UNIVERSITY OF
ILLINOIS LIBRARY
AT URBANA-CHAMPAIGN
BOOKSTACKS
Digitized by the Internet Archive
in 2011 with funding from
University of Illinois Urbana-Champaign
http://www.archive.org/details/japanseconomicgr54ujii
Faculty Working Papers
Japan's Economic Growth and Balance of Trade
Junichi Ujiie and Patrick Yeung
#54
College of Commerce and Business Administration
University of Illinois at Urbana-Champaign
FACULTY WORKING PAPERS
College of Commerce and Business Administration
May 23, 1972
Japan's Economic Growth and Balance of Trade
Junichi Ujiie and Patrick Yeung
#54
1
I . INTRODUCTION
In the literature on the relation between economic growth
and balance of payments, the familiar two-country framework
is often used for analytical purposes. The countries may be
used to represent a particular country whose balance of pay-
ments is under consideration and the rest of the world.
Within this framework, it might be expected that, other things
being equal, if country A's income rises due to productivity
increases, country A's demand for country B's exports would
rise, and the balance of payments would turn against country
A. Sir John Hicks [4 J, however, has pointed out that in
practice, the opposite result is possible. Such a "weird
case" may be attributable, according to Hicks, to the presence
of biased rather than uniform growth.
It is the purpose of this paper to study the case of
post-war Japan to shed some light on Hicks' argument. The
Japanese case is interesting because, while it has been grow-
ing faster relative to the rest of the world, it has a
tendency to' experience trade surpluses. .
A
2
Our paper is organized as follows: Section II presents
our theoretical model, extending a familiar aggregate mod-^i
to disaggregate levels. The method of estimating the parameters
of our model is explained and the- empirical results given in
Section III. Section IV summarizes our findings with some
concluding remarks .
•r - ■ ■« -r I_i
II. THE MODEL
Models on the trade balance of a country are formulate<
generally in terms of income and relative price effects on
imports and exports. Specifically,
X
M
2.1 T - v £
2.2 ir ■ =-.
P*
2.3 M-f(i.Y)
2.4 X - g (it , Y*>
where
T = export ratio
M = imports
X = exports
P = price level
Y = income
ir - terms of trade between exports and
imports
and the star ( *) denoting the rest -of -the -world terms.
According to Professor Harry Johnson's celebrated "basic
equation" [6 ] :
2.5 Rj, = [(l-n*-n) r^] + [e* R* - eR]
R = rate of growth of Y
R_ = rate of change of T
r = rate of change of the terms of trade ir
7T
€ = income elasticity of demand for impcits
t\ = price elasticity of demand for imports
(* denoting rest -of- the -world items)
The two bracketed terms on the right-hand side of Equation
2.5 are the "price term" and the "income term" respectively.
Recalling the case cited in the previous section where
r = R* = 0, Equation 2.5 reduces to R-, = -eR, whereby one
might expect the trade balance in the growing country to turn
unfavorable .
Continuing to make the simplifying assumption that there
is no change in prices (r = 0), if both countries enjoy
economic growth, Equation 2.5 becomes R_ = e* R* - eR. If
R > R* (as in the case of Japan), for VU to be positive would
imply that e < e*.2
In consisting of a basket of goods, M is unlikely to be
an inferior "good," and € is expected to.be non-negative.
2
It is reasonable to assume a priori that € and e* are
both positive .
Sv-
5
So far, imports and exports have been treated in the
aggregate. Let us re-write their functions in explicit form:
2.3a M = a (^ Ye eu
IT'
* e* *
2.4a • X = b Tr11 y* eu
u and u* being stochastic terms.
Now
2.6 ZP M, f P*M
m. J.
2.7 ZP X, = PX
xj J
where Pm M. and Px X. are respectively the value of imports
of the i " good and the value of exports of the j good.
Assume
*1j e u
2.8 M. = a. 7T, y e 1
*i i i
and
* *
2.9 X. = b. t. y e
J J 3 y
2.10 ir< z P /P
i mt
2.H 7T* ; P /P*
fch
where P = import price of the i good
mi
fch
P = export price o2 the j good,
Then by Equations 2.6 and 2.7:
P
mi
0(M, — - ) • Pm M
- Y9M , Y^ 1 P* Y?JHe ^k
z M9Y • M " ^y " M p* iY
P M.
m. i
P*M
P M.
. (?) ?M ' 7. „ ^\
71 = u~ m"z
p / d7ri
M *ty V H d(|)
dP
mi dP*
Assuming = 0 and -rr— = 0,
dP atf
_i p mi P
p* J J\ \\ d7ri d(— ) mt
9^ P ir± d(f) d(p-) P
/X\ X P « M P P M4
. _ - (F g M . t , m4 \Mi mi „ mi *
M gtL\ M P* Ti p 1 p*M
Similarly, it can be shown that
J. PX ■ ' • 'J px
P M.
m. 1
e = 26,
P M
n 7 sn±
p M.
m. i
P*M
2.12
P X,
€ = 2e — »—
PX
P.. X,
# * * I
J PX
The model is further extended by considering that the
income and price elasticities are not necessarily constant
over time. In Professor Johnson's formulation already
referred to, the bigger the income elasticity of world demand
for domestic exports (e*) is than the income elasticity of
domestic demand for imports (e), the more favorable is the
domestic trade balance (other things being equal). In a
The allowance for disproportionate changes in Pm. and in
P as well as this additional consideration make our model
different from that in S. Y. Kwack [7] which was applied to
the case of the U.S.
dynamic context, the income -elasticity differential will
increase if the income elasticity of world demand for
domestic exports grows faster than that of domestic demand
for imports .
