Chapter 17 of Keynes’s General Theory is titled “The Essential Properties of Money”. Perhaps it should have been entitled “The Essential Properties of Liquidity” for that is its key theme. Having argued how essential the desire for liquidity (relative to its supply) was to understanding how economic systems function in the preceding 16 chapters. Keynes turns to the question of what it is that makes an asset liquid. He provides a detailed discussion of the two key features that he argues determines an assets liquidity: its “elasticity of production” and its “elasticity of substitution”. Keynes considers that both of these must be extremely low – zero or near zero – for an asset to be liquid. Any asset that possesses these two things, along with a low carrying cost, will tend to be the most desired store of value in times of increased uncertainty and low confidence. The cul-de-sac of own rates of interest Keynes begins chapter 17 with a discussion of own rates of interest. He takes the concept from Sraffa. The concept is introduced in Sraffa’s March 1932 Economic Journal article. Sraffa uses this concept (which he calls the “natural” rate of interest on a particular commodity) to describe – in terms of rates of return- the process of restoring Marshallian long-run equilibrium once this equilibrium has been disturbed. In point of fact what Sraffa describes is the classic case of a shift of preferences away from certain products or commodities toward others in an economic system where constant cost industries prevail. 2 What is extremely odd, and at root one of the principal causes of confusion surrounding this chapter is that while Sraffa’s “natural” rate of interest describes the different rates of return (expressed in monetary terms) that obtain in the transition from one long-run Marshallian equilibrium to a new one what Keynes describes is really a real rate of return. 3 In other words, equating long-run minimum average cost with what both Keynes and Sraffa refer to as a forward price and the short-run equilibrium price that results after a shift in demand as they spot price, Sraffa uses equation (1) below while what Keynes refers to is instead equations (2A), (2B), or (2C) (all equivalent to one another). Sraffa’s natural rate of interest: israffa=(Pspot/ Pforward)-1 (1) Keynes’s own rate of interest: ikeynes==(Pspot/ Pforward)*(1+im)-1 (2A) . ikeynes=im-pj (2B) ikeynes=(qf/qs)-1 (2C) Where im = rate of interest charged on a one year loan ikeynes = Keynes’s own rate of interest israffa = Sraffa’s natural rate . pj= the rate of change in the price of good or commodity j. Flatly rejecting the feasibility of index numbers early on in the General Theory, Keynes’s real rate of interest is expressed, as equation (2B) indicates, as the money rate of interest relative to the expected appreciation or depreciation of a particular product. Then, as in chapter 17, if the rate of interest on a one year loan is 5% and wheat prices are expected to rise 7%, - perhaps because the demand for wheat is down and the market for corn is up – the money rate of interest, expressed in terms of wheat, would be -2%. Although not found in chapter 17 if the anticipated appreciation in wheat was due to a 4 restoration of the wheat market to its long-run equilibrium it is of course possible that the money rate of interest expressed in terms of corn might be +13%. This would due to the fact that as wheat farmers switch over to corn and the market supply of corn increases, the short-run spike in corn prices subsides and long-run equilibrium is restored. (The above implies an 8% drop in the price of corn over the period of time it takes to restore long-run equilibrium, or 5%-(-8%) =13%). Meanwhile, although Sraffa’s “natural’ rates of interest diverge (in my wheat and corn example here israffa=7% for wheat and israffa for corn -8% or israffa in general equals pj) it would be possible to express the wheat rate in terms of corn, 7-(-8%) =15% or the corn rate in terms of wheat, -8-7%=-15%. In the transition these are different too and the interpretation of such numbers, with some patience, is relatively straightforward. For example, if I buy wheat at its current depressed price and wait to sell it once its long-run equilibrium price is restored I will, all else equal, have 7% more to spend. With corn prices down 8%, again once long-run equilibrium is restored, I will have 7% more to spend on something that is 8% cheaper or a combined gain of 15%. The question remains where does Keynes’s playing with this misinterpretation of Sraffa’s “natural” rate of interest get us? Basically to the simple point that if an initial equilibrium is disturbed, rates of return, however expressed, will diverge and that the process of adjustment must continue until rates of return are once again brought back into balance. The essential properties: A Marshallian view: As I stated at the outset of this essay, the two features that Keynes argued determined the liquidity of an asset were its elasticities of production and substitution. 