Displacement effect on government expenditure and government revenue in Japan, 1957-1991-香川大学学術情報リポジトリ

(1)

Kαgawa Universify EconomI(Review

V 01 72， N 0 2， September1999， 367-388

D

i

s

p

l

a

c

e

m

e

n

t

E

f

e

c

t

on Government

E

x

p

e

n

d

i

t

u

r

e

and Government Revenue

i

n

}apan

，

1

9

5

7 -

1

9

1

Abstract

Takeyuki H

i

r

a

i

Masuo N

omura

Since the seminal work by Peacock and Wiseman (1961)， many researchers have examined the question of the existence of the displace -ment effects on government expenditures. Recently， in these studies， the displacement effect is interpreted econometrically as the existence of structural change.. However， very few studies have examined the displace -ment effects on govemment revenues. In this paper， we examine the displacement effects on government revenues and expenditures for the period 1957-1991 in Japan It is shown that the analyses of the displace -ment effects on revenues are also meaningful..

1 . Introduction

Since the seminal work by Peacock and Wiseman (1961)， many researchers have examined the question of the existence of displacement effects on govemment expenditures generated as a consequence of the W orld Wars or the Great Depression O'Hagan (1980) examined the dis -placement effect caused by the subsequent recession after the first oil crisis over the period 1953to 1976， and recently N omura (1991) has examined the

(2)

-368ー Kagawa University EconomたReview 676

displacement effects caused by the subsequent recession after the first oil crisis and the subsequent recession after the second oil crisis

Diamond (1977)， Watt (1978)and Nomura (1991)discussed an inter -pretation of the displacement effect as a structural change" The displace -ment effect on government expenditure is considered as a consequence of change of a number of exogenous factors， that is， tastes， preference， pro -duction technology， institutional arrangements and so on (assumed con -stant for the purpose of empirical testing)" Thus， the displacement effect is a change of structure which is a set of specified parameters by the model of variable GDP (or GNP) related to government expenditure (See Diamond (1977)) As mentioned in Diamond (1977)，羽Tatt(1978)and N omura (1991)， the assumption of equal variances of error terms may be unrealistic“

Many authors have examined the displacement effects on government expenditures Following Peacock and Wiseman (1979)， the notion of toler -able levels of taxation argued that government expenditures reflect both supply influences and demand influencesれ Thesupply influences mean the

raising of government revenues and the demand influences mean the demand by citizens for government services“ They discussed that

econometric tests used in the literature of the displacement effect have tended to concentrate on the demand side of the problem， to the relative neglect of supply" Very few studies have examined the displacement effects on government revenues"

A few types of studies have examined the relationship between the government expenditures and government revenues" In particular， many researchers have examined the intertemporallinkages between government expenditures and revenues， Following Holtz-Eskin， Newey and Rosen

(3)

-Displacement Effect on Government 677 Expenditure and Govemment Revenue in Japan， 1957-1991 369-eses have been advanced: (a) Revenues change concurrently with expenditures.. This situation resem -bles the textbook case in public finance in which voters simultaneously select expenditures and taxes using the standard calculus for weighing marginal benefits and costs (b)Taxes change before spending.. To see how this situation might emerge， consider a government controlled by individuals who want to expand its size beyond that desired by the citizenry In the presence of statutory or constitutional rules prohibiting deficits， public sector man-agers must wait for revenues to increase， and then increase expendi -tures .. (c) Spending changes before taxes. In this situation， taxes gradually adjust to expenditures

In the paper of von Furstenberg， Green and J eong (1986) (hereafter FGJ)， they used vector autoregressions (VARs) to analyze expenditure and revenue data of the federal government for the US Their finding was that spending did not respond to taxes， but there was support for the reverse situation (c) that spending helps to predict taxes.. Holtz-Eskin， Newey and Rosen (1989) (hereafter HNR) used V AR techniques to data from individual local government to study the revenues-expenditures nexus.. They have found that past revenues help to predict current expenditures， but past expenditures do not alter the future path of revenues.. From the studies of FGJ and HNR， there are intertemporallinkages between expend -itures and revenues.. The existence of such linkages suggests that hypothe -sis (a)-revenues change only concurrently with expenditures-is inconect This finding casts doubt upon the interpretation of standard regressions that examine only contemporaneous relationships among fiscal variables“

