View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

MOD ELLI NG U.S. SERVICES TRA DE FLOWS:
A COINTEGRATION-ECM APP ROA CH

by
Juann H. Hung and Sandra Viana

Federal Reserve Bank of New York
Research Paper No. 9518

August 1995

This paper is being circulated for purposes of discussion and comm
ent only.
The contents should be regarded as preliminary and not for citatio
n or quotation without
permission of the author. The views expressed are those of the
author and do not necessarily
reflect those of the Federal Reserve Bank of New York or the
Federal Reserve System.
•

Single copies are available on request to:
Public Information Department
Federal Reserve Bank of New York
New York, NY 10045

Modelling U.S. Services Trade Flows:
A Cointegration-ECM Approach

by
Juann H. Hung•
Sandra Viana
July 15, 1995
International Macroeconomics Function
Federal Reserve Bank of New York
33 Liberty Street
New York, New York 10025

* The authors thank colleagues at FRBNY, especially Howard Howe, Tom Klitgaard,
and Jim Orr for helpful comments. The opinions expressed in this paper are those of the
authors and do not necessarily represent those of the Federal Reserve Bank of New York
or
of the Federal Reserve System.

Modelling U.S. Services Trade Flows:
A Cointegration-ECM Approach

Abstract

The U.S. service surplus soared from near zero in 1985 to about
$60 billion in 1992,
offsetting about two thirds of the goods trade deficit. Could this
merely reflect improvement
in data collection? Or does this mean U.S. services industries
are more competitive
internationally than goods industries? Is the services surplus likely
to continue to rise?
This paper estimates a forecastable model of U.S. services trade
to address the above
questions. We find that data improvement actually had a negat
ive net impact on the services
surplus, since it affected imports more than exports. Instead,
the surge in the services surplus
was mainly due to strong foreign growth and, to a lesser exten
t, dollar depreciation. An
increase in either outward or inward foreign direct investment
asset (FDIA) has a significant
and positive impact on both exports and imports of other privat
e services, but has only a
modest net effect on the U.S. services balance. Thus, the outloo
k for the U.S. services
balance largely depends on the growth prospect of foreign econo
mies.

Keywords: Services trade; Tourism trade; Other private servic
es; Royalties and license fees;
Merchandise trade; Foreign direct investment

Modelling U.S. Services Trade Flows:
A Cointegration-ECM Approach
Juann H. Hung

Sandra Viana
I. Introduction
As services industries grow more important to natio
nal economies, they also feature
more prominently in international transactions and nego
tiations.' U.S. services industries, in
particular, have complained loudly in recent years abou
t foreign trade barriers. Despite the
perception of trade barriers, however, the U.S. servi
ces surplus has soared since 1985, both in
nominal and real terms (Charts I). Rising from near
zero in 1985, the services surplus
reached about $60 billion in 1992, offsetting about
two-thirds of the merchandise trade deficit.
Since l 992, the services surplus has managed to stabi
lize around $60 billion, while the
merchandise trade deficit has deteriorated sharply as
a result of weak foreign demand and a
robust U.S. economy.
In contrast to merchandise trade flows, which are the
subject of voluminous studies,
U.S. services trade flows have received relatively little
empirical analysis. 2 Prior to the mid1980s, research on U.S. services trade may have been
scarce because of the noneventful
movement and small magnitude of such trade. The
unprecedented surge in the U.S. services
surplus since 1985, however, has changed the scena
rio completely. Given the present

For example, the subject of removing barriers to trade
in financial services featured
prominently in the Uruguay Round world trade nego
tiation and continues to be a central issue
confronting the World Trade Organization.
1

2

The services model run by the staff at the Board of
Governors of the Federal Reserve
Systein is one of the rare exceptions. See Helkie and
Stekler (I 987) and (1992).

•

magnitude of the services surplus, a comprehensive study of services trade flows enhances not
only our overall understanding of U.S. international competitiveness, but also our ability to
forecast U.S. external balances.
This paper uses data since the advent of the floating exchange rate (1973) to estimate
a forecastable model of real nonmilitary and nontransportation services trade between the
United States and the rest of the world. The services account consists of five components:
military, transportation, tourism, other private services, and royalty and license fees (Table I).
Among them, tourism, other private services, and royalty and license fees are the fastestgrowing components. Consequently, our study focuses on identifying the determinants of
tourism, and other private services (including royalty and license fees). 3 Military and
transportation services are briefly described in the appendix.
One question often raised regarding the rapid rise of the services surplus has been
whether it mainly reflects better data collection. · There is no question that, because of
improvements made by the Bureau of Economic Analysis (BEA) in recent years, services
trade data has frequent breaks. Unlike merchandise trade data, which are collected when
goods enter and leave a country, services data largely rely on surveys or estimates and thus
are subject to frequent revisions. Moreover, these revisions tend to be carried back over a

3

Because the distinction between other private services and royalty and license fees is
not clear in some areas, we decided to combine the two subaccounts in our analysis. For
example, beginning in 1986, certain management fees received from or paid to unaffiliated
foreigners--amounting to less than $25 million for receipts and less than $5 million for
payments--have been removed from royalties and license fees and included in other private
services.

2

limited number of years, thereby causing data discontinuity. 4 On the whole, the revision of a
service subaccount tends to expand its scope of coverage, causing an upward break.
To account for the impact of data improvement, we examine the BEA's data
documentation and include a dummy variable for each data break in our regression analysis.
Our analysis uses the two-step (cointegration and Error Correction Model (ECM)) approach
suggested by Granger and Engle (1978) to avoid spurious regressions, since both (real)
service exports and imports appear to be nonstationary variables. We find that tourism trade
is primarily driven by economic activities and the real exchange rate. The movements in
other private services are mainly a function of economic activities, the real exchange rate,
outward direct foreign investment assets (FDIA), and inward FDIA. Overall, our models fit
the data very well within sample and out of sample.
Data improvement is estimated to have applied more to imports than to exports.
Thus. it has had a negative net impact on the services surplus. Had the data coverage and
estimation method remained the same as in 1982 over the entire sample period, the services
surplus rise would have been even greater than what is reported. Once data breaks are
accounted for, the surge in the U.S. services surplus since 1985 appears to be mainly the
result of strong foreign growth and, to a lesser extent, dollar depreciation. An increase in
outward or inward FDIA has a positive impact on both exports and imports of other private
services. Consequently, the net effect of inward and outward FDIA was much more modest
than that of economic activity, even though inward FDIA grew dramatically in the second

4

Typically, in June of each year, estimates of U.S. international transactions are revised
to incorporate new source data, improved methodologies, and changes in definitions. In
addition. a new benchmark survey is usually conducted every five years.
3

half of the 1980s.
The next section of the paper describes the stylized devel
opments in the fastestgrowing components of services trade since the mid-1980s.
Section ill estimates the
fundamental determinants of the trade flows of tourism and
other private services (including
royalty and license fees). Section IV presents conclusion
s.

II. The Stylized Facts

This section examines developments in the three fastest grow
ing service components
since the mid-1980s: tourism, other private services, and
royaliies and license fees.
II.A. Tourism5
On the export side, tourism has surpassed all other servic
es components since 1986,
accounting for about 40 percent of total service receipts
in 1993 (Table 1). The rise in
tourism imports has been somewhat less spectacular. Neve
rtheless, tourism imports remain
the largest component of service imports since the late 1970s
, accounting for roughly 40
percent of total service imports in 1993.

