Contents
(A) The Multiplier To establish I = S as an equilibrium behavioral equation, J.M. Keynes (1936: p.61-65, Ch. 7) postulated that the economy, when adjusting towards an equilibrium position, did not necessarily simultaneously achieve a full employment level of output. So it might be useful to realize that the purpose of this portion of the General Theory was to postulate a new set of mechanisms which would show the determination of the general level of output and employment. This, it was hoped, would prove the possibility of market failure (or rather, the inability to reach a full employment output level) as a natural consequence in a fully-working economic system. Keynes's proposition centered around the idea that prices, normally the adjusting mechanism in the (non-Marshall) Neoclassical model, were not the equilibrating mechanism for excess demands. Output, Keynes claimed, deserved that title. To explain this position, Keynes borrowed Richard Kahn's (1931) multiplier and placed it at the heart of his new behavioral system where investment demand determined savings rather than vice-versa. To illustrate Keynes's (1936: Ch. 10) multiplier process, assume that there is an injection of investment into a system by the ubiquitous entrepreneur. Accordingly, investment will translate into income - a portion of which will be consumed and the rest saved. Therefore, assuming a marginal propensity to save of, say, 0.1, then a 10 unit investment injection will result, in the first round, into an increase in income by 10. Consequently, consumption will increase by 9 and savings by 1. It seems, therefore, that savings (1) are in fact not equal to investment (10). Are we then violating the identity of S and I as Keynes had earlier done in his Treatise (1930)? Actually, we are not and we can base this answer on the idea of involuntary (or "unplanned") savings. To understand this better, let us proceed by one more round. The 9 units consumed themselves translate into higher output and income by 9 units which are, in the second round, divided into consumption (8.1) and savings (0.9). Therefore, at the end of this second cycle, we have total savings of 1.9 (1 + 0.9), total consumption of 17.1 (9 + 8.1) and total income of 19 (10 + 9). Our initial investment (10) is still greater than savings (1.9). However, let us reflect for a moment on what is happening. By arbitrarily slicing through the multiplier process (as we have done after the second round) we have in fact prevented the circulation to occur fully. In other words, the 8.1 that is going to be consumed in this second period has not, in fact, been consumed yet. Momentarily, it is still in the consumer's wallet - and therefore, it is "involuntarily saved". Total savings, at this time, are thus: planned savings (1.9) + involuntary savings (8.1) = total savings (10). Therefore, total savings will actually be equal to total investment. Thus, even though there is a behavioral relationship between I and S, the accounting identity, I = S, is necessarily kept at all times (Lerner, 1936, 1938, 1939, 1944) Only when the multiplier works its way through entirely will total savings be composed of planned savings (10) and no involuntary savings. We will, simultaneously, move to a higher level of economic activity (higher output and consumption) according to the size of the propensity to consume. By our calculation of a 0.1 marginal propensity to save (s), then output and demand will increase by (1/s)(change in I), or rather, (1/0.1)10 = 100. So, in the final instance, output and demand increase by 100 units and savings by 10 units - all as a result of an increase in autonomous investment by 10 units. (or, as Ohlin (1937) would call it, the planned savings are the "ex ante" savings and the actual savings are the "ex post" savings) Hence, as insinuated earlier, we clearly see that investment determines savings in this system. This may seem counter-intuitive since conventional wisdom holds that we cannot invest what we do not save. But Keynes, following the classics, regarded savings as "real" goods (Keynes, 1936: p.81-2). One cannot effectively save what does not exist. Therefore, to expand savings, one must expand investment first to produce the goods which can then be saved. How does one "invest" if savings do not exist to back them up? To answer this, Keynes (1936: Ch. 7) merely invoked the institutional framework of the monetary economy which made this possible. Investment, he proposed, is dependant merely upon the availability of finance - regardless of whether or not "savings" existed to back it up. Investment is a mere bookkeeping entry to begin with (Keynes, 1937). Hence an entrepreneur, in an effort to expand his production, will be given "credit" which he will then use to employ more labor to increase production. The extra income generated from this injection is then resolved into consumption and profits (ploughed-back earnings) - which are used to repay the loans. So production is increased by an entrepreneur on the basis not of existing savings but rather on the promise of profit to replenish his credit. [There is nothing extraordinary about this assertion - but, for some bizarre reason,
this obvious truth is far from universally accepted. The older Ricardians would disagree about the process of
credit e.g. J.S.Mill's claim that "Credit has a great but not, as many people seem to
suppose, a magical power; it cannot make something out of nothing" (J.S. Mill, 1848: p.348). In this, the normally cautious Mill
is completely out-of-touch with the workings of the real economy. This inexplicable and
peculiar state of denial is surprisingly also held by modern Neoclassical economics (e.g. Solowian growth theory)! However, it should be
