RESEARCH DOSSIER
Justice and Social Evaluation in Economic Theory
Abstract. Justice in economic theory concerns the terms on which persons share the benefits, burdens, opportunities, risks, and authority generated by economic cooperation. Social evaluation compares policies, institutions, or attainable social states through explicitly specified information and evaluative judgments. This monograph examines the relationship between economic efficiency and distributive justice, the informational requirements of interpersonal welfare aggregation, resource equality, priority to the disadvantaged, capabilities, rights, procedural fairness, and institutional feasibility. A comparative reading of foundational economic and philosophical literature is combined with independently constructed formal counterexamples, sensitivity calculations, and bounded interpretation of empirical evidence on neighborhood exposure, health insurance, and multidimensional welfare measurement. The analysis establishes why identical outcomes can receive different justified rankings under different evaluative premises, why arithmetic aggregation requires a defensible scale, and why the institutional production of opportunities matters to the assessment of distributions. It examines the implications for transfer policy, public investment, social insurance, taxation, and common-resource governance. The principal analytical contribution is an explicit separation of empirical identification, informational selection, ethical judgment, and institutional implementation, coupled with a demonstration of the consequences of varying these elements in otherwise identical policy comparisons. The study is a critical synthesis of established scholarship rather than an assertion of a novel theorem or a new named theory.
Keywords: Distributive justice; social evaluation; welfare economics; interpersonal comparisons; Pareto efficiency; social welfare functions; equality of resources; capabilities; rights; priority; social choice; public economics; inequality measurement; institutional design
Introduction
Economic justice is a property of the arrangements through which persons acquire claims on economic resources, participate in cooperation, and encounter the consequences of social decisions. It encompasses income and wealth, but its object also includes security, time, access to institutions, bargaining power, and the actual freedom to choose among valuable ways of living. An economy distributes more than the commodities appearing in market transactions: it distributes opportunities to produce, opportunities to refuse, obligations to bear costs, and possibilities of avoiding harms. Each distribution is partly constituted by rules of acquisition, exchange, inheritance, employment, taxation, public provision, and collective decision.
Social evaluation is the reasoned comparison of social alternatives. An alternative might be an allocation of consumption goods, a tax schedule, a health program, a system of property rights, or a sequence of institutional changes. A well-defined comparison requires a population, a time horizon, a feasible set, a description of the consequences for affected persons, and a criterion that establishes which differences count as improvements. Some comparisons remain incomplete because available information or justified value judgments do not settle every pair. Incompleteness can be a scientifically intelligible result.
The subject is central to Economic Theory and Economic Methodology because positive analysis describes mechanisms and estimates consequences while evaluative reasoning establishes the significance of those consequences. Suppose a subsidy raises aggregate output and also raises the cost of housing for renters. Whether it improves the economy in a socially relevant sense depends on which households gain, which lose, the duration of both changes, and what rights and opportunities are implicated. An econometric identification strategy can help establish the magnitudes; the strategy cannot supply the weights or rights with which the magnitudes are judged.
The inquiry uses historical and conceptual interpretation, formal analysis of explicitly constructed examples, and selective examination of research evidence. All numerical allocations introduced as examples are hypothetical and employ arbitrary units unless specified otherwise. They clarify dependence on assumptions rather than estimate any national distribution. The empirical studies are published historical investigations selected for the way their design illuminates competing evaluative criteria. No new survey, microdata analysis, causal estimator, or comprehensive systematic literature review is claimed. Access to some classic works is restricted to publisher records, excerpts, or corroborating scholarship; conclusions that require a full-text interpretation beyond that access are avoided.
The object and history of economic justice
From material distribution to the terms of cooperation
Classical political economy directed attention to the division of the social product among classes, the conditions of productive activity, and the institutional origins of property and exchange. Later marginal analysis supplied techniques for describing individual preferences, substitution, and decentralized allocation. The development of welfare economics made it possible to ask whether individually chosen exchanges produce socially desirable outcomes, while social choice theory made the aggregation of individual assessments a formal object. The resulting history includes successive changes in what counts as evidence of social improvement: material surplus, satisfaction, feasible exchange, individual rankings, interpersonal welfare, and wider conceptions of advantage.
Abram Bergson’s (published as Abram Burk, 1938) reformulation of welfare economics is a foundational bibliographic landmark in the explicit treatment of social welfare judgments. A social welfare function expresses the desirability of social states under an adopted evaluative specification. Its mathematical existence under particular representation assumptions does not establish the legitimacy of its ethical content. The economic problem becomes one of articulating the relationship between feasible production and allocation on one side and the criterion for ranking outcomes on the other.
The subsequent development of collective choice, interpersonal comparability, and fairness brought into view four independent questions. First, can a distribution be altered so that at least one person gains and no one loses? Second, if gains and losses coexist, how are they compared? Third, what claims arise independently of the aggregate pattern of gains? Fourth, by what institutions can an acceptable allocation be reached? The history of the field is intelligible as an expansion of these questions. The narrower and broader conceptions coexist because they solve different analytical tasks.
Units of justice
An individual is often the elementary unit of moral concern. Economic statistics, however, are frequently collected for households, firms, geographical areas, and income classes. An analyst who uses household income as the measure of an individual's advantage implicitly adopts a theory of intra-household sharing and needs. Household members may possess different control over income, different medical expenses, and unequal access to time. Averages over these units can hide individual deprivation. A group may also be a relevant unit for detecting discriminatory institutions, even if the underlying ethical claim concerns each of its members.
The appropriate unit depends on the question. Rights to political participation attach to individuals; access to many public goods is territorial; some economic decisions are made jointly by households; and risks can be pooled across generations. Justice includes the relationship between these levels. A national average cannot alone establish fair treatment of persons, and a person-level index cannot alone explain the institutional mechanisms that generate their opportunities.
The information underlying social comparisons
Let X denote the set of socially feasible alternatives and N a finite population of persons. Each state x in X has a consequence description for every person: consumption, resources, risks, capabilities, rights, and other relevant characteristics. A social ordering is a binary relation on X. Completeness and transitivity give a conventional ordering, but a partial ordering can preserve defensible comparisons where others remain unsettled. A real-valued representation, when one exists, is a convenient description of the relation rather than an independent justification of it.
The identification of relevant information comes before its aggregation. A consumption measure, a subjective well-being measure, a measure of health, and an index of effective opportunities each answer a different question. The same distribution of income can be associated with different transport access, morbidity, housing conditions, and opportunities for education. A policy ranking based on disposable income consequently answers an income-specific evaluative question even when the social issue is wider.
