦

Example 7. The ordering 1α defined in Example 3 satisfies noninterference, up-

per semicontinuity, and prioritarian threshold (with respect to α), but it violates
separability. ♦

Example 8. The ordering α,β defined in Example 4 satisfies separability, upper semi-
continuity, and prioritarian threshold (with respect to α), but it violates noninterfer-
ence: when T = 2, α = 0.5, β = 0.3 one has (0.6, 0) α,β (0.4, 0.4) but it is false that
(0.7, 0) α,β (0.6, 0.4). ♦

B.3 Independence of axioms used in Theorem 4

The following examples prove that the properties in Theorem 4 are independent.

Example 9. The ordering I defined in Example 2 satisfies individual neutrality, and

separability but not prioritarian threshold (with respect to any β). ♦

Example 10. The ordering 1α defined in Example 3 satisfies individual neutrality and
prioritarian threshold (with respect to α), but it violates separability. ♦

Example 11. The ordering α,β defined in Example 4 satisfies separability and priori-
tarian threshold (with respect to α), but it violates individual neutrality: when T = 2, α =
0.5, β = 0.3 one has (0.6, 0) α,β (0.4, 0.4) but it is false that (0.7, 0) α,β (0.6, 0.4). ♦

References
Alcantud, José Carlos R. (2013), “Liberal approaches to ranking infinite utility streams:
When can we avoid interference?” Social Choice and Welfare, 41, 381–396. [1532]
Anderson, Elizabeth S. (1999), “What is the point of equality?” Ethics, 109, 287–337.
[1547]
Arneson, Richard J. (2000), “Perfectionism and politics.” Ethics, 111, 37–63. [1539]
Theoretical Economics 17 (2022) Sufficientarianism 1555

Asheim, Geir B. (2010), “Intergenerational equity.” Annual Review of Economics, 2, 197–

222. [1549]
Axelsen, David V. and Lasse Nielsen (2015), “Sufficiency as freedom from duress.” The
Journal of Political Philosophy, 23, 406–426. [1531]
Baigent, Nick (2011), “Topological theories of social choice.” In Handbook of Social
Choice and Welfare—Vol. II (Kenneth J. Arrow, Amartya K. Sen, and Kotaro Suzumura,
eds.), 301–334, North Holland, Amsterdam. [1546]
Benbaji, Yitzhak (2005), “The doctrine of sufficiency: A defence.” Utilitas, 17, 310–332.
[1530, 1547]
Benbaji, Yitzhak (2006), “Sufficiency or priority?” European Journal of Philosophy, 14,
327–348. [1530, 1546, 1548]
Borgers, Tilman and Yan-Min Choo (2017), “Revealed relative utilitarianism.” CESifo
Working Paper Series No. 6613, Available at SSRN, https://ssrn.com/abstract=3035741.
[1531]
Bossert, Walter, Kohei Kamaga, and Susumu Cato (2021), “Critical level sufficientarian-
ism.” The Journal of Political Philosophy, forthcoming. [1530, 1531, 1546, 1547]
Bossert, Walter, Kohei Kamaga, and Susumu Cato (2022), “Thresholds, critical levels,
and generalized sufficientarian principles.” Economic Theory, forthcoming. [1530, 1531,
1546, 1547]
Brown, Campbell (2005), “Priority or sufficiency . . . or both?” Economics and Philoso-
phy, 21, 199–220. [1530, 1547]
Casal, Paula (2007), “Why sufficiency is not enough.” Ethics, 117, 296–326. [1529, 1530,
1547]
Crisp, Roger (2003), “Equality, priority, and compassion.” Ethics, 113, 745–763. [1530,
1543, 1547]
Dhillon, Amrita and Jean-François Mertens (1999), “Relative utilitarianism.” Economet-
rica, 67, 471–498. [1531]
Fleurbaey, Marc and Philippe Mongin (2005), “The news of the death of welfare eco-
nomics is greatly exaggerated.” Social Choice and Welfare, 25, 381–418. [1544, 1545]
Fourie, Carina (2016), “Sufficiency of capabilities, social equality, and two-tiered health
care systems.” In What Is Enough? Sufficiency, Justice, and Health (Carina Fourie and
Annette Rid, eds.), 185–204, Oxford University Press, Oxford. [1530]
Frankfurt, Harry G. (1987), “Equality as a moral ideal.” Ethics, 98, 21–43. [1529, 1531,
1539]
Frankfurt, Harry G. (1997), “Equality and respect.” Social Research, 64, 3–15. [1534]
Frankfurt, Harry G. (2000), “The moral irrelevance of equality.” Public Affairs Quarterly,
14, 87–103. [1529, 1534]
1556 Alcantud, Mariotti, and Veneziani Theoretical Economics 17 (2022)

