CENTRAL BANK SECRECY, INTEREST RATES, AND MONETARY CONTROL
THOMAS F. COSIMANO and JOHN 6.VAN HUYCK*
We construct a dynamic rational expectations model of thefederalfinds and deposit market that provides a rationale for central bank secrecy about current monetay aggregate objectives. In this analysis, the Trading Desk values secrecy because it reduces the influence of monetary control policy on interest rates. We then examine actual U.S. experience with monetary control and determine that the reserve bias predicted by the model is present in the datafrom 2978 to 1985.Finally, we demonstrate that central bank secrecy may not lower the value of commercial banks.
I.
INTRODUCTION
Central banks are notoriously secretive not only about the policy making process but also about the actual policy adopted. For example, the Federal Open Market Committee of the Federal Reserve System (hereafter denoted FOMC) does not keep minutes of its policy meetings and does not disclose the Domestic Policy Direc tivethe adopted policyuntil the direc tive is no longer current. While it is easy to rationalize secrecy about the policy making process, intuitively, secrecy about the adopted policy is puzzling since the uncertainty created might be expected to inhibit a central bank's ability to achieve the adopted policy. The FOMC provided several arguments for its policy of nondisclosure in Merrill v. FOMC, one of which was the "inappro priate market reaction" defense. The "in appropriate market reaction" defense ar
Department of Finance at the University of Notre Dame and Department of Economics at Texas ALM University. Ron Balvers, Michael Dotsey, Burkhard Drees, Marvin Goodfriend, Raymond Lombra, Thomas Mayer, Carlos Pinerua, Richard Sheehan,and Carl Walsh made helpful commentson earlier versions of this paper. Lisa Narbut provided research assis tance.
Economic Inquiry VO~.XXM,July 1993,370382
370
gues that disclosure of the current policy would cause markets to move interest rates in a manner contrary to the central bank's objectives.' This paper uses a choicetheoretic framework to formally examine the "inappropriate market reac tion" defense of central bank secrecy. Specifically, this paper provides a ratio nale for secrecy about _{a} central bank's current monetary aggregate objectives in a dynamic rational expectations model of the federal funds and deposit market. In the model, secrecy forces commercial banks to solve a bivariate signal extraction problem in which they use the current federal funds rate and the current deposit rate to estimate both the current reserve target and future reserve targets. The sig nal extraction problem results in the fed eral funds rate being less sensitive to the current level of reserves. Central bank concern for the level of interest rates can induce it to consistently set total reserves above its current reserve target. Secrecy, by reducing the sensitivity of the interest rate to the current level of reserves, lowers the marginal interest rate cost of moving towards the central bank's
1. See Goodfriend [1986]and Mayer [1987l.
@WesternEconomic Association International
COSIMANO I% VAN HUYCK: CENTRAL BANK SECRECY
_{3}_{7}_{1}
reserve target and, hence, lowers the up ward bias in total reserves. Consequently, the central bank prefers secrecy to disclo sure. In this analysis, the Trading Desk values secrecy because it reduces the in fluence of its monetary control policy on interest rates. Under both secrecy and disclosure, our analysis predicts that the Trading Desk will exhibit an upward bias in its choice of total reserves and, hence, in its choice of monetary aggregates such as _{M}_{1} and M2. In section IV, we attempt to determine if M1 was above target on average from 1978 to 1985. The mean percentage differ ence between M1 and the midpoint of the ranges for M1 contained in the Domestic Policy Directive over the period 1978 to 1985 was 1.6 percent per year, which is greater than zero at the 5 percent level of statistical significance. Conventional wisdom holds that the welfare of commercial banks is lowered by central bank secrecy. The influence of cen tral bank secrecy on the variability and predictability of interest rates is a common object of analysis. A standard result is that secrecy reduces the predictability of inter est rates.2 However, a second more subtle effect of secrecy is that the information regime changes the covariance structure of the economy. Given the choicetheoretic framework of this paper, it is possible to examine the conjecture that central bank secrecy low ers the value of commercial banks. In general, it is not possible to order the relative value of commercial banks under the alternative information regimes. The
2. See Dotsey (1987, Tabellini [1987, and Rudin [1988]on the relationshipbetweensecrecy and interest rate variability and predictability. See Lombra and Struble [1979] on the relationshipbetween interest rate predictability and the welfare of financial institutions. Our manuscript, Cosimano and Van Huyck [1991], demonstrates that there is no necessary relationship between interest rate predictability and commercial bank value. See Hakansson et al. [1982] for an abstract discussion of necessary and sufficient conditions for information to have social value in pure exchange.
information regime influences the value of commercial banks through two channels. Secrecy motivates the trading desk to sup ply less reserves on average, which re duces commercial bank profits; but se crecy increases the covariance of deposits with the spread between the loan rate and the deposit rate, which increases commer cial bank profits. The analysis reveals that for reasonable parameter values central bank secrecy can increase the value of commercial banks. Consequently, it is pos sible that both the central bank and com mercial banks prefer secrecy to disclosure of the current monetary aggregate objec tive.
