21
THOMAS F. COSIMANO University of Notre Dame DENNIS W. JANSEN Texas AL-M University Federal Resewe Policy, 7975- 7985: An Empirical Analysis* This paper examines the operating procedures of the Federal Reserve System over the last decade using VAR models. From April 1975 to September 1979, we check to see whether or not the Fed was using a federal funds operating procedure. We examine the time period from October 1979 to September 1982 to see if the Fed was using a nonborrowed or borrowed reserve operating procedure. The final pe- riod from October 1982 to October 1985 is examined to determine whether or not the Fed was using a borrowed reserve operating procedure. 1. Introduction The Federal Reserve System’s operating procedures have been subject to much scrutiny over the past several years.’ While it is generally accepted that the Fed pursued an intermediate money target throughout the seventies and eighties (see Feige and McGee 1977), debate centers on the nature of the operating procedures used to achieve the money target. Prior to October 1979, the Fed described its operating procedure as a federal funds rate procedure. In October 1979, the Fed announced a change to what has been called a nonborrowed reserves procedure. In this procedure, the Fed calculates the quantity of total reserves consistent with the money target. The Fed forecasts the level of borrowed reserves and sub- tracts this value from desired total reserves to give the nonbor- rowed reserve path. Several critics of the Fed, including Poole (1982), Brunner and Meltzer (1983), and McCallum (1985), argue that the Fed was using a borrowed reserve operating procedure after the October I979 an- nouncement. This procedure works as follows. Under lagged re- serve accounting, in effect until January 1984, the level of required *The authors thank two anonymous referees for helpful comments. Any re- maining errors are our responsibility. ‘Poole (1975) provides a description of the pre-1979 policy. Wallich (1984), Gil- bert (1985), and Rasche (1985) discuss the 1982 change in policy; while Poole (1982), Goodfiiend et al. (1986), Lindsey (1984), Gilbert (1985), Rasche (EXE), and McCdum (1985) discuss the 1979 change. ]ournal of Macroeconomics, Winter 1988, Vol. 10, No. 1, pp. 27-47 27 Copyright 0 1988 by Louisiana State University Press 0164-0704/88/$1.50

Federal Reserve Policy, 1975–1985: An empirical analysis

Embed Size (px)

Citation preview

THOMAS F. COSIMANO University of Notre Dame

DENNIS W. JANSEN Texas AL-M University

Federal Resewe Policy, 7975- 7985: An Empirical Analysis*

This paper examines the operating procedures of the Federal Reserve System over the last decade using VAR models. From April 1975 to September 1979, we check to see whether or not the Fed was using a federal funds operating procedure. We examine the time period from October 1979 to September 1982 to see if the Fed was using a nonborrowed or borrowed reserve operating procedure. The final pe- riod from October 1982 to October 1985 is examined to determine whether or not the Fed was using a borrowed reserve operating procedure.

1. Introduction The Federal Reserve System’s operating procedures have been

subject to much scrutiny over the past several years.’ While it is generally accepted that the Fed pursued an intermediate money target throughout the seventies and eighties (see Feige and McGee 1977), debate centers on the nature of the operating procedures used to achieve the money target. Prior to October 1979, the Fed described its operating procedure as a federal funds rate procedure. In October 1979, the Fed announced a change to what has been called a nonborrowed reserves procedure. In this procedure, the Fed calculates the quantity of total reserves consistent with the money target. The Fed forecasts the level of borrowed reserves and sub- tracts this value from desired total reserves to give the nonbor- rowed reserve path.

Several critics of the Fed, including Poole (1982), Brunner and Meltzer (1983), and McCallum (1985), argue that the Fed was using a borrowed reserve operating procedure after the October I979 an- nouncement. This procedure works as follows. Under lagged re- serve accounting, in effect until January 1984, the level of required

*The authors thank two anonymous referees for helpful comments. Any re- maining errors are our responsibility.

‘Poole (1975) provides a description of the pre-1979 policy. Wallich (1984), Gil- bert (1985), and Rasche (1985) discuss the 1982 change in policy; while Poole (1982), Goodfiiend et al. (1986), Lindsey (1984), Gilbert (1985), Rasche (EXE), and McCdum (1985) discuss the 1979 change.