Consider two social states with identical resources for each of two persons. In state A both can access a public school, whereas in state B the second person faces an inaccessible building. A resource statistic that counts only private income assigns identical positions. A capability-sensitive evaluation recognizes a difference. A rights-sensitive evaluation may also distinguish them if the inaccessibility violates a claim to equal access. The different rankings arise from different state descriptions rather than a computational error.
Marc Fleurbaey and François Maniquet's fairness and social welfare research (2011; online edition 2013) places fair resource allocation and social choice within a common analytical inquiry. This connection makes explicit the importance of the information retained in social ordering. In practical assessment, it matters whether resource access, individual characteristics, preferences, and opportunity sets are permitted to influence a comparison. Restricting the information class can make a formal exercise tractable and simultaneously narrow its substantive domain.
A distinction between observed outcomes and feasible options is essential. If two persons both travel to work by bus, their observed choices coincide. One might have several affordable alternatives and voluntarily choose the bus, while the other lacks any alternative. Revealed behavior alone identifies the selected option under actual constraints, and additional information is needed to describe freedom. Economic justice frequently concerns those constraints because the rules that determine feasible sets precede the observed choices within them.
Efficiency, compensation, and distributive judgment
Pareto comparison and the distribution of gains
A state x Pareto dominates y if everyone is at least as well off in x as in y and at least one person is strictly better off, according to the preferences or welfare comparisons specified for the exercise. Pareto efficiency is the absence of a feasible dominating state. The criterion is powerful when unanimous improvements exist, and indeterminate over many competing distributions.
Consider a feasible set consisting of allocations of 100 identical units between two persons whose well-being strictly increases with their own consumption. Every allocation that exhausts the 100 units is efficient, including (99, 1) and (50, 50). A transfer from the first person to the second reduces the former's consumption and increases the latter's. The allocation (50, 50) may be favored by an equality criterion, but Pareto efficiency alone ranks neither distribution above the other. Efficiency in this model identifies the boundary of the feasible set; justice supplies an assessment of positions on that boundary.
The first welfare theorem associates competitive equilibrium with Pareto efficiency subject to its conditions, including price-taking behavior, well-defined goods and local nonsatiation in the standard private-goods model. The second theorem associates Pareto-efficient allocations with price support under further assumptions, notably convexity and suitable conditions for decentralization. Neither statement guarantees that initial endowments are fair. The distinction between theorem and institutional implementation becomes important when lump-sum transfers cannot be implemented, markets are incomplete, or private choices create external effects. These facts alter the attainable distribution and the meaning of the efficiency comparison.
Potential and actual compensation
Nicholas Kaldor's (1939) welfare propositions belong to the historical development of compensation tests. A monetary test may find that the gainers could compensate the losers, given specified valuations and feasible transfers. The existence of potential compensation does not establish that it is paid, that the parties possess equal purchasing power, or that all losses are ethically compensable.
Suppose a project creates monetary gains valued at 120 units for one group and losses valued at 80 for another. Under the assumed money metric there is a potential surplus of 40. If no transfer occurs, the second group still bears its loss. If 85 units are transferred at zero administrative and behavioral cost, the resulting money changes become +35 and +5. A Pareto improvement in the relevant money metric may then occur, provided that the losses are fully captured by money and no third-party effect remains. The example demonstrates why a compensation test and a compensated allocation are different objects of analysis.
Willingness to pay also reflects capacity to pay. Two persons may attach similar urgency to a safe dwelling while their incomes differ considerably. Their measured monetary valuations can differ even when the need is similar. Using unweighted monetary benefits as social benefits adopts an evaluative convention with a particular relationship to wealth. Distributional weights can change the ranking, but they introduce an ethical and measurement obligation of their own. The weights must be explained in relation to the unit of advantage and the counterfactual distribution.
Welfarism and interpersonal comparison
Utility as information and as a scale
Welfarism evaluates social states through individual welfare information. A conventional additive specification is W(x) = sum over i of a_i u_i(x), with positive social weights a_i. Its use depends on the interpretation of u_i, the comparability of welfare between persons, and the properties of the social weights. Harsanyi's (1955) cardinal welfare analysis is a classic contribution to the formal problem of interpersonal utility comparisons and additive social assessment. The relevant article record is directly accessible; this monograph does not reconstruct its individual theorem numbering or claim a complete proof from a restricted publisher page.
An ordinal utility representation preserves preference rankings under any strictly increasing transformation. Adding two ordinal numbers from different persons generally gives a result that changes with the arbitrary choice of numerical representation. Suppose person 1 prefers A to B and person 2 prefers B to A. Represent these rankings by u_1(A)=4, u_1(B)=0, u_2(A)=0, u_2(B)=3. Addition favors A. Multiplying the second person's utility representation by two produces 4 for A and 6 for B, reversing the social ranking while preserving both individual orderings. Thus an additive ranking requires a justified common scale or a specified transformation-invariant construction.
Even cardinal utility under expected utility is ordinarily unique only up to person-specific positive affine transformations. The choice of interpersonal units is a further issue. A common multiplicative change of units retains the ranking of an unweighted sum, whereas independent rescaling can change it. An analyst using measured life satisfaction, equivalent income, or a health utility index must explain why the particular common scale is defensible and what information is lost in moving to that scale.
Preference satisfaction and ethical relevance
A preference may concern one's own consumption, the outcome for a friend, the choices of another person, or an institutional arrangement. Its existence is evidence about the person's valuation; whether and how it should enter a public decision remains an ethical question. Discriminatory preferences, adaptive expectations, and preferences about other persons' private decisions create particularly important difficulties for a welfare criterion. The choice of a preference domain is itself part of the evaluation.
Preferences can also adapt to longstanding restrictions. An individual who has never had access to higher education may report low demand for it. The report may accurately reveal current expectations and yet fail to identify the advantage that would become attainable after the restriction is removed. Economic welfare analysis can examine both observed choices and counterfactual options. The broader analysis requires assumptions about opportunities and information rather than treating reported satisfaction as a direct and exhaustive measure of justice.
Equality, priority, and sufficiency
Equality concerns comparative differences in a specified dimension. Priority assigns greater evaluative importance to improvements for persons at lower levels of advantage. Sufficiency focuses on whether people reach a defensible level of advantage. They can converge in policy recommendations and differ in important cases. Lamont and Favor’s (2017) distributive justice survey surveys the diversity of metrics, recipients, and principles through which such judgments are formed.
