Open market operations, base money, and deposits

An active controversy has proceeded for some years now on the effectiveness or otherwise of open market operations in the context of institutional arrangements in the United Kingdom.¹⁴ In this section an attempt will be made to examine some of the relevant issues and to synthesize the extensive literature on this question. To simplify the discussion that follows, and to keep the discussion in line with the literature, it is assumed through most of the section that c = 0, so that no complications arise from currency demands by the nonbank public. This assumption effectively means that variations in base money are absorbed only by the banking system. Our procedure will be to set up a simple model and then to modify the model in various ways in order to assess the impact of open market operations under alternative assumptions.

We begin by making the following assumptions:

(1) The monetary authorities sell short-term securities to the nonbank public (the banking sector includes the discount houses).

(2) Bank cash is maintained rigidly at 8 per cent of deposits, and liquid assets (including cash, call loans to the discount houses, and bills) are maintained rigidly at 28 per cent of deposits. We simplify slightly by disregarding the existence of a special deposit ratio.

(3) The Bank of England charges a penal rate for loans to the discount houses. A penal rate is generally taken to mean a rate in excess of the treasury bill rate.¹⁵

(4) The discount houses are in “disequilibrium” (with marginal costs in excess of marginal revenue) as long as they are in penal debt to the Bank of England and therefore will take some steps to liquidate the debt.

(5) The discount houses are able to liquidate any penal debt with the Bank of England by a combination of the following: (a) increasing borrowings from the nonbank public, (b) shifting some of their treasury bill holdings to the nonbank public—either by direct sales or by lowering their bid price at the weekly tender, (c) reducing their holdings of commercial bills, (d) selling short bonds to the nonbank public.

Table 1, Situation 1, shows the commercial banks in initial equilibrium. Now we suppose that the Bank of England sells 10 units of short-term bonds to the nonbank public. Cash and deposits of the banking system fall by 10 units, as shown in Situation 2. The liquidity ratio is now below 28 per cent, and the banks must now take steps to restore their cash and liquidity ratios.¹⁶ In Situation 3 we represent the banks as being in equihbrium again. Deposits will have fallen by about 35.7 units (10 × 100/28). The question now is whether the banking system as a whole, including the discount houses, is in full equilibrium. The banks are shown as having called in 7.1 units from the discount houses. By assumptions (3) and (4), the discount houses will then be in disequilibrium. By assumption (5), the discount houses will take steps to liquidate the debt of 7.1 with the Bank of England. Any of the possible lines of action under assumption (5) will result in a reduction in the cash deposits of the banking system by 7.1 units. In Situation 4 the banks are again out of equilibrium, and the discount houses are in equilibrium. Once again, to restore their liquidity ratio, the banks will need to call in more loans from the discount houses. The new position is shown in Situation 5, but here again the discount houses are out of equilibrium. Once again then the discount houses will act in such a way that the cash and deposits of the banks will fall by a further 5.1 units (not shown in the table). It is clear that since full equilibrium requires that both the banks and the discount houses be satisfied with their portfolio position, the end result must be the position shown in Situation 6 in the table. Here deposits have fallen by 125, which represents 12.5 times the original sale. What is interesting about this end result is that cash obtained indirectly from the Bank of England by the banking system is not permanent in the sense that steps will soon be taken to return the cash to the Bank of England.¹⁷ The result therefore would have been nearly identical if we had represented the banks as, in effect, selling the bills to the nonbank public in the process of adjustment.

Table 1.

Impact of Open Market Operations of the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

Table 1.

Impact of Open Market Operations of the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

	Situation
	1	2	3	4	5		6
Deposits	1,000	990	964.3	957.2	938.9	…..	875
Cash	80	70	77.1	70.0	75.1	…..	70
Call loans, etc.	200	200	192.9	192.9	187.8	…..	175
Investments and advances	720	720	694.3	694.3	676.0	…..	630

View Table

Table 1.

Impact of Open Market Operations of the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

	Situation
	1	2	3	4	5		6
Deposits	1,000	990	964.3	957.2	938.9	…..	875
Cash	80	70	77.1	70.0	75.1	…..	70
Call loans, etc.	200	200	192.9	192.9	187.8	…..	175
Investments and advances	720	720	694.3	694.3	676.0	…..	630

The effect on interest rates may be dealt with briefly. The original sale of short-term bonds will probably have raised the yield on bonds; to the extent that banks liquidate some of their security holdings, the effect on the yield of bonds is strengthened directly. If the discount houses shift treasury bills to the nonbank public, some direct pressure will be exerted on the bill rate. These results are of course consistent with expectations that a drop in the money supply tends to be accompanied by a rise in interest rates.

We must now proceed to modify the assumptions and to examine some of the consequences. Suppose, first, that banks hold a liquidity ratio in excess of the prescribed minimum of 28 per cent. (We assume throughout that they hold 8 per cent cash.) It is easy to demonstrate that the final equilibrium level of cash and deposits shown in Situation 6 will be the same, although, of course, the dynamics of adjustment will be different. On the assumption that the banks cannot obtain permanent additional cash, therefore, the drop in deposits must again be 12.5 times 10. The only modification in this case is that a larger burden of adjustment will fall on the liquid assets of the banks and a smaller burden on investments and advances. Where excess liquidity is substantial, the whole burden could fall on liquid assets without contravening the 28 per cent requirement. We conclude then that, irrespective of the liquidity ratio of the banking system, open market operations will be successful in permanently reducing base money, notwithstanding institutional arrangements whereby banks can indirectly replenish their cash base, provided that assumptions (3), (4), and (5) hold.

We now retain assumption (2) and examine more carefully assumptions (3), (4), and (5). Suppose now that assumption (3) does not hold. Then assumptions (4) and (5) will also not apply. Assumption (3) may be relaxed in the extreme example by supposing that the Bank of England accommodates the discount houses (or banks) at the “back door” at “market” (nonpenal) rates.¹⁸ In effect, any rate that does not raise marginal cost above marginal revenue is nonpenal. Under these conditions, Situation 3 in Table 1 will represent the final full equilibrium for the banks and the discount houses. The discount houses, by definition, will have no incentive to take any further action. In this instance, then, some permanent cash can be absorbed from the Bank of England. It is interesting that here the relevant multiplier (i.e., the change in deposits divided by the initial open market sale) is 100/28, or about 3.5, and not 12.5 as in the previous example. In principle, though, the same ultimate reduction in deposits can always be achieved by engaging in a much larger initial open market sale (e.g., in this case, just under 36).