(4)

-370ー Kagawa UniversiかEconomicReview 678

From the above discussion， the analyses of the displacement effects on the revenues may be different from those on the expenditures， The pur

-pose of this paper is to examine the displacement effects on the revenues and expenditures for the period 1957-1991 inJ apan Our definition of government expenditures include expenditures of general government (cen -tral government， local government and social security funds)

II. The Empirical Model and Statistical Testing

II.A. An Empirical Model

Gupta (1967) devise statistical tests for the displacement effect fitted a double logarithmic function which is linear in logarithms(log-linear) for different sub-periods， before and after major social upheavals， Gupta

(1967)， Nagarajan (1979， 1983) and Nomura (1991) used a log-linear func -tional form， The government expenditure equation is defined by the

equation，

loge

=

al

+

b1logz 十 UI， (1) where e and z denote per capita government expenditure and per capita GDP (or GNP) at constant prices respectively" In this paper， log stands for the natural logarithm of the variable，αIdenotes the regression inter -cept， the slope regression coefficient， bl， is the GDP (or GNP) elasticity of per capita government expenditure with respect to per capita GDP (or GNP)， and UI is the usual stochastic term We modify the government revenue equation used by Thorn (1967) The government revenue equation used in this paper is defined by the equation， logr

=

~

+

bzlogz 十 Zら (2)

where r and z denote per capita government revenue and per capita GDP (or GNP) at constant prices respectively" In this equation， ~ denotes the

(5)

Displacement Effect on Government 679 Expenditure and Government Revenue in Japan， 1957-1991 -371-regression intercept， the slope regression coefficient， b2， is the GDP (or GNP) elasticity of per capita government revenue with respect to per capita GDP (or GNP)， and zら isthe usual stochastic term“ Following Abizadeh and Y ousefi (1988)， the equation (2) can be re-expressed by the following equation : log(R/ Z)

N1

+

1見 Following Nomura (1991)， we introduce the two-stage test of the equality of the full set of regression coefficients after a pre-test for homos-cedasticity in two linear regressions. We can use the Chow test to test null hypothesis，品:β1=β2，against the a!ternative，日 : β1司とβ2. Ifσ12

=

0'22 under the null， the Chow test statistic is distributed as an F distribution with degrees of freedom k and

(6)

680 Kagaωa University Economic Review 372-We can S22 be that of

σ

♂ Lets/be usual estimate ofσ12 and N-2k.. Sl2/s/to test null hypothesis，品。:σ12=σ

乙

use the test statistic凡 = The test statistic凡isdistributed as against the alternative， H01 :σ12ヰσ22 Thus， there anF distribution with degrees of freedom N1 -k and N2 - k Diamond (1977)argued that the assumption of equal error variances can be unrealis -tic in some situations areN1 observations before the displacement and N2 afterwards The Chow test is not robust in the presence of Watt (1979)proposed a Wald test which is valid for heteroscedasticity，

cases of both equal and unequal error variances， and Honda and Ohtani

(1986)proposed a modified Wald test to control the size of the test for small Since the data of government expenditures are small samples， samples"

h t

t i

-l i

l i

-j i

-i -i

g

Nomura (1991)suggested use of the modified Wald test in examining the Following Nomura (1991)， we set the critical values displacement effect of the modified Wald test with reference to a central chi-square distribution withk degrees of freedom under the nulL Following Nomura (1991)， testing procedure for the equality of the full set of regression coefficients in two regressions is therefore as follows， The Chow test statistic is used to test Ho if品。 (σ12=σゾ)is not rejected in the pre-test and the modified Wald test statistic is used to test品 if品。is The modified Wald test statistic will be used in the studies of the rejected displacement effect for heteroscedasticity， since the Chow test is not robust to heteroscedasticity even if it is mild and the sample sizes are small in our examination of the displacement effect II. C. Statistical Estimation of the Structural Chang-e-Point