5 Again Keynes’s claim was that both of these had to be low, zero or near zero, for an asset to be liquid. Consider an economy that produces two types of goods (as in the passage taken from Sraffa above). Imagine that the demand for the one type of good, which currently serves as a medium of exchange among other things, rises and the demand for other goods falls. Sraffa’s “natural” rate of interest on the medium of exchange good, call it gold, rises while the “natural” rate of interest on all other commodities falls. Two things happen in the market for gold. First, as the overall demand for gold increases and the excess demand puts upward pressure on its price the quantity of gold demanded falls. Secondly, in the short-term there is an increase in the quantity of gold supplied. This is augmented over the long-run by the entry of producers into gold mining. They are at one and the same time drawn into the gold industry by higher rates of return as well as pushed there by lower rates of return, i.e. negative “natural” rates of interest, elsewhere. Save for frictions that stand in the way of easy entry and exit from one part of the economy to another employment and the overall standard of living would not be affected by this change. Here both the elasticity of substitution and production are high. Despite the low carrying cost of gold and the surge in the demand for gold, which keep in mind serves as a medium of exchange, the gold (i.e. money) rate of interest rises in the short-run but, over time, falls back to its original level. In contrast to this, consider what happens when both the elasticities of substitution and production are low. To begin with assume that the quantity of money, the supply of spending power, is absolutely fixed in terms of units of currency available. In short, assume that the elasticity of production is zero. Next, assume that these units of currency 6 can either be held as a store of value – to buy goods of some sort, some time in the future – or they can be used as a means of payment to buy other things. In essence, when someone holds, or wishes to hold, a greater amount of money they are, each day they choose to hold the cash, buying the freedom to manoeuvre later on. If under these circumstances, the demand for money rises because more seek to hold money as a store of value, the “natural” rate of interest on money rises as the rate of return or the “natural” rate of interest on other products falls. Following the previous example, the above normal returns on supplying money should encourage those currently supplying this product to supply a greater quantity. Similarly the above normal returns together with the negative returns elsewhere should encourage entry into the monetary sector of the economy. That the resources released from producing goods cannot be reabsorbed to “produce” money should be evident. As the prices of goods fall and therefore the opportunity cost of holding cash rises why would one not expect to see the money being held used to buy these goods? To begin with, what if those holding money as a store of value expect for some reason prices to continue to fall. The drop in prices that tempts me to spend now encourages me to wait a bit longer if I am fairly confident that the current drop is part of a broader downward trend. Secondly, if I were not myself to spend out of these balances but rather trust them to someone who borrows the cash from me I may rightly assume that the current drop in prices will lower the value of the collateral a borrower might offer as security (if it hasn’t done so already). The end result is perhaps at most a slight movement, a very slight one, toward the earlier equilibrium, or “natural” rate of interest for money. Firms employing resources 7 and producing goods and services will have to adjust to the higher “natural” rate of interest. In an economy in which all stores of value had relatively high elasticities of production and substitution rates of return would adjust to one another. Here the increased desire to hold liquid stores of value raises the standard that, at the margin, anyone is willing to accept if they are to part with liquidity and grant credit. All else equal, projects deemed worthwhile under more liberal standards will be delayed or cancelled. Liquid stores of value The very thing that makes liquid stores of value socially costly in terms of employment, production, and growth is what makes them attractive to individuals, especially when confidence erodes, despite these social costs. To see this return to the gold mining example given earlier where both the elasticities of production (perhaps both short-term and long) and substitution were relatively high. A rise in the desire for gold affords a premium to those who quickly and aggressively move, either to buy gold before its prices has peaked at Pspot or to supply it before the market return to Pforward. The more slowly the process unfolds, the slower supply moves up (either in terms of quantity supplied in the