]ournal of Macroeconomics, Winter 1988, Vol. 10, No. 1, pp. 27-47 27 Copyright 0 1988 by Louisiana State University Press 0164-0704/88/$1.50

Thomas F. Cosimano and Dennis W. Jansen

reserves is predetermined. The Fed subtracts the policy-deter- mined level of borrowed reserves from the predetermined level of required reserves to obtain the target level of nonborrowed re- serves it will supply. “Thus, fluctuations in required reserves are transmitted dollar for dollar to the Feds target for nonborrowed reserves until the Fed makes a policy decision to change its target level of banks borrowing at the discount window” (Poole 1982, 577). Under this procedure, an increase in the equilibrium money stock increases required reserves and, hence, nonborrowed reserves as the Fed maintains the level of borrowed reserves.

The Fed announced the adoption of a borrowed reserve pro- cedure in October 1982. This is the procedure that the critics men- tioned above claimed has been in effect since October 1979. The Fed claimed to abandon the nonborrowed reserve operating pro- cedure because the procedure depended on an Ml target that be- came less meaningful due to financial market deregulation.

The purpose of this paper is to bring empirical evidence to bear on the question of which operating procedure was in effect before and after the October 1979 and October 1982 announce- ments. This question has been previously investigated for the Oc- tober 1979 announcement by Bradley and Jansen (1986). They look at the periods June 1971 to September 1979, and October 1979 to October 1982. In each period, they estimate vector autoregressions (VAR’s) containing the money stock, federal funds rate, and non- borrowed reserves. Then, using a methodology introduced by Feige and McGee (1977), they find that their estimated VAR’s are con- sistent with simple money-market models of a federal funds rate operating procedure before October 1979 and a nonborrowed re- serve procedure for the October 1979 to October 1982 period.

We adopt the Bradley and Jansen (1986) approach in order to investigate Fed policy procedures before and after the October 1982 announcement. We introduce a fourth variable, the spread between the funds rate and discount rate, into the VAR’s we estimate. This variable should control for changes in the discount rate. We intro- duce the discount rate in this form so as to be able to make in- ferences about borrowed reserves from the behavior of the funds rate-discount rate spread under our maintained hypothesis that bor- rowed reserves are a positive function of this spread.’

*We choose to analyze the federal-funds-rate-discount-rate spread instead of the discount rate partly for ease of interpretation of the empirical work. An analysis of the discount rate in conjunction with the other three variables yields similar dy-

28

Federal Reserve Policy

The next section of this paper reports the results of our in- vestigation of Fed operating procedures over the following periods: April 1975 to September 1979, October 1979 to September 1982, and October 1982 to October 1985. We begin the first period in April 1975, the month the Fed began announcing money targets under Congressional mandate. The October 1979 announcement separates the first and second periods, and the October 1982 an- nouncement separates the second and third periods. Some claim that the second Fed policy change actually occurred earlier, in the summer of 1982; but we have chosen our sample to correspond to Fed announcements.

2. Empirical Analysis During the last decade, the Fed claimed to have used oper-

ating procedures based on the federal funds rate, nonborrowed re- serves, and borrowed reserves. Each of these operating procedures has definitive predictions about the relation between the money stock, federal funds rate, nonborrowed reserves, borrowed reserves, and the funds-rate-discount-rate spread. In this section, we examine the historical evidence concerning the relation between these variables for the time periods identified earlier. We examine VAR models for each time period. These models include the log of Ml (M), the federal funds rate (F), the log of nonborrowed reserves (N), and the interest-rate spread (S).3 We choose these variables because they

namics to those reported in the text, but these are difficult to interpret in terms of our money-market models. Our use of the interest-rate spread solves this prob- lem and also allows us to draw inferences about borrowed reserve behavior.

Some may argue that the discount rate changes infrequently, so that changes in the interest-rate spread will be dominated by changes in the federal funds rate. While this has some appeal, the discount rate actually changes frequently for an administered rate. Over the April 1975 to September 1979 period, the discount rate changed 22 times in 53 opportunities. Over the October 1979 to September 1982 period, the discount rate changed 21 times in 36 months, and the adjusted discount rate (see footnote 3) changed 29 times. Finally, in the October 1982 to October 1985 period, the discount rate changed 11 times in 37 months.