In studies of the structural change， many researchers have examined However， the In exact position of the change-point may not be clearly indicated a priori the structural change at a given structural change-point

(7)

Displacement Effect on Government

681 Expenditure and GovernmentR巴venuein Japan， 1957-1991

373-this case， the position of the structural change-point should be estimated For this purpose， we use the estimation procedure proposed by Nomura (1991)

II. D. Data Sources and Definition of Variables for Japan

Our definition of government expenditures include expenditures of general government (central government， local government and social security funds). Since government expenditures are sensitive to seasonal fluctuations， many researchers have used annual data in examining the government expenditures.. We wi11 use calendar yearly data .. Transfer payment is defined as follows : transfer payment of general government =social security benefits

+

social assistance grants

+

cunent trans -fers to private non-profit institutions serving households

+

unfund -ed employee welfare benefit We wi11 consider three types of government expenditures and three types of government revenues“ These are defined as fol1ows :

(8)

374 Kagawa University Economic Review 682

former EJ does， and the latterEz does not， involve the direct use of

resources by govemment. Next， total govemment revenueR3 is divided into govemment tax revenue RJ and social security revenue (revenue of social security funds) Rz The former RJ is collected by govemment compulsorily， and the latter Rz is also collected similarly under the social security program. Therefore， we should note that public debt is not included in the total government revenue R3 Note also that interest payment of public debt is not included in the total govemment expenditure f斗 Infact， it is difficult to divide the total expenditure and the total revenue respectively in order that the expenditure may strictly correspond to the revenue according to purpose.. Though the transfer expenditure is financed by a part of the tax revenue as well as by the social security revenue under the above division， we consider that there may be rough correspondences between the govemment purchasesEJ and the tax revenue

RJ， and between the transfer“expenditure ιand the social security revenue

Rz The data for govemment final consumption expenditure at constant price (base year1985)， gross domestic fixed capital formation of general government at constant price (base year1985)， transfer expenditure ι， govemment tax revenue RJ， social security revenue Rz and deflators for private final consumption expenditure (base year1985)are drown from Annual Report on N ational Accounts and data for population is drown from Monthly Report on Current Population Estimates N ote that the transfer expenditure includes transfer payment and subsidies for general govem-ment. We deflate the transfer expenditureι， government tax revenueRJ，

social security revenueRz by using deflators for private final consumption expenditure (See Beck (1979)) Thus， we consider real dataゎ Alldata

(9)

683 Expenditure and Government Revenue in Japan， 1957-1991 -375-NEEDS Database..

III.Tests of the Displacement Effects III. A. Location of Structural Changes In this section， we examine the displacement effects on government expenditure and govemment revenue. As mentioned in previous section， the displacement effect is intepreted econometrically as the existence of structural change.. So， first of all， we must estimate the position of struc -tural change because the exact position of the change-point is not clearly indicated a priori For this purpose， following Nomura (1991)， we succes -sively compute the modified Wald test statistic for all possible time-points and regard the time-point which maximizes a value of the test statistic as a candidate for the location of the change-point. However， it is possible that multiple structural changes may have occurred during the period 1957 -1991 We， therefore， divide the period 1957-1991 into sub-periods: the period 1957-1979 and the period 1974-199L In Table 1， TM W denotes the time-point which presents the maximum value of the modified Wald statistic during the period cosidered in our estimation procedure as stated above.. Furthermore， in the table， the modified Wald， the Chow and the凡teststatistics are shown for each time -point TM W respectively. For the expendituresEJ，

ι

，

ι

， and the revenues Rb Rz， R3' the candidates for the location of the change-point are detected atTM W during the period considered respectively III. B. Existence of Structural Changes for Government Expenditure We examine the displacement effects on govemment expenditure fol -lowing our estimation procedure in sectionII“ From the results for the govemment purchasesEJ in Table 1， we detect a structural change-point

(10)

