**O**VER THE YEARS the Fund has endeavored to develop various models that would indicate, if only in an approximate way, the likely relationship between credit creation in a member country and its balance of payments. One characteristic of this effort at model building has been that it has aimed at simple models. The advantage of simple models, if they are well designed, is that they will bring out clearly the main relevant features and interrelationships of a particular economy and thus point to certain major relationships important for policy purposes. More refined models may take into account a much wider set of variables and relationships, and may thus, in principle, serve to answer a wider variety of questions over a broader range of circumstances; but in practice such models often fall between two stools, being at the same time insufficiently articulated to catch aspects of reality that were not specifically considered at the time of their construction and too complicated to be readily tapped for significant answers, except perhaps by the most skilled operator.

Notation used

*Y*= gross national product*C*= consumption*I*= investment*A*= autonomous investment*G*= government expenditure*Mo*= money supply*R*= rate of interest*X*= exports*V*= autonomous capital inflows*M*= imports*K*= net capital inflow of the nonbanking sector*B*= change in net overseas assets of the banking system (central bank and commercial banks)*D*= total domestic credit of the banking system*Q*= autonomous influences on the interest rateZ = (A + G + X)

W = (X + V)

## Models for Developing Countries

The attempts at simplicity in the construction of the Fund’s models have necessarily involved recognizing the fact that models for different types of countries had to bring out different main features. In models for developing countries it was reasonable to assume that there was little organized capital market, that credit was rationed, that holdings of money were predominantly transactions balances and insensitive to interest rates, that capital flows were largely autonomous, and that the money supply tended to respond to the overseas balance. Moreover, lack of data in these countries has made it difficult to test hypotheses with any degree of refinement. Given these conditions, the following model could be developed:^{1}

Equations (1) and (2) are the two behavioral equations in the model, the first reflecting the constant velocity assumption ^{2} and the second explaining imports in terms of income. Equations (3) and (4) are identities, the first of these showing that the increase in the money supply is partly of foreign origin (B) and partly of domestic origin (ΔD), and the second identifying the components of the balance of payments.

For reasons of simplicity, this model makes no explicit allowance for reaction lags. The dynamic character of the model derives from the fact that it contains both Mo and ΔMo. There is an implicit lag in equation (1), income being a flow during a period and money a stock at the end of that period. ^{3}

These equations may be combined to yield solutions in discrete time periods for income and the change in net foreign assets of the banking system.

^{4}

For purposes of this paper the time period will be treated as one year. These equations show the first-period effects on the balance of payments and income of changes in the exogenous variables in the model (K, X, D), e.g., a unit increase in domestic credit will reduce the net overseas assets of the banking system in the period by _{-1} and hence also that ΔY = ΔY_{-1.} The long-run solutions that satisfy these conditions are

We will assume that changes in the exogenous variables are once for all, i.e., that exports, net capital flows, or credit all rise or fall to a new level that is then maintained in subsequent periods. In these conditions the coefficient in the long-run solution is the sum of all the effects in the individual periods on the balance of payments and the change in income, e.g., a coefficient of ^{5}

The interesting features of this model are as follows:

(1) The proportion of an increase in credit, exports, or capital inflow that leaks overseas in the first year in the form of imports is positively related to the size of the propensity to import and to velocity; the higher these are, the greater will be the leakage. Only the product of the two coefficients *(mv)* appears in the formulas explaining the balance of payments. This product represents

(2) A once-for-all increase in exports or net capital flows will continue to improve the reserves of the banking system until imports rise to match the higher level of exports and net capital flows. In equilibrium, income, the money supply, and imports will all be higher and reserves will have leveled off at the new higher level. ^{6}

(3) A once-for-all increase in credit will all leak out in the long run in the form of increased imports. In the transitional period income will be higher, but in the long run both income and imports will revert back to their original levels. ^{7} This case demonstrates that, over time, an excess of credit will result in a loss in reserves of the same order. The important policy implication here is that, given expected receipts from exports and net capital flows, the appropriate rate of credit expansion is determined by the desired change in the net foreign assets of the banking system.