However, theory does not suggest any particular form of
the transformation function for incorporating this feature
into our model . We therefore adopt the simplest (linear)
form
2.13 e = € + Cjt
2. .H e*=e* + e. t
o 1
where t = time
To be even more general, we may also write
2.15 TJ = T1Q + TJjt
2.16 n* = n* + \t
Finally, the dynamic formulation of the income and price
elasticities are also similarly extended to their disaggre-
gate counterparts.
9
III. ESTIMATION AND EMPIRICAL FINDINGS
Our empirical analysis is organized according to (1)
whether the aggregate or disaggregate approach was used, and (2)
whether the elasticities are hypothesized to follow secular
trends (static or dynamic formulations).
The Method of Ordinary Least Squares (OLS) was applied
to multiple regression equations in the double -logarithmic
form. The data we used consist of times -series figures on
the following dependent variables:
M -- Japan's total import quantum index
M. — Japan's quantity of import index of the i good
X — Japan's total export quantum index
* •
fch
X. — Japan's quantity of export index of the j good
and figures on the following independent variables:
P — Price index of Japan's imports
m
th
P — Price index of Japan's imports of the i good
P — Japan's wholesale price index
P -- Price index of Japan's exports
fch
P — Price index of Japan's exports of the j good
Xj
10
p * 1
x — "Competitor's export price index"
Y — Index of Japan's real GNP
Y* — Real gross domestic product index of world
Figures for M, M . , X, X. , P , P , P , P are obtained from
*■ j in m. x x»
Bank of Japan, Economic Statistics Annual, and figures for
P , P *, Y and Y* are taken from the UN Statistical Yearbook
W A
for the period 1953-1967, and are shown in our statistical
appendix at the back of this paper. (See Tables A-l, A-2,
and A-3). The former source categorizes imports (M. ) into 8
groups (Foodstuffs, Textile Materials, Chemical and Allied
Products, Machinery and Equipment, Mineral Fuels, Metal Ores
and Scrap, Other Crude Materials, and Miscellaneous), and
exports (X.) into 7 groups (Foodstuffs, Textiles, Chemicals
■I
and Allied Products, Machinery and Equipment, Metals and
i
Metal Products, Non-metallic Mineral Products, and Miscell-
aneous). Since all the variables are given as indexes, we
have uniformly converted the base year to I965.
This is actually the unit value index of exports of
manufactured goods . This index is a weighted average of
11 industrial countries. No adjustment has been made about
the inclusion of Japan due to economy of research time .
11
A. Income Elasticities
(i) Aggregate approach
Equations 2.3a and 2.^a may be estimated in the double-
logarithmic form
3.1 log M = log a + € log Y + T) log (P /P ) + u
3.2 log X = log b + e* log Y* + tj* log (P /P* ) + u*
The results are shown in Table 1. Table 1 not only shows
our estimates, but also those of Houthakker-Magee [5]
compared with T. C. Chang's estimates of the pre-war period
Bl.1
Since the period covered by Houthakker-Magee (1951-1966)
are approximately the same as ours (1953-1967), our results
are expected to be similar to theirs . While the data we usca
are slightly different from theirs, and further data refine-
ment is always desirable, the similarity in our results with
theirs further justifies our reliance on the proxy measure
of P *.2
x
Another related study for the post-war period has been
made by Baba-Tatemoto [1].
p
See note 1 on page 10 above . This proxy variable is
different from the corresponding one used by Houthakker-Magee
5
I
a
o st
st o
• •
CM r-4
O 00
r-( VD
• •
CM O
CM
pd
jd- o oo oo oo oo
CO t- 0\0\ U\0\
*
pr
o
vO
o
I
st
■8
*
CO
o
m
CO
in
s
00
O r-H
I I
o o
CM O
• •
C CVl
I I
t-r-J
d d
m cm
LOCO
i-l CO
0O4- coco
COvD
CVS O
r-H CO
CO r-»
cnt-
i-I CO
r-4
0)
60
H
60^
1 ' — »
00^—
C 00
J-J ^
C C-
nJ CO
q vo
3 VD
x: on
■^ ON
^ °>
O r-t
.Si rH
JM r-i
i
CtJ I
1 1
• St
Xi r*
<U CO
O CM
jj m
•H lf\
ON
3 OY
•n on
H --'
o
32
JJ
u
I
JJ
nj
CO
•H
JJ
C
0)
•H
o
•H
M-l
«J-I
<U
o
o
■8
as
CD
o
iH
cu
CO
•H
CO
0)
4J
C
(U
u
a.
c
•r(
a)
|
CD
JJ
O
Table 1 shows that the post-war estimates of e* are
approximately three times as high as their pre-war counter-
parts, while the level of € has apparently remained at about
the same level. This suggests that we should consider changes
in the income elasticities over time.
In incorporating secular trends into our analysis,
Equations 2.13 and 2.14 are substituted into Equations 2. 3a
and 2.4a, and the following regressions were run:
3.3 log M = log a + c log Y + €nt log Y + tj log (P /P ) + u
3.4 log X = log b + e* log Y* + e*t log Y* + n* log (P /P*) .
w J- X JZ*
The results are presented in Table 2 where it can be seen
* 1
that both e, and e, are not significant at the 5% level.
Substituting Equations 2.13 and 2.14 together with
Equations 2.15 and 2.16 into Equations 2-3^ and 2.4a was als
tried:
3.5 log M = log a + e log Y + e-t log Y + n log (P /P„)
O JL \J III w
+ ijjt log (Pm/Pw) + u
3.6 log X = log b + e* log Y* + e*t log Y* + n* log (P /P*)
w JL \J X X
+ T)*t l0g(Px/P*) + U*
The results are poor due to the existence of multicollinearity
between the first and the second independent variables as well
as serious autocorrelation in the export function. It was
therefore deemed unnecessary to reproduce these results in dc it
Table 2
e ,e *
o o
e e *
1» 1
n, n*
R2
D-W
M
2.644
(*.W)
-0.023
(-2.012)
0.387
(1.016)
•98
2.0:
X
1.426
(0.542)
0.018
( 0.667)
0.015
(0.011)
.98
0.72
However, we do not consider this evidence as conclusive because
the real relationships may be hidden by the process of aggre-
gation. We therefore turn to the disaggregate approach.