Consider A=(20,80), B=(40,60), and C=(40,100). Equal-weight total advantage values are 100, 100, and 140 respectively. Absolute inequality is 60, 20, and 60. A maximin ranking gives B and C equal first rank because each has a minimum of 40, while A has a minimum of 20. C Pareto dominates B if the coordinates are comparable individual advantage levels. A purely gap-minimizing criterion favors B over C even though no person is worse off in C. This exposes the substantive difference between an evaluation of inequality and an evaluation of social states that also values gains.
A prioritarian specification can take the form W = sum g(a_i), where a_i now denotes a comparable level of advantage and g is increasing and concave. Marginal gains at low levels receive a greater contribution to W. In the example, taking g(a)=square root of a yields W(A) approximately 13.416, W(B) approximately 14.071, and W(C) approximately 16.325. The specification favors a more equal allocation at equal totals and accepts an improvement for a better-off person when no one loses. In this illustrative exercise, concavity encodes priority at the adopted advantage scale. If the advantage measure already embodies a concave transformation of resources, applying another concave function adds a further normative choice.
A threshold of 35 yields one person below sufficiency in A and none in B or C. A simple headcount treats B and C alike; an evaluation of the depth of deprivation distinguishes a shortfall of one unit from a shortfall of fifteen units. A threshold criterion may be supplemented with priority above or below the threshold. Its relevance depends on the reason for selecting that level: a biological need, a legal minimum, a socially defined living standard, or another public justification. In a multidimensional assessment, attainment in one dimension may fail to compensate for severe deprivation in another.
These examples are independent calculations. They demonstrate why the expression “reducing inequality” needs an explicit measure and why a plausible distributive judgment should disclose whether it values equality itself, gains to the disadvantaged, attainment of a floor, or a combination of these considerations. A policy comparison with all persons above a threshold may still matter to equality and welfare; an inequality-reducing change caused by deterioration at the top may leave the position of the disadvantaged unchanged.
Justice as fairness and institutional structure
John Rawls's A Theory of Justice (1971; revised edition 1999) develops justice as fairness around the basic institutions of society and a hypothetical choice of principles under impartial conditions. Its historical contribution is especially important to economics because it treats the institutional terms of cooperation as the object of distributive justification. The familiar first principle protects equal basic liberties, while the second addresses fair opportunity and the distribution of social and economic advantages. The account developed further in later work is summarized in Leif Wenar's study of Rawls (2021 substantive revision; Winter 2022 archive).
Rawls's difference principle evaluates institutional inequalities through their effects on the least advantaged, within a structure that gives priority to basic liberties and fair equality of opportunity. An economic illustration compares feasible institutional outcomes A=(30,30), B=(45,90), and C=(40,150), interpreted narrowly as two representative positions on an agreed primary-goods measure under otherwise equal liberties and opportunity conditions. A maximin proxy ranks B first because 45 exceeds 40 and 30. The unweighted sum ranks C first. B is unequal and nevertheless improves the lower position relative to A. This exercise illustrates the reason that justice as fairness cannot be reduced to equality of realized income.
The full Rawlsian account concerns the basic structure, social primary goods, and institutional conditions. The smallest observed income in a survey is neither its complete metric nor necessarily its representative least-advantaged position. Applying the difference principle to tax policy requires examination of how taxation, public services, labor institutions, property, and opportunity affect lifetime prospects. A static two-person maximin calculation is a limited pedagogical approximation, not a comprehensive representation of Rawls's philosophy.
The original position is a device of justification. Its impartiality is achieved through restrictions on information concerning one's particular social position. It is not an observed labor-market negotiation or a survey of choices under risk. Different informational and decision-theoretic assumptions about a hypothetical chooser may support different principles. The philosophical argument therefore cannot be settled by showing that a group of experimental subjects chose one lottery over another. Such an experiment can inform a claim about human judgments under specified conditions, while the normative argument addresses what citizens have reason to accept as fair.
Resources, responsibility, and fair allocation
Resource-based evaluation focuses on the means at people's disposal and the institutional rules governing their use. This permits attention to the distinction between a disadvantage generated by circumstances and a consequence of an agent's own choice. A conception of equal resources, associated with Ronald Dworkin's (1981) Equality of Resources, uses the idea that resources have opportunity costs for other persons. His formulation is identified through the journal record and indexed text; the precise original argument is not reproduced as a fully reviewed textual exegesis.
An allocation may be called envy-free if no participant prefers the bundle allocated to another participant, assessed according to their own preferences. Consider two divisible goods and two persons, one with a relatively strong preference for good A and the other for good B. Giving both the same bundle is envy-free under identical bundles but may leave mutually beneficial exchanges available. A differentiated allocation can be both envy-free and efficient if the bundles reflect preferences and feasible resource totals. Envy-freeness is a criterion of fair allocation under specified rights and information; it does not independently identify how initial ownership, production costs, or needs should be treated.
An economic distinction between choice and circumstance becomes difficult when choices are shaped by prior institutions. A worker may accept an unsafe occupation because safer employment is unavailable in the region. It is analytically unsound to treat the decision as evidence that safety carries little value without examining the choice set. Moreover, a policy that rewards effort requires a measure of effort and a defensible treatment of productive capacity, family circumstances, and chance. Equal treatment of equal effort can conflict with incentives when effort is unobserved and outcomes also depend on skill or luck.
Resource equality and responsibility can support differentiated transfers. A person with unavoidable expenses may need more resources to possess a comparable real opportunity. A person who voluntarily spends a comparable amount on an optional luxury presents a different question. The classification is not self-executing. Administrative investigations of responsibility impose costs, can misclassify needs, and may intrude on privacy. A resource-sensitive system must consider these consequences as part of its institutional evaluation. It can adopt broad eligibility rules, individualized allowances, or a combination; the tradeoffs are matters of positive analysis and ethical justification.
Capabilities, agency, and opportunity
Amartya Sen introduced the capability-focused question in Equality of What? (lecture delivered 1979; published 1980). His subsequent chapter Capability and Well-Being (1993) develops an approach centered on valuable beings and doings and the real opportunities to achieve them. Functionings are realized achievements, while capabilities refer to attainable combinations of functionings. A person may possess a capability without exercising it.
Income and goods are means to achievements, and conversion varies with personal conditions and the environment. Two persons receiving identical food expenditure may achieve different nutrition because of metabolism, age, care responsibilities, or access to safe water. Two persons with equal nominal transport allowances may have different mobility because route availability differs. Equal income can coexist with unequal substantive freedom in each case. For economic analysis this creates a mapping from resources through conversion conditions to feasible achievements. The mapping is empirical and institution-dependent; the selection of valued capabilities requires a normative judgment.
