U.S. Direct Investment Receipts and Payments: Models and Projections

International Finance Discussion Papers Number 140

May 1979

U.S. DIRECT INVESTMENT RECEIPTS AND PAYMENTS: MODELS AND PROJECTIONS

by

Lois Stekler

NOTE: International Finance Dicussion Papers are preliminary materials circulated to stimulate discussion and critical comment, References in publications to International Finance Discussion Papers (other than an

acknowledgement by a writer that he has access to unpublished materials) should be cleared with the author or authors,

U.S. Direct Investment Receipts and Payments: Models and Projections by

* Lois Stekler

The purpose of this study was to analyze the factors influencing direct investment receipts and payments and to make projections of net investment income. The models and techniques used are very similar, although the receipts side is disaggregated by industry, while the payments side is not. The paper is divided into three parts: direct investment receipts are analyzed in Part I, direct investment payments are analyzed in Part II, and the implications for net direct investment income are discussed in Part III,

I. Direct Investment Receipts

Direct investment receipts, as reported by the Department of Commerce, are shown on Table 1, In 1977 28 per cent were from petroleun, 37 per cent were from manufacturing, and 35 per cent were from all other, Receipts were disaggregated by industry so factors specific to an industry

‘ could be considered, and because relationships need not be identical across industries, The particular industry breakdown was dictated by the availability of data. The model is presented in the first section, followed by a summary of the statistical results, The third section outlines the assumptions used in making the projections, and the last section analyzes the projections.

* The analysis and conclusions of this paper should not be interpreted as

representing the views of the Board of Governors of the Federal Reserve System or anyone else on its staff,

- Q-

Table 1. U.S. Direct Investment Receipts (millions of $)

Dividends, interest

+ earnings of Reinvested Year Receipts affiliates Earnings 1960 3,621 2,355 1,266 1961 3,822 2,768 1,054 1962 4,242 3,044 3,044 1963 4,636 3,129 3,129 1964 5,105 3,674 3,674 1965 5,505 4,963 3,963 1966 5,258 . 3,467 3,467 1967 5,604 3,847 3,847 1968 6,591 4,151 4,151 1969 7,649 4,819 4,819 1970 8,168 4,992 2,671 1971 9,159 5,983 2,641 1972 10,949 6,416 2,949 1973 16,542 8,384 4,330 1974 19,157 11,379 7,356 1975 16,595 8,547 7,644 1976 18,999 11,303 8,955

1977 19,851 , 12,540 10,881

= 37 a) The Model

Actual receipts on U.S. direct investment (YR) in a particular year are the product of both long run or expected forces (EYR), and temporary or

unexpected forces (Xi's). If nothing unexpected happens, the Xi's will be defined

equal to one and actual receipts will equal expected receipts,

YR = (EYR) a ae

In order to explore what determines expected receipts, consider a specific direct investment project: It requires a certain dollar expenditure

this period (1), and is expected to produce a certain level of average earnings,

2/

constant in real terms (EN) in perpetuity. The expected real rate of return

on the project (r) equals annual earnings in foreign currency (EN) divided by

3/

the foreign currency value of the initial investment {I/1,)~ » or

EN = r e I/n, | (2)

If inflation and exchange rate changes do not alter real rates of return on

investment in a predictable way, then nominal earnings in a currency would be

4 expected to increase at the same rate as average prices.

Po 7% Be (2a)

,

Expected nominal earnings converted to dollars would equal the rate of return times the initial investment adjusted for subsequent exchange rate and

price changes,

1/ An alternative approach would be to assume that expected receipts equal the current rate of return on direct investments times the current market value of assets, The major problem with this approach is that we have no data on the current market value of direct investment assets, and only assumed proxies for the current rate of return,

2/ Lags will be considered later. The assumption of a perpetual income stream is not as far-fetched as it first appears if capital consumption allowances accurately reflect depreciation in the economic sense, The Department of Commerce data excludes depreciation from both income and new investment, The reinvestment of capital consumption allowances maintains the productivity of the capital stock.

