Anglo-French Conference
Disequilibrium macroeconomics

'Disequilibrium' macroeconomics is often seen as an alternative to neo-classical approaches, in which the prices of goods and factors of production are flexible and adjust markets in each period. French and British economists met in Manchester on 15-16 April, to evaluate the contribution of disequilibrium macroeconomics to our understanding of the world. Financial support was provided by the Economic and Social Research Council, the French Cultural Delegation and the Maison des Sciences de l'Homme. Papers presented to the conference covered 'sticky' wages and prices, the implications of imperfect competition and the incorporation of rational expectations, and disequilibrium macroeconomics in open economies. The Centre is currently planning a further, complementary conference in Spain in 1987, which will focus on the empirical applications of disequilibrium models.

'Disequilibrium' models differ from conventional (Walrasian) equilibrium models in their assumptions concerning price behaviour. Equilibrium models generally assume that prices of goods and factors of production adjust rapidly to equate supply and demand in each market. Prices in disequilibrium models do not adjust rapidly to clear markets, and individuals are unable to carry out their desired transactions in all markets. Individuals may be rationed in the labour market, unable to work as many hours as they wish at the current real wage.

For simplicity, disequilibrium models often involve only three goods: output, labour, and money balances. This gives rise to a variety of quantity-constrained equilibria, and these may be characterized according to whether there is excess supply or excess demand in the labour market and whether there is excess demand or excess supply in the market for output. The resulting equilibria can be characterized according to this pattern of excess supplies and demands, and four 'regimes' have been identified. When there is excess supply in both the output and the labour markets we have 'Keynesian unemployment': workers are unable to work as many hours as they would like, and firms do not produce more because effective demand is too low. If there is excess supply in the labour market and excess demand in the output market, the resulting unemployment is termed 'classical' and is often attributed to a level of real wages which is 'too high'. It is also possible that demand exceeds supply in both the output and the labour markets; prices are assumed to be rigid in the short run, and the equilibrium is described as one of 'repressed inflation'. Finally, in some versions of the model a fourth, 'underconsumptionist' or 'anti-classical' regime is identified, in which there is simultaneously an excess demand for labour and an excess supply of output.

The conference was opened by Peter Neary (University College, Dublin and CEPR), who presented a paper entitled 'Neo-Keynesian Macroeconomics in Open Economies: A Survey'. Neary examined the common features of models that applied 'disequilibrium' theory to open economies, in order to see which features of the models give rise to distinctly Keynesian aspects. Neary attempted to identify, within the simplest possible model, which assumptions generate distinctively 'disequilibrium' phenomena, which he associated with such traditional Keynesian effects as the multiplier and the perverse effect of wage cuts on output and employment.

Neary compared a Walrasian model of a one-good, small open economy with the same model under a fixed-wage assumption. He also assumed that labour supply was fixed and that profits were distributed immediately, so as not to pre-dispose the model to give Keynesian results. In general, he found that Keynesian results required the interaction of two constraints; the expedient of appending a fixed-wage assumption did not of itself generate specifically Keynesian effects. The existence of rationing in at least two markets was an essential feature of models which produced 'Keynesian' results, Neary argued. Early work in the area, following the lead given by Dixit, assumed that the goods market was cleared by foreign trade. Any unemployment must then be of a classical nature, essentially caused by real wages which were 'too high', since firms would not be constrained on the goods market. The introduction of non-traded goods will not fundamentally alter this conclusion if the market for these goods also clears. Interestingly, government expenditure on non- traded goods can affect employment in such circumstances, but it does so for the eminently classical reason that by raising the price of non-traded goods, it effectively reduces the real wage facing employers producing such non-traded goods. In summarizing a lengthy paper, the main conclusion was that Keynesian results could only emerge from a model that allowed all agents to be constrained, and this required rationing on at least two markets.