Suppose two individuals have 100 monetary units and face a cost of 2 and 5 per unit of an essential service because of location or accessibility. Subject to no other expenditures, the feasible service quantities are 50 and 20. Equalizing the feasible quantity at 35 would require resources of 70 and 175, totaling 245 rather than 200. If service provision can instead lower the second individual's cost to 2, both can obtain 50 with the initial incomes. The arithmetic illustrates why equality of opportunity may involve changes in public infrastructure rather than only a redistribution of monetary resources. It also illustrates that the technology of provision changes the attainable set.
Agency extends beyond an individual's own achieved welfare. An agent may value participation in a community project or accept a personal cost for the well-being of another. A public evaluation of agency can consider opportunities for voice and collective action. Treating such choices exclusively as changes in personal satisfaction may miss the institutional significance of the action. The evaluative metric must be fitted to the object of inquiry: freedom, achievements, subjective welfare, or another specified dimension. The capability perspective is therefore especially useful when a policy changes the range of available actions rather than simply transferring purchasing power.
Rights, liberty, and collective evaluation
The limits of ordinal aggregation
Kenneth Arrow's Social Choice and Individual Values (1951; 2012 edition) established the formal problem of constructing a social ordering from individual preference orderings. For at least two individuals and three alternatives, unrestricted domain, a complete and transitive social ordering, weak Pareto (unanimous strict preference), independence of irrelevant alternatives, and nondictatorship cannot be jointly maintained in the standard ordinal aggregation setting. The theorem's statement and assumptions are corroborated by List’s (2022) social choice theory survey. The result concerns this specified class of aggregation rules, not the logical impossibility of democratic decision or rational evaluation in general.
A simple majority cycle illustrates one difficulty, though it is not Arrow's general proof. Three voters have rankings A>B>C, B>C>A, and C>A>B. Pairwise majorities rank A above B, B above C, and C above A. The resulting relation is cyclic. If a committee's agenda first compares A against B and then compares the winner with C, C wins. If it first compares B against C and then the winner with A, A wins. Outcomes depend on an agenda rule. The observation creates an institutional question about the design and control of agendas, separate from the distributive question of which alternative is best.
Possible responses to the general impossibility include restrictions on preference domains, broader interpersonal information, altered independence conditions, or decision procedures that serve a more limited purpose than producing a complete social ordering. Each response changes the original problem. A common-scale welfare sum can supply an ordering of social states under its own information and ethical conditions; it does not refute Arrow's theorem. The intellectual value of the result lies in making the sacrificed conditions visible.
Individual decisiveness and Pareto conflict
Sen's (1970) Paretian liberal impossibility demonstrates a possible conflict between a minimal allocation of individual decisiveness and unrestricted preference-based Pareto assessment. The exact formal statement is checked against the cited social-choice survey, because the journal article itself is access restricted. An independent illustration uses three states A, B, and C and two persons whose rankings are: person 1, A>B>C; person 2, B>C>A. Assign person 1 the decisive choice between A and B and person 2 the decisive choice between C and A. These two rights require A>B and C>A. Both persons prefer B to C, so Pareto requires B>C. The three strict social comparisons form a cycle. The example is designed solely to demonstrate the inconsistency of the stipulated pairwise requirements.
The result clarifies why a theory of rights cannot always be represented as unrestricted interpersonal preference aggregation with an additional rights constraint. A right can instead be interpreted through a legally protected domain of actions, a procedural constraint on decision-making, or a restriction on admissible reasons for collective interference. These interpretations may alter which states are compared or which preference information is relevant. Their institutional meaning should be evaluated explicitly rather than assumed by the mathematical notation.
A useful applied distinction is between a social score calculated for an alternative and its admissibility. If a public policy would violate a protected claim under the chosen legal or ethical framework, it may be excluded from the feasible evaluative comparison even when its aggregate measured benefit is high. A constrained optimization over admissible alternatives differs from an additive welfare score in which every right has a finite compensatory price. Deciding which claims belong in such constraints is a substantive question for political and legal justification.
Measurement and distribution-sensitive comparison
Inequality indices and equally distributed equivalent income
Anthony Atkinson's (1970) inequality measurement study connects the evaluation of income distributions to explicit distributive judgments. The operational formula below is corroborated by John Creedy’s New Zealand Treasury analysis (2014). For positive incomes y_i, mean μ and inequality aversion ε≥0, let EDE be the power mean with order 1−ε:
EDE(ε) = (mean(y_i^(1−ε)))^(1/(1−ε)) for ε≠1; EDE(1) = exp(mean(log y_i)).
Atkinson's inequality index is A(ε)=1−EDE(ε)/μ. At ε=0 this family gives A=0 for every positive distribution; as ethical sensitivity to low incomes increases, unequal distributions receive lower equivalent incomes relative to their mean. In the standard separable specification, EDE represents the equal income level yielding the same evaluated welfare as the observed unequal distribution. The formula requires attention to zero and negative income and to the chosen population weights; the positive-income examples below avoid those boundary problems.
For two equally weighted persons with incomes (20,80), μ=50. With ε=1, EDE=sqrt(1600)=40, so A=0.20. For (40,60), EDE=sqrt(2400)≈48.9898 and A≈0.0202. For (40,100), μ=70, EDE=sqrt(4000)≈63.2456 and A≈0.0965. The third distribution is more unequal than the second by this index but carries higher evaluated welfare on the EDE scale. Inequality and the level of advantage are distinct information. The result also shows why comparisons should report both the index and its underlying income level.
Sensitivity and measurement choices
A more demanding comparison uses D=(20,100) and E=(45,55). Their means are 60 and 50. At ε=0, EDE equals the means and D ranks higher. At ε=1, EDE(D)=sqrt(2000)≈44.721 and EDE(E)=sqrt(2475)≈49.749, so E ranks higher. At ε=2, the harmonic means are 33.333 and 49.5, respectively. This reversal isolates the significance of a normative parameter under otherwise fixed information. It is not evidence that a particular value of ε should be adopted in national policy.
The Gini coefficient is a summary of relative inequality and differs from a poverty rate or an individual welfare index. It can be expressed for a finite positive-income population as the sum of absolute pairwise differences divided by twice the square of population size times mean income. A relative inequality measure is invariant under proportional scaling: the distributions (20,80) and (40,160) have the same Gini coefficient, even though every person has twice as much income. A poverty headcount measured against a fixed real threshold can change. A threshold that moves with median income instead reflects a relative conception of poverty. The analyst must match the measure to the question and preserve its interpretation when comparing countries or time periods.