‘Before the VAR’s were estimated, a time trend was removed, as well as a sea- sonal pattern by regressing the variables on a quadratic time trend and 11 seasonal dummys. Nonborrowed reserves were adjusted for changes in reserve requirements by the Fed. From 1980 to 1981, the discount rate was subject to a surcharge for borrowing by institutions with over $599 million in deposits. To take account of this surcharge, we calculate an adjusted discount rate defined as adjusted discount rate = discount rate + (surcharge * percent of borrowed reserves covered by sur-

29

Thomas F. Cosimano and Dennis W. Jansen

include a measure of the (intermediate) target of monetary policy, the money stock, as well as measures of possible Fed policy in- struments-nonborrowed reserves, the funds rate, and a measure of the Fed’s stance in the borrowed reserve market.

April 1975 to September 1979 In Table 1, we report estimates of the VAR identified by the

Final Prediction Error (FPE) criterion of Hsiao (1981) for the vari- ables M, F, N, and S for the period 1975:io to 1979:ix.4 This FPE procedure is essentially a method of empirically identifying the lag lengths and causal structure of VAR’s. We employ it to conserve degrees of freedom by empirically identifying restrictions on gen- eral VAR representations, including zero restrictions that allow in- ferences about the causal structure of the model. Further descrip- tion of this methodology is contained in Bradley and Jansen (1986). To aid our discussion of the empirical results, we present the fol- lowing simple model which highlights the characteristics of a federal funds operating procedure.

M, = A0 - h,F, + l yt ; (1)

Nt = PO + P,Mt - P,Ft - PA + EM ; (2)

F, = (~0 + q(M, - M*) - cx3St + l Ft ; and (3)

(Note cont. from p. 29) charge). The spread we use is now given by the difference between the federal funds rate and this discount rate.

The quadratic trend is included to capture any change that may have occurred in the rate of acceleration (or deceleration) of the variables in the different periods. Presumably, part of the Feds reason for changing operating procedures was to alter the growth path of the money aggregates. The quadratic terms were usually sta- tistically significant at the 5% level. (An alternative to quadratic detrending as a method of inducing stationarity is second differencing. Nelson and Plosser (1982) argue strongly for this latter procedure. In applied work, authors such as Sargent (1978) and Sims (1980) still detrend their data.) Here detrending heightens com- parison with prior work by Feige and McGee (1977), and Bradley and Jansen (1986).

4Following Hsiao, we over-fit the model chosen by the FPE criterion. For the first period, we tested the model in Table 1 against six lags of all the variables and obtained the F-statistic on the null hypothesis that the FPE specification is correct of F~s.~~ = 1.3481, Pr > F = 0.1022. For the second period, we tested the model against three lags of all the variables and under the null hypothesis that the FPE specification is correct, we obtain the F-statistic F3,,sD = 1.4310, Pr > F = 0.1028. For the last period, our model was tested against four lags of all the variables, yielding the F-statistic Fd8,@ = 0.8723, Pr > F = 0.6892.

30

Federal Reserve Policy

S, = t& + &F, + &A’, + ~st . (4)

Here the superscript “*” denotes the Feds target, and eit, i = M, N, F, and S, are stochastic changes in the relevant variable at time t.

Equations 1 and 2 are the demand for money and nonbor- rowed reserves, respectively. Equation 3 represents our hypothesis about the Fed’s policy for controlling the federal funds rate. If the money stock is above the Fed’s target, then the Fed increases the funds rate to move back toward its money target. If the spread between the federal funds and discount rates rises (that is, in a funds rate procedure, the discount rate decreases), then borrowed reserves increase and the federal funds rate decreases to counteract this effect. Equation 4 represents the Fed’s policy for setting the spread between the funds rate and the discount rate (that is, under a funds rate procedure, the policy for setting the discount rate).5 If the funds rate increases, a contractionary policy, the Fed increases the discount rate by a smaller amount so that it is easier for the banks to adjust to the contraction in its reserves by borrowing from the Fed. We also assume that the Fed keeps the ratio between nonborrowed and borrowed reserves fairly constant, which implies that borrowed reserves, and hence the spread, increase whenever nonborrowed reserves increase.

In row one of Table 2, we present the predicted effect of shocks to M, N, F, and S on each dependent variable in our model. We also calculate the estimated total effect as in Feige and McGee (1977) and Bradley and Jansen (1986). To motivate the total effect, con- sider a VAR containing the endogenous variables (W,X,Y,Z). The total effect of X on Y measures the effect on Y of a permanent shock to X after all subsequent adjustments in the vector of endogenous variables take place. The total effects calculated for a federal funds rate operating procedure appear in row two of Table 2.

