It has been shown by Kaldor (1939, especially pp. 8-10) that the degree of price stabilizing speculation in a market depends upon two elasticities: (i) the elasticity of expected price with respect to current price, and (ii) the elasticity of speculative excess demand with respect to the difference between current and expected price. Speculation will generally stabilize the current price about the expected price, since a wide gap between the two gives rise to a counteracting pressure on the current price from the speculative excess demand. The smaller the first elasticity the smaller will be the fluctuations in the expected price resulting from fluctuations in underlying factors of supply or demand. The larger is the second elasticity the more closely will fluctuations in the current price reflect fluctuations in the expected price. Kaldor's analysis, like much of the more recent literature on speculation and stability (Friedman (1953), Telser (1959)) addresses the question of whether or not speculation will serve its traditionally cited role of ironing out some of the fluctuations in prevailing market prices that would occur in its absence. The question of speculation and stability has been phrased somewhat differently in the literature on general competitive analysis (Hicks (1939), Enthoven and Arrow (1956), Arrow and Nerlove (1958), Arrow and Hurwicz (1962), Arrow and Hahn (1971, 309-315)) where it is asked whether or not the presence of speculation makes it more likely that the process of market adjustment will converge asymptotically upon an equilibrium. In both of these branches of the literature the focus has been upon the first of Kaldor's elasticities, and the results have confirmed Kaldor's analysis. Putting the question either way, we may say (roughly) that speculation exerts a stabilizing influence if the elasticity of price expectations is less than unity, or if expectations are formed adaptively (Cagan (1956), Arrow and Nerlove (1958)), it exerts no influence on stability if the elasticity is unity, and it exerts a destabilizing influence if the elasticity exceeds unity or if expectations are formed extrapolatively (Arrow and McManus (1958)). The purpose of the present paper is to examine the importance of the second of Kaldor's elasticities for the convergence of the market adjustment process. In the context of a single market it is clear that if this elasticity is large enough the price-adjustment process will be stable, provided that the formation of expectations is not destabilizing, because, regardless of the slope of the non-speculative excess demand curve, a large enough elasticity of the speculative excess demand curve will imply a downward slope to the market excess demand curve. The central question of the present paper is whether or not the analogous result holds in the multi-market economy of general competitive analysis. The answer to this question is vital to the broader issue of the influence of speculation on stability, because the size of this elasticity can be interpreted as defining the existing degree of speculation. The characteristic feature of speculation that distinguishes it from similar activities, such as hedging or investing, that also may be influenced by future price expectations, is that speculation is primarily motivated by the expectation of capital gain, not by the desire to consume or otherwise transform commodities, to avoid risk, or to

This paper is concerned with evaluating the non-market-clearing (NMC) approach of Robert Barro and Herschel Grossman and others from a purely positive point of view. That is, it deals with the broad question of the extent to which the approach provides a theoretically satisfactory explanation of certain stylized facts characterizing the dynamic behavior of aggregate output and the price level. It does not deal with the important and difficult normative questions involving stabilization policy that are often associated with the approach. From this viewpoint the main strength of the NMC approach is its compatibility with the evidence that 1) fluctuations in aggregate output are closely (positively) correlated with fluctuations in aggregate demand, 2) output appears to respond with a much shorter lag than does the price level to changes in aggregate demand, and 3) changes in output are serially correlated from quarter to quarter. The main weakness of the approach is its failure to provide any satisfactory account of how markets are organized. For example, it offers no explanation of how prices are formed, beyond the crude hypothesis that they move in the direction of excess demands, despite the fact that the assumption that prices fail to respond quickly enough to clear markets lies at the heart of the approach. Nor does it explain why agents should be constrained to trade at these prices, even though these constraints are what ultimately produce the multiplier process of the approach. This inattention to the details of market organization also appears to be responsible for the curious multiplier, according to which an increase in aggregate demand, from an initial position of generalized excess demand or even of full employment equilibrium, causes a decrease in output-a prediction that threatens to undermine the compatibility of the approach with the positive correlation between aggregate demand and output unless some reason can be found why excess demand should be less common than excess supply. This shortcoming does not imply that the NMC approach is not useful for many purposes, nor that its predictions are inconsistent with the evidence (except for the predictions of the supply multipliers). But to be consistent with the evidence is not to explain it. What the approach lacks is a satisfactory theoretical underpinning that would at least make it consistent with the same notions of rational self-interest that underlie the rest of economic theory. This leaves us with the question of whether a satisfactory underpinning can be provided to the approach. In other words, can the approach be revised or replaced in such a way that the resulting theory contains a more satisfactory account of market organization, and explains the above mentioned stylized facts in a way that closely resembles the NMC approach. This question cannot now be answered with a great deal of confidence because no one has yet developed a satisfactory theory of market organization. However I think that an affirmative answer is likely, and that the key to developing the answer lies in recognizing that different markets are organized in different ways. In particular some markets, such as those for many labor services, personal credit, and heavy capital goods, are organized on a highly personal basis with individually negotiated contracts, whereas other markets, such as those for widely traded financial assets and for most consumer durables, are organized on a less personal basis by trading specialists like retailers, wholesalers, jobbers, brokers, and stock market specialists. The rest of this paper attempts to shed some light on the question of providing a satisfactory theoretical underpinning for the NMC approach by investigating how a market organized by such specialist *University of Western Ontario. I am indebted to David Laidler for helpful conversations on the topic of this paper, to Robert Solow for his critical comments, and to the Humanities and Social Sciences Research Council of Canada for financial support.