the impact of money on short-term interest rates

THE IMPACT OF MONEY ON SHORT-TERM INTEREST RATES

WILLIAM REICHENSTEIN'

This study reviews empirical evidence from four research methods related to the impact of money on short-term nominal rates. The studies consistently fail to find evidence supporting the much hypothesized short-term, negative relationship between money and nominal rates since at least April 1975. Rea- sons for the absence of a negative relationship include the tendency of financial markets to anticipate corrective action by the Fed whenever M1 deviates from targeted growth ranges and a rapid adjustment of inflationary expectations to changes in money growth.

There is a general perception that the Federal Reserve can significantly lower short-term interest rates, at least for a while, by increasing the growth rate of money. In a review of studies on this issue from four research methods, this paper finds no consistent empirical evidence (since at least April 1975) to support this negative relationship between money and interest rates. The four research methods include two families of interest rate models and two rational expectation-efficient market models.

I. ECONOMIC THEORY

The theoretical effect of a monetary acceleration can be examined in the Hicksian IS-LM framework. Assume the Fed permanently accelerates money growth and that prices, output, and inflation are not immediately affected. The acceleration shifts the LM curve, increases real balances, and reduces nominal and real interest rates. This is the liquidity effect.

The combination of lower real rates and higher real balances stimulates spending, which leads to an increase in nominal income. This is the income effect; it increases the demand for money and increases nominal rates.

Monetary acceleration also has an impact on inflationary expectations. In- creased inflationary expectations raise the IS curve, causing an increase in nominal rates.

Key to the popular argument that the Fed can lower nominal rates at least for a while through a monetary acceleration are three assumptions: that prices do not adjust immediately to the new money supply, and that inflationary expectations and money demand are not affected quickly by changes in money. It is clear from the literature that these assumptions have been widely

*Associate Professor, North Texas State University. The author wishes to thank J. Walter Elliott, Gary J. Santoni, Richard J. Sweeney, and two anonymous referees for valuable comments on earlier drafts.

67 Economic Inquiry Vol. XXV, January 1987, 67-82

68 ECONOMIC INQUIRY

held. For example, in 1968 Milton Friedman [1968] wrote that the initial impact of monetary expansion is to lower interest rates below the level they would have been for a period of perhaps “a year or two.” Until recently, the positive impact of money on rates was assumed to occur only after considerable lag. Friedman [1964] found money growth to be most highly correlated with economic activity at least a year later. It was assumed that inflationary expectations adjusted to actual inflation over at least a two-year period, and Yohe and Karnosky [1969] found inflation itself adjusted slowly to monetary growth. In such an environment, it is easy to see how the Fed may have considerable short-run influence on interest rates.

Two developments in the financial markets, however, inhibit the current ability of the Fed to lower rates for more than a very short time. The first is related to the establishment of target growth ranges for the monetary aggregates, which causes a policy anticipation effect. In April 1975 the Federal Reserve System began announcing twelve-month target growth ranges for the monetary aggregates. At the announcement of a larger-than-anticipated increase in the money supply, investors bid down the price of fixed-income securities in anticipation of a subsequent monetary deceleration (and vice versa). A larger-than-anticipated money supply announcement will increase the demand for money. This demand for money is similar to Keynes’ spec- ulative demand; the anticipation of higher future rates leads investors to sell bonds and move into money, which drives up interest rates. The policy anticipation effect, however, makes the net impact of an increase in the money supply very hard to determine after a period of only two weeks, even if one assumes price, output, and inflationary expectations are as yet unaffected.

The second reason to question the Fed’s short-run influence on interest rates is empirical evidence from Blejer [1978], Cornell [1983b], and Mehra [1985] indicating that in recent years the lag from money to inflationary expectations has considerably shortened.2 Friedman [ 19681 argues that at high inflation rates a monetary acceleration may quickly and strongly increase inflationary expectations, thereby off setting any short-run liquidity effect. Higher inflation rates in the 1970s may account for the elimination of the short-run negative relationship between money and interest rates experienced in earlier decades3

The empirical evidence reviewed in this paper supports the argument that the policy anticipation and short-run inflationary expectations effects offset the liquidity effect. Since April 1975 (and possibly before), a monetary acceleration has had an essentially neutral net impact on short rates in the month of the acceleration.

1. The market response is not dependent upon a quantitatively measureable short-run, negative relationship between money and rates. It is dependent upon the expectation that such an effect exists. The strength of the response should adjust over time with expectations.

2. K. M. Carlson [1980] and Friedman [1984] present evidence that the lags between money and prices and money and income have shortened.