Frankfurt, Harry G. (2015), On Inequality. Princeton University Press, Princeton. [1529]

Huseby, Robert (2010), “Sufficiency: Restated and defended.” The Journal of Political
Philosophy, 18, 178–197. [1530, 1548]
Karni, Edi (1998), “Impartiality: Definition and representation.” Econometrica, 66, 1405–
1415. [1531]
Lombardi, Michele and Roberto Veneziani (2016), “Liberal egalitarianism and the harm
principle.” Economic Journal, 126, 2173–2196. [1532]
Mariotti, Marco and Roberto Veneziani (2009), “Non-interference implies equality.” So-
cial Choice and Welfare, 32, 123–128. [1532]
Mariotti, Marco and Roberto Veneziani (2011), “Allocating chances of success in finite
and infinite societies: The utilitarian criterion.” Journal of Mathematical Economics, 48,
226–236. [1531, 1532]
Mariotti, Marco and Roberto Veneziani (2013), “On the impossibility of complete non-
interference in Paretian social judgements.” Journal of Economic Theory, 148, 1689–
1699. [1532, 1542, 1544]
Mariotti, Marco and Roberto Veneziani (2018), “Opportunities as chances: Maximising
the probability that everybody succeeds.” Economic Journal, 128, 1609–1633. [1531]
Mariotti, Marco and Roberto Veneziani (2020), “The liberal ethics of non-interference.”
British Journal of Political Science, 50, 567–584. [1543]
Moulin, Hervé (2000), “Priority rules and other asymmetric rationing methods.” Econo-
metrica, 68, 643–684. [1548]
Nussbaum, Martha (1988), “Nature, function, and capability: Aristotle on political dis-
tribution.” Oxford Studies in Political Thought, suppl. 1, 145–184. [1531, 1547]
Nussbaum, Martha (1990), “Aristotelian social democracy.” In Liberalism and the Good
(R. Bruce Douglas, Gerald M. Mara, and Henry S. Richardson, eds.), 203–252, Routledge,
New York. [1531, 1547]
Roemer, John E. (2004), “Eclectic distributional ethics.” Politics, Philosophy and Eco-
nomics, 3, 267–281. [1529, 1535, 1537, 1547, 1548]
Rubinstein, Ariel (1984), “The single profile analogues to multi profile theorems: Math-
ematical logic’s approach.” International Economic Review, 25, 719–730. [1544]
Samuelson, Paul A. (1977), “Reaffirming the existence of ‘reasonable’ Bergson–
Samuelson social welfare functions.” Economica, 44, 81–88. [1545]
Satz, Debra (2007), “Equality, adequacy, and education for citizenship.” Ethics, 117, 623–
648. [1530, 1547]
Satz, Debra (2010), “Ideals of egalitarianism and sufficiency global justice.” Canadian
Journal of Philosophy, Supp. 36, 53–71. [1530]
Theoretical Economics 17 (2022) Sufficientarianism 1557

Segal, Uzi (2000), “Let’s agree that all dictatorships are equally bad.” Journal of Political
Economy, 108, 569–589. [1531]
Sen, Amartya K. (1985), Commodities and Capabilities. North Holland, Amsterdam.
[1531]
Shields, Liam (2012), “The prospects for sufficientarianism.” Utilitas, 24, 101–117. [1530,
1546, 1547]
Temkin, Larry S. (2003), “Egalitarianism defended.” Ethics, 113, 764–782. [1530]
Vandamme, Pierre-Étienne (2017), “Why not more equality? Sufficientarianism and in-
equalities above the threshold.” Law, Ethics and Philosophy, 5, 130–141. [1530, 1547]

Co-editor Ran Spiegler handled this manuscript.

Manuscript received 31 March, 2020; final version accepted 18 October, 2021; available online 26
October, 2021.