short-run or overall supply with the passage of time) and the more slowly the quantity demanded moves down the more likely one is to gain. The risks of getting caught suddenly by the door slamming shut once opened are, as it were, are far less. Although earlier I had used gold in the example of what would happen in an economic system that had a store of value (an item with a low carrying cost) with relatively high elasticities of production and substitution, gold in reality has been 8 considered a good store of value precisely because in times of uncertainty its premium is likely to remain high rather than relatively quickly disappear. This is because of its relatively low elasticities of production and substitution. Similarly, as Keynes points out, land has served historically, especially in economies that have not fully monetized, as a liquid store of value. That both land and gold, especially land, have relatively low elasticities of production as well as low carrying costs is easily understood. The question is what about a low elasticity of substitution? First consider a flight into liquidity where the demand for land surges. Uncertain as to what the future might bring (the reason for the increased demand for land in the first place), the question becomes why someone should sell land and give up the security which comes from having collateral of above average value. Only if the threats of new land being cleared or new territories being opened up present themselves will the attraction to holding land as a store of value fall. To see more clearly why a low elasticity of substitution (i.e. a low price elasticity of demand) plays an important role, compare the two assets A and B in figure 1 below. Assume that both have zero elasticities of production (i.e. that the price elasticity of supply is zero) in both the short-run and over time. Say that in times of panic you believe both will see an equal increase in demand and that you have the option of buying either A or B at the current price P0 or failing that at a price only slightly above P0 if you move quickly enough. Given increased demand and assuming zero elasticity of production both assets, A and B, will increase in price and retain their value over time. However, A is much more attractive because its value will increase a great deal more, this is true even if 9 panic buying were somehow evenly divided between A and B. However, one should assume that given the conclusion just reached that in times of panic that demand should not be evenly divided between A and B. Instead, the bulk of increased demand would be directed toward A. This would result even more of an increase in the value of A making A far more desirable, all else equal, in times of increased uncertainty. S PA,PB figure 1 P0 DB DA QA,QB Consider the value of 2 assets A and C, both of which have the same low elasticity of substitution but while in A’s case the elasticity of production is zero in C’s the elasticity of production (again this means the price elasticity of supply) is relatively high in the short-run and, because C is a constant cost industry, supply is perfectly elastic in the long-run. The results are displayed in figure 2 below. 10 SA PA’ SC SC’ PC’ P0 DA’=DC’ DA=DC Figure 2 Here not only will A appreciate more in value but it will maintain this value indefinitely. On the other hand, C will appreciate less in the short-run as suppliers already producing increase production to satisfy increased demand. In the longer-term, as new producers come into the market to supply C the above normal returns from holding C, Sraffa’s “natural” rate of interest, will disappear. An investor who bought C at PC’ and planned to hold it as a store of value stands to lose (P0-PC’)÷PC’ per dollar invested. Compare this to someone who bought A at P’C and then after C has returned to P0 is able to sell A at PA’ . Again, rather than equal increases in the demand for A and C in times of increased uncertainty all involved can reasonably expect the demand for A to increase by far more than the demand for C. 11 The price elasticity of demand for money Once one has reached the point we are at now, a serious practical difficulty in understanding chapter 17 is basically how to express the price of money. What does one write on the vertical axis of a money demand diagram? There are really two options., In figure 3 below, I draw the demand for money curve with the nominal rate of interest, Keynes’s money rate of interest consisting of the liquidity premium alone on the vertical axis. Here note that despite the demand for money being relatively elastic with respect to the nominal rate of interest, a fall in the price of other goods (brought on by the initial surge in the demand for money and, as noted often, raises the opportunity cost of holding money) leads to little change in the demand for money. As a result of the demand for money being relatively inelastic with respect to the price level, the money rate of interest is very slow to fall. It will not, barring a change