Foreign direct investment in the United States in 1983 Essay

Last year, the foreign direct investment position in the United
States increased 9 percent, to $135.3 billion, following a 15-percent
increase in 1982 (table 1). In both years, the position grew at a much
slower rate than in the 4 years prior to 1982, when the average annual
rate was about 30 percent.

The slower growth in 1982 and 1983 primarily reflected the U.S.
recession, which lasted into the first half of 1983. Slack demand
substantially reduced U.S. affiliates’ earnings and, thus, funds
available for reinvestment. At the same time, because prospects for
future earnings were uncertain, foreign investors had little incentive
to expand their existing U.S. operations or to acquire or establish new
ones. Capital inflows for both equity and debt fell sharply from their
peaks in 1981, when foreign takeover activity was strong. Earnings of
U.S. affiliates improved in 1983, as the U.S. recovery began. However,
uncertainty about the recovery’s duration probably continued to
dampen investment throughout much of the year.

Other factors may also have contributed to the slowdown in
investment. Slack demand in most other developed countries hurt
earnings of foreign multinational companies and limited the funds
available for investment. In particular, weak worldwide petroleum
markets restrained U.S. investments by major foreign oil companies and
oil producing countries; their investments had contributed significantly
to the rapid growth in earlier years. Also, the strong appreciation of
the U.S. dollar against major foreign currencies during 1982-83 raised
the foreign currency cost of U.S. assets and lowered the dollar cost of
U.S. imports, both of which tend to encourage production abroad, rather
than in the United States, to serve the U.S. market. Finally, a surge in the U.S. stock market, beginning in mid-1982, raised the cost of
acquisition and, together with high interest rates here and abroad, may
have diverted some foreign capital inflows from direct investment to
more liquid porfolio investment in U.S. stocks and bonds.

Direct Investment Position

More than two-thirds of the year-end 1983 position of $135.3
billion was accounted for by European parents. Canadian and Japanese
parents accounted for 8 percent each, and parents in “other”
countries–mainly the Netherlands Antilles, Panama, and Kuwait–for 15
percent. By industry of the U.S. affiliate, 35 percent of the position
was in manufacturing, 15 percent in wholesale trade, 14 percent in
petroleum, and 36 percent in “other” industries, mainly real
estate, banking, and insurance.

Nearly three-fourths of the position was accounted for by equity
investment, including retained earnings (table 2). Net outstanding
intercompany debt owed by U.S. affiliates to their foreign parents
accounted for the remainder. The $35.5 billion of outstanding
intercompany debt was the net of U.S. affiliates’ payables due to
the foreign parents, $46.1 billion, and U.S. affiliates’
receivables due from their foreign parents, $10.6 billion.

The $11.7 billion increase in the position in 1983 was accounted
for by capital inflows of $11.4 billion and valuation adjustments of
$0.3 billion (tables 3). Capital inflows consisted of equity capital
inflows of $6.8 billion, intercompany debt inflows of $4.0 billion, and
reinvested earnings of $0.6 billion.

Compared with 1982, capital inflows fell $2.4 billion. Net equity
capital inflows fell $2.8 billion, and net intercompany debt inflows
fell $2.7 billion, mainly because of an increase in U.S.
affiliates’ receivables due from their foreign parents. Partially
offsetting the reduced equity and debt inflows was a $3.0 billion shift
to positive reinvested earnings, from a negative $2.4 billion in 1982.
The shift entirely reflected an improvement in affiliate earnings in
1983; distributed earnings remained virtually unchanged (table 4).

By country of foreign parent, by far the largest increase in
position–$4.1 billion–was for the United Kingdom. The increase was
concentrated in manufacturing and real estate. The position of parents
in the Netherlands increased $2.8 billion, mainly in petroleum and
manufacturing, and the position of Japanese parents increased $1.5
billion, all in wholesale trade.

In contrast, the position of Canadian parents fell $0.3 billion,
the third consecutive annual decrease. In all 3 years, negative
reinvested earnings contributed to the decreases. Also, in 1981 and
1983, large negative valuation adjustments were made to the position for
Canada to reflect a change in the manner in which direct investors held
their U.S. investments. In 1982, the sale of a Canadian company’s
U.S. mining operations to a French company contributed to the decrease.

By industry, the largest increase in the position–$4.9
billion–was in “other.” The increase was down substantially
from that in 1982, despite smaller negative reinvestment earnings.
About one-half of the increase was in real estate. The positions in
banking and insurance also showed strong increases.