3/ ™% equals the exchange rate (cents per unit of foreign currency) in the year when the investment was made, Ty equals the current exchange rate,

4/ If inflation or exchange rate changes alter relative prices, in particular the price of the product produced by the investment relative to the costs of inputs, then real rates of return would be affected.

re az as) 4 im rd 1

nL rn nn: (2b)

Total expected U.S. receiptSon direct investment (EYR) would equal the sum over all previous investments of the rate of return times the initial amount

invested adjusted for subsequent price and exchange rate changes.

n Po TM i

If r is redefined as some average rate of return for all investments, then expected

receipts equal average r times the "current" dollar value of previous investments.

EYR,=4r- v4 Li [Pe . rt | =r AR, (3a)

This variable, AR.» does not correspond to the Department of Commerce

data on the U.S. direct investment position.2/ The variable ARt was calculated

by starting with the Department of Commerce 1957 benchmark survey, and each

year (1) inflating the asset position by world average price increases adjusted 6/

for exchange rate changes and (2) adding the current direct investment outflow. Since assets acquired by the end of the year have not been held on average for the full year and since newly acquired assets may not earn at their full rate

7 immediately, new capital outflows are assumed to effect earnings with a lag.~

5/ The Commerce data is the sum over past years of nominal investment flows, it is not in constant or current dollars,

6/ A geometric average of 10 countries was used, with weights dependent on shares in U.S. direct investment according to Department of Commerce data for selected years between 1960 and 1977, _ Separate weights were calculated for petroleum, manufacturing, and ' ‘other"

7/ Assets acquired during year t were assigned a weight of .25 and assets acquired during year t-1 a weight of .50. Assets acquired before year t- 1 were assigned a weight of 1.00.

Actual receipts may not always equal expected receipts. Short run or unexpected factors (Xi,s) influence profits by altering demand for the firms output or by changing the relative prices of inputs and outputs, While many such factors influence receipts, only two were included in this study: fluctuations in the growth of aggregate demand and the influence of OPEC price and tax decisions on petroleum receipts, If imvestors increase capacity on the assumption that aggregate demand and sales will continue their past trends, then if aggregate demand falls below its trend value, profits will be lower than expected, On the other hand, if aggregate demand is above the trend value profits will be higher than expected?! The variable (CW) has been included to take this into account. Each country's actual GNP in constant prices is divided by the trend value for 1960-1977; then the country values are averaged .2/ Nominal petroleum income is likely to rise faster than the world average rate of inflation (adjusted for exchange rate changes) if petroleum

prices rise more than other prices. Petroleum income is also affected by

8/ If capacity is no longer growing at historic trend rates then the assumption

that it is will lead us to underestimate profits and U.S. direct investment receipts. 9/ A geometric average of eleven countries (including the U.S.) was used with multilateral trade weights, Trade rather than direct investment weights were used en the grounds that sales and hence profits of U.S. direct investment frequently depend on the world market and not just on the domestic market of the country where

the investment was made. The trade weights are shown on Table A6, the CW values on A7.

-6changes in royalties and tax payments to producing country governments, An additional variable, S, was included in the petroleum regressions to take these influences into account. The variable S equals the price of petroleum per barrel adjusted to subtract royalty and tax payments per barrel divided by the world average general price index adjusted for exchange rate changes. (See Table A7) The special problems encountered when analyzing the petroleum industry are discussed in greater detail in Appendix B.

A dummy variable, DR, was also included to take into account the break in the Department of Commerce direct investment income series, 1965-1966 22!

In summary, nominal income on U.S. direct investment is a function of

the rate of return, the current value of assets, world capacity utilization, a.

11/

dummy and a petroleum price variable for that industry— The regression equations are as follows: logYR = a, + ay log AR + ay log CW + a3 DR + ay log S$ (4)

The asset coefficient ay is expected to be around 1. It would equal 1 exactly if there were no measurement or specification errors, if we had accounted for all factors which would cause actual earnings to deviate from expected earnings, and if there were no fluctuations or systematic trends in the rate of return on investments. In that case, the average expected real rate of return (r) on U.S.

direct investment would equal the antilog of the constant a,«

i0/ The Commerce Department 19¢6 D.I benchmark survey led to a substantial revision

“downward in their estimates of aggregate receipts. The series was revised back to 1966 but the years 1960-1965 were left unrevised. The break in individual industry series was even larger, since after 1966 many firms with a variety of activities were included in the "other" category, rather than having their earnings broken down by industry.