II. THE TRADING DESKS MONETARY CONTROL PROBLEM
The monetary policy process in the United States can be characterized as fol lows. The FOMC solves a decision prob lem involving objectives such as high em ployment, economic growth, price stabil ity, and exchange rate stability. The FOMC's decision problem results in a di rective to the Trading Desk at the Federal Reserve Bank of New York, involving growth rate targets for monetary aggre gates and a tolerance range for the federal funds rate.3 In this section we formalize a Trading Desk's monetary control problem. Sup pose the Trading Desk engages in open market operations in an attempt to achieve the intermediate targets contained in the policy directive from the Central Bank. The Trading Desk knows the current reserve target, which is contained in the Policy Directive to the Trading Desk, but future reserve targets are uncertain. _{A}_{s}_{} sume that the Trading Desk treats interme diate targets as exogenously given and that the reserve targets, _{R}_{f}_{,} evolve accord ing to the following stochastic process:
3. See Cosimano and Van Huyck [1989].
372
ECONOMIC INQUIRY
where wt is normally distributed with
mean zero and variance
itive and less than the reciprocal of the Trading Desk’s discount rate. The stochas tic nature of the reserve targets reflects the Central Bank’s response to changing mac roeconomic conditions. In this paper, se crecy refers to an information regime in which the Central Bank does not disclose the current reserve target, RT, to the public until it is no longer current. The Trading Desk chooses total re serves, R, to minimize the following loss function:
u$ and _{a}_{,} is pos
m
(2)S1= C6’’ET “(1 hXR,  RT)*]/2 + hf,),
tT
where fi denotes the federal funds rate in period t, 6 denotes the Trading Desk’s time discount rate, and Eldenotes the expecta tion operator conditional on the Trading Desk’s information set at time _{T}_{,} which contains realizations of variables dated _{T} or earlier. The parameter h formalizes the trade off between the reserve target and federal funds rate objective. A concern for low interest rates, h > 0, weakens the Trading Desk’s incentive to achieve the reserve targets exactly. Lombra and Moran [1980] and Karamouzis and Lombra [1989] document the FOMC‘s practice of choosing inconsis tent targets for monetary aggregates and interest rates from the menu of policy options compiled by the staff. Karamouzis and Lombra [1989, 61 conclude that ”this tendency to select seemingly inconsistent shortrun alternatives calling for ‘low’ money growth and ‘low’ interest rates in order to secure a unanimous or near unan imous vote on the Directive seems to be an enduring characteristic of Fed
p~licymaking.”~For whatever reason, it appears that the Trading Desk is con cerned with both reserve and interest rate objectives and, hence, for descriptive pur poses we treat h as greater than 0 but less than _{1}_{.} In order to formalize the constraints on the Trading Desk’s decision problem, we need to specify how the federal funds rate is influenced by the Trading Desk’s open market operations. It seems reasonable to believe that the monetary control problem is a dynamic problem, that is, the expected supply of reserves in the future will influ ence the current federal funds rate. Con sideration of portfolio adjustment costs (see McCallum [1985,584]) or loan adjust ment costs (see Cosimano [1988, 1201) would produce such dynamics. Section VI constructs an explicit dynamic rational expectations model of the commercial banking industry in which the dynamics arise because of deposit adjustment costs. We delay our discussion of the commercial banking industry in order to focus on the Trading Desk’s monetary control problem. Assume that the federal funds rate is de creasing in current reserves, but increasing in lagged and expected future reserves. Specifically, assume that the federal funds rate is related to reserves by the following equation:
The parameters pOI pl, p2 are positive constants, which are functions of the parameters of the supply and demand for deposits. As shown in section VI, the contemporaneous effect of reserves on the funds rate is greater than the lagged or
4. See Kane [1980]and Woolley [1984]for a dis cussion of political pressure on the Federal Reserve System for low interest rates.