We see that an increase in the federal funds rate decreases both the money stock and nonborrowed reserves. The money stock decreases because of the reduction in the demand for money when the federal funds rate increases. Nonborrowed reserves are reduced when the funds rate increases partly because of a decrease in the demand for excess reserves and partly because of the induced re-

‘The Fed has not made explicit its guideline for adjusting the discount rate. Thus, our Equation (4) is one possible model of this behavior which is consistent with our empirical results. Other models of Fed behavior at the discount window are possible, but an investigation of this issue is beyond the scope of this paper.

31

TABL

E 1.

19

75:iv

-197

9:ix

8.07

6 .

1O-6

1.

917.

1O

-5

7.60

8 *

lO-‘j

-1

.186

.

1O-4

3.

539

* lo

-” 4.

314

* 1o

-4

1.40

8.

lo-’

7.82

5 .

1O-5

-1

.576

. 1O

-4

3.35

8 *

lo-’

1

ail(l

) =

0.47

6 .

L (4

.20)

xq

l =

0.47

6

q2(6

) =

-0.0

0272

. L

- 0.

0030

9.

L2 -

0.

0016

3.

L3 +

0.

0024

5.

L4

(1.2

3)

(1.1

5)

(0.8

4)

(1.2

6)

+ 0.

0031

0~

L5 -

0.

0035

0 * L

6 ZC

Q, =

-0.

0053

9 (1

.47)

(2

.03)

a,

(2)

= 0.

0012

2 * L

+

0.00

491

* L2

%q,

=

0.00

613

(0.4

5)

(1.7

6)

az1(

6) =

11.

97.

L +

10.3

4 * L

2 -

17.2

4 * L

3 -

6.28

. L4

(1

.10)

(0

.87)

(1

.53)

(0

.58)

+

12.4

6 * L

5 +

22.1

2 * L

6 Za

=

33.3

7 (1

.19)

(2

.41)

II II II II II II II

TABL

E 2.

Pr

edic

ted

and

Estim

ated

R

espo

nses

to

Sh

ocks

U

nder

Al

tern

ativ

e O

pera

tine

Proc

edur

es

NonB

orro

wed

Federal Reserve Policy

duction in the money stock and, hence, in required reserves. While the predicted effect of a change in the funds rate on the spread is ambiguous (the Fed increases S in response to the higher F, but reduces S in response to the lower N), the estimated total effect is negative, indicating that the Fed is more interested in maintaining the ratio of nonborrowed to borrowed reserves than in helping banks adjust to a change in the funds rate.

The estimated responses to a money demand shock show that the federal funds rate increases, implying that the Fed is most in- terested in increasing the federal funds rate when the money stock rises above target. Since nonborrowed reserves are estimated to de- cline in response to an increase in the federal funds rate, these results indicate that the funds rate is the most important effect on the demand for nonborrowed reserves. We also see that the spread decreases, implying that the decline in nonborrowed reserves is large relative to the increase in the federal funds rate.

A shock to nonborrowed reserves raises the spread so that borrowed reserves maintain the same proportion of total reserves. The subsequent increase in the spread decreases the funds rate, which in turn increases the money stock.

The estimated total effects of an increase in the spread on the money stock and nonborrowed reserves accords with the predic- tions of the model. The increase in nonborrowed reserves implies that the effect of the funds rate and the money stock is stronger than the effect of the spread on nonborrowed reserves. There is, however, an inconsistency between the predicted (negative) and calculated (positive) response of the federal funds rate. An expla- nation is that borrowed reserves depend on current and future in- terest-rate spreads due to Fed credit rationing. Following Good- friend (1983), the expectation of a higher future spread reduces today’s borrowed reserves, which in turn increases the funds rate.

In general the empirical results correspond to the predictions of our simple model of the money market. These results confirm the previous findings of Bradley and Jansen (1986). Their results are not overturned by our choice of a different sample period and our inclusion of the interest-rate spread to control for the Fed’s stance in the borrowed reserve market.

October 1979 to September 1982 Table 3 reports the VAR estimates for the second period. These

results can be interpreted with a simple money-market model in

35

+

II II II

Thomas F. Cosimano and Dennis W. Jansen

which the Fed uses a nonborrowed reserve rule to hit its money- supply target.