The position in manufacturing increased $3.7 billion, about the
same as in 1982. The increase was concentrated among affiliates with
parents in Europe, mainly in the United Kingdom, Netherlands, France,
and Switzerland. Several acquisitions of U.S. companies resulted in
sizable equity and intercompany debt inflows. Reinvested earnings were a
negative $0.3 billion, substantially improved from a negative $1.6
billion in 1982. Within manufacturing, increases in the position were
largest in chemicals ($1.7 billion), food ($0.8 billion), and
“other” ($1.0 billion).

As in manufacturing, the increase in the position in wholesale
trade–$2.3 billion–was little changed from that in 1982.
Japanese-owned affiliates accounted for two-thirds of the increase and
British-owned affiliates from most the remainder.

The increase in the position in petroleum was $0.8 billion, down
substantially from that in 1982. Reinvested earnings fell sharply, to
$0.8 billion, due to deterioration in petroleum affiliates’
earnings. Also, because of a large increase in affiliates’
receivables due from their foreign parents, intercompany debt
transactions resulted in net outflows of $0.2 billion. Most of the
increase in receivables reflected transaction of a Japanese-owned
affiliate engaged in worldwide petroleum trading.


Direct investment income, the return on the position, more than
doubled in 1983, to $6.4 billion (table 5). The sharp increase followed
a similarly sharp decline in 1982 and largely reflected the improvement
in U.S. affiliates’ earnings, related to strong U.S. economic
growth in 1983.

Direct investment income consists of U.S. affiliates’ earnings
(that is, foreign parents’ shares in their U.S. affiliates’
net income after U.S. income taxes), less U.S. withholding taxes on
affiliates’ distributed earnings, plus interest (net of withholding
taxes) on intercompany debt (table 6).

Of the $3.4 billion increase in income in 1983, $3.1 billion was
due to an increase, to $4.0 billion, in U.S. affiliates’ earnings
(table 7). Capital gains, which are included in earnings, increased
from near zero in 1982 to $0.5 billion in 1983, mainly because of
appreciation in the value of stock portfolios held by affiliates in

Earnings before capital gains rose $2.5 billion, to $3.5 billion.
The rise reflected shifts, from losses in 1982 to positive earnings in
1983, in manufacturing, wholesale trade, and “other”
industries. The shift was largest–$1.4 billion–in manufacturing.
Within manufacturing, earnings improved in each subindustry; however,
only in chemicals and food were earnings positive for the year. Many
manufacturing affiliates continued to have losses, though smaller than
those in 1982. As the impact of the U.S. recovery becomes more
widespread, shifts from losses to positive earnings can be expected to
provide a strong boost to both earnings and reinvested earnings after
1983. (Preliminary estimates indicate that, in the first half of 1984,
earnings of manufacturing affiliates were more than three times the 1983
annual total.)

In wholesale trade, affiliates had positive earnings of $0.9
billion in 1983, after small losses in 1982. The increase in earnings
was led by Japanese-owned affiliates, particularly those selling motor
vehicles and consumer electronics; it mirrored the upturn in demand for
consumer durables that was associated with the U.S. economic recovery.

In contrast to the increases in other industries, earnings in
petroleum fell, for the second consecutive year, to $1.8 billion, from
$2.2 billion. The decline reflected weak worldwide petroleum markets
resulting fro recession-dampened demand in most developed countries.
Despite the decline, earnings of petroleum affiliates were larger than
the earnings of all other affiliates combined.

Net interest payments increased $0.3 billion, to $2.5 billion; the
increase was concentrated in real estate. Net payments consisted of
U.S. affiliates’ gross payments of interest to the foreign parents
of $3.1 billion and gross receipts of interest from their foreign
parents of $0.5 billion.

Fees and Royalties

U.S. affiliates had net receipts of fees and royalties from their
foreign parents of less than $0.1 billion in 1983, unchanged from the
previous year (table 8). Affiliates’ gross receipts of $1.7
billion were almost offset by gross payments to their foreign parents.

In the 4 years for which data on gross payments and receipts are
available, gross receipts have risen much more rapidly than gross
payments. Receipts, unlike payments, are concentrated among relatively
few affiliates. Among them are several engaged in petroleum-related
service activities. U.S. affiliates of foreing auto makers also had
large receipts that reflected remittances from their foreign parents for
warranty work performed by the affiliates.

