NOTE-257: Sen's Capability Approach

Contribution

This is a compact account of the capability approach as Sen built it, separate from Nussbaum’s theory. It is organised around Sen’s motivations, his critiques of rival accounts, the internal structure of evaluation in his approach, the criticisms made of it with Sen’s replies, and the theories others built to fill its deliberate gaps: Robeyns, Alkire, Anderson, Alexander and Nussbaum. It is the only source in this batch that sets out Sen’s mechanism for reaching partial agreement on weights (§3b), and the only one that pairs each criticism with Sen’s response (§5).

Key insight

Sen’s approach has two stages.

• Valuation. Which functionings matter, and how much? This is a social-choice exercise of public reasoning and democratic decision (§3b).
• Evaluation. What capability do people actually have? This is a second-order mapping that “takes the value of functionings as given” (§3c).

Its under-theorisation is deliberate. On the entry’s reading it follows from Sen’s concern that evaluation should not “take a short-cut by excluding all sorts of information from consideration in advance” (§1b(4)). Incompleteness is therefore designed in, not left over: agreement may be partial, and it can still guide action.

Assumptions

The premises and authorities the entry relies on:

• Sen’s four motivating concerns (§1b). The entry lists them without citation:
  1. People differ in converting resources into functionings.
  2. Adaptive preferences make subjective metrics unreliable.
  3. Options not taken still matter. The example is fasting versus starving.
  4. Evaluation should not exclude information in advance.
• A value-laden definition (§3a). “Capability refers to the set of valuable functionings that a person has effective access to”, and it represents “the effective freedom … to choose between different functioning combinations … that she has reason to value”. This definition is value-laden (see Connections).
• Primary sources relied on. Sen 1979, 1985 (Commodities and Capabilities), 1989, 1999 (Development as Freedom, pp. 70–71 for the five sources of variation) and 2004a/b. Almost every claim is attributed to “Sen” without a page reference.

Key results

What the entry reports, by section.