3. Friedman [1980] recently argued that money affects inflationary expectations within a month or two.

REICHENSTEIN: MONEY AND INTEREST RATES 69

II. EMPIRICAL EVIDENCE

The ZS-LM Method

Studies by Makin [1983], Mehra [1985], and Wilcox [1983] assess the short- term impact of a monetary acceleration on nominal rates by specifying a complete interest rate model, where the impact is measured by the coefficient on the money term. These studies measure unanticipated monetary changes, under the hypothesis that anticipated changes should not matter in efficient markets4 The interest rate models are the reduced form specification of IS, LM, and aggregate supply equations. While equations and right hand side variables vary among studies, all share the common feature of specifying the nominal rate as a function of unanticipated money growth, inflationary expectations, and other variables.

The monetary coefficient in this methodology estimates the liquidity effect for the period before April 1975, the date the Fed began announcing year- ahead growth targets, and the combined liquidity and policy anticipation effects thereafter. Assessment of the full effect requires measurement of the monetary policy influence on inflationary expectations and other right hand side variables. Makin estimates anticipated monetary increases by an ARMA (0,8) model, while Wilcox and Mehra use annualized six-month money growth less annualized growth rate over the last three years as the measure of unanticipated money g r o ~ t h . ~

Two of the three-Makin and Mehra-find evidence of a significant negative liquidity impact for the pre-1970s period. Makin estimates that a 1 percent unanticipated monetary acceleration would have reduced short rates about three basis points within a quarter. Wilcox does not find a significant negative impact in the pre-1970s period, which Mehra attributes to his estimation techniques.

Mehra and Wilcox do not find a significant negative monetary coefficient when the estimation period includes much of the 1970 decade. The monetary coefficient varies among regressions from positive to negative, but is always within one standard deviation of zero. This is consistent with the hypothesis of offsetting liquidity and policy anticipation effects since April 1975. Makin, using an alternative monetary measure, continues to find a significant but weak negative monetary coefficient for the 1970-1981 period. This need not imply a negative short-term relationship between money and interest rates for the entire period. The influence of the policy anticipation effect on the monetary coefficient was not consistent throughout the period. Furthermore, the short-term impact of money growth on inflationary expectations must be considered as well.

4. The use of unanticipated monetary changes is a standard feature in most recent econometric work. However, Sweeney [1984] has shown that anticipated monetary policy can affect real variables even in models with perfect foresight.

5. J. Carlson [1979] uses the same specification with success.

70 ECONOMIC INQUIRY

The Gibson Method The second interest rate model regresses interest rates on current and past

observations of monetary growth rates to examine both direct and indirect influences of money. This type of study may have been used first by Gibson [1970], Cagan [1966], and Cagan and Gandolfi [1969], and has received sub- stantial renewed interest through the current work of Brown and Santoni [1983], Melvin [1983], and Hoehn [1983]. The methodology does not provide a complete model of the determinants of interest rates. A complete model might postulate interest rates as a function of liquidity, income, inflationary expectations, and other factors. The first three factors can be represented jointly by current and lagged observations of money growth, while the other factors are assumed to be uncorrelated with the money terms.

The general form of the equation is n

TB, = a, + C MI,-, + e, i=O

where

TB = the end-of-month three-month Treasury bill rate (measured by the closing market yield on the last Friday of the month),

M1 = the growth rate of end-of-month, seasonally adjusted M1 (measured by the weekly average M1 for the Wednesday before the bill rate), and

n = the length of the monetary lag.

The estimates of the bt’s will be unbiased to the extent that the monetary variables are uncorrelated with the missing factors. Missing factors introduces autocorrelation of unknown degree and suggests the estimation of equation (1) using an autoregressive technique. Failure to remove the autocorrelation would bias the standard errors and invalidate significance tests.

Estimation of equation (1) using the Cochrane-Orcutt technique revealed an autocorrelation coefficient insignificantly different from one at the 5 percent level. For the regressions in this section the first order autocorrelation coefficient is set at one, which is equivalent to estimating equation (1) in first difference form. The lag length is allowed to vary from four to twenty-four months in units of four.6

The empirical evidence reveals an unstable money-interest rate relationship between the 1965.2 (February 1965) to 1975.3 and 1975.4 to 1983.3 periods7 In fact, for the 1965.2 to 1975.3 periods, the monetary aggregates when

6. Initial lag length was set at four months, with additional months added four at a time until an F-test failed to justify the extension. The twelve-month lag length is also justified when the lag is set initially at twenty-four months, with four months removed at a time.

7. We found no evidence of instability in the M1-Treasury bill relationship between the 1975.4 to 1979.9 and 1979.10 to 1983.3 periods. The F,4,88 = .67 which is well below usual significance levels.


TABLE I Institutional Changes Potentially Affecting Money-Interest Rate

Relationship

August 1971: April 1975:

Closing of the gold window Fed began announcing twelve-month target growth

ranges for monetary aggregates Federal Reserve’s Daily Instrument of Monetary Control

Federal funds rate

Eclectic approach with emphasis on nonborrowed re-

Before October 1979: October 1979-October 1982: Nonborrowed reserves After October 1982:

serves and federal funds rate

considered as a group fail to reveal a significant relationship with the dependent variable. This confirms a similar finding by Brown and Santoni for the 1954.1 to 1970.12 period.