COSIMANO & VAN HUYCK CENTRAL BANK SECRECY
_{3}_{7}_{3}
expected future effect, that is, p1 > p2. The parameter p is the representative commercial bank’s discount rate. The random variable ut denotes an industry wide cost shock and the random variable e, denotes a deposit supply shock. Let ui denote the variance of ut and 0: denote the variance of e,. The industrywide cost shocks and the deposit supply shocks are assumed to be uncorrelated. The term Ei is the expectation operator conditional on the representative commercial bank’s information set at time _{t}_{.}
Secrecy denotes the information regime in which commercial banks do not observe the current innovation to the reserve tar get, yt. Disclosure denotes the information regime in which commercial banks do ob serve the current innovation, y.~r.Let _{f}_{i} de note the representative commercial bank’s information under secrecy, and let If de note the information set under disclosure. Equation (3) reflects the commercial banking industry’s response to Trading Desk behavior and is a constraint on the Trading Desk’s decision problem. An im portant characteristic of equation (3) is the dependence of the federal funds rate on expected future total reserves, which the Trading Desk can and does influence. This influence will depend on the information regime. Substituting the federal funds rate equation, (3), into equation (2) gives the Trading Desk‘s constrained loss function:
(4)S, = C
m
t1
6’’El ([(l  h)(R,  Ry)2]/2
The analysis derives a discretionary equilibrium under both a regime of disclo sure and a regime of secrecy. A discretion ary equilibrium consists of the following features: an expectations function, which determines the commercial banks’ expec tation of future reserves, _{(}_{E}_{!}_{R}_{,}_{+}_{l}_{}}_{r}_{,} as a function of the commercial banks’ current information, either I; or If; a sequence of decision rules for the Trading Desk, which determines the Trading Desk’s choice of total reserves as a function of the Trading Desk’s current information. In a discre tionary equilibrium the sequence of deci sion rules minimizes the Trading Desk’s constrained loss function, given the com mercial banks’ expectation function, and the commercial banks’ expectations are the best linear unbiased forecast of total reserves, given the Trading Desk’s deci sion rules.5 The alternative discretionary equilibria discussed in the following sec tions result from alternative assumptions about the information available to the commercial banks.
111.
DISCLOSURE
To focus the analysis and provide a useful benchmark case, suppose that, con trary to actual practice, the Central Bank disclosed the current reserve target, _{R}_{:}_{.} In
a linearquadraticgaussian framework, a
reasonable conjecture for the functional form of the Trading Desk’s decision rule, which turns out to be correct, is
_{(}_{5}_{)}
R, = uo + n,RT,
where uo and al are parameters to be de termined. If the conjectured decision rule
where R1,is an initial condition. The final component needed to complete the ana lytical framework is the determination of the commercial banks‘ expectation of fu
ture reserves,
[E!R,X.
*
.
.
,
I.
5. See Cosimano and Van Huyck [1989]for an anal
ysis of dynamic monetary control when the Trading Desk can make commitments.Secrecy underminesthe ability of the Trading Desk to make commitments. An interesting result is that the value of the Trading Desk’s loss function can be smaller under a regime of discre tion and secrecy than under a regime of commitment and disclosure.
374
ECONOMIC INQUIRY
satisfies the conditions of a discretionary equilibrium, then the commercial banks would use equations (5) and (1)to forecast total reserves. Substituting (1)into (5) and taking ex pectations conditioned on _{I}_{f}_{,} which in cludes RT, gives the following expecta tions equation for the commercial banks:
for t = T, ~+1, . Under disclosure the Trading Desk's current choice of total re serves, R,, does not influence the commer cial banks' expectations of _{R}_{,}_{+}_{,}_{,} because the commercial banks know the current total reserve target, _{R}_{T}_{.} The final step in deriving a discretion ary equilibrium under disclosure is to ver ify that the conjectured decision rule, (5), minimizes the Trading Desks constrained loss function, (4), subject to the conjec tured expectations equation, (6). Substitut ing (6) into (4) and taking the derivative with respect to R, gives the Euler condi tions for this problem, which can be re written as follows:
for t= T, T+ 1,
conjectured functional form of equation (5) where uo equals h(p,  6p2)/(l  h) and u, equals 1.Since the contemporaneous ef fect of reserves on the funds rate is greater than the lagged effect, that is, _{p}_{1} _{>} _{p}_{2} and
since 6 is less than 1 by assumption, uo is a positive constant. The Trading Desk's decision rule, (7), prescribes setting total reserves above the current reserve target by some constant amount. The analysis predicts that the Trading Desk will exhibit an upward bias in its choice of total reserves and, hence, in its choice of monetary aggregates such as M1 and M2. The size of the bias de pends on how much weight the Trading
Equation (7) has the
Desk puts on keeping the federal funds rate low, that is, parameter h. _{I}_{f} h equals 0, then the Trading Desk achieves the reserve target in every period and there is no bias. The greater the concern for low interest rates the larger the bias.6
IV.
SECRECY
In reality, central banks almost never disclose their current policy objectives and often go to great lengths to conceal them. Consequently, although providing a use ful benchmark case, the analysis of dy namic monetary control under disclosure fails to explain observed central bank se crecy. Suppose that the Central Bank does not disclose the current reserve target con tained in the policy directive and that this target is not inferable from known Central Bank objectives and perceivable macro economic conditions. Under secrecy commercial banks fore cast R,,, using their knowledge of current observable variables and of the Trading Desk's decision problem. Again, a reason able conjecture for the functional form of the Trading Desk's decision rule, which turns out to be correct, is _{R}_{,} _{=} _{b}_{o} _{+} b,RT, where bo and b, are parameters to be determined. Given the conjectured deci sion rule and equation _{(}_{l}_{)}_{,} the expecta tions function for commercial banks under secrecy is
6. A referee has pointed out that there is a parallel
between the reserve bias demonstrated in our discre tionary analysis of the dynamic monetary control
problem and the inflation bias demonstrated in Barro and Gordon's (19831discretionary analysis of the re peated inflation surprise game.