11/ pr equals 1 for the years 1960-1965 and zero for the years 1966-1977.

-quaToTyye0o yoee moTeq sTseyqueied uy savedde o73872e38 ,,3,, OUL /¥

oer canaienenemntnnineniepeneneeinmmmmninastentielteie hehe

(9°02) (7$") (42°) (€z"T)

90° Le°l 66° S8" €0"- ) ie 9e°T 72430 (29° 6T) (61°) (61° 4) (€6°T-)

60° 67°T 86° 00°T Le° OT’e ZY T~ Zutanjoesznuey

Coz"#7) (S€°7T) (87° €) (zs"4 ) (LE*Z=)

Ol OT’? 96° ce’ 80°T L¢e° ee’ Ge°v- und TOt3eg

ere snares spite his re el et iP shi PP Ss S uot #eZTTTIN swooUurl “was “M’a Y [orjed soOSssy Aun Aqtoedeg 4% 7uUB suo) quowjseauy 29e1TC

en

suotjenby uotsserZey eyl °*7 eTqeL

-i-

b) The Regression Results

Annual data for the years 1960-1977 were used in the regressions. Table 2 summarizes the results for the three industry breakdowns: petroleum, manufacturing, and other. The R2's are all high; the standard errors of the regressions range from .06 to .10. The actual receipts, fitted values and residuals for each industry are shown on Tables Al-A3.

The asset coefficients are significant in all cases: the values vary around 122 An asset coefficient exactly to 1 would mean that receipts increased proportionately with assets. The capacity utilization coefficients are significant in the petroleum and manufacturing industry regressions, but not in the "other" industry. A one per cent increase in theCW variable produces a 3 to 4 per cent increase in manufacturing or petroleum receipts. The petroleum relative price variable S is also significant, indicating that for each per cent increase in petroleum prices above general inflation, petroleum receipts will increase approximately by .30 per cent.

The average (pre-tax) real rates of return for each industry implicit in the regression results are shown on Table 3 for 1960 and 1977. Currently the real rate of return on petroleum assets is somewhat higher than for other industries. Over the period studied the rates of return appear to have changed

12a/ appreciably only in the "other" case.

12/ The standard error in the other industry case is ,04; the asset coefficient is significantly different than 1,

12a/The rates of return on Table 3 are not comparable to what the Department of Commerce calls the rate of return on the U.S. direct investment position for: several reasons, The Department of Commerce makes no effort to take into

account inflation and how it has affected the value of assets from

the date of the initial investment. The Department of Commerce rates are therefore substantially higher than the rates on Table 2. In addition the Department of Commerce uses actual receipts each year to calculate the rate of return. The

rates on Table 2 are calculated using a regression estimate of what receipts would have been if the CWvariable equaled 1, its average value. The rates shown on Table 2 abstract from the effects of aggregate demand fluctuations.

-9-

Table 3. Average Real Rates of Return on US DI by Industry

industry 1960 1977

petroleum 10.2 10.0

manufacturing 8.1 8.0

other 11.7 8.5

- 10 -

c) Assumptions Used in Projecting Receipts In order to use the regression results, presented in section 2,

to project U.S. direct investment receipts for 1979 and 1980, forecasts must first be made for the following variables: the increase in the real value of the U.S. direct investment position in 1979 and 1980, the average change in world prices and the value of the dollar, and the level of average world capacity utilization for those years.

| The real value of the U.S. direct investment position was projected

13/ We

first, by calculating the log linear trend from past data (1960-1977). assumed that asset growth over the next several years would be below the historic trends for both theoretical and empirical reasons, First, both the relative slowing of economic growth in developed countries abroad and the recent exchange rate changes were expected to make investments in productive facilities in the U.S. relatively more attractive. Second, the trend predicted increase in U.S. direct investments abroad implied a higher capital outflow than the rate observed in the first half of 1978. As a result we projected thatU.S. direct investment assets would grow at 4/5 their historic trend rate for 1978, 1979, 1980.