Technical Note

Since August 1983, when the last annual article on foreign direct
investment in the United States was published, three major changes have
been made to the data series. First, the series for 1980 forward have
been revised to incorporate the results of BEA’s 1980 benchmark survey of foreign direct investment in the United States; previous
estimates for these years were based on the 1974 benchmark survey.
Second, capital inflows for 1981 forward have been revised to include
certain data from BEA’s survey of new foreign direct investments in
the United States. Third, unincorporated affiliates are now required to
report as much detail as incorporated affiliates. As a result, the
definitions of some of the components of the direct investment position,
capital inflows, and income have been changed.

Preliminary estimates for 1980 forward, incorporating the bulk of
these changes, were published in two articles–one on U.S. international
transactions and the other on the U.S. international investment
position–in the June 1984 issue of the SURVEY. The estimates published
in this article supersede the June estimates. In June, the process of
incorporating the 1980 benchmark survey results into the estimates was
well underway, but was not completed. To avoid delaying publication of
the international transactions accounts and the international investment
position, preliminary direct investment estimates, at a highly
aggregated level, were made. Since June, the benchmarking process has
been completed, and revised estimates for 1980 forward prepared.
Although the revised estimates are presented here, they will not be
incorporated into the international transactions accounts or the
international investment position until June 1985, when both will
undergo customary annual revisions.

Benchmark revisions

Coverage of the estimates.–All series now incorporate information
from the 1980 benchmark survey of foreign direct investment in the
United States. The benchmark survey covered the universe of U.S.
affiliates. The larger affiliates in the universe were required to file
complete reports, which included the balance of payments and direct
investment position data used to derive the estimates in this article.
The remaining affiliates filed only partial reports, which did not
include the balance of payments and position data.

Complete reports were required from affiliates that had total
assets, sales, or net income of $1 million or more or that owned 200 or
more acres of U.S. land. Out of a total of 12,510 U.S. affiliates,
7,676 filed complete reports. Although these affiliates accounted for
only 61.o percent of the universe in terms of number, they accounted for
virtually all of it in terms of value–99.7 percent of total assets,
99.9 percent of sales, 100.6 percent of net income, and 99.3 percent of
acres of U.S. land owned. (The percentage for net income exceeds 100
because affiliates that filed only partial reports had, in the
aggregate, a net loss for the year.)

In this article, the universe consists of affiliates that filed
complete reports–and thus provided balance of payments and direct
investment position data–in the 1980 benchmark survey. For the
nonbenchmark years after 1980, balance of payments and position data
were reported, but by only a sample of these affiliates, in BEA’s
quarterly survey of foreign direct investment in the United States. To
obtain universe estimates of most items in nonbenchmark years, an
estimates had to be derived for affiliates that were in the universe but
not in the sample. The estimation was done by estrapolating forward the
data of each of these affiliates based on the movement of the reported
sample data. For 1981-83, the quarterly sample consisted of U.S.
affiliates that had annual assets, sales, or net income of $5 million or
more; for 1984 forward, the exemption level was raised to $10 million.

Comparison of the quarterly and benchmark survey data.–For
affiliates that reported in both the 1980 quarterly sample survey and
the 1980 benchmark survey, the first step in the benchmarking process
was a comparison and reconciliation of the data from the two surveys.
It should be noted that, to ease the reporting burden on respondents,
the benchmark survey data were collected on a fiscal-year, rather than a
calendar-year, basis (see the next section for further discussion).
Thus, the sum of the data for the four quarters corresponding to an
affiliate’s 1980 fiscal year, as reported in the quarterly survey,
was compared with the fiscal-year total reported in the benchmark
survey. Any significant discrepancies were investigated and
resolved–usually in favor of the fiscal-year total from the benchmark
survey, which was reported later and was subject to more edit checks
during processing than the quarterly survey data. To obtain quarterly
estimates, the fiscal-year total (including revisions, if any) was, in
most cases, allocated among the quarters in the same proportions as in
the quarterly sample data.

For affiliates that did not report in the quarterly survey, such a
comparison between benchmark and quarterly survey sdata could not be
made. The fiscal-year data reported in the benchmark survey were
accepted without revision. Quarterly estimates were obtained by
dividing the annual total by 4.

