In comparison with other problems of development the question of the proper management of the economy’s money may at first glance appear to be of secondary importance. Yet the history of money is full of examples of monetary disorders which have nullified the economic efforts of a whole nation for months and years. Even in less extreme cases inappropriate monetary policies may deprive an economy of part of the fruits of its development effort, no matter how excellent the development program is in other respects. For this reason it is important to incorporate considerations of monetary management into the general economic development plans. This in turn requires a quantitative assessment of the relations between money and other economic magnitudes.

In comparison with other problems of development the question of the proper management of the economy’s money may at first glance appear to be of secondary importance. Yet the history of money is full of examples of monetary disorders which have nullified the economic efforts of a whole nation for months and years. Even in less extreme cases inappropriate monetary policies may deprive an economy of part of the fruits of its development effort, no matter how excellent the development program is in other respects. For this reason it is important to incorporate considerations of monetary management into the general economic development plans. This in turn requires a quantitative assessment of the relations between money and other economic magnitudes.

I. Money, National Income, and the Balance of Payments

The last decade has seen a revival of interest in monetary theory and policy. Modern monetary economics is leaving behind not only the simplifications of the original quantity theory (‛a change in money affects only the price level’), but also the confining special case which Keynes’ early followers had pulled out of the General Theory (‛a change in money affects, in the first place, only the interest rate’). The new approach is more general. Money is one of a number of real and financial assets which households and businesses hold. Starting from an analysis of the factors on which the public’s demand for money and for alternative assets depends, one can describe the probable reaction of the private sector of the economy to changes in the supply of money.

A. The demand for money.—The public holds money balances (currency and demand deposits at commercial banks) for two principal reasons. First, cash balances must be held by households and business firms in order to bridge the temporal gap between receipts and payments, the size of the cash requirements for this ‛transactions’ purpose depending on the customary payments arrangements in the economy–the frequency of wage payments, and of payments for rent, food, and so on. Changes in the demand for transactions balances may be presumed to be more or less proportional to the value of the national income at current prices, or better, to that portion of it which is bought and sold in the market1

Second, individuals and businesses may hold a part of their assets in the form of money balances. Such idle balances, in contrast to other assets, do not earn interest, dividends, or profits, but compensate the holder for this loss of potential income by their perfect liquidity.2 The importance which the individual holder attaches to the liquidity of the marginal dollar diminishes as the proportion of his assets held in the form of money increases. Consequently, given their income and the total value of their assets, economic units will be willing to increase their holdings of cash balances if the importance they attach to liquidity as such rises, say, because of increased uncertainty about asset prices in the future, or if the expected yields on alternative real and financial assets (including expected changes in their capital values) decline.

The demand for money as an asset depends thus on the aggregate value of all assets and on the yields of assets other than money. Given the payments customs of the society, the total demand for money, both for transactions purposes and for purposes of asset holdings,3 rises with an increase in income and in the aggregate value of assets, and declines with a rise in the yields or the expected appreciation of other assets. It is, however, not possible, in practice, and perhaps not even in principle, to distinguish the effect of the income level on the demand for money from the effect of the total value of assets, for—given the yields of the assets—income and the value of assets are closely related. The level of assets is thus best left out of consideration; its influence will then be reflected in the two remaining factors, income and asset yields.

B. The supply of money.—We must now turn to the question of how the money supply is determined. Since a later section reviews the instruments of central bank policy, the discussion can be held brief at this stage. Provided that there is sufficient demand for credit at the going interest rate at which the commercial banks make loans to their customers, the volume of loans and thus of bank deposits will be a certain multiple of the reserve assets of the banks. This multiple depends on the statutory or customary reserve ratio to which the banks adhere. The reserve assets of the commercial banks usually comprise mainly their deposits at the central bank and their foreign exchange assets. Where the banks are required to surrender most or all of their foreign exchange holdings to the central bank or to a separate exchange control authority, they obtain additional central bank deposits in return. In either case commercial bank reserves will vary with changes in the country’s foreign exchange reserves.

The other set of influences on the reserve assets of commercial banks, and therefore on the money supply, can be subsumed under the heading of central bank credit. By making loans to, or buying assets from, the commercial banks, the public, or the government, the central bank can bring about an increase in the money supply either directly or via an increase in commercial bank reserves, and by reversing these operations it can accomplish a reduction in the money supply.4

It is important not to exaggerate the precision with which the central bank can control the money supply. Apart from the effect of the balance of payments, the reaction of the commercial banks to central bank credit changes must be taken into consideration.5 If the commercial banks always extended credit to the full extent permitted by their reserve assets, we could link the money supply by a simple formula to the combined amount of the nation’s foreign exchange reserves and central bank credit. In practice, however, banks will expand the volume of their loans in the short run only if there is sufficient demand at the going bank loan interest rates, and will reduce these rates in order to stimulate demand only gradually if excess reserves persist for some time. For now we may content ourselves with the statement that the money supply will in the long run depend on the commercial bank reserve ratio and on the sum of the central bank credit and of the country’s gold and foreign exchange reserves; and while it can never exceed the level so determined, it may in the short run occasionally remain below it.