One interesting problem raised by this particular model is the following: The objective of back-door accommodation is to prevent any disturbance in the bill rate, but at the same time deposits are allowed to fall from 1,000 to 964.3, putting some of the burden of adjustment therefore on investments and advances. If only advances are allowed to fall and advances directly affect expenditure, it is conceivable that the rate of interest will not be affected but, to the extent that short-term bonds are sold, the yield on bonds will tend to rise. Hence, where securities are sold by the banks, as is normal, the model assumes that the authorities will allow a rise in the bond rate while vigorously maintaining the bill rate unchanged.

If the liquidity ratio is in excess of the prescribed 28 per cent, deposits may fall by only 10 units. Banks would obtain the required cash to satisfy the 8 per cent requirement, but, since they are holding liquid assets in excess of the prescribed minimum, no further action will be necessary.

Now retain assumptions (2) and (3) and relax assumption (4). It has been argued by some that the discount houses may borrow at a penal rate, yet need not view themselves as being out of equilibrium; hence, they would not be anxious to take steps to liquidate the penal debt. Jasay, for example, argues that there is an implicit agreement between the banks and the discount houses according to which the discount houses agree to maintain some debt at a penal rate as part of a general package deal that embraces a number of arrangements within the banking system.¹⁹ If we accept this line of reasoning, then again Situation 3 in Table 1 may represent full equilibrium in the banking system.

However, in these circumstances it is always open to the monetary authorities to reinforce the original sale of securities to the nonbank public by affecting further sales of securities equivalent to any cash made available at penal rates. In other words, if the discount houses are unwilling to take steps to absorb bank cash by liquidating their debt, the monetary authorities can always bring about the same ultimate result by an active open market policy. A policy of this type by the monetary authorities has come to be known as “netting” of open market operations.²⁰

Now retain assumptions (2), (3), and (4) but relax assumption (5). This is an interesting case, because here one is involved in saying that the discount houses are anxious to get out of debt but are inhibited by difficulties in implementing any of the four steps indicated under assumption (5). They may be reluctant to unload their limited stock of short-term bonds. They may not be able to obtain any sizable nonbank funds, the scope for reducing their commercial bill holdings maybe quite limited, and, finally, most important, the nonbank market for treasury bills may be thin and relatively inelastic to changes in the bill rate. To the extent that these restrictions apply, an uneasy equilibrium may settle in Situation 3. But here again the monetary authorities may pursue a policy of netting their open market operations and thus ensure that Situation 6 is ultimately realized.

Finally, note that assumption (1) has been retained up to now. Would the results be different in any way if we supposed that the monetary authorities began the process by selling treasury bills to the discount houses?²¹ We will not develop this case in any detail. For example, where the banks hold no spare cash, as is normal, the discount houses will finance the purchase of new treasury bills by penal borrowing, but they will then take steps to liquidate this debt, and in the end deposits will have fallen by 12.5 times the initial sale of bills. If, however, assumptions (3), (4), and (5) do not hold, there may be some difference between security sales to the public and sales of bills to the discount houses. For example, if the discount houses take no action to repay the debt, then the bank may experience no change in cash or deposits.²²

We may now briefly summarize our results. We concluded that if the discount houses are accommodated at a penal rate and if, either at the initiative of the discount houses or at the initiative of the monetary authorities (“netting” operations), cash equivalent to penal borrowings is absorbed, then the relevant multiplier is 12.5; i.e., the drop in deposits will equal 12.5 times the original sale. This result also holds irrespective of the state of bank liquidity. If borrowing is nonpenal, liquidity ratios are rigidly maintained, and sales are made to the nonbank public, the relevant multiplier is probably 3.5. If there is excess liquidity, in these conditions, the relevant multiplier is probably 1.

A number of different models have been examined, each model incorporating a particular set of assumptions. Clearly, a most important question concerns the empirical relevance of the alternative assumptions. Unfortunately, on this point nothing very conclusive can be said. First, it is a fact that the liquidity ratio has on frequent occasions been in excess of the prescribed minimum. Second, it is also a fact that the Bank of England has at various times accommodated the banking system at non-penal rates. Third, it remains a matter of some controversy whether the discount houses make an active effort to liquidate debt at penal rates. On the whole, what evidence is available appears to be consistent with the postulate that penal debt does represent a disequilibrium situation.²³ Fourth, there is also considerable disagreement as to the empirical relevance of our assumption (5). Since treasury bills represent the most important asset held by the discount houses, the public’s elasticity of demand for treasury bills at different interest rates becomes a particularly critical consideration. On this point there is now evidence to indicate some flexibility in the public’s demand for treasury bills.²⁴

To conclude, then, our discussion of open market operations, it is a reasonable hypothesis that base money is in the main controlled by the monetary authorities in the United Kingdom,²⁵ provided, of course, that the authorities are prepared to accept corresponding changes in the yields on securities. Perhaps the most significant result is that the change in base money is not a simple accompaniment to open market sales. Sometimes an “active” netting policy on the part of the authorities may be required. In every instance the dynamics of adjustment is complicated, with changes occurring stage by stage to bring the result to an ultimate conclusion.

In this section, we excluded altogether currency demands by the non-bank public. When currency demands are allowed for, the relevant multiplier (i.e., the change in deposits divided by the initial sale of securities)²⁶ is considerably reduced, to something like 2.9, which means that an open market sale of one unit will on average reduce deposits by something like 2.9 units and not by 12.5 units as suggested in the simplified examples in this section.

The bank rate

Consider a rise in the bank rate with no accompanying open market sales. A number of rates conventionally tied to the bank rate will also rise. For example, the rate on bank advances has been, for most customers, about 1 per cent above bank rate subject to a minimum of 5 per cent; the basic call loan rate is about 1 to 1¾ per cent below bank rate, while the deposit rate is about 2 per cent below bank rate. With higher call money rates and also the threat of higher penal rates,²⁷ the discount houses will lower their bid price on bills at the tender; at the same time, with higher rates on deposits now offered, the nonbank public will require a higher yield on bills. Hence, the bill rate may be expected to rise. The rise in the bill rate in turn is likely to spread to the bond rate as well as to other rates in the economy. Some rates will adjust only slowly, so that in the short term some funds may switch between different assets; e.g., building society share rates may rise only after a lag during which building loans may decline.