(ii) Disaggregate approach
First, the disaggregate static formulation of the incor.
elasticities was estimated as follows:
3.7 log M± = log at + e± log Y + t\± log (Pm#/Pw) + ^
3.8 log X. = log b + £* log Y* + n log (P /P*) + u
J J J J x* x J
The results are presented in Table 3, where it can be seen
that all the e.'s and the e?'s are significantly different
from zero at the 5$ level.
Table 3
nl
M2
M3
M,.
H6
"7
«8
1.20
(6.83)
0.356
(3.043)
1.73
(8.902)
1.89
(3.926)
2.045
(16.617)
1.608
(4.473)
1.409
(52.676)
2.002
(10.856)
0.99
(1.285)
-0.753
(-2.181)
0.334
(0.456)
-1.641
(-1.86 )
0.0239
(0.089)
-2.471
(-2.118)
0.385
(2.204)
1.699
(1.859)
Bfc
.83
.86
.98
.81
.99
.79
.99
.91
D-W
0.59
2.21
1.435
0.811
1.807
2.5^
2.384
2.405
Al
h
2.05
(10.155)
-2.161
(-4.635)
1.2053
(3.777)
-2.172
(-2.52 )
4.147
(9.684)
-0.466
(-1.464)
5.^7
(13.693)
0.226
( 0.423)
2.999
(8.637)
-1.857
(-3.849)
2.361
(20.578)
-3.317
(-5.64 )
3.287
. (15.763)
-2.41
(-4.148)
.89
.92
.99
.98
.98
.97
.95
D-W
1.887
1.299
2.007
0.896
1.625
1.839
1.202
Next, Zd.lner's Seemingly Unrelated Regression Method
( SUR) [13] w_ j tried on the assumption that certain dis-
turbance ele. ants (such as dock strikes, changes in tariff
structure, e;c.) are contemporaneously common Co the M.'s
as a group and to the X.'s as a group. Regressions were run
using the joint GLS method (see [11, pp. 298-302]). The
results, however, were quite similar to those obtained fror.
the separate regressions. This need not imply the absence
of contemporaneous covariances in the disturbance terms, due
to the fact that in the joint estimation of the M. equations,
Y was an important variable common to all of them, and the
same thing was true of Y* in the case of the X. equations,
so that the matrix of independent variables is dominated by
similar rather than different variables. For economy of
space, we will not report our experimentation with the STIR
method any further.
Before going further with analyzing the implications of
the results of Equations 3-7 and 3'8, we investigated whethc*
the dynamic formulations of the elasticities might be sta-
tistically more fruitful by means of Equations 3-9 and 3. 10
and Equations 3.11 and 3.I2:
3.9 log M. = log a. + € log Y + e1 -tlogY + tj. log (P /P )+Uj
i i i
3.10 log Xj = logbj. + e*.logY* + e*tlogY* + n^log (px./px) + uj
17
3.11 log M± = log a± + eQ logY + ex .tlogY + t\ log ( P /P )
i i
3.12 log X. = log b. + e* logY* + e- tlogY* + r\ log (P /P*)
j J Oj ij Qj Xj x
+ < tlog (P /P*) + u*
Serious multicollinearity resulted from the strong cor-
relation between the first and the second independent
variables in Equations 3-9 and 3 '10 and again between these
variables as well as between the third and fourth independent
variables in Equations 3-H and 3-12. The statistical
results were therefore poor relative to those of Equations
3.7-3.8, so that we decided to discontinue the line of investi-
gation involved in Equations 3-9-3-12 and revert back to
analyzing more deeply the results of Equations 3-7~3-8.
Moreover, the computed coefficients of the trend term in
Equations 3-9-3-12 were found to be insignificantly different
from zero. We therefore assume that there is no trend in
each of the commodity group's income elasticity.
Now, assuming constancy of the income elasticity in each
commodity group, we can compute the aggregate elasticities
18
using Equation 2.12. The shares of Japan's imports and
exports are shown in Table A~3« These aggregate elasticities
shown in Table k are found to change over time due to changes
in the share of each commodity group in total imports and
*
exports. Moreover, it can be seen that e grows faster than
t
e . When we fit linear trend lines to these elasticities,
we can see the difference in the rate of increase in e
and e* more clearly.
3.13 et = e + "« t +tt|. R2
1.22 0.025 0.90
(60.08) (11.18)
* «. „* T>2
3.H «;- e*t + cjt t +u; K
2.50 0.077 0.95
(60.59) (17.04)
Equations 3-13 and 3-14 show that e grows approximately three
times faster than e .
1 *
A simple statistical test of the hypothesis that e
grows faster than e may be made with
3.15 <- et = (/t - 6ot) + (.Je - «lt)t + v
1.28 0.052
(27.^3) (10.20)
The result leads us to accept the hypothesis.
19
Table 4
Year e e*
1953 1.22 2.71
1954 1.22 2.59
1955 1.25 2.65
1956 1.30 2.86
1957 1-43 2.92
1958 1.39 2.99
1959 1.42 3.06
i960 1.44 3.04
1961 1.46 3. 18
1962 1.50 3.18
1963 1.49 3-30
1964 1.52 3-39
1965 1.52 3.50
1966 1.53 3.58
1967 1.57 3- 81
Average 1.42 3-12
The arithraatic means of the e 's and e*'s in Table 4
t t
are respectively 1.^2 and 3-12, which are close to the
figures for e and e* in Table 1 obtained directly from
Equations 3-1 and 3.2.