In net, tourism has been the fastest growing component of the
services balance since
1986, reaching $22 billion, or 38 percent of net U.S. servic
e exports in 1993 (Table 1).
According to reported data, however, the prominence of
net tourism is a relatively recent
phenomenon. Until 1989, U.S. net tourism had been runni
ng a deficit (Chart 2).
The rise in U.S. tourism exports occurred mainly vis-a-vis
Western Europe, Canada,
Japan, and Mexico (Table I). In 1986, Japan was the only
country with which the United
~

The tourism trade is the sum of two accounts: travel and
passenger fare.
4

States ran a bilateral tourism trade surplus. However, West
ern Europe and Canada have been
catching up rapidly since the late 1980s. In fact, the Unite
d States has been running a
tourism surplus with the other three regions as well since
I 992.
II.B. Other private services
"Other private services" (OPS) is a conglomerate account
that covers transactions with
both affiliated and unaffiliated foreigners in services that
are not related to either tourism,
military, transportation, or royalties and license fees (Tabl
e 3). 6
Net OPS was the largest surplus component of U.S. servic
es balance before the late
I 980s; however, mainly because of the surge in net touris
m, its share of total services balance
has gradually declined. Nevertheless, net OPS continues
to account for about 39 percent of
total services balance in 1993, exceeding all other servic
es components.
Net OPS has continued to rise steadily since the late 1980s
, albeit not as strikingly as
net tourism. This rise reflects the persistence of net expor
ts to unaffiliated foreigners. Most
of the unaffiliated accounts have run a surplus since I 986,
with telecommunications,
insurance, and advertising being the only exceptions. Net
exports of the "business,
professional and technical services" account have been the
fastest-growing since 1986,
accounting for roughly half of total net exports to unaffiliate
d foreigners in 1992. Net exports
of education services have also risen steadily since 1986.
In comparison, net exports of
financial services, which grew solidly during 1986-89, have
declined modestly since 1989.
6

Exports of "other private services" to affiliated foreigners
refers to the services of
tangible property supplied by U.S. parent companies to
their foreign affiliates and by U.S.
affiliates to their foreign parents. Exports of other priva
te services to unaffiliated foreigners
refer to a host of services listed in Table 3: education, finan
cial services, insurance,
telecommunications, and business, professional, technical
services.
5

Net OPS exports to affiliated foreigners have also been steadily running a surplus
,
even though their importance has been declining (relative to net exports to unaffil
iated
foreigners). It is not surprising that U.S. parent companies' exports to their foreign
affiliates
have constantly exceeded their imports from their foreign affiliates. Interestingly,
foreign
companies' affiliates in the United States have also been net exporters of service
s to their
parents. U.S. parents' net exports to their foreign affiliates accounted for about
80 percent of
total net OPS exports to affiliated foreigners since 1986, while U.S. affiliates'
net exports to
their foreign parents made up the remaining 20 percent.
11.C. Royalties and license fees
Similar to the "other private services" account, the "royalties and license fees"
(RLF)
account covers transactions with affiliated and unaffiliated foreigners in a numbe
r of services
(Table 4 ). 7 Net RLF ran a steady and solid surplus not only throughout the 1980s
but also
into the 1990s. Over the 1992-93 period, net RLF reached an average of $14.6
billion -about 26 percent of the U.S. services balance.
Over the 1990-92 period, net receipts from affiliated foreigners accounted for
about 83
percent of total net RLF, while net receipts from unaffiliated foreigners accoun
ted for the
remaining 17 percent. Table 4 shows that in 1992 the bulk of net receipts from
unaffiliated
foreigners (71 percent) came from industrial processes.
Foreign parents have generally been net exporters of royalties and license service
s to

7

Receipts from affiliated foreigners stem from the sale and rental of proprietary
rights
and intangible assets supplied by U.S. parents to their foreign affiliates and by
U.S. affiliates
to their foreign parents. Receipts from unaffiliated foreigners stems from sales
of patterns,
trade secrets (related to industrial processes), and copyrights (for books, records
and tapes),
royalties on broadcasting and recording of live events, and franchise fees.
6

their U.S. affiliates. Similarly, U.S. parent companies have been
net exporters to their foreign
affiliates. However, net RLF received by U.S. parents (from their
foreign affiliates) have far
exceeded that received by foreign parents (from their U.S. affilia
tes). Indeed, mainly because
of the substantially higher net receipts by U.S. parents (than that
by foreign parents), the
United States has historically run a persistent surplus of royalt
ies and license fees.
Given the less-than-ideal state of intellectual property right protec
tion so far, it is
perhaps of no surprise that net RLF on "books, records, and tapes"
have only registered a
dismal $0.1 billion over the.1989-92 period. More surprising
is that the United States has
been a net payee of "broadcasting and recording of live events
" fees to unaffiliated foreigners,
not a net recipient as commonly held. This puzzling phenomeno
n may have to do with the
fact that the sales of "broadcasting and recording of live events
" to foreigners are largely
transacted by foreign affiliates of U.S. companies. While the
size of these transactions is
potentially large. they are not recorded in the balance of paym
ent accounts because they are
not cross-border transactions.

III. Empirical Analyses
This section estimates the fundamental determinants of trade flows
of tourism and
other private services (including royalty and license fees). Becau
se all the dependent and
independent variables are I( I) variables (Table 5), we use the
two-step estimation method
suggested by Granger and Engle ( 1987) to avoid spurious regres
sion analysis.
In the first step, long-run equations are estimated using ordina
ry least squares (OLS)
regression. The residuals of these regressions are tested to see
if they are stationary. If the

7

null hypothesis that the residuals are nonstationary can be rejected, the long-ru
n equation is
considered to be cointegrated. 8 In the· second step, after long-run cointegrated
equations are
estimated, an error correction model (ECM) is estimated to study short-run dynam
ics between
the variables. In each regression, we use a constant dummy for each documented
data break.
III.A. Tourism
Demand for tourism is assumed to be determined primarily by economic activity
and
the real exchange rate, much in the same way as demand for goods. That is,
demand for
exports is expected to increase as foreign economies grow and/or as the real dollar
depreciates; demand for imports is expected to increase as the U.S. economy
grows and/or the
real dollar appreciates.
Regression analyses of both real tourism exports and imports are conducted over
the
1973Q4-1993Q4 period. To deal with data breaks, two constant dummies are
included in
both exports and imports regressions. The first one, for the 1984Ql-1993Q4
period, accounts
for the effect of new estimates introduced for passenger fares (for both payme
nts and receipts)
and travel (only receipts). 9 The second, for the 1991Ql-1993Q4 period, accoun
ts for the
effect of improved estimates for travel from Canada. 10
For real exports, two measures of the real exchange rate are used: the relative
export

8

In a cointegrating regression equation, the nonstationary dependent variable and
nonstationary independent variables drift together over time, so that the unexpl
ained
"residuals" of the regression equation are stationary over time.
9

10

See Survey of Current Business, June 1993.
See Survey of Current Business, June 1994.
8

price (= U.S. service export deflator in foreign currency terms/foreign
GDP deflator) 11 and the
relative CPI (foreign CPI/US CPI in foreign currency terms). The regres
sion results using
both measures are reported in Table 6a. Regression I (first column) shows
the regression
results using the relative export price, while regression 2 (second colum
n) shows the results
using the relative CPI.
Table 6a shows that different measures of the real exchange rate do
not make a
marked difference in the regression results. Both long-run regressions
are reasonably
cointegrated: one has an Augmented-Dickey-Fuller (ADF) statistic of
4.92, the other has an
ADF of 4.67. Both equations have an Adjusted-R2 = 0.99. A I percen
t increase in foreign
domestic demand raises real tourism exports by about 2.2 percent in
regression 1, and by
about 2.3 percent in regression 2. A I percent increase in the relativ
e export price lowers real
tourism exports by about 0.42 percent, while a I percent decrease in
the relative CPI (or real
dollar appreciation) lowers real tourism exports by about 0.36 percen
t, in the long run.
For forecasting purposes, we also estimate the error correction mecha
nism (ECM) for
both regressions. To see which regression has better forecasting perfor
mance, we reestimate
each equation using data over the l 973Q4-l 99 I Q4 period, and use the
resulting coefficient
estimates to make out-of-sample forecast (over !992Ql-1993Q4). We
then compute both the
(out-of-sample) root-mean-square error (RMSE) and Theil's inequality
coefficient (Theil's

11

Ideally, one should use the relative tourism export price( = U.S. tourism
export deflator
in foreign currency terms/foreign GDP deflator). However, tourism
export deflator is not
reported.
9

U)

12

for each model--the combination of the long-run equation and ECM. The results indicate

that the two models' forecasting performances are equally good by Theil' s U standard,
although model 2 forecasts slightly better than model I by the RMSE standard. The insample and out-of-sample forecasts of model 2 are plotted in the upper panel of Chart 3.
For real imports, two measures of the real exchange rate are also used: the relative
import price (= U.S. service import deflator/U.S. GDP deflator), and the relative CPI (foreign
CPI/US CPI in foreign currency terms). The regression results using both measures are
reported in Table 6b. Regression 1 shows the regression results using the relative import
price, while regression 2 shows the results using the relative CPI.
Table 6b shows that, in contrast to tourism exports, using different measures of the
real exchange rate does make a noticeable difference in the regression results of tourism
imports. Although the two regressions do not differ notably in terms of cointegration
statistics and goodness of fit, 13 the coefficient estimates of the fundamental variables in the
two regressions are quite different. A I percent increase in U.S. domestic demand raises real
tourism imports by about I percent in regression I, and by about 0.8 percent in regression 2.
A 1 percent increase in the relative import price lowers real tourism imports by about I
percent, while a I percent increase in the relative CPI (or real dollar depreciation) lowers real
tourism imports by about 0.4 percent, in the long run.