Sweder van Wijnbergen (World Bank and CEPR) argued that many of the results in the area depended on the assumption that decision- making was static in nature, and that many Keynesian results disappeared as soon as one considered even simple two-period models. Neary agreed, but so far only for special cases, involving perfect foresight for example. According to van Wijnbergen the role of capital-market imperfections offered a more interesting focus for research than did imperfections in the goods market. Robert Nobay (Liverpool University) suggested that the recent work on contracts could be integrated with the theory that Neary surveyed, since this would have implications for intertemporal behaviour. Jean-Pascal Benassy (Centre des Etudes Prospectives d'Economie Mathematique Appliquees a la Planification (CEPREMAP), Paris) argued that the fix-price assumption introduced by Neary should be dropped, even in the short run. Prices were not really fixed, but instead failed to reach their market-clearing levels. The explanation of this was an urgent task, not only for 'disequilibrium' theory, but for economic theory in general. The discussion centred around the difficulties involved in formulating a satisfactory theory of price formation, a theme that recurred throughout the conference.

The next paper, 'International Transmission of National Disturbances' was presented by Sweder van Wijnbergen (World Bank and CEPR). He pointed out that the standard framework for the analysis of open economies, the Mundell-Flemming model, implied that monetary shocks in one country had a negative spillover effect on other countries. This implied, for example, that a money-induced depression in the United States would stimulate the economies of European countries. Van Wijnbergen argued that this was implausible. He proposed as an alternative an intertemporal 'cash-in-advance' model, in which individuals needed cash balances in order to purchase goods. Market operations were sequential, in the sense that the goods and money markets were open on different days.

Markets failed to clear in this model, because of this 'cash-in- advance' constraint and because firms were assumed to set prices before production occurred so as to maximize their stock-market valuation. Since there are two constraints, the model satisfies Neary's criteria for a true Keynesian model. Van Wijnbergen completed the model with an assumption that exchange rates responded to the relative changes in national money supplies. The model generated constrained equilibria which could fall in three regions. In one region there was a constraint on goods supply, in the second a constraint on money balances and in the third the expectation of lower prices in the future led to larger money stock holdings in the present.

Van Wijnbergen then examined the spillover effects of a monetary disturbance in this model. Spillover effects would be negative if consumption could not easily be switched between periods, but positive with high intertemporal substitution. The Mundell- Fleming result, therefore, turns on the size of the intertemporal substitution effect.

Robert Nobay wondered whether van Wijnbergen's results depended on the sequential structure of the model, since individuals could not change their consumption plans and asset holdings simultaneously. It was also pointed out that changes in monetary policy did not alter budget constraints because of the sequential nature of market openings. Peter Neary commented that the assumption of monopolistic competition seemed to play only a small role in the workings of the model. It was also pointed out that the price-setting rule was relatively unimportant to the result, although it would be significant for fiscal policy analysis.

In his paper 'A Diagrammatic Introduction to Disequilibrium Macroeconomics', Paul Madden (Manchester) presented a simple short-run model of macroeconomic disequilibrium which allowed an easy diagrammatic analysis of quantity rationing. Madden used a simple model with three goods, output, labour, and money balances, within which a representative firm maximizes a well- behaved production function, workers supply a fixed quantity of labour and the representative consumer maximizes a utility function with output consumed and real money balances. Instead of assuming that wages and prices adjust quickly so as to establish a Walrasian equilibrium, Madden assumed that they remain 'stuck' in the short run at values which do not allow markets to clear. There is no guarantee that demands and supplies will be equal, and in attempting to trade, firms and households will discover quantity constraints on the amounts they wish to transact.

There are a number of concepts of quantity-constrained short-run equilibria: Madden adopted a variant of the 'Benassy equilibrium' in which decisions were taken in a sequential manner. Workers wish to supply a fixed quantity of labour, irrespective of the real wage. Trade only occurs in the goods market once employment is known, and firms' output decisions are taken on the basis of the level of demand they expect for their output.

Madden's model gave rise to the four regimes familiar in the disequilibrium literature. These corresponded to the various possible combinations of excess demand and excess supply in the goods and labour markets. Madden demonstrated that disequilibrium behaviour within this model could be illustrated by a series of simple diagrams. He showed, for example, that the model would generate a unique equilibrium if and only if the marginal propensity to consume (MPC) is less than one.