COSIMANO & VAN HUYCK CENTRAL BANK SECRECY
_{3}_{7}_{s}
for t = T, ~+1, .Unlike disclosure, under secrecy the banks can only estimate _{R}_{T} based on knowledge of the model and on contemporaneous observation of the fed eral funds rate and the deposit rate, that is, the commercial banks confront a bivar iate signal extraction problem. The bivariate signal extraction problem is solved in the appendix using the com mercial banking model developed in sec tion VI. Observing the federal funds rate and the deposit rate when combined with the model and the history of all stochastic forcing processes allows commercial banks to infer two signals
The final step in deriving a discretion ary equilibrium under secrecy is to verify that the conjectured decision rule mini mizes the trading desk’s loss function sub ject to the conjectured expectations equa tion. Substituting (9) and (10) into (4) and taking the derivative with respect to _{R}_{,} gives the Euler conditions for this prob lem, which can be rewritten as follows:
and
where tf denotes the deposit rate and Qil denotes all variables dated t1 or ear lier. The linear least squares projection of wt on 4and s; is
(It will be useful to know that 0 <dp, <1 in what follows.) Under secrecy, commer cial banks’ perception of the current inno vation to the reserve target E[y,lZ;] is a function of current reserves, _{R}_{,}_{.} The Trad ing Desk’s current choice of total reserves influences the federal funds rate, which in fluences the commercial banks’ perception of the innovation to the reserve target, and, hence, influences the commercial banks’ forecast of total reserves in period t+l, that is,
for t = T, ~+1, . Equation (11)is the best response to the commercial banks’ expec tation equations and it has the conjectured functional form of the decision rule used to derive the expectations equation: b,
_{e}_{q}_{u}_{a}_{l}_{s} h[p1 (1alPp2d)
bpd/(lh)
and bl
equals
positive, the analysis predicts a positive reserve bias under secrecy. Comparing the Trading Desk‘s decision rule under disclo sure, (7), with its decision rule under se crecy, (ll),reveals that b, = al and b, = uo
1. Since it can be shown that b, is
 a1P PZPldh/(W. The term %P P2Pldh
/(l/z) is positive and reflects the influence of current reserves on expected future re serves under secrecy. Since _{a}_{l}_{p}_{p}_{2}_{p}_{l}_{d}_{h} /(lh) is positive, the analysis predicts that secrecy reduces the upward bias in the Trading Desk’s choice of total reserves. A reasonable conjecture, given the smaller bias under secrecy, is that the Trading Desk would prefer to keep the reserve target secret. To confirm this con jecture, calculate the value of the Trading Desk’s loss function under secrecy. Substi tuting the stochastic process for the re serve target, (l),the forecasting equations (9) and (lo), and the Trading Desk’s deci sion rule, (ll),into the constrained loss function, equation (4)and after some al gebragives
376
ECONOMIC INQUIRY
where S; is the value of the Trading Desk's loss function under disclosure. Inspection of equation (12) reveals that S: < S:, be cause b<l and alp,pf/2<1. Conse quently, the value of the Trading Desk's loss function is smaller under secrecy. A concern for the level of interest rates results in the Trading Desk choosing ac tual reserves greater than target reserves. The size of this reserve bias depends on the sensitivity of the federal funds rate to the current level of reserves. Because com mercial banks cannot perfectly distinguish between changes in reserves and random shocks to the federal funds market, se crecy reduces the sensitivity of the federal funds rate to the current level of reserves. Hence, secrecy lowers the marginal inter est rate cost of moving towards the Trad ing Desk's reserve target and, thus, lowers the upward bias in actual reserves. The Trading Desk values secrecy because it reduces the influence of its monetary con trol policy on interest rates. Specifically, the Trading Desk does bet ter under a regime of secrecy because the resulting bivariate signal extraction prob lem reduces the elasticity of the federal funds rate with respect to current total reserves, R,. Substituting the relevant ex pectations function and decision rule into the federal funds rate equation, _{(}_{3}_{)}_{,} gives the reducedform expression for the funds rate under disclosure and secrecy. The sensitivity of the funds rate to current total reserves under disclosure and secrecy is
The funds rate is less sensitive to total re
serves under secrecy. This result depends on three features of the analysis. First, the reserve targets must be positively au
tocorrelated, that
is, _{a}_{,}_{>} 0. Second, the
commercial banks must solve a dynamic decision problem, that is, _{p}_{2}_{>} 0. Third, the
current reserve target must be secret, that
is, (6>0.
Since the funds rate is less sensitive to total reserves under secrecy, a given bias in the supply of total reserves is associated with a lower interest rate. When the objec tives of the Trading Desk include a con cern for the level of interest rates, _{h} _{>} 0, secrecy lowers the value of its loss func tion. In this analysis, the Trading Desk values secrecy because _{i}_{t} reduces the in fluence of its monetary control policy on interest rates.