B. Price Elasticities
Estimation of the price elasticities was undertaken via
the same approaches used for the income elasticities .
20
(i) Aggregate approach
From Equations 3.1-3-2 (see Table 1), che aggregate
static price elasticities were obtained. One-tail t-tests
at the 5$ level for the significance of tj and r\* were
rejected against the alternative hypotheses that they are
both negative. In comparison, the same tests were applied
to Houthakker-Magee's results (also shown in Table 1), with
13 instead of our 12 degrees of freedom, and showed that both
their r\ and n* were significant.
From Equations 3-3-3.1* (see Table 2), we again found
tj and tj* to be insignificant. Insignificance was also
indicated by the same one -tail test when the dynamic formu-
lation of r) and r\* was added to that of e and e*.
(ii) Disaggregate approach
From Equations 3.7-3.8 (see Table 3), it can be seen
that 3 of the 8 tj. 's and 5 of the 7 rjf's have the right signs
■*- J
p
and are significantly different from zero.
Equations 3-H-3-12 already discussed above generally
failed to establish statistical significance for the rj-.'s'
Also see note 1 on page 13 above.
2
The statistical problems with estimating Equations 3-9-
3-10 have already been noted above.
21
and t\ ' s. We have therefore relied on the results of
Equations 3.7 and 3.8.
By using the same method as in the case of income
elasticities, each year's aggregate price elasticities tj
and tj were computed indirectly from the disaggregate price
elasticities recorded in Table 3. Those price elasticities
statistically not significantly different from zero were
assumed to be zero. The results are shown in Table 5.
Table 5
*
year Tjfc t^
1953 -0.49 -1.77
1954 -0.49 -1.81
1955 -0.45 -1.82
1956 '-0.61 -1.70
1957 -0.65 -1.64
x 1958 -0.55 -I.63
1959 -O.63 -1.61
1960 -0.6? -I.62
1961 -0.70 -1-55
1962 -0.63 -1.55
1963 -0.57 -1.49
1964 -0.55 -1.44
1965 -0.53 -1.36
1966 -0.52 -1.36
1967 -0.54 -1.12
Average -0.57 -I.56
22
Recall that in Equation 2.12 the aggregate price
elasticity was approximately equal to the weighted sum of
each commodity group's price elasticity. From the table,
we see that the arithmetic means of the elasticities r\
and t) are respectively -O.57 and -I.56, which have the
right sign, and are bigger in absolute terms than those
obtained by the aggregate approach.
IV. SUMMARY At© CONCLUDING REMARKS
Several tentative conclusions can be drawn from our
analysis .
First, we have observed that the disaggregate approach
is superior to the aggregate approach in terms of revealing
the true income and price elasticities. With the disaggre-
gate approach, we have shown that the derived aggregate
elasticities are not constant over time. On the other hand,
not only does the aggregate model not show the true elas-
ticities, but it also hides the effect of changes in the
composition of imports and exports. The use of the disaggre-
gate approach is therefore consistent with Rick's suggestion
of biased growth in considering the "weird case" of changes
in the trade balance .
Second, while Houthakker-Magee pointed out the difference
between e and e* in the post-war Japanese economy, we con-
sidered additionally the difference between e and e* in the
pre-war period and these parameters in the post-war period.
We found that e and €* generally increased over the fifteen
years under consideration, and that the rate of growth of e*
has been faster than that of e. This means that the difference
24
between e and e* has become bigger and bigger over time.
Although we do not have any exact quality measurement of
Japanese products, it seems reasonable to consider that
improvement in the quality of Japanese exports has contri-
buted to the increase in e*. Another way of interpreting
our finding is that Japan's experience of rapid growth has
contributed to the possibility of a smooth adjustment in
terms of her resource allocation toward promising industries
characterized by high income elasticities. In other words,
it has been possible for the rapid change in the composition
of Japanese exports to be accomplished smoothly because of
the rapid growth of Japan's economy as well as the growth
of 6*.
Third, our derived aggregate r\ appears to exhibit some
kind of declining trend over time, but the aggregate r)
appears to be relatively constant over time. Within the
limited framework of the Marshall -Lerner Condition that
I7! + T { > 1 f°r a successful revaluation of the yen,
our finding shows that this condition is satisfied, although
over time there appears to be some tendency for it to be
weakened due to the reduction of the absolute value of the
sum of r\* and n. However, for considering Japan's role in
the recent world currency realignment, our finding is only
tentative and should be further verified.
25
Finally, it should be mentioned that no distinction has
been made in our paper between the private and government
sectors in Japan. Under this simplified setting, the
question of trade restrictions has been ignored from our
analysis .
REFERENCES
[1] M. Baba and M. Tatemoto, "Foreign Trade and Economic
Growth in Japan" in K. Ohkawa and L. Klein, eds .
Economic Growth; The Japanese Experience Since
Meij . Era (New Haven: Yale University Press, I968) ,
pp. 162-196.
[2] Bank of Japan, Economic Statistics Annual (various
issues) .
[3] T. C. Chang, Cyclical Movements in the Balance Payments
(Cambridge: Cambridge University Press, I95I).
[4] J. R. Hicks, "The Long-run Dollar Problem," Oxford
Economic Papers, n. s. vol. V, (June 1953), PP •
117-135.
[5] H. S. Houthakker and S. P. Magee , "Income and Price
Elasticities in World Trade," Review of Economic.
and Statistics, vol. 51 (May I969 ) , pp. 111-125-
[6] H. G. Johnson, "Increasing Productivity, Income-Price
Trends and the Trade Balance," Economic Journal,
vol. 64 (Sept. 1954), pp. 462-85.
[7] S. Y. Kwack, "The Determination of U. S. Imports and
Exports: A Disaggregated Quarterly Model, i960 II
I967 IV," Southern Economic Journal, vol. 38
(January 1972), pp. 302-314.
