Research " Monetary Policy , the Banking System, and Short -term Money Instrusments " pot

Hamilton, Chair Chapter I examines two countries, the US and Japan and make some comparisons in the following four topics: 1 money markets and reserve systems, 2 practices of market ope

UNIVERSITY OF CALIFORNIA, SAN DIEGO

Monetary Policy, the Banking System, and Short-term Money Instruments

A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in

2000

Beil & Howell information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, Mi 48106-1346

iv

To my parents, Koichiro and Keiko

and

To Noriko

P.jv 0x Tố ẻẻee iv

8 ăt, mẽ .Ặ nể vi ID Ầ e viii Acknowledgmentts ccccccccsecccecssscccecccecessesscerensnesseesseebspessvessneses ix

Vita, and Fields of Study ccccecscsecscsccecescncescsscecscesceceseeseesessnenes xi ve

À0 xi

An Institutional Comparison of Money Markets and Market Operations Between the

US and Ïapan HQ nọ HH mm nh ni nh my nh nen l

2 Features about short-term money markets and reserve systems .+ 3 2.1 Money markets overall .cccscoccccecscscscsceccrscvcscscnessenseeesees 3 2.2 The Federal Funds market (US) and the call market (Japan) 5

- 2.3 Reserve SystOms s.secrccscscsvcnscscecesescccecesescrerscessensscnenes 7

3 Practices of the daily implementation of monetary policies:

the Fed and the BOJ .- - Q1 nh mm 10 3.1 Variation of market operation instruments

and the discount window lending - - se 10 3.2 Frequency of market operations _ l§ 3.3 A day in market operation desks - óc - nen se 17

4 Objectives of the daily implementation of monetary

policies .- -_ -QQ nọ KỲ kh tin hp 20 4.1 Relation between the market operations and other variables 20 4.2 Operating targets .ccsscccsccssscevevcecresersceseeesseeronsesssenees 24

5 Implications on the liquidity effect .-< 26

› ƯA" 33 Measuring the Liquidity Effect: A Comparison Between the US and Japan 34

L Introduction .:.csecseccscscssccecesesececceseecnssecsscnsensenssesssenenes 35

2 Institutional Background .-.ccscesccecsecsvescccssccscnesersnsesssoess 40

3 Estimation Methodology .-.csescssecsccsccscceecssssrsesenereceeeees 44 (1) Using instrumental variables {e2 44 (2) Are these good instruments? Ăn 45 (i) Correlation between the instruments and the reserves 45 (ii) Exogeneity with respect to reserVes 46

4 Estimation ResulÌts -Q BH nh kh nh nh nh tư 48 (1) The model of NBR (US) and TR (Japan) - - 49 () A basic model - -QQ QQ Q1 n1 ni 49 (ii) A model with responses of NBR(TR) to previous day’s error in

forecasting exogenous variables 51

(iii) A model with anticipated fluctuations in the exogenous variables 53 (2) The model of change in the Federal Funds rate (the call rate) 54 (i) Arbasic model nh e 54 (ii) A model with anticipated changes in the exogenous variables 55 (3) Quality of forecasts by the BOJ .c.ccscscesevcsevereceseerseeeneees 57

5 Comparison ofthe liquidity effect Q Quy 60

Monetary POlicy .ccseccrcsserscccrerecnevcesesensenseeseneneeness 93

$ Conclusion ee 96

Ref€r€nc€s - - QQ no HH HH ni Hi ni Ki nh KH i09 102

Chapter I

Table 2.1: Descriptions and size of instruments in the short-term money markets 4 Table 2.2: Reserve Systems .csccsccesceececcccsceccccnceeccescscneceesesscesscecessences 9 Table 3.1: Maturity and characteristics of each instrument - ll Table 3.2: Frequency of implementation of each market operation instrument 15 Table 3.3: Time schedule of market operations_ -. nen 19 Table 4.1: Balance sheet of the central banks .ccssessececceccrecesceseeceeceeee 20 Chapter IT

Table 2.1: The balance sheet for the BOJ, 30" of April, 999 5.522 2< << 65

Table 2.2: Contents of BOJ’s daily report .ccecccsccssscrsceecceseceeesecereeeceeee 65 Table 2.3: Relation between the BOJ balance sheet items and the three categories

