Keywords

JEL Classifications

Methodological individualism is a doctrine in the social sciences according to which a proper explanation of a social regularity or phenomenon is one that is grounded in individual motivations and behaviour. In other words, according to this methodology, individual human beings are the basic units from which we must build up in order to understand the functioning of society, economy and polity. We may not in all our research succeed in doing so but to committed methodological individualists such research must be viewed as provisional and ideally be accompanied by a slight feeling of inadequacy on the part of the researcher.

The social scientists who have been the focus of much of the debate on methodological individualism and, paradoxically, also the ones least touched by the debate are economists. Economists are typically held up as examples of the most unbending methodological individualists; and, on the rare occasions when economists have joined this debate, they have tended to agree with this. The difference is that most non-economists mean this as criticism, whereas most economists take it as praise.

At first sight this characterization of economics seems right. Textbooks of microeconomics almost invariably begin by specifying individual utility functions or preference relations and asserting that human beings are rational in the sense that they behave so as to maximize their own utilities. They then build up from this to explain market phenomena, make claims about social welfare and discuss prospects of national economic growth. In some macroeconomic models economists are unable to build all the way up from individual behaviour and use aggregate behaviour descriptions as the starting point. But these models are almost always accompanied by an effort to ‘complete’ them with proper micro-foundations; and the profession regards these models as somewhat incomplete and awaiting the definitive work.

That economics may not actually be quite as methodologically individualistic as often presumed by both the discipline’s admirers and its critics is a matter to which I return below. What is interesting to note here is that the debate on methodological individualism has been a surprisingly cantankerous one that has spawned enemies and intrigues. Some social scientists have sworn by it: no other method is worth its salt. Others have castigated it as an instrument of exploitation and maintenance of the status quo. Concepts and categorizations have multiplied over the years. We have come across methodological holism, methodological solipsism, atomism, ‘MIs’ (that is, methodological individualisms) of different types – 1, 2, 3… – creating the impression that the British intelligence had somehow got involved in the quest to understand this elusive concept.

One cause of the controversy is the confounding of positive and normative social science. To some commentators, methodological individualism implies that it is fine to leave it all to individuals, and by implication it amounts to an argument against government intervention. Friedrich von Hayek (1942) and James Buchanan (1989), for instance, have taken this line, as have some sociologists, who felt that the conservatism of traditional economics is founded in its adherence to methodological individualism. But this happens because of a possibly logical error, a failure to appreciate Hume’s law, namely, that a normative proposition cannot be derived from a purely positive analysis. Kenneth Arrow (1994) has rightly criticized the tendency of some writers to treat methodological individualism and ‘normative individualism’ as inextricably linked. Similarly, Marxists often link methodological individualism automatically to certain ethical implications. Roemer (1981) and Elster (1982) argue that this is not a valid link. In what follows I treat the two as separate and assume that methodological individualism has no automatic normative implications.

Origins

The term ‘methodological individualism’ was probably used for the first time in the English language in 1909 by Joseph Schumpeter. Even if that is not so, Schumpeter certainly thought so, and he pointed out in his paper in the Quarterly Journal of Economics that year that he had actually coined the term in German the previous year. But methodological individualism had been practised from much earlier, at least as early as Adam Smith (1776), and was described as a deliberate methodology, though without the term itself being used, by Carl Menger in 1883 (Menger 1883). Max Weber’s later exposition of it was published posthumously in 1922 (Weber 1922).

From the perspective of economics it seems reasonable to treat Menger as the first proponent of methodological individualism. He did so vociferously, dismissing the German historical school of economists and their methods as outdated and flawed. He advanced the idea of ‘spontaneous order’ in society, which sprang from atomistic individual behaviour, reminiscent of Adam Smith’s ‘invisible hand’ and the efficiency of markets that was an outcome of rational, self-interested behaviour on the part of individuals. Menger not only failed to acknowledge that some of his ideas on spontaneous social order were already there in Adam Smith but wrote in a tone almost suggesting that Smith had taken those ideas from him.

A distinction is often drawn in philosophy between methodological individualism and ‘atomism’. The latter is treated as a more extreme version of individualism, in which it is possible to characterize each individual fully without reference to society and then explain social behaviour by simply imagining such individuals being brought together in one society. Since the proponents of these ideas did not really define terms with that much care – and when they did, they went on to write in a way that disregarded their own definitions – I shall refrain from drawing fine distinctions and treat these neighbouring terms as all representing the broad idea of methodological individualism. Moreover, concepts like these are probably innately indefinable. They are understood through a combination of approximate definitions and repeated use.