12

Theil's inequality coefficient (Theil's U), which always falls between 0 and I, can be
considered as the normalized RMSE. If U = 0, there is a perfect fit. If U = I, the predictive
performance of the model is as bad as it possibly can be.
13

Regression 1 has an ADF = -5.62 and adjusted-R2 = 0.99, while regression 2 has an
ADF = -5. 17 and adjusted-R2 =0.98.
10

Again, we estimate the ECM for both import regression
s, and reestimate each equation
using data over the I 973Q4- I 99 I Q4 period to forec
ast over I 992Q 1- I 993Q4. Both the
RMSE's and Theil's U's indicate that model 2 forec
asts slightly better than model 1. The insample and out-of-sample forecasts of model 2 are plott
ed in the lower panel of Chart 3.
To ascertain why the two measures of the real exch
ange rate make a greater difference
on the import regressions, we estimate the following
two typical exchange-rate-pass-through
equations:

(1) The tourism export price equation:
ln(PxS) = 0.1 I + 0.98 ln(S) + 1.04 ln(P) -0.02
ln(P' )

( 1.38)

(62.98)

(34.40)

(0.96)

Adjusted-R 2 = 0.999; Phillips-Perron (Z,) = -3.34, ADF
= -2.97.
(2) The tourism import price equation:
ln(Pm) =

1.78 - 0.52 ln(S) + 1.03 ln(P) + 0.13 ln(P' )
(14.8) (21.6)
(21.89)
(4.81)

Adjusted-R 2 = 0.995; Phillips-Perron (Z,) = -5.16, ADF
= -4.95.
Where.
pxs=

U.S. service export deflator in foreign currency terms
;

S=

the effective foreign currency/dollar exchange rate based
on tourism export weights;

P=

U.S. GDP deflator;

p'=

tourism-export-weighted foreign GDP deflator; and

pm=

U.S. service import deflator.

11

The tourism export price equation suggests that, holding U.S. and foreign prices
constant, a I percent dollar appreciation will raise the relative export price by about I percent
over time. Such an estimate suggests that nominal exchange rate changes are nearly
completely passed through to tourism export prices in the long run. It also helps explain why
the relative CPI and the relative export price are estimated to have roughly the same impact
on tourism exports over time.
The tourism import price equation suggests that, holding U.S. and foreign prices
unchanged, a I percent dollar appreciation will raise the relative import price by about 0.5
percent over time. The low exchange-rate-pass-through coefficient in the import price
equation helps explain why the relative CPI is estimated to have much less impact on tourism
imports than the relative import price.
Several features of the tourism trade regressions are worth noting. First, foreign
income elasticity for exports is twice as large as U.S income elasticity for imports in tourism
trade. This favorable asymmetry of income elasticities suggest that real net tourism exports
will have a tendency to grow unless the U.S economy sufficiently outgrows its trading
partners' economies.
Second, compared with merchandise trade, tourism trade seems to be less sensitive to
exchange rate changes. 14 Tourism exports, however, respond much more strongly to foreign
growth than merchandise exports. Indeed, the favorable income elasticity asymmetry in
tourism trade is the opposite of the case in merchandise trade, which is known to have a

14

The average estimate is about I for both U.S. merchandise export price elasticity and
import price elasticity.
12

larger income ell)sticity (of demand) for imports than for exports. 15
Third, U.S. import price elasticity is more than two-times larger than export price
elasticity, suggesting U.S. tourists are much more price conscious than their foreign
counterparts. However, because export prices (in foreign currency terms) respond
more
completely to dollar exchange rate changes than do import prices, the net effect
of a dollar
exchange rate change on tourism exports is roughly equal to that on tourism imports
(in
absolute terms).
III.B. Other private services
This subsection estimates equations of real other private services (including royalty
and license fees) exports and imports. For easy reference, real other private service
s
(including royalty and license fees) will be referred to as OPS in the subsequent
discussion.
OPS had three major data breaks since 1973. The first occurred in 1982QI,
when
receipts and payments of royalty and license fees began to be reported on a gross
basis rather
than on a net basis. 16 The second occurred in I 986Q I when (I) coverage of
education
services and business, professional, and technical services were included for the
first time,
and (2) certain portfolio incomes were reclassified to other private services. 17
The third
occurred in I 991 QI, when benchmark survey results led to new estimates for
other private
services receipts and payments. 18
15

See Goldstein and Khan ( I 985) for a survey of price and income elasticities of
demand
for U.S. merchandise exports and imports.
16

See Survey of Current Business, September 1994, footnote 5.

17

See Survey of Current Business, June 1989.

18

See Survey of Current Business, September 1994.
13

Among t~e three data breaks, the first one is potentially the most difficult to deal
with. We thus run both real OPS exports and imports regressions over the 1982Ql-1993Q4
period. A constant dummy is used to account for each of the remaining two data breaks.
We assume that demand for OPS is a function not only of economic activity and the
real exchange rate, as in tourism, but also of outward and inward FDIA. OPS exports is
expected to be a positive function of both outward and inward FDIA. Growth in outward
FDIA is expected to increase OPS exports, since exports of professional, insurance, and legal
services may tend to rise as U.S. parent firms expand the facility and operation of their
foreign subsidiaries; growth in inward FDIA is also expected to increase OPS exports, since
U.S. affiliates of foreign companies may tend to export cutting-edge technology and services
know-how to their companies. 19 By the same token, OPS imports are also expected to be a
positive function of both outward and inward FDIA. 20
Consequently, we run four regressions for both OPS exports and imports (Tables 7a
and 7b ). Each regression contains ~he same set of dummy variables, but a different set of
other independent variables. On the export side, regression 1 includes only the relative export
price and foreign domestic demand as explanatory variables. Regression 2 has the same
specification as regression 1 but uses the relative CPI rather than the relative export price as
the real exchange rate measure. Regression 3 includes outward FDIA and inward FDIA, in

19

As discussed in Section 11.B, exports to affiliated foreigners is the sum of(!) U.S.
parents' exports to foreign affiliates, and (2) U.S. affiliates' exports to their foreign parents.
20

As discussed in section 11.C, imports from affiliated foreigners is the sum of (1) U.S.
parents' imports from their foreign affiliates, and (2) U.S. affiliates' imports from their
foreign parents.
14

addition to the relative export price and foreign dome
stic demand. Regression 4 drops
foreign domestic demand but keeps outward and inwa
rd FDIA and the relative export price.
The regression results of real OPS exports are repor
ted in Table 7a. None of the four
regressions appear to be cointegrated, even at a 20 perce
nt significance level. These poor
cointegration results suggest that the relationships being
analyzed have not been sufficiently
stable. Of course, the relatively short sample period
may have contributed to the finding of a
lack of cointegrated equation for OPS exports. In any
case, the high adjusted-R2 in all four
regressions suggest that the four included variables
-- foreign domestic demand, the real
exchange rate, outward FDIA, and inward FDIA -are pretty much the main variables driving
the movements in real OPS exports, once data impr
ovements are taken into account.
Regressions I and 2 show that different measures of
the real exchange rate make little
difference on the estimation results. Both long-run
regressions are not cointegrated, and have
an Adjusted-R2 = 0.98, A I percent increase in forei
gn domestic demand raises OPS exports
by 1.57 percent in regression I, and by 1.62 percent
in regression 2. A I percent increase in
the relative export price lowers OPS exports by abou
t 0.23 percent, while a I percent
decrease in the relative CPI (or real dollar appreciatio
n) lowers OPS exports by about 0.22

percent, in the long run.
Regression 3 shows that, when inward and outward
FDIA are included as regressors
along with the relative export price and foreign dome
stic demand, foreign domestic demand is
estimated to have a wrong sign: A I percent increase
in foreign domestic demand decreases

15

OPS exports by 1.11 percent. 21

The coefficient on the relative expo rt price is estim
ated to be

some what great er (=0.31) than in the case of regre
ssions I and 2. The coefficient estim ate on
outward FDIA is estimated to be 0.61, slightly highe
r than the coefficient estim ate on inwa rd
FDIA (=0.56).
Regression 4 drops foreign domestic demand from
the list of regressors included in
regression 3. The coefficient estimates on outw ard
FDIA and the relative expo rt price in
regression 4 do not differ much from that in regre
ssion 3. However, the coeff icien t estimate
on inward FDIA in regression 4 (=0.25) is less than
that in regression 3 (=0.56).
To comp are the forecasting performances of these
four regressions, we estim ate an
ECM for each long-run exports equation. We then
estimate RMS E and Theil's U for each
model--the combination of the long-run equation
and ECM. Both in terms of RMS E and