Affecting FESEFVES .ccsceccnsecceecccercescecencceeccecceescenescerscescescesees 66 Table 2.4: Elements included in the treasury and others factor in Japan 66 Table 3.1: Correlation between the prediction errors and reserV€S 66 Table 4.1: Modeling differences .ccesecossceccecececcescescescescesceescreesereeseees 67 Table 4.2: Estimates of the basic modeÌ_ - nen ca 68 Table 4.3: Estimates for the prediction error in the previous đay 69 Table 4.4: Estimates for the three kinds of variables - - 70 Table 4.5: Estimates of the basic modeÌ - - con nn ng ng ng 71 Table 4.6: Estimates for three clustered date dummies with the prediction errors 72 Table 4.7: Estimates for the anticipated changes in the exogenous variables 72 Table 4.8: Estimates for the unanticipated and the anticipated changes

in the exogenous variables .ccecccesescceccceccessscscssccceseeecerersens 73 Table 4.9: Estimates for the lagged prediction errors .2.ccsscceseeecesececececeeeceees 73 Table 4.10:Estimates for the possible explanatory variables 74 Table 4.11:Comparison of the estimates between the model with the original forecast

and that with the improved - SH kh ve 75 Table 5.1: Semi-elasticity of the interest rate to reserVes 76 Table 5.2: Change of the interest rate to one negative standard deviation shock 76 Chapter III

Table 4.1: Estimates of the predictability - nhe ceee 98 Table 4.2: Estimates of asymmetric predictability - s 99 Table 4.3: Variances of the expected change in the interest rates 99 Table 4.4: Evidence on the Balduzzi, Bertola, and Foresi hypothesis .0000 100

LIST OF FIGURES Chapter I

Figure 3.1: | Borrowed reserves change — .ssssccesececeeceeeseceerscceccsecssceeens 30

Figure 3.2 (a): Banknote factor .cccccscesscesessccescnccccscvccsscccvsccsccecescssnscessee 30 Figure 3.2 (b): Treasury factor .cccceececccecevsvsccesccecsecscsscecsccsececevcecescessens 31

Figure 3.3: Total reserves change -QQ TQ nh nh nu ng nen cưy 32 Figure 5.1: Demand and supply schedules in the reserve market 29 Chapter II

Figure 1.1: | Demand and supply schedules in the reserve market 36 Chapter III

Figure 4.1: | Changes in # across different estimates 101

I am deeply grateful to my advisor, Professor James D Hamilton for his patient guidance, warm encouragement, and enduring support through all stages of the research His keen intuition in monetary economics and vast knowledge in econometrics played a key role in making good research progress Without his presence, this goal could not have been achieved

I am grateful to Professors Wouter Den Haan, Marjorie Flavin, and Takeo Hoshi for their valuable comments and suggestions I also would like to thank Professors Clive Granger, Walter Heller, Bruce Lehmann, Allan Timmerman, and Valerie Ramey for interesting discussions and suggestions

I would like to express my sincere thanks to the late Professor Tsuneo Ishikawa of University of Tokyo, who passed away two years ago at the age of 51 He not only wrote a recommendation letter for me to study here but also showed me how we should study economics during my undergraduate days Without his instruction and encouragement, I would not pursue the doctoral degree in economics

During the detailed investigation of institutional facts and data for the dissertation,

I was helped much by many professionals, including Masahiro Aramaki, Takayuki Asanuma, Doctor Vladimiro Ceci, Professor Gunter Dufey of University of Michigan, Doctor Hiroshi Fujiki, Shigeru Fujita, Kazushige Kamiyama, Tomotake Kurosaki and Manabu Ueno I greatly appreciate their help

It is almost impossible to thank all of the friends whom I met in the last three years Among them, especially, I wish to thank Yoichi Arai, Max Auffhammer, Jeff and Hiromi Bethard, James Brennan, Andrew Caffrey, Jin Seo Cho, Michael Davis, Shigeru Fujita, Rafaella Giacomini, Tomoo Inoue, Isao Ishida, Yongil Jeon, Garett Jones, Takayuki Kamae, Dong Heon Kim, Frank Kim, Natalya Lebedeva, Zhigang Li, Simone Manganelli, Andrew Patton, Bradley Paye, Binney Putnam, Piers Redmore, Lorien Rice, Sivan Ritz, Kevin Sheppard, Makoto Shimoji, Ricardo Suganuma,

Yueting Tong, Michael and Lisa Wilson, Guy Yamashiro, and Zhiwei Zhang

I am also grateful to Guy Yamashiro, Lynn Fisher, and Professor Margaret Loken for proofreading and editing this dissertation

Finally, I would like to thank my parents, Koichiro and Keiko Uesugi and my fiancée-to-be, Noriko Matsunami for their warm and continuous support Without

them, I could not have survived here

Chapter IT has been submitted to the Journal of Money, Credit, and Banking and I

am the sole author of the submitted paper

Ph.D., University of California, San Diego

FIELDS OF STUDY

Major Field: Economics

Studies in Monetary Economics

Professors James D Hamilton, Wouter Den Haan, Marjorie Flavin, and Takeo Hoshi

Studies in Econometrics

Professors Clive W.J Granger, James D Hamilton

ABSTRACT OF THE DISSERTATION

Monetary Policy, the Banking System, and Short-term Money Instruments

by lichiro Uesugi

Doctor of Philosophy in Economics University of California, San Diego, 2000

Professor James D Hamilton, Chair

Chapter I examines two countries, the US and Japan and make some comparisons in the following four topics: (1) money markets and reserve systems, (2) practices of market operations and discount window lending by the central banks, (3) objectives of market operations, and (4) implications of institutional differences and similarities on the liquidity effect