V. WAS M1 ABOVE TARGET ON AVERAGE
FROM 1978 TO 19857
The previous sections provide a simple analysis of the Trading Desk's Monetary Control problem. The analysis has ab stracted from many considerations some of which may be important. This section examines actual U.S. experience with monetary control to determine if the pre dicted bias is observed. The M1 targets are constructed from the M1 target ranges contained in the Domes tic Policy Directive reported in the Federal Reserve Bulletin as follows. First, calculate the midpoint of the _{M}_{1} target ranges to obtain an annual growth rate target. Sec ond, convert the annual growth rate target to growth rate target for the intermeeting period. Third, obtain the money supply figures available to the FOMC at the time of their meeting7 Fourth, calculateusing continuous compounding the M1 target
7. The authors thank Richard Sheehan for provid
ing this data.
COSIMANO & VAN HUYCK: CENTRAL BANK SECRECY
377
consistent with the best available figure for M1 and the intermeeting growth rate target. Table I reports the constructed time series for the M1 target, the actual value of M1, the difference between the actual and target value of M1,and the percentage difference. While M1 is not uniformly above target, the mean percentage difference is 0.2 per cent over an intermeeting period. The null hypothesis that the mean percentage dif ference is not positive generates a tstatis tic of 1.71, which exceeds the critical value of 1.658 at the 5 percent level of statistical significance. Given an average of eight meetings a year, the mean percentage dif ference between M1 and the midpoint of the ranges for M1 contained in the Domes tic Policy Directive over the period _{1}_{9}_{7}_{8} to 1985 was 1.6 percent per year. Hence, the data does not contradict the model’s pre diction that a concern for the level of interest rates, which is also expressed in the Domestic Policy Directive over this period, biases monetary aggregates above the relevant monetary targets.
_{V}_{I}_{.} CENTRAL BANK SECRECY AND THE CDMMERCIAL BANKING INDUSTRY
While something like the federal funds rate equation used in our analysis could be consistent with a number of commer cial banking models, this section derives it from an explicit model. This will not only allow us to derive the federal funds rate equation and formalize the signal extraction problem, but also allow us to formalize secrecy’s influence on the value of the commercial banks. The banking industry is composed of n competitive commercial banks. The ith commercial bank chooses deposits, Di, loans, Li, and federal funds, 4, to maxi mize the present value of expected profits,
Vi:
where p is one over one plus the bank’s return on equity, 6is the loan rate, $ is the deposit rate, pi is the random compo nent of the marginal cost of deposits, and all C~Sare positive parameters. Banks hire labor and capital to provide liquidity services to their depositors.s Hence, the marginal cost _{o}_{f} deposits de pends on the cost of these resources. The quadratic term in the level of deposits represents the increasing cost _{o}_{f} providing liquidity with fixed banking locations and check clearing facilities. The marginal cost of deposits also contains a random com ponent, pf, which represents both bank specificand industry wide shocks. (Notice that without these cost shocks the com mercial banks could infer the current in novation in the reserve target from the funds rate and deposit rate signals and, hence, we would not have a meaningful distinction between disclosure and se crecy.) The value function is linear in rev enue from loans and federal funds. The model’s dynamics arise from the quadratic cost of adjustment term in equa tion (14).The adjustment term could rep resent the cost of adjusting labor and cap ital to provide changing levels of liquidity (see Sargent [1987] on labor and capital adjustment costs) or the adjustment term could represent transaction specific in vestment in customer relationships. _{I}_{f} a bank wants to attract a customer from a competitor, it must pay a portion of the cost of setting up a customer relationship. This cost includes the loss of interest in come on float during the change, evaluat
8. See Saving [19n], Cosimano [1987], or Sweeney
[1988].