C. Money and spending.—In tracing the effects of changes in the money supply on the economy it is convenient to take as point of departure a hypothetical equilibrium situation in which the money supply is equal to the ‘long-run normal’ demand for money. In such a situation the real national income will, by hypothesis, be at its long-run normal level, given the country’s resources, its stage of development, and the average levels of its exports and of its capital imports, if any. The price level and the foreign exchange rate will be constant, and all asset yields and interest rates will equal the marginal productivity of capital goods minus the appropriate liquidity premium for each type of asset.

Such a long-run equilibrium situation is unlikely ever to exist in reality, since the monetary equilibrium is constantly disturbed by changes in the money supply or by short-run changes in the factors which determine the demand for money. But these disturbances will produce reactions in the economy which will tend to re-establish monetary equilibrium either by changing the demand for money or its supply, or both. It is, therefore, possible to describe the course of economic events in a country as a path along which the sectors of the economy react to disturbances of the monetary equilibrium.

The mechanism by which this readjustment takes place is simple: whenever the quantity of money actually held exceeds (or falls short of) the long-run normal demand for money, individuals and businesses will attempt to reduce (or increase) their cash balances by purchasing (or selling) other assets.

The adjustment process is as follows:

  • 1. To the extent that the additional money is spent directly on foreign securities or foreign goods, foreign exchange reserves and the money supply will be reduced. (In the unlikely event that the entire addition to money would be spent on imported goods and securities, the original money supply would be restored and this would be the end of the story.)

  • 2. To the extent that the additional money is spent on domestically produced goods (producers’ or consumers’ durables or inventories of non-durables to be held in business or household stocks), money national income will rise, either as a result of a rise in output or of price increases, with three consequences: first, a leakage into imports as under (I); second, an increased demand for cash balances;6 third, higher prices for real capital assets implying reduced yields and thus a larger demand for cash balances, provided that expected future prices of these capital goods and of the goods they help to produce are not revised upward in proportion to the rise in present asset prices.

  • 3. Finally, to the extent that cash holders attempt to spend the additional money on domestic securities, they will bid up their prices and reduce their yields without, of course, being able to reduce the cash balances which they, as a group, hold.7 At the lower interest yields of securities the demand for money will be increased. This is the pure Keynesian short-run case, where interest rates fall until the demand for money has increased sufficiently to absorb the rise in the money supply. But even in this case one should expect two further reactions in the longer run: first, the lower domestic interest rates may tend to induce a capital outflow or a reduction in the capital inflow with consequences as under (I) above. Second, the lower yields on securities will lead to additional spending on real assets (say, capital goods and inventories), as long as the marginal productivity of real capital is unchanged. In other words, there will be a tendency to return to an equilibrium between the yields on financial and real assets through sales of securities and purchases of capital goods, including inventories, with further consequences as under (2).

A similar analysis would show how monetary equilibrium tends to be re-established following different types of disturbances, such as a loss of export income, an increase in capital inflows, a budgetary deficit, or an autonomous wage increase. The only difference in the analysis would be that one would have to allow for the initial impact of these disturbances on national income and on asset yields, and thus on the demand for money, before tracing the further monetary consequences in the manner exemplified above.

D. Summary.—The scheme of monetary analysis sketched in this section is not, by itself, sufficient for a full appraisal or a forecast of short-term economic processes. As we have seen, there are several ways in which the public may react to, say, an increase in domestic credit creation or to a decline in the capital inflow. In order to go beyond a listing of possible outcomes, or an explanation from hindsight of events in the past, it is necessary to obtain some quantitative knowledge of the likely reactions of households and businesses to changes in the magnitudes which the discussion has singled out for special consideration.

Some of the quantitative knowledge may be quite easily accessible in the case of a particular economy at a given time. For instance, if capital movements are tightly controlled, the probable capital outflow as a result of an increase in the money supply may be put down as zero. Again, in many of the less developed countries the market for securities is quite ‘thin’ or non-existent, and the main alternative assets are money, foreign exchange assets, and real assets (real estate and goods); in these cases a whole class of possible reactions to an increase in the money supply can probably be ignored.8 By making use of known features of this sort it is usually possible to restrict the range of possible outcomes of a given situation. Beyond this it is necessary to base the analysis on experience as reflected in the available data and on one’s judgement as to the acceptability of past experience as a guide to the future.

When it comes to an assessment of some of the longer-run consequences of changes in the factors on which a country’s economic activity depends, we are on somewhat firmer ground. The reason for this is that the final outcome does not depend on the precise magnitude or timing of the shortrun reactions of the economic units. We know, for instance, that (under fixed exchange rates) the money supply will tend to increase (decrease) as long as there is a balance of payments surplus (deficit), unless these externally caused changes in money are constantly offset by changes in central bank credit. We know further that it is impossible in the long run to offset the effect of a continual external deficit on the money supply through central bank credit, since in that case the country must soon run out of foreign exchange reserves. Nor can the monetary effect of a persistent external surplus be offset indefinitely, at least in practice, since the central bank would sooner or later run out of domestic assets to sell.9 It follows that in the long run imports will equal exports plus capital inflows, money income will tend to a level such that imports conform to this equality, and the money supply will adjust itself to a level at which it satisfies the demand for money at that money income (and at the long-run level of the yield of capital).