The inflation and exchange rate projections used are based on the assumption that relative purchasing power parity will hold in 1979 and 1980. We assume that foreign prices adjusted for exchange rate changes will increase as rapidly as U.S. prices. U.S. prices are assumed to increase 7.5 per cent in both 1979 and 1980, 44/

The world-wide average capacity utilization variable is assumed to stay at the 1978 level in 1979 and 1980. The petroleum S variable is calculated assuming a 15 per cent average price hike in 1979, and no further change in the real price of petroleum in 1980. 13/7 The average percentage increase in real assets over the period 1960-1977 were 6.0 per cent for petroleum, 9.6 per cent for manufacturing and 7.7 per cent for

“other''. The annual percentage increases are shown on Table AS.

14/ See the Wall Street Journal, March 9, 1979, page 1 for a discussion of Administration inflation estimates.

- ill -

These assumptions are undoubtedly subject to a substantial margin of error. Therefore we will also explore the sensitivity of the projections to variations in these assumptions,

d) The Projections

The projections of percentage increases in direct investment receipts based upon the regression coefficients and the assumptions specified in the previous section are shown on Table 4. Overall, the model projects continued rapid growth in U.S. direct investment receipts. The projected percentage increase in receipts in 1979 is particularly large for the petroleum industry.

These projected percentage increases would, of course, be altered by any change in the assumptions specified in part 3. A rough idea of the affect can be obtained by using the regression coefficients on Table 1. For example, if either real asset growth or world prices were 1 per cent higher or lower, direct investment receipts from manufacturing would change by roughly 1 per cent. If OPEC price increases in 1979 were limited to the 10 per cent average originally announced then petroleum receipts would increase only about 14 per cent in 1979 43/ Petroleum profits are also likely to be sensitive to OPEC decisions on output and taxes.

Projections of the value of direct investment receipts in 1979 and 1980 can be obtained by adding the model's projected percentage annual increases to the fitted value for 1977. These projections are also shown on Table 4. They are based upon the fitted rather than Department of Commerce value for 1977 for several reasons. First, the Commerce data are still preliminary; they have, not yet been revised on the basis of firms' annual reports. Second, realized and unrealized capital gains (losses) cause temporary fluctuations in reported

157 This is the effect, assuming no change in inflation rates, exchange rates, and capacity utilization.

-12-

Table 4. Projections of Direct Investment Receipts

industry 1978 1979 1980 bil. of $ per cent bil. of $ per cent | bil of $ per cent change change change petroleum 5.8 +6 .8 6.7 +15.6 7.7 +13.9 manufacturing 8.4 +12.7 9.5 +13.7 11.0 +15.7 other 8.2 +14.2 9.2 +11.9 10.3 +12.0

Total 22. 11.7 25.5 ‘13.6 29.0 13.9

-13-

receipts. Using the fitted value for 1977 rather than the Commerce figure will give more accurate projections, if the difference between the two is accounted for by temporary factors.

The treatment of capital gains (losses) presents special problems, If the subsidiary of a U.S.firm bought an asset ten years ago and sells it today at a higher nominal price, but the same real price, the Department of Commerce data will shown higher income for this year. Although some of the variables in the model might influence these capital gains and although the receipts data used in estimating the coefficients does include them, it is unlikely that the model will accurately predict the size or timing of these realized gains.

The same is generally true for unrealized capital gains or losses due to- exchange rate changes, reported in receipts since 1976. Since the currency denominations of firm's assets and liabilities can be altered, one cannot necessarily project that what happened in the past will be repeated as the result of a similar exchange rate movement.

The influence of these temporary factors is one possible reason why the model's projections for 1978 differ from the Department of Commerce's preliminary data. Annual data by industry is not yet available, but the Department of Commerce preliminary aggregate estimate is $24 billion, in contrast to the models’ projected $22.4 billion. On the basis of three quarters, it appears that the model's projection for petroleum will be slightly above the Commerce Departments figure, while the model's projection for manufacturing will be substantially below, In the case of petroleum, there is partial evidence that receipts, as reported by the Department of Commerce

16/

were temporarily depressed by foreign currency translations losses =~ In the

16/According to the Wall Street Journal (January 26, 1979, p. 6) Exxon reported translation losses of $307 million in 1978 and Texaco reported $105 million.