An unstable money-interest rate relationship is not surprising. Changing institutional arrangements and Fed practices over time may dramatically alter the interest rate response. Brown and Santoni, for example, report little faith in Gibson-type results for the period before the August 1971 closing of the gold window. They feel that the Gold Reserve Act loosely tied the dollar to gold before that date. Melvin also argues that the money interest rate relationship changed during the period. Before about January 1973, the Fed commonly believed that the appropriate response to an interest rate increase was to accelerate money growth; since then, the Fed has been more likely to react to higher rates by restraining money growth in the belief that high rates are attributable to excess money growth and inflation in the past. This paper suggests that the announcement since 1975 of year-ahead desired growth ranges for the monetary aggregates is what changed the money-interest rate relationship. Regardless of the exact date, all studies conclude that the different institutional arrangement and Fed practices that have operated since the early to mid-1970s have affected the money-interest rate relationship. Table I presents dates of potential importance to the money-interest rate relationship.

For the 1975.4 to 1983.3 period, the M 1 variables are significantly related to the Treasury bill rate (Table 11). The estimates in Table I1 imply that a 1 percent monetary acceleration is associated with (a) an insignificant 1.9 basis point increase in the T-bill rate in the current month, (b) a significant rise in the bill rate after the first lagged month, and (c) an 80 basis points increase after one year. The estimates are similar to results of the 1970.1 to 1983.2 regression from the Brown and Santoni study. Using the commercial paper rate and monthly-averaged M they estimate (a) a significant 2.64 basis point decrease in the current month, (b) a net increase of 4.94 points after the first lagged month, and (c) a net increase after one year of approximately 1 percent.

72 ECONOMIC INQUIRY

Table I1 Interest Rates and M1 Growth-1975.4 to 1983.3

Equation: TB, = a + bfM1,-, + e, 12

f=O

Indep. Indep. Variable Coeff. t Variable Coeff. t

M1t .019 1.40 MIt-7 .068 2.43 M1t-I ,068 3.85 Mlt-8 ,049 1.82 Mlt-2 ,090 4.20 MA-* ,035 1.46 Mlt-3 ,089 3.76 M1t-10 ,026 1.14 MIt-4 .088 3.35 Mlt-11 ,043 2.31 MI,-5 .091 3.29 Mlt-12 .039 2.81

, Mlt-6 ,092 3.24 R2 = .33*; Zb, = 0.797 P I = 1; DW = 1.84** * To remove residual correlation equation (1) was regressed using the Cochrane-Orcutt

technique. The estimate of the first-order autocorrelation coefficient, pI, was approximately one in all regressions examined. Equation (1) was then reestimated in first difference form, which sets the constant equal to zero. Of course, the coefficient of determination is substantially reduced when the regression is run in first difference form.

** The Durbin-Watson test is not directly applicable for a regression without a constant.

The Melvin study regresses first differences in the commercial paper rate on first differences in monthly-average old M2 for the 1973.1 to 1979.12 period. His estimates of bo vary with the assumed lag length from an insignificant -2.59 to a significant -5.10 basis point change. Finally, Hoehn regresses the change in the log of the federal funds rate on the change in the log of M1 for the three-year periods before and after the October 1979 change in Federal Reserve operating procedure. In the six regressions examined, he fails to find evidence of a negative relationship between money and rates; the current month coefficient is consistently positive, but never significant.

The significant negative current month coe€ficient from the Brown and Santoni and Melvin studies should be judged with caution. The test periods include years before the announcement of year-ahead growth targets. The negative coefficients are consistent with a significant liquidity effect before April 1975 and roughly offsetting liquidity and policy anticipation effects after that date. The studies also use monthly averaged data, which Working [ 19601 shows can produce misleading results. Substituting end-of-month Treasury bill and M1 data while otherwise repeating these regressions pro- duces insignificant current month coefficient estimates.

The Mishkin Method The third research methodology measures the relationship between unan-

ticipated changes in money (and other demand for money factors) and unanticipated changes in interest rates using autoregressive and multivariate models. The leading examples of this work are those of Mishkin [1981; 19821, whose efficient market models build upon the work of Barro [1977; 19781 and


Muth [1961]. Mishkin finds no evidence for the 1959-76 period to support the hypothesis that unanticipated increases in the quarterly money stock are associated with unanticipated decreases in short or long rates. In fact, the evidence indicates that an unanticipated increase in the quarterly money stock is significantly associated with unanticipated increases in short rates within the quarter.

This paper extends Mishkin’s work by updating the time period examined and estimating the models with monthly instead of quarterly data.

We begin with the rational expectation-efficient market relationship:

DRS, = a + p m ( M , - M:) + c, (2)

where DRS is the unexpected change in short rates, M the growth rate of money, the superscript denotes expected values conditional on all past information, and a and Brn are regression parameters.