M, = ho - X,F, + l MMt ; (5)

F, = (Y,, + cxlMt - a2Nt + l Ft ; 63)

N, = PO - &(M, - M*) + eNt ; and (7)

Equation (5) is the money demand equation, (6) is the inverse de- mand for reserves. In this period, the Fed determines nonborrowed reserves and the funds rate is then determined by the demand for reserves.

Equation (7) represents the policy rule used by the Fed dur- ing this period. If the money stock is above target, the Fed reduces nonborrowed reserves. Equation (8) represents the Fed’s attempt to protect banks from abrupt changes in the money stock. Under lagged reserve accounting, an increase in the money stock increases required reserves leading banks to borrow reserves to counteract the contraction in nonborrowed reserves via Equation (7). In order to reduce the cost of this adjustment, the Fed increases the spread.

In rows three and four of Table 2, we report the predicted and estimated response of the money market for a random shock to each of the dependent variables in the model. An increase in the federal funds rate reduces money demand, which the Fed coun- teracts by increasing nonborrowed reserves and reducing the spread. Note that those critics who argue that the Fed accommodated in- creases in required reserves in this period would expect a decrease in nonborrowed reserves as the money stock declines, a result that is not obtained. In addition, we find that an increase in the money stock raises the funds rate and lowers nonborrowed reserves. Under a borrowed reserve procedure, the amount of nonborrowed re- serves should rise in this case because of the increase in required reserves. This increase in nonborrowed reserves would then reduce the funds rate and the spread. Thus the empirical results are in- consistent with a borrowed reserves procedure.

An increase in nonborrowed reserves is predicted to reduce the funds rate based on our model of the money market. We es- timate that the federal funds rate rises in response to an increase

38

Federal Reserve Policy

in nonborrowed reserves. Bradley and Jansen (1986) also report this finding and suggest that this result may be a manifestation of the announcement effect examined by Cornell (1983); Nichols, Small, and Webster (1983); and Boley (1983). The increase in the money supply and spread as a result of the increase in nonborrowed re- serves corresponds to the prediction of the theory linking increased money stock and reserves to a future contraction by the Fed, and hence, higher future interest rates. Higher future interest rates lead to higher current rates.

Finally, we find that changes in the spread do not influence the money stock, funds rate, or nonborrowed reserves. This result corresponds to our argument that banks must borrow the difference between required reserves and nonborrowed reserves horn the Fed, and thus the demand for reserves is independent of the spread. If the spread does not influence the demand for reserves, then it does not effect the funds rate, or the money stock. These results are also inconsistent with the contention of Poole (1982) and others that the Fed followed a borrowed reserve procedure in this period. In a borrowed reserve procedure, the spread does affect reserve de- mand.

The variance of the one-period forecast error of the federal funds rate, as measured by the mean-square error, MSE, increased by twenty-fold compared with the first period. In addition, there is a fourteen-fold increase in the MSE of the interest-rate spread. These qualitative results are not unexpected under a nonborrowed reserve operating rule, since the policy constraint placed on non- borrowed reserves implies greater fluctuations in the interest rates. We also see that the MSE of the money stock and nonborrowed reserves increases by a factor of two. This larger MSE may reflect an increase in the Feds desire to hit its money-supply target at the expense of larger fluctuations in the interest rates.6 Tinsley, von zur Muehlen, and Fries (1982) argue that the attempt by the Fed to return more quickly to its money target implies an increase in the variance of the money stock and nonborrowed reserves in the short term. This increase in the Feds willingness to maintain its money- stock target could also lead to larger fluctuations in the federal funds rate and interest-rate spread.

Our results are similar to those reported by Bradley and Jan-

‘The lo/79 to 9/82 period is also the period of the Carter credit controls. These controls were imposed in March 1980. They were relaxed in May 1980 and phased out over June and July. Some have claimed that much of the observed increase in

39

Thomas F. Cosimano and Dennis W. Jansen

sen (1986), and support their claim that the Fed adopted a non- borrowed reserve rule in October 1979. Our results do not support the view that the Fed adopted a borrowed reserve procedure in this period.

October 1982 to October 1985 During this period the Fed claimed to adopt a borrowed re-

serve operating procedure similar to the procedure that some argue was adopted in October 1979. Table 4 contains estimates of the VAR for this period. These results can be explained with the aid of the following simple money-market model of a borrowed reserve rule.