This is an interesting outcome, because it means that interest rates in general may rise, and the bank rate therefore may exercise a deflationary impact, even without any fall in the money supply.²⁸

It might be thought that the rise in the bank rate might itself be responsible for a fall in the money supply, but it is difficult to envisage a reliable mechanism that could bring about such a result. For example, it may be argued that a rise in the bank advance rate might reduce the demand for advances and the fall in advances might reduce deposits, but this argument is weak in at least two respects. First, if the market for advances is a disequilibrium one, with demand in excess of supply at the going interest rate, an increase in bank advances is a more likely outcome, with the banks switching from investments to advances to take advantage of the higher rate. Second, even if the public were to reduce their advances and if deposits also fell, the banks would now be holding excess cash and would be anxious to expand their investments.

A second example of possible causation running from the bank rate to the money supply may be more interesting. Consider a case where the discount houses are in debt to the Bank of England, yet they have taken no measures to liquidate the debt. The higher bank rate will become applicable after a lag and will increase the cost of carrying debt. It is conceivable that the discount houses will now have an incentive to sell securities and to repay some of this debt. These sales and repayments will then influence the money supply in ways suggested earlier.

In a recent contribution, Walters suggests a mechanism whereby a rise in the bank rate might actually increase spending. “. . . the higher rates of interest will encourage those who hold money balances to put them to work—to invest them in financial institutions, in hire purchase debt, etc. The higher interest rate will tend to increase the supply of credit. Now if many of those who seek credit are not deterred by the higher rates of interest (and this argument runs through much of the evidence in the Radcliffe Report) and if their spending is primarily limited by the quantity of credit they can obtain, the raising of Bank Rate will actually increase spending.”²⁹ This sequence of events would appear to depend on two assumptions, both of which may be open to question. First, it seems to rely on the unsupported notion that, notwithstanding a rise in the deposit rate, there will remain a net incentive to shift out of money balances. Second, it also assumes that there is an excess demand for funds from a number of financial institutions and that the increased availability of funds will now absorb part of the excess demand.

Special deposits

Since 1959 the Bank of England has had the power to require the banks to hold with it in cash a certain percentage of their deposits. The required percentage has varied on a number of occasions since 1960 from 0 to 3. These special deposits, which are not counted as liquid assets, earn a return that is quite close to the return on treasury bills. From the beginning the Bank of England decided that the cash required by the banks to accommodate the special deposit requirement would be provided by the Bank by the purchase of treasury bills. It appears that the authorities did not anticipate that the special deposit scheme would have any marked effect on the level of deposits but hoped that the pressure on bank liquidity as a result of a rise in special deposits would induce the banks to restrict their advances. We will, in what follows, briefly examine the likely effects of a rise in the special deposit ratio.

Suppose that the initial position of the banks is as shown in Table 2, Situation 1, where the special deposit ratio is assumed to be 0.01; the banks are shown as holding exactly 8 per cent of deposits in cash and 28 per cent of deposits in liquid assets. It seems clear that if banks hold liquid assets in excess of 28 per cent then a rise in the special deposit rate is almost certain to have no effect whatsoever on cash, deposits, or investments and advances. In this instance the banks will simply exchange excess liquid assets (treasury bills) for cash, which would then be used to meet the special deposit requirement. Suppose now that the special deposit rate rises to 0.02. It is worth considering two alternative solutions corresponding to different reactions by the banking system. We would expect the banks in the first place to obtain cash by selling some of their bills. The banks will then be short of liquid assets. They may respond in one of two ways: they could first reduce their deposits and investments and advances to accommodate the lower level of liquid assets. This is the solution shown in Situation 2a. One can envisage this solution coming about in the following way. As investments and advances and deposits fall, “excess” cash is created. This excess cash is then lent to the discount houses who in turn either “pay off” some debt with the Bank of England or, alternatively, buy up treasury bills from the Bank of England operating through the back door. In this way, some cash would be lost to the banking system.

Table 2.

Impact of Changes in Special Deposit Ratios by the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

Table 2.

Impact of Changes in Special Deposit Ratios by the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

	Situation
	1	2a	2b
Deposits	1,000	966.7	1,000
Cash	80	77.3	80
Liquid assets	200	193.3	200
Special deposits	10	19.3	20
Investments and advances	710	676.8	700

View Table

Table 2.

Impact of Changes in Special Deposit Ratios by the Bank of England on the Equilibrium of Commercial Banks in the United Kingdom

	Situation
	1	2a	2b
Deposits	1,000	966.7	1,000
Cash	80	77.3	80
Liquid assets	200	193.3	200
Special deposits	10	19.3	20
Investments and advances	710	676.8	700

More likely, however, they can make up for the loss of liquid assets by selling investments and advances (10 units) and buying bills from the public (10 units)—Situation 2b. In this case deposits will be unchanged.³⁰ To the extent that advances fall, the objective of the monetary authorities might still be satisfied. It has been argued, however, that the major part of the adjustment is likely to fall on investments and not on advances, and indeed that advances may even increase following a rise in the special deposit ratio.³¹ The rationale here is that since the liquid asset ratio plus the special deposit ratio must now rise, the mean return on the banks’ portfolio will have fallen; the banks then will have an incentive to substitute higher earnings assets (advances) for bonds, or at least to minimize the drop in higher earnings assets.

If bonds are liquidated and bills are bought, at a given level of deposits, the term structure of interest rates may change. With given supplies of short-term bonds and bills, a fall in the demand for bonds and a rise in the demand for bills will tend to raise the bond rate relative to the bill rate. Hence, we conclude that the most important effect of a rise in the special deposit rate is likely to be a change in the term structure of rates and not a reduction in the level of advances. These effects appear to be different from those intended by the authorities.

The liquid asset ratio

The Bank of England has powers to vary the liquid asset ratio of the banking system. The prevailing rate is 28 per cent. The effect of a rise in the liquid asset ratio is almost identical to the effects of a rise in the special deposit rate. The level of deposits is likely to remain unchanged with a switch from investments and advances to bills. Again there will be an incentive to minimize the switch out of higher earnings assets (advances), and the term structure of rates is likely to move in favor of bonds.