Data operations matter. A comparison of annual household disposable income should identify the treatment of taxes, cash transfers, household composition, inflation, and equivalence scales. Imputed public services, informal earnings, and housing costs can alter the interpretation. Missing top incomes can understate concentration; missing homeless individuals can obscure severe deprivation. Economic justice cannot be inferred from a single index calculated on an incompletely described statistical population.
Taxation, transfers, and incentive compatibility
Tax and transfer policy connects distributive judgments with private responses to rules. James Mirrlees's (1971) optimum income taxation is a foundational contribution to the incentive problem under informational constraints. The verified article record supports the historical attribution; the following screening example is an independent construction, not a reproduction of estimates or a claimed derivation of his full model.
Suppose worker type θ determines productivity and gross earnings y=θ l, where l is labor. A tax schedule T(y) yields consumption c=y−T(y). With utility c−l²/2, a worker selects y to maximize y−T(y)−(y/θ)²/2. At an interior differentiable optimum, 1−T'(y)=y/θ². The equality illustrates how a marginal tax rate changes the individual incentive to generate earnings under these particular utility and production assumptions. It is not a general labor-supply elasticity estimate. A social planner choosing schedules needs information about the distribution of θ, the utility specification, public expenditure, and the normative value assigned to different persons' welfare.
If productivity or effort is unobserved, a schedule cannot directly assign a tax by each hidden characteristic. The observed earnings of two workers can be equal even if one works longer at a lower wage. A tax on observed earnings treats them similarly, while a responsibility-sensitive criterion could make a different comparison if effort were observable. A policy mechanism must respect both its informational feasibility and its distributive aim. The incentive constraints and budget constraint delimit implementable alternatives; the social welfare criterion ranks them.
Tax incidence introduces further distinctions. A statutory tax liability may be borne through wages, prices, rents, profits, or other margins of adjustment. Its distribution depends on behavior and equilibrium responses. A transfer's beneficiary may face changes in eligibility, stigma, administrative cost, or housing prices. A static comparison of statutory payments is useful for identifying direct legal claims and insufficient for establishing ultimate economic incidence. The choice of horizon is equally important: consumption, labor supply, capital accumulation, and intergenerational inheritance respond over different periods.
A universal transfer may be more accessible and administratively simple than a targeted payment, while requiring more gross expenditure for the same payment to the disadvantaged. Targeting can use resources more directly but creates exclusion errors and incentives around eligibility boundaries. These are contingent consequences, not universal rankings. A defensible evaluation compares complete financing and delivery arrangements, including administrative burden and take-up, rather than assigning justice to the label of the instrument.
Opportunity and the evidence of neighborhood exposure
Opportunity-sensitive evaluation requires evidence about the constraints that shape later achievements. The Moving to Opportunity experiment investigated the consequences of offering housing vouchers to eligible families in high-poverty U.S. public housing contexts. Chetty, Hendren, and Katz's (2016) neighborhood exposure study reports improved college attendance and later earnings for children who moved to lower-poverty neighborhoods before age 13, with slightly negative results for adolescent movers. The finding is assessed at the level of the retrieved published abstract. No numerical treatment estimate is imported.
The evaluative relevance is the relationship between childhood surroundings and later choices. A person's earnings at age thirty are partly produced through earlier circumstances, and a policy changing those circumstances can affect advantage long after the initial transfer. A contemporaneous measurement of household income would miss some of this mechanism. Likewise, an evaluation of equal formal access to housing vouchers needs information about actual take-up, barriers to moving, and which neighborhoods families can reach.
Randomized offers identify contrasts under the allocation design. Actual moving is a behavioral response to the offer rather than a randomly imposed action. Distinguishing intention-to-treat effects from estimates for participants induced to move requires appropriate instrumental-variable assumptions, including a first-stage relationship and restrictions on the ways assignment affects the outcome. Under standard conditions the induced-treatment effect applies to a particular participating subpopulation. It need not equal the average effect of relocating all families in every housing market. The distinction matters for any claim about distributive effects of a scaled program.
A further difficulty arises from general equilibrium. A housing program offered to a limited number of families can alter their opportunities without changing the broad supply of affordable dwellings. Large-scale expansion can change rents, school congestion, and neighborhood composition. A fair evaluation of expansion must incorporate these responses and examine effects on existing residents as well as new entrants. The published experiment provides evidence for a channel affecting opportunity; it supplies neither a complete national welfare calculation nor a decision about the normative priority of that channel.
Health insurance and the plurality of policy outcomes
The Oregon Health Insurance Experiment illustrates how policies produce multiple, potentially noncommensurate outcomes. Baicker and colleagues (2013), in The Oregon Experiment, use random allocation of lottery selection in the 2008 Medicaid expansion context to study coverage and outcomes after about two years. Their reported results distinguish increased health-care use, reduced financial strain and fewer positive depression screens, and lack of statistically significant improvement for the measured physical-health outcomes over that horizon. The evidence does not establish that the true physical-health effect is exactly zero.
From a financial perspective, insurance can protect household consumption against medical expenditure. From a health perspective, access and treatment may change morbidity and mortality over different horizons. From a capability perspective, security and access may expand feasible life activities even when a measured clinical variable has changed little. From a procedural or rights perspective, the distribution of eligibility and barriers to care may matter independently of the estimated average outcomes. Each evaluative basis identifies an aspect of the policy and calls for a distinct measure.
Consider two hypothetical programs with the same resource cost. P reduces a family's probability of catastrophic expense by a specified amount without a measured short-term clinical improvement. Q produces a small average clinical improvement but leaves catastrophic expenditure risk unchanged. A financial risk-protection criterion can favor P; a clinical criterion can favor Q. To combine them, an analyst needs comparable units or a deliberative rule for weighting dimensions. Even if an aggregate index is adopted, reporting separate dimensions prevents a single score from obscuring major distributional changes.
Statistical uncertainty enters the comparison directly. An estimate may be uncertain because the study has limited power to detect effects within the available period. A nonsignificant estimate is compatible with a range of effects, and a normative evaluation using estimates should preserve this range. Longer follow-up, improved measurement, and larger samples may alter the empirical understanding. The Oregon evidence is bounded by its study population, actual coverage response, institution, and observation period. Its value for economic justice is methodological: financial protection, utilization, clinical health, and subjective well-being have distinct meanings and should remain visible throughout policy assessment.