M, = ho - h,F, + eyt ; (9)

F, = a0 + alMt - aZNt - c& + ‘+t ; (10)

(Note cont. from p. 39) variability over this period is due to these credit controls and not to the Fed’s operating procedure.

We investigated this issue by examining the contribution of these five months to the total sum of squared residuals for the second period.

Sum of Squared Residuals

M F N S

1. 3/80-5/80 .00005232 6.049 .002139 0.6717 2. 6/80-7/80 . oooo2994 1.513 .0001156 0.7614 3. 10/79-g/82 .0006769 23.395 .005375 16.6036

Clearly these five months contribute much more than a 5/36 or 13% share to the total sum of squared errors (SSE) for F, N, and S over the lo/79 to 9/82 period. For the funds rate, the first three months of controls contribute about 26% of the SSE. These three months contribute about 40% of the SSE for nonborrowed re- serves, and about 40% of the SSE for the funds-rate-discount-rate spread. Inter- estingly, this period does not contribute even a 13% share of the variance of the money stock.

We have estimated our VAR for the lo/79 to 9/82 period with dummy variables to capture any effects of the credit control period on the level of the variables involved. This does not change our calculated total effects or the order of magni- tude of these forecast error variances reported in Table 3. Moreover, the order of magnitude of these forecast errorvariances would not be changed by eliminating the credit control period from the sample. Thus, although the credit control period contributes strongly to the reported variance estimates, it does not by itself cause the ordinal variance rankings among the periods to change.

40

Federal Reserve Policy

N, = R0 + RiM, + eNt ; and (11)

s, = a0 + 6,(M, - M:) - 6,Nt + ~st . (12)

The demand for money and the inverse demand for nonborrowed reserves are the same as under the nonborrowed reserve rule.

Under the borrowed reserve procedure, an increase in the money stock leads to an increase in total reserves, and hence, an increase in the amount of nonborrowed reserves supplied by the Fed. If the Federal Open Market Committee (FOMC) has called for restraint, the open market desk increases the amount of bor- rowed reserves by increasing the interest-rate spread in order to partly meet the higher demand for total reserves. In this case, an increase in the money supply increases the spread; that is, reduces the discount rate. The idea behind restraint is that borrowed re- serves are viewed by banks as less permanent than nonborrowed reserves; so that by partially meeting the increased demand for re- serves with borrowed reserves, the Fed is being less accommoda- tive. However, the lower discount rate increases the total supply of reserves today in the face of an increase in the money supply. In this sense, the Fed is accommodating the expansion of the money supply when the FOMC calls for restraint.

Equations (11) and (12) capture the above elements of a bor- rowed reserve procedure. Nonborrowed reserves (11) increase when the money stock increases so as to meet the higher level of re- quired reserves. Borrowed reserves and the spread increase with the money stock due to the restraint argument. In addition, with money constant, an increase in nonborrowed reserves will be offset by a reduction in the spread and, hence, borrowed reserves.

Rows five and six of Table 2 report the predicted and actual responses to an unanticipated change in the dependent variables of our simple money-market model under a borrowed reserve oper- ating procedure. An increase in the federal funds rate causes a de- crease in the demand for money which results in a decrease in the amount of nonborrowed reserves supplied by the Fed. The de- crease in the money stock reduces the spread so that borrowed re- serves fall, while the decrease in nonborrowed reserves increases the spread since the Fed has raised the borrowed reserve target. The empirical results imply that the second effect dominates.

An increase in the money stock is accommodated by the Fed in the form of larger nonborrowed reserves. The increase in the money stock and nonborrowed reserves has opposite effects on the

41

TABL

E 4.

19

82x-

1985

:x

Mt

c(4)

~4

) 0

[1[

Ft

= 0

Nt 0

St

dl)

~(1)

45

5)

0

-1.0

62

* 1o-

4 1.

978

* 1o-

5 2.

835

*

9.40

1.

1o-2

-3

.161

. 1O

-3

5.96

7.

lo-”

4.51

3 * 1

o-4

-2.4

75

* 4.

960

- 1o-5

I

1O-3

lo

-”

a,,(4

) =

0.75

6.

L +

0.04

3.

L2 +

0.

480.

L3

-

0.49

8.