-14-

case of manufacturing, it is possible that firms had capital gains associated with the depreciation of the dollar, but there is no firm evidence at this date,

Another possible explanation of why the model's projection of receipts might be too low has to do with the CW variable. Profits and hence receipts are assumed to vary with real GNP relative to its trend value. In recent years real GNP growth has been substantially below previous historic trends in many countries. As a consequence, it is not unlikely that investors would revise their plans downward. Capacity would grow more slowly. Capacity utilization and hence profits would not be as low as implied by the CW variable. If we assume that CW in 1978 stayed at its 1977 level, instead of falling, the model's projection of receipts for 1978 would increase $.7 billion.

Although it is not yet possible to judge whether the difference between the model's forecast for 1978 and the preliminary Commerce estimate is due to permanent °F temporary factors, we can calculate how much higher receipts would be if the shift were assumed permanent. The projection for 1979 would be $27.3 billion and 1980 would be $31.0 billion. II. Direct Investment Payments

The framework used in analyzing and projecting direct investment payments is very similar *o that used for recaipts. The major difference is that payments are not disaggregated by industry: data by industry are not readily available on a consistent basis for years before 1974. The model and the data are discussed in detail in the first section, followed by a summary of the statistical results. Then there are sections discussing the assumptions

used in making the projections and the projections themselves.

a) the model and the data The payments-model is analogous to the receipts mdel presented in part I, section a. Actual payments are assumed to be the product of both expected and unexpected factors. Expected nominal payments (EYP) equal the sum over all previous investments of the rate of return times the initial

investment times the change in the price level since the initial investment

17 was made! EYP = r,-I,+ P 5 Po If r is redefined as an average rate of return then pyPe sre My Wy Pe wt APE (6) Po.

Actual payments (YP) will not always equal expected payments because of U.S. business fluctuations. The higher real U.S. GNP compared to its trend value (CUS), the higher business profits and payments are expected to be.

Because the Department of Commerce payments data is not a consistent series, a dummy variable (DP) was included in the regression equation and certain adjustments were made in the data. The line between portfolio and direct investment was redrawn in 1974 to include as direct investment any interest greater than 10 per cent (instead of 25 per cent). Some increase in payments would be expected as a result, but it is not possible to use overlapping data to judge how much 22! A dummy variable equal to zero from 1974 was included in the regression equation to take into account this change in definition. 17/Recent capital flows are assumed to have the same lagged affect on payments as on receipts.

18/ The overlapping data is strongly affected by a revision in the treatment of a petroleum company.

-16-

From 1974 on, unrealized capital gains (losses) are included both in payments and capital inflows. Table 5 shows payments and capital inflows including and excluding estimated capital gains (losses). These capital gains and losses largely reflect changes in the values of insurance companies' portfolios. In the statistical analysis and projections, consistent series excluding capital gains have been used for both direct investment payments and capital inflows. In order to project the Department of Commerce direct investment payments number, capital gains on insurance company portfolios would also have to be projected 2!

b) the Regression Equations

The regression equation was

log YP = ay + ay log AP + a, log CUS + a, DP (7)

The regression results using annual data since 1962 are shown on Table 6. The R is .99; the D.W. statistical does not indicate serial correlation in the residuals. Both the asset and the income coefficients are statistically significant. The asset coefficient is just slightly below 1, indicating that the increase in payments has been almost proportionate with the increase in assets. The income coefficient indicates that payments are rather sensitive to U.S. business fluctuations. The 1977 implied real "long run" rate of return on foreign direct investments is approximately 8 Der cent 22 c) The Projections

In order to use the model to project future direct investment

payments, projections are needed of (1) the real growth of foreign direct

investment in the U.S.,(2) U.S. price changes, and (3) U.S. GNP growth

relative to trend,

19/These capital gains or losses would reflect stock and bond price movements and

the skills of portfolio managers. According to anecdotal evidence, these managers have adjusted portfolios to avoid repetition of the 1974 losses,

20/ The “long run" rate of return is the rate for 1977 (before-taxes), if U.S.