noted that many old Neoclassicals, such as Knut Keynes's argument on the credit-investment decision is familiar to any businessman and is based on the old Wicksellian conception. Furthermore, it is precisely reminiscent of the microeconomic concept of the entrepreneur as an organizer rather than an owner of resources (even though the Neoclassicals muddled this quite a bit): an entrepreneur might not own anything other than his enterprising skills, but, in principle, on the basis of promised repayment, he can rent as much labor, capital and land as he needs to produce as much as he wants. A penniless entrepreneur can conceivably create a billion-dollar industrial empire on the basis of credit alone. He is not personally constrained by any endowment. Similarly, investment occurs on credit and not a given endowment of "savings". The savings will come to meet the original investment:
From here we get two paradoxes. Firstly, there is the Keynesian "Paradox of Thrift": increasing savings exogenously is not possible without reducing consumption. If consumption is reduced, then firms' sales will be less. As inventories pile up, firms will shelve their investment plans and cut back production and reduce employment. So contrary to the popular pre-war prescription, exhorting a population to save more and spend less is in fact inducing it to move to a lower level of output. Thus Keynes' famous saying, "whenever you save five shillings, you put a man out of work for a day" (Keynes, 1931: p.152). As we have seen before, the Neoclassicals, following Say's Law, considered overconsumption (insufficient savings) to be the true cause of unemployment. With Keynes, underconsumption rather than overconsumption is the root evil. (a position Keynes (1936: p.358-61) traces back to Mandeville (1723) and Malthus (1820)). Secondly, in the Neoclassical loanable funds theory, interest rates will adjust to bring savings and investment into equality at full employment equilibrium. However, as we just noted, S = I at any level of output. This is brought about not by interest rate changes but by definition. There can be no separate savings and investment schedules. They are an identity and thus irrelevant to interest rate determination. (Keynes, 1973: p.552). However, the equality of voluntary (i.e. "planned" or "ex ante") savings and investment is brought about by changes in output. Investment generates output and hence income which then generates the necessary voluntary savings. However, during the process, total saving (voluntary plus involuntary) is equal to investment. B. Ohlin (1937) and D.H. Robertson (1937, 1940) took Keynes to task on this point. Investment must necessarily be financed but at any one time, only voluntary savings are available - involuntary savings are not loanable (they are in the wallet and not at the bank). So, if we are at the first round, where voluntary savings are only 1 unit and investment is 10 units, then 9 units of investment remain unfinanced. Of course, we can, as before, appeal to credit creation by financial intermediaries to make this extra 9 units possible and indeed, that is what Keynes (1937) does by appealing to a finance motive in the demand for money.
However, this causes further complications since it implies indirectly that interest rates do adjust to bring the supply of loanable funds (ex ante savings) into equality with the demand for loanable funds (investment) at any point in time. This does break with the structure of the General Theory - however, since it is a post-1936 modification, we shall ignore it here. Returning to the General Theory, we can note that the one crucial difference between Keynes and the Neoclassicals, exemplified by Pigou, Cassel, etc. seems to be the utilization of output and income adjustments to bring savings into equality with investment rather than the interest rate (a price). Not surprisingly, there really is no theory of prices in the General Theory at all (which is not the case of the Treatise). The logical conclusion of this system, therefore, is that expenditure determines output. But this is not, by itself, a novelty for any Marshallian demand schedule shows that demand determines output. Furthermore, Marshall (1890) was quite at home with quantity adjustments in response to disequilibrium in demand and supply. Keynes' innovation, then, was his use of investment as an independent variable and the subsequent use of Kahn's multiplier rather than Wicksell's "cumulative process" to bring savings into equality with investment. This was possible, as we have seen, insofar as credit exists in a monetary system. It is in this respect that he was able to exclude the price system in the process and use output as the sole adjustment mechanism. Realizing this, we are immediately inclined to ask what determines this level of investment. Keynes confronted this difficulty in a strange manner. His answer was to suddenly leap from the aggregate to the individual world and invoke the "marginal efficiency of capital" (MEC) schedule of the entrepreneur. The MEC schedule drawn by Keynes (1936: Ch. 11) incorporated the idea that investment and the rate of return on investment (MEC) are negatively related. Is Keynes relying on the marginal productivity theory? Not apparently. Rather he relies on a "quasi-rent" conception of a falling MEC curve. He gives no fundamentally intricate reason for this, but the intuition he offers seems, at first glance, quite simple: the volume of investment is dependent on the existence of profitable opportunities and the costs of pursuing them. As Keynes had earlier claimed in his Treatise, turning Ricardo on his head, "the engine of Enterprise is not Thrift but Profit" (Keynes, 1930). As such, initial investment will naturally take up the most profitable opportunities (those with the highest returns to investment, MEC). As investment increases, the most profitable opportunities are already used up and the entrepreneur must start taking on projects with increasingly lower returns. The second half of the story is that as investment increases, the entrepreneur will eventually reach a point where all remaining investment opportunities yield a return which is