Based on the above institutional investigation, Chapter II measures the liquidity effect in Japan, and compares it with the US Since the institutional features are similar across these

countries, we apply Hamilton’s (1997,1998) methodology to obtain the estimates In addition,

the detailed daily data supplied by the Bank of Japan enables us to obtain more accurate

in these countries that the liquidity effect is larger and more statistically significant toward the end of the period

Chapter III revisits the relationship between the forward interest rate and the spot interest rate at the shortest maturities This set of interest rates has not been fully utilized in the literature We demonstrate two kinds of asymmetric predictability of the forward interest rates The first asymmetry occurs when the forward rate minus the current spot rate is positive or negative The other asymmetry occurs between the two periods: periods right after a policy change and the following period up to the next policy change The varying term premium and erroneous anticipation of future monetary policy may explain each of these asymmetries, respectively.

Chapter I

An Institutional Comparison of Money Markets and Market Operations Between the US and Japan

To understand the effects of monetary policy in a country, we have to know

institutional differences in financial markets, practices of monetary policy, and so forth between countries Without this knowledge we may misinterpret results from econometric models or may misspecify the models themselves

We examine two countries, the US and Japan and make some comparisons in the following four topics: (1) money markets and reserve systems, (2) practices of market operations (including discount window lending) by the central banks, (3) objectives of market operations, and (4) implications of institutional differences on the liquidity effect Each topic is covered in its own section and related to the others The structure

of money markets and reserve systems affects practices of market operations, which are treated in the third section These operation practices are based on how the central banks see money markets and respond to them, which is the topic in the section four Section five discusses the implications on the liquidity effect based on the facts

presented in the first three sections

2 Features about short-term money markets and reserve systems

2.1 Money markets overall

Short-term money markets in this paper are defined as those in which assets have a maturity of less than one year and are traded by market participants The call, bills (Tegata), securities with repurchase agreements (securities with Gensaki), certificate

of deposits (CDs), commercial paper (CP), treasury bills (TBs) and financial bills (FBs) fall into this category in Japan, while the Federal Funds (FF), Repurchase Agreements (RPs), CDs, CP, bankers acceptances (BAs), and TBs do in the US Descriptions and the market size of each instrument are shown in Table 2.1

When we compare these markets in the US and Japan, there are two major

differences between them First, the size of the markets in Japan as measured by the ratio of outstanding amounts of the traded instruments to GDP is smaller than that in the US (23.9% in Japan compared with 37.6% in the US at the end of 1996) Second, the size of short-term government paper relative to the total outstanding amount of the market is smaller in Japan than that of the US (11.6% in Japan and 27.0% in the US at the end of 1996) Though the outstanding amount of short-term government debt has been recently increasing, private liabilities, especially bank liabilities, such as call money, Tegata, CDs still dominate the short-term money markets in Japan

FF and | FF 1 day Depository institutions 700.3

RPs | TRPs | I day, 2daysto3 | Banks, security dealers, other

months owners of securities,

6 month or more is nonfinancial corporations,

less typical overnments

cD 1 to 6 months or Depository institutions 590.9

ed Nothing for more

than a week

Uncollatera | 1 day

lized Almost nothing is for

more than 4 months

| Tegata 1 week to 3 months Depository institutions 95.3

CD 2 weeks and longer Depository institutions 276.0

CP up to 9 months Financial and business 93.5

enterprises

TB 3 or 6 months Government 111.2

Securities with 5 days to 3 months Banks, securities dealers, 103.0

governments, government- related institutions

* Offshore market and Eurodollar(yen) market are excluded here because they are not primary concern

of central banks which implement market operations only in the above domestic money markets

**Excluding TB and FB transactions with Gensaki Repo market was formally established in April

1996 following the abolition of old regulations for the Repo transactions It is functionally the same as the market for securities with Gensaki in that cash and securities are exchanged for a limited period of time However, Repo market has an advantage over Gensaki market since securities (not TB or FB) transacted in Gensaki market are subject to securities transaction taxes Therefore, the Repo market is expected to replace for the securites with Gensaki market in the near future.

