Journal
of International
“BALANCE
Economics
1 (1971)
103-117.
@ North-Holland
AN EMPIRICAL
TEST OF THE
OF PAYMENTS
STAGES”
Publishing
HYPOTHESIS
Company
*
Nadav HALEVI The Eliezer Kaplan School of Economic and Social Sciences, The Hebrew University, Jerusalem, Israel
1. The concept
of balance of payments
stages
The idea that a country goes through a number of distinct balance of payments stages, starting as an immature debtor and ending as a mature creditor, goes back at least to Cariness (1874). Discussing borrowing by colonies, Cairnes traces the changes in the balance of payments from an import surplus financed by borrowing to an export surplus in goods to pay interest charges, and on to repayment of debts and lending abroad (Cairnes, 1874, pp. 355-63). The essential element of all later refinements is a three-fold breakdown of the balance of payments - net goods and services excluding investment income, net investment income, and net capital transfers - to define balance of payments stages. Some writers list four (e.g. Bach, 1960, pp. 699-700), some five (e.g. Meier and Baldwin, 1957, pp. 8 l-82) and some six or seven stages (e.g. Enke and Salera, 1947, pp. 638-41). Crowther (1957), in what is probably the most detailed discussion of this concept, presents the following six-stage classification: 1. Immature debtors-borrowers: a deficit in goods and services excluding investment income ’ and a deficit in investment income, financed by net capital receipts. * This is a summary of part of a research project financed by the Truman Peace Research Institute of the Hebrew University. I am indebted to D. Landau and J. Mokyr for technical assistance in compiling the data, and to J.N. Bhagwati, E. Kleiman, M. Michaely, M. Sarnat and two referees for helpful comments on earlier drafts of this paper. i For the sake of brevity, in what follows the term “goods and services” refers to net transactions in goods and services excluding investment income, and the term “current account” refers to all goods and services, including investment income.
104
N.Halevi, The “balance of payments stages” hypothesis
2. Mature debtors-borrowers: a surplus in goods and services, but smaller than the deficit in investment income; therefore, still net capital receipts. 3. Debtor-lenders and Deb tor-repayers: the export surplus exceeds the deficit in investment income; therefore, net capital outflow. 4. Immature creditor-lenders: export surpluses in both goods and services and investment income; therefore, net capital outflow. 5. Mature creditor-lenders: a deficit in goods and services, more than covered by investment income; thus, still a net outflow of capital. 6. Creditor-drawers and borrowers: a deficit in goods and services, larger than the surplus in investment income; therefore, net capital inflows. If a country borrows, it must expect a growing interest burden and ultimate debt repayment. Much has been written about the credit worthiness of developing countries and the capacity to repay. 2 To the extent that net new lending could continue indefinitely to cover interest charges, a country need not move from Crowthers’ stage 1 to stage 2. 3 But even assuming that borrowing countries must ultimately move to stage 2, and on to stage 3 (debtor-repayer), two basic questions remain. First, why must countries move on to the creditor stages? Second, even accepting the balance of payments stages as a time sequence, why should this movement be related to the process and level of economic development. 7 4 The neo-classical answer, re-iterated most eloquently by Crowther, applies to both questions: the poor country needs the excess saving of the rich, and capital movements are in response to interest rate differentials. Traditionally, these differentials - and “need” for foreign capital - have been explained in terms of differences in the marginal productivity of capital. Thus, the balance of payments stages are implicitly viewed as a function of relative levels of development. 5 The neo-classical balance of payments stages thesis can be seen as arising from a two-country analysis: a developed country and a underdeveloped country or colony with clear potential for develop2 See, for example, Avramovic (1964). 3 Domar (1960) has shown that this depends on the relative size of interest and amortization charges and the rate of growth of new investments. 4 Taussig, for example, while clearly describing the balance of payments stages, does not connect them to stage of development (Taussig, 1928, pp. 130-31). s Nurkse (1958, p. 134) points out that return capital flow cannot be expected before creditor and debtor have changed place in their scale of development.
N.Halevi,
The “balance
of payments
stages” hypothesis
10.5
ment, which attracts capital. Ultimately the two must reverse position if the later stages are to be reached. The application of this thesis to a multi-country world, where all but the extreme income level countries have economic relations with both poorer and richer countries, and where not all poor countries are capital-attracting, is not obvious. In fact, one would not expect cross-section data for many countries to show the described relationship between level of development and balance of payments stages.