Although this simple scheme leaves a number of important questions unanswered, there are others that it answers quite adequately: (a) What ever the influence of monetary policy may be in the short run, in the long run it cannot have a large role to play in the determination of income and imports, given the restraints imposed by the foreign balance. This is not to deny that inappropriate monetary policies during successive short-run periods may disturb the adjustment of the economy to its long-run equilibrium, i.e., to its growth path, and thus make matters worse than they would otherwise have been. (b) A permanent rise in exports will raise imports by an equal amount, 10 and it will raise money income by an amount which depends on the income and price elasticities of import demand, but does not depend on the conventional Keynesian multiplier. If output would have been at its full capacity level without the rise in exports, the increase in the money income will be mainly in the form of price increases. (c) A permanent rise in capital imports, say, in the form of foreign aid, will have the same effect on imports and money income as a permanent rise in exports. (d) A permanent rise in the propensity to import will bring about a decline in money income, which may manifest itself either as a reduction in (the growth path of) output or as a decline in the price level, or a combination of both.

As J. J. Polak has pointed out,11 these are essentially the conclusions of the classical economists analyzing foreign trade and lending under gold standard conditions. This is not surprising since at the present level of abstraction the classical assumptions are indistinguishable from those underlying the long-run scheme set out above: fixed exchange rates, the money supply determined by the foreign balance, money income determined by the money supply, and imports determined by money income. But in the analysis of year-to-year changes in money, income and imports some of these assumptions must be modified. We return to this topic in Section IV.

II. Empirical Evidence

A. The relation between imports and income.—The relation between imports and income has been extensively investigated,12 both for the interwar period and the post-war period. As a result of these studies there can be no reasonable doubt that, barring unusually incisive exchange control measures, imports rise with income. Furthermore, the income elasticity of imports is typically greater than unity. Table I shows the marginal pro-

Table I

Marginal Propensity to Import and Income Elasticity of Import Demand, Twenty-One Countries, Generally 1949-60a

article image

Data: Commodity imports deflated by import prices; GNP deflated by cost of living index. Source: International Financial Statistics.

Regression coefficient b of the function M = a + bY, where M is real imports and Y is real GNP.

Regression coefficient b of the function log M = a + b log Y.

Note: In all cases R2 exceeds 0.85 (for most countries it exceeds 0.90), and the regression coefficients are at least five times as large as their standard errors (in most cases they are ten to twenty times as large).

pensity to import and the income elasticity of imports in twenty-one countries for which these values could be reliably estimated 13 from postwar data.

National income data for longer periods of time are not available for most countries. Moreover, the structural changes which occur over a longer time period, especially during and after major wars, make it difficult to estimate the long-run relation between imports and income.14

B. The long-run relation between money and income.—In this area, too, it is difficult to obtain consistent time series covering a longer period. From the data which they have been able to piece together, some students of this field have concluded that money tends to increase faster than income and that the income velocity of money declines with the secular rise in income. There is little doubt that this conclusion is justified in the case of the early development of the United States, where the ratio of income to money has declined considerably from 1800 to the present century. The evidence on the behaviour of the income velocity is less conclusive with respect to the last sixty or eighty years.15 In accordance with the reasoning in Section I we should, however, not look for a stable ratio of money to income but for a stable relation of money to income and the long-term yield of capital.

For the United States reasonably consistent data on money (currency plus demand deposits) are available for several decades. A long-run demand function making the demand for real cash balances depend on real gross national product and the long-term corporate bond yield (as an admittedly imperfect substitute for the yield on capital assets in general) has been fitted to the annual data for 1900 to 1961:


where L = money, Y = gross national product, P = price level (GNP price deflator, 1929 = 1), and i = bond yield in per cent.16 The equation shows that the income elasticity of the demand for money was almost precisely 1 (with a negligibly small standard error of ± 0*02), while the interest elasticity of the demand for money was—0*65 (with a standard error of less than one-tenth of the size of the coefficient).17 Over the last six decades the demand for real cash balances in the United States tended thus to vary in proportion to real income and was inversely related to the long-term interest rate. At an earlier stage of development, however, when the whole financial system develops more rapidly than the rest of the economy, the demand for money is also likely to rise faster than income.18

III. Economic Growth, Money, and Prices

In a closed economy the money supply should increase at a rate which is sufficient to satisfy the growing demand for cash balances which the rise in real income with stable prices brings about. But in an economy in which foreign trade plays a large role an additional factor must be considered. With a given domestic cost and price level, the rate of growth of a country’s exports is not necessarily equal to the rate of increase in its imports which will result when output and employment are maintained at satisfactory levels. In the construction of long-range plans it is, therefore, important to allow for this possibility, and to establish overall guidelines for monetary policy which are compatible with the appraisal of long-term domestic and foreign developments.

Given a long-run demand function for money of the type discussed in the preceding section, there is a rate of growth of the money supply which will tend to maintain domestic price stability. Suppose that real income grows at a rate of y per cent and that the marginal productivity of capital and the level of interest rates remain constant over time.19 If the income elasticity of demand for real cash balances is g, then the rate at which the money supply must grow to maintain a constant domestic price level is yg. As was argued in Section II, it is likely that in the early stages of development g is greater than one, since the financial institutions will develop more rapidly than the rest of the economy and since a growing portion of the nation’s output will be exchanged against money at some stage of the process of production and distribution. In this case the money supply should grow faster than output in order to prevent deflationary tendencies. At a later stage of economic development the value of g is likely to be closer to unity and the money supply should then grow at approximately the same rate as output.

