Contents (A) The Problem of Macroeconomics
"Almost two thirds of the General Theory is in fact devoted to microeconomics" (Frank, Hahn, 1985) (A) The Problem of Macroeconomics Generally speaking, macroeconomics differs from micro analysis in focusing on the dynamics of output and employment in the economy as a whole rather than on the competitive process which establishes the prices and output levels in different markets. The relationship between the two types of theory has been a controversial one throughout the history of economics. The main questions invoked in any appreciation of macroeconomic models often involve the conception of equilibrium. To a large extent, the points for comparison refer to whether or not prices are determined simultaneously with output (notably at full employment) and, as is more often of concern, whether movements in either prices or quantities (or both) will guarantee a restoration or at least initiate a movement towards an equilibrium with full employment. However, this implies, by its very premise, that there indeed exists some gravitation point and if, somehow, that gravitation force fails to pull the economy towards itself, then we must ask whether such a development is inherent in the system or due to some external imperfection. "Macroeconomic" theory before the mid-1920s was usually shared by three types of analysis: business cycle theory, growth theory and monetary theory. As we have noted elsewhere, there was usually nothing in these theories that was distinctively Neoclassical. They were theories shared by the Classical School of Ricardo, Mill and Marx. The Classical conclusions were simply adopted by the Neoclassicals without much thought as to whether they were compatible (the noble exception was Léon Walras (1874)). In the macroeconomics of the pre-1920s, there was no apparent role for Neoclassical price theory to play. In the late 1920s and 1930s, this story changed somewhat with the introduction of explicitly Neoclassical theories of the macroeconomy -- such as that by Irving Fisher (1930), Friedrich Hayek (1928, 1931) and John Maynard Keynes (1930) himself. Only then was there a significant effort to pursue an integrated approach to economic theory. Their "Neoclassical Macromodel" was really merely a simplified version of Walrasian general equilibrium theory. Now, as regards the general equilibrium theory developed by Walras (1874), equilibrium prices and output are indeed determined simultaneously. When translated to any market, this means that the theory of price and the theory of output are one and the same and full employment prevails. If any excess demands or supplies exist, following Walras' theory of tatonnement, we might expect prices to rise or fall in response. If, for instance, labor supply exceeds labor demand, we would expect wages to fall. However, expecting wages to fall and to have them actually fall are different stories entirely. Heroic assumptions about the properties of aggregate functions and the ugly implications of the Debreu-Sonnenschein-Mantel theorem all conspire against it. Nevertheless, the "Neoclassical Macroeconomics" of Fisher, Marshall, Pigou, Cassel, Robertson and other pre-Keynesian theorists, held stability, as represented by the tatonnement or some analogous process, to be true in the aggregate world. After all, what meaning can be given to an equilibrium, whether in the aggregate world or the partial, if it is not a point of gravitation? That they did not bother to question the conditions under which gravitation is ensured was not really a slip. The proposition of stability, they felt, was so intuitively tied up with the idea of an equilibrium that it would not make sense to speak of one without the other. After all, they noted, economists since Adam Smith have defined normal or equilibrium prices as the prices which competition brings about. Thus, equilibrium prices are themselves defined as a result of a dynamic process. This is not due to ideological intransigience or naivete but rather to theoretical sloppiness. Consequently, the Neoclassical macroeconomists argued that any failure to achieve an equilibrium must be solely due to the fact that competition, the force of economic gravitation, is somehow being prevented from doing its job. Whatever the market, they argued, disequilibrium was a result of wrong prices and, furthermore, this can only be sustained if competitive pressures are externally constrained. According to this, then, phenomena like involuntary unemployment, which is defined as an excess supply of labor, can only exist and persist if and only if prices are not adjusting to bring labor demand into equality with supply. John Maynard Keynes, in his General Theory (1936) is presumed to have changed this state of thinking. How? Essentially, he argued that the chain of reasoning followed by the Neoclassical macroeconomists was not self-evident. Specifically:
In other words, the establishment of a so-called equilibrium prices by competition is only half of the question. The other question is what determines the level of output and employment in an economy as a whole. This was, of course, a question which almost by definition is inadmissable in Neoclassical economics: the setting of equilibrium prices brings demand into equality with a given endowment, i.e. prices and quantities are determined simultaneously. Keynes had several objections to this formulation - which we shall review below. He also proposed his own system which sought to dissociate these two sides. The basic story is this: Keynes argued that it is quantities (and not prices) that adjust to eliminate excess demands or supplies. One may argue that prices also adjust, but their role is limited to eliminating excess profits - they have nothing to do with eliminating excess demand. At this point, this distinction seems innocent enough. After all, Alfred Marshall (1890) himself had quantities adjusting and Morishima (1977) shows that Walras clearly also allowed for it. However, it is in translating this proposition to the aggregate that the issue gets tricky. First, let us recognize that:
the aggregate excess demand for goods, i.e. the difference between aggregate demand (Yd) and aggregate supply (Ys), is merely the difference between investment and savings. This arises from simple definitions of national accounts: aggregate demand is defined as Yd = C + I + G + (X-M), where it is composed of consumption spending (C), investment spending (I), government expenditure (G) and the trade balance (X - M) while aggregate output, in turn, translates into income which is either consumed, saved or taxed away, Ys = C + S + T. Then, assuming a balanced government budget (G=T) and a closed economy (so the trade balance is zero, X - M = 0), then the result follows that Yd - Ys = I - S. In equilibrium, Yd - Ys = 0, excess goods demand is eliminated - or, alternatively, I = S, investment is equal to savings. The issue is what adjusts to bring I and S into equality so that aggregate demand is equal to aggregate supply? The Neoclassical macroeconomists had prices (more precisely, interest rates) adjusting to do so. Keynes opted for his output adjustment process so that output would adjust to bring savings into equality with the given investment. To say that aggregate demand is equal to aggregate supply may seem to indicate that all markets are in equilibrium. Keynes, however, removed the labor market from this aggregation. Or rather, he proposed that for a given level of output (determined by demand) there is, simultaneously, a given level of employment determined by technological conditions. Now if output can be anything (since aggregate demand, being independent, can be anything), then the level of employment may or may not be equivalent to full employment. Thus, if aggregate demand is low, then aggregate supply is low and, concurrently, employment may be below full employment. In other words, we can have unemployment due to insufficient aggregate demand. For the Neoclassical macroeconomists, this case is unfathomable. The reasoning is that unemployment implies an excess supply of labor. Now, by Say's Law, if anything is being supplied there must be a corresponding demand. If economy-wide demands are composed of goods demand plus labor demand and economy-wide supplies are goods supply plus labor supply, then it must be that by Say's Law (as defined by Lange (1942)):
were Nd is labor demand and Ns is labor supply, or:
so that if Yd = Ys, which is what is implied by the condition I = S, then it must be, by definition of Say's Law, that, Nd = Ns, i.e. there is no excess supply of labor, there is no unemployment. In other words, the Neoclassicals argued, aggregate goods demand may indeed fall and that may indeed bring aggregate goods supply down with it, but this implies concurrently, that labor supply has fallen too. But that does not mean that population has suddenly disappeared nor does it mean that no one is idle. Rather this implies that laborers have decided to work less and are thus supplying less labor to the market. There may be unemployment but it is entirely voluntary. There can be no involuntary unemployment by Say's Law. Suppose there was. In other words, suppose that Nd < Ns. By Say's Law, this implies that Yd > Ys, or, in our aggregates, I > S, there is excess investment. Thus, for the Neoclassical macroeconomists, the cause of unemployment is not insufficient aggregate demand for goods but rather quite the opposite: there is excess aggregate demand. Excess demand for goods rather than excess supply, overconsumption (or overinvestment) rather than underconsumption (or oversaving), is what the Neoclassicals considered to be the chief cause of unemployment. The Neoclassical macroeconomists, then, argued that to restore full employment, either the rate of interest would rise (choking off excess aggregate demand for goods) and/or the real wage (w/p) would have to fall (thereby increasing labor demand). The only cause of persistent involuntary unemployment, they felt, was if interest rates and/or the real wages were prevented from adjusting by some imperfection of some sort. This was the gist of the thesis of the Neoclassical macroeconomists:
Keynes's General Theory was thus geared to demonstrating two points. Firstly, Say's Law has to be abandoned. In other words, he wanted to show that it is possible that Yd = Ys (or I = S) and, at the same time, Nd < Ns. Secondly, he had to show that adjustments in the real wage will not cure the latter. This last element is implied by the structure of Keynes' system. Real wages are given according to profit-maximizing technological conditions at a given level of employment. Since the latter is fixed by aggregate demand then, the real wage is also fixed. Flexible real wages have nothing to do with employment since the direction of causation in the General Theory runs from aggregate demand to real wages and not vice-versa. If there is insufficient aggregate demand, there will be less output and hence less employment - and there is nothing that prices can do to cure this. Lower wages will not convince employers to higher more laborers since they will not be able to sell their output. Consequently, the economy enters into a vicious equilibrium. Workers do not get hired since there is low aggregate demand, but there will never be higher aggregate demand unless there is greater spending. However, workers will only spend more in aggregate if more of them are hired or wage income rises, but wages will never rise under unemployment and no more will be hired due to low aggregate demand. Thus, there will be no endogenous dynamic process which will bring one out of this state, therefore unemployment can be maintained indefinitely. There is, in effect, an "underemployment equilibrium". To break this unfortunate state of affairs, the extra spending (i.e. extra aggregate demand) must come from elsewhere, e.g. from higher autonomous investment or higher government expenditure. At this juncture, it may seem that the Neoclassical and Keynesian cases are incompatible. However, neither Keynes nor the Neoclassics believed this to be the case. Rather, they each believed the other was a special case of their own system. In Keynes's view, his theory was the "general" one, the Neoclassical case to be applicable only under certain conditions - full employment. In his own words:
But the Neoclassicals would have none of this. As Pigou writes, "Einstein actually did for Physics what Mr. Keynes believes himself to have done for Economics. He developed a far-reaching generalization, under which Newton's results can be subsumed as a special case." (A.C. Pigou, 1936), but they did not accept that Keynes's did the same. Quite the contrary, the Neoclassicals believed that Keynes' theory was merely an imperfectionist version of their own more general theory. As noted, the Neoclassical theory of employment is also, in circumstances of disequilibrium, a theory of unemployment. In this respect, if wages or prices or interest rates are "sticky" or if individuals are unresponsive to market signals that would otherwise move them into equilibrium, then we have a market "failure" and, by that, the seemingly "Keynesian" result of protracted unemployment. Therefore, the Neoclassical economists perceived Keynes' work as stemming from some imperfection in their own model. In this view, Keynesian situations were only special cases of the general Neoclassical model. Who is right? From the outset, this is not obvious because it is difficult to fathom whether Keynes perceived his situation as one of "protracted disequilibrium" or an "underemployment equilibrium". The difficulty lies in that both can imply the same situation. Strictly speaking, an "equilibrium" is a state of affairs where nothing changes or rather there are no incentives in place which will lead to endogenous changes. By this, then, we may, if we so wish, regard an equilibrium as a "rest point". A disequilibrium is presumably "not equilibrium" but it may be a rest point if the disequilibrium is maintained indefinitely. However, by our understanding, both could ostensibly describe the same situation. The difficulty lies largely in the uncertainty surrounding one issue: whether Keynes was looking at the long run or the long period. Strictly speaking the "long run" is really only a extensive amount of time, it is long in a historical sense. The "long period" has a more specific meaning: it is when all prices and quantities have adjusted to their "normal" or "natural" values. What was Keynes' view? His famous "in the long-run we are all dead" remark clearly reveals that he was more concerned with the long run in a historical rather than logical sense. Technically, the logical long period can emerge after a short or long amount of time - indeed, for some modern New Classical theorists, the long period is established in an instant. In contrast, Marshall (1890), who introduced the terms, believed that the "long run" and the "long period" coincided. In other words, he believed that a sufficiently long amount of historical time will allow prices and quantities to adjust to their natural levels. If Keynes had a long-period conception, then he was seeking to find a "underemployment equilibrium" and the issue is then to find the long-period conditions, starting from primitive data, which lead to such an equilibrium. If he followed the long-run definition, then our attention should turn to discussing the protraction of disequilibrium and the failure of adjustment mechanisms to bring us to the long-period, full employment equilibrium. Our premature instincts tend towards the belief that Keynes worked logically in the short period but historically in the long run. We ground our claim on his belief in the permanence of investment. If investment is conceived of in a Hayekian sense as a movement in the stock of capital towards an equilibrium level of capital, then obviously, in the long period, there cannot be investment. However, if we follow the Robinsonian conception of investment as "what entrepreneurs do" as a matter of duty or desire or what not, this is not immediately clear. However, there is no indication of stocks of capital actually changing in the General Theory. For all intensive purposes, capital stock is "given" and normalized out of the system. Furthermore, as one eye-witness points out, "Keynes was interested only in very short-period questions" (J Robinson, 1962: p.80). It is interesting to note that different time references of short and long run were also behind much of the General Glut Controversy between Ricardo and Malthus. The coincidence is doubled since this is often regarded as a dress rehearsal for the Keynes-Classics battle over a century later. (B) Say's Law and Unemployment Keynes (1936) concentrated his fire on a motley crew he called the "classics". Who they were is not precisely made clear. Certainly Pigou, Robertson, Cassel and Edgeworth were suspects, and so, by inference, were Walras and Pareto. But was it fair for Keynes to also refer to his old master, Marshall? What about other economists, such as Wicksell or, for that matter, the Austrians, all of whom had reservations, at one point or another, about the structure of Neoclassical macroeconomic theory? Or what about the Classical political economists, such as Ricardo and Mill, whom had contributed only a few elements to Neoclassical theory but also held against involuntary unemployment? Clearly it is preposterous to lump all economists before Keynes as the "classics", but to concentrate exclusively on a handful of them, say Pigou and Cassel, would permit too many others to escape. Therefore, let us follow Sweezy (1946) in claiming that Keynes's critique was directed at those who maintained firm belief in the validity of Say's Law of markets. True, this will permit some to elude the Keynesian critique (perhaps deservedly, such as Wicksell) but we will have far greater bounty by nabbing others who might not be so obvious (such as Ricardo). Most, but by no means all, Neoclassicals accepted the general validity of Say's Law -- like Ricardo and Mill before them. As stated by Lange (1942), we can think of this way. Suppose we have n goods. An optimizing consumer's budget constraint, with non-satiation, necessarily means that the value of their demands for goods will be equal to the value of their supplies of endowments. Now, suppose we add up demands and supplies for the whole economy. Then we are effectively adding up budget constraints for all consumers - and thus end up with the conclusion that the value of total demands for all goods equal the value of total supplies of goods. This is Walras's Law. Thus, Walras's Law is obtained by adding up all budget constraints, and agents demand and supply things that are not necessarily "goods". Consider factors such as labor, for instance. Or money - that too is demanded and supplied. Thus, it must be that, in aggregate (and suppressing prices and the stock-flow distinction), I + Nd + Md = S + Ns + Ms, everything that is demanded must be equal to everything that is supplied. We can rewrite this as:
But what is Say's Law in this context? Recall that Say's Law states that total demand can never exceed total supply. Are we saying Say's Law and Walras's Law are one and the same thing? Not quite. Say's Law, as defined by Lange (1942) states that the supply of things other than money is equal to the demand for things other than money. Thus, Say's Law only requires that I + Nd = S + Ns at all times. But, by Walras's Law, that means imposing Say's Law is effectively the same as claiming that Md = Ms, i.e. that we have a monetary equilibrium. But since Say's Law applies at all times, then, by extension, Say' Law implies that there is monetary equilibrium at all times. There never is nor can ever be an excess demand or excess supply of money under Say's Law, regardless of whether or not the economy is in equilibrium. The absurdity of Say's Law was famously noted by Patinkin (1956). When Say's Law is imposed, he noted, there is no monetary theory - and indeed the very Quantity Theory ceases to work. Why? Recall that the quantity theory argues that increases in money supply ought to increase the price level. But how is that to happen? If money supply increases, the imposition of Say's Law implies that immediately money demand will rise to meet it - because there must be monetary equilibrium at all times. But for the Quantity Theory to work, excess money supply must exist which will lead to excess goods demand. It is this very excess goods demand which will lead to a rise in prices for all goods - and that rising price which then brings real money supply back down. But, with Say's Law in operation, there can never be an excess supply of money, nor can increasing the supply of money ever increase the demand for goods. Consequently, there is no mechanism by which prices rise. In fact, as we can immediately see, by imposing Md - Ms = 0 at all times (from Say's Law), then the price level is automatically indeterminate. We always have a monetary equilibrium, so it does not matter what the price level is. The Quantity Theory cannot work - indeed, no monetary theory can work - when Say's Law is imposed. Recognizing this (and calling for the abandonment of Say's Law) was the main consuquence of the Patinkin Controversy. So why impose Say's Law if it is so damning? The reasoning can be seen immediately from our equation. Suppose we did not have Say's Law, but only Walras's Law. Then, in this case, it is entirely plausible that a sudden excess demand for money (Md > Ms) can be offset not by excess supply of goods but rather by an excess supply of labor, i.e. unemployment (Nd < Ns). Or let us take it further, suppose that the excess demand for money is so great that we require both an excess supply of goods (I < S ) and an excess supply of labor (Nd < Ns). to offset it. Here we have unemployment and underconsumption (or underinvestment)! This the Neoclassicals could not allow. The Fisher-Pigou-Cassel system argued that it was overconsumption and not underconsumption that caused unemployment and yet here, having broken Say's Law, we have both! This should help us realize the centrality of Say's Law in pre-Keynesian Neoclassical macroeconomic theory. With Say's Law, (Md - Ms) is immediately set to zero so that all that is left is that:
so that unemployment (Nd This intimate connection between a monetary economy and unemployment was stressed by the early underconsumptionist writers, Foster and Catchings:
Say's Law and theories of money are incompatible. This was recognized as early as Knut Wicksell (1906: 159-60). But if Say's Law was abandoned, so that the Quantity Theory could work as we described it before, there are other resulting problems. Firstly, as noted, there is the Foster-Catchings possibility of overproduction and unemployment as a result of an excess demand for money. For another, there are problems regarding the homogeneity postulate of excess demand functions (that they must be homogeneous of degree zero in money prices). Some of these problems were corrected by Patinkin (1956) with his `real balance' effect. But, in 1936, the position of Say's Law at the heart of Neoclassical theory was essential - and this is precisely what Keynes set out to demolish. As Sweezy notes:
In short, Keynes's theory presupposes the existence of a monetary economy and thus presupposes the non-applicability of Say's Law. The (Neo)Classical system began with the proposition that under full flexibility of prices and interest rates, aggregate goods demand must equal aggregate goods supply (Yd = Ys) which, in turn, is the same thing as saying that savings equal investment (S = I). Say's Law, by prohibiting general gluts, in turn, implied that goods market equilibrium will ensure that the labor market clears. Keynes (1936) was faced with the function of breaking Say's Law so that, as explained earlier, we could have, simultaneously, a goods market equilibrium (S = I) with the labor market in excess supply. To do so, the system of reasoning about the economy would have to be rethought. Firstly, Keynes sought to cast this equality of savings and investment into doubt - or rather, sought to transform what had effectively been treated as an accounting identity (S = I) into a behavioral one as well. Such a manoeuvre is quintessential Keynes. Already in the Treatise on Money (1930), he treated his "fundamental equations" (which were also self-evident) in the same spirit. Now, one of the fundamental equations was that:
where P is the price level, E is effective demand, Y is output, I investment and S savings. This "behavioral" equation arises from a series of identities and it is really only a transfiguration of the MV = PY equation of exchange. In this equation, (I - S) essentially refers to "windfall" gains. Now, I and S were assumed by Keynes (1930) to be functions of the rate of interest in Wicksellian manner. So, if the rate of interest equibrated S and I, then the equation reduces to:
However, Keynes had no explanation for E/Y. The interesting thing to note is that at this point, S = I, but effective demand (E) need not necessarily equal full employment output. In fact, E could very well be below Y (i.e. insufficient aggregate demand) even though there is a capital market equilibrium (S = I). One of the purposes for the initiation of his thesis in the General Theory was precisely to explain what determines the level of effective demand, E. One does not need to look too far back to realize where the spirit of transforming identities into behavioral equations came from: Alfred Marshall and the "Cambridge" tradition. It might seem odd for us to present this proposition here given that Marshall and the Cantabrigians were explicitly named by Keynes (on page 3) as the "classics" against whom the General Theory was directed. But there is sufficient evidence to indeed propose that while Keynes might not have sought to reinstate Marshall at the helm of economic theory, he did attempt to reintroduce some crucial, overlooked Marshallian conceptions into general economic theory. That, in the process, he happened to stumble upon a manner by which to stretch economic theory itself is but a tribute to his scholarship. Indeed, we shall argue that the attempt to reinstate the Marshallian notions of equilibrium in a general context without assuming away such vital concepts of behavior and historical time, might perhaps have been the most clear purpose underlying both the creation of the Treatise and the General Theory (although by no means their only purpose) and indeed as far back as the Tract (1923) during the