Reconstructing Keynesian Economics with Imperfect Competition
Marris’s motives for writing his 1991 book stemmed ultimately from his Cambridge education in the 1940s, which made him a lifelong advocate of Keynesian theory as a basis for policies to improve the world. He had met Keynes, been taught by Kahn, and knew Joan Robinson. The subsequent attacks on Keynesian macroeconomics—as well as some of the defences—seemed to him misguided. In the early 1980s, he had started to write, with his then colleague Dennis Snower, ‘a world-shaking new textbook of macroeconomics with explicit microfoundations’ (Marris 1991: xvi). The collapse of that project led him to write his own book.
The central argument of the book is that Keynesian macro theory makes sense only when combined with the theory of imperfect competition. Only if firms set their own prices, rather than being price-takers in perfectly competitive markets, will the main effect of a fall in aggregate demand be to reduce total output and employment, which was Keynes’s key assertion, rather than to lower the general price level. Yet imperfect competition is completely absent from The General Theory, which is puzzling, since Kahn had played key roles in the development of both this book and Robinson’s book on imperfect competition. What kept the strands apart, Marris argues, were tensions in the triangle of personal relationships between Kahn, Keynes, and Robinson.
Marris reviews the passages in The General Theory on microeconomic mechanisms, which he thinks were hastily inserted at a late stage and should not be taken seriously (ibid.: 180). These passages mainly imply that the immediate impact of a fall in aggregate demand is to lower prices, as in perfectly competitive markets, which leads firms to cut output. This mechanism is guaranteed to work only if nominal wages are fixed, which Keynes denied (Marris 1998b: 228), and if wages are flexible it works only on fragile assumptions (Marris 1999b: 25). Perfectly competitive markets are also rare in the sort of modern economy that Keynes had in mind.
The type of market structure described by Robinson and Chamberlin, which Marris eventually came to call ‘imperfect polipoly’, provides a simple alternative mechanism that is both more realistic and technically more robust. A fall in aggregate demand shifts each individual firm’s demand curve inwards, inducing it to cut its output. The firm will not alter its price, which is set to maximise profits as a mark-up on marginal cost, unless there is a change in the elasticity of demand for its product (which there is no obvious reason to expect) or its marginal costs change (which might happen later if nominal wages fell).[1] Had the macro model of The General Theory been connected to this micro model of imperfect competition (a possible chapter structure is suggested in Marris 1997), Keynes’s book would have been far more internally consistent.
Marris explains in detail how this connection failed to be made. In a 1929 dissertation that he did not publish for 60 years, Kahn created the theory of imperfect competition before working closely with Robinson on her book (Marris 1992). Kahn was also the creator of the multiplier, and he worked closely with Keynes on The General Theory, providing extensive comments and possibly doing some drafting. But Kahn was not ‘prone to meta-theorising’ (Marris 1991: 183). Moreover, Keynes and Robinson did not get on, as Marris observed (ibid.: 187), and were competing for Kahn’s loyalty and affection.[2] Keynes had told Ohlin that he could not ‘see how on earth imperfect competition comes in’ (Keynes quoted in ibid.: 185, fn. 1). The person who did see immediately how it came in was Kalecki, in a 1936 review of The General Theory in Polish, but he never followed up, possibly for ideological reasons (ibid.: 188-197). Marris’s book includes many other perceptive memories of Cambridge economists and their relationships.
One chapter of Marris (1991) is a comprehensive and valuable review of alternative ways of modelling imperfect competition, divided into three parts. The first considers alternative approaches to ‘structure’—are there assumed to be distinct industries, and within industries are goods homogeneous or differentiated? The second part examines alternative formalisations of the competitive process based on conjectures about the reactions of rivals or on game theory (as a deeper rationalisation of conjectures). The first two parts are summarised in a decision tree for choice of model (ibid.: 152), based on observed circumstances. The third part compares assumptions about market entry for small groups and large groups of firms. A key conclusion is that other models of imperfect competition provide basically the same underpinning as imperfect polipoly for Keynesian economics, though they are more complicated.
The book’s exposition of its reconstruction of Keynesian economics is mainly in the form of a simulation model—the subtitle of the book is ‘a desk-top simulation’ and the picture on the front of the dust jacket is a screenshot of the results of a run of the model. The book contains many other screenshots, tracing the evolution of a key set of variables through time in response to specified shocks, much as in the output of many later dynamic stochastic general equilibrium (DSGE) models. Simulation modelling was not new, but this book was one of the first to come with a disk containing the model, for which Marris had spent two years writing the code (on an IBM AT), and with straightforward instructions to the reader about how to run the model on a personal computer. Variation in any or all of 18 key parameters permitted a wide range of experiments.