It has sometimes been argued that a rate of growth of the money supply which induces some price inflation may also raise the rate of growth of output. But while it may be possible for a relatively short period to increase the levels of saving and capital formation through the redistribution of income which price inflation brings about (i.e., through the socalled process of’forced saving’), this cannot result in a permanent increase in the rate of economic growth. Inflation is a relatively inefficient form of taxation, since the speed of inflation would have to continue to accelerate merely to provide a constant amount of additional saving. An accelerating rate of inflation must soon result either in the application of restrictions or in the breakdown of the market system and what little may have been initially gained will then be lost as the economy is subjected to a painful readjustment process.

These considerations apply particularly strongly to economies in which foreign trade plays a large role. Table I (Section II) shows that many countries are highly dependent on imports, and a computation of the proportion of national income derived from exports would, of course, give a similar picture. Exports are often concentrated in a few specialized products. Only a small portion of the output of these industries can be utilized by the domestic economy. In these circumstances domestic cost and price inflation would rapidly lead not only to a loss of the country’s foreign exchange reserves, but to a reduction in output and employment in many of the export industries as well.

As already shown, in an economy which is highly dependent on foreign trade, as most of the underdeveloped economies are, the money supply will in fact tend to grow at a rate which, over the long run, maintains the foreign balance roughly in equilibrium. The monetary authorities can intervene to smooth short-run fluctuations in the money supply, but they cannot indefinitely counteract the long-run trend unless they are also prepared to alter the foreign exchange rate from time to time. The alternation of ‛stop’ and ‛go’ signals which domestic producers receive under a policy of domestic price inflation with intermittent exchange depreciation favours neither development nor the inflow of foreign capital. In the long-run development effort monetary policy must remain a relatively passive factor, conducted with the aim of avoiding disruptive monetary disturbances both of the inflationary and of the deflationary varieties, rather than with the objective of making it an independent stimulant to economic growth.

IV. Money, Income, and the Foreign Balance in the Short Run

Monetary policy is more effective in influencing the level of real income in the short run because of the absence of some of the restraints which arc operative in the long run. First, in the short run imports need not equal the sum of exports and foreign receipts on capital account. Second, although the foreign balance does affect the money supply, this effect can be offset by central bank policy, at least for some time, provided (in the case of a foreign deficit) that the country has an adequate level of foreign exchange reserves. Third, in the short run the level of income and employment is subject to fluctuations, either due to variations in exports or to changes in domestic conditions, which monetary policy can attempt to counteract, whereas it cannot be nearly as effective in influencing the long-run growth path of economic activity.

A. Features of a simple short-run model.—To assess the influence of a change in the quantity of money on national income during a short period, one must estimate primarily two behavior relations. First, one must know how much of the additional money supplied will tend to be wiped out through an induced increase in imports and a concomitant decline in exchange reserves. Second, one must find out the extent to which the remaining addition to the money supply will increase the public’s spending on additional domestic output (investment goods, inventories, consumer goods, etc.), and to what extent it will merely lead to higher prices, and thus reduced yields, of financial and real assets.

These estimates must then be combined in a fairly intricate manner: the drain on the money supply due to the induced rise in imports cannot be known until we determine the rise in income, which in turn is due only to the portion of the additional money supply not drained away through higher imports. In other words, imports, income, and the total money supply are simultaneously determined by a number of other variables which we may take as more or less independent, such as exports, domestic credit creation, and in some cases 20 perhaps also the net inflow of foreign capital.

In principle there is no limit to the number of refinements one can introduce into such a model. But if data are limited, as well as computing facilities, it should be kept as simple as possible. Moreover, there is virtue in starting with a simple model and expanding it only gradually as one gains experience in its application.

Guided by these considerations, J.J. Polak has constructed a model 21 in which the income velocity of circulation is assumed to be constant (at its average level for the post-war period) and in which money imports are determined by money income. The estimated annual values of income and imports move in response to changes in the annual total of exports, net capital inflows, and changes in domestic credit creation. The key assumption of this model is the constancy of the income-money ratio. With a given long-term yield of capital, the constant income velocity can be interpreted as the long-run demand function for money. It is, therefore, not surprising that the estimates for income and imports obtained from the Polak model are found to correspond quite well to the average level of the actually observed values of income and imports over several years, but that they do not follow the year to year variations in the observed values very closely. In other words, the estimated values give the appearance of moving averages of the actual values. In order to get a closer explanation of the year to year changes in income and imports we must try to estimate more nearly the short-run relation between money and income. We can do this only at the sacrifice of some of the simplicity of the Polak model.22

B. Money and expenditure in the short run.—In the familiar Kcynesian model private expenditure (i.e., consumption plus investment) is assumed to depend on income and the rate of interest. An increase in the quantity of money, other things being equal, is expected to affect private expenditure by first depressing the interest rate, which in turn will stimulate (investment) expenditure. In other words, the public first bids up security prices but not the prices of any other assets and it is only the concomitant reduction in the cost of borrowing23 which leads to an increase in private expenditure. Since long-term borrowing rates are notoriously inflexible, many economists have come to doubt that an increase in the money supply will have an appreciable influence on spending and thus on income.