[8] T. Negishi and F. Watanabe, eds., The International
Trade of Japan (In Japanese), (Tokyo: Iwanami-
Shoten, 1971).
[9] M. Tatemoto, "Productivity Growth and Trade Balance,"
Osaka Economic Papers, vol. V (1957), PP • 16-23-
[10] M. Tatemoto, T. Uchida and T. Watanabe, "A Stabilization
Model for the Postwar Japanese Economy: 1954-62,"
International Economic RggifiK, vol. 8 (February 1967),
pp. 13-44.
[11] H. Theil, Principles of Econometrics (New York:
Wiley, 1971 )
[12] United Nations, Statistical Yearbook (various issues).
[13] A. Zellner, "An Efficient Method of Estimating Seeming"1
Unrelated Regressions and Tests for Aggregation
Bias," Journal of the American Statistical Associa,'-
vol. 57 (June, 1962), pp. 3^8-368.
Statistical Appendix
X
r-l 01
(0 -o
a) c
•a <u
+-
K
>-
-i e
*J o
o o
s c
•H
to X
0)
P TJ
O *-< C
4J M -H
K X
■H O
-4—
CL,
*-i a. a>
-r-
a) X o
a. a) -w
E i-i
o a.
cj
X
Ol 0>
£0 H T3
- tfl C
C co -w
S
a) a)
4-
+-
PL,
O..H a)
to o u
*-l rC -H
3 n
a.
X
at
CO T3
- r-l C
C CO -H
c« 0)
+-
fH
a, )-> at
•¥-
at g
*-> o
u
c
•H
X
Ol
(0 TJ
- 4J C
C W -H
X
ed O
+-
PU
c c o
ffl !«! O
1-3 0) -H
Ij
D.
X
CU
CO 13
- u c
C U i4
E
« o
+-
(X
B.C. a
« s o
<-} -H t-I
. t-l
a.
X
in 0)
o -o
CO c
- >-,•*
C *J
+-
X
cfl -H 4-1
a. 4-1 i_
CO C O
•"J CO O.
3 X
a- oi
X
>4-l 01
o "a
(0 c
- ^-W
C 4J
4-
£
CO -H 4-1
a. *j i-<
to c o
•-) co a.
3 E
cr-n
no
ON
NO
co
M3
00
r->.
O
ON
co
00
ao
o
ON
vO
ON
o
o
o
nO
o
o
CM
o
00
CM
CM
r-l
CM
CM
o
o
o
o
O
o
o
o
o
o
o
o
ON
H
ON
CM
ON
vO
ON
00
ON
r--
ON
NO
ON
00
ON
ON
ON
ON
ON.
o
o
i-H
o
o
NO
o
O
CO 00
ON
ON
ON
nO
ON
NO
ON
o
ON
oo
ON
ON
On
O
O
CO
ON
o
o
o
o
o
rH
o
CO
o
m
o
o
-3"
oo
ON
m
NO
00
o
ON
o
o
l~t
o
CO
o
00
o
CO
sr
U0 CM vO oo ON
•J CO
00
CO
ON
ON
CO
^r
o
r^
o
o
o
o
o
H O
o
r-l
o
ON
o
o
o
ON
o
o
o
o
o
-i rH
r-l
r-l
■H
r-l
rH
H
rH
i-H
r-1
H
1— 1
o
r>»
1/0
<r
NO
T-i
CO
NO
NO
CO
o
-<r
\D
CO
o
uo
r-l
■— 1
r-l
CM
r-l
r-.
t-H
NO
CM
M0
O
o
c
O
o
o
CO
ON
o
o
CM
O
O
o
m
o
CO ON
00
CM O
CO CO
o
CM
I —
CM
in
CO
CO
o
m
o
NO
ON
m
ON
oo
o
o
CM
o
nO
CO
00
O
ON
H ON
vO
CO
CM
<r
CO
NO
NO
*o
o
rH
00
o
ON
-1 r-1
0 CO
CM
CO
o
o
uo
CM
CO
m
m
NO
m
00
00
o
o
m
m
r— 1
CO
ON
\0
CO
-T
m
NO
r~-
CO
ON
o
r-l
CM
co
<r
m
nO
r-
m
m
in
in
m
in
m
NO
NO
NO
NO
>o
NO
NO
NO
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
On
ON
ON
co
CU
g
•
co
X— N
co
03
•H
0)
3
co
CO
3
CO
O
•r-l
•H
Li
CO
CS
d
>
o
v^^
•r-i
Li
r-l
cfl
cd
>
3
> ^
C
c
Jd
<
O
O
CO
43
o
Li
•r4
cti
4J
CU
CO
>^
•H
4-1
rH
c«
ctf
4J
U
CO
•H
i-l
a
CO
•rH
•H
e
4-»
o
ctf
c
4-1
o
CO
o
_
td
CO
**
c
c
O
CO
■t4
a.
4-»
cfl
CO
>->
2
iw
T5
O
CU
u
Jid
•rH
c
C
nJ
3
«
. .
CU
CU
o
o
Li
Li
3
3
O
O
CO
CO
4-
+
+
TJ
c
CO
<r
00
en
a\
O
<?
<r
CO
<r
CO
in
<r
CM
O
en
o
>* u
e
r~
-H
o
CO
VO
r^
CM
no
MO
r—
r-l
CTi
o
m
o
u c
a.
<r
m
m
<r
m
m
vO
■X)
vO
vO
r-»
r—
o
ON
ON
<x> a)
r-|
c e
#
•H &
.C -H
a s
co cr
S W
<f
o
r-i
LO
no
rH
CM
i-t
CM
rH
00
in
NO
o
CM
o
•
s
r-
co
NO
00
co
<r
ON
NO
r--
r^
i^-
vO
o
en
<r
<r
-<r
<•
en
<r
r-~.
