I. Financially Repressed Economies

Many less developed countries have no organized money markets and have negligible capital markets. Savings are invested directly or through banks. When these conditions are combined with “effective” capital controls, money and interest rate targets can be set independently. Such countries often set interest rates below their equilibrium (market clearing) levels, which unfortunately precludes the subsequent development of markets in these financial instruments.¹ These conditions make it possible to independently set monetary and below-equilibrium interest rate targets. This is not possible if rates are set above equilibrium, because at such levels banks may not be able to lend an amount consistent with creating the targeted level of deposits.

The adverse effects on economic development of below-equilibrium interest rates, that is, of financial repression, are well known (McKinnon, 1973; Shaw, 1973; Coats and Khatkhate, 1980). But even in the most backward economies, market forces (i.e., profit incentives) tend to erode repressive controls. Significantly below-equilibrium interest rates leave money borrowers unsatisfied by official domestic credit institutions, who then turn to “unorganized” or black markets for funds at significantly above-equilibrium rates. This stymies the development of organized financial markets, thus impeding economic development in general and diminishes the effectiveness of the very interest rate controls causing the problem in the first place. In addition, repressed interest rates and the nonprice rationing that must accompany them create an incentive for borrowers to find credit abroad and for savers to lend abroad. The capital controls that inevitably accompany repressed interest rates are effective in reverse proportion to their need. The more out of line domestic rates are with exchange rate adjusted world rates, the more capital controls will be evaded. Therefore, even in the most extreme case of a financially repressed, closed economy, there are limits on the independence of monetary and interest rate objectives.

II. Economies with Financial Markets

It is easier for governments to regulate interest rates in primary credit markets, for example, on bank deposits and loans, or on the initial issues of securities, than in secondary (resale) markets. Attempts to regulate sales in secondary markets are generally enough to prevent such markets from existing (Khatkhate, 1977; Porter, 1973). The development of secondary markets in financial instruments is desirable because they generate better (more accurate) information on the opportunity cost of funds, improve the efficiency with which funds flow from savers to investors, increase the attractiveness of saving in general, help rationalize the structure of investment by increasing the probability that what is saved will be made available to the most productive investments, and provide the monetary authorities with greater independence from the finance ministry. Well-established money markets make possible central bank open market operations and potentially weaken the link between government budget deficits and central banks’ creation of base money.

Active markets in financial instruments, both primary and secondary, and the market-determined interest rates they require, eliminate whatever limited independence there may have been without them between monetary and interest rate targets. This is true even in the closed economy case, where monetary authorities are free to set either a monetary or an interest rate target independently of conditions in the rest of the world. This well-known proposition is graphically revealed by the famous Hicksian IS-LM curves (Hicks, 1967, Chap. 7). For the monetary authorities to hit an interest rate target, they must accept whatever quantity of money is consistent with monetary equilibrium (LM curve) at the targeted interest rate and the level of real income consistent with real sector equilibrium (IS curve). A monetary target requires accepting whatever interest rate results from the specified quantity of money (see Figure 1).²

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

As Hicks pointed out, and as the experience of the past decade has made very clear, this analysis exaggerates the monetary authority’s ability to set interest rate targets even in the closed economy context by ignoring the labor market and the eventual adjustment of the price level. That is, in Hicks’s corrected model (Chap. 8), the real rate of interest is not a monetary variable at all. The price level will adjust to any nominal quantity of money so as to intersect the IS curve at the full employment levels of income (y_f) and the interest rate (see Figure 2). Furthermore, continuous growth in money in excess of the growth of income (a money growth rate target) will lead to continuous inflation, which will drive a wedge between market and real interest rates but will not otherwise affect real rates.³ Once the price level is allowed to adjust in the model, the effort to hold the real rate of interest below r_f in Figure 2, which requires the IS-LM curves to intersect to the right of full employment income (y_f), will require accelerating injections of nominal money. However, these temporary increases in real money balances will precipitate exploding increases in the price level. Like a dog chasing its own tail, the process must explode, since there is no market reconciliation possible between the monetary authorities’ interest rate target and the rate required for market equilibrium (r_f). Needless to say, nominal rates will actually rise (with inflation) further and further above the target value the harder the authorities press to achieve it.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

In the open economy case with a high degree of capital mobility, interest arbitrage ensures interest rate parity with the rest of the world (adjusted by the expected rate of change in the exchange rate, which, in general, will equal expected inflation differentials). In the fixed exchange rate case, independent monetary targets are also ruled out, since the addition of commodity arbitrage to interest arbitrage links domestic prices to world prices, making the equilibrium nominal money stock endogenous.