-376 Kagawa University Ewnomic Review 684 Table 1

Search of the Strucutral Change Time-Points of Government Exprenditur巴and

Revenue Equations for the Periods 1957-1991， 1957-1979 and 1974-1991

Data Period TM W Chow Mod Wald Fo

E1 1957-1991 1959 1.193 205..811a O. OOOb (2，31) (1，30) 1957-1979 1959 0..648 108 161a O. OOOb (2，19) (1，18) 1974-1991 1978 47.684a ₅_8._.₄₆₀a ₁_.₁₃₂ (2，14) (3，11) E2 1957-1991 1973 91.251a 281724a 0.246a (2，31) (15，16) 1957-1979 1973 151. 646a ₁₉₈_.₅₈₂a ₀_.₄₈₉ (2，19) (15，4) 1974-1991 1980 24.195a ₃₉_._.₀₅₃a ₃_._.₉₇₃ (2，l4) (5，9) E3 1957 -1991 1971 26.367α 175..804a ₀_.₀₈₇a (2，31) (13，18) 1957-1979 1974 62.984a ₁₈₆_.₈₆₃a ₁₀_.₄₇₈ (2，19) (16，3) 1974-1991 1978 45.321a ₄₇_._.₀₃₆a ₂_.₇₄₃ (2，14) (3，11) R1 1957-1991 1975 21. 8川α 113.271 a ₇_.₂₂_7α (2，31) (17，14) 1957-1979 1969 4.092b ₁₀_._.₀₆₁_a ₀_.₃₃₅ (2，19) (11，8) 1974-1991 1986 1.299 3.914 10.346 (2，14) (11，3) R2 1957-1991 1974 94.903α 154..952 a ₃_.₄₇₆b (2，31) (16，15) 1957-1979 1974 53.676a ₁₃₉₇₀₀a ₄_.₃₆₁ (2，19) (16，3) 1974-1991 1981 3.630 6.. 648b ₇ 市291 b (2，14) (6，8) R3 1957-1991 1969 76.247a ₁₄₅_._.₂₉₇a ₀_.₇₀₁ (2，31) (11，20) 1957-1979 1969 15.097a _{34..066α0.336} (2，19) (11，8) 1974-1991 1986 L025 2..982 6.313 (2，14) (11，3) Notes:TM W denotes the time-point that the modified Wald test statistic has the

maximum value for the period considered and Mod. Wald denotes the maximum value of the modified Wald test statistic over k <爪 <N -k. Chow denotes value of Chow test sta tistic

a denotes significance at the 1% leγel and b denotes significance at the 5% level The modified Wald test statistic are conducted by reference to a central chi-square distrib叫tionwithk = 2 degr四sof freedom The results ofFotest are obtained by the

(11)

Displacement Effect on Government 685 Expenditure and Government Revenue in Japan， 1957-1991 -377 -1959 for the periods 1957-1991 and 1957-1979 In this case， the test statis -tics凡 revealthe presence of heteroscedasticity and the results of the modified Wald test statistics are significant at the 1% level This struc -tural change might be caused by the recession of 1957-1958 During the period 1957-1974， a transition from the postwar economy to the high growth economy has occurred at the recession of 1957-1958 inJ apan.. For the period 1974-1991， a structural change for the government purchases is detected in 1978 since the test statistic Fo reveals the presence of homos-cedasticity and the result of the Chow test statistic is significant at the 1% level“ This structural change might be caused by the recession which occurred after the second oil crisis. Similarly， from the results for the government transfer expenditure

ι

， we detect a structural change-point 1973 for the periods 1957-1991 and 1957-1979.. In this case， considering that the transfer expenditure is closely related to the social security program， the displacement effect may be caused by both the reformation of social security program in 1973， which is called“fukushi gannen

ぺ

andthe first oil crisis For the period 1974-1991， a structural change for the transfer expenditure is detected in 1980.. This structural change may be caused by the recession after the second oil crisis ..

For the total government expenditure広， three structural changes are detected in 1971， 1974 and 1978 for the periods 1957-1991， 1957-1979 and 1974-1991 respectively. These results may depend on various causes which are also related to the structural changes ofEland E2. We consider that the displacement effect in 1974 was caused by both the first oil crisis and the reformation of the social security program as stated above While the structural change in 1971 might be caused by the recession of 1970， the structural change in 1978 might be caused by the recession after the second oil crisis..