CO
r-.
CO
CM
rH
m
r-i
in
r-l
r-(
o
r-l
r-l
rH
m
r-t
en
u
•a o
C 3
CO -o
on
m
NO
CO
ON
CM
on
r-l
00
a\
m
CM
o
Or
r-l
O
CO
U) u
£
in
i-i
en
CO
no
r-~
CM
r-t
o
CM
<r
r-~
o
o
ON
H P-.
FU
r-i
CM
CN
CM
CM
i-H
rH
r-l
r-l
o
o
o\
o
O
ON
at
rH
^1
rH
rH
r-l
rH
rH
rH
r-l
r-l
r-l
r-{
rH
U 13
•H CD
6 "H
CD i-H
X rH
O <
en
r-»
CO
en
rH
<r
00
ON
m
m
m
m
m
o
en
o
•
2
o\
oo
CM
r-l
in
<r
r--
oo
<r
r-l
no
m
o
r-l
1-1
CO
.-i
i— i
CM
en
m
en
-*
m
r-»
r»-
co
r-l
o
CM
m
CM
I
<
CD
u
CD
±->
2:
01
p-l
CO -U CM
at
m m
ON
m
en
in
ON
CM
r~-
ON
o
r~~
00
CM
ON
CM
CN
en
CM
CM
CM
o
ON
ON
o
m
en
m
r--
en
m
vO
m
r-l
CO
en
o
ON NO ON
00
o
o
o
o
ON
ON
o
no
00
o
o
m
ON
o
o
o
o
CO
o
m
m
ON
NO
r- 1
co
UH
to
T3
O
O
o
00
co
o
O
CM
<r
O
NO
CM
en
rH
o
CM
00
CM
CM
r-l
-3-
rH
rH
o
r-i
r-l
CO
o
rH
OO
o
r-t
ON
rH
rH
en
O
rH
ON
CN
en
ON
ON
o
rH
en
r-i
rH
o
o
rH
o
o
rH
CM
O
00
r~~
<r
CM
CO
en
i^
CM
rH
rH
r-~
-^r
o
oo
<T
en
00
en
CO
en
in
en
NO
o
en
CM
en
ON
en
00
in
r-.
en
00
o
o
en
T-i
ON
rH
en
<r
m
v£>
r-v
CO
ON
o
r-t
CM
m
<r
«
m
NO
r-~
m
m
m
m
in
m
in
vO
NO
NO
>J3
vO
NO
NO
NO
o\
CJ\
ON
ON
On
ON
ON
CT\
ON
ON
ON
ON
ON
ON
ON
rH
rH
rH
rH
rH
rH
•—I
r-t
rH
rH
rH
rH
rH
rH
rH
CO
3.
O
0)
c
e*
co co in
cm m vd
oo
vO
in
CN
•<r
r-.
CN
CN
est
CM
r-t
<T
o
O
<r
00
00
oo
oo
00
00
oo
00
00
CO
00
CO
o
rH
o
rH
o
r-t
o
r-t
CJ
co
•H
S
X
oo
CO
CM
co cm
CM
CO
o
-3"
ST
o
m
CO
CO
oo
00
m
00
o
o
rs
o
00
T-l
vO
co
<u
co
co
c«
>-
^s
<U
o
o\
m
00
r-t
CM
vO
st
<r
CJ\
rH
rH
o
CO
m
X)
c
v£>
•■
•H
X
00
r--
CO
m
00
m
m
v£>
v£i
CM
CO
m
o
rH
CO
4J
X
P-,
co
00
00
oo
00
co
00
CO
00
ON
ON
o\
o
o
o
c
rH
r-l
rH
o
o
CJ
•H
r-t
r-t
CM
1
ccj
J-l
0]
<
CJ
o
E
3
r^
CO
r- i
o-
CO
00
CM
00
o
st
rH
o
O
O
CM
cu
1
T3
^O
rH
c
O
X
vO
CM
CO
<r
m
CM
co
CO
St
vO
sO
\o
o
\o
00
^O
o
u
CM
CO
<r
m
m
■n
vO
r-
r^
r--
co
cr>
o
o
o
«
23
(U
r-l
r-t
rH
H
\£>
CO
O
•r-l
U
ca
>
at
3
C
C
<
CO
CJ
•H
cO
CM
<j-
CO
vO
00
r-t
o
ON
rH
<T
rH
■«■
o
00
CO
4-t
m
01
X
ON
r>»
r^
on
o
CM
o
<r
00
r~
m
CO
o
vi3
rH
A
Ph
rH
O
O
CM
<r
rH
rH
rH
o
ON
ON
ON
o
ON
O
■H
rH
r-t
r-t
rH
rH
rH
r-t
rH
rH
r-t
*o
c
CO W
u
CO CJ
r-i ZS
co -a
4J o
St
r>*
CN
CO
m
m
m
00
m
-3-
CI
r-t
o
o
CO
<U !-4
m
S 0.
X
on
CO
o
tn
CO
ON
rH
CO
o
<r
00
r-t
o
r-~
CM
rH
CM
rH
rH
i-i
CM
CN
CO
<r
^o
P>.
o
O
o
m
co
•iH
AJ
at
.u
c/3
o
•H
s
O
c
o
CJ
w
c
ex,
ctt
<4H
o
c
CO
PQ
5h
CtJ
CU
cu
w
«
CO
<J-
m
o
r^
CO
ON
o
r-t
CM
CO
<r
*
m
vO
•-^
m
m
m
m
in
in
m
\D
vO
O
o
vO
vO
vO
\0
cr\
ON
ON
ON
ON
ON
ON
o>
ON
ON
ON
ON
ON
CT\
cr>
rH
rH
r-t
rH
m
r-t
rH
rH
rH
r-t
r~i
rH
rH
rH
r-t
<u
o
u
o
c/)
CU
o
cr>
vO
00
<r
o
o
i— 1
vO
uo
o\
o
CO
o
oo
iO
-o
a
ra
-a-
rH
CO
.-1
CM
rH
en
o
CO
rH
rH
CM
H
CO
r-l
CO
H
CO
iH
CO
CM
rH
r-l
00
o
rH
o
rH
o
o
rH
rH
o
rH
O
rH
>.