Participants agreed that Madden had encapsulated the basic disequilibrium model in a convenient and revealing way. Could the analysis be extended to a situation where trade was assumed to occur simultaneously rather than sequentially? The discussion which followed focussed on the likelihood of an MPC exceeding one. Peter Neary pointed out that this might only be plausible in the absence of an intertemporal approach. Hamid Sabourian argued that a more plausible starting point might be a marginal propensity to consume leisure which was greater than one.

Economic theory suggests that if there are unemployed workers who are willing to work for less than the current wage, wage levels will be bid down by the unemployed. Hamid Sabourian (King's College, Cambridge), in a paper entitled 'Wage Rigidity and Involuntary Unemployment: Some Exercises in Labour Market Analysis', examined various reasons why wages might not fall in the presence of unemployment. Sabourian introduced an overlapping-generations model, in which 'insiders', those with a job, developed norms of behaviour which deterred the entry of outsiders either directly, by imposing costs on potential entrants or indirectly, by imposing costs on the firm.

In one model developed by Sabourian unemployed outsiders do not undercut wage levels because their entry is blocked by harassment of recruits by the insiders. He demonstrated that there existed a wage level at which insiders would harass new recruits and outsiders would stay out because of the costs of harassment. This behaviour was able to support a wage above the level which would clear the labour market. In a second model, insiders deterred the firm from employing outsiders by threatening strike action. So long as the costs of the strike action meant that profits without entry of new workers were greater than with entry of new workers at lower wages, it was optimal for the firm to concede a wage above the market-clearing level. Similarly there existed a wage above the market-clearing level which justified the costs of strike action to the insiders. As a result there existed a wage which exceeded the market-clearing level and which would be supported by the behaviour of both the firm and the insiders.

In the discussion of Sabourian's paper, Sweder van Wijnbergen pointed out that in the first story it was optimal to hire old workers who have no incentive to harass. This was agreed but it was pointed out that a variety of assumptions would negate this strategy. Alan Kirman (Groupe de Recherche en Economie Quantitative et en Econome{^p^h2}trie (GREQE), Aix-Marseille) suggested that the introduction of more than one firm would complicate Sabourian's model and might alter its qualitative conclusions. David Canning (Pembroke College, Cambridge) pointed out that these models typically generated multiple equilibria.

The notion of 'voluntary trade', that no one can be forced to trade more than they wish, is an important feature of rationing models. In his paper, 'Monopolistic Competition and Fix-price Equilibrium', Joaquim Silvestre (University of California, Davis) discussed how this condition could limit the range of possible equilibria in a situation of competition between producers of similar, but not identical products. Each producer has a 'perceived demand curve', which represents the change in demand the producer believes would result from any change in the price he sets. Silvestre discussed a range of examples that demonstrated that many of the approaches to price-setting considered by other researchers could be incorporated within his framework. He then discussed the technical conditions that the perceived demand curve must satisfy, and he showed that these conditions would ensure that any market equilibrium with rationing would automatically satisfy the voluntary trade condition. This result suggests that a wide range of price- setting behaviour can be incorporated into disequilibrium analysis.

Pierre Picard (Universite de Rouen and CEPREMAP, Paris) wondered whether the result applied to the case where the firm knew only part of the demand curve which it faced. Silvestre responded yes, since in this case the perceived demand curve would be the real demand curve in the region of the firm's current position. Picard then asked whether Silvestre could also demonstrate the existence of market equilibrium in the examples he discussed. Silvestre replied that demonstrating that trade would be voluntary in equilibrium was very different from proving the existence of such equilibrium. Previous research on such models suggested that general proofs were very complex, and he considered it best to explore each case separately.