Adjustment of 1980 data from a fiscal- to a calendar-year
basis.–Before universe estimates for 1981 onward could be derived, the
fiscal-year data from the benchmark survey (including any revisions) had
to be adjusted to a calendar-year basis–the basis required for
recording U.S. international transactions. An affiliate’s 1980
fiscal year was defined as its financial reporting year that had an
ending data in calendar year 1980. About two-thirds of the affiliates
had a fiscal year that coincided with the calendar year; these
affiliates’ data required no adjustment. For the remaining
one-third, the specific adjustment procedure depended upon whether or
not the affiliate reported in the 1980 quarterly sample survey.

For affiliates that did report in the quarterly survey, the data
dervived from the benchmark survey for those quarters of the fiscal year
that also fell in calendar year u980 were first isolated. These data
were then added to data from the quarterly survey for any remaining
quarters of the calendar year to obtain estimates for the calendar year
as a whole.

For affiliates that did not report in the quarterly survey, the
procedure differed depending on the item being estimated. For income
and its components, except capital gains/losses, and for fees and
royalties, the data from the benchmark survey for fiscal year 1980 were
used as the estimates for calendar year 1980. For equity capital and
intercompany debt–two components of total capital inflows–and for
capital gains/losses, the calendar-year estimates consisted only of the
amounts for those quarters of the affiliate’s 1980 fiscal year that
were also in the 1980 calendar year. For any remaining quarters of the
calendar year, the amounts were assumed to be zero.

The treatment of the latter items is consistent with that in
nonbenchmark periods, in which data for them are not expanded to
universe-levels, but are included in the estimates only as actually
reported in the quarterly sample survey. These items tend to be
volatile, with frequent sign reversals; thus, the reported sample data
do not provide a reliable basis for estimating unreported data.

The revisions to reported benchmark survey data and the adjustment
from a fiscal- to a calendar-year basis raised the direct investment
position 3.9 percent–from $80.0 billion to $83.0 billion (table 9).
Income was raised 19.4 percent–from $7.2 billion to $8.6
billion–mainly because of an exceptionally large capital gain that fell
in the calendar year but not in the affiliate’s 1980 fiscal year.
Capital inflows were raised 11.5 percent–from $15.2 billion to $16.9
billion. This upward adjustment partly reflected the impact on
reinvested earnings of the exceptionally large capital gain, and partly
a fairly sizable upward adjustment in intercompany debt.

Comparison of revised and previously published 1980 estimates.–In
table 10, the calendar-year estimates for 1980 based oin the 1980
benchmark survey are compared with previously published estimates for
1980 based on the 1974 benchmark survey. The previously published
estimates were obtained by extrapolating forward universe data from the
1974 benchmark survey (which were for the calendar year), using sample
data for interim years.

Differences between the estimates on the 1974 and 1980 bases are
quite large. On the 1980 basis, the direct investment position, at
$83.0 billion, is $14.7 billion higher than on the 1974 basis. Capital
inflows, at $16.9 billion, are $3.3 billion higher. In contrast, income
on the 1980 basis is $0.8 billion lower, and fees and royalties $0.1
billion lower, than on the 1974 basis.

All of the $3.3 billion upward revisioin in capital inflows was
accounted for by affiliates that reported in the 1980 benchmark survey
but not in the 1980 quarterly sample survey; revisions or corrections of
errors in the reported sample data were negligible. As noted earlier, in
the absence of a benchmark survey, no estimate of unreported equity
capital or intercompany debt inflows is made. Thus, the previously
published estimates, which were prepared before the results of the 1980
benchmark survey became available, included only equity capital and
intercompany debt inflows actually reported in the 1980 quarterly
survey. Of the unreported inflows, about three-fourths were accounted
for by affiliates that should have reported but did not, and about
one-fourth were accounted for by exempt affiliates. (Most of the former
affiliates, identifed during the benchmarking process, have now begun to
report quarterly.)

Similarly, a large share–about two-thirds–of the upward revision
in the direct investment position, which consists of cumulative capital
inflows and valuation adjustments, was due to unreported equity capital
and intercompany debt transactions from 1974 to 1980. Most of the
remaining one-third was due to revisions or corrections of errors in the
reported sample data.

For income and for fees and royalies, the downward revisions were
due both to revisions in reported sample data and to overesstimation of
the unreported part of the universe on the 1974 basis.