2. Empirical classification
by stages
The empirical classification uses published data on income per capita (usually 1963) and three-year average (generally 1962-64) balance of payments data. ’ Examination of the data shows that Crowther’s six-stage classification is still inadequate for classification of border-line cases: e.g., a country that no longer has an import surplus but is not yet a net exporter. For the preliminary grouping of countries, therefore, a twelve-category classification (summarized in table 1) is used. In category I are countries with a deficit in goods and services, excluding investment income, but which are not debtors in the sense that net interest payments are insignificant. This may be because countries are only starting to borrow, or because the current account deficits have been financed by unilateral transfers. Category II exemplifies the traditional immature borrower-debtor. III and IV still have a current account deficit, but due solely to interest charges on past borrowing. In V, exports cover interest charges, and in VI they are sufficient to decrease debts or lend abroad. Category VII includes both countries in transition from debtor to creditor and those who have not at all entered the lending-borrowing process. VIII is the traditional
6 Balance of payments data are not always reliable even for countries with sophisticated statistics. Particularly poor are estimates of capital transfers and investment income, the latter, unfortunately crucial for this type of classification, often is inconsistent, including various factor payments which are not strictly speaking capital charges. Thus, a careful country-by-country examination of the data would lead to some re-classification; probably Jordan and Malta would disappear from the “creditors” list.
106
N.Halevi, The “balance of payments stages” hypothesis
Classification
--___ Balance of payment category
Table 1 of balance of payments
categories.
______-
Net goods and services (excluding investment income) (1)
I II III IV V VI VII VIII IX X XI XII
scheme
Net investment income (2) -_____0 _ _ _ _ _ 0 + + + + +
Net total current account
Net capital and unilateral transfers
(3)
(4)
_
+ + t +
0 +
0 _
0 t +
0 _ _ _
0 _
0 t
_
+
lender-creditor; XI and X are current lenders by virtue of charges on past lending; the latter already has an import surplus. Category XI is a creditor but no longer a lender, and XII is eating up past foreign investments. Table 2 presents a frequency distribution of the 81 countries classified by these twelve categories and by income per capita.7 What is most striking is the small number of creditors and their classification: only one country is a mature creditor in the traditional sense, only one has a goods and services deficit covered by interest receipts, and three, including two problematic countries, are eating up past investments. Category I is of interest. Thirteen countries, ten of them at a very low income level, had goods and services deficits but not net interest payments. The net investment income item is a function of past borrowing, which our figures do not examine. However, the ten low 7 For such a classification an arbitrary decision must be made to determine when an item is “negligible”; for net investment income, negligible is set at less than two per cent of total imports of goods and services; for columns (l), (3), and (4) of table 1, negligible is less than fwe per cent of total imports.
N.Halevi, The “balance of payments stages” hypothesis
Frequency
distribution
of countries,
Table 2 by income per capita
and by balance
107
of payments
category
Balance of payments category
Income
per capita
class (in $)
33-98
110-196
202-286
327-390
441-477
542-832
996-1335
1483+Total
I II III
4 I _ _
6 6 1 _
_
_ 3 2 _
2 1 _ _
_ 3 2 _
2 _
3
2 _
1 1 _ -
_ 1 2 _ _ _
_ _ 2 1 _ _
_ _
_ _ _ _ _ _ _
1 _ 1 _ _
_ 12
IV V VI VII VIII IX X XI XII Total Source:
_
3 1 _ _ _ _ _
14
20
1 _ _ _
I 2 _
1 _
1 4 _ 1 _
1
_ _ _
_ 1
1 _
6
5
9
9
1 1 _ _ _
13 27 10 1 8 6 10 1 1 _
1
1 3
6
81
Computed from balance of payments data in International Monetary Fund, Balance of Payments Yearbook, Vol. 19, 1962-66, 9th monthly loose-leaf issue, June 1968, and from U.N. Yearbook of National Accounts Statistics 1966, pp. 730-34.
income countries here included were all currently receiving net foreign capital inflow in excess of five per cent of imports. Five were also receiving unilateral transfer in excess of long-term capital. The classification presented above hypothesizes that there is an orderly progression through balance of payments stages associated with increasing level of income. To test this relationship, some condensation of table 2 is necessary. The upper part of table 3 classifies all countries by income class and five balance of payments stages: immature debtor, mature debtor, debtor-repayer, neutral and creditor. The test applied is Kendall’s rank measures of association, which are rank coefficients of ~ correlation applied to data grouped in classes.8 The coefficients obtained are less than 0.2, which are too small to be statistically significant. The lower part of table 3 condenses the income classes into ’ See Kendall and Stuart (1967, pp. 562-565). a chi-square test. However, the immature-mature method in section 4 below.