Theil's U, model 2 appears to outperform the other
three models. The in-sample and out-ofsample forecasts of model 2 are plotted in the uppe
r panel of Chart 4.
On the import side, we also run four regressions
that parallel those on the expo rt side.
The regression results of real OPS imports are repor
ted in Table 7b.
Regressions I and 2 show that different measures
of the real exch ange rate do make a
noticeable difference on the estimation results. Regr
ession I is not cointegrated at a 15
percent significant level, while regression 2 is coint
egrated at a 10 percent significant level.
A I percent increase in U.S. domestic demand raise
s OPS imports by 2.47 percent in

Such a finding may be caused by the high correlation
s between foreign domestic .
dema nd and inward FDIA. In a regression of inwa
rd FDIA on a constant term and foreign
domestic dema nd, foreign domestic demand is found
to be positively correlated with inward
FDIA (coefficient estim ate= 3.7L I-sta tistic = 79.68
).
·
21

16

regression 1, and by 2.16 percent in regression 2. The coefficient estimate is
larger on the
relative import price (=0.78) than on the relative CPI (=0.43), but both are estimat
ed to have
a wrong sign: A real dollar depreciation is estimated to increase, not decrease,
OPS imports.
Regression 3 shows that, when inward and outward FDIA are included as regress
ors
along with the relative import price and U.S. domestic demand, the relative import
price
becomes insignificant. As a result, the coefficient estimate on U.S. domestic
demand
becomes much smaller. (=0.59) than in regression I. The coefficient estimate
on outward
FDIA (=0.87) is estimated to be much higher than on inward FDIA (=0.21).
Regression 4 drops U.S. domestic demand as a regressor, but keeps all the other
regressors included in regression 3. As in regression 3, both inward and outwar
d FDIA are
found to have a positive impact on OPS imports; the coefficient estimate on outwar
d FDIA is
noticeably greater in regression 4 (=1.14), while the coefficient estimates on inward
FDIA are
nearly the same as in regression 3.
Two results of regressions 3 and 4 are worth noting: (1) both inward and outwar
d
FDIA have a positive impact on OPS imports, and (2) the coefficient on the
real exchange
rate has a right sign when inward and outward FDIA are included. These results
help explain
the finding that the real exchange rate has a wrong sign in regressions I and
2. A dollar
depreciation may have three effects on OPS imports. The first is the direct and
negative
(price) effect: A dollar depreciation will raise the import price and thus lower
U.S. demand
for OPS imports. The second is the indirect and positive effect: A dollar deprec
iation may
cause inward FDIA to rise, thereby increasing OPS imports. The third is the
indirect and
negative effect: A dollar depreciation may cause outward FDIA to fall, thereby
lowering OPS

17

imports. The finding that a real dollar depreciation
has a positive impact on OPS imports
suggests that the net indirect effect was positive and
outweighed the negative direct effect
over the sample period. 22
To compare forecasting performances of these four
regressions, we estimate an ECM
for each Jong-run imports equation. We then estimate
the out-of-sample RMSE and Theil' s U
for each model over the 1991 Q4-93Q4 period. The
results show that model 2 appears to
outperform the other three models, both in terms of
RMSE and Theil's U. The in-sample and
out-of-sample forecasts of model 2 are plotted in the
lower panel of Chart 4.
We estimate the following two typical exchange-rate
-pass-through equations to find
out why the two measures of the real exchange rate
make a greater difference on the import
regressions. The specifications of both equations are
the same as in tourism, but the effective
nominal exchange rate and foreign GDP deflator are
constructed based on OPS trade weights.
(3) The OPS export price equation:
ln(P' S) = 0 015

(0.16)

+ 0.98 ln(S) + 1.17 ln(P) -0.12 ln(P' )
(65.94)

(16.97)

(2.39)

Adjusted-R 2 = 0.999; Phillips-Perron (Z,) = -3.53, ADF
= -3.14.

In a regression of inward FDIA on a constant term
and the real exchange rate, a 1
percent real dollar depreciation is estimated to be posit
ively correlated with inward FDIA
(coefficient estim ate= 2.69; I-statistic = 12.82). In
a regression of outward FDIA on a
constant term and the real exchange rate, a I percent
real dollar depreciation is also estimated
to be positively correlated with outward FDIA (coef
ficient estimate = 0.85; I-statistic = 9.67).
Inward FDIA grew 118 percent, while outward FDIA
grew 51 percent over the 1982-93
period.
22

18

(4) The OPS import price equation:
ln(Pm)

= 1.79 - 0.51 ln(S)
(16.38)

(27.17)

+ 1.00 ln(P)
(20.09)

+ 0.15 ln(P')
(4.85)

Adjusted-R2 = 0.997; Phillips-Perron (2i) = -6.44, ADP = -6.32.
Where variables are as defined for equations (I) and (2) except:
S =

the effective foreign currency/dollar exchange rate based on OPS export weights;

P' =

OPS-export-weighted foreign GDP deflator.

Equations (3) and (4) show that the elasticities of OPS export and import prices with
respect to exchange rate changes basically do not differ from those for tourism exports and
imports prices. This finding helps explain why the two measures of the real exchange rate
make little difference in both tourism and OPS exports, but result in noticeable difference in
both tourism and OPS imports.
Several features of the OPS trade regressions are worth discussing. First, both import
and export price elasticities are low--below 0.5 (in absolute terms), which is not only lower
than the price elasticities for merchandise imports and exports but also lower than those for
tourism exports and imports.
Second, a I percent increase in outward FDIA is estimated to cause U.S. OPS imports
to grow faster than exports. This unfavorable asymmetry, of course, may be a result of the
fact that U.S. exports have been larger than imports.
Third, the coefficient on inward FDIA in the imports equation is about the same size
as that in the exports equation. This finding suggests that a I percent increase in inward FDIA
will cause U.S. OPS imports to grow at about the same rate as exports. Given that U.S. OPS
19

exports were larger than imports in the early 1980s, this symmetry sugges
ts that the dramatic
growth in inward FDIA since the mid- I 980s may be partly responsible
for the recent surge in
the services surplus.
III.C. Identifying the forces behind the service sumlus surge
This subsection identifies the factors driving the services surplus rise
by netting out
each individual factor's direct contribution to the rise. To do so, we
use coefficient estimates
of regression I for tourism exports and imports (Tables 6a and 6b ),23
and estimates of
regression 4 for OPS exports and imports (Tables 7a and 7b). 24
The computation is done in two steps. First, we multiply the coeffic
ient estimate of
each variable with the actual growth rate of that variable over 1984-9
2 to see what
contribution each factor makes to the growth in each dependent variab
le -- namely, tourism
exports, tourism imports, OPS exports, and OPS imports. Second, based
on the share of each
component in exports and imports, we net out each factor's contribution
to the services
balance over the period.
Table 8 shows that strong foreign growth was the main driver behind
the surge in the
services surplus, followed by dollar depreciation. The seven major servic
es trading partners
grew about 28 percent over 1984-92, contributing 62 percentage points
to the 114 percent

23

We choose regression I (which uses the relative export price as the real
exchange
rate measure), rather than regression 2 (which uses the relative CPI as
the real exchange rate
measure), to avoid the complication arising from incomplete exchange-rate
-pass-through.
24

We choose regression 4 for three reasons: (1) it uses the relative export
price as the
relative price measure and thus avoids the complication arising from
incomplete exchangerate-pass-through; (2) it does not contain an economic variable that is
estimated to have a
"perverse" effect on demand for OPS: (3) it enables one to estimate the
"direct" effect of each
independent variable on OPS movements.
20

predicted growth in tourism exports, or 29 percentage
points to the 93 percent predicted
growth in total (tourism and OPS) exports. If we norm
alize the predicted growth in total
net exports to be 100 percent over the period, then
foreign domestic demand's
contribution to net exports amounts to 118 percenta
ge points.
The contribution of the 34 percent decline in the relati
ve export price amounted to
about 45 percentage points, while the contribution of
the 9 percent increase in the relative
import price equals about 26 percentage points. Toge
ther, the real exchange rate changes -mainly a result of the dollar depreciation over the perio
d -- contribute nearly 71 percentage
points to the (normalized) 100 percent rise of the
services surplus.
Because U.S. OPS exports have been larger than OPS
imports over the period, the
dramatic growth in inward FDIA (!02 percent) did
contribute modestly to the services surplus
surge, even though inward FDIA was estimated to have
roughly the same percent impact on
OPS exports as imports. For the same reason, the unfav
orably asymmetric percent impacts of
outward FDIA growth on OPS did not result in a majo
r drag on the net services exports.
The two major drags on net services exports were grow
th in U.S. domestic demand
and, surprisingly, data improvement. Mainly because
data improvement related to a larger
extent to imports than exports, it actually had a nega
tive net impact on the services surplus.