less than the prevailing rate of interest. Since investment needs to be financed, it will cost him more to borrow money to invest than he will obtain in return. Keynes then concluded that investment will occur until the marginal efficiency of capital (MEC) is equal to the rate of interest. Let us step back for a moment. Why do returns diminish as investment increases? Presumably, by Keynes' own argument, higher investment generates higher demand, thus the firm's revenue will rise as that higher demand is spent. Of course, not all the demand generated by, say, Ford Motor Company investment will be for Ford cars. They may very well be for Chryslers or any other products. Thus, increasing investment will not by itself create more investment opportunities for the individual firm. Granted, this may explain why returns diminish for a firm's investment, but in aggregate, the demand generated in an economy will all be for the economy's total product. There is no "leakage" out of it as there might be for an individual company. Thus, while this rationale implies investment of a single company might yield diminishing returns, the investment of the economy as a whole should not. A second argument to justify a falling MEC involves adjustment costs. As firms increase investment, supply-constrained capital goods industries cannot meet the extra demand immediately, resulting, therefore, in higher prices for more capital goods and hence higher costs for every additional unit invested. This would give us a falling MEC. However, this could only be true in the short-run. In the long-run, capital goods industries should be able to adjust their capacity to meet the increased demand for capital goods. Therefore, in the Marshallian long run, the marginal adjustment cost argument also collapses. It is here where the distinction between long run and long period becomes crucial. If one argues (as we have) that Keynes is developing a short-period/long-run theory, then this presents no difficulty. The long run in this case would be merely a succession of short periods where complete adjustment is not possible, where firms do run up against capacity constraints. However, if we follow others, such as Garegnani (1978), in the belief that Keynes' theory is applied to the long period (the length of time necessary to change all inputs), then capacity constraints cannot exist and the MEC should be horizontal instead. Hence, since the adjustment cost argument cannot apply in the long period, it cannot explain the falling MEC curve. This problem is underlined by Keynes' own resolution when he says that "Professor Fisher uses his `rate of return over cost' in the same sense and for precisely the same purpose as I employ `the marginal efficiency of investment'." (Keynes, 1936: p.141; also Keynes, 1937). In the long-period interpretation, this move was a big mistake. Let us see why by recalling what Irving Fisher's (1930) rate of return over cost actually is. In Fisherian theory, we are given two arguments for declining returns to investment: in one, it is assumed that as investment increases, the demand for investment goods rises and, capital goods firms being supply-constrained, cannot meet that higher demand and thus must raise price. This is the same as the "adjustment cost" version given above, essentially a short-run story, although Fisherians really make no such distinction. The second explanation given in Fisher's (1930) original exposition is simply based on marginal productivity. As more and more capital gets applied to a fixed amount of labor, returns for each extra unit of capital added will diminish. The Fisherian stories are very credible - but for two details: we have referred to "supply-constraints" and a "fixed amount of labor". In the former, it means that there is no excess capacity; in the latter, no more supply of labor is forthcoming. From whatever the vantage point, this is the definition of full employment! In unemployment, neither of the Fisherian arguments hold water. Thus, Fisher can derive the negative relationship between investment and rate of return only at full employment. We can see now how it might be illegitimate for Keynes to apply Fisher's explanations to a situation of underemployment equilibrium given that the premise for Fisher's arguments was full employment. Keynes' MEC curve, the investment demand curve, is illegally employed. As Garegnani (1978) argues, it cannot be downward sloping but rather must be horizontal. In what seems, then, to be a less coherent and less logical portion of the General Theory, Keynes failed to carry his thesis and create a distinct theory of investment for the long-period. Instead, in an effort to find a short-cut, he resorted to resurrecting the inverse relationship between interest rates and investment without any theoretical rationale behind it which is consistent with his theory of effective demand. This, more than anything, reveals that Keynes could not really have been working in the long period but rather in the short period - with the long run defined merely as a successive series of short periods. To some extent, this view is not very appealing since no one realy denies that Keynesian phenomena exist in the short run and one might indeed make the argument that capacity constraints are imposed imperfections. Indeed one might be tempted to fall into Jacob Viner's (1936) mistake and insist that Keynes's theory is only short run and that, in the longer run, things should be thought exclusively in terms of the classical models. Garegnani argues that the issue which Keynes is trying to grapple with, if he is to be theoretically interesting, is whether this phenomena will persist uncorrected in a fully-working, fully-flexible Neoclassical-type economy, i.e. in the long period, in its precise theoretical meaning. One plausible alternative would be to employ Kalecki's (1937) "principle of increasing risk". As investment increases, so does the debt burden and the