Funding

In general, funding involves any lengthening of the average maturity of the public debt. More particularly, in the U.K. context, the notion of funding is usually associated with a reduction in the floating debt matched by an increase in the availability of bonds. As a general principle, a policy of funding the public debt, particularly during inflationary periods, appears to have received a good deal of official support in the postwar years. The desirability of funding has been based on a number of considerations. One important argument was that the reduction in the availability of liquid assets to the banking system would force a contraction in deposits (deposits, on this view, being determined by the supply of liquid assets).³² Other arguments used in defense of funding were that funding itself entails a higher rate on longer term bonds, which may have a stabilizing effect during inflation; that a lengthening of the average maturity of the debt will make it more difficult for the private sector to sell securities to finance expenditure; that a given change, in whatever direction, in the interest rate will have a larger impact on capital values, the longer the debt, and this might influence expenditure on consumption and investment either directly or indirectly through its effect on lending by financial institutions; that a smaller floating debt means less continuing concern over the finance of maturing securities.

If funding was desirable in principle, it was thought that there were many practical difficulties in implementing such a policy. The difficulties were that raising long-term rates might not result in an increased demand for bonds if rates were expected to rise further; that it was important to maintain orderly markets, and therefore there were limits to acceptable fluctuations in rates; that a deliberate policy of forcing up rates at the long end represented an act of “bad faith” toward holders of securities in general; finally, that funding increases the cost of servicing the public debt.

A proper appraisal of all these arguments would involve an extended discussion of the principles of debt management that is beyond the scope of this paper. It will be sufficient in this section to examine the validity of the first argument only, i.e., the case for funding as a means of monetary control.

A strong case against this particular argument has been advanced recently by Crouch.³³ He argues that a policy designed to reduce the floating debt and to release more bonds will almost certainly leave the money supply unchanged and will raise long-term rates relative to short-term rates. There are several routes by which this outcome may be brought about. One mechanism suggested by Crouch is the following: the discount houses will maintain their holdings of bills simply by raising their bid price at the tender; this will leave a reduced supply of bills for the nonbank public, which will now use its increased cash to buy the more attractive bonds being offered. In the new equilibrium the banks are unaffected, the discount houses hold the same portfolio (with bills now carrying a lower return), and the public has replaced its bills with bonds. Crouch finds, in fact, that on the occasions when funding did take place, something like these results could be observed.

Whether a change in the structure of rates in favor of long-term rates is a desirable outcome will depend on the assumptions made about the state of the economy and the responsiveness to particular changes in rates. Consider an economy that is experiencing both inflation and balance of payments difficulties; suppose that both short-term rates and long-term rates influence expenditures but the long-term rate operates with a long lag that may be destabilizing. Under these conditions the change in the structure in favor of long-term rates will be clearly undesirable. Under a different set of assumptions a policy of “defunding” (buying long and selling short) may be desirable. Consider an economy with balance of payments difficulties and a high rate of unemployment; suppose now that the short-term rate has little effect on expenditures but has a strong effect on short-term capital flows; suppose also that long-term rates are stabilizing. Then raising short-term rates and lowering long-term rates (assuming the policy is feasible) may improve the balance of payments and conceivably may raise the level of economic activity. (This was the rationale of “operation twist” in the United States.)

Prescribing limits on advances

In the United Kingdom in the postwar years banks have been requested to limit their advances at various times. Suppose that advance limits are prescribed without corresponding limits on deposits. In this case banks can maintain their level of deposits by switching out of advances and into other assets, e.g., investments. It was argued elsewhere that even if this happened some net deflationary impact was still possible, i.e., the deflationary effect of the fall in advances exceeding the expansionary effect of the increased security holdings.³⁴ On the whole, banks would be expected to minimize their loss of earnings so that if they were holding excess liquid assets, they might well shift out of these liquid assets into the higher earning investments.³⁵ Interestingly, this would result in a change in the term structure but this time the change would be in favor of the bill rate, i.e., a rise in the bill rate relative to the bond rate.

Hire-purchase restrictions

Hire-purchase terms have been changed frequently in the postwar years. It is sufficient here to note that there is evidence that these changes have had strong effects on the demand for consumer durables.³⁶

The transmission mechanism—general

The transmission mechanism defines the way in which monetary changes may make their impact on expenditure. It is useful to distinguish five channels through which monetary policy may operate: an interest rate effect, a real balance (or wealth) effect, a “credit rationing” effect, a “quantity theory” effect, and an “announcement” effect.

The interest rate transmission mechanism

The control of the rate of interest

In the short term, the rate of interest may be influenced by a number of considerations: current and past values of the ratio of money to income (or, more generally, income and the money supply); expectations with respect to future short-term or long-term rates; the expected rate of price change; changes in the weights of particular sectors with different liquidity needs relative to their transactions; changes in the distribution of income (with different factor incomes having different liquidity needs); financial changes and innovations that may influence the demand for money; changes in controlled rates (e.g., the bank rate) tied to other rates in the system. If we isolate the role of the ratio of money to income (the Cambridge ratio) and assume that all other influences are constant, we can trace a relationship in the short run between the money/income ratio and the rate of interest. This relationship, assumed to be nonlinear, is shown in Figure 1. In order to induce the public to hold a lower ratio of money to income, a higher rate of interest will need to be offered. Moreover, the lower the ratio, the larger will need to be the rise in the rate of interest in order to bring about a given fall in the ratio. The lower the Cambridge ratio, then, the larger will tend to be the rise in the rate of interest for a given reduction in the money supply; in this limited sense, a low ratio is favorable to monetary policy.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

The crucial question for monetary policy is the extent to which the authorities are able in the short term to manipulate the rate of interest. Other things being equal, there are three ways in which the rate of interest may be directly sensitive to government actions: First, and most important, by changing the money supply. Second, by changing “controlled” rates in the system (e.g., a change in the fixed deposit rate relative to “outside” rates will increase the demand for fixed deposits and will result in a switch away from other claims or securities). Third, and most difficult, by operating on expectations with respect to future rates. If the authorities can induce an expectation that rates in the future will be higher, current rates may also rise. Whether the monetary authorities will in fact be successful in making an impression on the rate of interest will depend critically on the weight and dominance of the other influences (not responsive to direct government policy), some of which were listed at the beginning of this section. For example, if the Cambridge ratio-interest rate schedule is relatively stable in the short run, a change in the money supply will tend to bring about a predictable change in the rate of interest. If, on the other hand, the schedule is relatively unstable and shifts rather erratically in the short period, then a change in the money supply may be consistent with a whole range of interest rates.⁴³

So far we have considered the extent to which the monetary authorities are able to bring about a change in interest rates in general. We saw in PART A that monetary actions frequently have effects on the term structure of rates. The theory underlying the term structure is briefly discussed in the Appendix. A change in the term structure of rates as such may change both the composition of expenditure and the total of expenditure. Suppose that the long-term rate rises relative to the short-term rate; then investments financed by long-term capital will be harder hit than expenditures financed by short-term borrowings (e.g., for inventories and consumption). Suppose now, to take an extreme example,⁴⁴ that the short-term rate rises and the long-term rate falls, then the effect on total expenditure will be determined by the interest elasticity of demand of short-term and long-term rates.