This reasoning ignores the change in demand for assets other than securities which must have accompanied the rise in the demand schedule for securities unless some kind of irrational aversion towards real assets is assumed. Some of these other assets are, like securities, in temporarily or permanently inelastic supply (real estate, art treasures, etc.). But others, such as capital goods, including inventories of raw materials, and stocks of consumer goods have supply schedules whose price elasticity is always positive and increases with the length of the period considered. It is a well-known principle of demand analysis that when spending in general increases, more of the additional expenditure is devoted to a good in elastic supply than to one in inelastic supply, if the two goods are close substitutes for each other,24 since the prices of the former will rise less than those of the latter for every additional dollar spent on each.

When interest rates decline and land values rise as a consequence of an increase in the money supply, an increase in spending on other assets must occur simultaneously, unless the demand for these assets has by chance declined for reasons unrelated to the increase in the quantity of money. The significance of the interest rate in the Keynesian expenditure function is not so much that it reflects changes in the cost of borrowing, but that it is an index of changes in expenditures on all types of assets, including securities and investment goods. The reciprocal of a real estate price index would, except for market imperfections, serve equally well as a variable in the investment function.

Rather than include in the expenditure function an interest variable which reflects the extent to which an increment of money was not spent on real assets (but instead on securities or not at all), it may be more straightforward to include instead the increment of money itself. The coefficient of this variable will then indicate the additional spending on goods which was necessary during a given period of time in order to lower the yields of these real assets (by raising their prices or by increasing their volume, or both) by a sufficient amount so as to induce the public to hold the larger quantity of money. This additional expenditure may be a fraction or a multiple of the original excess supply of money, depending on the length of the period considered, on the initial ratio of money to total assets, and on the elasticities of demand for and supply of the various asset classes, including money itself. While these individual elasticities will remain hidden behind the scenes, this approach should furnish a clue to the effect of an increase in the quantity of money on spending and thus on income, even in countries where security markets are thin and where published interest rates are not representative of the true yields of real and financial assets.

C. The basic equations of the model.—Assuming the long-run yield on capital to be constant, the long-run demand for money can be expressed as


where L*are the desired cash balances which the public would wish to hold in the long run at an income level of Y, and where k is a constant. The deviation of the actual money supply from this desired level in any particular year t is Lt—L*t. Private expenditure, E, is linearly dependent on current and last year’s income and on the excess of actual over desired cash balances 25 some time ago.

Since Lt is the stock of money measured at a moment of time (at the end of year t) while Yt is the flow of income during year t, we express the average desired cash balances during year t as (L*t + L*t_1)/2 = kYt, and the deviation of actual from desired cash balances during year t as [(Lt + Lt_1)/2—kYt]. The private expenditure function is thus


if there is no lag of expenditure with respect to a change in the excess of desired over actual cash balances.26 If there is a one year lag, the term in square brackets is replaced by [(Lt-1 + Lt-2)/2 - kYt_1], and if the lag is one half-year it becomes [(Lt-1—k(Yt + Yt_1)/2]. The most promising form of the expenditure function would seem to be that incorporating a lag of one half-year or of one year, which gives adequate time for orders, price lists, production plans, and actual output to change.27

The model contains in addition an import function,


which expresses imports as a function of expenditure (rather than income). Government expenditures on goods and services, G, may be taken as autonomously determined or they may be related to income, recognizing the fact that G must depend to a considerable extent on tax revenue which is itself a function of income:


The model is completed by two identities defining income and the money supply:



where Xt = exports, Ct = net capital imports, and ΔDt= change in domestic credit creation (i.e., the change in money of domestic origin). These identities and the non-observable relation (IV-1) are the same for all countries. The behaviour equations (IV-2), (IV-3), and (IV-4) are given below for Norway, and in the Appendix (Tables A1 and A2) for four other countries (Costa Rica, Ecuador, Japan, and the Netherlands). The expenditure equation (IV-2) is estimated 28 in a form in which the coefficients of Yt or of Yt_x are combined into one coefficient, say, (a2—ka3)Yt_1.

D. An example: Norway, 1949-60.—The variables appearing below are measured in billions of kroner.


By combining these equations we can express each (current) dependent variable as a function of the independent and the lagged dependent variables. (The corresponding equations for the other four countries are given in Appendix Table A3.)


This formulation shows that a rise (or fall) in Norway’s current exports has tended to raise (or lower) current money income by about vS times the export change, and current imports by about one-half of the current export change. A rise (or fall) in the money supply has tended to raise (or lower) money income the following year by two-thirds of the change in the money supply, and imports by almost three-fourths of the money change.29 The change in money during a particular year depends, of course, itself on exports, capital inflows, imports, and the change in domestic credit creation in that year. The effect of a change in money in the preceding year on current money income and imports can be carried back to these other variables by substituting them for (Lt_1—Lt_2) in the two equations above. The import equation shows particularly clearly the extent to which it may be necessary to allow for the future effect of present monetary action on the foreign balance.

E. Conclusion——In Norway as well as in the other four countries, a change in the money supply appears to affect expenditure appreciably with an average lag of one year (one-half year for Japan). The statistical significance of the coefficient of the money variable is, however, at a lower level than that of some of the other coefficients of the model. In Norway, a rise in the money supply tends to raise private expenditure (consumption plus investment) in the following year by an amount which is about 90 per cent of the increase in money. In the Netherlands, in Costa Rica, and in Ecuador, this proportion was higher, while in Japan it was much lower (and not significantly different from zero in the probability sense).