This model can be tested statistically in two ways. ^{8} One way is to obtain estimates of velocity and the marginal propensity to import from data for a number of years. These estimates are used to calculate a priori coefficients attaching to the exogenous variables. These coefficients are then applied to data for the exogenous variables, and predictions for imports and income are obtained for each year. These predictions are then compared with the actual values of imports and income. A second way is to directly regress the dependent variables in the model (imports, income) against the independent variables (exports, net capital flows, and credit) in current and past years. The results of these tests, for both the developing and the developed countries, have been good. On the whole, imports and income have been explained well by this model. ^{9}

Given the two behavioral equations in the model, it follows that one would expect the model to perform better, the smaller the variability in the import and velocity functions. Studies within the Fund have shown that these functions tend to be more unstable for the less developed countries; ^{10} hence, this model should really give worse results for the latter group of countries. ^{11} But this greater variability in the less developed countries is due to reasons that are difficult to incorporate formally in models, e.g., for imports, the greater variability in the use of controls and duties, and for velocity, the greater instability (political, social, and economic) and—more important—the much larger variation in the nonmonetary (barter) sector.

## Models for Countries with Developed Financial Markets

The simplification involved in the model fails to reflect some important features of economies with highly developed financial markets. It seemed advisable to develop an alternative model that would bring out such features as the role of interest rates in the demand for investment, the demand for money, and the flow of international capital movements.

The following set of equations represents this alternative model. ^{12}

Gross national product, in equation (9), is the sum of consumption, investment, government expenditure, and exports less imports. Equation (10) shows consumption as a function of income. Equation (11) shows investment as a function of the rate of interest and an autonomous component (A). Equation (12) represents imports as a function of income. In equation (13) K, the capital balance, is assumed to be determined by an autonomous component (V) and the rate of interest. Equation (14) shows the rate of interest as determined by the level of income and the money supply. Equation (15) is the balance of payments identity, and equation (16) is the money supply identity. ^{13} This model, then, retains the feature of the first model that the money supply responds to the balance of payments. Exports and government expenditure are treated as entirely autonomous.

The model, although it contains the major behavioral features to be found in macromodels of the industrial countries, nevertheless lacks a number of refinements. In particular, there are no lags in the behavioral equations, the investment function has no real variables, and there is no explicit reference to the price level; also, if foreign asset portfolios are assumed to respond to the levels of interest rates, then the capital flow function should be related to the change in interest rates—not to the level of interest rates. ^{14} Wealth variables are omitted, and no distinction is made between short-term interest rates (influencing inventories and short-term capital movements) and long-term interest rates (influencing fixed investment and long-term capital movements). But, as indicated earlier, these refinements, while important, make the models difficult to understand and to apply; many important points may be demonstrated with the simpler models without at the same time aspiring to any precision in the mechanics and dynamics of adjustment.

## Comparison of the Models

A useful way to look at the two models is to consider the first model as a simplification of the second, a simplification that is acceptable in circumstances where particular complications are not important. There are two differences between the models. First, whereas capital flows are exogenous in the first model, they are assumed to be sensitive to interest rates in the second. The second difference lies in the money demand equation. Equation (14) may be rewritten as

The money demand equation in the first model is

(la) is then the limiting case where money is totally inelastic to interest rates, so that

It follows that the solutions to the first model may be derived directly from the second model by setting g = 0 (capital flows are unresponsive to interest rates),

The Two Models Where g = 0

For simplicity, let

The solutions to the second model when g = 0 are

*Short run*

*Long run*

These solutions may be compared with the solutions to the first model—equations (5)-(8). There are interesting parallels and differences between the results.

(1) In the short run the leakage into imports from a given expansion in credit is smaller in the second model than in the first. This is so because the increase in the money supply has a weaker effect on income in the second than in the first. ^{15}

(2) While increases (decreases) in autonomous expenditures (A + G) unambiguously worsen (improve) the balance of payments and raise (lower) income in the second model, variations in autonomous expenditures make no impact on either the balance of payments or income in the first model.

(3) In the long run an excessive expansion of credit will all leak out into imports in the two models, the only difference being that it will leak out more slowly in the second. In both models, income in the long run reverts to its original level after a loss of reserves equal to the additional credit created. ^{16}

(4) In the long run, changes in autonomous expenditures cannot permanently change income in the second model; hence, there is an important parallel between the two models. The reason for this result in the second model is that as long as income is higher imports will be higher and, with exports and net capital flows given, the money supply will be falling; this will continue to depress income until it reverts to its original level.