The rate of interest and investment

We will deal briefly with some of the considerations that have been thought to be relevant in determining the response of investment to a change in the rate of interest. First, other things being equal, long-term projects are generally thought to be more responsive to the rate of interest than short-term projects. This is usually explained on two grounds.⁴⁷ One is that interest payments represent a smaller proportion of amortization charges, the shorter the life of the project; the other is that the present value of an investment yielding a flow of receipts is more sensitive to a given change in the rate of interest, the longer the expected life of the investment. The latter explanation in particular has not gone unchallenged.⁴⁸ Second, the greater the risk and uncertainty attaching to a project, the weaker will tend to be the interest rate effect. If the required margin for risk and uncertainty is large, a given change in the rate of interest will represent only a small proportion of this margin. Third, a rise in the rate of interest may produce no change in the real rate of interest if it is accompanied by an equivalent rise in the expected rate of inflation. It is conceivable that periods of rising rates of money interest are also conducive to expectations of an acceleration in the rate of inflation. Fourth, it is sometimes argued that if changes in the rate of interest could be passed on by businesses, this might weaken the effects of monetary policy. Since, however, businesses are affected quite differently by a change in the rate of interest, it is difficult to see how, at least in the short run, this shift in the rate of interest could eventuate.

Fifth, the weight of internal funds in the finance of investment may be significant in determining the effect of the rate of interest (see Figure 2). DD₁ represents the amount of internal funds (depreciation and retained profits) available to the firm. The rate of interest (D) earned on internal funds is assumed to be lower (by D₁G) than the minimum rate imputed to external funds (G). This difference may reflect a gap between the firm’s lending and borrowing rate or, alternatively, it may reflect the firm’s reluctance to incur additional debt. The interest cost of external funds is shown as a function of the size of external funds. This can be explained either in terms of the firm’s unwillingness to incur additional debt or in terms of the lender’s unwillingness to continue to provide loans at a fixed rate of interest. MEC₁, MEC₂, and MEC₃ represent three alternative positions of the marginal efficiency of capital schedule (each schedule showing the expected gross return at different levels of investment). Suppose now that the rate of interest rises, and suppose that we disregard valuation effects (considered earlier). If the imputed rate of interest on internal funds is not correspondingly raised (the horizontal line DD₁ continues to apply), then investment will be unresponsive for both MEC₁ and MEC₂. If the rate of interest on internal funds is written up (line DD₁ shifts to EE₁, then investment in case 1 may be affected but investment in case 2 will remain unresponsive. If the firm operates at MEC₃, the investment will be affected in the normal expected way (with the rate of interest line shifting upward from F₁ to F₂). It follows, therefore, that the more that firms operate in the discontinuous portion of the schedule, the larger the gap between the internal and external rates of interest, and the less the inclination to raise internal rates in line with external rates, the weaker will tend to be the interest rate effect.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

What this shows is that a fall in internal liquidity effectively raises the weighted cost of capital to business. Since businesses tend to become relatively illiquid and to depend more heavily on external finance in the later stages of a boom, the effective weighted cost of capital would rise in this phase of the cycle. If, in addition, the market rate of interest is allowed to rise, the effects on the cost of capital may be magnified.

Sixth, it is also argued that the higher the rate of tax on profits, the lower will tend to be the response to a given change in the rate of interest.⁴⁹ If the rate of tax is 25 per cent, the profit rate 10 per cent, and the rate of interest 2 per cent, the net profit rate will be 6 per cent. If the rate of interest rises to 4 per cent, the net profit rate falls by 1½ percentage points to 4½ per cent. If now the rate of tax is 50 per cent, a rise in the rate of interest from 2 to 4 per cent will lower the net profit rate by only 1 percentage point, from 4 to 3. In the form in which this argument is generally presented, it is incorrect. Suppose that t is the rate of tax, MEC the marginal efficiency of capital, and R the rate of interest, then the net profit rate will be (1—t) (MEC—R), where (MEC—R) represents the gross rate. It is true that the larger the t, the smaller the change in the net profit rate from a given change in the gross rate, but as long as the net rate is positive, the incentive to invest will remain. If the gross return is negative, no tax applies. Therefore, any project that was considered worthwhile at the lower tax rate must continue to be worthwhile at the higher tax rate. However, the argument could have some validity when risk, uncertainty, or income concepts for tax purposes are introduced.

Seventh, business sensitivity to the rate of interest may also depend on the techniques used to appraise investment projects. For example, it may be that the “payoff criterion” (requiring the receipts of an investment project in the first two to five years to cover the initial outlay) makes investment less sensitive to the rate of interest than, for example, a “discounted cash flow” technique.

Eighth, since government and public utility investment tends to be largely immune to monetary policy, the greater its weight in total investment, the less will be the impact of monetary policy.

Before concluding this section, a word should be said about Tobin’s approach to the question of the cost of capital.⁵⁰ He argues that the true cost of capital is represented by the equity yield, not the bond rate. It is the relation between the equity yield and the marginal productivity of capital that determines the rate of new investment.⁵¹ If monetary policy is to make an impact on expenditure, therefore, it must influence the equity yield. Monetary policy generally operates by changing the relative supplies of money and government bonds. The question then is how policies of this type may alter the equity yield. Tobin assumes the existence of three types of financial assets—money, bonds, and equities. He further supposes that while money and bonds are close substitutes, neither is a good substitute for equities. Consider now three types of monetary policy. First, an increase in the supply of interest-bearing debt as a result of an excess of government expenditures over tax receipts. In this instance the interest rate on bonds will rise and the equity yield will fall. The fall in the equity yield will be brought about by an attempt to shift out of bonds into equities. Here then we have a result that is contractionary in terms of the bond rate but expansionary in terms of the equity yield. Second, an open market operation that increases the supply of bonds and reduces money. Here both the bond rate and the equity yield will rise. Third, a reduction in the supply of deposits brought about by a change in reserve requirements. Again the end result will be some increase in both the bond rate and the equity yield. It does seem, then, that the most pronounced divergence between the Keynesian approach in terms of the bond rate and the Tobin approach in terms of the equity yield occurs in the first case, with the other two types of monetary policy yielding movements in the two indicators in the same direction.