risk of personal loss - thus, entrepreneurs throughout the economy discount future returns at a higher rate. This would give us a falling MEC curve. Since models based on uncertainty sit between the long run and the short (neither time nor input flexibility eliminate uncertainty), the principle of increasing risk might get us out of this debate. Uncertainty may be considered an "imperfection" and so, for long-period theorists, it is still unacceptable, however one cannot ignore the stress Keynes (1936: Ch. 12; 1937) placed on uncertainty in investment decisions (as many Post Keynesians have argued, e.g. Davidson, 1993).. The confusion surrounding the MEC curve will, of course, land Keynes in serious trouble in chapters ahead. For one, anyone familiar with simple economic theory will realize immediately that he is introducing a price (interest) into the system which can in principle change real phenomena (i.e. with a downward-sloping MEC, interest rate can change investment). This can have grave consequences on his thesis that an economy can be constrained to an underemployment equilibrium. Primarily, we know that there must be some level of investment which will be equal to the full employment level of investment. Having introduced a price, Keynes must now show that the interest rate will fail to bring investment into equality with the full employment investment level or else his whole system will fall apart. He must, in effect, neutralize the variations in the interest rate somehow and so prevent the interest rate from acting as an adjustment mechanism. He does so by extracting the interest rate from the goods market entirely - introducing the "theory of liquidity preference" (Keynes, 1936: Chs. 13-15; also Keynes, 1937). Having interest rates determined in the money market, Keynes hoped to be able to show that the interest rate would not be a real market adjustment mechanism as in the Neoclassical model. The crucial component of this theory, of course, is the speculative demand for money. However, the speculative demand is a function of two interest rates: the present interest rate and the future expected interest rate. If underemployment equilibrium is to persist indefinitely, then these must be equal - so that there will be no change in holdings (and hence no movements in the interest rate which might bring us to full employment). In effect, a long-run theory of the interest rate is necessary. However, Keynes provides two quite different theories - neither of which is very convincing: the long-run interest rate is either "psychologically" determined or it is "conventionally"- determined (in fact, on opposing pages, 1936: p.202-203). In addition, Keynes has to effectively explain the relationship between the rate of return on money and the rate of return on real assets which govern investment decisions. Why and how should an rate determined in the money market ensure that the rate in the goods market be constrained from moving to their full employment equilibrium? To solve this, Keynes imposes upon us the idea of a "liquidity premium" - the money rate of interest. This premia, he claims, places both a floor and ceiling upon all other rates of return and thus is capable of preventing a movement of these real rates to the full employment equilibrium interest rate. In essence, the money rate of interest rules the roost (Keynes, 1936: Ch.17). This idea, in fact, merely adapts concepts already contained in both Hicks (1935) and his own Treatise (e.g.Keynes, 1930: Vol. I pp.127, 222-30). Namely, the optimal portfolio mix of assets would be one where the marginal revenues (rates of return or "own-rates" of interest) on different assets are equated. It is assumed that if the rates are altered or are not in equality, the portfolio holdings will change until they are. Letting there be three assets, money, bonds and capital, where rM, rB and rK are their respective rates of return, then the portfolio balance condition is:
Now, the "interest rate" which Keynes speaks so often about is really the rate of return on an aggregate he referred to as "alternative assets". In this case, there are two: bonds and capital. It is implicit in Keynes, that the "rate of interest", i, is really the return on bonds so rB = i. This implies that capital is a perfect substitute for bonds and so the interest rate on capital is a fixed differential above i. So, if the portfolio balance implies that: rK = i, then it must be that iK (the "interest rate" on capital) is a fixed differential (the `equity premium', d) above i so that:
the interest rate on capital (iK) is an equity premium (d) above the interest rate on bonds (i which is equal to both rB and rK in portfolio balance). Thus, we can rewrite the portfolio balance condition as:
so the rate of interest on bonds, i, is the "liquidity premium" (rM) and the rate of interest on capital is just a fixed differential above that. (indeed, Hicks applauds the treatment of interest as a liquidity premium "The rate of interest on these securities is a measure of their imperfection - of their imperfect moneyness" (Hicks (1939: p.163)). The crucial interaction for interest rate determination, then, is between the money and bonds markets. Letting Md, Bd, Kd be the demands for money, bonds and capital respectively, and Ms, Bs, Ks the supplies of money, bonds and capital, then wealth demand is
and wealth supply is:
Now, Keynes' theory involved a "dual decision": firstly, on the amount of consumption and saving (through the multiplier and the theory of effective demand) and, given that, on the allocation of accumulated savings (the stock of wealth) between money, bonds and capital. This implies that the flow decision (consumption/saving) precedes, and therefore is independent from, the stock decision (money/bonds/capital). The former determines the level of output and hence income and savings which then feed into wealth demands which, together with stocks, give us the equilibrium rate of interest (i) by portfolio balance, which will then give us the level of investment (via the MEC curve) which then gives us output, and so on until an equilibrium is reached. This implies that at any moment in time, wealth demand is equal to wealth supply, i.e. Wd = Ws. If not, then the shortfall might be met by changing the level of savings. This is not a relationship Keynes sought since it would imply that investment and the multiplier are not the only determinants of saving. Thus, Wd must be equal to Ws at every point in time (which, indeed, is what is implied if I = S, so that no changes exist to wealth supplies or demands). The Wd = Ws condition implies a portfolio stock constraint as:
or, since iK is a fixed differential above i, then bonds and capital are perfect substitutes so:
or, for simplicity we can drop the capital market altogether so the constraint becomes merely:
So, if there is excess demand in the money market, there must be a corresponding excess supply in the bond market. If this were true, then the rate of interest on bonds would rise, implying that the liquidity premium has risen (i.e. it will take more of a premium to get agents to surrender money and take up bonds). Similarly, if there is excess supply of money, there must be an excess demand for bonds so the rate of interest will fall (i.e. the liquidity premium need not be so high). Keynes' (1936: Ch. 17) specific proposition was that the money rate of return, (rM, the liquidity premium), as determined in the money market, would "rule the roost". In other words, rM would not adjust to get the optimal portfolio but would rather expect the holdings of other assets to do the adjusting - bringing their returns into equality with the money rate of return. (see also Kaldor (1960) and Kahn (1954)). How is this possible if we have identified rM with i? Well, this
identification is only true at equilibrium. Suppose rM is "given",
then will i adjust to become equal to rM or will the rM itself do
the adjusting? Firstly, Keynes proposed that money has a "zero elasticity of
production" (1936: p.230). This simply means that if the rate of return on bonds is
below the money rate of return, (rB = i Granted that the money rate cannot be adjusted downwards, can it at least be adjusted upwards? No. Keynes also claimed that money has a "zero elasticity of substitution" (1936: p.231). This means that money must be a part of every asset sheet by sheer need for its liquidity. There is no perfect substitute for money. So if the rate of return on other assets is higher than the rate of return on money, rB = i > rM, one cannot simply do away with all one's money holdings and substitute for alternative assets (thus bringing the money rate up). Rather, one must adjust the quantities of alternative assets down to the money rate. Therefore, by the zero elasticity of substitution assumption, the money rate also places a ceiling on all other rates. So far, there is nothing in liquidity preference that is so unappetizing. However, we have said nothing about the level of the money rate of return. The money rate places both a ceiling and a floor on all other rates, but if it is "given", if it is not itself determined by optimal portfolio allocation, where is it determined? For this, Keynes again takes refuge behind the idea of the speculative demand for money. He claims that the rate of return on money is determined by the expectation of a rise or fall in the rate of interest on bonds. But if the rate of interest on bonds is itself, by portfolio balance, the liquidity premium, then it seems we have really said nothing at all - leading Hicks to comment that:
Thus, we return to the question of the long-run theory of interest which Keynes left effectively unanswered. However unfortunate, at this point, the absence of a long-run theory of interest is not crucial. Other theories have been constructed to give us an interest-sensitive money demand curve and explain precisely how the rate of interest is determined by stock portfolio conditions. What is important, however, is to realize the implications of a rate of interest fixed outside the realm of the goods market. This rate of interest implies a portfolio stock equilibrium and thus has no endogenous tendency to change. However, it will determine, by the MEC schedule, the amount of investment in an economy. If the interest rate is constrained to give us a level of investment below the full employment level, this implies that aggregate demand and, hence, by the multiplier, output, is below full employment output. By the technical conditions of the production function, this will give us a level of employment which is below full employment. As long as the rate of interest fails to change, then investment, output and employment will remain constrained below full employment. And the rate of interest will not change as long as the portfolio balance condition is met - and it can be met at any level of output. If employment is below full employment, then the real wage, as determined by profit-maximizing conditions, is above the full employment real wage. The profit-maximizing, marginal productivity condition states (which Keynes accepted, 1936: p.17):