However, because income and imports will be higher in the transitional phase, there will be some cumulative loss in reserves

(5) The long-run effects on income of increases in exports or net capital inflows are identical in the two models, where imports must rise in equilibrium to match the increase in foreign receipts. However, the cumulative gain in reserves is larger in the second model because the adjustment mechanism is weaker. ^{17} In the extreme case in the second model where b = 0 (the money supply does not influence income), the accumulation of reserves will be infinite.

The Two Models Where g > 0

The solutions to the first model are now

*Short run*

*Long run*

The solution to the second model are

*Short run*

*Long run*

The solution to the first model is obtained by solving the second model, setting

Since we retain the constant velocity equation, in the absence of a change in the money supply the level of income is fixed. In these conditions increases in autonomous expenditures can have no effect on income ^{18} but will raise the interest rate. An increase in the supply of money unaccompanied by a change in autonomous expenditures (A + G + X) must lower the interest rate; in this respect the first model has a parallel with the second.

When these two models are being contrasted with one another the difference again lies solely in the assumption respecting the interest sensitivity of the demand for money. At the same time, since we now assume for both models that capital flows are interest sensitive, we may also note the changes for each model that this additional assumption introduces.

(1) In both models now the proportion of credit that leaks out in the first period is larger than where g = 0. ^{19} The reason is that in both models there is now an additional leakage into capital outflows. In the special case where g = ∞ (perfect capital mobility), the first year coefficient becomes 1, implying that all credit created leaks overseas, leaving the domestic money supply and income unchanged at the end of the period. This is the familiar special case where, under the assumption of perfect capital mobility, monetary policy has virtually no domestic impact but a strong external impact.

(2) In both models changes in autonomous expenditures influence income in the short run, but the transmission mechanism is quite different. In the first model increases in autonomous expenditures raise the interest rate, which pulls in money from overseas, and the increased money in turn raises domestic income. In the second model the increase in income is made possible by the rise in interest rates, which reduces desired money balances.

(3) In the first model an increase in autonomous expenditures (A + G) unambiguously improves the balance of payments by attracting capital inflows. In the second model the outcome is ambiguous: the increase in income will raise imports but at the same time interest rates will rise, attracting some capital inflow. If the former (latter) effect is stronger, the balance of payments will worsen (improve). ^{20}

(4) Again in the long run in the two models an expansion of credit, other things, being equal, will result in an equivalent loss in reserves. ^{21}

(5) There is now a permanent income effect, identical in the two models, from increases in autonomous expenditures {A + G). In the first model the initial outcome, as we have seen, is an improvement in the balance of payments as a result of the rise in interest rates. As long as there is a surplus the domestic money supply continues to grow; this forces down the interest rate, encouraging capital outflows, and also raises income and imports. Equilibrium is attained at a point where the interest rate is somewhat higher than initially and the additional capital inflow is exactly matched by the additional imports at the higher level of income.

In the second model there are two cases to consider. Where the balance of payments initially worsens (the increased imports exceeding the increase in the capital inflow), the money supply will begin to fall back progressively as the deficit persists; this will raise interest rates and improve the capital inflow but will lower income and imports. This will continue up to the point where the additional capital inflow equals the additional imports at a higher level of income. Where the balance of payments initially improves (the additional capital inflows exceeding the additional imports), the sequence that develops is more nearly parallel to the first model, with the money supply rising, interest rates falling, and imports and income continuing to rise up to the point where the capital inflow has so fallen and the imports have so risen that the balance of payments is again in equilibrium. This second case hence yields a higher equilibrium level of income than the first case. ^{22}

## Credit Rationing in Industrial Countries

The major contrast between the first model (at least in its simplest form, where g = 0) and the second model lies in the assumption respecting the degree of perfection in the capital markets. Whereas the first model was characterized as an extreme case where there was virtually no organized capital market, the second model was assumed to have a perfect capital market where the demand and supply of funds were cleared by an equilibrating interest rate. It is now generally agreed that even in industrial countries with the most highly developed capital markets the assumption of a perfect clearing mechanism is quite unrealistic. ^{23} Hence, it is necessary to recognize that in these economies there exists, side by side, an organized as well as an imperfect capital market.