L4

(6.2

1)

(0.2

7)

(2.9

8)

Za,,

= 0.

781

(4.5

9)

(~~~

(1) =

-0.0

0320

. L

(3.7

5)

I&x,

, =

-0.0

03

a&l)

= 1.

204

*L

(5.6

2)

h,,

= 1.

204

a=(2

) = -

6.55

.L

+ 11

.05.

L'

(2.3

6)

(4.9

9)

Lx,

= 4.

50

c&l)

= -0

.627

.L

(2.2

9)

h,

= -0

.627

(~~~

(1) =

-0.0

271.

L

(4.1

4)

Lx,,

= -0

.027

Q(1

) =

0.59

3. L

(6.1

8)

zcQ

3 =

0.59

3

c~(l)

=

6.09

. L

(1.6

1)

h,,

= 6.

09

(Y&(

l) =

0.45

5. L

(5.4

8)

Lx,,

= 0.

455

a,,(2

) = -

4.84

. L

+ 7.

61-L

’ (2

.40)

(5

.04)

C

a,,

= 2.

77

Thomas F. Cosimano and Dennis W. Jansen

demand for nonborrowed reserves, which yield an ambiguous effect on the federal funds rate. The empirical results show that the in- crease in nonborrowed reserves dominates, so the federal funds rate decreases. The increase in the money stock increases the spread because of the restraint called for by the FOMC in the face of an increase in the money supply. However, the larger supply of non- borrowed reserves reduces the spread. The empirical results show that the spread increased in reaction to an increase in the money stock. This suggests that the Fed was interested in restraint when the money stock increased.

An increase in nonborrowed reserves reduces the federal fimds rate and the spread. Subsequently, the demand for money in- creases, while the funds rate increases because of the fall in the spread. The empirical results suggest that the net effect of these changes is a decrease in the money stock and the federal funds rate. The initial increase in the demand for money tends to increase the spread, while the increase in the nonborrowed reserves reduces the spread. This first channel of influence has been titled “restraint” to indicate that borrowed reserves move to offset changes in nonbor- rowed reserves. Thus, restraint by the Fed tends to increase the spread. The empirical results indicate that the indirect effect of nonborrowed reserves via the funds rate and money stock domi- nants the direct effect of nonborrowed reserves on the spread.

An increase in the interest-rate spread reduces the demand for nonborrowed reserves and the federal funds rate. This reduction in the federal funds rate increases the demand for money which the Fed accommodates by increasing nonborrowed reserves. In this case, the empirical results coincide with the predictions of our sim- ple theory of the money market under a borrowed reserve oper- ating procedure.

These results for the 1982:x to 1985:x period correspond to the operating procedure the Fed announced in October 1982. We note that our results indicate that an increase in the federal funds rate leads to a decrease in both the money stock and nonborrowed reserves. This supports Rasche’s (1985) argument that the borrowed reserve procedure adopted by the Fed has the same implications for the money stock and nonborrowed reserves as the federal funds operating procedure. In addition, the MSE of the federal funds rate and interest-rate spread are in line with Rasche’s expectations.

Rasche argues that the attempt to control the money stock by a borrowed reserve procedure is essentially the same as a federal funds procedure, but with additional variation in the federal funds

44

Federal Reserve Policy

rate due to errors in the borrowed reserve relationship with the interest-rate spread. We see that the MSE of the interest-rate spread is fifty percent larger in the third period as compared to the first period, but the federal funds rate has an MSE that is three times as large. This supports Rasche’s contention that the borrowed re- serve function would introduce additional variation in the federal funds rate. However, the MSE of the money stock is virtually the same in the first and third period, which is not compatible with Rasche’s criticism of a borrowed reserve procedure. We also see that there is additional variation in nonborrowed reserves when we compare the first and third periods. On the other hand, the bor- rowed reserve procedure gives less variation in the money supply relative to the nonborrowed reserve operating procedure of the sec- ond period, which is consistent, via Tinsley’s et al. argument, with the Fed’s stated position that it was not as interested in hitting its money-supply target during this last period.