Public goods, commons, and procedural justice
Shared resources and institutions
The benefits of collective production and the distribution of common-resource access raise questions beyond private commodity allocation. A pure public good is nonrival and nonexcludable in the standard model; a common-pool resource has subtractable use and costly exclusion. An irrigation system, a fishery, and a publicly available warning service therefore present different problems of access, congestion, and financing. The physical characteristics of a resource influence the feasible governance arrangements, although they do not uniquely determine the just arrangement.
Elinor Ostrom's (2010) polycentric governance research is an important contribution to the institutional analysis of collective resource problems. The retrieved article record and associated research descriptions establish its scope; a complete comparison of the field studies is beyond the evidence assembled here. Polycentric governance directs attention to multiple interacting centers of decision and to the possibility of varied institutional arrangements in different resource environments. This perspective makes the distribution of decision authority an economic variable in its own right.
Imagine an irrigation community with ten members upstream and ten downstream. A rule permitting unrestricted withdrawal can be administratively straightforward yet impose a systematically greater risk of shortage downstream. An equal-volume quota may respond to this concern but be difficult to enforce or ill-suited to differences in seasonal land use. A differentiated quota with monitoring and appeal can address particular needs, while the authority administering it creates its own accountability problem. The institutional comparison must consider the allocation of water, the distribution of monitoring burdens, the reliability of supply, and the participation of those affected.
In a simple public-good model, if provision is continuous, everyone has quasi-linear preferences over a numeraire and the public good, and there are no income or external effects outside the model, an interior efficient quantity satisfies the sum of individual marginal willingness to pay equal to marginal cost. The equality specifies a provision margin. It does not determine the division of cost among beneficiaries. Charging the same amount to every person and financing provision through progressive taxation can support the same total output while producing different distributions of disposable resources. A social evaluation must account for both provision and finance.
Procedure as a dimension of evaluation
Procedural justice concerns how decisions are made, who can participate, how reasons are given, and how mistakes can be challenged. Procedural rules also affect outcomes through information revelation, compliance, and the ability to adapt to shocks. These instrumental effects can be examined empirically. The independent value of being heard or treated as an equal member of a cooperative scheme requires an articulated normative basis.
A rule announced in advance may provide greater predictability than an equally costly allocation decided arbitrarily after individuals have made investments. If the latter creates a risk of expropriation or favoritism, it changes incentives as well as procedural standings. When comparing institutional alternatives, analysts should distinguish ex ante commitments, realized outcomes, and the availability of review. A realized allocation that happens to be equal does not reveal whether the selection process offered equal standing or subjected some participants to avoidable uncertainty.
Procedural institutions also face participation costs. Attendance at a meeting can be more expensive for a worker with inflexible hours than for a manager with scheduling discretion. Formal inclusion can coexist with unequal effective representation. Providing accessible hearings, transparent records, and practicable appeals can therefore change both information and access to authority. The resource costs of these measures belong in feasibility analysis, and their value belongs in the explicitly stated evaluation of the institution.
International and intergenerational evaluation
Comparing distributions across countries
International assessment requires the distinction between a common statistical indicator and genuinely comparable advantage. Monetary measures need consistent concepts of resources and price adjustment. A given amount of income can provide different access to public services, infrastructure, housing, health care, and personal security across institutions. Purchasing-power adjustments address differences in the price of a specified basket; they cannot on their own establish equality in access to services or in political and social rights.
The United Nations Development Programme's inequality-adjusted Human Development Index incorporates distribution-sensitive adjustments to the human-development dimensions and uses a geometric mean to form the combined index. This illustrates an institutional effort to distinguish average attainment from its distribution. The index is an operational summary with a defined information base, rather than a comprehensive measure of every aspect of justice. Unobserved intersections among disadvantages and differences in data quality can matter when interpreting comparisons.
The OECD's (2024) How's Life? 2024 examines current well-being, inequalities, and resources supporting future well-being through a broad indicator set. Its account of post-2019 developments separates economic resilience from persistent pressures in housing, health, and social connectedness. The relevance here lies in the simultaneous examination of multiple dimensions, rather than any asserted current ranking of countries. The report's geographical coverage and dated observation periods delimit the inference. No 2026 trend is inferred from the 2024 report.
A comparison can therefore be structured around common outcome definitions, institutional context, and sensitivity to aggregation. For example, before interpreting a difference in disposable income between countries, a researcher should examine how much essential health and education expenditure is financed privately, how unemployment risks are distributed, and what public provision is available. Separate indicators can disclose divergences that an aggregate monetary comparison conceals. National means likewise need distributional breakdowns where data permit, especially when exposure to environmental hazards or access to services differs sharply within borders.
Present and future persons
Intergenerational assessment extends the population and horizon of social evaluation. Present production may deplete resources or create infrastructure whose benefits are received by later persons. A simple discounted sum of future assessed values incorporates a discount factor, but the rate's interpretation matters. A market return, an opportunity cost of public investment, and a pure ethical preference for earlier welfare are distinct quantities. Their numerical equality in a particular model requires arguments about the assumptions connecting them.
For a hypothetical improvement valued at 100 units in thirty years, discounting at 2 percent annually yields a present equivalent of approximately 55.21; discounting at 5 percent yields approximately 23.14. This arithmetic demonstrates sensitivity to a parameter, not an estimate of ethically appropriate discounting. When future individuals are expected to be poorer or more exposed to environmental hazards, a distribution-sensitive welfare assessment may place more value on their marginal resources. Uncertainty over population size, technological development, and irreversible loss creates further issues. A finite present-value comparison can be useful without settling whether ecological thresholds or protected claims should be treated as constraints.
International and intergenerational evaluation share an important challenge: the affected persons may lack standing in the decision procedure that selects the policy. The geographical boundary of government and the temporal boundary of electoral participation delimit actual authority. The ethical population used to evaluate external effects can be wider. A scientific analysis should describe the mismatch rather than treating the jurisdiction that makes a decision as the automatic boundary of all consequences that matter.