A second efficient market model regresses changes in short rates on unexpected changes in money and unexpected changes in the other components of the demand for money:

DRS, = a + @ m ( M , - M f ) + py(Y, - Y:) + p p ( P , - P f ) + tt (3)

where Y is the growth rate of real income, P is the rate of inflation, and a, pm, By, and p p are parameters.

The sign of pm is of prime importance. If unexpected money growth lowers rates, as traditional Keynesian models hypothesize, prn should be negative. The income and price level coefficients should be positive.

The empirical counterparts to equations (2) and (3) are

DTB, = a + b,,,(M1, - M I ; ) + e, (4) and

DTB, = a + b,(M1, - M I ; ) + by(ZP, - ZP:) + b,(CPZ, - CPZ;) + e,. (5)

The symbols include the following:

DTB = the change in the end-of-month three-month Treasury bill rate,8 M 1 = the first difference in logs of end-of-month, seasonally adjusted

MI, ZP = the first difference in logs of the industrial production index,

CPZ = the first difference in logs of the consumer price index.

Two models are used to estimate the expected growth of M 1 , ZP, and CPZ. The first model is a univariable autoregressive time series model of the form

4

x, = a, + six,_, + u, i = l

8. Mishkin uses a measure of the unanticipated change in rates, but the correlation between his dependent variable and the actual change in short rates is very high. Similar estimates prevail for the change in short rates. This study uses the change in rates to avoid unnecessary complication and to focus on the variable of interest.

74 ECONOMIC INQUIRY

where X is M1, ZP, or CPZ. The fitted values are used as measures of expected growth. The second model is a multivariate time series model based upon the multivariate model used by Mishkin [1982]. In an efhcient market the expectations variables will reflect any useful information contained in the past growth of money, real income, and prices, as well as any information contained in other publicly available data. Each variable-MI, ZP, and CPZ-is regressed on the four lagged values of each variable. In addition, the four lagged values of the following values were included to capture their possible information content:

M2 = the first difference in logs of end-of-month, seasonally-adjusted

UN = the unemployment rate, TB = the end-of-month three-month Treasury bill rate,

M2,

HEBS = the high employment budget surplus (spliced to monthly data), FMD = the first difference in logs of federal marketable debt,

GP = the first difference in logs of real government purchases (spliced

The four lagged values of each variable were retained in an equation only if they were jointly significant at the 5 percent level.

Univariate and multivariate time series models were developed using monthly data for the 1965.1 to 1983.3 period. The stability of the relationships was examined with Chow tests for the 1965.1 to 1975.3 and 1975.4 to 1983.3 periods. Stability could be rejected at the 5 percent level for CPZ in both the univariate and multivariate models. The CPZ models are separately estimated for each subperi~d.~

The multivariate models are summarized briefly in Table 111. It is interesting to note the dominance of demand for money variables. The only non demand-for-money variable is real government purchases in the CPZ regression for the 1975.4-1983.3 subperiod. In general, the fiscal policy variables did not fare well in the regressions.

The expectations variables are generated from the multivariate and univariate models. lo The models assume that individuals have prior knowledge of the structural relationships. Equations (4) and (5) are estimated for four time periods: the 1965.1 to 1983.3 period, the 1965.1 to 1975.3 and 1975.4 to 1983.3 subperiods, and the 1979.1 to 1983.3 period corresponding to the pe-

to monthly data).

9. Chow tests support the hypothesis of stability in the CPI univariate and multivariate relationships within each subperiod. The 1965.1 to 1975.3 period was split at its midpoint for the test and the 1975.4 to 1983.3 period was split at October 1979 to correspond with the change in Fed operating procedure.

10. Durbin [1970] and Sims [1977] indicate that using residuals from one regression as variables in another regression can lead to misleading results. Mishkin [1981] estimates his regressions using the approach followed in this paper and an approach designed to circumvent the potential problem. Results from both approaches were essentially the same. Thus, the two-step procedure is used in this study.


TABLE 111 Estimates of Multivariate and Univariate Time Series Models

Multivariate Models R2 DW

1965.1-83.3 M 1 = 41(M1, TB, I P ) IP = $,(M2, TB, I P )

CPI = &(M2, CPI, TB) 1965.1-75.3

1975.4-83.3 CPI = +&vf1, CPI, GP)

Univariate Models

.25 1.98

.48 2.00

.57 2.04

.66 2.17

(A) 1965.1-83.3 M 1 = yl(M1) I P = y,(IP)

CPI = y,(CPI)

CPI = rJCPI)

(B) 1965.1-75.3

(C) 1975.4-83.3

.06 1.98

.41 2.01

.47 2.01

.52 2.01

riod after a change in Federal Reserve operating procedure. The results are summarized in Table IV.