It is useful in an exposition like this to think of the polar opposite of the term under consideration. This is captured in the concept of ‘methodological holism’, developed (without endorsement) by the philosopher John Watkins. Methodological holism is the belief that there are ‘macroscopic laws that are sui generis and apply to the system as an organic whole’ (Watkins 1952, p. 187), and the behaviour of its components had to be deduced from it. In economics, this would imply beginning our analysis by stating the laws of an aggregate economy and, perhaps, the behaviour of prices and industries and, from that, deducing how individuals behaved and what motivated them. Stated in these terms, it immediately becomes clear from a perusal of almost any microeconomics textbook that economics belongs essentially to the methodological individualist end of the spectrum defined by methodological holism at one end and individualism at the other.

After these writings, interest in the subject flagged. Social scientists, especially economists, continued to do research without trying to explicitly articulate the method that they were in fact using. The feeling developed among economists that the issue of methodological individualism was either trivial or had been resolved in their favour.

In the early 1990s the economists’ gathering insouciance was challenged by Rajeev Bhargava (1993) and Kenneth Arrow (1994). Bhargava summarizes various points of view on the subject and then challenges the orthodoxy, especially within economics. But he also expresses well the philosopher’s inevitable anxiety in a debate like this, which stems from not knowing whether what one is grappling with is something profound or trivial. As he writes, ‘On reading the literature one is swung between exuberance and despair, from feeling that all problems have been resolved to one that none has ... Gradually an intense frustration overwhelms the reader: perhaps there was nothing worth discussing in the first place. What on earth was all the fuss about?’ (1993, p. 5).

What he settles for as the best face of methodological individualism is ‘intentionalism’. The intentional man is somewhere between the imaginary homo economicus and equally rare homo sociologicus. He can choose and decide individually but he is not a relentless, maximizing agent. He has psychology and a sense of social norms, which get in the way of selfish maximization. Bhargava then develops the idea of ‘contextualism’ as a challenge to methodological individualism, including intentionalism. The challenge consists of arguing that a variety of beliefs and practices in everyday life make sense only in the context of the society where they occur. Hence, in describing a society or an economy we are compelled to use concepts which are irreducibly social.

The reason why the assertion that certain beliefs and concepts are inextricably social is unlikely to stir a hornet’s nest is that, although many economists claim to be rigid adherents of methodological individualism, they do use and have always used social concepts and categories. This is convincingly argued by Arrow. He points out how a variable such as price in a competitive model is an irreducibly social concept. Each individual takes the price to be given but the price that comes to prevail is an outcome of the choices made by all the individuals. So economists constructing equilibrium models, who claim to be hardened methodological individualists, are actually not so, at least in the sense that they use some concepts that are irreducibly social. Knowingly or unknowingly they follow a method which uses social categories. In fact, this was explicitly recognized by Schumpeter in his classic essay on methodological individualism, where he noted ‘prices are obviously social phenomena’ (1909, p. 217).

Preferences and Groups

There are more contentious claims that one can make about the role of social concepts in economics. One of these relates to the permissibility of a certain class of propositions in social science, such as: ‘The landlord will undertake action A, because it is in the landlord’s class interest to do so.’ (Action A could, for instance, be: ‘refuse to hire a servant who has fled another landlord’s employment and offers to work for this landlord for a low wage’). Let me call this proposition P.

The bone of contention between neoclassical and traditional Marxist economists is frequently whether such propositions are permissible. Many neoclassical economists and some political scientists (especially those belonging to the positive political economy school) believe that P is not permissible – a person’s class interest must be not be treated as an innate characteristic in the same way that his self-interest may be. A small group of writers maintain that even Marxism is compatible with methodological individualism and that class and other aggregate behaviour should, ideally, be built from individual motivations and preferences (Roemer 1981; Elster 1982).

In any case, whether or not proposition P is wrong, mainstream economics certainly considers it so. If an economist were to use an axiom like proposition P, she would usually want to first satisfy herself why it may be in the landlord’s self-interest to behave in a way which is in his class interest. However, this does not negate the use of beliefs and other concepts and variables which are irreducibly social. It is not clear whether a researcher who does both (that is, resists explaining individual behaviour solely in terms of its ability to serve group or class interests but uses concepts and beliefs which are inherently social) is a methodological individualist. But this is a purely definitional matter and of no great consequence. The important and contestable question is whether assumptions like proposition P should or should not be used. I take the view that it is best to avoid such assumptions as far as possible, without making that into a dogma.

There are some fundamental ways in which modern economics has moved further away from methodological individualism than merely by using irreducibly social concepts, like prices, and even without using propositions, like P. I here mention two. First, most models of economics make use of the idea of ‘rules of the game’. In Cournot oligopoly, firms choose quantities and then wait for prices to form. In Bertrand oligopoly, firms set prices and then wait to sell what the market demands from them. In most real-life situations, these rules evolve over time through intrinsically social processes. We may not fully understand what these social processes are, but few individuals will deny their existence. Arrow (1994) has emphasized this and also the importance of ‘social knowledge.’