IV. Conclusion
This paper estimates a forecastable model of U.S. nonm
ilitary and nontransportation
services trade. Because the dependent and independe
nt variables appear to be nonstationary
variables, we use the two-step (cointegration and ECM
) approach suggested by Granger and

21

Engle (1978) to avoid spurious regressions. The mode
ls appear to fit the data very well both
within and out of the sample.
Using the coefficient estimates of the models and the past
grow th rates of the main
determinants of services trade flows, we find that the surge
in the services trade surplus since
the mid- l 980s is largely attributable to movements in
traditional macroeconomic variables.
Robust economic growth abroad and the substantial depre
ciation of the dollar in the secon d
half of the 1980s are the primary factors driving the rapid
rise in the services surplus.
Growth in either outward or inward foreign direct inves
tment asset has had only a modest net
effect on the services balance. The finding that data impro
vement has a negative net impa ct
on the services surplus suggests that the services surpl
us surge may in fact reflect a
comparative advantage in favor of the U.S. services indus
tries.

22

: Appendix: Transportation and Military Service Trade
Both transportation and military services command a relatively
small share of total
services trade, and have remained relatively stable over time
(Chart 2). Together, they run a
small deficit. The surge in the services surplus would have been
higher were these two
accounts not included.
Transportation
The transportation account includes receipts and payments for
services involving the
transport of merchandise goods via water, air, rail and pipeli
ne. The primary transportation
services are (1) freight charges, (2) operating expenses at foreig
n ports, and (3) vessel charters
and rentals.
Freight charges are incurred when the vessel operators and aircra
ft carriers of one
country carry goods to a foreign country and between foreign
ports. Also included are the
expenses of rail carriers for transporting goods between Canad
a, the United States, and third
countries and the costs to pipeline companies for moving gas
and oil from the United States
to Canada. Port services cover the expenditures of shipping,
air, and rail carriers at foreign
ports, including fuel, port call and cargo expenses, wages paid
to foreign residents, repair and
maintenance fees, and terminal services. Charter hire and rental
s are the expenses for the
charter of vessels and the leasing of aircraft and freight car to
foreigners.
An important factor in estimating transportation receipts and
payments is determining
the residency of a vessel operator, since it may differ from the
vesse l's country of registry.
U.S. companies, for example, commonly operate tankers that
are registered in flag-ofconvenience countries (for example, Panama, Liberia, Honduras),
either directly or indirectly
through foreign subsidiaries. If directly operated, U.S. comp
anies are treated as U.S.
residents. If indirectly operated, the foreign subsidiaries of U.S.
companies are treated as
foreign residents.
Military
The activity recorded in military services is based on who condu
cts the transaction,
not on what is traded. Commercial transactions between privat
e firms and governments
involving military type of goods or services are not included,
unless a U.S. military agency
participates.
The transfer of goods and services to foreign governments are
recorded as receipts,
and direct defense expenditures abroad are recorded as paym
ents. Because military
installations abroad are considered part of the U.S. economy,
the accounts include the receipts
and payments of U.S. military installations abroad. Receipts
are mostly transfers under
foreign military sales programs. Payments cover a broad spectr
um of activity: (1)
expenditures by U.S. personnel abroad, (2) wages paid to foreig
n residents, (3) construction
expenses, (4) foreign contractual services, (5) foreign goods
and services purchased for
military assistance programs, (6) NATO support project paym
ents, and (7) foreign goods and
services purchased by the U.S. Coast Guard.
23

Data Appendix
Tourism
U.S. real tourism export is nominal tourism exports divided by the service export deflator
(1985=100). Nominal tourism exports and the service export deflator are obtained from the
U.S. Department of Commerce (Survey of Current Business, various issues).
U.S. real tourism import is nominal tourism imports divided by the service import deflator
(1985=100). Nominal tourism imports and the service import deflator are obtained from the
U.S. Department of Commerce (Survey of Current Business, various issues).
Foreign Domestic Demand is a weighted index of the real domestic demand of the seven
major U.S. trading partners, based on share of tourism exports. Real domestic demand for six
trading partners is obtained from the Bank of International Settlements. Mexico's real
domestic demand is obtained from the Banco de Mexico. (Country shares of tourism exports
are shown in Table Al below.)
U.S. Domestic Demand is obtained from the U.S. Department of Commerce, Bureau of
Economic Analysis.
The relative export price is the U.S. services export deflator (in foreign currency terms)
divided by the foreign GDP deflator. The U.S. export price deflator in foreign currency terms
is the product of the U.S. export price deflator and the effective nominal exchange rate
(foreign currency/dollar). The effective nominal exchange rate is a weighted index of the bilateral exchange rates between the US and each of its seven trading partners, based on share
of tourism exports. Foreign GDP dejlator used in the relative price term is a weighted index
of the GDP deflators of the seven major trading partners. Foreign GDP deflator and bilateral
nominal exchange rate for six trading partners are obtained from the Bank of International
Settlements. Mexican data are obtained from the Banco the Mexico and the International
Monetary Fund.
The relative import price is the U.S. service import deflator divided by the U.S. GDP deflator.
U.S. GDP dejlator is obtained from the U.S. Department of Commerce.

Other Private Services and Royalties & License Fees (OPS&RLF)
U.S. real ops&rlf export is nominal ops&rlf exports divided by the service export deflator
( 1985=100). Nominal ops&rlf exports are obtained from the U.S. Department of Commerce
(Survey of Current Business, various issues).
24

US real ops&r lf import is nominal ops&rlf imports divided by the service import
deflator
(1985=100). Nominal ops&rlf imports are obtained from the U.S. Department
of Commerce
(Survey of Current Business, various issues).
Outward FDIA is the total real U.S. foreign direct investment assets abroad,
obtained by
cumulating real U.S. FDI outflows. Real U.S. FDI outflows are obtained by
deflating
nominal U.S. direct investment abroad by the weighted foreign producer price
index based on
the share of U.S. FDIA (1985=1). Nominal U.S. direct investment abroad is
obtained from
the U.S. Department of Commerce (Survey of Current Business, various issues)
.
Inward FDIA is the total real foreign direct investment assets in the United States,
obtained
by cumulating real U.S. FDI inflows. Real U.S. FDI inflows are obtained by
deflating
nominal foreign direct investment in the U.S. by the U.S. producer price index
(1985=1, U.S.
Department of Labor). Nominal foreign direct investment in the U.S. is obtaine
d from the
U.S. Department of Commerce (Survey of Current Business, various issues)
The relative export price is the U.S. services export deflator (in foreign currenc
y terms)
divided by the foreign GDP deflator. The U.S. export price deflator in foreign
currency terms
is the product of the U.S. export price deflator and the effective nominal exchan
ge rate
(foreign currency/dollar). The effective nominal exchange rate is a weighted
index of the
bilateral exchange rates between the United States and each of its seven trading
partners,
based on share of ops&rlf exports. Foreign GDP deflator used in the relative
price term is a
weighted index of the GDP deflators of the seven major trading partners. Foreign
GDP
deflator and bilateral nominal exchange rate for six trading partners are obtaine
d from the
Bank of International Settlements. Mexican data are obtained from the Banco
the Mexico and
the International Monetary Fund. (Country shares of ops&rlf exports are shown
in Table Al
below.)

The relative import price is the U.S. service import deflator divided by the U.S.
GDP
deflator.
US GDP dejlator is obtained from the U.S. Department of Commerce.

Table A I. Share of US services exports by country
US Major Trading Partners

Tourism

OPS&RLF

Canada
France
Germany
Italy
Japan
Mexico
United Kingdom

0.22
0.06
0.08
0.03
0.32
0.12
0.16

0.23
0.10
0.12
0.05
0.23
0.05
0.22
25

References
Campbell, J. and P. Perron, "What Macroeconomists Should Know About
Unit Roots," NBER
Macroeconomics Annual, forthcoming, 1991.
Clarida, R., "The Real Exchange Rate and U.S. Manufacturing Profits: A Theore
tical
Framework with Some Empirical Support," FRBNY Research Paper no 9214,
1992.
Dickey, D., and W. Fuller, "Likelihood Ratio Statistics for Autoregressive
Time Series with a
Unit Root," Econometrica, (49), 1981.
Engle, R., and C. Granger, "Co-integration and Error Correction," Econom
etrica, (55) 1987.
Engle, R., and B.S. Yoo, "Forecasting and Testing in Co~integrated System
s," Journal of
Econometrics, (35), 1987.
Goldstein, M. and Mohsin S. Khan, "Income and Price Effects in Foreign
Trade," in Ronald
W. Jones and Peter Kenen, eds., Handbook of International Economics. New
York: North
Holland, 1985.
Helkie, W., and L. Stekler, "Modeling Investment Income and Other Service
s in the U.S.
International Transactions Accounts," International Finance Discussion Papers
, the Board of
Governors. No. 319., 1987.
Kravis, Irving B. Services in the Domestic Economy and in World Transa
ctions, NBER
Working Paper No. 1124. Cambridge MA:NBER, May 1983.
Helkie, W., and L. Stekler, "Modeling Services in the U.S. International Transa
ctions,"
Manuscript, International Finance Division, the Board of Governors, I 992.