The coefficients on unexpected M1 are small, positive, and generally insignificant while the coefficients on unanticipated income and inflation are positive and often significant. The results fail to support the hypothesis that the net one-month impact of money on short rates is negative. The low coefficients of determination emphasize the weak short-term money-interest rate relationship. The results imply that money’s net impact on interest rates is essentially neutral during the month of the monetary change.” The monetary coefficients are higher for the period after April 1975. The pattern is consistent with the hypothesis of stronger policy anticipation and short-term inflationary expectation effects after April 1975.

An examination of the correlations among variables from the Mishkin-type regressions provides insights into the money supply-interest rate relationship (Table V). The correlations suggest that an unexpected increase in money is positively correlated with the change in the bill rate in the current and the next two months (DTB,,, and DTB,+2). The strength of the correlation increases in the post-April 1975 period, which is consistent with the policy anticipation hypothesis and the faster adjustment of inflationary expectations to monetary surprises.

The dating of the variables may account for the significant correlation

11. The same conclusion prevails when unanticipated changes in M 2 are substituted for unanticipated MI.

76 ECONOMIC INQUIRY

TABLE IV Estimates of the Efficient Markets Models. Dependent Variable = DTB

of ( M I - M I " ) (ZP-ZP) (CPZ-CPZ8) R2 Coefficient

Multivariate Models EQ. (4) 17.7 (1.47) EQ. (5) 14.4 (1.22)

EQ. (4) 23.6 (2.22) EQ. (5) 23.8 (2.28)

Univariate Models


EQ. (4) 28.4 (1.67) EQ. (5) 26.3 (1.64)

Univariate Models


EQ. (4) 41.2 (1.38) EQ. (5) 29.5 (1.01)

Univariate Models


EQ. (4) 17.3 (1.30) EO. 15) 18.3 11.36)

Univariate Models

(A) 1965.1-83.3

17.7 (2.26)

17.6 (2.43)

(B) 1975.4-83.3

38.0 (2.57)

37.3 (2.90)

(C) 1979.10-83.3

60.5 (1.99)

52.2 (1.99)

(D) 1965.10-75.3

2.1 (0.28)

-0.4 (0.05)

74.3 (2.62)

61.5 (2.52)

49.0 (0.95)

89.5 (2.15)

56.0 (0.59)

101.4 (1.38)

91.5 (3.15)

27.5 (1.03)

,010 ,063

.022 ,076

,018 .094

,029 ,156

,029 .132

,045 .195

.002 ,079

,014 ,022

t-statistics are in parentheses. 1.0 represents a one basis point change in the Treasury bill rate.

between a money surprise and next month's interest rate change. The Trea- sury bill rate is the closing rate for the last Friday of the month. The money supply represents the average for the week ending two days earlier. The end- of-month money supply figure is not announced until after the close of the month, so its impact should be partly reflected in next month's interest rate change, DTB,,

An examination of the variables correlated with the anticipated change in M1 suggests an interesting story. The M1" exhibits a strong negative correlation with the change in the bill rate from the previous two months. In addition, it is negatively associated with the unanticipated monetary change from

TAB

LE V

A

naly

sis

of C

orre

latio

ns

8 DT

BI+3

D

TB,,,

D

TB

,,,

DTB

, D

TB

,-,

DT

B,-

, D

TB,-3

(M

l-M

le)t

-i

3 2 el (M

I-M

I"),

.oo

.12b

.27"

.1

2b

.oo

-.04

.oo

.oo

r; 5 3 M

1;

.06

.12

- .0

2 .0

8 -.3

2"

-.37"

- 3

4"

- .3

4'

(M 1 -M

1 "),

-.1@

.1

0 .1

2 .09

. 1 8b

-.1

2 .0

3 .0

7 .e

(C) 1

975.

7-82

.12

z -.1

7 U

M

1;

.11

.08

.27'

.1

4 -.5

1"

- .5

0"

-.23"

m m M

lp

.24

.16

.33"

.1

9 -.5

5"

-.56"

-.2

3 -.

11

v)

(M 1 -M

1 ')*

-.01

.19

.40"

.2

4 -.1

2 -.

11

-.05

.01

3 2

(A)

1965

.4-8

2.12

M

1;

.10

.09

.19"

. 1

3b

-.43'

-.45"

-.1

8"

- .2

4"

(B)

1965

.4-7

5.1

* 3 (M

1 -M

1 e)t

.0

9 .1

5 .3

P

.16

-.09

- .0

4 - .

03

- .0

4 15

el

9

(D) 1

980.

1-82

.12

M1:

is ob

tain

ed fr

om th

e m

ultiv

aria

te m

odel

for

the

1965

.1 to

198

3.3

peri

od. T

hree

obs

erva

tions

are

lost

on

the

fron

t end

and

tail

end

of ea

ch p

erio

d du

e to

the

a In

dica

tes s

igni

fican

ce at

the 5

perc

ent

leve

l or

bette

r. le

ads

and

lags

of

DTB

. v)

Indi

cate

s sig

nific

ance

at th

e 10

per

cent

leve

l.