• §1a Origins.
  • First statement. Sen first introduced “capability” in the 1979 Tanner Lecture.
  • Lineage. The antecedents are Aristotle, Smith (relative poverty and “necessities”) and Marx (quoted on “replacing the domination of circumstances and chance over individuals by the domination of individuals over chance and circumstances”, uncited).
  • Main synthesis. Development as Freedom (1999) is called the most comprehensive account.
• §2a Utilitarianism. Sen rejects all three pillars.
  • Act-consequentialism. He replaces it with “comprehensive consequentialism”, which counts processes and principles. The example: equalising the capability for long life by biasing health care toward men would violate fairness.
  • Welfarism. It shows “valuational neglect” and “physical condition neglect”, and it is vulnerable to adaptive preferences. The entry quotes Sen 1985 (Commodities and Capabilities), pp. 21–22: “Our mental reactions to what we actually get and what we can sensibly expect to get may frequently involve compromises with a harsh reality … The deprivations are suppressed and muffled in the scale of utilities”.
  • Sum-ranking. It ignores the separateness of persons, and it moves resources away from inefficient converters. This point is overstated; see corrections.
• §2b Resourcism. Sen’s critique of Rawls stands in for all resourcism.
  • Heterogeneity. Equal primary goods do not give equal effective freedom when conversion differs.
  • Hard cases. Sen rejects Rawls’s strategy of settling the “normal” case first and extending to “hard” cases later.
  • Fetishism. Rawls “‘fetishize[s]’ resources as the embodiment of advantage”. This matches Sen 1979, p. 216: “an element of ‘fetishism’ in the Rawlsian framework”.
  • A place for resources. Resource distribution remains relevant to procedural fairness.
• §3a Functionings and capability.
  • The bicycle example. A bicycle does not provide transportation to a person without legs, even if she finds it delightful.
  • From one set to several capabilities. Later Sen speaks of “capabilities” in the plural, or of “freedoms”, rather than a single capability set.
  • Utility. Utility is both an output of functioning and a functioning in its own right.
• §3b Valuation.
  • Why Sen rejects a philosophically grounded list. It “denies the relevance of the values people may come to have and the role of democracy” (Sen 2004b). “The legitimate source of decisions … must be the people concerned.”
  • No complete ranking. Because of reasonable disagreement, social choice is unlikely to deliver a complete ranking of functionings.
  • Partial agreement. Action-guiding agreement is possible by intersection (different views reach the same judgement by different arguments) and by ranges of weights. With weights on literacy relative to health of ½, ⅓, ¼ and 1/5 across four views, all four agree that one unit of literacy plus two of health beats two of literacy plus one of health.
  • Basic capabilities. A subset of basic capabilities (education, health, nutrition and shelter at minimal levels) can be agreed relatively easily, for assessing poverty.
• §3c Evaluation.
  • Not a count of options. The value of a capability set is not the number of options: a tenth washing-powder brand is not the vote. Eradicating malaria expands capability although it removes an “option”.
  • Five sources of interpersonal variation (Sen 1999, 70–71): personal physiology, local environment, social conditions, relational perspectives (appearing in public without shame), and distribution within the family.
  • Coupled disadvantage. Disadvantages in conversion and in earning are often coupled.
• §4 Applications.
  • Global-local character. A capability is defined in the abstract, but its requirements are local. Relative income poverty can be absolute capability poverty.
  • Evaluation is multi-dimensional. Surplus in one dimension does not compensate for shortfall in another. The example is the sporting icon who cannot read.
  • Perspicuous contrasts. Philippines versus South Africa, 2010 HDR: GNI per capita (PPP) $4,002versus$9,812; life expectancy 72.3 versus 52 years; mean years of schooling 8.7 versus 8.2.
• §5 Criticisms, each with Sen’s reply.
  • Illiberalism (a). The Rawlsian neutrality objection. Sen’s reply: heterogeneity means equal means do not give equal freedom.
  • Under-theorisation (b). Sen names no capabilities and no distributive rule. This section also reports Pogge 2002 on institutional questions: weighting, basic priority, equality, cost, responsibility, and deprivations that cannot be remedied. Sen’s response is that the approach “is not a theory of justice but rather an approach to the evaluation of effective freedom”.
  • Individualism (c). Critics include Gore 1997 on irreducibly social goods; Nussbaum 2003 on the need to rank freedoms, since some, like racist speech, should be limited; and Giri 2000 and Gasper 2002 on thin, rationalistic agency. Sen’s replies:
    • the value of social goods is derivative on the reflective choices of those concerned (Sen 2004a, HDR 2004), so if people do not value a minority language on reflection, that “should trump the ‘right’” of the institution to continue;
    • freedom-to-choose is to be distinguished from the value of the options;
    • adaptive-preference and intra-household analyses go “some way” toward a situated sociology.
  • Information gaps (d). The HDI, built by Sen and ul Haq in 1990, is crude: three equally weighted dimensions. Sen was “won over” by ul Haq’s rhetorical argument for it. Sen’s replies are that much basic-capability data already exists, and that adopting the capability space will drive new data collection.
• §6 Filling the gaps.
  • Robeyns 2003. Five criteria for procedural list-making: explicit formulation, methodological justification, sensitivity to context, different levels of generality, exhaustion and non-reduction.
  • Alkire 2005. Finnis-style practical reason plus participation, yielding seven dimensions.
  • Anderson 1999. Democratic equality. It quotes her entitlements claim, cited to p. 317 (the SEP gives p. 316).
  • Alexander 2008. Republican non-domination.
• §7 Nussbaum. Her ten Central Human Capabilities are reproduced in full from Creating Capabilities (2011, 33–4). The entry covers her shift from an Aristotelian to a politically liberal justification, her threshold and dignity, and the critics Okin, Stewart, Menon and Alkire. Her list differs from Sen’s approach in four ways:
  • a theory of justice versus an evaluative framework;
  • dignity versus freedom;
  • rejection versus use of instrumental and feasibility analysis, where Sen’s “comprehensive consequentialism” makes room for it;
  • utopian thresholds versus incremental improvement.