(12)

686

Kagawa UniversilyEconomic Revieω -378ー

III. C. Existence of Structural Changes for Government Revenue In We also examine the displacement effects on government revenue. Table 1， from the results for the government tax revenueRl'we detect a

This structural structural change-point 1969 for the period 1957-1979 For the period 1957-1991， change might be caused by the recession of 1970 We consider that this structural a structural change is detected in 1975 For the period change is caused by the recession after the first oil crisis.. 1974-1991， however， we find that 1986 cannot be considered a point of structural change for the tax revenue because the result of the Chow test is not significant.

i

l

For the social security revenue R2' we detect a structural change-point 1974 for the periods 1957-1991 and 1957-1979 and a structural change-point These structural changes are dated one year 1981 for the period 1974-1991. later than the structural changes for the transfer expenditure ιfor each of So， we consider that， as for the periods consid -the periods respectively ered， the causes of these structural changes are same as those of the structural changes for the transfer expenditure respectively since the social security revenues are closely related to the transfer expenditures Furthermore， for the total government revenueR3' a structural change -Q U LU T point is detected in 1969 for the periods 1957-1991 and 1957-1979.. structural change might be related to that of the tax revenue R1 for the For the period 1974-1991， however， we cannot regard period 1957-1979.. 1986 as a point of structural change since the result of the Chow test is not significant III. D. Summury of Findings We pointed that there are rough correspondences between the govern -ment purchases E1 and the tax revenueRl'and between the transfer

(13)

Displacement Effect on Government 687 Expenditure and Government Revenue in Japan， 1957-1991 -37~少ー expenditureιand the social security revenueRz So， the findings of this section are as follows : (a) The structural change for the social security revenue is dated one year later than that for the transfer expenditure during each of the periods considered respectively.. The displacement effect on the former and the displacement effect on the latter might have occurred at the same time approximately since there is close relation between them (b) The structural change-points for government expenditures are not generally equal to those for government revenues. This means that the analyses of the displacement effcts on revenues are different from those on expenditures..

By the way， according to Peacock and Wiseman (1961)， people's ideas about the tolerable burden of taxation can be separated from their notions of the desirable level of public expenditure and there is likely to be a gap between the two concepts. Therefore， major social disturbances， such as World Wars， may create a displacement effect by shifting public expendi -ture to new levels， and such a shift may occurred because new ideas of tolerable tax levels emerge after the disturbance is over.. According to Gupta (1967)， a shift in people's ideas about the tolerable burden of taxa -tion is possibly of greater significance for a displacement effect of war and a shift in their ideas about the desirable level of public expenditure could be considered of greater significance for the displacement effect of the Great Depression.. So， in this paper， the structural changes for government expenditures and revenues are interpreted by adapting the above two concepts as follows. First， we interpret the structural changes for the expendituresEl'E;z， and 広 asthe shifts in people's ideas of the desirable level of government purchases， transfer expenditure， and total government expenditure respec

(14)

-380ー Kagawa University Economic Review 688

tively Next， we interpret the structural changes for the revenuesRl' R2'

and R3 as the shifts in people's ideas about the tolerable burden of tax revenue， social security revenue， and total government revenue respective -ly Peacock and Wiseman (1961， p.. xxiv) state that people's ideas about the tolerable burden of taxation are tranalated into ideas of reasonable tax rates.. Considering this statement in our model， we suggest that they can be formulated by the regression coefficients of government revenue equa -tions Since we have observed some possible structural changes for gov -ernment expenditures and revenues， we wi1lnow examine in some detail the shifts in these concepts by the regression analyses.

I

V

.