78 ECONOMIC INQUIRY

the previous month. These correlations are consistent with the assumption that the Fed feels more pressure to limit monetary increases and fight inflationary pressure following periods of rising rates. Furthermore, the Fed may be expected to partially offset unanticipated changes in money in the following month.

Failure to uncover the anticipated short-run negative relationship between money and interest rates with monthly data might suggest that we move to a shorter time frame, perhaps weekly data. Cornell [1983a] examined the correlations between weekly changes in the money supply and the monetary base and weekly changes in several interest rates for the January 1978 to August 1981 period and two subperiods. He concluded that “the data do not reveal the existence of an actual liquidity effect . . . . The only noticeable pattern in the reported correlations is that changes in M1B are positively correlated with changes in interest rates during the announcement week” [1983a, 201.

As for the possibility of a weak liquidity effect, he reports consistent negative correlations between changes in the monetary base and changes in interest rates three weeks earlier. Similarly, Urich and Wachtel [ 19811 present evidence that a $1 billion increase in money will reduce short rates 1.5 basis points about three weeks before the money supply announcement.

The Announcement Efect Short-term interest rates respond positively to the announcement of an

unexpected increase in the money supply.12 This response is consistent with a policy anticipation effect, an inflationary expectations effect, or both. Evi- dence to be presented below suggests that both effects are present. Their strengths vary with the operating strategy of the Federal Reserve, but Har- douvelis [ 19841, Judd [ 19841, and Sheehan [ 19851, find the policy anticipation effect to usually be the stronger of the two short-run effects.I3

Before October 1979 (regime I), the Federal Reserve used the federal funds rate as its daily instrument of monetary control. The Fed was not expected to move aggressively in this operating strategy to off set an unexpected change in money. Financial markets expected a relatively weak policy anticipation effect and a relatively strong inflationary expectations effect. With inflationary expectations being the weaker short-run effect, absolute change in interest rates are expected to be small. The Fed used nonborrowed reserves as its daily

12. See the following articles: Berkman [19781 Conrad [1979], Cornell [1982; 1983133, Grossman [1981], Hardouvelis [1984], Hein [1985], Judd [1984], Roley [1982; 19831, Roley and Troll [1983], Roley and Walsh [19841 Shilier, Campbell, and Schoenholtz [1983], Urich [1982; 19841, and Urich and Wachtel[1981].

13. Other hypotheses have been offered to explain markets’ reactions to money supply announcements. Mascaro and Meltzer [1983] and Tatum [1984] hypothesize that increased monetary or interest rate uncertainty leads to a higher risk premium, but Cornell [1983b] does not find the announcement data to be consistent with the theories. See Nichols, Small, and Webster [1983] and Sheehan [1985] for alternative hypotheses with little empirical support.

REICHENSTEIN: MONEY AND INTEREST RATES

instrument between October 1979 and October 1982. Markets expected the Fed to generally offset money shocks in this regime, in which case a strong policy anticipation effect should produce strong interest rate movement^.'^ Wallich [1984] describes the post-October 1982 strategy (regime 3) as a com- promise between regimes 1 and 2. The interest rate response should lie between that of the first two regimes.

Studies of the impact of money announcements on exchange rates, commodity prices, and other markets allow us to estimate the relative strengths of the policy anticipation and short-run inflationary expectations effects, since the effects predict different responses in these markets. Hakkio and Pearce [ 19851, Cornell [1982], and Hardouvelis [ 19841 generally find money surprises in regime 1 produce insignificant exchange rate responses, which suggests equal and offsetting short-run effects. Other regime 1 responses are generally consistent with this interpretation.

The announcement literature shows that the policy anticipation effect dom- inates in regimes 2 and 3. A positive money surprise is associated with the following responses predicted by the policy anticipation hypothesis: an appreciation of the dollar, a rise in long-term rates, no effect on distant forward rates, and a decrease in commodity prices.15

Even though the policy anticipation effect has been the stronger effect since October 1979, separate studies by Hardouvelis 119841, Judd [1984], and Shee- han [ 19851 conclude that the announcement literature best supports the existence of both effects. In related work, Mehra [1985] shows the Livingston survey measure of inflationary expectation to be very strongly influenced by money growth since the 1970s.

The research design of the announcement literature is as close to the ideal of laboratory control as possible in the social sciences. Factors besides the announcement are held “constant” by limiting the interest rate observation to a short period surrounding the money supply announcement (often as short as ninety minutes). The new information is introduced into the financial environment and the reaction is recorded.

The major contribution of the announcement effect literature is evidence that some of the positive impact of an increase in the money supply occurs within a few weeks. This evidence does not deny the existence of a liquidity effect. It does, however, indicate that the very short-term net impact of money on interest rates is uncertain. The short-term net impact depends upon the strength of the liquidity effect and the strengths of the policy anticipation and short-term inflationary expectation effects.

79

14. Hardouvelis [1984] and Frankel and Hardouvelis [1985] present evidence that financial markets’ confidence in the Fed’s ability and willingness to hit money supply targets increased in the nonborrowed reserve regime.