*->
u
C
<u
0)
a
e
■H
p.
.C
•H
o
CO
3
CO
<3\
m
r»
CO
St
O
cr.
<T\
rH
m
00
o
o
CO
£
w
-a-
X
in
UO
00
m
<-h
O0
rH
CM
CM
m
CM
CO
CO
o
CM
in
CO
o
o
r-l
CM
rH
CO
rH
to
4J
CN
CM
vO
TJ o
CO
•
•
•
C 0
X
CM
r-
r-~
ea T3
PH
M3
in
<r
O
rH
rH
t-\
ca u
rH d4
*
O TJ
•H 41
6 "-<
CU rH
XI rH
rH
rH
oo
u <:
CO
^r\
1
rsl
cn
in
oo
CM
CO
CO
<3"
o
o
On
On
CO
O*
r-
Or.
O
O
CO
cr.
vO
oo
oo
o
O
r~.
00
O
nO
rH
00
O
O
vO
oo
rH
rH
O
CM
CM
cr>
CM
co
CO
00
rH
CM
O
o
rH
CO
H
r-».
■<r
rH
v£> 00
•o
CM
X
■u .
CO
P*
G
0)
O
1-1
4J
X
CM
<u
<!
0)
CM
CM
iH
X
rD
CO
H
00
vO
CM
CM
co
M3
m
CM
o
o
o
o
O
ON
o\
o
— i
00
CM
ON
CO
vO CO
CM
00
in
On
CO
m
o
vO
00
ON
o
1-t
o
rH
o
rH
ON
ON
00
MD
UO
o>
vD
m
oo
O
o
CO
m
ON
o
co
IM
3
•u
(0
-o
o
o
Cm
X
«tf
O
oo
CM
\0
<*
r-~
m
CM
m
rH
CM
o
m
O
CM
ON
m
on
CM
oo
cN
oo
CM
00
CM
OO
CM
CO
oo
X>
CO
in
en
00
ON
o
o
rH
rH
rH
CM
rH
rH
CO CM
o
CO
co
00
CO
o
o
00
00
CO
co
vD
m
o
00
vO
00
00
00
oo
o
f-t
00
o\
o
rH
<T>
o\
CO
m
on
m
On
m
uo
sO
m
On
m
ON
00
m
On
ON
in
cr.
o
vO
vO
CTn
CM
ON
CO
vD
ON
*
-^ m vo
vO vD M3
CTl CT» CT\
vO
en
3
O
a>
c
to
00
c
co
NO
CO
00
00
o
co co
oo
on
as
oo
oo , no
O
on
on
oo
o\
vD
00
CO
on
o
o
00
NO
o
0)
u
CO
■H
co
oo
r-
r-~
ON
o
NO
~3-
<r
co
on
r-l
oo
ON
o
rH
CM
CN
rH
CN
rH
rH
i-l
-3-
CM
ON
00
r-l
as
CN
r--
<-*
NO
m
o
o
o
rH
m
CO
to
CO
■H
u
<D
u
CO
S
<u
3
W>
t)
!-■
<U
p-.
CM
o
en
i-i
o
00
\0
CM
CO
o
00
rH
ON
oo in m
CM ON
ON
CM
o
On
vO
on
m
on
-3-
o
o
<r en
o O
on on
vD
CO
vr
vO
CM
CO
r-«.
r-
CO
vo
vO
CN
CM
CO
CO
CO
CO
sr
UO
NO
00
00
1-^
ON
o
o
CO
CM
a.
•-N
CO
•a
u
—
a
4J
in
e
o
T3
o
c
N— '
CO
CN
u>
1
0)
<
W
a
<D
tH
i-i
.O
CO
CO
•u
H
at
2
vD
ON On NO 00
CO
00
vO
Pm
m
o
^3-
ON
CO
r-~
ON
p~
o
CM
r^
ON
m
o
CM
00
co
-3-
ON
ON
o
o
ON
00
ON
o
o
as
nO
h as
ON
vO
vO
CO
r^
m
-T
in
rH
m
o
CM
vO
D p»
H rH
rH
CO
CO
~T
m
CM
ON
m
vO
rH
as
rH
-3"
CO
ON
ON
o
o
iO
rH
ON
in
cu
3
00 CO OO
NO nO
CO ON
O co m
~T
rH
CM
O
CM
oc
CO
SO
o
00
vD
CM
o
P--
OO
<t
CO
CO
m
[■^
^3"
CM
rH
r-H
o
O
O
O
ON
ON
r-l
r-l
rH
r-l
rH
r-l
r-l
r-H
rH
rH
rH
rH
rH
CO
u
a>
c
•H
m
CO
in
<r
ON
CO
rH
P^
CO
r^
co
o
r»
o
o
CO
CM
rH
CM
rH
CO
rH
NO
rH
-3"
CM
rH
CM
NO
CM
ON
CO
rH
m
00
m
o
-3-
00
o
o
rH
ON
CO
CO
^3-
m
NO
r-
CO
ON
o
rH
CM
CO
-3-
*
m
NO
r-»
m
m
m
in
m
m
m
MO
vO
NO
NO
vO
NO
vO
vO
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON
CO
I
<
iH
CO
H
>R
CO
■u
Vj
O
i"
M
<L>
CO
<u
C
cO
CX
CO
—>
4-4
o
cu
u
CO
.c
CO
oo
s w(\it^trii-imir\scoNso\^a\w
.-I r-1
Or-tCT\OOrO^CVir-i t^.(7\CO^O0Or-<
S~ ^j- mo^MD co ia s s in vo vo lovdco t—
w h m h wintDomvod-wmsN
vO •
2: SSS^VDOOMlflVO CMHCUCMCMfO