The imperfect capital market has two important characteristics. First, the interest rate on credit in these markets (e.g., commercial bank loans and consumer or housing finance) tends to be relatively sticky, ^{24} with the demand for credit in excess of the supply at the prevailing interest rate. Second, at least part of the excess demand does not “spill over” into other sectors of the capital market but remains unsatisfied continuously, so that expenditure is being constrained not by its interest sensitivity but simply by the unavailability of funds.

As an illustration, consider what happens when monetary policy is restrictive. There is, first, a direct contraction in the supply of loans by the banking system (as well as perhaps some liquidation of securities). This will simultaneously discourage expenditure directly and raise market interest rates. ^{25} The rise in market rates will divert funds away from intermediaries whose interest rates are sticky; these intermediaries in turn will reduce the availability of credit, which again may have direct expenditure effects. Hence, monetary policy, by increasing the degree of credit rationing in these markets, has influenced expenditure directly as well as by an interest rate mechanism. ^{26}.

While these market phenomena are well known, it is difficult to deal with them formally in macromodels. Several possibilities present themselves. First, to the extent that the degree of credit rationing is positively related to the interest rate, the interest rate may simply be made to serve as a proxy for all “restrictive” or “expansionary” policies. Second, if it is true that some expenditure is being frustrated by a lack of finance, then variations in bank credit^{27} may be assumed to produce simultaneous, autonomous shifts in expenditure functions, e.g., in the second model an increase in bank credit accompanying an increase in the money supply would make its impact not only through the money supply but also through an increase in autonomous expenditure. ^{28} Third, bank credit as well as the interest rate could appear directly in the expenditure functions. ^{29} This would bring out clearly the fact that bank credit could influence expenditure independently of the interest elasticity (e.g., even in the special case where b in the second model would be zero). ^{30}

Perhaps one conclusion that emerges is that, insofar as these imperfections are important, the second model discussed may possibly understate the impact of monetary policy on expenditure, where monetary policy is implemented by changes in bank credit rather than by, say, open market operations. An implication of this would be, for example, that a somewhat larger proportion of credit created might leak overseas in the first year. The importance of this would vary widely among industrial countries, as would the method of monetary policy selected by them, and no generalization seems possible.

## Monetary and Credit Ceilings and Long-Term Rules

This section compares the built-in stabilizing properties of credit ceilings and monetary ceilings on output and the balance of payments. The theoretical framework will be a slight variation of the second model, discussed earlier. The credit and monetary ceilings would be determined, in the first place, on the assumption of given values of the autonomous elements in the model, including autonomous expenditures and exports. Obviously, if the “forecast” values are correct and the structural coefficients of the model known, the “desired” change in net foreign assets will be fulfilled; ^{31} in these conditions it becomes a matter of indifference whether a credit or a monetary ceiling is used. The interesting question concerns the behavior of the system when “errors” (in the form of “disturbances”) occur in the year for which a ceiling has been determined. We assume that the ceilings are maintained during the relevant period in the face of these disturbances. The alternative systems are then evaluated in terms of the size of the effect, stemming from the disturbance, on output and the balance of payments.

We consider the behavior of the system under four types of disturbance: (1) an unanticipated increase in autonomous expenditure; (2) an unanticipated increase in the demand for money (given income and interest rates); (3) an unanticipated increase in exports; (4) an unanticipated increase in the net capital inflow.

There are five cases to consider. The first disturbance has two possibilities. As we have seen, an increase in expenditure may either improve or worsen the balance of payments; hence, these two cases will be considered separately. Table 1 lists all five possibilities.

**Stabilizing Effects Of Credit Ceilings And Monetary Ceilings On Output And Balance Of Payments Resulting From Various Types Of Disturbance**

**Stabilizing Effects Of Credit Ceilings And Monetary Ceilings On Output And Balance Of Payments Resulting From Various Types Of Disturbance**

Policy That Is More Stabilizing For | ||||
---|---|---|---|---|

Type of Disturbance | Output | Balance of payments | ||

Domestic | ||||

An increase in expenditure: | ||||

balance of payments worsens | credit | credit | ||

balance of payments improves | money | credit | ||

An increase in the demand Tor money | credit | credit | ||

External | ||||

An increase in exports | money | credit | ||

A capital inflow | money | credit |

**Stabilizing Effects Of Credit Ceilings And Monetary Ceilings On Output And Balance Of Payments Resulting From Various Types Of Disturbance**

