Variability of the Japanese money stock

Variability of the Japanese money stock

Economics Letters North-Holland 14 (1984) 241-244 241 VARIABILITY OF THE JAPANESE MONEY STOCK Links to the U.S. and the October 1979 Policy Change ...

200KB Sizes 0 Downloads 39 Views

Economics Letters North-Holland

14 (1984) 241-244

241

VARIABILITY OF THE JAPANESE MONEY STOCK Links to the U.S. and the October 1979 Policy Change

Michael

T. BELONGIA

FedercrlRcwrw Received

of St.

Bunk

* Louis,

St. Louts.

MO 63166,

USA

26 July 1983

McKinnon’s currency substitution model predicts that variable U.S. money growth causes greater variability of foreign money growth. Contrary to this thesis, the variahllity of Japan’s monetary targets is found to be inversely related to the variability of U.S. Ml.

1. Introduction

McKinnon (1982) has argued that the monetary policies of major trading nations respond passively to the pattern of money growth established in the United States. If this is, in fact. the case - and other countries are attempting to peg the value of their currencies to the dollar _ changes in the variability of U.S. monetary aggregates should cause similar increases in the variability of foreign monetary aggregates. This paper explores such a hypothesis in a case study of U.S. and Japanese money stocks. In particular, the issue addressed is whether the variability of Japanese money stock changed after the focus of U.S. monetary policy shifted to aggregate targeting in October 1979.

2. Recent behavior The growth

of the aggregates

rates

of U.S. monetary

aggregates

have

become

more

* Acknowledgements are extended to Milton Friedman for sharing data provided to him by the Bank of Japan and R.W. Hafer for his comments on an earlier draft. The views expressed do not necessarily represent those of the Federal Reserve Bank of St. Louis or the Board of Governors of the Federal Reserve System. All remaining errors are solely the responsibility of the author. 0165.1765/84/$3.00

8) 1984, Elsevier Science Publishers

B.V. (North-Holland)

242

M. T. Belongiu

/

Vartahilrt~~ of monqv

stock

variable since October 1979. From 1970 to October 1979. the standard deviations of month-to-month percentage changes in MI and M2 were 4.21 and 3.99 percent, respectively. Over the more recent period of aggregate targeting ~ from October 1979 through May 1983 - the standard deviation of MI growth was more than double its earlier measure (8.84) while MI’s standard deviation increased to 5.99. Despite this increase in the variability of growth rates for U.S. monetary aggregates, Federal Reserve officials have argued that the U.S. money stock has grown at a more stable rate relative to foreign monetary aggregates [Volcker (19X2)]. This latter claim has been disputed by Friedman (1983), however, who argues that conclusions concerning superior U.S. performance are more related to a statistical artifact than to better monetary control. ’ Friedman’s analysis demonstrates that comparisons of money stock behavior across countries depend not only on the comparability of data series but also on the choice of a particular aggregate. The question that remains to be answered with respect to McKinnon’s model concerns whether changes in the variability of Japanese monetary aggregates are related systematically to changes in the variability of U.S. aggregates.

3. Model and estimation

results

To test this proposition, Granger-type causality tests are applied to a 1973-82 monthly sample of MI for the U.S. and M2 for Japan; these are the aggregate targets in each country. The Japanese data are the monthly averages of daily data employed by Friedman. The null hypothesis is that variability of the monetary aggregate targeted by the U.S. is unrelated to the variability of Japanese monetary target. The measure of variability chosen for estimation purposes is the squared log-differences of each series. Derived by Theil (1967). this ’ Frvzdman’s based

aggregates monthly

to show similarly

was less variable MI

is that

m other averages

expected Using

point

monthly

on one day of each month.

and

foreign of daily

data;

target

data

than monthly

on Japanese

construction

monetary

of the Japanese

IS not parallel

intuitively,

in the U.S. than

the Japanese

MZ in the U.S.

countrtes)

less variation constructed

reported This

to U.S. data

‘smoothed’

averages

a series of selected

indiwdual

averages

in Japan,

M2. Japanese

of daily

a result M2.

data,

which

exhibited

(and

days

data

arc

monetary

are constructed

of daily

would

as be

from each month.

Friedman

not unexpected

however,

aggregates

srrics

showed

since

that

MI

the U.S. targets

less variability

than

M. T. Belongia / Vuricrbibty of moneystock

Table 1 Results of Granger

243

tests.

Null hypothesis

F-statistic

(1) (2)

USA41 + JM2 JM2 + USMI

0.02 1.64”

(1’) (2’)

USMI + JM2 JM2 ---tUSMI

1.05 0.11 d

a Regression F-statistics indicate cantly different from zero.

that the explained

variation

of this model ia not signifi-

measure gives a component of each series’ log-variance. Rows (1) and (2) of table 1 show the results of Granger causality tests applied to U.S. MI and Japanese M2 after being transformed in this manner. Each model includes 12 lagged values of the measures of U.S. and Japanese monetary variability. * The hypothesized path of influence in McKinnon’s model would run from U.S. A41 to Japanese M2. Or, more directly, as the U.S. aggregate varied less widely, attempts by the Japanese to peg the value of the yen would cause their aggregate also to exhibit decreased variability. As this model’s F-test indicates, however, no causal relationship is found. In fact, the results of both Granger tests indicate that the variabilities of the U.S. and Japanese monetary aggregates are unrelated. To investigate the effects of the October 1979 policy change on these relationships, the same models were re-estimated including a dummy variable to mark this split in the sample (models 1’ and 2’). The most interesting result for these models is the suggestion that the change in Fed policy tended to reduce the variability of the Japanese aggregates. Although neither model indicated a causal relationship running between U.S. and Japanese aggregates, the dummy variable in each model was negative and significant at the 0.01 percent level. Contrary to results predicted by a McKinnon model, the variability of Japanese monetary aggregates was lower over a period in which the variability of U.S. aggregates increased.

* The Granger test results appeared to be invariant with respect to choice of log length. Models using six- and eighteen-month lags did not generate results quantitatively or qualitatively different from those reported.

of monq stock

M. T. Belongia / Variohilit_v

244

4. Conclusions

Recent work by McKinnon has argued that increased instability of U.S. monetary policy will cause greater instability in foreign monetary aggregates if these countries attempt to peg the values of their currencies to the dollar. Tests on comparable data for U.S. and Japanese measures of MI and M2, however, show no systematic relationship between the variability of the monetary aggregates targeted by the U.S. and Japan. Further, some evidence suggests that - even if the growth rates of the U.S. and Japanese money stocks were causally related ~ the increased variability of U.S. aggregates since the October 1979 change in Fed operating procedures has tended to decrease the variability of Japanese aggregates.

References Friedman, and

Milton. Banking

McKinnon.

Ronald

standard, Theil,

Housing

Monetary

1.. 1982.

American

Henri.

Volcker.

1983,

Currency

Economic

1967, Economics

Paul. and

variability:

U.S.

and

Japan.

Journal

of Money.

Credit

25. no. 3. forthcoming.

1982. Urban

Letter

and to

Affairs.

substitution

Review inflation

Senator Feb.

18.

and

instability

in the

world

dollar

72, no. 3. 328-337. theory Jake

(North-Holland.

Garn,

chairman.

Amsterdam). Committee

on

Banking.