Results of the Regression Analyses

In Table 2， we show the results of regression analyses for the sub periods of 1957-1979 and 1974-1991れ Theslope regression coefficient esti

-mates in the table have expected signs and are significantly different from zero at the 1

%

levelゎItis therefore interesting to examine the GDP elasticities of government expenditures and revenues which are given by the slope regression coefficient estimates of expenditure and revenue equations respectively in Table 2

IV. A. Examination of the GDP Elasticities of the Government Expenditures

We examine the elasticities of the government expenditures. From the results in Table 2， the elasticity of government purchasesE1 is relative

..984 2..024 ( -9632a₎ ₍₂₃₈₁₆a₎ R3 1957-1969 -2..269 L050 0..995 1. 095 (-8051a) (51.085a₎ 1970-1979 -7..716 L435

。

..913 L646 (-3.627a₎ ₍₉_.₇₆₁a₎ 1974-1991 -8 102 L463 0..988 L028 (-13.906a₎ ₍₃₆_.₉₅₀a₎

Notes: The values in parentheses are f-values. a and b mean that the variables are significant at the 1% and 5% levels respectively R2 _d_e_n_o_t_e_s_t_h_e_c_o_e_f_f_i_c_i_e_n_t_o_f_d_e_t_e_r_m_i

-nation. Adj肝R

(16)

-382 Kagawa Universiか EwnomicRevi仰， 690 relatively small for the period after the structural change-point 1978， and there might be a decrease in the desirable level of government purchases as a result of the displacement effect of the recession after the second oil crisis Similarly， the elasticity of government transfer expenditure 旦 1S relatively large for the period after the structural change-point 1973 con -sidering the sub-period 1957-1979 However， cosidering the sub-period 1974-1991， the elasticity of transfer expenditure is relatively small for the period after the structural change-point 1980.. Therefore， there might be some increase in the desirable level of transfer expenditure as a result of the displacement effect due to the reformation of the social security program as well as the first oil crisis， whereas the desirable level of transfer expendi -ture appears to decrease as a result of the displacement effect of the recession after the second oil crisis.. For the total goverment expenditure E3'while there might be an increase in the desirable level of total govern -ment expenditure as a result of the displacement effect due to both the reformation of the social security program and the first oil crisis， there might be a decrease as a result of the displacement effect due to the recession after the second oil crisis

IV.B. Examination of the GDP Elasticities of the Government Revenues We examine the elasticities of the government revenues from the results in Table 2.. The elasticity of the government tax revenueRl in -creased slightly for the period after the structural change-point 1969 con -sidering the sub-period 1957-1979. Such a result suggests a slight upward shift in people's ideas about the tolerable burden of tax revenue as a result of the displacement effectリ Insection IIL we observed that the structural

change-points for government purcasesElare not equal to those for government tax revenueRl' Therefore， there remains a gap between

(17)

Displacement Effect on Government 691 Expenditure and Government Revenue in Japan， 1957-1991 -383 people's ideas about the tolerable burden of tax revenue and their notions of the desirable level of government purchases even if a displacement effect on government purchases occurred N ext， the elasticity of the social security revenueR2 is relatively large for the period after the structural change-point 1974 considering the sub -period 1957-1979， and is relatively small for the period after the structural change-point 1981 considering the sub-period 1974-199L Therefore， for the revenueR2， while there might be an upward shift in the tolerable burden of social security revenue as a result of the displacement effect due to the reformation of the social security program as well as the first oil crisis， the displacement effect of the recession after the second oil crisis might give rise to a downward shift in the tolerable burden of social security revenue.. Such a shift has the same direction as a shift in the desirable level of transfer expenditureιfor each of the sub-periods 1957-1979 and 1974 -1991引 Insection III， we also observed that the displacement effect on the social security revenue and the displacement effect on the transfer expendi -ture might have occurred at the same time approximately for each of the sub-periods" Thus， we consider that a gap between people's ideas about the tolerable burden of social security revenue and their notions of the desirable level of transfer expenditure is likely to tend to be relatively small after the structural change Finally， for the sub-period 1957-1979， the elasticity of the total govern -ment revenue R3 is relatively large for the period after the structural change-point 1969， and there might be an upward shift in the tolerable burden of total government revenue.. In this case， similar to the relation between the government purchases and the government tax revenue， there remains a gap between people's ideas about the tolerable burden of total government revenue and their notions of the desirable level of total govern

(18)

-384- Kagawa UniversiかEじonomicRevieω 692 ment expenditure even if a displacement effect on total government expend -iture occurred.. IV. C.Summury of Findings From the results of the regression analyses， our findings of this section are as follows : (a) For the transfer expenditure ιand the social security revenueR2， the plausible explanation of the displacement effects on government transfer expenditure is based basically on the concept of the tolerable burden of social security revenue. A shift in the desirable level of transfer expendi -ture has the same direction as a shift in the tolerable burden of social security revenue.