The above features (especially the second one) have a lot to do with the market operation instruments the BOJ has today As we will mention in section 3, the BOJ uses Tegata and CP for market operations in addition to TB while the Fed does not use them as operation instruments This is due to the fact that there is not a core market for the BOJ to make market operations in Japan If there are government short-term bond markets which satisfy the following conditions, (a) appropriate market size, (b)

homogeneity of operational instruments, (c) same-day settlement, and (d) minimal administrative cost, there will be a core market like that in the US However, TB market in Japan lacks condition (a)' Some people criticize that the withholding tax system for the profit from redemption of TB is the obstacle for foreigners to purchase Japanese TBs

2.2 The Federal Funds market (US) and the call market (Japan)

Next, we look at the difference between the Federal Funds market and the call market The interest rates in these markets have the most importance both to the central banks and market participants

' Other instruments have other problems as well

(i) Tegata and CP

Tegata and CP are issued by various private companies Therefore, their creditworthiness is not homogeneous, and additional administrative work is necessary, making them not appropriate as primary tools of operation

(ii) Long-term securities with Gensaki

It is subject to a securities transactions tax Settlements are three days after transactions

Same-day settlement is not ensured.

financial institutions with shortages and those with excess reserves On the contrary in Japan, only transactions with brokers are allowed as transactions in the call market Direct transactions between depository institutions in the FF market most commonly consist of sales by small-to-medium sized institutions to larger correspondent banks A substantial share of large transactions is arranged in the brokers' market Trades

through the brokers are typically for $25 million or more, although smaller trades may

be executed on occasion In 1996, the daily volume of the FF trades arranged through brokers reporting to the Fed of NY averaged around $45 billion (No measure is available of the total volume of the FF transaction, including direct transactions.)

In the FF market, loans are unsecured, while in the call market in Japan there are two types of loans, which are collateralized (secured) and uncollateralized

(unsecured) However, when depository institutions make unsecured loans in either of these two countries, they establish credit limits for each potential buyer In Japan, although uncollateralized loans have dominated the call market recently (e.g 70.8% of total outstanding in the call market at the end of 1998), uncollateralized loans in the call market had been prohibited until 1985 due to historical events in the 1920s After the WWI economic boom in Japan, mainly due to the damages caused by the Great Tokyo earthquake in 1923, many banks had a large amount of bad loans A typical example was the Bank of Taiwan, a government controlled special bank which was established in 1899 after the Japan's acquisition of Taiwan This bank lent heavily to

Suzuki which was on the verge of collapse and was one of the largest concerns in Japan at that time To maintain solvency, the Bank of Taiwan borrowed a large

amount in the uncollateralized call market However, when the BOJ refused to

continue advances without a government guarantee, it failed and stopped the payback

of its call loans for more than three months Following this incident, dealers in the call market established the rules to prohibit the uncollateralized call loan

In both of these markets, ‘term’ transactions whose maturity ranges from a few days to (more than) one year do exist However, the term funds market in the Federal Funds market is considerably smaller than the overnight market; the volume of

activity varies, but the amount of term FF outstanding is probably on the order of one tenth of the amount of overnight funds arranged on a given day For the call market in Japan, overnight transactions dominate the entire call market, having a share of 67.3%

of total outstanding (sum of the collateralized and the uncollateralized) at the end of October, 1997

2.3 Reserve systems

The reserve system is another important factor determining the framework of daily implementation of monetary policy as well as money market structures It is not difficult to see the institutional differences in their reserve systems on Table 2.2 First, the Fed allows banks to carry-over certain portions of the reserve requirement or excess reserves to the next maintenance period, while the BOJ does not When banks are not allowed to have any carry-overs on the final day of a maintenance period, they

that are able to have carry-overs may change the demand for reserves on the final day When they think that the interest rate on the final day is too high, they may leave a portion of reserve requirement to be satisfied in the next maintenance period The reserve demand in this case is more elastic to the interest rate than the case without carry-overs Second, the length of a lag between the period for calculating the reserve requirement and the period for maintaining the required reserves are different between the two countries In the US, the lag is only for two days while it is half a month in Japan Third, applied vault cash is included in the reserve requirement in the US while

it is not in Japan

Aside from the above obvious differences, we briefly note two other points of interest The first is that the excess reserve ratio in the US is much higher than in Japan (1.225% in the US and 0.142% in Japan from 1967-87, 2.542% in the US and 0.412% in Japan from November 1996 to October in 1997) This might come from the carry-over system in the US, which is not allowed in Japan’

The second is that Japanese depository institutions do not seem to recognize the reserve system as partly contemporaneous though it has a half-month overlapping period The preliminary values of total deposits (and the required reserves) at the

? Although there are significantly large excess reserves in the Japanese banking system after late 1997, this is thought to be temporary since the BOJ tries to make the call rate to almost zero percent level by supplying a large amount of finds into the market.