Estimates for 1981 forward.–As noted earlier, for equity capital,
intercompany debt inflows, and capital gains/losses, the estimates for
1981 forward consist only of the sample data reported in the quarterly
survey; no estimates are made for affiliates that did not report in that
survey. In contrast, for income and its components, except capital
gains/losses, and for fees and royalties, the estimates cover all

For the latter items, the estimation procedure used is designed to
ensure coverage as complete as that in the 1980 benchmark survey. Thus,
estimates had to be made for affiliates that filed complete reports in
the 1980 benchmark survey but did not report in the quarterly sampel
survey, either because they were exempt from the quarterly survey or
should have reported in that survey but, for some reason, did not. The
esstimates for affiliates that did not report in the quarterly survey
are then added to the reported data of affiliates that did report to
obtain the universe estimates.

For most nonreporting affiliiates, the estimate for the current
period is the product of two factors: (1) the prior-period estimate for
the affiliate and (2) the ratio of current- to prior-period data for a
matched sample of affiliates–i.e., affiliates that reported in both the
prior and current periods–that are in the same industry as the
affiliate whose data are being estimated. The multiplication of the
first factor by the ratio essentially assumes that, in the given
industry, data for each nonreporting affiliate change at the same rates
as data for affiliates in the matched sample.

In four instances, the calculated ratio for the matched sample was
not used in estimating data for nonreporting affiliates. First, the
largest nonreporting affiliates were scrutinized individually and, where
available information suggested that applying the ratio would give an
erroneous result, an alternative, estimate was made. Second, if the
ratio was biased by the data for one or two large reporters, or was
unrepresentative because of low coverage, it was adjusted before being
applied. Third, if the calculated ratio for earnings was undefined
because it was derived from numbers of opposite sign, a substitute ratio
was developed. Fourth, if a ratio for earnings was derived from numbers
of the same sign, but was to be applied to numbers of the opposite sign,
the results would not have been meaningful; in these cases, the
calculated ratio was inverted before being applied, as discussed below.

This procedure for expanding sample data to universe levels
differs, in several respects, from that used in the past:

1. For all items except distributed earnings, the “prior
period” is now defined as the immediately preceding quarter; for
distributed earnings, it is the same quarter of the preceding year. In
the past, the “prior period” was the same quarter of the
preceding year for all items.

The past procedure took into account the impact of seasonal
variations on the estimates. Review of the estimates over time,
however, now indicates that, despite seasonal variations, current-period
data for all items except distributed earnings tend to be estimated more
accurately using data for the immediately preceding quarter that for the
same quarter of the preceding year. For distributed earnings, in
contrast, data for the current quarter bear a much stronger relationship
to those for the same quarter of the preceding year, because affiliates
often make earnings distributions only once a year and usually in the
same quarter each year.

2. The ratios of current- to prior-period data for affiliates in
the matched sample are now calculated for 14 industries, instead of for
a matrix of 11 industries of affiliate by 19 areas or countries of
foreign parent. The more aggregated level now used reduces the
likelihood that a ratio will be unrepresentative because it is based on
data for only a few affiliates. Also, ratios are no longer computed by
country, because an affiliate’s industry and the economic
conditions in the United States have been found to be more important
than the country of the affiliate’s foreign parent in explaining
changes in income and in fees and royalties.

3. In a given industry, a separate current-period estimate for each
nonreporting affiliate is now obtained by applying the ratio of current-
to prior-period data for affiliates in the matched sample to the
prior-period estimate of each nonreporting affiliate. Previously, an
aggregate current-period estimate for all nonreporting affiliates in a
country-industry cell combined was obtained by applying the ratio for
the given country-industry cell to the aggregate prior-period estimate
for all nonreporting affiliates in the cell. Thus, only an estimate for
the cell as a whole was obtained; no detail below that level was
available. The new procedure has the advantage that universe estimates
in full country and industry detail, and for any combination of
countries or industries, can be obtained.

4. As noted earlier, data for capital gains/losses are now included
in the estimates of earnings and income only to the extent they are
actually reported in the quartely sample survey. Previously, capital
gains/losses were expanded to universe levels as part of the expansion
of total earnings.

5. Separate estimates are derived for payments and receipts of
interest on intercompany debt and for payments and receipts of fees and
royalties. In the past, universe estimates were derived only for net
payments (that is, payments less receipts).