III. Interest Rate Volatility

In the light of these considerations, countries with developed markets in financial instruments have increasingly adopted monetary rather than interest rate targets as the central objective of monetary policy. As a result, the still considerable concern over the behavior of interest rates has tended to shift from their equilibrium level to their variations around that level, that is, to interest rate volatility. The desire to balance avoidance of “unnecessary” interest rate volatility against the achievement of monetary targets has focused attention on the horizon, over which control of the monetary supply should be sought. Closer control of money (i.e., a shorter control horizon) increases the variability of the yields on financial instruments. In addition to other objectives, central banks have traditionally sought to moderate or prevent short-term fluctuations in interest rates out of the concern that

the increased volatility of interest rates would disrupt the functioning of financial markets by increasing risks borne by lenders and borrowers. This, in turn, will spill over into the nonfinancial sector by disrupting the savings-investment process and thereby increase the instability of prices, employment, and output.—Lombra and Struble (1979), p. 285, who cite Hayes.

Most of the central banks of these countries have conducted their day-to-day operations in terms of interest rates, even when pursuing monetary targets in the medium or longer run. For example, the Federal Republic of Germany, the United Kingdom, and the United States (until October 1979), all of whose central banks set monetary aggregate targets, all operate by setting or manipulating interest rates so as to elicit the desired monetary behavior over the longer run. Short-run deviations from longer-run (usually annual) monetary targets have been tolerated on the grounds that (a) tight, very short-run (weekly, monthly, and perhaps quarterly) control of the money supply is not very important as long as the longer-run behavior is appropriate; (b) short-run deviations in money growth tend to be self-reversing; and (c) interest rates would be dramatically more volatile with rigid adherence to very short-run monetary targets, which in turn would have important adverse effects on financial markets.

An empirical assessment of the validity of each of these propositions is needed in order to determine the optimal control horizon. In the United States, these issues are discussed in connection with the choice of a monetary policy “operating strategy.” This and the next sections examine the “key issues involved in assessing the degree of [short-term] rate volatility likely to be generated by the attempt to exercise closer control over the monetary aggregate” (Lombra and Struble (1979), p. 286).⁴

Efforts to assess the likely short-run interest rate consequences of tighter money supply control have focused on the public’s demand for money and the need for interest rates to adjust to changes in the money stock for that market to clear. Ceteris paribus, a lower interest elasticity of demand for money and a slower adjustment of short-run demand (or of supply) to long-run demand mean a more volatile interest rate when controlling the money supply. A widely used formulation postulates a money demand function such as

in which the long-run demand for real money (m) depends on real income (y) and an interest rate (i), which is related to the money supply through a stock adjustment relation as in

This reflects the assumption that adjustment costs make gradual adjustment of supply to demand optimal.⁵ By solving equations (1) and (2) for the interest rate and assuming that in the short run real income and prices adjust little to changes in the money stock or shifts in money demand (ε_m), it is seen that larger changes in i are required to clear the market (i.e., to satisfy equations (1) and (2)) in the face of changes in the money supply, the smaller are the values of a₂ and λ.

Therefore, $\frac{\partial i}{\partial \ln M}=-\frac{1}{a_2\lambda }\mathrm{.}$

Estimated values of a₂ and λ are generally very small (on the order of 0.1), so that ignoring induced changes in y and P implies very large changes in i from changes in M.

A more thorough analysis would require solving a complete model for i. This would take account of the extent to which y and/or P respond to changes in M and thus diminish the work left for i. A partial step in this direction is to add the real sector (IS equation) to the above equations. To keep the analysis simple, the model continues to treat real output as exogenous, that is, the labor sector is omitted, so that it solves for the interest rate and the price level given real output, inflationary expectations, and the money stock. The following equations are added.

Aggregate demand:

where y is real income and r is the real rate of interest.

Fisher relationship:

where Π is the expected rate of inflation.

Partial adjustment of nominal income to demand:

or

where Y is nominal income. All coefficients are defined to be positive.