beginning of the reserve maintenance period (16" of a month) and the realized values

in the midst of the period (1* of the next month) do not differ significantly

Table 2.2 Reserve systems

Length of one 2 weeks 1 month

maintenance period (from Thursday to Wednesday two | (from 16" of month to 15" of the

(for the transactional weeks later) *** next month)

deposit)

Treatment of vault cash [| Applied vault cash is included in _| It is not included in required

required reserves reserves

Carry-over Possible (Up to 4% of required Not allowed

reserves)

Reserve-ratio 3-10% (for transaction accounts) 0.05-1.2% (for time deposits)

0.1-1.3% (for others) Penalty rate for the short | Discount rate + 2% Discount rate + 3.75%

of reserves

* From 7/30/1998, C is 16 days prior to M This means that there's no overlapping days between C and M now There is a complete lagged reserve system for the transactional deposits

** From 7/30/1998, C is completely overlapped with M

*** There is no change even after 7/30/1998

3 Practices of the daily implementation of monetary policies: the Fed and the BOJ

The fundamental features in the daily monetary policy implementations are the same in these two countries in that their central banks have used both of discount window lending and market operations to affect the money markets based on their reserve systems However, once we go into details, they are significantly different in what market operation instruments are available, how they see the discount window lending, and so forth These differences come partly from the different money market structures and the different reserve systems we discussed in the second section Based

on this understanding, we make comparisons between the Fed and the BOJ in the way they use monetary policy tools in practice

3.1 Variation of market operation instruments and the discount window lending The Fed implements its market operations mostly based on government securities (Treasury bill, note, or bond) while the BOJ still gives private sector liabilities

(Tegata, CP) large roles in its market operations This is due to the small size of Japanese government securities markets, especially TB market, which was already mentioned in the previous section

Maturity Active instruments Non-active instruments

Overnight Discount lending,

Tegata operation (sales)

TB purchase Less than 3 months Discount lending _———=

Tegata operation (purchase and sales)

TB purchase

CP purchase

FB sales * Repo operation **

Long-term bonds purchase with Gensaki *

More than 3 months Long-term bonds purchase ———_—_=

outright

US

Maturity Active instruments Non-active instruments

15 days and more Outright purchase of bonds Discount lending

* They are not used now (10/99)

** Repo operation: The BOJ borrows bonds from private banks and gives money as collateral to them (from 11/28/97) This has replaced the long-term bonds purchase with Gensaki since the repo is not subject to the securities transaction tax while Gensaki operations are

The next notable difference seems to be about the treatment of discount window lending It is classified as a non-active instrument in the US, in that the Fed passively supplies reserves when private banks come to the discount window On the contrary, the BOJ have attempted to directly control the discount window lending

"(For BOJ lending,) BOJ advances to banks are made and called back solely upon the BOJ's initiative; amounts also determined by the BOJ This makes BOJ lending a very flexible and attractive instrument for controlling the reserve supply in the money market." (Okina 1993, p43)

However, when we look closer at what each of these central banks will do with a reserve shortage in the market especially on the final day, their behavior through the

discount windows looks similar, although there are still several superficial differences between them’

In the US, the Fed waits for private banks to come to the discount window asking for borrowed reserves If private banks have good reasons to come to the discount window, such as an unexpected increase in the loan demand, a sudden drop in

deposits, or a temporary and unexpected difficulty to acquire funds, the Fed has to supply certain amount of borrowed reserves to them But the Fed is not completely passive in giving funds to the private banks It sets credit lines for each bank asking for the loans It does more intense surveillance to the banks with more borrowed reserves, which affects the frequency banks come to the discount window and the amount of borrowed reserves they ask for at the window

In Japan, the BOJ seems to take initiatives by making offers of the discount

window loans to private banks But the BOJ has to collect information about which private bank is in need of additional funds to initiate the discount window loans Once

it comes to know that there is a serious shortage of funds in some banks, it is pretty difficult for the BOJ to completely disregard the situation without making an offer of loans

From the above descriptions, we state that in both of the countries the decision about making discount window loans are firmly based on the needs for reserves of each private bank, which forces the central banks to respond They have to make

3 ‘There is some feature that makes the discount window loans in Japan look more active than those in

the US For example, the BOJ is able to change the interest rate it charges for the discount window loans by changing the term it lends funds It is possible since the BOJ charges interest rates at both ends

of a period fora loan Effective interest rate is higher as the term gets shorter.