The economic content and workings of the model are explained in two chapters of the book, first in general terms and then in technical terms. Each firm in the economy is modelled separately and in detail. Firms take turns in making price and investment decisions on the basis of adaptive expectations of demand with a view to maximising long-term profits (modified, if the model user chooses, by managerial preferences for greater size). After each such decision, all firms revise their outputs in light of levels of unsold stocks and unfilled orders, and aggregate variables are recalculated.
In contrast to the treatment of firms, other aspects of the model are highly simplified, partly because of technical constraints on programming or computer memory. There is an aggregated household sector that earns income, buys goods, saves and supplies labour, and government spending. But there are no taxes and no markets for money or financial assets. The economy is closed and there is no technical progress. Subject to these limitations, the uses of the model are illustrated in the final chapter of the book with a series of experiments that explore the effects of fiscal policy, the relationship between the real wage and the demand for labour, and stagfla?tion. As a bonus for the reader, there is also a fascinating chapter on the brain, computers, and economics.
Overall, this is an astonishing book, packed with ideas, insights, and information. Its two main arguments—that Keynesian macro theory makes sense only when combined with imperfect competition, and that this combination was omitted from The General Theory for personal reasons—are highly convincing. But although it was reviewed, in some respects enthusiastically, in the Economic Journal and the Journal of Economic Literature, the book had no discernible influence on mainstream macroeconomics. A telling illustration mentioned by Marris (1998b) is that the book is not cited in a 1994 survey of imperfect competition and macroeconomics (Dixon and Rankin 1994), even though one of the authors of the survey had reviewed it a couple of years earlier.
The reasons for its neglect are partly substantive. As Marris knew from his review of recent literature (Marris 1991: 213-233), the microfoundations of macroeconomics had become a crowded field—the survey that omitted him contained about 90 references. More basically, his main technical contribution— adding monopolistic competition—was already a key element of New Keynesian (NK) models.
Marris was aware of NK models but rejected them because of what he saw as their wrong assumptions about the labour market. In his model the labour market does not clear, as assumed by US NK economists, and nominal wages are either exogenous or respond to excess demand in the labour market. Nor in his model do real wages vary with employment on NK lines. However, European economists were modifying the NK model with a treatment of the labour market that was closer to Marris’s thinking. The book’s references include Layard et al. (1991), the leading contribution of this sort, but it had appeared too late to be discussed in the text, and Marris never identified himself as a European NK.[3]
Another important reason for the book’s lack of impact is that its exposition was tied to a simulation model. As a result, some important variables and relationships had to be omitted for technical reasons. The model also rapidly became obsolete in terms of both software and hardware. Moreover, as Marris recognised in a review of the pros and cons of computer simulation (Marris 1991: 107-113), it is easier to communicate in the language of mathematics than in the language of simulation. To explain his model, he could not use either algebra or programming code and had to make up his own ‘local notation’ (ibid.: 234). All this Marris knew: that he nonetheless pursued the simulation approach is probably mainly because he ‘fell in love with the microcomputer’ (ibid.: 2).
A further shortcoming of simulation models, which Marris does not recognise in his review, is that despite their seemingly objective technical nature, each model is just an extension of the brain of its author. The variables that are included and excluded and the causal links reflect what the author believes to be important or unimportant, which, apart from a limited range of parameter choices, the user must take or leave unless she wants to rewrite the code. As the reviewers pointed out, Marris’s model contained his own view of how the economy works. The qualitative results of his experiments, though not all predictable, arose largely by construction.
As a technical contribution to macroeconomics, Marris’s book alas sank without trace, and there is no possibility of it resurfacing. It remains of enduring relevance, though, as a contribution to the history of economic thought— flagging and explaining the puzzling omission of imperfect competition from The General Theory. Some of its thoughts on policy also remain relevant. Marris emphasised that macromanagement is difficult, especially in a democracy, and argued that its technical aspects should be ‘defended from the pollution of ideology’ (ibid.: 292), which in the UK soon came to pass with the creation of the Monetary Policy Committee and later the Office for Budget Responsibility. Although by then too ill to get involved, he must also have been pleased by the revival of Keynesian policies and thinking after the crash of 2008.
- [1] This mechanism differs from ‘price stickiness’ (mentioned by Keynes at one point), which implies thatfirms want to reduce prices but for some reason cannot do so (Marris 1991: 229-230).
- [2] The book’s public disclosure of the personal relationship between Kahn and Robinson upset peoplein Cambridge, where it was common knowledge, partly because Austin Robinson, Joan’s husband, was still alive.
- [3] In Marris (1998b), he attributes neglect of the book to errors in mainstream thinking. Marris (1997)reiterates its theoretical argument, although with no mention of the simulation model, and Marris(1999b) summarises it as part of an endorsement of the continuing relevance of Keynesian economics.