6. A new procedure has been established to deal with cases in which
the ratio of current to prior-period earnings of affiliates in the
matched sample is derived from numbers that are of the same sign, but
the ratio is to be applied to an individual affiliate’s earnings
that are of the opposite sign. In such cases, the results are not
meaningful. The new procedure requires that the ratio be inverted
before it is applied. The inverted ratio yields an estimate that moves
in the correct direction and that usually shows an appropriate degree of
change. In the past, a less formal procedure was used. Sign reversals,
however, were not as much of a problem in the past, when the ratios were
applied to aggregate data in a country-industry cell rather than to data
for individual affiliates; earnings of individual affiliates tend to be
more volatile than those for an aggregate.

Data on capital inflows to acquire or

establish new U.S. affiliates

BEA’s survey of new foreign direct investments in the United
States (the BE-13), which was instituted in 1979, collects data on total
outlays to acquire or establish new U.S. affiliates. These data are
used to supplement the data on capital inflows reported in the quarterly
sample survey. As noted earlier, no estimate of equity capital inflows
is normally made for affiliates that are exempt from the quarterly
survey. Because newly established or acquired affiliates tend to be
relatively small, many of them are exempt. The BE-13 data indicate that
equity capital inflows to acquire or establish such exempt affiliates,
although small for any given affiliate, were large in total during
1981-83. Therefore, quarterly estimates of these unreported inflows
were made utilizing the BE-13 data.

Total outlays to acquire or establish new U.S. affiliates would
constitute direct investment capital inflows if they were made directly
by foreign parents. (They would also constitute capital inflows if they
were made indirectly by the foreign parents through their existing U.S.
affiliates, but the inflows probably would have been reported in the
existing U.S. affiliate’s quarterly report.) The BE-13 survey
indicates that direct outlays by foreign parents to acquire or establish
U.S. affiliates that were exempt from the quarterly sample survey were
$977 million in 1981, $663 million in 1982, and $314 million in 1983.
These data were added to direct investment capital inflows reported in
the quarterly surveys for the 3 years; for any given affiliate, the data
were included in inflows in the quarter in which the acquisition or
establishment occurred. Similar adjustments, based on the BE-13 data,
will be made to capital inflows in future years as well.

Change in treatment of unincorporated affiliates

For 1980 forward, the treatment of unincorporated U.S. affiliates
has been changed to parallel of incorporated U.S. affiliates, because
BEA now obtains similar detail for both types of affiliates. This
change is possible because, over time, the accounting procedures of the
two types of affiliates have become more similar. The change in
treatment affects only individual components, not the totals, for the
direct investment position, income, and capital flows.

Previously, the direct investment position in unincorporated
affiliates was reported as a single summary account, including both
foreign parents’ equity in, and intercompany debt with, these
affiliates. Now, the components are reported separately. The equity
portion is combined with equity in incorporated affiliates and shown as
the foreign parent’s equity in all affiliates; the intercompany
debt portion is combined with intercompany debt of incorporated
affiliates and shown as the foreign parents’ intercompany debt
position with all affiliates.

In addition, reinvested and distributed earnings of unincorporated
affiliates previously were not reported separately. Instead, in the
income account, total earnings of these affiliates were treated as if
they were distributed; in the capital account, earnings actually
reinvested were treated as transfers of cash or other assets from
parents to affiliates adn were included, along with other capital flows,
in a single summary account in the “equity and intercompany
accounts” component of direct investment capital flows. The term
“reinvested earnings” referred only to incorporated
affiliates. Now, in contrast, reinvested and distributed earnings of
unincorporated affiliates are reported separately. The distributed
portion is included, together with dividends of incorporated affiliates,
in distributed earnings of all affiliates, and, in both the income and
capital accounts, the reinvested portion is included, together with
reinvested earnings of incorporated affiliates, in reinvested earnings
of all affiliates.

Finally, capital flows to unincorporated affiliates, which–as
noted earlier–were previously shown together as a single summary
account in equity and intercompany account flows, have now been split
into the equity, intercompany debt, and reinvested earnings portions,
and each portion is combined with that for incorporated affiliates to
obtain a total for all affiliates. In the few cases where equity
capital of an unincorporated affiliate could not be separated from
intercompany debt, the entire amount was considered equity capital.

Because the totals for the direct investment position, income, and
capital inflows are not affected by the change in treatment of
unincorporated affiliates, the estimates of these items for 1980 forward
are comparable with those for earlier years. However, the components of
these items are not comparable. Estimates of the components on the new
basis for earlier years cannot be made because the necessary detail for
unincorporated affiliates is not available. For 1980 forward, however,
estimates on the new basis are separately available for both
incorporated and unincorporated affiliates (table 11).