John Maynard Keynes and Methodology in Twentieth-Century Cambridge

The leading figure of this new generation was Neville Keynes’s son, John Maynard Keynes, who came to economics from the philosophy of mathematics. His explicit contributions to economic methodology were, as for Marshall, spasmodic rather than the kind of focused study his father undertook. Yet his distinctive methodology, as set out by Shackle (1974), Chick (1983: Chapter 2), and Harcourt (1987), has had a profound influence on PostKeynesian economics. The philosophical foundations of this methodology became a significant field of study, beginning with the path-breaking work of Meeks (1976), Carabelli (1988) and O’Donnell (1989). As Runde (1997: 240) points out, Keynes’s methodological position fell outside the positivist tradition which, as we have seen, was encouraged (however inadvertently) by his father. Nevertheless, many of Maynard Keynes’s methodological statements were expressed in the process of criticising the positivist mainstream, a phenomenon which has carried forward to modern-day heterodox economics.^[1]

Keynes’s earlier work on probability allows us to infer his philosophical approach when he came to economics, where his stated aim was not just methodological but epistemological: to ‘revolutionise...the way the world thinks about economic problems’ (Keynes 1935 [1973]: 492-493). In A Treatise on Probability, Keynes explored the grounds for belief as the basis for action under the general conditions of uncertainty which followed from his organi- cism. Where structures evolve and the interactions between their components evolve (i.e. the atomic principle does not hold and the physical or social system is open), probabilities cannot be quantified. But rather than falling into the philosophical dual of certainty/ignorance, Keynes explored the realm of uncertainty, outlining the mechanisms by which more or less reliable knowledge may be established with more or less confidence. Keynes was thus updating Hume’s epistemology as a resolution to scepticism with respect to reason on the one hand and the problem of induction with respect to an organic system on the other. This was Keynes’s own version of the middle way, in the tradition of Newton (and Hume and Smith), whereby the mind identified patterns from experience; the resulting generalisations were provisional in the face of an organic system and, for application (the purpose of theory), required close reference to that evolving reality (Carabelli 1988: Chapters 4 and 5). These generalisations about tendencies were to be reasonable even if they could not be demonstrated to be true.

Keynes explicitly challenged the prevailing positivist view of induction which involved gathering data on repeated instances (‘pure’ induction) as requiring empirical uniformity of nature. Rather he understood induction as being a logical process of employing negative analogy in order to uncover persistence in spite of difference. But Keynes departed from classical logic (which required certainty as to the truth of premises) and empirical logic (which required unambiguous facts). Keynes was thus diverging from the endeavours of Russell and (the early) Whitehead to build a complete mathematical system based on classical logic. His (human or ordinary) logic was more suited to conditions of uncertainty. As with Hume’s epistemology, Keynes founded his logic on common sense and convention and he employed ordinary language.^[2]

Just as the scientist organises observations according to prior conceptualisations and patterns, so the individual or group in society has to apply judgement to observation. Keynes was quite explicit about the ambiguity of evidence this entails. He referred to direct knowledge based on experience in the following terms: ‘Sensations which we may be said to experience, the ideas of meanings, about which we have thoughts which we may be said to understand, and facts and characteristics or relations of sense-data, or meanings which we may be said to perceive’ (Keynes 1921 [1973]: 12; italics in original). The theory of probability was thus subjective in the sense that the evidence brought to bear and its assessment in relation to other sources of knowledge involved judgement. But it was objective in the sense that anyone in the same circumstances and with the same understandings of the evidence would arrive at the same judgement.^[3]

When Keynes turned to economics, this epistemology had strong methodological implications (Chick 2003). First, he regarded economics as an art. While the father had privileged deductive theory as the core of the discipline, with values imported later and ‘unscientific’ methods only introduced at the stage of policy application, the son privileged the art of application at the core. For him, the requirements of practical application determine the methodology of theory development and, given the open-system nature of the subject matter, that methodology was pluralist. Ordinary logic required multiple strands of reasoning and evidence which could lend weight to argument, for economists as well as economic agents. Keynes was therefore highly critical of the monist methodology of positivism (O’Donnell 1989: Chapter 9).