Where dF/dN is the marginal product of labor, w is the nominal wage and p the price level. Under normal conditions, this implies that labor will be hired until the real wage (w/p) is equal to the marginal product of labor (dF/dN). However, in Keynes's system causality runs the other way. Demand determines output so we arrive at the labor market with the level of employment already determined. Thus, the marginal product of labor, dF/dN, is already given. All the profit-maximizing condition gives us, then, is the level of real wages, w/p consistent with this - which will be above the market-clearing real wage. Thus, there will be excess supply of labor (unemployment), but the real wage will not fall to its market-clearing level since it is already determined by demand conditions (Keynes, 1936: p.29). So far, we see that the real wage is fixed but it is fixed endogenously. There is no imperfection implied in this statement. However, early in the General Theory, Keynes asked us to assume that nominal (or money) wages, w, were also fixed. This is an imposed imperfection. But why did he impose it if he sought to have a fully working system? The reason for this is that Keynes sought to speak of a determinate price level. Although w/p is determinate, w and p themselves are indeterminate. If w/p = 10, then it could be that w = 100 and p = 10 but it could also be that w = 20 and p = 2 or w = 1000 and p = 100 - all of which would give us w/p = 10. Thus, the actual price level is indeterminate - and so too is the money wage. However, if we fix the money wage at, say, 50, then p must definitely be 5. It cannot be anything else. Thus, the sole purpose of assuming a fixed money wage is to pin down a positive, determinate price level. This assumption was not made, as later observers would consider it, to guarantee an unemployment equilibrium. Unemployment equilibrium, and the corresponding real wage, is determined by the demand conditions. Making the nominal wage flexible does not change this. It only changes the price level. The price level will adjust one-to-one with the money wage to keep the same real wage. In chapter 19 of the General Theory, Keynes tries to prove this by permitting a flexible money wage. Here is where Keynes's thesis runs into trouble. Early in the General Theory, Keynes makes it a point to argue that worker- employer bargaining occurs over the money wage (1936: pp.7-9). Now, in unemployment, with the real wage above the market clearing real wage and an excess supply of labor, there will be competitive pressures to bring down the real wage rate. However, since the real wage is fixed by demand conditions, the most that can be achieved is the bidding down of the money wage. As long as w/p is fixed by demand conditions above the market- clearing real wage, then w/p will not change by a fall in w but will be accompanied by a concurrent fall in p. In other words, as long as Y is fixed by demand conditions, then so is dF/dN. As dF/dN = w/p, then reducing w implies dF/dN > w/p, which is not a profit-maximizing solution under competitive conditions. Thus, microeconomic considerations require that p fall by exactly the same amount as w fell so that w/p is brought back into equality with dF/dN. However, as long as unemployment exists, there will be competitive pressures to bring the money wage down. Thus, over time, the money wage rate will fall to zero and so too will the price level. Thus, fixing the money wage, as Keynes did early on, gives us a positive price level which may be used as a numeraire. When he made the money wage flexible in Chapter 19, the possibility of being at or getting near a zero money wage and zero price level reemerged. What precludes the money wage from falling to zero? There must be some justification for the sustainability of labor market disequilibrium without zero wages. One plausible answer is, of course, that the wage is somehow fixed from the outside (just as the interest rate is fixed from the money market). However, recall that Keynes's thesis was that underemployment equilibrium was a natural and not an imperfectionist situation - the General Theory is supposed to hold true in a perfectly competitive situation. Of course, later Keynesians, through the Phillips Curve, were able to present a justification for the determination of the money wage (shifted by one derivative), but Keynes himself possessed no such device. Let us then follow Keynes (1936: Ch. 19) and suppose that the money wage is flexible so that it did fall. Now, a falling price and wage will influence three components which can affect output and employment: the multiplier (i.e. consumption demand), the marginal efficiency of capital (i.e. investment demand) and liquidity preference (i.e. money demand). Let us take each a step at a time. With a falling money wage, it is immediately apparent that workers will not be worse off if the price level falls proportionally. The real wage remains unchanged, and thus it is of no consequence to aggregate demand. It is plausible, as Keynes (1936: p.262) argues, that the falling prices will involve a redistribution of income from debtors to creditors, and thus, by different propensities, aggregate consumption and hence aggregate demand and output will fall (an argument akin to the "debt-deflation" hypothesis of Irving Fisher (1933) and early Keynes in his Essays in Persuasion (1931: Chs. II.2, 5, 7)). Since w/p is fixed by demand conditions, it will never be the case that w/p will rise (so output and aggregate demand rises). On the contrary, if anything, wages might fall faster than prices (since prices adjust in response to the wage rate) so that, only temporarily, w/p falls, then worker income and hence aggregate demand, output and employment would fall. Thus, the net effect of a falling money wage on output is that it remains the same. The only permissable lag (that of prices behind wages) and debt redistribution effects imply that, if anything, demand and output would fall. Thus, through this channel, flexible money wages do not improve upon employment. Through the investment channel, the result is somewhat the same. If wages and prices fall proportionally, the real wage is unchanged and nothing happens. If money wages fall more than prices, then we might expect investment to increase since there will be lower costs to production. But, recall that lower real wages imply lower aggregate demand. Either the forward-looking entrepreneur realizes this (and hence refuses to invest and/or actually divests) or else he learns it the hard way: i.e. invests but fails to find the demand to absorb not only the extra output but also the previous output, and thus is forced to cut back. Thus, in the case of investment, a falling