GNP had equaled its trend value. ©

-17-

Table 5. U.S. Direct Investment Payments and Capital Inflows Including and Excluding Capital Gains a/ (millions of dollars)

Pa ents! Direct Investment Inflow’! Year excluding inc luding excluding including C.G. C.G. C.G. C.G. 1962 399.3 346.4 1963 459.0 231.0 1964 528.7 321.8 1965 656.6 414.9 1966 710.9 425.2 1967 821.1 698.1 1968 876.3 807.3 1969 847.9 1,262.7 1970 874.5 © 1,463.5 1971 1,163.7 367.4 1972 1,255.7 . 949.0 1973 1,609.6 2,800.0 1974 1,908 .8 1,330.8 5,338.2 4,760.2 1975 1,891.3 2,234.3 2,260.0 2, 603. 0 1976 2,620.3 3,110.3 3,856.5 4, 346.5 1977 2,742.9 2 ,828.9 3,251.7 7

a/ Capital gains (losses) are approximate figures froma partial sample, provided by the Department of Commerce. They should be treated as rough estimates. ; b/ Survey of Current Business, Table 1, U.S. International Transactions,

line 27, Direct Investment Payments.

c/ Line 65, Direct Investment Capital Inflow.

-18-

Table 6. Regression Results Explaining U.S. Direct Investment Payments

Variable coefficient

constant

log nominal Direct Investment Assets 16.37 log U.S. GNP actual/trend 5.17 dummy 1.77 R? .99

DW 1.96

S.E.E. .07

Since 1962 the real growth of foreign direct investments

in the U.S. has averaged about 8 per cent per year. Since 1973 assets have increased more rapidly, about 12.5 per cent per year. This higher rate of increase has been used in the projections on the grounds that the forces that have attracted foreign investors to the U.S. since 1973 are likely to continue into 1979 and 1980 21/ These forces include exchange rate changes and the slowing down of economic growth relative to historic trends in Japan and many European countries. The nominal direct investment capital inflow for 1978 implied by a 12.5 per cent real growth in assets would be approximately $6 billion. U.S. prices are assumed to increase 7.5 per cent in 1978, 1979, 1980. U.S. real GNP is assumed to have grown 3.9 per cent in 1978, and capacity utilization is assumed to stay at the 1978 level in 1979 and 1980. Again, these assumptions may be subject to a wide margin of error. The reader can use the regression coefficients on Table 6 to adjust the projections for changes in assumptions. Combining these assumptions with the regression coefficients provides estimates of the percentage increase in U.S. direct investment payments for 1978, 1979, 1980., shown on Table 7. Also shown are dollar projections, calculated '21/ Since 1973 the annual percentage increases have varied widely (between 7 per cent and 22 per cent). If we are correct in assuming a distributed lag between new investments and payments, then the increase in payments will be steadier than the growth of real assets, and assuming that real assets are growing at the average

rate will not cause serious errors in projections of payments. See Appendix Table A5 for the yearly percentage changes.

- 19 -

by adding the predicted percentage changes to the model's fitted value for

1977. Table 7. Projected Direct Investment Payments Year Per cent change Payments (mil. $) 1978 22 3,644 1979 20 4,377

1980 20 5,258

The model projects continued rapid growth in direct investment payments , assuming no slowdown in U.S. economic growth. If, on the other hand, U.S. real GNP growth in 1979 were only 2 per cent, then the percentage increase in payments would fall to approximately 12 per cent. The payments projection is very sensitive to the U.S. real GNP growth assumption,

Revision of the inflation assumptions would also change the payments projection; for each additional per cent increase in U.S. prices, payments would increase -96 per cent.

The reader should keep in mind that the projections on Table 7 exclude capital gains or losses on firms' asset portfolios, It is therefore not possible to check the model's prediction for 1978 against the Department of Commerce figures, until estimates of capital gains and losses are also available. The preliminary Department of Commerce estimate for direct investment payments (including capital gains) is $3,730 million for 1978,

almost $100 million higher than the models projection (excluding capital gains).