Second, there is increasing recognition in economics that individual preferences are endogenous. They evolve over time and may be responsive to what happens in society at large. As Thorstein Veblen (1899) recognized, around the time when neoclassical economics was taking shape, human preferences for certain objects often depend on who else is consuming those objects. If a film star wears a brand-name shirt, you may be willing to pay more for that same shirt. If the elite likes a particular wine, then some people will acquire a taste for that wine; moreover, such people will be viewed by others as belonging to the elite because of their taste in wine. In other words, people often use goods to associate themselves with other people who use those goods (Basu 1989). These are obvious matters (though they were sidelined during the time of Veblen) and any economist whose ability to think is not damaged by excessive textbook education will recognize that these kinds of preference endogeneity exist. What is remarkable about Schumpeter’s (1909) essay is that he understood (admittedly in a somewhat inchoate way) that this recognition may cut into the methodological individualism of economics. He observed how, given the human tendency to conform to society, ‘there will be a tendency to give [each individual’s] utility curves shapes similar to those of other members of the community’ (1909, p. 219).

To see how this can ruffle methodological individualism, suppose that each person likes to wear jeans if more than 60 per cent of society wears jeans; more precisely, suppose that, if over 60 per cent of society wears jeans, each person is willing to pay for jeans more than the marginal cost of producing them; otherwise they are willing to pay less. This society will have two possible equilibria: one in which no one wears jeans and another (however revolting it may be to visualize this) in which everyone wears jeans. In models of this kind there is an interdependence between society’s behaviour and each individual’s preference. Once we recognize this, there is no reason to start our analysis by characterizing the individual. We may still do so through force of habit. But we could equally begin by considering a social behaviour postulate – for instance, that 50 per cent of people wear jeans. Then we work out how much each individual prefers to wear jeans (and so how much each is willing to pay for his or her jeans) and check whether the initial social postulate is sustainable. If it is, then we have found an equilibrium. If not (as in the above example), then the behaviour is not one that will prevail in equilibrium. This method is one of neither methodological individualism nor methodological holism. It is therefore evident that, as economics becomes more sophisticated, it is moving away from pure individualism towards this kind of a hybrid methodology.

Normative Statements

An interesting and unexpected area where methodological individualism is violated is in some of our normative statements. We often pass moral judgement on groups of people which cannot be reduced to the individuals in the group. Normative propositions of the following kind are common: ‘It is a shame that no one in your university does research on poverty.’ If you asked the person making this observation whether he was blaming you for not doing research on poverty, he would typically claim that he was not; in fact, he would deny that he was casting aspersions on any individual but criticizing the collectivity of individuals in the university. This amounts to an implicit rejection of individualism.

Methodological individualism in the context of normative statements like the above one has not been much analysed, but Ronald Dworkin has provided an interesting analysis. He argues that in situations of group responsibility it may be reasonable to personify the group. Thus, when a corporation produces a dangerously defective good but it is not possible to pin down the responsibility on any particular individual, we may need to treat the corporation as a moral agent and apply ‘facsimiles of our principles about individual fault and responsibility to it’ (Dworkin 1986, p. 170). And then, by virtue of the corporation’s responsibility, we may proceed to hold some or all of the individual members of the corporation responsible. This is interesting because it comes as close to Watkins’s ‘methodological holism’ as we are likely to encounter anywhere. Individuals are still essential units in Dworkin’s analysis but, unlike in standard methodological individualism, judgement of the group precedes judgement of the individual.

Dworkin argues that we unwittingly often use this method. This happens when we talk of the state’s responsibility for certain kinds of individual rights. Thus we talk of the state’s obligation to ensure that no one is assaulted by others. Moreover, we do this even before agreeing on how this responsibility is to be apportioned across various units and agents of the state, such as the police and the bureaucracy. Dworkin (1986, p. 171) points out how we discuss the community’s responsibility and ‘leave for separate consideration the different issue of which arrangement of official duties would best acquit the communal responsibility’ (emphasis added).

It is possible to criticize Dworkin’s line (see Basu 2000) by arguing that the personification of the corporation or the community has to be an interim construct. It will be sustained if we can then apportion the blame among the members of the corporation. If, however, we find that we are not able to spread the blame among the individuals in some reasonable way, then we may have to forgo our initial stand, which held the corporation responsible, or at least maintain that there is no way to take the next step of tracing the fault to individuals.

Interestingly, this brings us back to the kind of analysis defended in the case of endogenous preferences. And this suggests, once again, that what is needed for modern social science is neither holism not individualism but a hybrid methodology that, at least for now, lacks a name.

See Also