-1970: From Optimum Savings to Social Choice
It would be an interesting exercise to establish how many of Sen’s methods and ideas can be detected in some way in his doctoral Choice of Techniques: An Aspect of the Theory of Planned Economic Development (Sen I960)—for- mal proofs not allowed to drown out intuitive explanation, spurious mathematical precision challenged, utility measurement and comparison queried, divergence of market and social values explained, choice central with unavoidably political aspects recognised, assumed homogeneity of developing countries debunked, and more. The substance of the thesis is summed up with customary clarity by Atkinson:11 by creating an elegant analytical framework for discussing choice of technique in circumstances of scarce capital but plentiful labour, it establishes ‘a practical tool for investment appraisal’[1] [2] in the determination of a shadow price for labour that takes into account ‘the full economic implications of employing an additional per- son...[which may include] reduction in surplus for investment’ (Atkinson 1999: 184). With continued work in the 1960s on investment and distribution in neoclassical and Neo-Keynesian models of growth, Sen was the ideal editor of a collection of papers on growth theory in the Penguin Modern Economics Readings series (speaking to ‘both sides’ of the discipline now an advantage, together with rare skill in conveying mathematical results in non-technical language): Growth Economics (Sen 1970a) with its fine selection of papers and wonderfully clear and exciting Introduction became the indispensable guide to this subject.[3]
Meanwhile, Sen was vigorously pursuing his prime interest in social choice theory, publishing more than ten papers relating to it during the decade, crowned by the classic volume mentioned in the Introduction, Collective Choice and Social Welfare (Sen 1970b). Stylistically, it experiments with alternation of formal chapters proving results with informal ones explaining their signif- icance—surely a pronounced experimental success[4]—but that is not its key claim to fame. Having set himself the task of investigating ‘the relation between the objectives of social policy and the preferences and aspirations of members of a society’ (ibid.: 1) in the wake of what appeared to be the gloomy implications for this of Arrow’s Impossibility Theorem,[5] Sen identified precisely where and why such problems arose,[6] explored the consequences of relaxing each of the conditions underlying Arrow’s result (generating a substantial technical literature, some signalling new problems), and restored a degree of hope to social choice theory with a groundbreaking argument about the possibility of making partial interpersonal comparisons, a move sufficient to overcome the impossibility. He argued also for bringing other forms of information to bear,[7] and he introduced the concept of ‘non-basic’ value judgements in which the judgements reached are evidence dependent. The ‘far-reaching impact’ of this work was emphasised in the Nobel citation, to which the reader is referred for a fairly compact formal treatment (Nobel Prize website 1998)—and to Sen’s Nobel Lecture for a wider and more informal one (Sen 1999b); see also helpful coverage of this very extensive literature in Sen (1977a, 1986, 2011b), Atkinson (1987), Suzumura (2002), and Anand (2008). Sen concludes Collective Choice and Social Welfare with the comment that while ‘pure’ systems of social choice may have theoretical appeal, it may be more useful to work with the ‘impu?rities’ of ‘partial interpersonal comparability... partial cardinality ...restricted domains...intransitive social indifference.incomplete social preference...and so on.. .while purity is an uncomplicated virtue for olive oil, sea air and heroines of folk tales, it is not so for systems of collective choice’ (Sen 1970b: 200).
Two especially memorable illustrations of how supposed ‘impurities’ can factor in are selected here. First, on interpersonal comparisons, observing that influential previous discussion had failed to distinguish ‘between some comparability and total comparability of units’, Sen presses home the point by considering what might reasonably be concluded about aggregate Roman welfare when Rome was burnt ‘while Nero played his fiddle’. The judgement that ‘the sum total of welfare went down’ assumes some comparability, but the comparisons need not be precise, and ‘one-to-one correspondence’ of every Roman’s welfare units is certainly not required. (Similarly, neither precise knowledge of individual welfare functions nor exact correspondence between welfare units is necessary for it to be reasonable to assert that the sum total of welfare under the existing distribution of money income ‘is less than what could happen with a more equal distribution’ (ibid.: 99-100; italics in original).)
Second, the book’s discussion of ‘non-utility’ information[8] draws, among other cases, on a now famous example from a paper of the same year through which Sen had shown its relevance by presenting an impossibility result of his own. This was ‘The Impossibility of a Paretian Liberal’ (Sen 1970c). Suppose a prudish character who would prefer not to read an available copy of a licentious book nevertheless prefers reading it himself to opting for this access to depravity to be given to an individual who is already all too lewd. But the lewd person, though eager enough to read such a book, nevertheless would prefer that the prude be given so eye-opening an experience. Then it is apparent that if the book is to be read by one of them, both individuals happen to prefer that the reader should be the prude—a Pareto-superior outcome, then, to its being read by the lewd. Yet on liberal principles, what a person reads, there being no harm to others, is a matter for that individual’s own preference only, so the book would be read by the lewd person who relishes the prospect, not the prude who does not. In these circumstances, it is impossible simultaneously to meet both Paretian and liberal requirements. This hypothetical combination of attitudes neatly demonstrates then that, when preferences have what has become known as a ‘meddlesome’ dimension (here by including the private/personal reading materials of oth?ers within their domain), this means that the potential for conflict between Paretian considerations and liberal principles cannot be eliminated.