The statistical fit of the model as a whole is uniformly better for the five countries than that of the Polak model, as shown below:

Standard error of computed income*

(Per cent of average income)

article image

Root-mean-square deviation of computed from actual income.

These figures give a rough idea of the average percentage error that one would be likely to make in forecasting income, if exports, capital imports, and the change in domestic credit creation were known in advance. Since these values must themselves be estimated for any future period, the actual forecasting error will probably tend to be higher.

The model is, however, not specifically designed for the purpose of forecasting, but rather to give a general understanding of the magnitudes and the timing of the short-run effects of autonomous changes in the money supply (domestic credit creation), as well as of changes in exports and in the capital account of the balance of payments. Having satisfied ourselves that changes in the money supply do have an effect on spending, income, and imports in the short run, we turn now to a brief discussion of the means by which the central bank can influence the money supply.

V. Central Bank Control of the Money Supply

A number of underdeveloped countries do not have a national currency of their own but employ instead a foreign currency as the circulating medium. In some other instances the so-called currency-board system is in effect, under which local currency is issued in exchange for foreign currency and other foreign exchange assets. In both of these cases the amount of circulating currency is always equal to the country’s foreign exchange reserves. Moreover, the commercial banks tend to expand or contract their loans in accordance with variations in their reserves which consist of domestic currency and foreign exchange. In these circumstances, the money supply depends entirely on the foreign balance, not only in the long-run but all the time, and the direct effects of export fluctuations on national income tend to be magnified by the monetary effects of variations in the external balance.

As has already been indicated in Section I, a central bank can offset the influence of the foreign balance on the money supply by altering either its holdings of domestic assets or by changing the commercial banks’ ratios of reserve assets to deposit liabilities. There are three widely used instruments by which central banks influence the money supply. First, the central bank can vary the interest rate (’discount rate’ or ‘bank rate’) and other terms at which it buys (’discounts’) private short-term debt instruments (’commercial paper’) from the banks and at which it makes loans to the banks, thus influencing their loan policies.

Secondly, the central bank can raise commercial bank reserves by buying government securities (or other securities specified in the Central Bank Law) either directly from the commercial banks or from the general public, and it can achieve the opposite effect by selling such securities. If these ‘open market operations’ are carried out with the non-bank public rather than with the commercial banks, there will be a direct effect on the public’s cash holdings in addition to the indirect effect via commercial bank reserves.

Thirdly, in many countries the central bank may prescribe a minimum ratio of bank reserves to deposit liabilities and may vary this ratio within certain limits set by law, thus limiting the potential maximum volume of outstanding commercial bank loans at a given level of bank reserves.

The extent to which these instruments can be applied depends on the degree of development of the financial structure of the economy. In many underdeveloped countries the volume of commercial paper and of government securities is small or non-existent. The commercial banks in some countries are branches of foreign banking institutions which can augment their reserves by borrowing from their head offices and need not rely on credit from the central bank. Moreover, in underdeveloped countries a large part—say, between one-half and four-fifths and in some countries probably more—of the money supply is held in the form of currency, so that measures which affect primarily the volume of bank deposits will have a proportionately smaller effect on the total money supply.

While the scope for monetary policy is reduced by these conditions, there would seem to be some compensation: although it is more difficult for the central bank to change the money supply, a given change may have a larger effect on spending and national income than it would have with more fully developed financial institutions and money markets, because of the absence of a number of the types of financial assets which residents of more developed economies can hold as alternatives to money. The asset holders’ choice is thus restricted to holding either money or goods. There is, however, an important qualification to this statement. If the asset holders have unrestricted and convenient opportunities to buy or sell foreign securities, the range of their choice is at least as wide as it would be if there were a market for domestic securities. In fact, when businesses and the group of wealth-holding individuals are accustomed to holding a varying portion of their assets in foreign securities or foreign time deposits, the attempt to control the money supply may at times be frustrated by variations in capital movements which offset the effects of the central bank’s policy measures.30

The problem of monetary policy in a developed economy has been described by saying that ‘you can pull on a string but you cannot push on a string’. This means that it is possible to prevent the money supply from exceeding a desired level by restrictive central bank policies which force the commercial banks to restrain the volume of their loans, but that the central bank cannot force an expansion in the money supply in all circumstances, since it can neither force the banks to expand their loans nor make the businessmen borrow. When monetary and budgetary policies are considered jointly, one might say that the reverse difficulty exists in countries with underdeveloped financial markets. When monetary expansion is indicated, it can always be achieved by expanding government expenditures on one or several of the many pressing development projects and by financing the additional spending through an issue of government securities sold to the central bank. But when inflationary pressures make it desirable to restrict the money supply it may prove difficult for the central bank to sell government securities to the commercial banks or to the public. For the reasons already given, the attempt to reduce the money supply through the use of the other instruments may also be unsuccessful. This asymmetry adds to the inflationary bias which is already present in developing economies.