Combining equations (4) through (6) gives the IS equation:

Combining the IS and LM equations (equations (7) and (3)) so as to remove the endogenous price level, and solving for i gives

where d = a₂λ(1 – α) + αb₂

As real income is fixed, changes in Y reflect changes in P. equation (8) shows that the interest rate effect is reduced to the extent the price level responds to changes in M (i.e., to the extent that α > 0):

when b₂ > 0. Interest-sensitive aggregate demand (b₂ > 0, i.e., a flatter IS curve) also reduces the volatility of interest rates from changes in M. A full treatment would specify, that is, make endogenous, the determination of inflationary expectations (adaptive, rational, etc.) and real output. However, it is possible to specify the direction of the result of making these adjustments endogenous by inspecting equation (8), which includes them as exogenous. They both enter that equation with the opposite sign from that of money. Therefore, if, as is generally supposed, an increase in money increases either Π or y, the negative impact on i of that increase in M will be smaller than indicated by the above partial derivative.

Inferring interest rate volatility from equation (3) is equivalent to setting a to zero in equation (8), in which case the impact multipliers for M are the same in the two equations. The opposite assumption, namely, that α = 1, depicts full, unlagged adjustment (of prices) in the real sector. In this case, M’s impact multiplier is zero; a change in M has no effect on i (assuming no change in Π and y), because full and immediate adjustments in P (which leave M/P unchanged) keep the public on its money demand curve without the need for any change in i, as explained earlier in terms of Figure 2.

IV. Shortcomings of the Framework

Empirical estimates based on the foregoing types of equation suggest very high interest rate volatility. Several considerations cast doubts on this finding. In addition to the inadequate treatment of inflationary expectations, the coefficient estimates of previous studies generally suffer other biases. The treatment of the money stock as exogenous in money demand estimates biases the estimated interest elasticity of demand downward (Lombra and Torto, 1975), although the empirical significance of this bias seems to be small (Havrilesky and Boorman, 1978). However, a more serious downward bias is overlooked in the taxonomy of Lombra and Struble. The opportunity cost of holding money is the difference between money’s own rate of interest (i_d) and that on alternative assets. A change in i matched by an equal change in i_d (i.e., no change in i–i_d) will have little or no effect on the demand for money. Estimates of specifications like equation (1), which omit money’s own rate of return, will reflect the combined impacts of the change in i (i.e., ∂lnm^d/∂lni) and the change in i_d (to the extent i_d moves with i), that is, (∂lnm^d/∂lni_d) (∂i_d/∂i). If i_d is reasonably constant (or if its movements are independent of i) within the time unit used in estimating a₂, its omission (absorption into the constant term) does not bias the estimate. This assumption is obviously correct for the currency component of M but questionable for the deposit components when using monthly or longer data.

Interpreting the low estimates for a₂ as the impact interest elasticity of demand for money implicitly assumes that the own rate of interest on money is zero or constant, that is, that interest rate controls are fully effective. Klein (1974) has argued that the opposite assumption, that deposits yield a competitive return, is a better approximation of the truth. Using annual data, Klein finds that the estimated coefficient of the interest rate term in a money demand function is increased by a factor of four when i is replaced by i–i_d for the i_d proxy he has constructed.

In estimating the daily or weekly volatility of i, ∂i_d/∂i is likely to be approximately zero. Therefore, Klein’s dramatically higher elasticities, obtained from monthly or quarterly estimates of a₂ that include i_d, are more relevant in the very short run than are the lower elasticities that reflect the offsetting effects of both i and i_d.⁶

Another overstatement results from a misspecification of the stock adjustment process. equation (2) says that the supply of money adjusts to its demand with a lag. However, if rapid adjustments in the monetary sector are costly, then it should be the demand for (nominal) money that adjusts gradually to changes in its supply rather than the other way around.⁷ equation (2) says that a change in demand affects supply by only λ of the change in demand. Something close to equation (2) results from the assumption that the demand for money is a function of permanent rather than current measured income or interest rates. Given that interpretation, an exogenous change in M (or m, given P) will require a larger change in i the smaller is λ, because it takes a larger change in actual i to generate a given change in permanent i. However, this interpretation has nothing to do with adjustment costs.

A gradual adjustment of demand to an exogenous change in supply would dampen rather than multiply the necessary change in i as demonstrated by Starleaf. Treating money balances as a buffer stock and recognizing that adjusting that stock in the face of exogenous disturbances (shocks) requires a change in spending (on goods or bonds, etc.) means that if adjustments are costly it is optimal to adjust (spending on bonds, etc.) gradually. But gradual adjustment means that economic units are willingly (though only temporarily) holding a quantity of money that is not the long-run equilibrium quantity. This willingly held “disequilibrium” money is the extent to which a change in M exerts no market pressure on prices or interest rates. This is seen by replacing equation (2) by Starleaf’s specification:

and solving the monetary sector (the LM relation) for the interest rate. This gives

In the generally used stock adjustment of equation (3), a one-unit increase in M immediately decreases i by 1/a₂λ, while in equation (10) the same increase in M decreases i by λ/a₂. An interest elasticity of a₂ = 0.5 and a partial adjustment coefficient of λ = 0.2 imply money impact multipliers of 10 for equation (3) and 0.4 for equation (10).