_{3}_{7}_{8}
ECONOMIC INQUIRY
TABLE I M1 Target Derived from Domestic Policy Directive and Actual M1: 19781985
Date of 
Percent 

Meeting 
M1 Target 
Actual M1 
ActualTarget 
Difference 
Jan. 17,1978 
341.53 
336.10 
5.43 
1.6% 
Feb. 28,1978 
336.78 
335.90 
0.88 
0.3 
Mar. 21, 1978 
337.45 
341.70 
4.25 
1.3 
Apr. 18,1978 
343.34 
350.50 
7.16 
2.1 
May 16,1978 
352.35 
349.70 
2.65 
0.7 
June 20,1978 
351.79 
354.10 
2.31 
0.7 
July 18, 1978 
355.68 
353.10 
2.58 
0.7 
Aug. 15,1978 
355.14 
357.30 
2.16 
0.6 
Sep. 19, 1978 
359.22 
359.90 
0.68 
0.2 
Oct. 17, 1978 
362.15 
362.10 
0.05 
0.0 
Nov. 21, 1978 
363.49 
360.70 
2.79 
0.8 
Dec. 19, 1978 
362.64 
357.60 
5.04 
1.4 
Feb. 06,1979 
359.66 
360.10 
0.44 
0.1 
Mar. 20, 1979 
361.76 
359.40 
2.36 
0.6 
Apr. 17,1979 
361.47 
364.90 
3.43 
0.9 
May 22,1979 
366.13 
368.60 
2.47 
0.7 
July 11,1979 
370.19 
372.30 
2.11 
0.6 
Aug. 14, 1979 
374.45 
376.90 
2.45 
0.7 
Sep. 18,1979 
378.09 
377.80 
0.29 
0.1 
Oct. 06, 1979 
379.76 
379.20 
0.56 
0.1 
Nov. 20, 1979* 
382.07 
381.70 
0.37 
0.1 
Jan. 09, 1980 
383.35 
389.30 
5.95 
1.6 
Feb. 05,1980 
391.17 
391.30 
0.13 
0.0 
Mar. 18, 1980 
393.18 
390.40 
2.78 
0.7 
Apr. 22,1980 
391.90 
388.60 
3.30 
0.8 
May 20,1980 
392.90 
391.60 
1.30 
0.3 
July 09,1980 
394.35 
394.30 
0.05 
0.0 
Aug. 12,1980 
397.71 
403.00 
5.29 
1.3 
Sep. 16,1980 
405.51 
409.70 
4.19 
1.o 
Oct. 21, 1980 
411.49 
411.90 
0.41 
0.1 
Nov. 18, 1980 
413.65 
412.40 
1.25 
0.3 
Dec. 19, 1980 
414.87 
416.40 
1.53 
0.4 
Feb. 03,1981 
419.97 
420.60 
0.63 
0.2 
Mar. 31, 1981 
423.72 
430.90 
7.18 
1.7 
May 18, 1981 
432.75 
429.00 
3.75 
0.9 
July 07, 1981 
432.22 
433.70 
1.48 
0.3 
Aug. 18, 1981 
437.79 
431.70 
6.09 
1.4 
Oct. 06, 1981 
435.19 
433.20 
1.99 
0.5 
Nov. 18, 1981 
436.12 
440.60 
4.48 
1.0 
Dec. 22, 1981 
442.89 
450.50 
7.61 
1.7 
Feb. 01,1982 
450.50 
449.00 
1.50 
0.3 
COSIMANO & VAN HUYCK: CENTRAL BANK SECRECY
379
TABLE I continued
M1 Target Derived from Domestic Policy Directive
_{a}_{n}_{d} _{A}_{c}_{t}_{u}_{a}_{l} M1: 19781985
Date of 
Percent 

Meeting 
M1 Target 
Actual M1 
ActualTarget 
Difference 
Mar. 28,1982 
450.81 
449.90 
0.91 
0.2 
May 18,1982 
451.46 
452.50 
1.04 
0.2 
June 30,1982 
455.98 
453.40 
2.58 
0.6 
Aug. 24,1982 
456.02 
461.00 
4.98 
1.1 
Nov. 15,1982‘ 
 
 
  

Dec. 21, 1982 
 
 
  

Feb. 09,1983 
 
 
  

Mar. 28, 1983 
499.39 
506.90 
7.51 
1.5 
May 24,1983 
511.34 
508.30 
3.04 
0.6 
July 12, 1983 
512.41 
516.90 
4.49 
0.9 
Aug. 23,1983 
521.08 
517.80 
3.28 
0.6 
Oct. 04, 1983 
522.29 
515.70 
6.59 
1.3 
Nov. 15,1983 
518.19 
522.70 
4.51 
0.9 
Dec. 20,1983 
526.32 
523.60 
2.72 
0.5 
Jan.31, 1984 
529.14 
536.10 
6.96 
1.3 
Mar. 26, 1984 
541.46 
539.40 
2.06 
0.4 
May 22, 1984 
544.79 
544.60 
0.19 
0.0 
July 16, 1984 
547.47 
547.30 
0.17 
0.0 
Aug. 21,1984 
550.45 
549.60 
0.85 
0.1 
Oct. 02, 1984 
552.76 
544.70 
8.06 
1.5 
Nov. 07,1984 
546.58 
547.10 
0.52 
0.1 
Dec. 17, 1984 
552.99 
559.60 
6.61 
1.2 
Feb. 12,1985 
564.78 
570.60 
5.82 
1.0 
Mar. 26, 1985 
575.88 
577.60 
1.72 
0.3 
May 21,1985 
582.94 
596.00 
13.06 
2.2 
July 10, 1985 
599.15 
601.90 
2.75 
0.5 
Aug. 20,1985 
607.82 
610.40 
2.58 
0.4 
Oct. 01, 1985 
614.22 
613.60 
0.62 
0.1 
Nov. 04, 1985* 
617.85 
626.10 
8.25 
1.3 
*The definition of M1 changes between the Nov. 20,1979 and Jan.9,1980 meetings. The directive does not contain M1 target ranges for the following meetings: Nov. 15, 1982; Dec. 21, 1982; Feb. 9,1983. The FOMC ceased including M1 targets after the May 20, 1986 meeting.