26

TAB LE 1.
THE U.S. SER VICE S ACC OUN T BY COM PON
ENT
(Per cent Share)

Exports

!mR.Qrt~

(Billions of $)

N~t

198§

11!!!~

1!!e§

199~

1!!e§

1!!93

85.5

184.0

78.2

125.6

7.4

58.4

30.4

40.3

41.5

41.4

-6.4

22.2

10.0

6.2

17.6

9.7

-5.2

-0.8

Transportation

18.4

12.6

21.4

19.5

-0.9

-1.4

Othe r Private Services

31.9

29.8

17.8

25.6

13.4

22.8

9.3

11.1

1.8

3.9

6.5

15.6

Total Services (billions of$)
Tour ism
Milit ary

Roya lties & License Fees

Memorandum:
(billions of $)

Merc hand ise Trad e

~~R.Qr!~
1!!e§

H1!!~

1986

lm1>1:>!1~
1!3!3~

1!3136

19!:13

223. 3

456. 9

368.4

589.4

-145.1

-132.6

-151.2

-103.9

Net

Curr ent Account

TABLE 2
U.S. TOURISM TRANSACTIONS, BY REGION
(billions of dollars, balance of payments basis, annualized rates)

Tourism Exports

Western Europe
of which, U.K.
of which, Germany
Canada
Japan
Latin America, and Other Western Hemisphere
of which, Mexico
Other

Tourism Imports

Western Europe
of which, U.K.
of which, Germany
Canada
Japan
Latin America, and Other Western Hemisphere
of which, Mexico
Other

1986

198.9

1992

1993

26.0

46.9

71.3

74.2

7.6
n.a.
n.a.

14.7
4.8
2.4

24.0
7.6
4.9

24.6
8.2
5.3

3.3

6.2

9.3

8.6

4.3

9.6

13.8

14.4

6.9
2.2

10.0
4.3

15.0
6.2

16.3
5.7

3.9

6.0

8.8

9.7

32.4

41.7

49.6

52.0

12.1
n.a.
n.a.

15.9
4.5
3.2

18.2
4.8
3.0

19.6
6.0
3.3

3.3

3.6

3.8

4.0

1.8

2.4

3.0

3.4

8.2
2.9

11.2
4.8

13.5
5.8

14.2
5.8

7.1

8.4

10.3

10.4

TABL E 3
OTHE R PRIVATE U.S. SERVICES TRANSACT
IONS
(billions of dollars, seasonally adjusted)
EXPO RTS (RECEIPTS)

Total other private services exports (Imports)

Ue ltro..ml .11.!!ili§l!!!l1Qr!l!!lf!8~

IMPORTS (PAYMENTS)

1IDle

1989

1992

~

1W

1009

1992

:Im

27.3

36.5

51.0

54.9

13.9

19.9

26.6

32.1

(!~

1?~

. 16,1

Hl.0

3.9

7.9

1().0

US paren ts' exports to (imports from) their foreig
n affiliates

10.!,

5.4

9.1

10.2

10.5

2.4

4.8

5.3

5.6

US affiliates' exports to (imports from) their foreig
n paren ts

2.8

3.2

5.9

5.5

1.5

3.1

4.6

5.0

11!,1

.. g4,g

.. 34,9

38.9

10.0

1;!.0

111-?'

gq;

3.5

4.6

6.2

6.8

0.4

0.6

0.7

0.8

3.3

5.0

5.5

6.5

1.8

2.1

3.5

5.6

2.0

0.5

1.2

1.5

2.2

0.8

1.3

2.9

1.8

2.5

3.0

3.2

3.3

5.2

6.1

6.5

4.4
0.1
1.0
0.1
0.3
0.3
0.1
0.8
0.1
1.0
0.5
0.2

6.2
0.1
1.0
0.2
0.4
0.3
0.4
0.9
0.2
1.7
0.6
0.3

12.1
0.3
1.8
0.6
0.7
0.7
1.4
1.9
0.2
2.8
0.7
0.9

13.3
0.3
2.1
0.7
0.6
0.8
1.5
2.3
0.2
3.1
0.7
0.9

1.3
0.1

2.0
0.2
0.0
0.0
0.1
0.1
0.1
0.4
0.1
0.7

3.4
0.5
0.1
0.1
0.3
0.2
0.3
0.3
0.1
0.7

3.9
0.6
0.3
0.1
0.3
0.3
0.3
0.3
0.1
0.8

n.a.

n.a.
n.a.

n.a.
n.a.

n.a.
n.a.

4.1

5.4

6.9

7.5

1.1

1.4

1.6

1.8

n.a.

2.3

2.5
n.a.
n.a.
1.8
•: Exports include mainly expenditures of foreig
n governments and international organizations
in the US.
Imports include mainly wages of foreign reside
nts temporaily employed in the US, and of Canad
ian and Mexican comm uters in the US border
area.

0.1

n.a.

bil'
'·\ P.:·!llfr.QmlJ~!llll li!I~.!190.~
. f§

.

,.

Financial Services

Insurance
Telecommunicalions
Business, Professional, Technical services
Advertising
Comp uter & data processing services
Data base and other information services
Research, devel opme nt and testing services
Management, consuHing, and public relations servic
es
Legal servic es
Construction,engineering, architectural & mining
Industrial engineering
Installation, maintenance, and repair of equipment
Medic al servic es
Other

Adden dum:
Film and tape rentals

~\... .,

-

,,..1,_,_;,,,.:...,1,_..;._,...~ -----~~- ··---J, .. ;_..._, .. ;:.

Education

Other

'

o.o
o.o

0.1
0.1
0.0
0.3
0.1
0.5
n.a.

TABLE 4

U.S. ROYALTIES AND LICENSE FEES
(billions of dollars)

Exports (Receipts)

.T..2.~ffili~te9.fi21!!i9D~f§.,_ . .
-~ ···- - -~····-----··
U.S. parents' exports to foreign~----affiliates
U.S. affiliates' exports to their foreign parents
:Jg_Un~fmi~tE1.!!FQrE1,lgnE1,~
Industrial processes
Books, records, and tapes
Broadcasting and recording of live events
Franchise fees
Othe r
Imports (Payments)

IEr2m:A.ffmatedifloreignt1rfilc.c --~-- c.~-~ ··-~
-·U.S. parents' imports from their foreign affili
ates
U.S. affiliates' imports from their foreign pare
nts
IErP11LlJn.mtms~Eflr~l9n§fll
Industrial processes
Books, records, and tapes
Broadcasting and recording of live events
Franchise fees
Other

1fil36

1fillli

1992

~

7.9

13.8

19.9

20.4

. ,Q,Q .
5.8
0.2

. 11.9
10.6
0.3

1q.~
15.2
0.7

1f?.O
15.2
0.8

n.a.
n.a.
n.a.
n.a.
n.a.

1-~

2.9
2.1
0.1
0.1
0.2
0.4

4.0
2.5
0.2
0.1
0.3
0.8

4.4
2.8
0.2
0.2
0.4
0.8

1.4

2.5

5.0

4.8

Q,~

1.7
0.1
1.6

?,?

0.2
3.1

3.5
0.2
3.2

0.8
0.6
0.1
0.1
0.0
0.1

1.7
0.8
0.1
0.6
0.0
0.2

1.4
1.0
0.1
0.0
0.0
0.2

0.1
0.8

O.q
n.a.
n.a.
n.a.
n.a.
n.a.

Ta ble 5. Te sts of Ord er of Int
egr ati on
(Sa mp le Per iod :

yar iab 1a

197 4:Q l to 199 3:Q 4)

APl fk•J. 1 taa t sta tis tic

US dom est ic dem and
4US dom est ic dem and

0.5 2
4.9 7

for eig n dom est ic dem and
4fo rei gn dom est ic dem and

1.3 0
3.5 7

rel ati ve imp ort pri ce
4re lat ive imp ort pri ce

1.7 4
4.4 9

rel ati ve exp ort pri ce
4re lat ive exp ort pri ce

1.0 4
6.0 6

out wa rd FDI A
4ou twa rd FDI A

2.5 7
7.1 3

inw ard FDI A
4in wa rd FDI A

(k= 0)
(k= 0)

0.8 8
3.6 7

No tes :
1.

var iab les are in log ter ms .
I'

3.

is the num ber of lag s in the
ADF reg res sio n of var iab le
X:

Re lat ive imp ort pri ce =

us ser vic e import de fla tor
US GDP de fla ter

4.