15. Cornell [1982; 1983b1, Engel and Frankel [1984], Hakkio and Pearce [1985], and Hardou- velis [1984] find an appreciation of the dollar. Cornell [1983a; 1983bI and Roley and Walsh [1984] examine long-term rates, while Hardouvelis [1984] and Hein [1985] examine forward rates. Fran- kel and Hardouvelis [1985] examine commodity prices.

80 ECONOMIC INQUIRY

111. CONCLUSIONS

This study reviews empirical evidence from four research methods related to the impact of money on short-term nominal rates. The studies consistently fail to find evidence supporting the much hypothesized short-term, negative relationship between money and nominal rates since at least April 1975. The consistency of results across research methods makes us confident that the conclusion is not a byproduct of a particular research approach.

Since April 1975, and possibly before, the negative influence of a monetary acceleration on short rates has been, at best, weak and short-lived. Extensive empirical evidence indicates that a permanent acceleration of money growth will have an essentially neutral impact on short rates for the first month and even raise rates significantly within the quarter.

Reasons for the absence of a negative short-term relationship between money and interest rates include the tendency of financial markets to anticipate corrective action by the Fed whenever M1 deviates from targeted growth ranges and a rapid adjustment of inflationary expectations to changes in money growth. Under present institutional arrangements and Fed operating procedures, any attempt by the Fed to lower short rates through a monetary acceleration will soon prove counterproductive. The Fed appears to have little control over month-to-month changes in interest rates. It does appear, however, to be able to lower interest rates over a period of a quarter or longer by decreasing the growth rate of the basic money supply.

REFERENCES

Barro, R. J. “Unanticipated Money Growth and Unemployment in the United States.” American Economic Review, March 1977, 101-15.

-. “Unanticipated Money, Output, and the Price Level in the United States.” Journal of Political Economy, August 1978, 549-80.

Berkman, N. G . “On the Significance of Weekly Changes in Ml.” Federal Reserve Bank of Boston New England Economic Review, Maydune 1978,5-22.

Blejer, M. I. “Money and the Nominal Interest Rate in an Inflationary Economy: An Empirical Test.” Journal of Political Economy, July 1978, 529-34.

Brown, W. W. and G. J. Santoni. “Monetary Growth and the Timing of Interest Rate Move- ments.” Federal Reserve Bank of St. Louis Review, August/September 1983, 16-25.

Cagan, P. “The Channels of Monetary Effects on Interest Rates.” National Bureau of Economic Research, 1966.

Cagan, P. D. and Gandolfi, A. “The Lag in Monetary Policy as Implied by the Time Pattern of Monetary Effects on Interest Rates.” American Economic Reoiew, May 1969, 277-84.

Carlson, J. “Expected Inflation and Interest Rates.” Economic Inquiry, October 1979, 597-608.

Carlson, K. M. “The Lag from Money to Prices.” Federal Reserve Bank of St. Louis Review,

Conrad, W. E. “Initial Treasury Bill Market Response to Money Stock Announcements.” Pho-

Cornell, B. “Money Supply Announcements and Interest Rates: Another View.” Journal of Busi-

October 1980, 3-10.

tocopy, Federal Reserve Board, May 1979.

ness, January 1983a, 1-23.


-. “Money Supply Announcements, Interest Rates and Foreign Exchange.” Journal of International Money and Finance, August 1982, 201-08.

-, “The Money Supply Announcements Puzzle: Review and Interpretation.” American Economic Review, September 1983b, 644-57.

Durbin, J. “Testing for Serial Correlation in Least Squares Regression When Some of the Regres- sions are Lagged Dependent Variables.” Econornetrica, May 1970, 410-21.

Engel, C. and J. Frankel. “Why Interest Rates React to Money Announcements: An Explanation from the Foreign Exchange Market.” Journal of Monetary Economics, January 1984,

Frankel, J. A. and G. A. Hardouvelis. “Commodity Prices, Money Surprises, and Fed Credibility.”

Friedman, M. “The Fed and Inflation.” Newsweek, December 29, 1980, 57.

-, “Lessons from the 1979-82 Monetary Policy Experiment.” American Economic Review, May 1984, 397-400.

-. “The Monetary Studies of the National Bureau.” The National Bureau Enters Its 45th Year, 44th Annual Report, National Bureau of Economic Research, 1964, 7-25.

-. “The Role of Monetary Policy.” American Economic Review, March 1968, 1-17. Gibson, W. E. “Interest Rates and Monetary Policy.” Journal of Political Economy, May/June

Grossman, J. “The ‘Rationality’ of Money Supply Expectations and the Short-Run Response of Interest Rates to Monetary Surprises.” Journal of Money, Credit, and Banking, November

Hakkio, C . S. and D. K. Pearce. “The Reaction of Exchange Rates to Economic News.” Economic Inquiry, October 1985, 621-36.