in policy or investors’ sentiment, return to its original level of r0 . r1 r1’ r0 Md’ Md” Md QD$, QS$ 12 The other option is to draw the demand for money curve , Keynes’s reservations re index numbers notwithstanding, the real rate of interest, rr, on the vertical axis as in figure 4 below. Note that there are two demand for money curves drawn, DD’ and dd’ . The former traces out the impact of an increase in the real and the nominal rates, i.e. it assumes that prices do not change. The curve dd’on the other hand depicts the impact of an increase in rr because prices have fallen. The negative returns to businesses bringing their output to market (and as a result, the negative return to the assets they employ) along with the decline in the value of collateral means that the real rate of interest, the commodity rate of interest on money, rises sharply. The below average returns elsewhere, together with real returns that are well above average for money should draw resources into the money market. That money and more importantly credit are not “grown like a crop or manufactured like a motor-car” (Keynes, 1936, pp230-231) means that as the price of holding liquidity mounts the burden of adjustment falls on other markets, not financial markets. d’ d MS rr Figure 4 D’ D D” D’ d’ d” QD$, QS$ 13 Price inelastic demand for money and aggregate demand: the connection between chapters 17 and 19 of the General Theory. In Chapter 19 of the General Theory Keynes addresses the question of the efficacy of cutting wages to restore an economy, once pushed below its optimum whatever the reason might be (a decline in the overall mpc, a drop in the marginal efficiency of capital, or a rise in liquidity preference). I would like to briefly consider this question in light of the argument Keynes offers in Chapter 17 that essentially the demand for money is price inelastic and thus relatively insensitive to changes in the real rate of interest. To begin, I ask the reader to review how the textbook vision of the aggregate demand curve is obtained. Plotting the price level on the vertical axis and the total quantity of goods and services demanded on the horizontal, aggregate demand is derived from ISLM, varying the price level, observing its impact on the demand for money and therefore the position of the LM curve. As we have seen, the less sensitive the demand for money is to changes in the price level the less the demand for money curve shifts. As a result, the LM curve shifts by a smaller amount and the less any given fall in prices will lower interest rates and therefore stimulate spending. In short, the lower the elasticity of substitution for money is, the more price inelastic the aggregate demand a curve will be. Now imagine a negative demand stock that pushes AD to the left as in figure 5 below. In both the Alpha and Beta economies, equilibrium output, prices, and employment will fall. However, in the Alpha economy the wage cuts required to restore 14 equilibrium in both the goods market and the labor market and restore the economy to its optimum are considerably less than in Beta. Figure 5 P LRAS AS AS’ ADβ ADα AD’β AD’α Yfull QD, QS It is extremely surprising that Keynes does not connect chapter 17 to his discussion in chapter 19 of money wage cuts failing to restore full employment and the economy to its social optimum. And Keynes is flatly wrong, disregarding the argument so painfully developed in chapter 17 when he writes “(W)e can, therefore, theoretically at least, produce precisely the same effects on the rate of interest by reducing wages, whilst leaving the quantity of money unchanged that we can produce by increasing the quantity of money whilst leaving the level of wages unchanged.” (Keynes, 1976, p.266). In the context of figures 3 and 5 above, the rightward shift in aggregate supply required to bring the price level and hence the demand for money down, restoring the money rate of interest to r0 in figure 3 would be considerable. Because of this it would be far more 15 painful and therefore less likely to succeed than simply increasing the money supply. A fall in wages cannot produce the same effect on the money rate of interest that an increase in the money supply can. In passing Keynes does refer in chapter 17 to the impact of falling wages on the marginal efficiency of capital. In chapter 19 he much more fully develops this noting that the marginal efficiency of capital may fall by far more than interest rates as falling wages redistribute income from spenders to those with lower propensities to consume and as falling prices raise the burden if left. These would push an already relatively unresponsive aggregate demand curve to the left as wages fell. They would provide additional reasons for being skeptical about relying on labor markets to adjust to a suboptimal equilibrium. The shape of the aggregate demand curve in contrast flows directly from the existence of a safe haven, a liquid asset, as defined in chapter 17, that in times of uncertainty investors can turn.