Whether tax cuts or increased government spending provide the best means of pursuing full employment is at the heart of current UK policy debates. Neil Rankin (Queen Mary College, London, and CEPR), brought the first day of the conference to a close with his paper 'Taxation Versus Spending: A Disequilibrium Welfare Approach' (CEPR Discussion Paper No. 84, reported in Bulletin No. 13). Rankin used a disequilibrium framework, but he assumed that the money market is always in equilibrium. The success of alternative policies is assessed by their impact on the welfare of individuals, rather than on macroeconomic indicators: his conclusions therefore depended on the direct effect that government spending has on welfare. Rankin considered three cases, in which government policy is successively treated as 'waste', as a consumption good, and as an investment good. Since the government can finance expenditure by issuing bonds in Rankin's model, his conclusions also depend on whether budget deficits are permissible. Rankin concluded that while government spending is beneficial, tax cuts are a more efficient means of pursuing full employment. In general government spending should be set at a level which would be appropriate if there were full employment, with adjustments in tax rates being used to achieve this full employment.

Commenting on the paper, Sweder van Wijnbergen suggested that the analysis should take into account the distortionary effects of taxes. Geoffrey Harcourt (Jesus College, Cambridge) agreed in general with Rankin's conclusions and suggested that Keynes himself favoured policies that stimulated private activity over policies that relied on the direct effect of government spending on output and employment.

Policy analysis is often based on the notion that the price of a good is a satisfactory measure of its true resource cost to society. Pierre Picard (Universite de Rouen and CEPREMAP, Paris) began the second day of the conference with his paper, 'Public Policy and Shadow Prices in a Disequilibrium Model of an Open Economy'. If rationing is present, Picard argued, the price of a good will not reflect its true resource cost or 'shadow price', and this complicates policy-making. Picard discussed joint research with Claude Forgeaud (Universite de Paris I and CEPREMAP) and B Lenclud (Universite de Paris I). Consumers and firms calculate their 'effective' demands and in so doing produce a set of 'multipliers' that measure the severity of the constraints they face. Planners can use these multipliers to calculate a set of true shadow prices for the economy. Picard and his collaborators explored the complications introduced by shadow prices in applying cost-benefit analysis to an open economy, where there may be disequilibrium in the goods and labour markets as a result of short-run rigidities in wages and prices. Their model gives particular emphasis to foreign exchange constraints: inelastic foreign demand limits the quantity of exports, and the need to equilibrate the trade balance with a fixed exchange rate may lead to an equilibrium with unemployment.

Shadow prices can be computed by first calculating the quantities of labour services and importables that are directly or indirectly necessary to produce domestic goods, and then weighting these coefficients in light of the shadow wage rate or shadow exchange rate. These shadow prices allow planners to combine the criteria of profitability and reduction of disequilibrium in cost-benefit analysis. The valuation of increased employment takes into account the gap between market and shadow wage rates and must be added to the project's profit. Likewise, if a shadow exchange rate is used, foreign currency savings, weighted by the gap between market and shadow exchange rates, should also be added to the projected benefits. Shadow prices can also be used to appraise the welfare consequences of an exchange rate variation, in Picard's model. Additional foreign demand for exports should be evaluated in terms of the difference between market and shadow prices of exportables. These shadow prices can be used in cost-benefit analysis and are also potentially useful in the evaluation of policies aimed at reducing the severity of rationing. The calculation of the shadow prices requires fairly detailed information on the economic structure, but Picard presented an example applying the method to the French economy.

Peter Neary queried whether Picard's assumption that the domestic goods market cleared was reasonable. Van Wijnbergen argued that the government budget constraint must be explicitly introduced into the model. David Canning wondered to what extent Picard's conclusions depended on the assumption that production was subject to constant returns to scale, and he speculated that the optimal policy would change if there were increasing returns. Picard replied that this would make little difference since the quantity constraints dominated the short-run behaviour of the model. Van Wijnbergen wondered whether the assumed rigidity of prices was appropriate in an essentially long-run analysis. Picard pointed out that prices in the model were determined by technology, and that the only important rigidity was in the nominal wage. Kirman argued that the assumption of constant returns to scale was inappropriate if there were unused resources; a proper marginal analysis was necessary in such circumstances.