III. Projections of Net Investment Income

Combining the projections of direct investment receipts and direct investment payments produces the projections of net direct investment income shown on Table 8. Over the next several years payments are expected to grow

at a more rapid rate than receipts. But since the absolute size of receipts is

- 20 -

many times larger, the dollar growth in receipts is larger. Net direct investment

income is projected to increase in 1979 and 1980.

Table 8. Projected Net DI Income

DI DI receipts payments net DI income 1978 24.0 3.7 20.3 1979p 25.5 4.4 21.1 1980p 29.0 5.3 23.7

p = projected

Table Al. Petroleum Industry Receipts (millions of dollars)

Year Actual Fitted Residual

Year ss Actual Fitted = Residual 1960 1,309 1,314 -5 1961 1,497 1,420 77 1962 1,713 1,557 156 1963 1,835 1, 723 112 1964 1,821 1, 935 -114 1965 1,853 2,147 -294 1966 1,496 1,647 | -151 1967 1, 765 1,721 44 1968 1, 983. 1,915 68 1969 2,026 2, 245 -218 1970 2,456 2,448 8 1971 2,878 2,632 246 1972 3,095 3, 341 -246 1973 5,717 4,897 820 1974 6, 963 6,841 122 1975 4,795 4, 323 472 1976 5,123 6,007 -884

1977 5,481 5,450 31

Table A2, Manufacturing Receipts (millions of dollars)

Year Actual Fitted Residual 1960 1,181 999 182 1961 1,161 1,129 32 1962 1,265 1,318 -53 1963 1,527 1,524 3 1964 1,827 1,884 -57 1965 1,989 2,254 -265 1966 1,868 1,860 8 1967 1,863 2,083 | -~220 1968 2,411 2,434 -22 1969 3,113 2,895 218 1970 3,133 3,332 -199 1971 3,492 3,808 ~316 1972 4,740 4,526 214 1973 6,579 5,791 788 1974 6, 684 6,324 360 1975 5,998 6,305 -307 1976 7,223 7,009 214

1977 7,326 7,452 -126

Table A3 Other Industries' Receipts (millions of dollars)

Year Actual Fitted Residual 1960 1,131 1,119 12 1961 1,164 1,180 -16 1962 1,264 1,260 4 1963 1,274 1,356 -82 1964 1,457 1,453 4 1965 1,663. 1,566 ~ 97 1966 1,895 1,779 116 1967 1,977 1,974. 3 1968 2,198 2,173 25 1969 2,510 2,389 121 1970 2,580 2,692 -111 1971 2,790 3,070. -280 1972 3,113 3,537. -424 1973 4,246 . 4,119 : 127 1974 5,509 | 4,893 616 1975 5,802 5,806 -4 1976 6,653 6,575 78

1977 7,044 7,189. -145

Table A4, Direct Investment Payments (millions of dollars)

Year Actual Fitted Residual 1962 399 426 -26 1963 459 459 -0 1964 529 523 6 1965 657 610 47 1966 711 727 -16 1967 821 765 ; 56 1968 876 878 -2 1969 848 964° -116 1970 875 946 , -72 1971 1,164 1,044 120 . 1972 -1,256 1,266 -9 1973 1,610 1,547 62 1974 1,909 1,852 57 1975 1,891 1,875 16 1976 2,620 . 2,481 138

1977 ~ 2,743 3,011 -268

Table A5. Annual Percentage Increases in Real U.S, Direct Investment Assets Abroad and Foreigners' Direct Investment Assets in the U.S,

Year Foreigners’ Assets 1960 5.8 14.7 8.9

1961 8.6 8.2 7.4

1962 6.2 10.1 7.5 5.3 1963 7.1 13.8 5.0 3.3 1964 4,9 15.1 4.2 4.4 1965 6.4 13.6 . 9.1 5.3 1966 5.3 11.1 11.7 - 5.0 1967 6.7 7.9 7.6 7.7 1968 6.7 8.1 7.5 7.9 1969 3.9 10.0 7.9 10.9 1970 8.6 7.4 9.6 10.7 1971 6.8 7.7 6.8 2.3 1972 4.7 7.7 5.6 5.7 1973 3.8 9.5 8.5 15.0 1974 7,2 8.9 10.5 22.4 1975 8.5 5.1 6.9 7.1 1976 3.1 5.0 5.7 10.7 1977 3.6 3.8 6.2 7.6