The role of nonbank financial intermediaries

In recent years there has been considerable discussion of the situation in which nonbank financial intermediaries may weaken the effectiveness of monetary policy. It is useful to distinguish two major ways in which these intermediaries may be in a position partially to offset the impact of monetary policy.

First, during a credit squeeze they may raise the rate of interest that they pay on their liabilities relative to the rate of interest on time deposits,⁵² attract more funds from the public, and expand their lending. Finance companies, building societies, and savings institutions are examples of intermediaries capable of behaving in this way, in theory. In effect, these institutions act as intermediaries in channeling “idle” funds to final spenders. These activities may reduce the effectiveness of monetary policy in one of two ways. One way is to represent funds loaned by financial intermediaries as creating a direct effect on expenditure. This would be the counterpart of the credit rationing effect for the banking system. Essentially, this is an argument that would have some force in the short run (e.g., if in a particular period the flow of funds through certain intermediaries were stopped) and where capital markets are unsophisticated.

Another way is to represent the financial intermediaries as permitting a given transfer of funds to the spender with a smaller rise in the interest rate than in the absence of intermediary financing. The more liquid the claims created by the intermediaries and the greater the substitutability of their credit for bank credit, the greater the ease with which funds are channeled to final spenders. In terms of Figure 1 the existence of intermediary finance lowers the interest elasticity of the Cambridge ratio-interest rate schedule in the relevant range, so that a given percentage reduction in the money supply results in a smaller percentage increase in the rate of interest than otherwise.⁵³ This would complicate the problem for monetary policy only if there were certain costs involved in bringing about larger percentage changes in the money supply, e.g., the effect on bank profits or discrimination against the banking system.⁵⁴

Second, certain financial intermediaries (e.g., life insurance companies, superannuation funds) cannot expand their lending by creating new liabilities in the short run. They may, however, still be destabilizing to the extent that they switch between government securities and private securities during a squeeze. In this respect, banks behave in ways quite similar to financial intermediaries: by switching they do not create money but simply act as intermediaries in transferring “idle” funds from surplus to deficit units. Again, this switching operation might be represented as directly affecting expenditure or lowering the interest elasticity of the schedule in Figure 1.

If financial intermediaries were destabilizing in any of these ways, the monetary authorities could take a number of possible steps to deal with the situation. One might be to impose ceilings either on the rate of interest on intermediary claims or on the rate of interest at which they could lend to the public. In the latter instance, switching out of government securities would also be discouraged, since they would not have the incentive provided by the higher rate on private loans. Another step might be to impose secondary reserve requirements on both the banking system and those intermediaries in a position to engage in switching. Again, lending by financial institutions could be attacked directly by imposing ceilings on their loans.

In general, central banks are anxious to avoid resort to direct controls; for this reason they have tended to rely on two indirect methods of influencing intermediary lending. One is to raise the rate of interest on time deposits and so discourage shifts into intermediary claims. Intermediaries could of course respond by raising their rates further to attract funds but in the end the higher rates will tend to depress expenditure. A second indirect method is the “locking-in” effect. The argument here is that a rise in the rate of interest on government securities, engineered by monetary policy, might discourage switching by banks or insurance companies.⁵⁵ This discouragement will tend to be the more pronounced, the more rigid the rate of interest on private securities, the greater the reluctance to take capital losses, the greater the concern over portfolio liquidity, and the greater the expectation that the rate of interest will fall again shortly. On the other hand, if the rate of interest on private securities is flexible upward, and the demand for funds is relatively inelastic to the rate of interest, switching is not likely to be discouraged. More important, switching need not necessarily entail losses by these institutions. Banks can switch government securities that are currently maturing or that have a very short life; insurance companies can divert a larger proportion of the new inflow of funds into private securities.

In the end, many of the questions raised here can be settled only by empirical work. To what extent do financial intermediaries actually create claims that are close substitutes for money and increase their lending during a credit squeeze? Is switching accelerated by a credit squeeze? Does the interest rate on intermediary claims show a relative rise during a squeeze?

The role of lags

A most important issue is the length of the lag in the implementation of monetary policy. In general, the shorter the lag, the greater the likelihood that policy will be stabilizing; a long lag means that policy continues to make a significant contribution into the relatively distant future when conditions that originally provoked the policy change may have materially altered. There are several types of lag involved. The “inside” lag is made up of a “recognition” lag (i.e., between the need for a change of gear and the recognition of the need) and an “action” lag. The “outside” lag, which is the length of time from the policy action to the impact on production, may be broken down into a number of distinct lags: a bank reaction lag; a “financial intermediary” reaction lag; a lag in decision taking by households and business units; a production lag that makes its impact felt over several periods.

Given the number of lags involved, the “average” lag is bound to be variable depending on the specific conditions of a monetary change. For example, the lag may be shorter if banks restrict loans rather than reduce their holdings of securities; monetary policy will vary in its impact on different components of expenditure, and each component will have its own “production” lag (see below); bank and financial intermediary reaction will vary. For example, with banks where overdraft limits (approvals) are granted, there will be an element of inflexibility in the response of loans to a restrictive policy; a “bills only” policy will operate with a different lag from a policy that makes a direct impact at all ends of the market.

Given a decision to change expenditure, the distribution of the production effect (or lack of effect) over time will depend on the type of expenditure affected. One component of expenditure would have a shorter lag than another if, say, over six months a larger percentage of the “total” effect had been realized. Consumption and inventories may have short lags in this sense and residential investment a somewhat longer lag, while equipment and commercial construction may have the longest lags. Consider now the situation in which monetary policy makes its impact only on expenditures with the longest lags and in which these lags are too long for effective stabilization. Then the case against discretionary monetary policy for stabilization would be strong. Consider now the situation where monetary policy makes its impact on the whole range of expenditure. In these conditions it becomes uncertain whether monetary policy will have a net stabilizing effect. If destabilizing results are more likely from the long end of the market and stabilizing results from the short end, then there may be a case for concentrating on the short end of the market (an active short-term rate policy) and maintaining the long end relatively stable over cycles. (The feasibility of this solution is another matter.)

