【正文】 ovides an inherently more general theory.(pp. 458) As we shall see later, these two types of demand play different roles in the market exchange process and hence the distinction turns out to be necessary.Constraints for Economically Sensible SupplyRelated Demand. Demand always means a financial payout and hence for an agent it is a loss in his wealth. Thus before he pays out, the agent must make sure whether his payout is economically sensible. Then what do we mean by an economically sensible payout or an economically sensible demand? For supplyrelated demands, two constraints must be posited. I. Financial Constraint: The agent must have financial resource to pay for his purchase, as demand.II. Expectational Constraint: The agent must expect that he can sell the related supply so that the revenue from this selling can cover his current demand. As to the financial constraint, where the financial resource es from is not important. It can e from either previous selling or from credit. In the modern credit system, this constraint seems to be somewhat easily satisfied. However this does not mean that the constraint does not exist. For example, ine identification is often required by banks to issue a credit card. Nevertheless this is only a necessary constraint, not sufficient. Even if a producer has sufficient idle cash, he will not demand those inputs used to produce the modity that cannot be sold. Similarly, a consumer will not plan his consumption demand without knowing whether he can (or cannot) or how much he can sell his endowments. We will find that traditional equilibrium analysis lacks a consideration exactly on this constraint.The Expectation as to Sales. We now focus our attention on the expectation constraint. This constraint itself means that before paying for the supply related demand an agent should expect that he can sell the related supply.. In this sense, forming an expectation, as we mentioned early (see footnote 1), may also be regarded as a selling activity. Then where does the expectation e from? To make things clear, we may define an expectation function,(1) .Above is the expected sale。s multiplier analysis can be understood as an approach to describe the generation process of market exchanges. There are two ways of this exposition: one I call the forward exposition and the other the backward exposition. Then two possible doubts will be addressed, which seem to be, in the view of many economists, unsatisfactory to the multiplier theory. A mathematical model will follow to show how Keynesian macroeconomic relation can be specified in my micro model of market exchange. Finally, the relation between this multiplier approach to other approaches on this subject will be discussed.II. Demand and Supply, Which One is the First?Consider an agent who es to an economy. He certainly wants to have two types of exchange: buying and selling. Which one should he have first? The traditional equilibrium analysis does not offer an explicit answer. Implicitly, the system assumes that agents make their demand and supply decisions simultaneously. Then how can this simultaneity be rationalized in practice? One first finds that this simultaneity is indeed rationalized in the t_tonnement process. Agents, in the t_tonnement process, are supposed to put forward their demand and supply decisions simultaneously at quoted prices. These demand and supply decisions are not subject to actual exchanges. Actual exchanges will not take place until the equilibrium has been reached. Yet outside the t_tonnement, this simultaneity can hardly be rationalized. First, a decision without exchange is economically meaningless. Second, agents cannot have two types of exchanges, demand and supply, to take place at the same time. Naturally there is a problem which one is the first?. One should note that this issue, which one is the first?, has long been suppressed in economic literature. Weintraub (1977, pp. 4) once regarded the simultaneity as one of the anomalies of neoclassical system, but did not provide a solution. Benassy (1982) was even entangled. In chapter 4, he distinguishes the time ordering of an agent39。27 / 28The Multiplier Process as Market Exchange ProcessA Contribution to the Micro Foundation of Keynesian MacroeconomicsGang Gong. Dept. of economics, Graduate Faculty, New School for Social Research, 65 Fifth Ave. New York, NY 10003. 093992. Feb., 1995The author is very grateful to Edward Nell, Willi Semmler, Duncan Foley, David Colander, John Eatwell, Paul Davidson and William Milberg for their ments and suggestions on the early draft of this paper.The Multiplier Process as Market Exchange ProcessA Contribution to the Micro Foundation of Keynesian MacroeconomicsAbstractTraditional equilibrium analysis has been incorrectly founded once an ordering issue is concerned. To circumvent this problem, the autonomous demand, which has been missed in traditional microeconomic analysis, has to be introduced into the system as a starting point of a sequence of market exchanges. Following this direction, Keynes39。s multiplier analysis can also be viewed as a description of the process through which market exchanges are generated. This consideration provides us a new theoretical framework. The way an economy operates is pletely different from the way described by traditional equilibrium analysis. Even the concept of equilibrium has to be changed. Yet the new theoretical framework provided here turns out to be much superior in the sense that many mysteries of traditional microeconomics automatically disappear and further, the Keynesian macroeconomic relation is exactly specified in a micro economic context.The paper is organized as follows. First, I will raise an issue, which I believe has long been suppressed in economic literature. This issue is cruc