An integrated comparative policy exercise
Common feasible alternatives
The preceding chapters establish the need to compare evaluative rules on a common, explicitly described set of alternatives. Consider three equal-size groups, each represented by an individual, and assume their baseline advantage in arbitrary resource-equivalent units is (30, 60, 120). Four hypothetical net policy outcomes are given below. They incorporate all resource costs and behavioral responses by stipulation; no empirical claim is made about the feasibility of obtaining them in an actual jurisdiction. Assume initially that basic rights and institutional eligibility are identical, and the three coordinates are on a justified common scale.
| Alternative | Group L | Group M | Group H | Sum | Minimum | Persons below 50 |
|---|---|---|---|---|---|---|
| Baseline | 30 | 60 | 120 | 210 | 30 | 1 |
| Expansion | 35 | 65 | 145 | 245 | 35 | 1 |
| Targeted transfer | 50 | 60 | 105 | 215 | 50 | 0 |
| Broad provision | 45 | 75 | 110 | 230 | 45 | 1 |
The expansion alternative Pareto dominates the baseline for all three groups. The targeted transfer and broad provision both involve a loss for group H relative to baseline, and their superiority cannot be established by unanimity in this simplified metric. The total advantage ordering is expansion, broad provision, targeted transfer, baseline. The minimum ordering is targeted transfer, broad provision, expansion, baseline. The threshold count also favors the targeted transfer. If the public objective is to raise the least-advantaged position to at least 50, it is the sole feasible alternative among those listed.
Distribution-sensitive weighting
An additive evaluation using natural logarithms orders the states through their geometric means. The products of the three coordinates are respectively 216000, 329875, 315000, and 371250. The geometric means are 60, approximately 69.10, 68.04, and 71.87. The logarithmic evaluation thus gives priority to broad provision, followed by expansion, targeted transfer, and baseline. This is independently derived arithmetic from stipulated values. It reveals a ranking that neither the sum nor maximin produces.
Consider also a weighted linear valuation assigning each unit of advantage in group L weight w, and weight 1 to each unit of advantage in groups M and H. The scores for expansion, targeted transfer, and broad provision are 35w+210, 50w+165, and 45w+185. Expansion and broad provision are equal at w=2.5; broad provision and targeted transfer are equal at w=4. For w=1 expansion wins, for w=3 broad provision wins, and for w=5 targeted transfer wins. These comparisons describe a sensitivity exercise rather than a recommendation of group-specific weights. A public defense of such weights must explain why group membership tracks a morally relevant disadvantage and how the policy affects persons within each group.
The hypothetical ordering depends on the assumed resource-equivalent scale. If group L faces higher conversion costs for the relevant service, changes in advantage may differ from changes in resources. An assessment using a capability metric would require a revised consequence description. The exercise also assumes that rights conditions are identical. If expansion depended on violating a protected entitlement, its admissibility could change under a rights-constrained appraisal. An arithmetic score cannot decide that entitlement independently. The way available alternatives are described is thus as important as the aggregation formula.
Interpretation of the tradeoffs
The comparison identifies where disagreement lies. A disagreement about whether expansion raises group H's advantage to 145 is factual and can in principle be investigated through economic evidence. A disagreement about the use of a sum or a minimum is normative. A disagreement about the appropriate population or whether benefits should be measured as resources or capabilities concerns the informational basis. A disagreement about whether the targeted transfer can be implemented at its stipulated cost concerns institutional feasibility.
These dimensions interact. An analyst who discovers that a policy is legally unavailable changes the feasible set. An analyst who discovers unmeasured benefits changes the description of outcomes. A citizen who favors a sufficiency threshold can change the ranking without contesting either discovery. Public disagreement becomes more tractable when these differences are made explicit. The same quantitative evidence can support a range of conditional evaluations while the criteria remain contested.
This is the monograph's original comparative synthesis: economic justice is best assessed by keeping the chain from causal consequences to relevant information, evaluative comparison, and institutional implementation visible in the argument. The exercise demonstrates how each link affects actual conclusions. This is an interpretive synthesis of established work and independent examples. It carries no claim to a newly discovered mathematical theorem or a named authorial doctrine.
Methodological limits and open research problems
Identification and value judgments
An applied evaluation can specify potential outcomes A_i(1) and A_i(0) for individuals under policy and counterfactual conditions. The average treatment effect E[A_i(1)−A_i(0)] is one possible estimand. It does not determine the complete distribution of either potential outcome, the identity of every person who gains or loses, or the normative significance of those changes. Even if an experiment identifies marginal distributions under each assignment, the joint distribution of individual treatment effects generally requires further assumptions. A distributive evaluation must disclose what can be identified from its design.
An inequality-sensitive welfare comparison can in principle evaluate the distribution under each policy separately, provided the metric is comparable, the population is defined, and missing outcomes are adequately treated. But assessments that aim to compensate people whose individual circumstances deteriorate demand information about who loses, which may be unavailable. Administrative records, surveys, and structural models can supplement evidence, but each brings identification assumptions. The methodological objective is to distinguish sensitivity to empirical uncertainty from sensitivity to normative judgment.
Robust comparison is possible when two alternatives are ordered alike under a justified range of specifications. If a gains distribution satisfies dominance conditions relevant to the selected class of social welfare functions, varying the parameters within that class may preserve a ranking. When distributions cross, rankings may depend on the assumed curvature or weights. Reporting this dependence is an informative result. It identifies the magnitude of the judgment needed to favor one policy over another and permits public scrutiny without pretending that factual research alone selects the judgment.
Scenario analysis should also distinguish uncertainty about impacts from uncertainty about the decision procedure. A social insurance scheme may have a predicted mean transfer but uncertain participation and cost. Estimating alternative participation scenarios is an empirical modeling exercise. Choosing whether to protect persons against the worst credible outcome or to maximize expected assessed welfare requires an additional decision criterion. In the presence of irreversible losses, the appropriateness of treating all effects as compensable can itself become contested.
Open questions and boundaries
Three research directions follow. The first concerns measurement of effective opportunities, including the joint distribution of needs, services, and institutional restrictions. Separate indicators may fail to reveal people who experience several simultaneous disadvantages. Panel data and carefully designed measures can improve the empirical foundation, while the selection and interpretation of capabilities remain subject to public reasoning.
The second concerns institutional endogeneity. Policies change behavior, investment, bargaining, and the distribution of political influence. A static welfare comparison can omit feedback that subsequently alters the feasible set. Future research should trace the transmission channels, distinguish short-run incidence from long-run consequences, and examine which claims about institutional responses are supported by credible designs in specific contexts.
The third concerns procedural and substantive evaluation when preferences diverge and protected claims are disputed. Formal results identify incompatible requirements within stated aggregation frameworks. They do not supply a unique constitution or mechanism. Institutional studies can examine representation, appeal, accountability, and the ability of different groups to participate, while normative inquiry explains which features merit protection. Understanding the relationship between these modes of analysis remains an open scientific program.