In view of the limited applicability of the traditional instruments of central bank policy in the case of countries with undeveloped financial institutions, some countries have had recourse to measures designed to restrain the activity of particular sectors of the economy or to put maximal limits (’ceilings’) on the amount of credit which banks may make available to the economy.31 A particular technique which has been applied in a number of countries in recent years is that of advance deposits on imports.32 In the frame of this paper it is not possible to comment in detail on these special techniques. Credit ceilings may be useful in times of strong inflationary stress, but are not very helpful as a means of continuous rational control of the money supply. The experience with advance deposits on imports is not entirely favourable. While it may bring quick results by sharply reducing excess liquidity, the continuous application of this instrument requires delicate timing and presents many pitfalls. The experience with advance deposits on imports is evaluated in the first of the two articles given in the last footnote.


It will be seen, then, that there are many special problems confronting the authorities in a newly emerging country, if they wish to use monetary policy for short-run stability and to permit maximum growth. What may be called for in these cases is an initial co-operative effort of the central authorities and the commercial banks to establish a set of suitable procedures for monetary control. In the western literature on monetary policy such co-operation is sometimes listed as a separate type of policy instrument under the infelicitous name of ‘moral suasion’. 33 When the number of commercial banks is small it would not seem impossible to reach agreement on desirable changes in loan policies, on reserve ratios, or discount policy, on the holding of government securities by the commercial banks, and on similar matters. After some of the financial markets have developed, partly as a result of this co-operation, less reliance would have to be placed on the direct personal contact between the central bank and the commercial banks and increasing use may be made of the traditional instruments of central bank policy.


Tables AI and A2 give the estimates of equation (IV-2) to (IV-4) in Section IV for Costa Rica, Ecuador, Japan, and the Netherlands. Table A3 gives the reduced form equations for income and imports for the four countries.

Table AI

Expenditure Functions (IV-2)

article image

The money variable is Lt-1 in the case of Japan (implying a lag of one-half year).

Table A2

Imports and Government Expenditures

article image
Table A3

Reduced Form Equations for Income and Imports

article image

Coefficient of Lt-1


With kind permission of the publisher, Basil Blackwell, reproduced from Economic Development in Africa, ed. by E.F. Jackson (Oxford, England, 1965), pp. 254-75.


In developing economies part of the upward trend in the demand for money is due to the fact that, as development progresses, an increasing portion of output is channelled through the market and exchanged for money.


Perfect liquidity is the property of an asset that allows it to be used instantly for the purchase of any other asset without loss. The liquidity of assets other than money diminishes with the market loss that must be expected if they are to be sold instantly for the purpose of acquiring another asset. Thus government bonds are less liquid than money, but (ordinarily) more liquid than inventories of goods, which are in turn more liquid than the machinery in a factory.


In a very elegant derivation Milton Friedman subsumes the transactions demand under the demand for money as an asset since the former depends on the same factors as the latter. For example, higher yields on alternative assets will induce changes in payments habits designed to economize transactions balances. [Studies in the Quantity Theory of Money (M. Friedman, editor), University of Chicago Press, 1956. Chapter I.] From the long-run viewpoint Friedman is no doubt right, but since the payments customs normally change only slowly, the concept of a transactions demand may still have a place in short-run analysis.


Textbooks on money and banking give a precise formula for the ‘money multiplier’, in which the ratio of currency to demand deposits and other details are taken into account. The exact formula can be given only with reference to the banking legislation and the institutional features of the country in question.


See J. Ahrensdorf and S. Kanesathasan, ‘Variations in the Money Multiplier and Their Implications for Central Banking,’ IMF, Staff Papers, November 1960


The nominal cash balances demanded change in proportion to the price level so that the real cash balances (i.e., cash balances deflated by the price index) remain constant. If real income also rises, the demand for real cash balances will itself increase.


The attempt on the part of the cash holders to acquire additional real estate or (domestic) art treasures would have similar consequences as the attempt to buy additional securities.


See J. J. Polak, ‘Monetary Analysis of Income Formation and Payments Problems,’ IMF Staff Papers, November 1957, and J. J. Polak and L. Boissonneault, ‘Monetary Analysis of Income and Imports and its Statistical Application,’ IMF Staff Papers, April 1960. In Polak’s approach use is made of the assumption that changes in the desire to bold securities (and physical assets not currently produced, such as real estate) can be ignored even in the short run. This makes it possible to postulate that money income will change in proportion to the money supply. See also Section IV below.


In principle, a central bank could continue indefinitely to sell obligations of its own which are specially introduced for this purpose, or raise reserve requirements if necessary to 100 per cent or higher.


Note that the Keynesian model leads to the conclusion that imports will rise by less than exports provided that the marginal propensity to spend is less than unity. The difference is due to the fact that the Keynesian model does not allow for the effect of the foreign balance on the money supply.


J. J. Polak, op. cit., p. 29.


See the survey of statistical results by H. S. Cheng, ‘statistical Estimates of Elasticities and Propensities in International Trade: A Survey of Published Studies’, IMF Staff Papers, April 1959.


The criteria of reliability chosen were that income variations explain 85 per cent or more of the observed variations in imports, and that the standard error of the estimated coefficient be reasonably small relative to the value of the coefficient (one fifth or smaller).


It should be remembered that Eugene Staley has argued that commodity imports will tend to rise less fast than GNP as an economy develops, since the latter will contain an increasing proportion of services which are not internationally traded. (See his World Economic Development, International Labour Office, Montreal, 1944.) But it is, of course, entirely possible that imports will tend to rise more than in proportion to the nonservice components of GNP.