Combined with the IS sector, Starleaf s version gives

where z = a₂ (1–α) + λαb₂

and $\frac{\partial i}{\partial \ln M}=\frac{\lambda (1-\alpha )}{a_2(1-\alpha )+\lambda \alpha b_2}\mathrm{.}$

Equations (8) and (11) also allow an examination of the impact multipliers of i from “exogenous” changes in real output (y), inflationary expectations (Π), money demand (ε_m), and aggregate demand (ε_I).

Table 1 dramatizes the considerable differences these considerations make in assessing interest rate volatility. Taking the extreme cases: (a) ignoring the real sector and focusing only on the demand for money with the traditional stock adjustment lags, an interest coefficient for money of 0.2 and a speed of adjustment coefficient of 0.2 gives an impact interest effect of a change in the money stock of –25; (b) making the real sector (i.e., the price level) partially endogenous and using the Starleaf stock adjustment formulation, the higher interest coefficient of demand for money of 0.5, the same speed of adjustment in the money market (λ = 0.2), an adjustment speed in the real sector of 0.5, and a unitary interest coefficient for investment demand gives an impact interest effect of –0.29. However, overwhelmingly the greater part of the difference results from the shift to the Starleaf stock adjustment specification.

Table 1.

Impact Multipliers

Table 1.

Impact Multipliers

	λ = 0.2	λ = 0.5	λ = 0.2	λ = 0.2	λ = 0.2	λ = 0.2
	α = 0.2	α = 0.2	α = 0.5	α = 0.2	α = 0.2	α = 0.2
	a2 = 0.5	a2 = 0.5	a2 = 0.5	a2 = 0.2	a2 = 0.5	a2 = 0.2
	b2 = 1	b2 = 1	b2 = 1	b2 = 1	b2 = 0.5	b2 = 0.5
Equation (3)
$\frac{\partial i}{\partial \ln M}=\frac{1}{a_2\lambda }$	–10	–4	–10	–25	–10	–25
Equation (10)
$\frac{\partial i}{\partial \ln M}=\frac{\lambda }{a_2}$	–0.4	–1.0	–0.4	–1.0	–0.4	–1.0
Equation (8)
$\frac{\partial i}{\partial \ln M}=$	–2.857	–2.000	–0.909	–3.448	–4.444	–6.060
$\frac{\partial i}{\partial \mathrm{\Pi }}=$	0.714	0.500	0.909	0.862	0.555	0.758
Equation (11)
$\frac{\partial i}{\partial \ln M}=$	–0.364	–0.800	–0.286	–0.800	–0.381	–0.888
$\frac{\partial i}{\partial \mathrm{\Pi }}=$	0.090	0.200	0.286	0.200	0.048	0.111

View Table

Table 1.

Impact Multipliers

	λ = 0.2	λ = 0.5	λ = 0.2	λ = 0.2	λ = 0.2	λ = 0.2
	α = 0.2	α = 0.2	α = 0.5	α = 0.2	α = 0.2	α = 0.2
	a2 = 0.5	a2 = 0.5	a2 = 0.5	a2 = 0.2	a2 = 0.5	a2 = 0.2
	b2 = 1	b2 = 1	b2 = 1	b2 = 1	b2 = 0.5	b2 = 0.5
Equation (3)
$\frac{\partial i}{\partial \ln M}=\frac{1}{a_2\lambda }$	–10	–4	–10	–25	–10	–25
Equation (10)
$\frac{\partial i}{\partial \ln M}=\frac{\lambda }{a_2}$	–0.4	–1.0	–0.4	–1.0	–0.4	–1.0
Equation (8)
$\frac{\partial i}{\partial \ln M}=$	–2.857	–2.000	–0.909	–3.448	–4.444	–6.060
$\frac{\partial i}{\partial \mathrm{\Pi }}=$	0.714	0.500	0.909	0.862	0.555	0.758
Equation (11)
$\frac{\partial i}{\partial \ln M}=$	–0.364	–0.800	–0.286	–0.800	–0.381	–0.888
$\frac{\partial i}{\partial \mathrm{\Pi }}=$	0.090	0.200	0.286	0.200	0.048	0.111

It is possible that estimating an equation like (11) rather than (8) will yield different values for the underlying structural coefficient. In fact, Coats’s (1982) versions of equations (3) and (10) using quarterly U.S. monetary data (M1-B), gave values of a₂ = 0.010 and λ = 0.355 for equation (3)⁸ and of a₂ = 0.005 and λ = 0.218 for equation (10). These values imply money impact multipliers for the interest rate of 281.7 and 43.6 for equations (3) and (10), respectively.