conversion, jointly with Marshall, of still another identity into an behavioral equation: the Cambridge cash-balance theory. Time, in particular, is crucial for any treatment in the Marshallian vein. A review of most Walrasian models should reveal how time gets essentially removed from any operational significance. Without time, there is in fact no need for the concept of a gravitational equilibrium which is anything less than instantaneous. In the pure Walrasian context, everything which would otherwise be considered a "noise" point by the other classics is now part of the "equilibrium" configuration. General disequilibrium in the Walrasian world is not something that is observable: it happens before exchange (and hence real time) begins. Effectively, if we take a strict interpretation, at any point in real, historical time, the economy is already in equilibrium. Marshall (1890), of course, was far more cautious than this. Primarily, he allowed for disequilibrium to exist in the "real" world. He spoke of markets adjusting "over time", actual prices and actual output gravitating around market prices and output. Everything "tended" towards something - usually attained only in the logical long period - equated, by Marshall, with the historical long run. Marshallian equilibria were not, as in Walrasian theory, attained instantaneously but rather, were the result of a painstakingly long process often attenuated by uncertainty and recurrent shocks to the conditions which established them. However, all these concepts applied to single markets in Marshall - or at least to a subset of markets. Nowhere did Marshall systematically extend this to a fully general system of interlocking markets. As such, then, Walrasian conclusions of general economic stability were absent in Marshall's world. However, there was a group of economists in England who believed that such general conclusions could be grafted onto the matrix of Marshallian economics. Most notable, of course, was Arthur C. Pigou. Through most of his work, Pigou had drawn aggregate generalities into Marshall's economics, creating, in the process, what Hicks (1976) calls the "other kind of macro-economics". Samples of this exercise are found prominently in most of his major works. What Keynes seemed to do in his work was to show that Pigou had gone about it all wrong. He seemed to have been arguing that if one wishes to draw Marshall logically from the particular to the general, one does not reach Walrasian-style conclusions as Pigou had done (Keynes, 1936: p.259-60; also Keynes, 1933, 1937). This misses the entire richness and realism of Marshall's analysis - not to mention rests in complete ignorance of Marshall's ideas of fluctations in effective demand (Marshall, 1890:p. 591) and multiple and unstable equilibria (Marshall, 1890: p.665). What Keynes perceived, then, was that the General Theory, rather than Pigou's Theory of Unemployment (1933), was the accurate conclusion to be reached from any such extension Walrasian conclusions can only be derived from Walrasian equations. If one begins with the type of "realism" embedded in Marshall's system, i.e. time-consuming production, uncertainty, money, expectations, out-of-equilibrium exchange, there is really no guarantee that the Walrasian conclusions will hold. Of course, the verity of this is undeniable, as decades of research into general equilibrium theory have proven. As stressed by a particularly famous Neo-Walrasian couple:
However obvious this is in retrospect, one must admit that none of the Pigovian Cantabrigians had Arrow or Debreu handy to spell out the conditions of existence and stability of equilibrium - conditions which we know today are not met in systems plagued by the type of "real world" details of Cantabrigian economics. How, then, could Keynes be sure that Pigou's extension was logically inconsistent? It is probably true that Keynes could have guessed it although, of course, he had nothing more than a hint that it would not work. The hint was that Pigou's extensions of Marshallian partial economics to a general system neglected completely to discuss items which are irrelevant in partial systems but absolutely essential in aggregate systems, notably, the determination of the general level of employment and output. Without these concepts, then not much can be said at the aggregate level as we then have a "fallacy of composition". As Keynes writes:
It is this lack of apparatus on the part of Pigou and company that is crucial. The Pigovian style of general economic reasoning can be characterized as merely "gluing" partial markets together. There are cross-effects between markets, certainly, but within each market, the Neoclassical market-clearing apparatus works almost in isolation. Now, this apparatus rests on the idea that prices clear markets where aggregate output is pre-determined as data in the model. Now, if one were to construct a macroeconomic theory, by which we mean a theory which purports to explain how output and employment are determined, then it is obvious that such a construction will not do. Keynes claimed that the general level of output enters as data in Pigou's system, therefore, he cannot claim that it is endogenously determined by the system. Consequently, the Pigovian Neoclassical theory has "nothing to offer when it is applied to the problem of what determines the volume of actual output as a whole" (Keynes, 1936: 260). Why should this matter? Why not admit that the general level of output and employment is pre-determined and get it over with? Well, for two reasons: one, it is incapable of answering crucial macroeconomic questions and, second, it can hardly be a "general" theory if nothing explains where the data comes from. The first objection is obvious. The result of assuming that output (and employment) is given and the price level irrelevant is that such a theory cannot explain fluctuations in output, employment and the price level. In short, it does not explain the important phenomenon it was presumably constructed to address. The result, then, is that Pigou's extensions disallowed the endogenous existence of "general" phenomena such as unemployment, inflation, slumps and booms, all of which are an integral part of economic reality. These situations are not permissible in Neoclassical theory in a fully-working system (supporting the old jibe that "economists, unlike other scientists, find empirical phenomenon and wonder if they can work in theory"). In the rigid Walrasian world, and in Pigou's slightly- less-rigid world, there is no theory of fluctuations. Rather, there is an "untheory", their existence entirely justified as a result of imperfections