Source: Federal Reseive Bulletin, various issues.
ing the cost and benefits of doing business with a new bank, and learning the loca tions, hours, and procedures of the new bank. Additionally, it is costly to establish the new customer’s account. Flannery [1982]presents evidence that deposit ad
justment costs are an important source of dynamics. The loan market will only be used to close the model in order to calculate the value of commercial banks under disclo sure and secrecy. The cost of loans in
380
ECONOMIC INQUIRY
creases at an increasing rate, which repre sents the resource cost of evaluating and monitoring loans. For simplicity, the anal ysis assumes that loan demand is inelas
tic.9
Assume that a bank holds neither ex cess nor borrowed reserves. Relaxing this assumption merely obscures the analysis of secrecy in this paper without contribut ing insight. Consequently, the balance sheet constraint of a bank reduces to Fi = (1O)Di  Li, where 8 is the required reserve ratio. The Euler equation for a bank's deposit demand is found by substituting the bal ance sheet constraint into equation _{(}_{1}_{4}_{)} and taking the derivative with respect to D:. Let D, denote the sum of the deposits supplied by the n banks. Summing across the n Euler equations gives the aggregate deposit demand:
_{(}_{1}_{5}_{)}
D, = (n(lO)f,  nt+ c2D,,
where u, is the sum of the pi and is serially uncorrelated with mean zero and finite
variance.10
To complete the market for deposits, assume that the aggregate supply of de posits depends linearly on the spread be tween the federal funds rate and the de
posit rate and on a random shock to de posits:
where mo and ml are positive parameters, and E, is a serially uncorrelated random variable with mean zero and finite vari ance.,' The deposit rate can be found by using the deposit demand equation (15) and the deposit supply equation (16) to solve out for current deposits, which gives
_{(}_{1}_{7}_{)}
where
= (1ndO)f, + dc2D,l + dc2PE;Dt+,
d = 1/ [ n + [cl + c2(l+p)]ml 1.
The equilibrium condition for the fed eral funds market is obtained by adding together the balance sheet constraints of the individual banks and by noting that L,  D, equals the total reserves deter mined by the Trading Desk, R,. The federal funds market clears when OD,= R,. Together, the equilibrium conditions for the deposit market and the federal funds market yield the federal funds rate that simultaneously clears the federal funds and deposit market, which was denoted equation (3) in section 11:
9. See Cosimano [1988] for a more developed model of the loan market. 10. The signal extraction problem arises because commercial banks do not observe PI. If was observ able, then commercial banks could infer the current innovation to the reserve target from the deposit rate and the federal funds rate. In their analysis of the FOMC's decision problem, Cukierman and Meltzer [1986]and Lewis [1991]argue that imperfect monetary control causes private agents to solve a signal extrac tion problem. Interestingly, the Trading Desk has no incentive to create ambiguity through imperfect mon etary control in order to hide the current reserve target. In fact, it is possible to show that the Trading Desk prefers perfect monetary control in the framework of this paper.
where
+ p2R,,
 u, + e,
11. Our analysis focuses on the Trading Desk's short term monetary control problem. _{A} more complete model would include the influence of the price level on deposit demand and supply. However, we do not believe that these long term considerations will elim inate the effect of secrecy illustrated by the analysis.
COSIMANO & VAN HUYCK: CENTRAL BANK SECRECY
381
Deriving equation (3) from a choicethe oretic analysis of the commercial banking industry allows us to determine the rela tive magnitudes of the parameters. _{F}_{o}_{r} ex ample, we know that _{p}_{1} is less than one and greater than p2 as was claimed. By substituting the Trading Desk's de cision rule, the market clearing conditions, the balance sheet condition, and the first order conditions into the ith bank's value function, (14),it is possible to calculate the value of a commercial bank, which under disclosure is
equation (18) and (19).Notice that because p,pf <1 this second influence increases a bank's value. Under both regimes the vari
ance of the reserve target increases the cost of deposit adjustment and, hence, lowers
the value
However, secrecy increases the covariance between deposits and the spread between the loan rate and the deposit rate, which increases the value of a commercial bank. When the Trading Desk has little con cern for the level of the federal funds rateh is smallthen the steady state levels of reserves under disclosure and secrecy are nearly equal and the covari ance influence of secrecy dominates. Hence, for reasonable parameter values central bank secrecy can increase the value of commercial banks. Consequently, it is possible that both the central bank and commercial banks prefer secrecy to disclo sure.
of a commercial bank: c2 _{>} 0.
and under secrecy is
(19) E<
= [ ii + cl[%/(l~) + b,I2/2e2
where SE denotes those terms that do not change with the information regime. In general, it is not possible to order the relative value of the ith bank under disclo sure versus secrecy. The information re gime influences the value of a commercial bank for two reasons. First, the Trading Desk chooses a smaller steady state level of total reserves under secrecy, b, _{<} _{u}_{,}_{,} which reduces a bank's value; see the middle term of the righthand side of equations (18) and (19). Second, secrecy increases the covariance between deposits and the spread between the loan rate and the deposit rate. This second more subtle influence of secrecy on a bank's value is reflected in the final term of the righthand side of
APPENDIX
This appendix derives equation (lo), which is the optimal forecast of yt based on the cur rent information available to the commercial banks under secrecy: the deposit rate and the
funds rate. Let
= fi,#,Qil), where Qf1 de
notes all variables dated _{t}_{}_{1} _{o}_{r} earlier. By equa tion (3) the federal funds rate allows commer cial banks to infer the following signal:
and by equation (17) the deposit rate allows commercial banks to infer the following signal:
(A2)
s';= 8 E[8 I Qi11 =ylut + Y2et,
where y1 = nde and y2 = (1nd)e.