(b) For the government purchasesEland the tax revenueR1， there remains

a gap between people's ideas about the tolerable burden of government tax revenue and their notions of the desirable level of government purchases even if the displacement effects on government purchases occurred (c) For the total expenditure~ and the total revenue凡， there also remains a gap between people's ideas about the tolerable burden of total government revenue and their notions of the desirable level of total govern -ment expenditure even if the displacement effects on total government expenditure occurred Itseems likely that the displacement effect on the transfer expenditure associated with the reformation of the social security program and the recession after the first oil crisis occurred because some new expenditures became highly desirable. This upward shift in the level of transfer expend -iture also occurred because people got accustomed to a higher burden of social security revenue after the reformation of the social security program and the first oil crisis Thus， such a shift in people's ideas about the

(19)

Displacement Effect on Govemment

693 Expenditure and Government Revenue in Japan， 1957-1991 -385-tolerable burden of social security revenue may shift the transfer expendi -ture to new levels. This explanation is also given for the displacement effect due to the World War， which is discussed in Peacock and Wiseman

(1961)

But， the plausible explanation of the displacement effects on the gov -ernment purchases has to be provided by some other relevant factors except the concept of the tolerable burden of tax revenue. Considering that the deficit financing has had important role in the central government revenue since1965in apanJ. ， the debt financing may determine the level up to which the changed ideas about the desirable level of government purchases are implemented.. The burden of debt financing is considered almost zero during a period of severe depression This explanation is given for the displacement effect due to the Great Depression as seen in Gupta (1967)

However， dissimilar to Gupta (1967)， we examine the displacement effects on government revenues as well as those on government expenditures.

As for the relation between the total government expenditure広 and the total government revenueR3' we can discuss similarly. The plausible explanation of the displacement effects on the total expenditure is not provided only by the concept of the tolerable burden of total revenue.. The level up to which the changed ideas about the desirable level of total expenditure are implemented may depend on the debt financing as well as the tolerable burden of total revenue. Thus， there remains a gap between people's ideas about the tolerable burden of total government revenue and their notions of the desirable level of total government expenditure after the recessions due to the two oil crises as well as the Great Depression..

V.

Conclusion

(20)

-386'ー Kagawa University EconomiιReview 694

revenues and expenditures using the annual data from 1957 to 1991 in Japan From the results， we observed that structural change-points for expendi -tures are not generally equal to those for revenues. This means that the analyses of the displacement effects on expenditures are generally different from those on revenues. Thus， analyses of the displacement effects on revenues also are meaningful though very few studies have examined the displacement effects on revenues

Particularly， in this papar， we interpreted the structural changes for expenditures as the shifts in people's ideas of the desirable level of expendi -tures and the structural changes for revenues as the shifts in their ideas of the tolerable burden of revenues.. Thus， there is likely to remain a gap between the two concepts even if the displacement effects on expenditures occurred引 In this case， the plausible explanation of the displacement

effects on expenditures is not provided only by the concept of tolerable burden of revenues. The debt financing as well as the tolerable burden of government revenues may determine the level up to which the changed ideas about the desirable level of government expenditures are imple -mented.. Acknowledgement An earlier version of this paper was pasented at the Western meeting of the Japan Association of E乙onomicsand Econometrics held at Fukuoka University on June， 3， 1995..We aa grateful to ProfessorHideo Nakai， the discussant of the meeting for his helpful comments Comments and advice by Professor Kazuhiro Ohtani are also grate -fully acknowledged Refer'ences Abizadeh， S.. and Y ousefi， M， (1988)，“An Empirical Reexamination of Wagner's Law