Claims

id	claim	strength	support
C1	For Sen, selecting and weighting functionings is a social-choice exercise requiring public reasoning and democratic decision	moderate (as a report)	citation of Sen 2004b (§3b); no page reference or quotation
C2	Partial agreement via intersection and ranges of weights can be action-guiding without a complete ranking	moderate	worked numerical example (§3b), attributed to Sen without a citation; it is consistent with Sen 1979’s “partial ordering … supplemented by … conventions of relative importance” (p. 219)
C3	Sen names adaptive preferences as a defect of welfarism	strong (as a report)	direct quotation, Sen 1985 (Commodities and Capabilities) pp. 21–22 (§2a.ii); the quotation was not checked against that source. Its seed is visible in Sen 1979 p. 217 (the contented cripple with a “low aspiration level”)
C4	Utility-sum maximisation moves resources away from the disabled	weak as stated	assertion (§2a.iii); Sen 1979 pp. 203–204 makes it conditional on a lower marginal utility at the relevant points
C5	Sen’s response to under-theorisation is that the approach is not a theory of justice	weak	assertion (§5b); no citation
C6	Sen holds that the value of social goods is derivative on individuals’ reflective choices	moderate (as a report)	citation of HDR 2004 ch. 1 (§5c)
C7	Sen is a “comprehensive consequentialist”	weak	assertion (§§2a.i, 7c); no citation; the SEP (§4) reports whether the approach is consequentialist as contested
C8	Nussbaum’s list is almost unchanged by her cross-cultural dialogue and reflects American liberal values	weak (as a report)	reported critics (Okin 2003; Stewart 2001) (§7b)

Concepts

• functionings: states of “being and doing”, distinct from the commodities used to achieve them (§3a).
• capability: the set of valuable functionings a person has effective access to, that is, effective freedom to choose among combinations she has reason to value. The definition is value-laden (§3a).
• basic capabilities: the easily agreed subset tied to basic needs, used for assessing deprivation (§3b).
• adaptive preferences: preferences adjusted to a harsh reality, so that deprivation is “muffled in the scale of utilities” (§§1b, 2a.ii).
• comprehensive consequentialism: Sen’s position, on the entry’s account, that consequences and processes or principles together determine evaluation (§2a.i).
• valuational neglect and physical condition neglect: Sen’s two charges against welfarism (§2a.ii).
• intersection and ranges of weights: Sen’s devices for partial, action-guiding agreement (§3b).
• perspicuous contrasts: comparisons that show different capability outcomes at similar income levels (§4).

Connections

Where the IEP and SEP entries disagree (the SEP entry is Robeyns & Byskov, “The Capability Approach”):