Policy That Is More Stabilizing For | ||||
---|---|---|---|---|

Type of Disturbance | Output | Balance of payments | ||

Domestic | ||||

An increase in expenditure: | ||||

balance of payments worsens | credit | credit | ||

balance of payments improves | money | credit | ||

An increase in the demand Tor money | credit | credit | ||

External | ||||

An increase in exports | money | credit | ||

A capital inflow | money | credit |

The relative effects on the balance of payments of monetary and credit ceilings can be dealt with first. The difference between the two is that an unanticipated improvement in the balance of payments will raise the money supply with a credit ceiling and leave it unchanged with a monetary ceiling, while an unanticipated deterioration in the balance of payments will lower the money supply with a credit ceiling and again leave it unchanged with a monetary ceiling. Since an increase (decrease) in the money supply unambiguously worsens (improves) the balance of payments, it immediately follows that, whatever the disturbance, a credit ceiling must be a better stabilizer of the balance of payments than a monetary ceiling. This result is shown in the last column of the table.

The relative effects on output are somewhat more complicated, and each case will be dealt with individually. Consider first Case 1. Suppose that credit and money stay “on target” and expenditure increases. By assumption, the balance of payments will worsen. This will tend to reduce the money supply with a credit ceiling, which is stabilizing with respect to output, since it will dampen the increase in expenditure. Hence, in this case, a credit ceiling is unambiguously the more stabilizing policy for both output and the balance of payments.

In Case 2 an increase in expenditure will improve the balance of payments so that the money supply will be larger with a credit ceiling than with a monetary ceiling. A larger money supply will be destabilizing with respect to output, since it will raise output further.

In Case 3 an increase in the demand for money raises interest rates and hence unambiguously improves the balance of payments. The resultant addition to the money supply with a credit ceiling will dampen the fall in output. In Case 4 the increase in exports will raise the domestic money supply with a credit ceiling, thus accentuating the impact on output. The same result holds for Case 5, where there is an unanticipated capital inflow. Where the money supply is unresponsive, output will be unaffected and the balance of payments will improve by the amount of the capital inflow. Where the money supply is allowed to increase, output will increase.

These results are summarized in the last two columns of the table. The conclusion then is that a credit ceiling is unambiguously a better stabilizer of the balance of payments than a monetary ceiling. On the output side, the outcome depends very much on the type of disturbance and the assumptions made about the sizes of the parameters. If increases in autonomous expenditures worsen the balance of payments (and a reduction improves the balance of payments), then for all *domestic* disturbances a credit ceiling is also a better stabilizer of output. If increases in autonomous expenditures improve the balance of payments, then a monetary ceiling is the better stabilizer of output for this disturbance. For external disturbances, a monetary ceiling is unambiguously a better stabilizer of output.

If, then, the balance of payments has overriding priority, a credit ceiling would appear to be the better policy; even if strong priority is attached to minimizing fluctuations in output, a credit ceiling may still be the more efficient policy where external disturbances tend to be relatively less important than domestic disturbances (and particularly so when increases in expenditures tend to worsen the balance of payments). Some conflict may arise in countries that have a large and unstable foreign sector; then the cost Of minimizing fluctuations in international reserves may be an accentuation in the volatility of output.

The analysis, which compared a monetary ceiling with a credit ceiling in the framework of a decision period of approximately one year, could easily be extended to the evaluation of a monetary versus a credit growth rule. Friedman, in his many writings, has proposed that, instead of implementing a discretionary monetary policy, money should be allowed to grow at a fixed rate, year after year. ^{32} An alternative long-term rule would be to allow total domestic credit to grow at a constant rate. The difference would be that under the credit rule the rate of growth of money would be sensitive to fluctuations in the balance of payments. Given that over the long term there were no misalignments in exchange rates, the optimal rate of growth of credit would be the same as the optimal rate of growth of money. This optimal growth rate would be determined by considerations of secular velocity, the real growth of capacity, and the rate of inflation considered to be acceptable. ^{33} The text, in effect, argues that, under certain conditions concerning the size and nature of the disturbances to the system, a credit rule could unambiguously outperform a monetary rule. If these conditions were not met, the choice between the two rules would then depend on a value judgment between a credit rule, which was a better long-term stabilizer of the balance of payments, ^{34} and a monetary rule, which was a better long-term stabilizer of output.