13

certain amount of discount window loans to accommodate the reserve demands of each bank whether the central bank makes the first move or not

As stated above, the difference between whether discount window loans are

considered active or not is not ascribed to the difference in the fundamental

characteristics between countries The two countries are the same in that their central banks have to passively respond to the demand for reserves of each bank with the discount window loans to some extent Rather, difference in the treatment of these discount loans seems to come from other reasons, such as, how each central bank has

to rely on the loans

In the US, the Fed has relied heavily upon and has long used the RPs operation that buys government bonds or bills (TBs or FBs) with a resale agreement as an overnight adjustment tool However, until recently in Japan, the discount window loans have been the primary instrument for the BOJ to fine-tune the overnight supply-demand gaps TB in Japan appeared in 1986 but its transaction market is not as fully developed

as the one in the US Fluctuations in borrowed reserves in Japan was much stronger than in the US until the middle of 1995 reflecting the heavy usage of the discount window loans in Japan When we see the change in borrowed reserves from January

1990 to August 1995 on a weekly basis, the fluctuation in Japan is far larger than the one in the US (Figure 3.1) There is no change in borrowed reserves on the first week

of a maintenance period in the US while there are always large changes in Japan The fact that the BOJ has almost nothing other than discount window loans to adjust funds overnight may well be the reason that it considers loans as an active instrument If it

were considered non-active, the BOJ may feel reluctant to use discount window loans

as heavily as it did in the past

However, recent developments in the TB market in Japan make the treatment of discount window loans similar to those in the US After the middle of 1995, the BOJ stopped giving large discount window loans partly due to the situation the discount rate being higher than the overnight call rate, which reduced the demand for the

discount loans to almost zero On January 16, 1996, the BOJ announced that it would | abolish the credit lines for the loans with the recent development of market operations that will substitute for the discount window loans

There was a break in this scheme for a while from November 1997, when several major Japanese banks and securities companies went out of business and the short- term money market was on the brink of malfunctioning forcing the BOJ to lend funds through the discount window But the basic framework of market operations in Japan, which gives priority to the market operations over the discount window loans, has not been changed and become closer to the one in the US

15

3.2 Frequency of market operations

For the frequency of the market operations, temporary operations such as RPs and MSPs are dominant

A more striking point is the difference in the number of operations between the US and Japan The BOJ does more than three times as many operations as the Fed does in one year There are two possible reasons that explain this conspicuous difference

Table 3.2 Frequency of implementation of each market operation instruments

no market operations or discount loans by the central banks So one of the main tasks

of central banks is to counteract the above exogenous factors Therefore, as these exogenous factors fluctuate more, the size and the frequency of market operations by central banks are larger and higher Figure 3.2 is the graph comparing the fluctuations

of the total of the exogenous factors and its two major components (banknote and treasury) on a weekly basis We see in the figures that the BOJ is faced with higher volatility in these factors than the Fed

There are two major causes for the large gaps in fluctuations The first is about the practice of payment that affects the banknote factor It is a common knowledge that the other forms of payment such as cards or personal checks are far more prevailing in the US than in Japan The second is for the treasury factor and is related to the

existence of the system called treasury tax and loan note option (TT&L) in the US The TT&L accounts are held in the private banks by the treasury department and one

of their main functions is to smooth out the fluctuation of the treasury deposit in the Fed What is done on a daily basis is as follows:

Each morning, the Treasury, the New York Reserve Bank, and the Board

staffs evaluate the estimated flows through the Treasury's Fed account The

Treasury may decide to transfer funds to the Fed by making a "call" on the

TT&L accounts if estimates suggest its balance would otherwise be below the target balance or to transfer funds to the TT&L accounts by making a "direct

investment" to the accounts if the estimated balance would otherwise be higher than desired (Meulendyke 1998 p156)

Because of the above ‘buffer’ transferred between the TT&L accounts and the Treasury balance in the Fed, there is less fluctuation than there would be without the transfers Japan does not seem to have the similar system to smooth out the treasury balance in the BOJ Therefore, whether there exists a buffer to cancel out the possible

in the exogenous factors affecting reserve levels, the number and the amount of

market operations may be small if the central bank does not want to smooth out the reserve path Figure 3.3 is the graph comparing the fluctuations of the total reserves Here, we observe that the BOJ tends to smooth out the total reserve path on a weekly basis

Therefore, we state that the BOJ has to make market operations more frequently

than the Fed due to the two reasons mentioned here

3.3 A day in market operation desks

In both of these two central banks, market operation managers implement market operations which have the same-day settlement in the early moming when money markets are active In the US, RPs and MSPs are offered right after the conference call meeting around 10:30, and operation with same-day settlement (Tegata and TB

operations) are offered at 9:20 in Japan

A major difference lies in the way each central bank makes public announcements The BOJ announces information about market condition (prediction, preliminary, and realized) and market operation (preliminary and realized) every day while the Fed does not The BOJ says that it will be crucial for market participants to form

appropriate expectations about the market behavior and that this will prevent

unnecessary fluctuations in the money market

Table 3.3 Time Schedule of Market Operations

JAPAN

9:20 Tegata operation (same day settlement)