Re lat ive exp ort pri ce =

us ser vic e exp ort de fla tor
for eig n GDP de fla ter

Table 6a. Real U.S. Tour ism Es:po rts

(t•stati stics in parent hesis,

Long -Run Equa tion

■ample

(1)

period : 1973Q4 - 1993Q4 )

(2)

Regres sor

Coeffi cients

Coeffi cients

Consta nt

-3.59 (-6.87)

-4.32 (-8.81)

Foreig n domes tic deman d

2.19 (24.01)

2.28 (24.25)

Relativ e export price

-0.42 (6.06)

Relativ e CPI

0.36 (4.113)

D84
D90

0.20 (7.84)
0.18 (7.44)

0.18 (6.69)
0-19 (7.59)

Adjust ed-R2

o.99
4.92
5.16

0.99
4.67
4.89

ADF
Phillip s-Perro n(Zt)

Error Corre ction Mech anism
Regres sor

Coeffi cients

Coeffi cients

Residu altt-1 l

-0.51 (-4.56)

-0.46(-4.42)

..lForei gn domes tic deman d(t-1)

2.45 (3.63)

2.57 (3.81)

Adjust ed-R2

DW

0.28
2.10

0.27
2.15

RMSE
Theil U

1.07
0.019

1.03
0.019

Notes:
J. DS4 =
2,
3.
4,
5.
6.

(all the variab les are in Jog terms, except the dumm y variab
les)

a consta nt dumm y for 1984:1-1993:4.
D90 = a consta nt dumm y for 1990:1- 1993:4.
Relativ e export price= USd service export deOato r in
foreign curren cy terms.l foreign GDP deflato r.
Real CPI = foreign CPVUS CPI in foreign curren cy terms.
RMSE & Theil U measu re out-of- sample foreca st based
on regres sion using 1973Q- 91Q4 sample period .
<>I(t ► = x(t)-x(t
-1).

Table 6b. Real U.S. Touri sm Impor ts

(t-statis tics in parenth esis, sample period: 1_973Q4 • 1993Q4
)

Long-Run Equat ion
Regres sor

(1)

(2)

Coeffic ients

Coeffic ients

Consta nt

-4.11 (-5. 71)

-4.90 (-5.21)

US domest ic demand

0.99 (11.18)

0.82 (8.04)

Relativ e import price

-1.07 (-9.26)

Relativ e CPI

-0.40 (-6.02)

D84
D90
Adjuste d-R2

ADF
Phillips -Perron (Zt)

0.36 (12.44)
0.07 (4.40)
0.99
-5.62
-5.91

0.44 (14.66)
0.10 (4.41)
0.98
-5.17
-5.30

Error Corre ction Mecha nism
Regres sor

Coeffic ients

Coeffic ients

Residua l(t-1)

-0.38 (-2.66)

-0.30 (-2.49)

LiUS domest ic demand (t-1)

1.11 (2.20)

1.31 (2.55)

Adjuste d-R2

DW

0.09
1.92

0.08
2.09

RMSE
Theil U

1.50
0.018

1.45
0.017

Notes:
(all the variabl es are in log terms, except the dummy variabl
es)
1. D84 = a constan t dummy for 1984:1-1993:4.
2. D90 = a constan t dummy for 1990:1-1993:4.
3. Relativ e import price = US service import deflator /US GDP
deflato r in foreign currenc y terms.
4. Real CPI = foreign CPI/US CPI in foreign currenc y terms.
5. RMSE & Theil U measur e out-of-s ample forecas t based
on regress ion using 1973Q-9 1Q4 sample period.
6. Llx(t) = x(tJ-x(t .l).

Tab le 7a. Rea l US Ezp orts of
Oth er Priv ate Serv ices

(inc ludi

ng roya lty
(t.aa tiatic a In paren theai a, 8"1Dple perio
d: 1982 Ql • 1993 Q4)

Lon g-R un Equ atio n
Regr essor
Cons tant

(1)

(2)

(3)

Coef ficie nts

Coef ficie nts

Coef ficie nts

-1.49 (-1.80)

-1.74 (-1.92)

-4.02 (-2.18)

-6.G8 (-3.41)

0.81 (3.32)

o.88 (3.G5>

0.G8 (2.82)

0.25 (3.89)

0.Sl (3.77)

0.83 (3.92)

Inwa rd FDIA
-0.23 (2.40)

Rela tive CPI.

0.22 (2.17)

Fore ign dome stic dema nd

1.G7 (11.21)

1.62 (12.00)

D86
D91

0.11 (3.84)
0.15 (8.55)

0.10 (3.33)
0.14 (7.85)

Adju sted- R2
ADF
Phill ips-P erron (Zt)

(4)

Coef ficie nts

Outw ard FDIA

Rela tive expo rt price

& lice nse fees )

. -1.11 (-1.88)
0.11 (4.112)
0.05 (2.27)

0.10 (4.35)
0.07 (3.04)

0.98
-3.12
-3.24

0.98
-3.10
-3.22

Coef ficie nts

Coef ficie nts

Coef ficie nts

Coef ficie nts

-0.36 (-2.80)

-0.38 (-2.97)

-0.42 (-2.24)

-0.48 (-2.81)

.1.Ou twurd FDIA ft-1)

0.87 (2.11)

0.87 (2.14)

0.78 (1.83)

0.80 (1.95)

.lOPS expo rts(t- 1 J

0.14 (1.22)

0.14 (1.27)

0.20 (1.62)

0.19 (1.83)

0.02
1.80

0.04
1.81

-0.03
1.83

0.02
1.82

1.79
0.014

1.76
0.013

1.96
0.015

1.89
0.014

0.99
-3.80
-3.83

0.99
-3.42
-3.G2

Erro r Cor rect ion Mec hani sm
Regr essor

Resid uallt- 1)

Adju sted- R2

DW
RMS E

Thei l U

Note s:

(all the varia bles are in log term s, exce
pt the dumm y varia bles)
a cons tant dum my for 1986: 1-199 3:4,
a cons tant dum my for 1991: 1-199 3:4.
Rela tive expo rt price ., US servi ce impo
rt defta tor in forei gn curre ncy term s/for
eign GDP deOa tor.
Real CPI = forei gn CPI/U S CPI in forei gn
curre ncy term s.
RMS E & Thei l U meas ure out-o f-sam ple
forec ast based on regre ssion using 1973
Q-91Q4 samp le perio d.
<l.x(t) = x(t)-x (t-1).

1. D86 =
2. D91 c

3.
4.
5.
6.

Tabl e 7b. Rea l US Expo rts of Othe r Priv ate Serv
ices (incl

udin g roya lty & licen se fees)

(t-ata tiatico ill paren theaio , samp le period : 1982Q
l - 1993Q 4)

Long -Run Equ ation
Regre ssor

Cons tant

(1)

(2)

(4)

Coeff icient s

Coeff icient s

Coeff icient s

Coeff icient s

-17.24 (9.87)

-12.37 (-8.88)

-15.29 (-6.40 )

-14.33 (-6.08)

0.87 (2.88)

1.14 (4.81)

0.21 (2.89)

0.25 (3.67)

-0.14 (-0.64)

-0.44 (2.62)

Outw ard FDIA
Inwa rd FDIA

Relat ive impo rt price

(3)

0.78 (3.09)

Relat ive CPI

0.43 (3.74)

US dome stic dema nd

2.47 (11.84 )

2.16 (13.42 )

0.69 (1.68) )

DSG
D91

0.11 (2.52)
0.14 (5.49)

o.u (2.69)
0.14 (6.25)

D.18 (4.90)
0.02 (0.63)

0.22 (7.88)
o.oo (-0.13)

0.98
3.45
3.69

0.98
4.77
3.45

0.99
3.89
3.83

0.99
3.76
3.89

Adjus ted R2
0

ADF
Philli ps-Pe rroo(Z t)

Erro r Corr ectio n Mec hani sm
Regre :ssor
Resid uaHt• l)

~Inw ard FDIA( t-1 J

Coeff icient s

Coeff icient s

Coeff icient s

Coeff icient s

-0.29 (-2.83)

-0.28 (-2.67)

-0.43 (-3.31)

-0.39 (-2.95)

0.60 (3.59)

0.35 (1.60)

0.64 (3.90)

o.64 (3.64)

_\Out ward FDIA (t-1, t-2)

1.01 (!.65)

~OPS impor ts(t-1 )

0.07 (1.53)

0.07 (1.38)

0.08 (1.92)

0.09 (1.95)

Adjus ted-R 2
DW

0.03
1.87

0.06
1.93

o.os
1.85

0.04
1.86

RMSE
Theil U

0.81
0.016

0.70
0.014

0.85
0.017

0.89
0.017

~

(all the varia bles are in log terms , excep t the
dumm y variab les)
a const ant dumm y for 1986:1 -1993: 4.
a const ant dumm y for 1991:1 -1993: 4.
Relat ive impo rt price = US servic e impo rt deflat
or/US GDP deflat or.
Relat ive CPI = foreig n CPVU S CPI in foreig n
curre ncy terms .
RMSE & Theil U meas ure out-o f•sam ple forec
ast based on regre ssion using 1973Q~91Q4 samp
le perio d.
iix(t) = z(t)-s( t-1 ).