Table A6. Trade and Direct Investment Weights

Direct Investments Petroleum Manufacturin

Canada Japan

U.K, Germany France Italy Belgium Netherlands Switzerland Sweden

U.S,

Trade Weights

072 -107 094 - 164 . 103 O71 050 2065 .028 .033 213

Table A7, Variable Values: S, CW, CUS

Year i) CUS CW 1960 1.50 96 96 1961 1.42 96 97 1962 1.31 98 98 1963 1.17 -98 98 1964 94 1.00 1.01 1965 89 1.02 1.01 1966 86 1.05 1.01 1967 80 1.04 1.01 1968 -68 1.05 . 1.02 1969 64 1.04 1.04 1970 225 1.00 1.04 1971 50 1.00 1.03 1972 209 1.02 1.03 1973 89 1,04 1.05 1974 2.26 99 1.02 1975 .73 94 95 1976 1.40 97 95

1977 -86 98 94

Appendix B: The Petroleum Industry: Special Problems

The variable S is included in the regression analysis, as an index of petroleum companies' revenues per barrel produced, This improves our ability to explain fluctuations in receipts from petroleum, but many problems remain,

First, the measure of company revenues per barrel produced is crude, Only one petroleum price series is use (light crude, Ras Tanura, Saudi Arabia). In the early years of this study, transactions did not necessarily take place at the posted price, so the actual average price is difficult to calculate! No account is taken of divergent price movements or shifts in proportions between different grades, No account is taken of differences in taxes or other charges between Saudi Arabia and other producer governments,

Second, revenues depend on output as well as profits per barrel, U.S, firms do not have the same percentage interest in all producing countries, When world demand fluctuates, U.S, receipts will also depend on which producing countries cut or expand production, In the OPEC context, this is a political decision and not necessarily predictable on the basis of past behavior, Similarly, the impact of the current crisis in Iran on U,S, petroleum receipts will depend not just on what happens to petroleum prices and producer country taxes, but also on how large the U.S, interest is in the countries that expand output,

*/ Edelman, M,A,, The World Petroleum Market Johns Hopkins University Press, Baltimore Maryland, 1972 pp.160-194

Third, recorded direct investment capital flows in the petroleum industry do not always reflect changes in claims on profits, Because of the accounting practices of certain companies, fluctuations in the rate of production cause large recorded capital flows that do not necessarily cancel each other out within a calendar year.

Fourth, takeovers of increased ownership by producing countries may not be accurately measured in recorded capital flows and if accurately measured would not affect U.S. receipts with the same distributed lag that is assumed for increases in U.S, assets abroad, For example, Venezuela has yet to complete compensation payments to petroleum companies for takeovers in 1976,

Fifth, the FASB accounting standards in force starting 1976 require reporting unrealized capital gains or losses as part of income.

The petroleum companies typically have assets denominated in dollars and liabilities denominated in foreign currency, As a result, when the U,5, dollar depreciates, petroleum receipts appear temporarily depressed compared to what they otherwise would have been,

Finally, it is puzzling why the asset coefficient in the petroleum regression is greater than 1, This implies, in a sense, that the marginal return on investment has been greater than the average during the period studied, One possible explanation is that the variable 5 does not adequately deal with the shift in earnings due to OPEC actions from 1973 on. An alternative hypothesis might be that the petroleum companies’ perception of the risk of investment in many countries has increased over the period

studied, and hence the rate of return necessary to induce new investment

has increased, A second tentative hypothesis might be that there has been a shift in the mix of petroleum companies’ activities away from capital intensive production to services and distribution, Many producing countries have taken over ownership of production assets,

but the companies have continued to make profits from activities in these countries, It is possible that in the future petroleum receipts will not

continue to grow-more rapidly than assets, despite the regression

coefficient of 1,08,