As with Marshall, Keynes’s ideas were then reduced by mainstream interpreters to their mathematical versions within a static equilibrium framework. Also like Marshall, Keynes himself used mathematical formulations, but only as a contributor to the overall argument. Keynes applied the same logic to the testing of theory by means of econometrics, a logic which he made explicit in his critique of Tinbergen. His primary critique was of econometric analysis which requires an invariant structure; he argued that the onus should be on the econometrician to demonstrate that a particular case reasonably approximated a fixed structure, so that regression analysis was warranted. O’Donnell (1989) shows that, as with his objections to mathematics, Keynes’s objections were to Tinbergen’s specific techniques in relation to the subject matter, not to econometrics per se.

While rational economic man had increasingly taken hold as the axiomatic foundation of deductivist theory, Keynes was particularly critical of the fictional nature of the concept. In line with his organicist view of society, Keynes rejected an atomistic representation of economic agents. Rather he emphasised man’s social aspect, not least when it comes to epistemology and the reliance on social conventional knowledge. Here again Keynes pursues a middle way, made explicit in his movement away from the reason-emotion dual represented by Russell and D.H. Lawrence, respectively (Keynes 1938 [1972]). The methodological implications of this epistemology are most evident in Chapter 12 of The General Theory (Keynes 1936 [1973]), where he analyses the determinants of the macroeconomic variable which was central to his theory of effective demand: investment (Runde 1997). His objections to the rationality assumptions arose from philosophical argument rather than explicit evidence; consistent with his theory of probability, Keynes’s methodology centred on logic. Nevertheless, his philosophy was founded in turn on an organicist ontology which arose in an inchoate way from ‘deep background’ experience (Searle 1995). While Keynes did not seek out evidence to the same systematic extent as Marshall, his human logic was informed by experience.

Chick (2003) spells out further the connections between Keynes’s ontology and epistemology and his methodology.^[4] From the former he set out to develop an open theoretical system which was general in its openness; the certainty of the closed mainstream system was a special case of the more general open system, characterised by uncertainty. This system therefore had at its heart a ‘method of expectations’, formed under uncertainty, within different time frames. Time played a crucial role, understood in logical terms, referring to causal sequence, or in historical terms, both of which gave equilibrium meanings beyond the limited mainstream meaning of a solution to simultaneous equations. Further, because money had a special role in decisionmaking with respect to an uncertain future, monetary and real factors were to be treated in an integrated way. But, contrary to the reductionism of the mainstream approach, some of the analysis was to be conducted at aggregative levels without an explicit derivation from individual decision-making. In any case, again like Marshall, Keynes was conscious of the important role of the evolution (in historical time) of the social conventions and institutions which provide the framework for decision-making. Like Marshall he too was motivated to improve society. While he apparently separated off his social philosophy (consistent with Neville Keynes’s normative-positive distinction) in advocating that the ‘economic problem’ be addressed first in order then to promote the Good Life, solving the economic problem itself involved moral judgements, for example, about the need to prevent unemployment and the need for institutions to promote the public interest (Chick and Dow 2013). Economics was a moral science.

Even though Keynes’s methodology was misunderstood when interpreted from the prevailing positivist standpoint of the mainstream, there was nevertheless a Keynesian methodological revolution from that perspective (i.e. within a narrow understanding of methodology) (Dow 2010). On the one hand, Keynes sparked off the development of macroeconomic theory as a field distinct from microeconomics. On the other hand, he provided the related impetus to the development of large databases and econometric techniques to analyse them, and indeed provided an agenda for the development of econometrics in attempting to deal with the problems raised in his debate with Tinbergen. But Keynes had a more enduring impact among those who took his alternative methodology seriously, forming under the banner of PostKeynesianism with its distinctive methodology (Chick 1995). This approach includes Keynes’s typically Cambridge requirement for abstractions to take the form of simplification rather than fiction, with a view to allowing application to the more complex reality; for a focus on the passage of historical time with its implications for the uncertain basis for decision-making; and for aggregation to allow macroeconomic analysis distinct from microeconomics, paving the way for analysis of growth and distribution.