The evidence selected for this monograph supports conceptual, mathematical, and methodological synthesis. The article-level access limitations recorded in the evidence dossier constrain historical interpretations of particular classic works; several classic theoretical records are available only through bibliographic landing records, excerpts, or corroborating sources. The study does not reconstruct national inequality series, estimate optimal tax rates, or infer a universal causal effect from the cited experiments. Those are distinct research projects. The formal examples are reproducible from their stated assumptions and all numerical values are illustrative. The empirical conclusions are bounded to the designs and scope of the cited investigations.
Conclusion
Justice in economic theory concerns the institutional and distributive terms of cooperation among persons. Social evaluation gives analytical structure to the comparison of those terms by specifying the alternatives, relevant information, affected population, and evaluative criteria. This structure clarifies the importance and the limits of efficiency, utility aggregation, resource equality, priority, sufficiency, capabilities, rights, and procedural fairness.
The Pareto criterion identifies unanimous improvements and efficient boundaries; conflicts of advantage require further judgment. Cardinal and interpersonal welfare aggregation requires justified units and weights. Inequality measures describe dimensions of a distribution under specified assumptions and should be interpreted with information about its level. A capability assessment considers effective opportunities and heterogeneous conversion conditions. Rights and procedures concern the permissible and publicly justified exercise of authority. These dimensions can combine within particular evaluations when their compatibility is defended. They can also generate divergent rankings of the same alternatives.
The study's independently constructed comparisons identify the sources of disagreement without treating every disagreement as empirical. Formal examples show why sums, minimum positions, thresholds, and distribution-sensitive indices can rank identical feasible allocations differently. The research on housing opportunity and medical insurance demonstrates the need to preserve distinct outcomes and causal boundaries. Institutional analysis of taxation and common resources establishes that feasibility and the distribution of authority are substantive parts of economic assessment.
The scientific task is to articulate the premises of comparison, investigate actual consequences, examine the institutional mechanisms that produce them, and state the resulting normative judgments with their grounds and limits. An evaluation becomes more informative when it reveals the circumstances under which its ranking changes. Economics can contribute rigorously to justice by combining its explanatory and empirical resources with explicit, reasoned evaluation of social arrangements.
References
Arrow, Kenneth J. (1951; 2012 edition). Social Choice and Individual Values. New Haven, CT: Yale University Press. Foreword by Eric Maskin in the 2012 edition. Publisher record.
Atkinson, Anthony B. (1970). On the Measurement of Inequality. Journal of Economic Theory 2(3), 244–263. DOI: 10.1016/0022-0531(70)90039-6. Bibliographic record.
Baicker, Katherine, Sarah L. Taubman, Heidi L. Allen, Mira Bernstein, Jonathan H. Gruber, Joseph P. Newhouse, Eric C. Schneider, Bill J. Wright, Alan M. Zaslavsky, and Amy N. Finkelstein, for the Oregon Health Study Group (2013). The Oregon Experiment — Effects of Medicaid on Clinical Outcomes. New England Journal of Medicine 368(18), 1713–1722. DOI: 10.1056/NEJMsa1212321. Published article PDF hosted by J-PAL.
Bergson, Abram (published as Abram Burk, 1938). A Reformulation of Certain Aspects of Welfare Economics. Quarterly Journal of Economics 52(2), 310–334. DOI: 10.2307/1881737. Journal record.
Chetty, Raj, Nathaniel Hendren, and Lawrence F. Katz (2016). The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment. American Economic Review 106(4), 855–902. DOI: 10.1257/aer.20150572. Publisher article record.
Creedy, John (2014). Interpreting Inequality Measures and Changes in Inequality. Treasury Working Paper 14/20. Wellington: New Zealand Treasury. Working paper.
Dworkin, Ronald (1981). What Is Equality? Part 2: Equality of Resources. Philosophy & Public Affairs 10(4), 283–345. Journal archive.
Fleurbaey, Marc, and François Maniquet (2011). A Theory of Fairness and Social Welfare. Cambridge: Cambridge University Press. Online edition 2013. DOI: 10.1017/CBO9780511851971. Publisher preface.
Harsanyi, John C. (1955). Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility. Journal of Political Economy 63(4), 309–321. DOI: 10.1086/257678. Journal record.
Kaldor, Nicholas (1939). Welfare Propositions of Economics and Interpersonal Comparisons of Utility. Economic Journal 49(195), 549–552. DOI: 10.2307/2224835. Journal record.
Lamont, Julian, and Christi Favor (2017 substantive revision). Distributive Justice. Stanford Encyclopedia of Philosophy. Stanford University, Metaphysics Research Lab. Encyclopedia entry.
List, Christian (2022 substantive revision). Social Choice Theory. Stanford Encyclopedia of Philosophy. Stanford University, Metaphysics Research Lab. Encyclopedia entry.
Mirrlees, James A. (1971). An Exploration in the Theory of Optimum Income Taxation. Review of Economic Studies 38(2), 175–208. DOI: 10.2307/2296779. Bibliographic record.
OECD (2024). How's Life? 2024: Well-being and Resilience in Times of Crisis. Paris: OECD Publishing. DOI: 10.1787/90ba854a-en. Report.
Ostrom, Elinor (2010). Beyond Markets and States: Polycentric Governance of Complex Economic Systems. American Economic Review 100(3), 641–672. DOI: 10.1257/aer.100.3.641. Article record.
Rawls, John (1971; revised edition 1999). A Theory of Justice. Cambridge, MA: Belknap Press of Harvard University Press. DOI for digital edition: 10.2307/j.ctvkjb25m. Book archive.
Sen, Amartya (1970). The Impossibility of a Paretian Liberal. Journal of Political Economy 78(1), 152–157. DOI: 10.1086/259614. Journal record.
Sen, Amartya (1980; lecture delivered May 22, 1979). Equality of What? In The Tanner Lectures on Human Values, Volume 1. Official lecture archive.
Sen, Amartya (1993). Capability and Well-Being. In Martha C. Nussbaum and Amartya Sen (eds.), The Quality of Life, pp. 30–53. Oxford: Clarendon Press. DOI: 10.1093/0198287976.003.0003. Chapter.
United Nations Development Programme (n.d.). Inequality-adjusted Human Development Index (IHDI). Human Development Reports. Indicator documentation.
Wenar, Leif (2021 substantive revision; Winter 2022 archive). John Rawls. Stanford Encyclopedia of Philosophy. Stanford University, Metaphysics Research Lab. Archived encyclopedia entry.