TB purchase (same day settlement) *

CP purchase

BOS lending

10:00 Realized fund surplus/shortage, BOS operations

(for the previous day)

Securities outright purchase (4"" day settlement)

Securities with Gensaki purchase (3 day settlement)

TB purchase (3 day settlement)

11:00 Tegata operation (next day settlement)

13:00 < Tegata clearing and domestic exchange settlement >

BO} lending

15:00 < Settlement at 15:00>

BO] lending

17:00 <Final settlement >

17:30 Prediction fund surplus/shortage (for the next day)

Preliminary fund surplus/shortage, BOJ operations (for the day)

US

10:20 The conference call with FOMC representatives

10:30 RPs, MSPs **

Outright purchase (at a variety of times)

(generally 3™ day settlement)

* In Japan, after | 1/28/1997, the BOJ makes additional offers of operations with the same-day settlement around the midday

** For many years, regular entry time was around [1:30 am It was moved to around 10:30 am primarily because the RP market is most active in the morning Sometimes the Desk may operate before the conference call Expectation of collateral shortage and planned early closings of the markets ahead of major holidays led to early entries.

4 Objectives of the daily implementation of monetary policies

In the previous section, we discussed what and how market operations (including discount lending) are implemented in practice It is also useful to discuss the

objectives of monetary policy implementation on a daily basis First, we develop a simple formula for each country to explain the relation between market operations and other variables With these formulas, it is possible for us to better understand what the central banks have in mind when they take actions such as open market operations or discount window loans Then, we move on to discuss the operating target for each central bank and see if the setting of operating targets affects the monetary policy practices

4.1 Relation between the market operations and other variables

First, we discuss the formula the Fed is supposed to have when it does market operations A convenient starting point is the balance sheet of the central banks, shown

in Table 4.1 From the equalities of total assets and liabilities, we have:

BL (discount window lending) R (member-bank deposits)

S (security holdings) CU (currency in circulation)

FL (float) DG (treasury deposits)

OA (other assets) OL (other liabilities)

This is written as:

21

(4.2) NRA =R-BL =(S + FL + OA) - (CU + DG + OL)

where NRA is non-borrowed reserve accounts This NRA is not equal to non borrowed reserves (NBR) since reserves that banks are required to hold against their deposit liabilities may be held as reserve balances in the Fed or as part of the vault cash in the banks Therefore, to obtain total NBR, vault cash used to satisfy reserve requirements (VC) must be added on both sides of the equation

(4.3) NBR = NRA + VC =(S + FL + OA) - (CU+ DG + OL) + VC

The RHS is arranged further to separate the system's securities portfolio from other assets and liabilities

(4.4) NBR = S + (FL + OA) -(CU+ DG + OL) + VC

By taking the first difference, we obtain an identity involving open market operations, defined as OMO = AS:

where RF is reserve factors defined as RF = (FL + OA) - (CU + DG + OL) + VC The Fed derives its objectives for NBR or ANBR by estimating the sum of required and excess reserves which is then adjusted by the assumption of borrowed reserves This sets the 'dynamic' objectives of the FOMC The second term, ARF’, in addition to being volatile and uncertain, is believed by many central banks to be beyond their control in the short run Therefore, it would be best to estimate as precisely as possible the fluctuations in RF and offset them using open-market operations in order to avoid

There is one thing to be noted about VC Banks may satisfy all or part of their reserve requirements during a given reserve maintenance period with vault cash held approximately two weeks earlier However, for those banks whose vault cash exceeds their reserve requirements, only the amount of vault cash equal to their reserve

requirements (called applied vault cash) is included in the definition of total NBR That is, the applied vault cash is uncontrolled reserve factor noted above as VC We know that total vault cash amount (not applied vault cash) is fixed before the

maintenance period begins due to the two weeks lag Nevertheless, during the

maintenance period, the reported data for total vault cash must be adjusted by

removing an estimate of vault cash not applied toward reserve requirements Hence, how much of the total vault cash is to be removed is unknown until total required reserves are calculated after the computation period ends (only two days before the

end of the reserve maintenance)’

Next, we discuss the formula the BOJ is supposed to have starting again from the Table 4.1 Corresponding formulas to (4.4) and (4.5) for the BOJ are:

“ The words "dynamic" and “defensive” operations were used in Meek (1982), for example

* However, after 7/30/98, the computation period and the maintenance period with vault cash are

completely overlapped In this case, we have no difficulty to know exactly the amount of applied vault

cash (VC) before the reserve maintenance period begins Therefore, AVC is already fixed with

certainty in (4.5) and different from other items in RF that is unknown.