The literature on the Cambridge figures who contributed to this development is vast. Here we pick out just some of the important contributions specifically on methodology rather than the theory of Keynes’s followers at Cambridge (the Circus) who perpetuated and developed aspects of his methodological legacy. Joan Robinson (1962: Chapter 4) pinpointed Keynes’s departure from the mainstream approach in terms of a shift in the subject matter to the capitalist system as a phase in history, bringing to the surface moral issues (with respect to the free operation of markets) and time. Keynes’s followers added to this a shift away from concern with the short period, where income distribution and power relations could be taken as given. Many of his followers in Cambridge were influenced by Marx and his classical concern with distribution and economic processes over the long period analysed in class terms.

Among the Circus, Joan Robinson made the most explicit methodological contribution. Her early book, Economic Philosophy, had a particular impact in the 1960s when there was a receptive audience for a reflection on the foundations of economics. Her methodological stance shifted somewhat over her career (Harcourt 1996a; Salanti 1996). But Robinson, like Keynes, believed that economics should address real issues, and take account of institutions (the ‘rules of the game’), and therefore that theoretical assumptions be simplifications rather than fictions. As with Keynes, too, she challenged apparently- false assumptions on the basis of her ontology rather than specific empirical study. She also shared Keynes’s views about the uncertainty of knowledge and thus the impossibility of establishing economic laws. She made a particular contribution in her clear distinction between a logical-time framework, which aims to establish causal mechanisms, and an historical-time framework, which aims to establish irreversible processes such that equilibrium is a position to which a process tends in the long run.

While Kaldor (1972) too criticised the ahistorical mainstream version of equilibrium, he also contributed to the methodological approach to evidence, developing the notion of stylised facts (as clarified and extended by Lawson 1989). Given the open nature of the economic system, detailed facts reflect a variety of tendencies at work under particular circumstances. By abstracting from this detail, stylised facts can form the basis of hypotheses. Here we see a version of the interaction of the mind with observed experience as a way of addressing the problem of induction, in the Newtonian/Smithian tradition. The Department of Applied Economics was established at Cambridge in 1945 to provide the empirical material on which such abductive reasoning could build.

In some respects Robinson could be said to have adopted a position similar to that of Neville Keynes: what Salanti (1996: 292) refers to as ‘empirical apriorisim’.^[5] To the extent that she aimed her critique of the mainstream at the realism of assumptions and the internal consistency of its logic, she was implicitly accepting the underlying positivist methodology. This was perhaps most evident in the ‘capital controversies’, where Robinson challenged mainstream marginalist analysis of value and production, such that the controversy came to be epitomised for many by the ‘reswitching’ problem which demonstrated that capital and its return need not be inversely related (Harcourt 1972). The fact that the force of this critique did not seem to be felt by the mainstream illustrates well the insufficiency of empirical apriorism as a methodology. In fact, while the debate was widely interpreted in the terms of the mainstream approach, the differences over capital theory arose from different ontologies and epistemologies. Robinson’s views on method were in fact changing as she moved away from Marshall’s influence and pursued the methodological implications of analysing a dynamic process over time. She argued further that ideology was embedded in mainstream theory, contrary to professed mainstream methodology. While her efforts to strip it out have been seen by some to reflect a Keynesian view—of the John Neville type— that ideology-free theory was indeed possible, Pasinetti (2008: 218) rather emphasises Robinson’s position simply that ideological views need to be made explicit since economics was ideologically non-neutral.