of one sort or another in specific markets. Such may not, in itself, be an illogical proposition. However, it should force us to reconsider the question of what an equilibrium consists of and whether observable economic activity is amenable to such a paradigm. This situation, perhaps more than anything, discomforted Keynes. Surely, one cannot accept such an "untheory" of fluctuations and most certainly it is not much of an accurate reference for reality. This was an old concern of Marshall's for as Keynes notes, "Marshall...arrived very early at the point of view that the bare bones of economic theory are not worth much in themselves and do not carry one far in the direction of useful, practical conclusions." (Keynes, 1933: p.170). However, this is an objection immediately accepted (and even proposed) by modern general equilibrium theorists themselves (e.g. Hahn, 1973, 1974). However, Pigou and the other Neoclassicals were not as prone to concede the point. Instead, they believed that their hybrid model was an explanation of the real world phenomenon of unemployment and macrofluctuations. Keynes regarded this as evidence of poor scholarship. To take an analogy, the explanation of military wars as an "absence of goodwill" does not describe how they come about - no self-respecting political theorist would let himself get away with that. Similarly, the existence of such economic phenomenon does not reveal so much a "market failure" as it does a "theoretical failure" in explaining them. The theory of perfectly competitive exchange is interesting but cannot itself provide any basis for a theory of fluctuations (although, one must admit that the original Walrasians had intended to extend it there). Therefore, one cannot employ it to describe them, as Pigou had most clearly attempted to do. Keynes regarded underlining the inadequate treatment of such phenomena on the part of the Neoclassicals as central to his own thesis. However, and it cannot be stressed often enough, the General Theory is not an instruction book. It is not geared to merely recommending what should be done in any of these situations by a governing body - however influential that might have been as an impetus and however widespread its later interpretation. Rather, as Minsky (1975) stresses, the General Theory lends itself to the not-so-simple task of describing how such phenomena come about - not as "untheory", but as "theory" proper. At least in this respect, Keynes has a remarkable methodological ally in the modern New Classical theorist, Robert Lucas (1987: p.48-53). To this end, Keynes sought to define a new general system to stand against Pigou's own synthesis. The difference was that instead of appending Marshallian concepts to what was essentially a Walrasian general system of markets (as Pigou had done), Keynes proposed to begin from a Marshallian environment: we take off from a "real" world, we have time, capital, money, uncertainty, expectations, etc.; given that, what can we say about the workings of general systems of markets? To this system, he had to perform the ritual inquiries: whether it is stable or not, whether it is unique, how it is determined, how it is supposed to look like and, once the rudiments of that are established, how it explains erratic phenomena such as unemployment and cyclical fluctuations. The second objection to the Pigovian extension was that it does not make up its mind whether output is determined by the model or a datum in the model. To understand, let us recall the foundations of the Neoclassical system in a multi-market system. By this, we mean that different, conflicting objectives of agents could be reconciled - the crucial word here being "conflicting", which implies that the problem is one of finding an allocation of resources on the basis of an initial set of conditions. Since we have an allocational problem, we are essentially referring to an allocation of a given set of endowments - and, through technology, outputs. The data, then, are the preferences of individuals (by which to identify and weigh the "conflicting objectives"), technological possibilities (to convert factor services to consumable goods) and the initial endowment of goods and services and their distribution among agents (which will give us exchange as the operator of the "allocation" problem). The crux of Neoclassical theory, of finding the set of prices which reconciles different objectives, confers these values on the basis of nothing other than scarcity of given factors and goods. If there were no scarcity of any good, then there would be no "conflict" in agents' attempts to fulfill their objectives. So, the entire Neoclassical theory of price rests on an assumption that endowment is given and all that follows is the question of how to allocate it. This is the data of Walras and he solved his problem on the basis of some quite primitive, individual level propositions concerning utility-maximization and substitution. These gave demand functions for individuals their familiar shapes. All that happens then, is that by a process of groping, demands are brought into equality with endowments. Now, the aggregate output of the Walrasian system is derived from the stock of endowments which were there initially (as Wicksteed reminds us, the demand for goods is indirectly a demand for endowments). There is no flexibility in aggregate quantity other than via changing labor supply. The mistake of Pigou, Keynes argued, is that he took this "foundation", the demand curve, and applied it to such highly aggregated markets to answer the questions they were not equipped to address. Among the important questions they sought was precisely what determined the level of output and employment. But the answer to this question is, in Keynes's view, already given by the theory: it is pre-determined. But surely Cassel and Pigou, trained economists in the finest of Neoclassical traditions, must have realized this, no? Perhaps. At fault lies not their economic beliefs but rather their methodological system. Before the Cassel-Pigou macroeconomics, there did indeed exist two classes of theorists which sought to explain such general phenomena: Business Cycle theorists and monetary theorists (mainly of the Quantity Theory variety). But neither the Business Cycle theorists nor indeed the Monetary Theorists based their models on the Neoclassical micro-system. Rather, their observations about the general economy were nested in inductive reasoning from empirical data rather than deductive reasoning from anything akin to Walrasian axioms. The Cassel-Pigou brand of macroeconomics certainly operated largely with the type of inductive reasoning that was