The interest rate transmission mechanism

The role of other liquidity variables in investment and consumption

The rate of interest is not the only liquidity variable found to be significant in studies of investment and consumption. A brief review of these other liquidity variables is therefore important.

Quantity theory studies

We have seen that one possible rationale for using liquid assets in expenditure functions (in particular, consumption functions) is a “quantity-theory” type effect. Apart from studies using money balances in expenditure functions, two other types of study are relevant in appraising the significance of a “quantity theory” effect. One relates income directly to the money supply in current and earlier periods. For the United Kingdom the work of Walters⁸¹ and Crouch⁸² is relevant. Walters relates income to the money supply in the current and five previous quarters (in level and first difference forms) in the years 1955–62. Many of the coefficients are insignificant with some quarters showing positive and others negative coefficients.⁸³ For the level equations the coefficients sum up to +1.7, suggesting a six-quarter elasticity of well over 1 for income with respect to money.⁸⁴ Crouch, in his monetary model, has estimated the following equation for quarterly data for the years 1954–65 (y = income, D = public cash + deposits):

According to this equation, income is a function of deposits in the previous quarter and all earlier quarters, with declining weights attached to the earlier periods (assuming that the coefficient for y_t–1 can be taken seriously as reflecting a distributed lag function). Interestingly, although Crouch’s equation is essentially different in construction from Walters’, the long-run elasticity of income with respect to deposits (measured at the means) is 1.80, a result quite similar to Walters’.⁸⁵

Most of the empirical work on the relevance of a quantity theory effect in the United States has been inspired by Friedman. Reviewing the historical behavior of money and income, Friedman finds that the amplitude of income has been considerably greater than the amplitude of money. He explains this difference in amplitude in terms of his permanent income hypothesis: Money balances are adjusted to permanent income, and the amplitude of permanent income is considerably less than the amplitude of measured income. The rise in velocity in the upswing and the fall in velocity in the downswing are taken as evidence of the powerful effects of money.⁸⁶ He estimates in fact that the elasticity of income on average with respect to money approaches 2, so that a 1 per cent increase in the money supply has tended to raise income by something like 2 per cent.⁸⁷ One difficulty with this analysis is its relevance to the postwar years. The high elasticity of income with respect to money is the result of severe fluctuations in measured income. Since in the postwar years cycles have been considerably milder than in the prewar years, the elasticity of income will presumably have fallen well below 2. Indeed as cycles are eliminated and differences between permanent and measured incomes disappear, the elasticity of income will be less than 1. This is because Friedman treats money as a luxury good in the long run, with desired money holdings rising at a faster rate than income.

Another group of studies inspired by Friedman is concerned with comparing the long-term relative efficiency of Keynesian and quantity theories in explaining the behavior of income. Friedman and Meiselman initiated this exercise for the United States for the period 1897 to 1958.⁸⁸ They concluded that, in the main and taking the period as a whole, the behavior of consumption is better explained by money than by “autonomous” expenditures (the latter representing the Keynesian explanation of “induced” expenditure).⁸⁹ On the whole they tended to play down the role of autonomous expenditures and took the view that money tended to be the most important determinant of expenditure.⁹⁰ A number of doubts were subsequently raised about these results.⁹¹ For the United Kingdom, Barrett and Walters applied rather similar tests for the period 1891 to 1962 including a number of subperiods.⁹² They defined autonomous expenditures as the sum of private domestic investment, government expenditure, and the excess of exports over imports. Over the whole period, autonomous expenditures appear to provide a slightly better explanation of consumption than money; but for certain sub-periods (e.g., before World War I) this result is reversed. The final conclusion, which is not surprising, is that “both money and autonomous expenditure have been important factors in the determination of aggregate consumption.”⁹³ A typical equation for the most recent subperiod (1948–63) using both money and autonomous expenditure is the following (c = consumption, a = autonomous expenditure, m = money supply):

The author has himself experimented with equations of this type for the United Kingdom, and it may be useful to report the results of the best equation estimated in first difference form from annual data for 1951–67.

Money here is narrowly defined to exclude deposit accounts, and A represents the sum of government expenditures, investment in fixed capital, and exports. The explanatory power of the equation is good for first differences. Autonomous expenditures in the same period are insignificant and have the wrong sign. One surprising result is that autonomous expenditures operate with a longer lag than money.

Three brief comments on the Friedman and Meiselman type of tests are in order. First, few types of expenditure can be unambiguously classified as either “induced” or autonomous of income. Private domestic investment, treated as autonomous in these studies, is obviously related in a complicated way to the behavior of income (e.g., an accelerator-type function). Consumer durables and the external balance are also partly induced and partly autonomous. Second, sophisticated Keynesianism does not imply a simple relationship between “autonomous” expenditure and consumption of the type tested in these studies. Third, where the money supply and autonomous expenditure are both quite significant, as in the above function, this can easily be shown to be consistent with a simple variant of Keynesianism. For example, all expenditures (which make up income) are a function of income, the rate of interest, and autonomous expenditures. The rate of interest in turn is determined by income and the money supply. In this simple scheme, income is then determined by autonomous expenditure and the money supply (a reduced form of a simple Keynesian system).⁹⁴

A recent study for the United States by Anderson and Jordan has aroused considerable interest and comment.⁹⁵ Anderson and Jordan regress changes in income against an indicator of monetary policy and an indicator of fiscal policy. The study is very much in the spirit of the original Friedman and Meiselman contribution, since it is designed to appraise the relative merits of a monetary model against a Keynesian-type model (this time represented by fiscal policy instead of autonomous expenditures). Their own results showed, first, that monetary policy exerts a larger influence on economic activity than fiscal policy; second, that the response of economic activity to monetary policy is more predictable and reliable than the response to fiscal policy; third, that there was no evidence that fiscal actions were quicker acting than monetary actions. These results, then, were strongly monetarist and attributed rather weak effects to fiscal actions.

Later work for the United States, particularly by de Leeuw and Kalchbrenner, threw some doubt on these results.⁹⁶ These authors questioned the assumption that money was truly exogenous, since it also responded to movements in income. More important, they ran regressions for alternative indicators of monetary and fiscal actions.⁹⁷ The most striking difference was obtained with a different indicator of monetary policy:⁹⁸ now monetary policy becomes substantially weaker and fiscal policy much stronger and more reliable. A similar study has also been applied to the United Kingdom recently.⁹⁹ If we take this study seriously, the results for the United Kingdom turn out to be the exact opposite of those obtained by Anderson and Jordan. Fiscal policy now performs better and with a smaller lag.