money wage will not affect it or, if it does, it will be to lower investment. The case of liquidity preference is a little less ambiguous. As money wages fall, the transactions demand for money falls. This is true regardless of whether the real wage remains unchanged or not: not only is nominal income lower but so too are prices, hence less cash is required for transactions purposes. A fall in the transactions demand for money thereby depresses interest rates. Lower interest rates, by Keynes' own thesis, imply higher investment and thus a rise in both output and employment. This conclusion goes right against the purpose of the General Theory! Keynes expects there to be no movement towards full employment and yet here we see that permitting wages to be flexible, there is indeed such a movement. This is what Patinkin (1948) christened the "Keynes Effect" (actually, Patinkin argues that we can conceive of the falling w (and hence p) leading to an increase in the real money supply - which has the same result). In an effort to preclude this, Keynes explained that a fall in money wage rates could generate two other scenarios. If the wage reduction is very small, then this would fail to affect anything. If, however, the wage reduction was large enough, Keynes proposed that it could have such a depressing effect on consumer confidence that most would increase their precautionary demand for money in the hope for better times. Also, given deflation is usually followed by further deflation, then the speculative motive would also imply a higer demand for money (since the value of money would be expected to rise in the future). Thus by acting on confidence and expectations of agents, falling money wages and prices would increase money demand. These increases in money demand would drive up interest rates and thus stifle investment and hence kill the movement towards full employment. This was later formalized by Robert Mundell (1963, 1965) and James Tobin (1965, 1993) and has become known as the "Tobin-Mundell Effect". The interpretation of the manner in which Keynes chose to close this issue depends on our position in the long period-long run debate. On one side, we can argue that Keynes was trying to obtain an entirely non-imperfectionist system, by which, of course, we should eliminate all inklings of rigidities, uncertainty and other intangibles from the arguments which may give us a view of sustained unemployment as an "imperfectionist" model. In this view, unemployment is an equilibrium situation that arises in the long period from a perfectly working system. In this context, Keynes' closure on the issue of confidence seems to be a woefully inadequate escape hatch. Clearly, by resting this part of his thesis on ill-defined "psychological propensities", Keynes had been unable to set his system firmly on a set of economically primitive propositions as might have been originally intended. On the other hand, we must not forget that Keynes also wrote a series of other books and articles, from his 1921 Treatise on Probability to the QJE article (1937), all of which, almost without exception, invoke the important role of historical time and expectations and attempt to work in the short-period. Whether the Keynes of these other writings is the same as the Keynes of 1936 is, of course, an issue which cannot be resolved easily - but we would be hard-pressed to believe otherwise. As such, then, some theorists, such as Tobin (1975, 1980, 1989, 1993), Davidson (1972, 1992), Minsky (1975) and Palley (1992, 1993, 1994), indeed are inclined to rest the Keynesian system on precisely "real world" effects such as these. This need not necessarily be with an eye for a Shackle-style collective losses of confidence as such, but for more precise, reasonable justifications for a permanent failure of traditional adjustment processes. Of course, this interpretation requires that we substitute the idea of (presumably long period) underemployment equilibrium for one of (possibly long time or even permanent) disequilibrium. Tobin stresses this unambiguously:
and:
Which of the two interpretations might be correct? Intuitively, the idea of a long period underemployment equilibrium would be more akin to the the grandiose purpose Keynes alluded to. From this vantage point then, the General Theory remains somewhat unconvincing in its conclusion. However, the parallels between the "protracted disequilibrium" version of the General Theory and his other work, before and after 1936, is striking. To hinge the closure of his thesis on these conceptions is not without earlier and indeed concurrent foundation. Furthermore, it supports the idea of the Marshall-Keynes "behavioral" connection we proposed earlier. We shall then conclude our interpretation with a compromise. We propose that the General Theory was indeed a stab at constructing a "new" theory of demand-determined general equilibrium. It is, as a consequence, also a theory (as opposed to an untheory) of economic fluctuations. However, it has a language of its own: it is not expressed in terms of stale, deterministic relationships in logical time (as of the Walrasian veriety), but, rather, it is imbedded in a matrix built by the type of rich, historical Marshallian behaviorism which infused all of Keynes' other work. Keynes might have garnered illusions of grandeur in assuming he could play Einstein (i.e. that the Neoclassical version would emerge as a special case) and thus led many to assume that his system could be expressed in the static and deterministic language of Walras and still give us an underemployment equilibrium. This is virtually impossible for the construction of the Walrasian system is precisely about market-clearing. As Lucas notes:
In other words, placing Walrasian theory and unemployment together into the same pot results precisely in the type of "untheory" Keynes fought against. The problem, of course, is that Keynes was not really combating Walras but rather Pigou, Robertson, Hawtrey, Lavington and the rest of the Cambridge crowd, the "original" macroeconomists. All of these, at the time, were really Marshallians and not Walrasians but nevertheless committed the fatal error (in Keynes' view) of employing "general" essentially Walrasian resolutions to allocation problems to determine the general levels of output and employment. What Keynes really sought to show was that these theorists were incorrect. The Walrasian question, the question of allocation, is wholly unrelated to the issue of the determination of output and employment. Thus, what might have been Keynes impulse in 1936, as we argued earlier, was to show Pigou that if one carries Marshall's behavior-laden partial analysis to the general context and then on to its logical conclusion, one does not obtain Walrasian results. As Hicks explains in his review:
However, in the middle of correcting the Pigovian generalization of Marshall, Keynes must have felt he was on the verge of a theory which would revolutionize all economics - including the original marginalist thesis. In a way it did, but Keynes's system of protracted disequilibria had to be expressed in Marshallian language. He might not have thought twice about it until after publication, his repeated closures of difficult issues with "expectations" and "conventions" might have seemed natural to him, but the economic world of Cambridge is not that of Lausanne. Just as Pigou might have made a mistake in transposing Walrasian conclusions to Cambridge economics, Keynes did the complementary mistake in alluding to the idea that Cambridge economics could be brought into the marginalist, Walrasian system. But the Walrasian virus he retained inside the system would soon explode into a rash and bring the entire General Theory down. Thus, both interpretations have room for maneouvre here. The underemployment equilibrium interpretation believes that the effort of translating Keynes to Walras should be completed: either by expressing Keynes in Walrasian terms (as the Clower-Leijonhufvud- Post-Walrasian tradition would have it), or by destroying Walrasian theory on its own and letting Keynes's supplant it (as Garegnani, Pasinetti, Eatwell and the Neo-Ricardians have attempted). In the protracted disequilibrium interpretation, efforts to translate Keynes into the language of Walras should be kept to a minimum. The importance of the General Theory is held to be its explanatory power on the real world - defined in the real-time, behavioral sense, with fluctuations in theory rather than untheory. It is through this, rather than internal theoretical logic, that Keynes' system will replace the Walrasian goggles. This type of interpretation unites "fundamentalist" Post Keynesians such as Weintraub, Davidson, Minsky etc., with some of the latter-day Old Keynesians (notably, Tobin and other reformed synthesists). The research program in this interpretation - especially that of the Post Keynesians in particular - is to take Keynes at face-value and attempt to give his work its own distinct foundations which Keynes originally neglected or only hinted at. Both the interpretations of the Keynesian system, then, can be derived. The main difference lies in the stress placed on different parts of Keynes' system: either on his claim to replace all other theory, or in his claim to explain fluctuations with theory. Both, of course, are not incompatible, but one must make clear, in any work involving the General Theory, if the stress lies on "translation" or on "explanation". Nevertheless, the contributions Keynes made to economic analysis as a whole were more profound than merely attempting to find a "market failure" in itself. High among its long list of achievements was the provision of a novel conception of equilibrium and adjustment. Primarily, instead of assuming a "gravitation" towards some unique equilibrium position, the mechanics in Keynes' system allowed for a continuum of gravitation points to be determined by the level of demand. In addition, for each of these points, output (treated here as determined by monetary-cum-demand phenomenon) is the adjustment mechanism rather than the interest rate (a price). These crucial points break away from the Neoclassicals' use of a supply-constrained equilibrium. Furthermore, in spite of all the shortcomings, Keynes never did away with any fundamental economic concepts except Say's Law. On the other hand, he kept vital Marshallian "real-world" components which had been abandoned in the generalization of the classical system. Nevertheless, while Keynes might not have successfully carried out his own mandate in an entirely cohesive manner, he dropped in his path sufficient (perhaps too many) clues and insights on how to close his system. In the establishment of any research program, we should not expect all answers to all questions to be met within the covers of one book - especially when only one, unrevised edition of it is available. True, there are many questions left hanging in the air but they are no more nor loss than the ones placed before any other pathfinder - such as Walras (1874) himself. The difference, however, is that Walras made no deference to previous authorities: he built up his system from scratch. Keynes, on the other hand, selectively altered particular sections of the Classical system to forge his own without changing the underlying classical relationships. Perhaps this was precisely his most crucial error. As Hahn notes:
Keeping one foot in the old Neoclassical paradigm and the other foot in his own system was an invitation to incoherence and contradiction. The disaster became obvious as the Neoclassical-Keynesian Synthesis was launched, believing it could indeed explain Keynes with Neoclassical microfoundations only to find that it would have to make some heroic assumptions and dubious reinterpretations of Keynesian concepts. Was this necessary? Perhaps not.
|
All rights reserved, Gonçalo L. Fonseca