## APPENDIX

The cases discussed in the last section in the text may easily be demonstrated more rigorously. The only extension to the model is to allow for changes in interest rates as a result of autonomous changes in the demand for money. The interest rate equation (14) may now be written as

where Q represents all influences on the interest rate other than income and the money supply. The solutions for income and the balance of payments for money and credit are as follows:^{35}

Case 1: m > ge Δ(A + G)

ForB:

For Y:

Case 2: m < ge Δ(A + G)

For B:

For Y:

Case 3: (ΔQ)

ForB:

For Y:

Case 4: (ΔX)

For B:

For Y:

Case 5: (ΔV)

For B:

For Y:

^{}1

he model is developed in J.J. Polak, “Monetary Analysis of Income Formation and Payments Problems,” Staff Papers, Vol. VI (1957-58), pp. 1-50, and J.J. Polak and Lorette Boissonneault, “Monetary Analysis of Income and Imports and Its Statistical Application,” Staff Papers, Vol. VII (1959-60), pp. 349-415. The reader is referred to these articles for a detailed discussion of the theoretical implications and empirical applications of the model. Limitations involved in the assumption of constant velocity and an analysis of factors influencing velocity are to be found in Polak, op. cit., pp. 37-41. While the price level is not explicitly mentioned in the model, this is not to say that the applicability of the model is limited only to situations below full employment. The import equation is expressed in nominal terms, implying that a change in income owing to a change in the price level would generate the same effect on imports as a real change in income. This assumption is also examined in Polak, op. cit., pp. 17-18.

^{}2

In the original model, the time period was defined in such a way that a change in the stock of money was associated with an equal increase in income. This meant defining the time period for any country as the fraction of the year presented by the inverse of velocity. This expositional device appeared useful at the time; however, all the theoretical aspects can be demonstrated by the simpler version used here.

^{}3

This lag could have been eliminated by relating income to the mean money in the period, for which (M_{ot} + M_{ot-1})/2 might be used. It should be noted that as a result of the implicit lag the numerical results are somewhat affected by the choice of the time period. A substantive question not explored here is the length of any actual lag between Y and Mo. These issues are dealt with in S.J. Prais, “Some Mathematical Notes on the Quantity Theory of Money in an Open Economy,” Staff Papers, Vol. VIII (1960-61), pp. 212-26.

^{}4

Combine equations (1), in the first-difference form, and (3); then substitute equation (5) for B.

^{}5

_{-1}= 0 and ΔY = ΔY

_{-1}= 0. Here, foreign assets and income both level off in equilibrium. This is dealt with in the text when changes are once for all. As an illustration, consider the effect of a once-for-all change in credit on the balance of payments and the change in income. The total long-run effect is the cumulative sum of the effect in each period.

_{-1}≠ 0 and ΔY = ΔY

_{-1}≠ 0. This is so where the change in reserves and income is constant in equilibrium. This equilibrium is possible only when the exogenous variables are changing by a fixed amount, period after period. For example, suppose that the volume of credit increases by 10 units in each period. Then it can be shown that, other things being equal, in the new equilibrium the loss in reserves will be equal to 10 units, with imports, and hence income, permanently higher; the money supply will cease to grow, since the addition to credit is exactly absorbed by the deficit in the overseas balance. This is the interpretation of the unit coefficient in equation (7); since income moves to a higher level, the change in income in equilibrium is zero. This case is discussed in Polak,

*op.cit*., pp. 26-27'.

^{}6

From equation (8) we can see that the cumulative increase in income is

^{}7

The long-run coefficient for domestic credit is 1 for the balance of payments and zero for income. See Polak, *op. cit*., p. 10.

^{}8

See Polak and Boissonneault, op. d/.; J. Marcus Fleming and Lorette Boissonneault, “Money Supply and Imports,” *Staff Papers*, Vol. VIII (1960-61), pp. 227-40; an unpublished paper by Lorette Boissonneault and Joseph O. Adekunle, “Monetary Analysis of Imports and Income: Further Investigations.”*

^{}9

An interesting difference in the result of the two methods of testing the model is that the latter (the direct regression approach) tends to yield a higher firstyear coefficient with respect to imports, implying that a larger proportion of credit, exports, and net capital flows tends to leak out than is implied by the model. This is noted in Boissonneault and Adekunle, *op. cit*. The reasons for this are discussed in Fleming and Boissonneault, *op. cit*.