23

where RF'= (FL + OA) - (CU + DG + OL)

Some of the differences between the Fed and the BOJ operating procedures are observed here The reserve factor in Japan, RF’ , does not have VC in it while the RF

in the US has VC This is because applied vault cash is not allowed to be used to meet legal reserve requirements in Japan while it is in the US (See Table 2.2)

Another difference between equation (4.5) and (4.7) is that BZ is not subtracted from R to arrive at NBR In fact, the BOJ has never used the concept of NBR in the past This reflects the BOJ's perception that the borrowed reserve (BR) is a

controllable variable (See Section 3.1) However, after the middle of 1995 the BOJ tries not to use the BOJ lending as an instrument to adjust daily fund shortage/surplus

in the market and (4.7) may be modified to explain the recent behavior of the BOJ The BOJ calls the ARF’ term the surplus (or shortage when it is negative) of funds

in the money market The "defensive" operations of the BOJ are directed to offset this term The BOJ devotes considerable effort to estimate this term and reports its

prediction to the public every day Funds are supplied either through the BOJ's

discount window, BL, or by OMO For “defensive" operations, both instruments have been used for a long time

The difference between the total supply of funds from the BOJ and the amount of

"defensive" operations is "dynamic" operations of the BOJ, and this difference

determines the change in reserves

4.2 Operating targets

It is well known that the Fed targeted the FF rate until 1979, NBR from 1979 until

1982, BR from 1982 to 1989 After 1989, the Fed is thought to have returned to

targeting the FF rate, although there has been no announcement about the shift in the target However, it seems that the difference lies more in emphasis than in substance Obviously, the Fed is not able to set targets for reserves on a day-to-day or even month-to-month basis and hit them exactly If the Fed tries this, it would create

enormous volatility in the FF rate Targeting reserves merely means more frequent adjustments of the "dynamic" part in response to the deviations of actual reserves from the targets, and consequently, more fluctuations in the FF rate than in the case of targeting the FF rate The BR targeting which was adopted in 1982 has since virtually transformed to the FF targeting as mentioned above It is due to the looser correlation between the FF rate and BR in the late 1980s

However, this implicit shift in the targeting regime has brought some anomalies in the way the FF rate behaves Since October 1979, market operations by the Fed are offered and implemented only once a day in principle and as a result of this, the

number of operations by the Fed is much smaller than that by the BOJ (We already observed this point in Section 3.2.) This seems to reflect the idea that it does not need

to make multiple market operations in one day provided that its official target is reserves (NBR or BR), but not the FF rate When it targeted the FF rate before

October 1979, it made multiple operations in one day to have the rate within a target range The present situation is that although the Fed actually targets the FF rate, it

25

continues to have the old market operation practice of reserves targeting days by limiting the number of operations in one day Therefore, the FF rate fluctuates much more than the days before 1979 to deviate from the targeting level by a large margin What has been the target of the BOJ's operations? The BOJ never seems to have targeted bank reserves or high-powered money Month-to-month control over bank reserves is almost impossible because of the lagged reserve system and relatively small size of excess reserves (For 1967-87, excess reserves were 1.225% of required reserves in the US and 0.142% in Japan.) Since the mid-1970s, the BOJ has paid attention to the broader monetary aggregates as intermediate targets of monetary policy However, it seems that it has never used information on monetary aggregates

to calculate target levels for bank reserves or the call rate

The BOJ has kept the overnight call rate as a short-term operating target A change

in the stance of monetary policy creates new target levels for the call rates New targets are achieved by "dynamic" operations We have not known the precise targets

of the overnight call rate until recently, but the precise target levels of the call rate have been disclosed since April 1995

5 Implications on the liquidity effect

In the previous sections, we have seen the institutional differences in the money markets or in the implementation of monetary policies between the US and Japan However, our primary concern is about the effect of the monetary policy, especially the liquidity effect Therefore, we discuss the implications of the institutional

differences on the liquidity effect

Basically, the mechanisms to have the liquidity effect are expected to be the same

in these countries, even though there are many institutional differences The most fundamental settings that make the effect possible are:

(1) Multiple days in one maintenance period

(2) Reserve requirement that must be satisfied by the end of a period

Because of these settings in the reserve market, the US and Japan have the same set

of two diagrams of demand and supply schedules for required reserves depending on the date in a maintenance period

On the final day of a period, private banks have to satisfy their reserve

requirements by any means and may bid up the interest rate to acquire necessary funds

to complete their requirements Therefore, the demand schedule is perfectly inelastic

to the interest rate and vertical at the point of the required reserves on the final day In this case, only the central banks have the funds to fulfill the private banks' need for reserves and they charge whatever the interest rate provided that the private banks accept it