3. Conclusion In this paper, we use VAR models to help identify the op-

erating procedure the Federal Reserve used to control the money stock during three periods over the last decade. The results of this work suggest that the Fed used a federal funds operating procedure to maintain its money-stock target from April 1975 to September 1979. From October 1979 to September 1982, our results suggest that the Fed was using a nonborrowed reserve procedure. In par- ticular, our results do not support the claim that the Fed was using a borrowed reserve procedure during this period. Finally, we pre- sent evidence that the Fed was using a borrowed reserve operating procedure from October 1982 to October 1985. The results during this last period generally support the claim that a borrowed reserve procedure is, in practice, similar to a federal funds operating pro- cedure.

Received: September 1985 Final version: April 1987

References Bradley, M.D., and D.W. Jansen. “Federal Reserve Operating Pro-

cedure in the Eighties: A Dynamic Analysis.” Journal of Money, Credit and Banking 18 (August 1986): 323-35.

45

Thomas F. Cosimano and Dennis W. Jansen

Brunner, K., and A.H. Meltzer. “Strategies and Tactics for Mon- etary Control.” Money, Monetary Policy, and Financial Znstitu- tions, edited by K. Brunner and A.H. Meltzer, 59-104. Car- negie-Rochester Conference Series on Public Policy, 18. Amsterdam: North Holland, 1983.

Cornell, B. “The Money Supply Announcement Puzzle: Review and Interpretations.” American Economic Review 13 (March 1983): 144-57.

Feige, E., and R. McGee. “Money Supply Control and Lagged Re- serve Accounting.” Journal of Money, Credit and Banking 9 (November 1977): 536-56.

Gilbert, R.A. “Operating Procedures for Conducting Monetary Pol- icy.” Federal Reserve Bank of St. Louis Review 67 (February 1985): 13-21.

Goodfriend, M. “Discount Window Borrowing, Monetary Policy, and the Post-October 6, 1979, Federal Reserve Operating Pro- cedure.” Journal of Monetary Economics 12 (September 1983): 343-56.

Goodfriend, M., G. Anderson, A. Kashyap, G. Moore, and R.D. Porter. “A Weekly Rational Expectations Model of the Nonbor- rowed Reserve Operating Procedure.” Federal Reserve Bank of Richmond Economic Review 72 (January/February 1986): 11-28.

Hsiao, C. “Autoregressive Modelling and Money-Income Causality Detection.” Journal of Monetary Economics (January 1981): 85- 106.

Lindsey, D.E. “The Monetary Regime of the Federal Reserve Sys- tem.” Paper presented at a conference on Alternative Monetary Regimes, sponsored by the Ellis L. Phillips Foundation and Dartmouth College, August 22-24, 1984.

McCallum, B.T. “On the Consequence and Criticisms of Monetary Targeting.” Journal of Money, Credit and Banking 17, pt. 2 (No- vember 1985): 570-97.

Nelson, C.R., and C.I. Plosser. “Trends and Random Walks in Macroeconomic Time Series.” Journal of Monetary Economics 10 (September 1982): 139-62.

Nichols, D.A., D.H. Small, and C.E. Webster. “Why Interest Rates Rise when an Unexpectedly Large Money Stock is Announced.” American Economic Review 73 (June 1983): 383-88.

Poole, W. “The Making of Monetary Policy Description and Anal- ysis.” Economic Inquiry 13 (June 1975): 253-65.

-. “Federal Reserve Operating Procedures: A Survey and Evaluation of the Historical Record Since October 1979.” Journal

46

Federal Reserve Policy

of Money, Credit and Banking 14, pt. 2 (November 1982): 575- 96.

Rasche, R. H. “Interest Rate Volatility and Alternative Monetary Control, Procedures.” Federal Reserve Bank of San Francisco Economic Review (Summer 1985): 46-63.

Roley, V.V. “The Response of Short-Term Interest Rates to Weekly Money Supply Announcements.” Journal of Money, Credit and Banking 15 (August 1983): 344-54.

Sargent, T. J. “Estimation of Dynamic Demand Schedules Under Rational Expectations.” Journal of Political Economy 86 (Decem- ber 1978): 1009-44.

Sims, C.A. “Macroeconomics and Reality.” Econometrica 48 (Jan- uary 1980): l-48.

Tinsley, P.A., P. von zur Muehlen, and G. Fries. “The Short-Run Volatility of Money Stock Targeting.” Journal of Monetary Eco- nomics 10 (September 1982): 215-37.

Wallich, H.C. “Recent Techniques of Monetary Policy.” Federal Reserve Bank of Kansas City Economic Review (May 1984): 21- 30.

47