t— lr-1 r-4 .— I r-i r-4 r-4 r-4 1— t r— I 1 — I
o *-* [--cocno n m o mo s a\ o\ m
LT\
s whh wmsin^D^oocosaico^
r_4r_(,_|r_tl_|r_)r_lrHr-|i-ir-4r-<r-4i-4r-4
C—sJ-sf-Ob-OOCOOOOMDOA-st-t^OO
St-
S MDt-LPvLnvo>-io>cTiOror-ioooco<T\
r- 1 r— I i— I 1— I r-4
o\ a\ m <-* ooiA>-t ctvCO m in a\ o cm oa
ro
S oon4-in4-in^ininininmif\inin
^ co n m vo oc^oinr-<r-4o^-t>-t---
cu t^-Lapocooooc^-c«-voro<,r>'-<oo>>-
S CMCVJCMCM«HCUr-lr-4r-4r-«f-4iHr-4
o\ cu cn n tn m no cm ur\ h r-4 m o v«o ir\
,-»
2 msmstot-ooNHon^soosin
CVI CU CVl t — t 1 — l 1 — I 1 — * t — i 1 — I 1 — i 1 — I 1 — 1 1 — I 1 — i i — 1
-69-
cc
r-4 -a
CO C
•U CO
O CO
H 3
B
4J
t— sf- O .=}-
co o ro 00
\D ^r -=f t~-
LT> r-4 tMtM^Si^CTiCUS
C\IC?\0OPO0J0O.=»-r-IOCO
r-4 st- r-4 j* LncotnoN-o
ct\o\'-1 ct\ c> n*"* ca --i ouovo t>- cr\ oj 00
o eft c— cu co 00 (?\ ca h mcornvo C\Jv£>
d-fO*wc\ioin4-coy3Ncr\'~|tnu3
<\lCVlC\Joos^^oc,os^^r\^nu^^--coo^'--,
u oo^mvof-oocTvOi-tcMrojd-Lnvoi^-
co inmininininin^ujvD^^^^^o
qj a\c^c7\C7NC^o>CT\o^c^cr\c^cTsa\a\a>
>1 ,_),_( ^.-(.-^r-tr-li-^r-li-ir-lr-irHr-lr-l
X
o c\ <-* \o \o <j\ en o m oj oo o o co vo
.=*• CO in in lf\ ^d co oo co co c— c— y) vo h
x°
O t— VO 0\ iTl rl r-i
in 4- 4- if Jt 4- *
cMocoovor-Hcr^co
.=t-.=f-cOitcococMCM
VO CM [V if
\R
CO
■U
u
o
•
cu
T3
X
w
4J
C
0)
O
w
O
CD
>^
c
CO
co
a
1
cO
<
►->
CO
u-i
r-l
o
CO
cu
H
L|
fl
s:
CO
LO
X
in
iC\
(Tl
CO
r-t
CO
r-l
if
CO
m
t-
00
o
CO
t>
r-l
r-l
r-l
r-i
i-l
r-l
r-l
r-l
r-l
■—I
H
r-l
CM
r-l
r-4
CO
0)
3
CO
cr.
m
•=t
La
CM
o
VO
(T\
r-i
r-l
O
CO
CO
co
r-l
— f-
CO
X
j4-
OJ
CVJ
a\
<M
CM
CO
Cm
VO
m
t-
o\
r-l
co
CM
•H
r-l
r-l
r-l
r-l
CM
CM
CM
CM
CM
CM
CM
CM
CO
CO
it
CO
3
O
•H
U
oS
vO
m
i-l
VO
c*-
r-l
r-l
m
CO
co
CO
t-
m
CO
vO
>
en
**^>
X
m
m
m
if
if
m
m
j4-
zt
m
m
m
vo
vO
vo
r-l
0
c
c
<
<-i
CO
CO
00
m
o
CO
CM
co
VO
a\
^
t>-
O
CO
t
CM
CO
X
VD
o
r-~
«si*
LO
r-l
£
O
is-
in
CM
r-l
co
CO
VO
°,\
en
if
co
CO
CO
CO
CO
CM
CM
CM
CM
r-i
r-l
r-l
•rl|
X
-Cr>
CO
X)
c
co
3
O
H
^ VO CO oc
0\ IS- vO vC
VO
C\ CM CO CS CTk CO CO r-i C\ VO
is t-- vo m VO in 4 4 CO CO
CO VO O VI") CO
if CO O CO r-l
CO CM VO vO O
o cm r- vo c\ vo
VO 0> CO ON >n r-l
in 4 m m r-i r-t
o\
CM r-l CM
4 CA N
o- co in
it CT\
o
o
00
vo
vO
it
m
vo
CM
co
1-1
VO
1-1
t- CM
r-l
o
m
t-
m
m
CO
r-l
in
s-
m
IS
if
CM VO
O
in
00
co
if
o
CM
o\
.=!-
vo
i»-
ts
if
CMCMCMCMCOifititmvOOOONO
•u
CO
•H
U
R)
■U
CO
CJ
•r-l
6
O
c
o
o
w
c
cd
a,
ct)
•-1
u-i
O
.y
G
CO
s-i cOij-mvor-oo^Or-icMcOitmvor—
co inmLnmmmmvovovovovovovovo
JH r-li-lr-lr-lr-lr-lr-lr-lr-lr-lr-lr-lr-lr-lr-l
CU
a
u
3
O
CO