1. D86 ""
2. D91 =

3.
4,
5,
6.

Table 8. Explaining the Sur ge in Rea
l U.S. Services Surplus, 1984-92

(nonannualized, percent)

Actual growth

I
I
I

i>redicted Gro wth

Other Private
Services & ALF
Exports Imports

73

85

75

85

foreign dom esti c demand (28)

I

Tou rism
Exports Imports

Total Tourism
OPS & ALF
Exports Imports

128

55

99

66

33

114

60:

93

68

25

100

29

31

I

62

I

U.S. dome~tic demand (22)

29

26:'

17

'

relative export price (-34)

9

relative imp ort price (9)

Balance

Percentage of
Predicted Total
Ser vice Growth
Exporie Imports

14
-4

11
-8

-17
11

-7

7

27

35

-8

13

14

100

Balance

100
118

25
12

-68
45

-10

26

29

51

-33

-2

14

21

-6

13
9
4
>tes:
Growth rates of explanatory varibles ove
r 1984-92 are shown in parenthesis.
relative export price = export price defl
ator x exchange rate/foreign GD P defl
ator.
relative imp ort pric e= import price defl
ator/U.S. GD P dellator.
Average sha re of tourism exports in tota
l (tourism, OPS & ALF) services exports
is 47% over 1984-92,
Average share of tourism imports in tota
l (tourism, OPS & ALF) services imports
is 66% over 1984-92.
A 1 percent doll ar appreciation is esti
mated to increase relative export pric
e by about 1 percent,
and decrease relative import price by
0.5 percent ove r time, holding US and
foreign prices unchanged.

14

13

18

data imp rove men t
outward FDIA (37)
inward FDIA (102) _

17

I

I

24
25

21
421
261

38

I

42'I
'

CH AR T 1: U.S. GOODS AN D SERVICES TR
AD E BA LA NC E

billions of$, BOP basis, saar

100 , - - - - - - - - - - - - - - - ,
A. NOMINAL

billions of 1987$, BOP basis, saar

100

8. REAL

Services Trade

50

0

:,,

'
I I

'

0

I

\
\

I

-50

-

"

I

r

I

I

I ..,
II
II
I I
11
I
I
f
1,1
•••••• •••••• ••• •••••• •••••• • . L •
••••• •••.•• ••••• .••••• ••.••. ••.

~

I
I

I

-50

I

I

'

,.
I

,,,

I

Merchandise Trade

'

'

/

I

I

',,

...... ...... . - ...... ... - ...... . - ...... . ·'·,. -:\

,

',, ,/

I I
...... ... I I.

,,

I\

'

.I .

I
II
I

I I

-100

11

',

..

.,. ~.

~ I

Merchandise Trade •
I

-'

,,

I

I

'

-150

II
I

-'".
I

I I
I

,11

I•

'

I

I
I

I I •

·•- •- - • -•- .•
fl I

"''

I I

.,

I I L
I , '
I
I
I
I
I
I

\

••••• •.•••• ••••• ••.••• ••••• ••••• I •••••• •••••• •••••

\

\

j

\

i

-150

I

'

,, ' ' ,, '
f I

I

-10 0

Services Trade

50 ,-... ..... ..... ..... ..... ... .

1' I

,;
\

I

. \.

. J""

\

\

I

I

\ I

I I I I I ~ \ I ,·11h» ~~1.,, 1,.,1,
-200 I '"""'1'~~r,lw/9';~""
.. (
"19s(""~~ ... "l'9~
I 6 199.1

-200

, 1""', jlj~;" ,"'1tj7~
, '""j9
, , /\i"
, ,'"1tjg~ '""
, l~s~"\...'I1>'\\II"'"199.1
' '" ,

Chart 2:

U.S. SERVICE ACCOUNT BY COMPO
NENT

billions of$ , BOP basis, saar

50

40

........ - ................
........

.........................

.........................

. - .......................
...

......

..... ..... ..... ..... ..... ..... ...

. . . . . . . . .. . ..

..... .....

'
I

. ..... ..... .

.. .,,.

--- -------------'

Net Tou rism

'

,

. .. ·;- "~- ..... ... .

-

... ,_ ...

- _.

-

........

·--·-··-"·

. ,,

.. -,..---JI'•'-.;, .. .;.-... -~."": ..

....-··-··--.

_

,'

,-.----------,_._,'

..... ..... ..... ..... ....
. . . . . .. . . .

••

I

...........

I

10

, '

' -····
····· ····· ;··~ ~--

.-.........

,•

0

,,, .\ . ,

Net Oth er Private Services
& Royalty & License Fees

30
20

•

. . . . . . . - . . . . . . . . . 1·\. , . . .
.... .

- ..., --/-··''-./,_,,,. ,

..._
....... ..._
·/v \/··
·.

'

,,,,

·.~r-··
V

-10

1..

\·

·.
r·/\·.J.·~.

.

/

.

\

:

I /

=-v:

Net Military

-20

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

..
.._,

~

.

Chart 3: Actual and Fitted Tourism Exports and Imports
billions of US$
160

r--------------------------,------,----,
A. EXPORTS

,-

140 r--·········· ............. ............. ·············· ·············· ····· ..
120

1- ........................................................................................................actual ..

100 r-- · · .. · · .. · · · · · .. · .. · · · · · · · · · · .. · · · · · · · · · · · · · · · · · · .. · · · · · · · .. · · · · · · · · · · ·

fitted .. ·~· ·

80 r--············ ·············· ··········· ·············· •············ ····
60r--····· ············ ············ ············ ············ ············ ············ ~
40 r········
20

120

1973

out of sample
. . . . . • • . . . .

1975

..,

. • . . . . • . . . . ! ""'!,

1977

1979

1981

1983.

1985

1987

1989

1993

1991

.-----------'-----------------~-7
B. IMPORTS

.,

100 r············ ············· ············· ············· ···········

~-:-,..

80 r············ ············· ············· ···· ·············· ··

·.·

actual
60 r············ ····

·············· ..

...

·········

...

...

1979

1981

~~

40 I-········=- ~"'"········ ············· ····· .. ~

20

1973

1975

•

1977

out of sample

fitted

1983

1985

1987

1989

1991

..,

1993

•

Chart 4: Actual and Fitted OPS&RLF Exports and Im
ports
billions of US$
160.------

A.EXPORTS

------------

~----,

140 t-···· ······ · ······ ······ ······ ······ ······ ······ ······
······ ··· ..... ..... ..... .. .
120

f-··· ······ ······ ······ ······ ······ ······ ······

,,, r
00

a~al2

. ················

1- ·· ······ ······ ···· .... .. ...... ....
--

> - ---c..._... ..,,
__ -

-

-

······ ······ ······ ······ ······ ······ ··· · · ·
····~·"'············

----

1985

4.6 . - - - - - - - - - - -

1989

1991

-------------

~... JM.:PQ~T.S. ............................................... .

4.2 t-···· ······ ······ ······ ······ ····· ·····

--- - - ... --

3.8 ,- · · · · · · · · · · · · · ·

,

1-··· ······ ·····.

·····actual · · ·/~ ·tt ~d

--------

--'-~ ....... ...,'-- :-'-;: -~

I 993

------,

...... ...... ...... ...... ... .

4 t-···· ······ ····· ······ ······ ······ ··

3 .6

l···

--'-- ~............~ ~ ~ ~ . . . . . . . . i . . . ~ ~.........~----

1987

log
4.4 1- ..

..... .

~··

_,' fitted

60 ~-.. __._ _~~ ..... ..._ __._ _.__ ._~ -.._
._~

I 983

out of sample

.. - .. ·...: ·.,;;. ·..,;,_;f.,- ,,, ,.

'

out of sample

3.4 I-···· ······ ······ .·· ...... .... ,fr.~.... .
3.2
3

...._._._..._L-:;1:':::9-;t83::-'-_._._~...._,_-1-=-98;,-..,5,.._.i-~__,__1~9-8_7
...._.__~~....L-.l-9-89~L........~.......J...-I<-::-c99~J........1.~~.....J....,I"""9~9-:::---'3.