Another overstatement results from ignoring the interest elasticity of the money multiplier. Central banks do not control the money supply directly. A more practical short-run strategy might be the adherence to a bank reserve or monetary base target. The deposit or monetary response to a change in reserves (B) is dampened by the shock absorber roles of banks’ excess reserves and nondeposit (managed) liabilities. This dampened monetary response in turn dampens the interest rate response to changes in reserves. For example, adding an interest-sensitive money supply:

1nM = c_o + lnB + c₂i

to equations (1) and (2) gives

and

when c₂ > 0.

Federal Funds Rate, January 4, 1978–October 14, 1981

(Seven-day average for week ended Wednesday)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Federal Funds Rate, January 4, 1978–October 14, 1981

(Seven-day average for week ended Wednesday)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

Federal Funds Rate, January 4, 1978–October 14, 1981

(Seven-day average for week ended Wednesday)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

Moreover, the change to a monetary or reserve strategy will itself very likely lead to adaptive changes in the banking sector’s behavior. One of the most likely structural changes in response to the adoption of a reserve strategy is the increased use by banks of excess reserves and managed liabilities as reserve shock absorbers. In other words, the adoption of a reserve strategy itself is likely to eventually raise the interest elasticity of the money multiplier, thus further dampening the interest rate response to a change in reserves.

In general, these and other structural adjustments would probably overshadow other factors tending to smooth interest rate behavior, but they are the most difficult to quantify or even foresee. Volatile interest rates contain profitable opportunities to innovate in ways that smooth rates. Given time, the banking system and financial markets and instruments can be expected to respond to these opportunities. However, as shown above, even the existing structure, when properly and fully elaborated, gives hope that increased interest rate volatility from the adoption of a reserve strategy may not be as great as existing estimates seem to suggest.

Monetary Growth Target, M1–B

(Percentage annual growth rate of monthly average, seasonally adjusted)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Monetary Growth Target, M1–B

(Percentage annual growth rate of monthly average, seasonally adjusted)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

Monetary Growth Target, M1–B

(Percentage annual growth rate of monthly average, seasonally adjusted)

Citation: IMF Staff Papers 1982, 001; 10.5089/9781451956634.024.A002

• Download Figure
• Download figure as PowerPoint slide

The large increase in interest rate volatility in the United States, following the Federal Reserve System’s adoption of a weekly reserve targeting strategy in October 1979, seems to undercut this somewhat cheerful assessment (see Chart 1). However, the more recent period has been a turbulent one for many reasons that would have made interest rates more volatile even without the Federal Reserve’s new reserve targeting. Even the money supply itself was more volatile despite intensified efforts to smooth it out (see Chart 2). Monetary shocks to interest rates should fall primarily on very short-term rates, while more permanent and real shocks would affect short-term and long-term rates equally. In fact, both short-term and long-term rates have become considerably more volatile. This question of “whether the observed increase in interest-rate volatility stemmed from the change in monetary policy on October 6, 1979” is addressed by Evans (1981, p. 5).

[He] finds that this policy change produced only about 30 per cent of the increased volatility in long-term interest rates, and that the rest came from sources not directly under Federal Reserve control.

V. Conclusion

The interest rate consequences of targeting money depend on the economic environment in which the policy is undertaken. The simultaneous pursuit of interest rate and monetary targets (by means of monetary policy alone) is not generally possible except in a limited way even in the most isolated, financially undeveloped economies. For this reason the major interest rate issue associated with targeting money for those countries with developed financial markets is its consequence for the volatility of rates. In examining the underlying relationships that link monetary and interest rate behavior in an effort to assess the magnitude of the impact of monetary targeting on interest rate volatility, several shortcomings of the existing literature are presented. When these shortcomings are taken into account (or are corrected) the earlier assessments that interest rates would be “much” more volatile are “greatly” attenuated.