This result led to much discussion. Various escape routes from the impasse, of varying subtlety, were mooted and are reviewed in Sen (1983, 1999a: 363-364, 2009: 309-314). One line would be just to side with the stated preferences: if this is what the individuals do in fact prefer, would it not be reasonable—even liberal—to indulge them? But, as Sen points out, without entering into some form of contract—peculiarly in the context—the prude might cheat by merely trying to look as if he is reading; and also this ‘solution’ involves sidestepping the issue by means of a reinterpretation of what liberalism requires. (Equally one could simply side with liberal principles, but again this seems something of an evasion.) Sen argues that the resolution is to recognise the conflict and face up to evaluating priorities, accepting that a political dimension enters in (Sen 1999b: 364).
An alternative path some others had taken was to eliminate the blockage by disallowing meddlesome—anti-liberal—preferences in welfare calculations. This move chimes in with a more general tendency to resort to preference ‘laundering’ or ‘cleansing’ to rescue utilitarian thought from problems it can face over inclusion not only of illiberal but also of misinformed, miscalculated, short-term, externally swayed, antisocial, even vicious, preferences. There is a fine account, complete with devastating thrusts, of this sort of procedure in the Introduction to Utilitarianism and Beyond which Sen edited jointly with eminent Cambridge philosopher Bernard Williams (Sen and Williams 1982: 1-21). In the case of misinformation or lack of care in reasoning, the adjustment of an individual’s preferences towards what are deemed her ‘true’ preferences leaves a very significant problem—‘a “true” preference is not necessarily an actual preference at all’ (ibid.: 10)—while more radical laundering in line with what are conceived to be preferences consistent with morality begs more questions still.
The Sen-Williams Introduction gives a succinct and pivotal insight into Sen’s dissatisfaction with the utilitarian underpinning of the foundations of much of economic theory, building on earlier themes and presaging future ideas, with a focus on the informational limitations of both utilitarian and Rawlsian theories. In the utilitarian context, with respect to judging states of affairs solely in terms of ‘people getting what they prefer’ (ibid.: 3),[9] several perils were identified. One was that of reducing ‘all interests, ideals, aspirations, and desires’ to factors that are regarded as ‘on the same level, and all representable as preferences, of different degrees of intensity perhaps, but otherwise to be treated alike’ (ibid.: 8). Another came in characterising utility in a dual way, defined on the one hand in terms of choice and on the other as having ‘a particular content...usually in the form of maximising the person’s “self-interest’’, or “well-being”’ (ibid.: 12). A third was the loss of information on welfare levels through focus on sum-ranking only. And a fourth was the malleability of preference under social conditioning. In the Rawlsian context, while the shift away from derived utility to a concentration on the availability of rights and opportunities was welcomed, an informational gap came in insufficient recognition of differences in people’s needs. I take up Sen’s development of these themes below.
- [1] See also the informative summary in Riskin (2006: 541-542).
- [2] Followed up in a UNIDO manual of 1972 that Sen co-authored.
- [3] Including as it does a key extract from the other ‘bible’ on this, the Hahn and Matthews survey ofgrowth theory (Hahn and Matthews 1964).
- [4] As the Nobel Committee judged, ‘the style.. .provided a new dimension’ (Nobel Prize website 1998).
- [5] The proof that, for a finite set of individuals and a finite set of at least three social states, if certain seemingly mild and reasonable conditions (unrestricted domain [sometimes expressed through two axioms],Pareto efficiency, independence of irrelevant alternatives, and non-dictatorship) are to be simultaneouslysatisfied, there exists no social choice procedure for aggregating individual preferences into preferenceorderings for society as a whole.
- [6] This included, for instance, establishing necessary and sufficient conditions for the existence of a choicefunction under majority rule (possible inconsistencies in the majority method having been exposed in theeighteenth century by Borda and Condorcet).
- [7] See Morris (2009: 5).
- [8] Though as Sen allows, this description might be challenged under the broader utilitarianism ofJohn StuartMill, that champion of liberty—the broadening indicating a recurrent form of defence of utilitarianism.
- [9] The brand of utilitarianism most familiar in modern economics.