• Definition.
  • IEP. Capability is defined in terms of valuable functionings and what one has “reason to value” (§3a).
  • SEP. The value-laden definition is one side of a debate. The SEP authors argue for a value-neutral definition at the framework level (SEP §2.1).
  • Which is Sen’s. The IEP’s definition is closer to Sen’s own formula as the SEP itself reports it (SEP §3.3).
• How Sen selects capabilities.
  • IEP. It credits Sen with a positive procedure: a social-choice exercise, intersection, ranges of weights, and basic capabilities.
  • SEP. Sen “has never explained in detail how such a selection could and should be done” (SEP §3.3). “Reason to value” admits three incompatible readings.
  • The difference. This is a real difference of emphasis. The IEP gives Sen more of a method than the SEP allows.
• Agency.
  • SEP. It reports Sen’s well-being/agency and freedom/achievement two-by-two (SEP §2.5).
  • IEP. The distinction is absent.
• Consequentialism.
  • IEP. It asserts that Sen is a “comprehensive consequentialist”.
  • SEP. Whether the approach is consequentialist is an open debate (Qizilbash, Kanbur, Suzumura).
• Rawls.
  • IEP. It presents Sen’s critique of Rawls without the Rawlsian rejoinder.
  • SEP. Rawlsians (Pogge among them) argue that Sen overstates the critique, and not all their replies have been rebutted.
• Pogge.
  • IEP. Pogge 2002 raises institutional questions.
  • SEP. It uses Pogge 2002 for the demarcation of capabilities and for the claim that Rawls accommodates special needs.
  • Neither states Pogge’s core argument.
• Individualism. The two entries complement each other.
  • IEP. Gore’s irreducibly social goods, and Sen’s reply that social goods are derivative on reflective choice.
  • SEP. Collective capabilities, and Robeyns’s reply distinguishing ethical from methodological individualism.
• Dating the origins.
  • IEP. The approach was “first articulated … in the 1980s”, while 1979 is named as the first statement.
  • SEP. The origins are Sen 1974 and 1979.

Against the primary texts in this batch. Sen’s “Equality of What?” (1979) confirms the IEP’s attribution of “fetishism” (p. 216). It qualifies the sum-ranking claim (pp. 203–204; see corrections). It also shows that Sen already had the partial-ordering idea behind §3b in 1979 (p. 219). The IEP gets the pleasure-wizard structure right, where the SEP inverts it. It does not use the term, but its sum-ranking paragraph has the more efficient converters gaining at the disabled’s expense.

Other batch items.

• Nussbaum 2003. Nussbaum’s “Capabilities as Fundamental Entitlements” is used in §5c (freedoms must be ranked, and some limited) and annotated in the bibliography as the “too vague” critique.
• Robeyns 2017. Robeyns’s later Wellbeing, Freedom and Social Justice supersedes the 2003 criteria given in §6a.
• Valentini and Zwolinski. Valentini’s critique and Zwolinski’s review concern The Idea of Justice (2009), which the entry does not cover. It predates the entry, but the entry never mentions it.
• Sen’s social-choice papers. The Nobel lecture and the Paretian-liberal paper are not cited. §1a mentions only that Sen won the 1998 Nobel for social choice.

In this record. LIT-202 (Pettigrew, “Consent and the Formation of Preferences”) works on the adaptive-preference problem that §§1b and 2a.ii put at the root of Sen’s rejection of welfarism.

Bearing on the record

What the entry says about Sen’s paywalled texts. For those texts it is a second-hand source. Here is every claim it makes about them, by section:

• Dewey Lectures (1985). No claim. It is not cited and not in the bibliography, and the well-being/agency distinction is absent.
• “Capability and Well-Being” (1993). No claim. It is not cited.
• “Capabilities, Lists, and Public Reason” (2004; the IEP’s “2004b”).
  • §3b. Sen rejects a philosophically specified list because “it denies the relevance of the values people may come to have and the role of democracy”. The legitimate source of valuation “must be the people concerned”. Sen therefore “proposes a social choice exercise requiring both public reasoning and democratic procedures of decision-making”.
  • The bibliography calls it “an interview … in which he elaborates on his rejection of a fixed list”.
  • Uncited claims. §3b’s intersection, ranges-of-weights and basic-capabilities points are not explicitly cited to it. Their source in Sen is unverified from this entry.
• Human-rights papers (“Elements of a Theory of Human Rights” 2004, “Human Rights and Capabilities” 2005). No claim. Neither is cited. “Human rights” occurs once, in §7a, about Nussbaum’s list entering “international human rights legislation”.
• Other paywalled Sen texts.
  • Commodities and Capabilities (1985). pp. 21–22, the adaptive-preferences quotation (§2a.ii). The bicycle example (§3a) is uncited but is Sen’s from that book.
  • Development as Freedom (1999). pp. 70–71, the five sources of variation (§3c).
  • HDR 2004 ch. 1. Social goods are derivative on reflective choice (§5c).
  • “Development as Capability Expansion” (1989). Listed only, with no in-text claim.