Hardouvelis, G. A. “Market Perceptions of Federal Reserve Policy and the Weekly Monetary Announcements. ” Journal of Monetary Economics, September 1984, 225-40.

Hein, S. E. “The Response of Short-Term Interest Rates to Weekly Money Announcements: A Comment.” Journal of Money, Credit and Banking, May 1985, 264-71.

Hoehn, J. G. “Recent Monetary Control Procedures and the Response of Interest Rates to Fluc- tuations in Money Growth. ” Federal Reserve Bank of Dallas Economic Review, September

Judd, J. P. “Money Supply Announcements, Forward Interest Rates and Budget Deficits.” Federal Reserve Bank of San Francisco Economic Review, Fall 1984, 36-46.

Makin, J. H. “Real Interest, Money Surprises, Anticipated Inflation and Fiscal Deficits.” The Review of Economics and Statistics, August 1983, 374-84.

Mascaro, A. and A. H. Meltzer. “Long- and Short-Term Interest Rates in a Risky World.” Journal of Monetary Economics, November 1983, 485-518.

Mehra, Y. “Inflationary Expectations, Money Growth, and the Vanishing Liquidity Effect of Money on Interest: A Further Investigation.” Federal Reserve Bank of Richmond Economic Reoiew, March/April 1985, 23-35.

Melvin, M. “The Vanishing Liquidity Effect of Money on Interest: Analysis and Implications for Pohcy.” Economic Inquiry, April 1983, 188-202.

Mishkin, F. S. “Monetary Policy and Long-Term Interest Rates: An Efficient Markets Approach.” Journal of Monetary Economics, January 1981, 29-55.

-. “Monetary Policy and Short-Term Interest Rates: An Efficient Markets-Rational Expec- tations Approach.” Journal of Finance, March 1982, 63-72. A longer version of this paper is available as 1981 National Bureau of Economic Research Working Paper.

31-9.

Journal of Money, Credit and Banking, November 1985, Part I, 425-38.

1970, 431-55.

1981, 409-24.

1983, 1-10,

Muth, J. F. “Rational Expectations and the Theory of Price Movements.” Econometrica, July 1961, 315-35.

a2 ECONOMIC INQUIRY

Nichols, D. A., D. H. Small, and C. E. Webster, Jr. “Why Interest Rates Rise When an Unex- pectedly Large Money Stock is Announced.” American Economic Reoiew, June 1983, 383-88.

Roley, V. V. “The Response of Short-Term Interest Rates to Weekly Money Announcements: A Note.” Journal of Money, Credit and Banktng, August 1983, 344-54.

. “Weekly Money Supply Announcements and the Volatility of Short-Tern Interest Rates.” Federal Reserve Bank of Kansas City Economic Recdew, April 1982, 3-15. - and R. Troll. “The Impact of New Economic Information on the Volatility of Short-Term

Interest Rates.” Federal Reserve Bank of Kansas City Economic Review, February 1983,

- and C. E. Walsh. “Unanticipated Money and Interest Rates.” American Economic Re- view, May 1984, 49-54.

Sheehan, R. G. “Weekly Money Announcements: New Information and Its Effects.” Federal Reserve Bank of St. Louis Reotew, August/September 1985, 25-34.

Shiller, R. J., J. Y. Campbell, and K. L. Schoenholtz. “Forward Rates and Future Policy: Inter- preting the Term Structure of Interest Rates.“ Brooktngs Papers on Economic Activity 1, 1983, 173-217.

3-15.

Sims, C. A. “Comment.” Journal of the Amesican Statistical Association, January 1977, 23-4.

Sweeney, R. J. “Anticipated Counter-Cyclical Monetary Policy.” Economic Inquiry, January 1984, 28-36.

Tatum, J. A. “Interest Rate Variability: Its Link to the Variability of Monetary Growth and Economic Performance.” Federal Reserve Bank of St. Louis Review, November 1984, 31-47.

Urich, T. J. “The Effect of Inflation and Money Supply Announcements on Interest Rates.” Journal of Finance, September 1984, 1177-88.

-. “The Information Content of Weekly Money Supply Announcements.” Journal of Mon- etary Economics, July 1982, 73-88.

and P. Wachtel. “Market Response to the Weekly Money Supply Announcements in the 1970s.” Journal of Finance, December 1981, 1063-72.

Wallich, H. C. “Recent Techniques of Monetary Policy.” Federal Reserve Bank of Kansas City Economic Review, May 1984,2140.

Wilcox, J. A. “Why Real Rates Were So Low in the 1970s.’’ American Economic Reoiew, March

Working, H. “Note on the Correlation of First Differences of Averages in a Random Chain.”

Yohe, W. P. and D. Karnosky. “Interest Rates and Price Level Changes.” Federal Reserve Bank

1983, 44-53.

Econometrics, October 1960,916-18.

of St. Louis Reoiew, December 1969, 1936.

the impact of money on short-term interest rates

Documents