The next paper, 'On Price Making in Disequilibrium: A Bertrand- Edgeworth-Chamberlin Model', was presented by Jean-Pascal Benassy (CEPREMAP, Paris). He focussed on the problem of analysing price formulation, arguing that it was the single most important issue facing disequilibrium theory and indeed economic theory generally. The essence of the problem, according to Benassy, was not that prices are fixed, but rather that they do not reach their market-clearing values.

Traditional models of markets with differentiated products assume that firms maximize profits subject to a demand constraint, taking other firms' prices as given. This approach has the unsatisfactory feature that no (Nash) equilibrium exists as the degree of substitutability between goods increases. To explain this phenomenon, Benassy considered a model of an imperfectly competitive industry in which a range of firms produce goods that are partially, but not completely, substitutes for each other. Benassy outlined an alternative model, in which equilibrium is determined by two parameters, the degree of substitutability and the number of firms. He showed that for any degree of substitutability there exists a number of firms which would permit an equilibrium and that for any number of competitors there is a degree of substitutability guaranteeing an equilibrium. Benassy then presented a series of results that showed that under certain conditions there exists an equilibrium which will converge to the Walrasian market-clearing equilibrium as both the number of competitors and the degree of similarity between goods becomes very large.

Harcourt pointed out in his discussion of Benassy's paper that firms earned positive profits in the model and that therefore we should expect new firms to enter the industry. Benassy agreed, but noted that the analysis was at a preliminary stage and that the basics of the model must be properly understood before complications were introduced. Sabourian argued that the model should allow firms' behaviour to evolve over time in a more sophisticated fashion: the assumption of 'Nash' behaviour by firms was rather unrealistic. Benassy agreed that this was possible but again suggested that such developments would not be useful unless the mechanics of the simple model were fully understood. Silvestre suggested that a partial answer to the problem of making entry to the market endogenous would be to allow firms to suspend production in a given period and then to re-enter the market at a later date.

The next paper, 'Towards Disequilibrium Dynamics: A Model of Temporary Equilibrium with Stochastic Quantity Rationing' was presented by Alan Kirman (GREQE, Aix-Marseille). Kirman generalized the three-good disequilibrium model introduced by Barro and Grossman and Malinvaud, by introducing inventory holdings by firms, which resulted from a lag between production and sales. It is an explicitly intertemporal model, in which firms produce 'today' for 'tomorrow' and consumers can spend 'today' what they earned 'yesterday'. Agents form expectations one period ahead but expect no changes thereafter. Kirman argued that this accords with intuition. He also allowed wages and prices to evolve over the longer run. Rationing was stochastic in the model: individuals do not know in advance whether they can carry out their preferred trade plans.

Kirman's model treated decisions made over a sequence of time periods. Decisions in different time periods were linked by inventories and also by expectations concerning both future prices and future rationing. Kirman explored whether an equilibrium with rationing existed in this general model and, if so, whether it was unique. Equilibria could be assigned to the four regimes common in disequilibrium theory, corresponding to the various combinations of excess supply and excess demand for goods and for labour. Kirman found that, in general, equilibria are not unique, and in fact the same set of wages and prices can give rise to different outcomes which fall in different regimes. There seemed to be no intuitive and plausible way to limit the number of equilibria, and this makes the analysis of the dynamics of the system extremely difficult and unrewarding.

Harcourt suggested that individuals could be assumed to maintain short-term buffer stocks, which would allow them to delay their responses to shocks, and Kirman agreed. Harcourt also speculated upon the possibility of specifying the system of prices in some way, rather than solving the model in order to find them. Nobay was interested to know what would happen if everyone expected prices to reach their market-clearing levels in every period: it appeared that this would be of some help, but that the model would nevertheless prove very difficult to analyse. Van Wijnbergen questioned whether such an assumption would make sense in a disequilibrium model. Neary argued that the imposition of some structure onto the expectations functions might nevertheless simplify the model, and the discussion focussed on how this might be achieved.