embdued in Business Cycle and Quantity Theories, rather than the Walrasian type. Individual demand curves were aggregegated into market demand curves without much hesitation. The latter, in turn, were added together to give us aggregate demand curves, also without reservations. They then assumed that aggregate demand curves operated in the general economy much like market demand curves in a single market and individual demand curves at a much lower level. But the properties of aggregate demand curves are not analagous to individual demand curves - at least not without much exploration and explanation. The properties of individual demand curves, as noted, are derived from the idea of competing desires for a fixed set of factors. They are downward sloping for each individual reflecting each one's preferences and substitution possibilities. At the aggregate level, who is in "conflict", what is being substituted and whose preferences are being respected in order for demand curves to be downward sloping? And how can a fixed set of factors suddenly become variable? How does the data, the general level of output and employment, suddenly become something to be determined? The Business Cycle theorists, who had answered questions at the aggregate level, did not impose marginal, individualist properties on their system, thus they did not generally speak of "equilibria". Their objective was to describe movements in the most important data of the allocation problem. To describe it via the allocation problem, they justly felt was impossible. Thus, it is in principle true that for most pre-Keynesian theorists, aggregate, Business Cycle-type theories were wholly unrelated to the individual, equilibrium allocation theories. Nowhere is this dichotomy more visible than in the creator of the allocation question: W.S. Jevons presented his equilibrium allocation system in his 1871 Principles and his business cycle theory in his 1884 Investigations. The modes of thinking about the two problems were so completely unrelated that Jevons cannot bring himself to give an economic explanation to business cycles - relying instead on "sunspots" (no metaphor intended). Similarly for the Quantity Theorists: it is almost impossible to read Irving Fisher's Mathematical Investigations (1892) and his Purchasing Power of Money (1911) and come to believe they were written by the same person - the first consolidated the marginalist principles of allocation, the latter is merely an inductive discourse of a monetary appendage. So, one must not be blamed if upon encountering the Cassel- Pigou macrosystem, where the marginalist allocation principles are grafted upon a discussion of the determination of macroeconomic variables, one is grabbed by a sense of contradiction - as indeed Keynes was. But it is not so much that there is contradiction as there is a simple uneasiness in the manner which Pigou and Cassel just plug marginalist results into macrosystems without explaining or even wondering if they are compatible. Indeed, Cassel (1918) neglected entirely to explain the utility-maximizing foundations of his demand curves. Were these freely drawn from the air? Not quite. Anticipating Friedman, he justified this on the basis of the "empirical irrefutability" of the idea that price and demand were negatively related. No microfoundations were offered as to why this happens, no deductive reasoning was evident. It is simply assumed that these properties will be present. Pigou was as guilty of this type of reasoning as was Cassel, but given his work on welfare economics, he had less of an excuse. It is therefore no surprise that Keynes concludes (repeating our earlier quote) that
The result which Keynes argued was that Neoclassical theory was in need of a more "general" theory. This theory would explain the data (general levels of output and employment) which could then be plugged into the allocation problem of Neoclassical theory. It is in this sense that Keynes constructed a more "general" theory. One can suspect immediately that Keynes bit off more than he could chew. Or, rather, he took upon himself really two tasks: showing why the Pigovian system is wrong and then explaining his own system. One task was largely "critical", the other "positive". However, as we shall see, it was precisely the manner in which he shuffled the two tasks in the pages of the General Theory which made his central message so hard to decipher. The dual objectives of the book seriously marred the explanatory power of his thesis. With critique and contribution muddled together, comparing the "explanatory" power of the two theories became virtually impossible. There was really no discerning instrument. Thus, what later Keynesians extracted to explain real world phenomena were different amalgams of Neoclassical and Keynesian concepts. They did not fall either as Neoclassical theory proper nor as Keynes' theory. The lack of explanatory power and inconsistencies of such mongrel models was thus understandable. As explained, it was precisely Keynes' positive task, the theoretical foundations of his new "general" system, that yielded his best and most original contributions. It is for this reason that policy-effectiveness, an unfortunate modern synonym of the Keynesian revolution, should be disregarded as frequently as possible. Considering the Keynesian revolution from the angle of policy can be quite misleading for not only does it obscure his theoretical contributions but neither do the policy recommendations presuppose the theory nor the theory require the policy. In fact, as exemplified by Jewkes (1948), it is very possible to follow the Keynesian system without ever having to take any of his policy conclusions to heart. To take Keynes' statement:
At any rate, Keynes had already formed his basic views on policy which were not insubstantial and had been popularly disseminated after the First World War. Neither he, nor any of his contemporaries, including Pigou and, most famously, the Chicago School, needed the General Theory to understand interventionist policy implications or even to justify them - all one had to do was appeal to imperfections after all. But Keynes had a rather peculiar attachment to economics. He was not, as many economists later were, driven to the discipline as a "technician" with social questions. Rather, when 1936 arrived, he already had all the "social stuff" figured out - now, he only had some technical questions. These questions, as we have attempted to show, revolve around the ahistorical, moneyless Neoclassical system and its "untheoretical" explanation of far too much economic reality. He sought a theory with explanatory power and historical content, where personae are colored distinctively rather than taking on the mask of homogeneous, representative agents, where time is relevant and uncertainty prevalent, where the institutional structures, particularly that of the financial sector, are pivotal. In short, he wanted a vision of the economic world to explain what had previously been theoretically impossible - or rather, handled through "untheory". (D) Stability and Indeterminacy One way to understand the implications of Keynes's theoretical analysis is to concentrate on the issue of "stability" - as stressed by Leijonhufvud (1968). We need to recall that in Neoclassical stability theory, trade does not occur before the equilibrium set of prices is reached by either some recontracting process (from Edgeworth) or an auctioneer (from Walras). But in Marshall's system, stability requires time: with quantities adjusting to meet excess prices, production decisions are pivotal yet less amenable to instantaneous changes. This is perhaps where Keynes begins in full force. As surmised, Keynes drew Marshall's operational conception of equilibrium away from partial and into a general analysis - while, at the same time, avoiding some of the Neoclassical pitfalls which had already been documented assiduously by other economists such as Wicksell (1901-6). In order to keep the behavioral nature of Marshall's propositions, he was compelled to generalize without eliminating historical time, money and other important concerns. In order to keep historical time, he would have to permit out-of-equilibrium exchange otherwise, there would always be general equilibrium - which would destroy the need for time-consuming behavior and quantity adjustments as preached by Marshall. The Walrasian paradigm avoids the possibility of exchange out of equilibrium precisely because this would affect the conditions of economic agents and so change the nature of economic relationships. Exchanging endowments before equilibrium would change the purchasing power of agents and hence the "equilibrium" that was being groped for to the point where not much of an equilbrium can be said to exist. In essence, as Leijonhufvud reminds us, it is the groping process itself which had to be reevaluated:
The question that might be posed here is to discover which of the following two propositions is true: (1) whether these "out-of-equilibrium" movements are of a temporary nature and that the changing conditions will eventually work themselves out so as to draw the whole economy more closely towards some single (usually full employment) gravitational equilibrium or (2) whether these out-of-equilibrium exchanges so alter the conditions of the agents, both from the demand and supply-side, such that a unique gravitation point can not really be deemed to exist (i.e. there is a continuum of possible equilibria). Marshall (1890), surprisingly enough, seems at times to fit the latter more than the former. In order to allow for "evolutionary" processes, economic relationships must be altered - and this is only possible by out-of-equilibrium exchange. But if this evolution is to be sustained, structural changes must change the long-run gravitation point - so much so that it cannot be certain to exist in a unique form. In Marshall's own words:
However, exactly what these changes were and how exactly they were brought about was a point which Marshall consistently tried to avoid - relying instead on short-term, partial analysis and defining the longer term in terms of secular trends. General equilibrium was too potent a hurdle and too dangerous to the rest of his system. At least for Marshall, albeit not for his upstart pupil. It can be argued that Keynes sought to recast the concept of equilibrium as a continuum rather that a single gravitation point - the particular choice of which depended upon a structural component he discovered to be effective aggregate demand. For this Keynes had to take several steps. The first was aggregation. This simple methodological procedure would permit Keynes to treat general equilibrium without the trappings of Marshall's partial analytics (namely, industrial structures which had proved to be, and still are, highly problematic) and steer clear of the dreaded conceptual consequences of Walrasian-style equations. In forging these aggregates, Keynes gave birth to a distinct type of economic analysis more amenable to the problems he wished to grapple with. However useful, his aggregation was not pivotal. What was crucial was his description of that elusive structure which would determine the short-run general equilibrium from a continuum of possible ones. But that equilibrium must not be trapped until eternity. It must be able to "evolve" if necessary (and possible), "always affecting and being affected by the character and the extent of demand" (Marshall, 1890: p.306). The driver of that change would be none other than the protagonist of Marshall's Principles: the Victorian entrepreneur and his private investment, in effect, the quantity-changer. This would provide the underlying determinism of the general system. In his criticism of Pigou he notes:
However, if Say's Law is kept, the entrepreneur's activity is constrained by the volume of savings and thus cannot be given the independence required to serve as the "driver" of the evolution of the economic system (a point anticipated by Knut Wicksell, 1906: p.159-60). It is for this reason, as can best be surmised from the propositions made, that Keynes reversed the direction of causality of the macroeconomic identity, S = I, and drove economics into a conceptual revolution - and, in the process bit his old master's hand. This perhaps unwarranted detour has only been an extended stab at that age-old concern of "what Keynes really meant to do". We say "meant", for it is actually more difficult to assess what he actually did do. The difficulty lies more in the "first-draft" nature of the General Theory and the lack of substantially clear and coherent post-1936 explanations by Lord Keynes himself (a second edition might have relieved much trauma). It is clear that while his intuitions are already spelled out elsewhere and he was probably certain of what he desired to achieve, he was generally thinking up his specific strategy as he went along - pursuing little insights here and there, some (or many) of which might be actually inconsequential to his thesis but which are nonetheless surprising and instructive. For the purposes of continuation, let us then follow the General Theory along the broad lines of his own strategy (a self-contained analytical account shall be saved for later).
|
All rights reserved, Gonçalo L. Fonseca