^{}10

On the relative variability of the import function, see Boissonneault and Adekunle, *op. cit*. The variability of velocity after the removal of trend is strikingly higher in the less developed countries. See also Yung Chul Park, “The Variability of Velocity: An International Comparison,” *Staff Papers*, Vol. XVII (1970), pp. 620-37.

^{}11

The evidence on this, however, remains ambiguous. See Boissonneault and Adekunle, *op. cit*., and Fleming and Boissonneault, *op. cit*.

^{}12

This is the model in Victor Argy, “Monetary Variables and the Balance of Payments,” *Staff Papers*, Vol. XVI (1969), pp. 267-88; the model is examined in some detail in this article.

^{}13

In this model capital flows must refer to the nonbanking sector. Net overseas assets of the commercial banks may, of course, also be sensitive to interest rates. To the extent that this is true, there will be some reshuffling of net foreign assets between the commercial banks and the central bank; in other words, this would affect only the form in which the balance of payments is financed.

^{}14

See, for example, Thomas D. Willett and Francesco Forte, “Interest Rate Policy and External Balance,” *The Quarterly Journal of Economics*, Vol. LXXXIII (1969), pp. 242-62.

^{}15

*f* and remembering that in the first model

^{}16

Compare equation (19) with equation (7) and equation (20) with equation (8). In both models the long-run coefficient for credit is 1 for the balance of payments and zero for income.

^{}17

In the second model the cumulative gain in reserves is larger for an increase in the capital inflow than for an increase in exports (

^{}18

In this respect the model may be said to represent the so-called monetarist school. The model itself may have little relevance to developing countries, but it does have a number of interesting features worth pursuing.

^{}19

For the first model,

^{}21

This will not hold, of course, to the extent that there is a “return flow” from other countries but, with the possible exception of the United States, this return flow will tend to be negligible.

^{}22

From equation (28) the cumulative increase in income is

^{}23

Franco Modigliani, “The Monetary Mechanism and Its Interaction with Real Phenomena,” *The Review of Economics and Statistics*, Vol. XLV (Supplement, February 1963), pp. 79-107; Assar Lindbeck, *A Study in Monetary Analysis* (Stockholm, 1963); Frank de Leeuw and Edward M. Gramlich, “The Channels of Monetary Policy,” *Federal Reserve Bulletin*, Vol. 55 (1969), pp. 472-91; Donald P. Tucker, “Credit Rationing, Interest Rate Lags, and Monetary Policy Speed,” *The Quarterly Journal of Economics*, Vol. LXXXII (1968), pp. 54-84.

^{}24

The stickiness may be due to institutional or legal constraints or to considerations of “good will” toward customers. This amounts to saying that a discriminatory pricing system prevails.

^{}25

An implication of this approach is that the composition of bank assets may be important in influencing expenditure. There are difficulties in empirically testing this hypothesis. For one attempt, see William L. Silber, “Velocity and Bank Portfolio Composition,” *The Southern Economic Journal*, Vol. XXXVI (October 1969), pp. 147-52.

^{}26

Modigliani, *op. cit*., p. 100, interprets this mechanism as follows: “…the capital rationing mechanism provides a plausible way of reconciling moderate fluctuations in market rates with a widely shifting and interest-inelastic investment schedule.”

^{}27

More correctly, the part of the total credit of the banking system that influences expenditure directly.

^{}31

The appropriate credit ceiling that would reach the balance of payments target is not necessarily one that would generate a satisfactory level of income. Resort to a second instrument (e.g., fiscal policy) might be necessary under these conditions.

^{}32

See Milton Friedman, *Essays in Positive Economics* (University of Chicago Press, 1953). One of the authors examines the proposal in some detail in Victor Argy, “Rules, Discretion in Monetary Management, and Short-Term Stability,” *Journal of Money, Credit and Banking*, Vol. Ill (1971), pp. 102-22.

^{}33

See Argy, “Rules, Discretion in Monetary Management, and Short-Term Stability” (cited in footnote 32), Appendix 1, pp. 117-20.