In short, this entry adds §3b’s partial-agreement account and §2a’s adaptive-preference quotation to what the SEP gives. It adds nothing on the Dewey Lectures or the human-rights papers.

It carries nothing for ML practice. There is nothing for the Anthology of the SOTA.

Limitations

• Dated and unrevised. It dates from 2012 and was never revised. It predates The Idea of Justice reception, and later literature such as Robeyns 2017 and Byskov’s work on selection. One empirical aside, on Saudi Arabia, is now false.
• Thinly referenced. Most attributions to Sen carry no page reference. Several key ones carry no citation at all: comprehensive consequentialism, intersection and ranges of weights, “not a theory of justice”, and the Marx quotation.
• Sen as a closed system. It omits agency freedom and achievement, the human-rights work, and Sen’s social-choice results. It reads Sen’s under-theorisation sympathetically, as deliberate.
• Pogge in outline only. The “Rawlsian critique” of Pogge 2002 is reduced to a list of questions.

Open questions

• Where do intersection and ranges of weights come from in Sen? Which of his texts states them: Inequality Reexamined, Development as Freedom, or the 2004 Feminist Economics piece? Settling it needs the primary text.
• Is “Capabilities, Lists, and Public Reason” an interview or an authored article? The IEP and the other sources disagree. Settling it needs sight of the Feminist Economics 10(3) item.
• Is Sen’s reply to the individualism critique his own? The entry says the value of social goods is derivative on reflective choice (HDR 2004). It is a strong claim, and it would need checking against the HDR 2004 chapter.

Corrections to the seeded skim

• none (there was no seed or dossier)
• The scope report says the entry is “undated in page”. The date is recoverable. It was originally published on 24 April 2012 and last updated on 25 April 2012, according to the Wayback capture of 1 May 2012. The text is unchanged since. Internal evidence agrees: it uses 2010 HDR data (Figure 2) and cites Nussbaum 2011. It still says women in Saudi Arabia are “presently” not permitted to drive, which has been out of date since June 2018.
• The scope report says the criticisms section “covers illiberalism, under-theorisation, individualism and information gaps”. That is correct, but it does not cover the agency distinction. The entry never mentions well-being freedom versus agency freedom, or the Dewey Lectures. “Agency” occurs twice: once as “individual agency” in the act-consequentialism paragraph, once as Sen’s “‘thin’ agency based picture of persons” in a critique. So the entry cannot serve as the second-hand source for the Dewey Lectures.
• The entry contradicts itself on the date. It says the approach “was first articulated … in the 1980s” (preamble; §1a says “developing the Capability Approach during the 1980s”). It also says Sen “first introduced the concept of capability” in “Equality of What?” (Sen 1979) (§1a). That lecture was delivered on 22 May 1979 and published in 1980.
• Sum-ranking (§2a.iii) drops Sen’s own qualification. The entry says utility maximisation would move resources “away from the sick and disabled to people who are more efficient convertors”. Sen 1979 (pp. 203–204) explicitly limits this to people with a disadvantage “in terms of both total and marginal utility at the relevant points”. He notes the opposite could hold at equal incomes, in which case “even utilitarianism would give the cripple more income”.
• Citation labels. The IEP’s “Sen 2004b” is “Capabilities, Lists, and Public Reason” (Feminist Economics 10(3)), whereas the SEP labels that paper 2004a. The IEP’s “Sen 2004a” is chapter 1 of the UNDP Human Development Report 2004, “Cultural Liberty and Human Development”. The IEP bibliography describes 2004b as “an interview with Amartya Sen”. The scoping report and the SEP list it as Sen’s article. I have not seen it, so whether it is an interview is unverified.