The role of nonbank financial intermediaries

The only fairly comprehensive study of financial intermediaries in a period of monetary restriction in the United Kingdom is by Clayton.¹⁰⁰ His method is to examine the actual behavior of a number of institutions that are potentially destabilizing. He finds that building society interest rates tend to be sticky, so that in restrictive periods the flow of funds to these institutions is actually slowed down. The interest rate that finance companies pay on deposits tends to be a fixed margin over the bank rate; since the commercial bank deposit rate is also tied to the bank rate, the absolute margin between the bank deposit rate and the finance company rate is constant. This means that as interest rates rise there is in general no incentive to move into finance company deposits.¹⁰¹ Turning to insurance companies, he does find some actual switching out of government securities in restrictive periods. He also finds that investment trusts may have been destabilizing because they had switched substantially from fixed-interest securities into equities; but since the fixed-interest securities were, in the main, private issues, it is not clear that this would necessarily have been destabilizing.

Another, more abstract and far less satisfactory, way to attack the problem is to examine the extent to which the growth of money substitutes by financial intermediaries may have affected the demand for money. Other things being equal, one would expect, for example, that if these money substitutes were at all significant, they would help to explain the behavior of interest rates in the postwar years, i.e., a broader concept of liquidity that embraces these money substitutes ought to provide a better explanation of interest rate movements than a narrow bank concept of liquidity.¹⁰² There is some evidence from studies by Ball¹⁰³ and Ford and Stark¹⁰⁴ for the United Kingdom that money substitutes have not played a significant part in interest rate determination; however, this evidence should not be taken too seriously.¹⁰⁵

The evidence for the United States on this question is far less relevant for the United Kingdom. First, the types of institution involved are different in the two countries. Second, arrangements and conventions respecting interest rates offered on intermediary claims will be different in the two countries. Third, in the United States reserve requirements are lower for fixed than for current deposits, and this fact is important in analyzing the effects on the supply of loanable funds of shifts from deposits to intermediary claims. For what it is worth then, in the United States detailed studies of financial institutions do not reveal any marked destabilizing behavior.¹⁰⁶ The evidence also suggests that money substitutes need not be explicitly allowed for in studies of the demand for money.¹⁰⁷

The significance of lags

The Radcliffe Report suggested that monetary policy might be too slow to be useful for short-term stabilization.¹⁰⁸ Unfortunately, there has been virtually no research on U.K. lags to permit some judgment on this question. The only research that has any bearing on this issue at all is not very helpful; nevertheless, a brief review of the evidence will be given.

A typical method of estimating the “inside” lag is the following: Consider the equation P_t = aW + bX + cY + dZ + eP_t–1, where P is the monetary instrument, W, X, Y, and Z represent policy targets, and P_t–1 is the distributed lag variable for the policy targets. A recent study by Fisher applies this method for the United Kingdom, using a number of monetary instruments and policy targets.¹⁰⁹ The targets that are significant in some of the equations are the price level, international reserves, and unemployment rates. The e coefficient is large (between 0.6 and 0.9), indicating that much of the response is delayed.

Walters has also tried to measure the “inside” lag by relating money (m) to prices (p) and the deficit in the balance of trade (d).¹¹⁰ The two relevant equations for quarterly data 1955–62 (in logarithmic form) are

It is extremely difficult to extract any meaning from these equations: the signs are different (the first equation, for example, implying that the government increases money as the level of prices rises and the deficit increases); many of the coefficients are insignificant; other possible indicators of government policy (e.g., unemployment rate) are left out.

A direct measure of one “outside” lag is suggested by the Klein (and others) econometric model. In the investment functions, the debenture yield, which as we saw was significant in the later version, appears with a four-quarter lag. In a study by Walters referred to earlier, income is related to the money supply in the current and five previous quarters, but many of the coefficients are insignificant, and some coefficients are positive while others are negative.¹¹¹ Crouch’s equation for income and money, given earlier, suggests that income is a function of money in the previous and all earlier periods.¹¹²

For the United States there is now considerable empirical literature but the position remains somewhat unsettled. Kareken and Solow¹¹³ and Brunner and Meltzer¹¹⁴ estimated the total inside lags (recognition and action) to be in the vicinity of 9 months, on average.¹¹⁵ On the other hand, some writers, including Hendershott,¹¹⁶ Wiles,¹¹⁷ and Havrilesky¹¹⁸ all estimate the lag to be considerably less than that (from zero to 2 to 3 months).

The “outside” lag is more complicated, with lags varying depending on the type of expenditure, e.g., inventories, residential construction, and plant and equipment investment. Kareken and Solow looked at the lag from changes in free reserves to changes in bank interest rates, and then from changes in bank interest rates to inventory investment. The first lag proved to be quite long (with about one fifth of the effect on interest rates occurring after six quarters),¹¹⁹ and the second averaged (meaning 50 per cent of effects felt) about 8 or 9 months. The average lag for the change in the bond rate to the change in producer durable equipment turns out to be between 14 and 18 months, about twice as long as the inventory lag. (No estimate is given of the time it takes for the bond rate to respond.) Kareken and Solow concluded that plant and equipment investment, while probably ultimately responsive to monetary policy, changes too slowly to be of any use for countercyclical policy of the postwar variety. Mayer’s study arrives at only a slightly less pessimistic result.¹²⁰ He concludes that “monetary policy is too inflexible to reduce the fluctuation of industrial production by more than about 5 to 10 per cent on the average.”¹²¹ Ta-Chung Liu also finds the monetary lags to be somewhat long in his econometric model of the U.S. economy.¹²² Two experiments carried out by Liu are relevant for us. In one experiment he traces the effects of a maintained increase in the short-term rate of interest (effectively controlled by the monetary authorities) on gross national product (GNP). He finds that the maximum effect does not eventuate until eight quarters later. In another experiment he supposes that a given error in actual GNP (excess or deficiency) is fully corrected by adjusting the interest rate in the next quarter. It turns out that this policy increases the severity of fluctuations and thus worsens the situation. Friedman’s own work also supports the “long lag” hypothesis.¹²³ He finds that the lead of the money series over business (national bureau reference figures) has tended to average between 12 and 16 months.¹²⁴ Several recent writings also support the “long lag,” although there are significant differences in detail between these studies.¹²⁵ It is difficult then to escape the conclusion that, insofar as fixed investment is concerned, the lag may be somewhat on the long side.¹²⁶