See, e.g., Ernest M. Doblin, ‘The Ratio of Income to Money Supply; An International Survey’, Review of Economics and Statistics, August 1951, and Richard T. Selden, ‘Monetary Velocity in the United States’, in Studies in the Quantity Theory of Money (M. Friedman, editor), University of Chicago Press, 1956.


The data for L/P and Y/P are taken from Edward S. Shaw, ‘Money Supply and Stable Economic Growth,’ Chapter 2 of United States Monetary Policy, The American Assembly, 1958, pp. 55-56 for 1900-28 and from the Federal Reserve Bulletin (L) and the Survey of Current Business (Y and P, which were then converted into 1929 prices) from 1929-61. The yield, i, on 20-year highest-grade corporate bonds is taken from William Fellner, Trends and Cycles in Economic Activity, Holt, New York: 1953, for 1900-45, and from the Federal Reserve Bulletin for 1946-60.


For a different approach see Milton Friedman, The Demand for Money: Some Theoretical and Empirical Results, Occasional Paper 68, National Bureau of Economic Research, New York, 1959. Our results are in line with the findings of Henry A. Latane, ‘Income Velocity and Interest Rates: A Pragmatic Approach’, Review of Economics and Statistics, November i960. Latane’s article also contains a discussion of Friedman’s approach.


See J. G. Gurley and Edward S. Shaw, Money in a Theory of Finance, Brooking’s Institution, Washington D.C., 1960, especially Chapter IV, and Graeme S. Dorrance and Eckhard Brehmer, ‘The Growth in Liquidity in Selected Industrial Countries’, IMF, DM 62/19, May 1962.


This assumption, merely made for simplicity of exposition, is warranted if there are underemployed labour resources available which are drawn into the productive process as development continues, or if technological improvements continue to counterbalance any tendency towards diminishing returns to capital a: they have done historically in the industrially advanced countries.


For instance, when the net capital inflow consists chiefly of foreign aid and does not depend on domestic interest rates.


J. J. Polak, op. cit., the model has been applied to thirty-six countries in J. J. Polak and L. Boissonneault, op. cit.; see also J. M. Fleming and L. Boissonneault, ‘Money Supply and Imports", IMF Staff Papers, May 1961; S. J. Prais, ‘some Mathematical Notes on the Quantity Theory of Money in an Open Economy", ibid., May 1961; and S. Kanesathasan, ‘Government Imports and Import Taxes in Monetary Analysis of Income and Imports’, ibid., December 1961.


The Polak model is fully set out in the published sources given in the preceding footnote


It is probably fair to say that the interpretation of the role of the interest variable in the investment function as representing the ‘cost of borrowing’ was a red herring supplied by some of Keyncs’ followers rather than by Keynes himself.


I.e., if their price elasticities of demand exceed unity. We are in this context not concerned with the case of different income elasticities of the demand, or complementarities among goods.


For a discussion of such a model in terms of differential equations rather than difference equations see S. J. Prais, op. cit.


That there is n o substantive difference between the classical and Keynesian approaches, when both are formulated in appropriately general terms (i.e., without assigning specific causal significance to one particular variable or another), can be seen in this way: the Keynesian expenditure function (ignoring time lags) is E = b1Y—b2i + b0 and a linear approximation of the liquidity preference function is L* = c1Y—c2i + c0. Substituting for i from the liquidity preference function into the expenditure function and combining terms in Y results in E = (bl—c1b2/c2) Y + (b2/c2)L* + constant, which is the form of (IV-2) in the text with (b2/c2) equal to the coefficient a3 in (IV-2). A Keynesian could thus interpret a3 as the ratio of the effect of a change in interest on expenditure to the effect of a change in interest on the demand for money. For a fuller statement of the equivalence of the two systems, if properly stated, see J. M. Fleming, ‘The Determination of the Rate of Interest’, Economica, August 1938.


The first reaction of an increase in spending will be a reduction in inventories accompanied perhaps by a rise in some prices. Since the model outlined here is in current prices (because of data limitations rather than by choice) the rise in prices will be reflected in private expenditure. But we should allow a long enough lag for inventories to be restored to their desired level, so as to express the effect of a change in money on both the price level and on the level of output.


The equations have been estimated by the two-stage least-squares method from annual data for 1949-60. The standard errors of the regression coefficients are given in parentheses below the estimated coefficients


This lag in the effect of a change in the money supply on income and imports is, of course, somewhat artificial. The adjustment of spending to the changes in the stock of money goes on all t he time and some effect will already be present in the current year. W e should really have an equation showing a distributed lag response of expenditure to changes in money. But the available data (short time series and annual values) make the fitting of more complicated functions impracticable for many countries.


This difficulty is, of course, not restricted to underdeveloped countries.


See G. S. Dorrance and W. H. White, ‘Alternative Forms of Monetary Ceilings for Stabilization Purposes’, IMF, Staff Papers, November 1962.


See E. A. Birnbaum and M. A. Qureshi, ‘Advance Deposit Requirements for Imports’, IMF Staff Papers, November 1960; and I.O.W. Olakanpo, ‘Monetary Management in Dependent Economies’, Economica, November 1961.


There is, of course, nothing moral about such a policy, and the persuasion has, in fact, often been of the ‘ … or else’ variety.