David Canning (Pembroke College, Cambridge) then presented a paper on 'Money, Expectations and Employment'. Canning attempted to reexamine in a formal framework Keynes's advocacy of controlling unemployment via government demand management rather than by varying money wages. The model was a disequilibrium one, with wages fixed in the short-run first period but adjusting in the second period, and expectations were taken to be rational. The source of Keynesian phenomena was the effect of expected prices on activity, particularly investment, in the current period. Canning argued that this was an important theme in Keynes's own writings: the government attempts to influence expectations of future prices in order to achieve full employment today. Canning made a number of assumptions in order to simplify the analysis. Individuals are assumed to live for two periods, working only in the first period: this generates a demand for money as a store of value. The behaviour of the young in the current period will depend on what they expect prices to be in the next period. Canning found that equilibrium in the model was sensitive to assumptions concerning future expected prices; in particular the model can produce involuntary unemployment.

Changes in expectations due to informational shocks will be destabilizing, but the effects can be cancelled by the expectation of appropriate government action. Thus policy effectiveness is established in a rational-expectations framework without recourse to any informational or operational advantage. This follows, argued Canning, because for any given level of excess demand with a fixed price structure, there will exist expected future prices which will clear the market: government policy can be designed to generate this expectation.

Canning demonstrated this by considering an informational shock concerning the future productivity of capital. This will change expected wages and prices and generate a portfolio adjustment, leading to changes in investment and therefore output and employment. Government may choose a monetary policy which creates an expectation that future wages and prices will be steady, thereby cutting off any portfolio adjustment. In this context it is the rational expectation that the government will adhere to this policy which provides the leverage in the current period. The government is able to commit itself to these policies in a time-consistent way whereas the market cannot. This demonstration supported Keynes' intuition as to the efficacy of government intervention.

Commenting on the paper, Nobay was concerned that in his model Canning had eliminated wealth effects, which he argued were an important feature of the real world and should be incorporated. Canning replied that they had been eliminated because otherwise analysis of the model would have been very complicated. Canning agreed that a complete model should allow for a range of effects. His paper, however, focussed on the important link between expected future prices and current employment. All simplifications were made in order to bring this point out clearly, and their removal would complicate the model without altering the basic result.

Many less developed countries face limits on the external credit they can obtain, and it is often argued that this has adverse effects on output and employment. Arvind Subramian (St Edmund Hall, Oxford) presented the final paper of the conference, entitled 'Intertemporal Credit Rationing in a Disequilibrium Open Economy: Characterization and Consequences'. Subramian examined the effect of an exogenous constraint on the amount of external borrowing, using a two-period model with a traded and a non- traded good. In the first period the price of the non-traded good is fixed but in the second period the market clears. Profits are distributed in the period in which they are made, and a capital market allows the smoothing of consumption between periods.

The primary focus of the analysis was on the current account. Subramian argued that in general the introduction of a ceiling on external borrowing had the effect of making 'classical' models less classical and 'Keynesian' models less Keynesian. A wage increase can have favourable welfare effects even in a regime of classical unemployment, when the real wage is often considered too high. The introduction of a traded-goods sector into a disequilibrium model often causes the model to behave in a more classical fashion. In Subramian's model, however, we need to consider the effect that any policy measure will have on the availability of credit, because this will limit the availability of traded goods. Therefore a cut in real wages may have a net contractionary effect. To predict the response to any policy change, it was important to take into account the entire structure of the economy. In addition, it is shown that the behaviour of the model under credit rationing depends on the structure of the tradable-goods sector; there can be no general presumption that a credit-rationed model will behave either in a more Keynesian or in a more classical manner.

Neary thought that Subramian had identified an interesting avenue for further research. The possibility of 'perverse' responses in Subramian's model was perhaps not that suprising. Credit rationing was equivalent to a failure of the traded-goods market; two markets were therefore constrained and 'disequilibrium' characteristics emerged. In fact the results complemented Neary's introductory survey.

Overall, the conference identified a range of issues of common interest across the broad spectrum of work on disequilibrium theory. There was broad agreement with Neary that rationing in at least two markets was essential in order to obtain 'Keynesian' results. A satisfactory theory of price formation was also an urgent theoretical requirement. The conference was useful in bringing together a range of research in a difficult area and identifying some common priorities for future work.