Other key contemporary figures in Cambridge were Michal Kalecki and Piero Sraffa, who both came to Cambridge from different methodological traditions. Both influenced others at Cambridge and spawned schools of thought developing their ideas within their distinctive methodological frameworks. Both focused on issues of production and distribution in historical time, influenced by Marx. Kalecki drew on empirical evidence, filtered through his careful conceptual classification of variables, in order to establish stylised facts with respect to the short period (such as all wages being used up in consumption). Sraffa differed from Keynes and Kalecki in focusing on the long period and in presenting a formal mathematical system (rather than partial mathematical arguments). But, since the content was not marginalist, Sraffa’s mathematics was different from the mainstream calculus, employing the mathematics of the classical period (Velupillai 2008). While the Ricardian tradition might seem to depart from both Malthusian and Post-Keynesian methodology, Sraffa traced his concern with struggle over the surplus in historical time back to both Ricardo and Malthus (via Marx).^[6] This interpretation of Ricardo as not being so different from Malthus was distinguished by Dobb (1931, 1959) from Marshall’s version, which was arrived at through Mill (see Harcourt and Kriesler 2016).

The resulting strands of Post-Keynesianism differ in part in terms of how much of the analysis was amenable to formal mathematical expression (echoing the debates between Malthus and Ricardo). But others have used formal mathematical models of the long period which are more in the Marshallian tradition of specifying equilibrium in gravitational terms.^[7] There are nevertheless common elements to the methodologies emanating from the key

Cambridge figures which are in accord with what we have identified as a Cambridge methodological tradition. Thus, Pasinetti (1974: 43-44) identified common features between the Ricardian approach and Maynard Keynes’s methodology, while Dutt and Amadeo (1990) set out the common methodological ground between Post-Keynesians and the neo-Ricardians who followed the Sraffian approach.^[8] Arestis et al. (1999) likewise identified the coherence of Post-Keynesianism, in spite of its different strands, in terms of methodological approach. The rather vexed question of assessing differences relative to commonalities needs to be considered in terms of a pluralist, non- dualistic approach to epistemology. Phyllis Deane (1983) criticised Marshall’s and Neville Keynes’s search for synthesis, advocating instead methodological pluralism. Her history of economic thought followed the Cambridge tradition we have identified by focusing on methodology from the perspective of (analytical) economic history (Deane 1978). Like Robinson, she kept philosophical issues to the fore at a time when she could reach a receptive audience.

Geoff Harcourt’s history of economic thought also took a methodological perspective, charting the contributions of the Circus (see, for example, Harcourt (1972, 2006) and Harcourt and Kerr (2009) on the capital controversy). His own work has employed a pluralist methodology to theory development, governed by a close understanding of real circumstances and by policy concerns. He has made a particular contribution to methodology by articulating and justifying his ‘horses for courses’ approach whereby selections are made from a plurality of methods according to the problem at hand, rather than according to some internal requirement (Harcourt 1996b). In particular, it involves a continuation of the Cambridge critique of the mainstream requirement to express all arguments in terms of formal deductive mathematics. The ‘horses for courses’ approach contributes to the increasing focus in heterodox economics on pluralist methodology, not as anything goes, but as an ontologically grounded selection of methods (Lawson 1997a; King 2002).

• [1] See Gerrard (1997) for a comprehensive review of Keynes’s methodology.
• [2] This theme was developed by Shackle (1983: 116) for whom mathematics was insufficient, being toorestrictive a language compared to verbal argument: ‘[L]anguage at its full compass, where words arefingers touching the keyboard of a hearer’s mind’.
• [3] On Keynes’s debate with Ramsey over subjectivity, see Carabelli (1988: 96—97).
• [4] Consistent with our characterisation of a Cambridge methodological tradition, Chick (2003: 311)characterises Keynes’s theory of rational belief under uncertainty in terms of a ‘third way’.
• [5] The continuing influence of Neville Keynes may have been due in part to the fact the Maynard Keynesnever produced a systematic account of his position on methodology.
• [6] This reading of Ricardo implies that the methodological differences between him and Malthus were lessmarked than is implied by their own debate, or that the differences were more of degree than of kind,where the latter might characterise the differences between marginalist economics and Cambridge political economy.
• [7] See, for example, Harcourt’s (2006: Chapter 5) discussion of Marglin (1984).
• [8] Note that their title is Keyness Third _Alternative?