The linear least squares projection of yt on 4 and s; is
_{(}_{'}_{4}_{3}_{)}
E[ytId] = d4  cp'd,
_{3}_{8}_{2}
ECONOMIC INQUIRY
where
and
See Dotsey and King (1986, 411 on the solution of bivariate signal extraction problems. Equa tion (10) in the text follows from substituting (Al) and (A2) into (A3).
REFERENCES
Barro, Robert J., and David Gordon. "A Positive The ory of Monetary Policy in a Natural Rate Model." Journal of Political Economy, August 1983,589610. Cosimano, Thomas F. "The Federal Funds Market
under Bank Deregulation."
Journal of Money,
Credit, and Banking, August 1987, 32639. "The Banking Industry Under Uncertain Mon etary Policy." Journal of Banking and Finance, March 1988, 11739. Cosimano, Thomas F,, and John 8. Van Huyck. "Dy namic Monetary Control and Interest Rate Stabi lization." Journal of Monetay Economics, January 1989, 5363. "Central Bank Secrecy,Interest Rates, and Mon etary Control." Manuscript, April _{1}_{9}_{9}_{1}_{.} Cukierman, Alex, and Allan H. Meltzer. "A Theory of Ambiguity, Credibility, and Inflation under Dis cretion and Asymmetric Information." Econometrica, September 1986,10991128. Dotsey, Michael. "Monetary Policy, Secrecy, and Fed eral Funds Rate Behavior." Journal of Monetary Economics, December 1987, 46374. Dotsey, Michael, and Robert G. King. "Information Im plications of Interest Rate Rules." American Eco nomic Review, March 1986,3342. Federal Reserve Bulletin. Various issues.
.
.
Flannery, Mark J. "Retail Bank Deposits as Quasi Fixed Factors of Production." American Economic Review, June 1982, 52736. Goodfriend, Marvin. "Monetary Mystique: Secrecy and Central Banking." Journal of Monetay Eco nomics, January 1986, 6396. Hakansson, Nils H., J. Gregory Kunkel, and James A. Ohlson. "Sufficient and Necessary Conditions for Information to Have Social Value." Journal of Fi nance, December 1982, 116981. Kane, Edward J. "Politics and Fed Policymaking: The More Things Change the More They Remain the Same." Journal of Monetary Economics, April 1980,
199211.
Karamouzis, Nicolas, and Raymond Lombra. "Federal Reserve Policymaking: An Overview and Analy sis of the Policy Process." CarnegieRochester Conference Series on Public Policy, 30, 1989. Lewis, Karen K. "Why Doesn't Society Minimize Cen tral Bank Secrecy?" Economic Inquiry, 29(3), Oc tober 1991, 40315. Lombra, Raymond, and Michael Moran. "Policy Ad vice and Policymaking at the Federal Reserve," inMonetary Institutions and the Policy Process, Car negieRochester Conference Series on Public Pol icy, 13, 1980. Lombra, Raymond, and Frederick Struble. "Monetary Aggregate Targets and the Volatility of Interest Rates." journal of Money, Credit, and Banking, Au gust 1979, 284300. Mayer, Thomas. Disclosing Monetary Policy. Unpub lished monograph, 1987. McCallum, Bennett T. "On the Consequences and Crit icism of Monetary Targeting." jourrial 01Money, Credit, and Banking, November 1985, 57097. Merrill, D. R., Plaintiff v. Federal Open Market Com mittee of the Federal Reserve System, Defendant. U. S. District Court for the District of Columbia, Civil Action no. 750736, March 1976. Rudin, Jeremy R. "Central Bank Secrecy, 'Fed Watching', and the Predictability of Interest Rates." journal of Monetary Economics, September 1988, 31734.
Sargent, Thomas J. Macroeconomic Theory, 2nd ed. Bos ton: Academic Press, 1987.
Saving, Thomas R. "A Theory of the Money Supply with Competitive Banking." Journal of Monetary Economics, July 1977, 289303.
Sweeney,Richard J. Wealth Effects and Monetary Theory. New York: Basil Blackwell, 1988.
Tabellini,Guido. "Secrecy of Monetary Policy and the Variability of Interest Rates." jourrial of Money, Credit, nnd Banking, November 1987, 42536.
Woolley, John T. Monetary Politics. Cambridge: Cam bridge University Press, 1984.
Гораздо больше, чем просто документы.
Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.
Отменить можно в любой момент.