ぺ

(21)

695 Expenditure and Government Revenue in Japan， 1957-1991 387 -Economiιs Letters， V 01.26， N o. 2， pp. 169-173 Beck， M ， (1979)，“Public Sector Growth: A Real PerspectiveぺPublic Finance / Finances Publiques， V 01 34， N 0 3， pp. 313-356 Beck， M ， (1985)，“Public Expenditure， Relative Prices， and Resource AllocationぺPub -lic Finance / Financω Publiφas， Vo140， No 1， pp 17-34 Diamond， J， (1977)，“Econometric Testing of the Displacement Effect: A Reconsidera -tion"， Finanzarchiv， Vol.35， No 3， pp 387-404

Gupta， S P， (1967)，“Public Expenditure and Economic Growth: A Time-Series Analy -sis"， Public Finanじe/ Finances Publiques， V 01.22， N 0 4， pp.. 423-466

Holtz-Eskin， D， Newey， W. and Rosen， H. S， (1989)，“The Revenues-Expenditures Nexus: Evidence from Local Government Data"， International Bιonomic Review， Vol 30， No.2， pp.415-429

Honda， Y and Ohtani， K， (1986)，“Modified Wald Tests in Tests of Equality between Sets of Coefficients in Two Linear Regressions under Heteroscedasticity"， The ManchesteァSchool0/Economic and Social Studies， Vol 54， No. 2， pp..208-218 Jou!faian， D and Mookerjee， R， (1990)，“The Intertemporal Relationship between

State and Local Government Revenues and Expenditures: Evidence from OECD Countries"， PublたFinance/ Finances Publique.s， Vo145， No 1， pp. 109-117 Nagarajan， P.， (1979)，“Econometric Testing of the Displacement Effect Associated

with a Non-Global Social Disturbance in India"， Public Finanα/ Financιs Pub -liques， Vol.34， No 1， pp.100-113

Nagarajan， P， (1983)，“Displacement Effect in Government Spending in Sweden: A ReexaminationぺPublicFinance / Finance.s Publiques， Vol.38， N 0 1， pp 156-162 Nomura， M ， (1991)，“The Displacement Effect on Govemment Expenditure of Two Oil

Crises: Japan， the United Kingdom and the United States"， The Manchester School 0/Economic and SoじialStudies， Vol 59， No 4， pp..408-418

O'Hagan， J羽r，(1980)，“Demonstration， Income and Displacement Effects as Determi -nations of Public Secter Expenditure Shares in the Republic ofIreland"， Public Finance / Finances Publiques， V 01 35， N o. 3， pp 425-435

Peacock， A.T. and Wiseman， J， (1961)， The Growth 0/Public Eゆenditure in the United Kingdom， Princeton， N..J， Princeton University Press

Peacock， A.T. and Wiseman， J ， (1979)， ‘'Approaches to the Analysis of Govemment Expenditure Growth"， Public Fi仰 nceQω情的， Vol.7， No..1， pp.3-23

Ram， R， (1988)，“Additional Evidence on Causality between Govemment Revenue and Govemment Expenditure"， Southern Economic Journal， Vo154， No.. 3， pp.. 763-769. Thom， R S， (1967)，“The Evolution of Public Finances duringEconomic Development"

(22)

-388 Kagawa UniversiかEwnomicReview 696 von Furstenberg， G M， Green， R.l， and .Jeong， J， (1986)，“Tax and Spend， or Spend

and Tax ?"， Review oj Ewnomics and Statistics， VoL 68， No..2， pp.179-188

羽Tatt， P A， (1978)，“Econometric Testing of the Displacement Effect:A Note"， Fi -nanzarchiv， Vol.36， No..3， pp..445-448

Watt， P A， (1979)，“Tests of Equality between Sets of Coefficients in Two Linear Regressions when Disturbance Variances are Unequal: Some Small Sample Prop -ertiesヘTheManchester Schoo!0/Ewnomic and Socia! Studies， VoL 47， No. 4， pp 391-396