These measures debtor position
are more suitable here than is also tested by the latter
N.Halevi, The “balance of payments stages” hypothesis
108
Summary
frequency
Table 3 of countries, by income ments category.
distribution
per capita
and by balance
Income per capita class (in $)
Immature borrowers
Mature borrowers
Debtorsrepayers
Neutral
Creditors
(I, II)
(III,W,W
WI)
(VII)
(VIII to XII)
33-98 1 lo-196 202-286 327-390 441-477 542-832 996-1335 1483+
11 12 I 3 3 3 1 0
Total
40
19
33-286 327-832 996+
30 9 1
Total
40
Source:
of pay-
Total
1 1 1 0 0 2 4 1
0 0 0 1 0 1 2 2
14 20 12 6 5 9 9 6
6
10
6
81
10 S 4
3. 2 1
3 2 5
0
2 4
46 20 15
19
6
10
6
81
2 4 4 2 1 2 1 3
Same as table 2.
three - poor, medium and rich - in order insure that random factors have not washed out the debtor-creditor status of countries with relatively small differences in per capita income. The rank measures of association are all under 0.2. Excluding the creditor countries and running the same test on the 7.5 debtor and neutral countries does not improve the results. Clearly, the concept of an orderly progression through balance of payments stages (or even through the first stages) as one moves up the income per capita scale is not borne out by the cross-section data. It may be argued that income per capita is not a satisfactory criterion for determining level of economic development, and ranking countries by stage of development. Granting that this is true for many purposes, it is difficult to see what other development criterion has significance for examining the hypothesized relationship between stage of development and of the balance of payments. Nevertheless, an attempt was made to test the relationship, using a more sophisticated system for classifying countries by level of development.
N.Halevi, The ‘balance of payments stages” hypothesis
Frequency
distribution
of countries,
by
stage
and
by
balance
of payments
Stage of development
Balance of payments category
Total B
C
8 4 1 0 0 0 0
12 2 2 3 0 0 0
16 7 1 0 0 0 0
1 4 1 I 1 1 2
37 17 5 10 1 1 2
13
19
24
17
13
A
Immature borrowers (1, II) Mature borrowers (III, IV, V) Debtors-payers (VI) Neutral (VII) Immature creditors (VIII) Mature creditors (IX, X) Creditors-withdrawers (Xl, XII) Total Source:
Table 4 development category.
109
Same as table 2, and Adelman
D
and Morris (1967).
The classification scheme and empirical classification adopted for this purpose was that of Adelman and Morris. 9 Applying their factor analysis to forty-one economic, social and political variables, these authors classified 74 developing countries into three groups. lo Table 4 presents a frequency distribution of 56 of the countries so grouped for which we had balance of payments data, plus 17 richer countries which are not included in their list, but which they would probably rank as developed. Applying the same rank measures of association to this classification gives very low - insignificant - coefficients. Condensing the table to only four balance of payments categories, i.e., combining all the creditors, does not give a better result. Excluding the developed countries (D) and running the same test on 56 countries gives no higher coefficients. The creditors are usually the richer, more developed, countries; but there is no visible clear progression through balance of payments stages associated with an increase in the stage of development. Thus, the statistical evidence fails to show the pattern predicted by traditional theory.
9 The fullest treatment and Morris (1967). lo Ibid., p. 170.
of their system,
and the source
for the data here used, is in Adelman
N.Halevi, The “balance of payments stages” hypothesis
110
3. Income per capita and deficits Fundamental to the traditional thesis is the relationship between level of income and deficits in the balance of payments. To examine this relationship, table 5 presents a frequency distribution of countries classified according to income per capita and balance of payments position. For purposes of classification the “deficit” and “surplus” categories include only cases where there are deficits or surpluses in each of the three years covered and the deficits (or surpluses) are at least five per cent of total imports (of “goods and services” in table 5, A, and the total current account in table 5, B). All other cases The number of clear-cut surplus are classified as “not clear-cut”.
Frequency
distribution
Income per capita class (in 8) b
of countries,
Table 5 by income per capita
Deficit
A. Goods
and by balance
Not clear-cut
Surplus
and services (excluding
capital
of payments.
Total
charges)
33-98 110-196 202-390 441-832 996-2562
11 9 9 6 2
1 5 1 2 2
2 6 8 6 11
14 20 18 14 15
Total
37
11
33
81
14 20 18 14 15 81
B. Current
a
account
33-98 110-196 202-390 441-832 996-2562
11 13 13 8 4
_
1 2
3 5 5 5 9
Total
49
5
2-i
2
a Classified as deficit (surplus) if there was a deficit (surplus) in each of the three years which averaged at least five per cent of imports (excluding capital charges - panel A; including capital charges - panel B). Not clear-cut - all other countries. b The range of each class is the actual lowest and highest income per capita of the class. Source: Same as table 2.
N.Halevi, The “balance of payments stages” hypothesis
111
countries is small - eleven according to the “goods and services” definition, and only five according to the total “current account” definition. The small number of observations in certain cells (zero in some) made it necessary to condense the frequency tables in order to apply a chi-square test of significant differences. The “surplus” and “not clear-cut” categories were combined to give just two categories: deficit and other. Alternative condensation of the income per capita classes were made. The following results of the chi-square test were obtained: (1) A three-fold division of income classes - up to 390 (poor), 441-832 (medium), and 996 and above (rich) - shows that the “deficit’‘-“other” distribution is not independent of income class, at the 95 per cent level of significance for panel A (the goods and services definition) and at the 99 per cent level of significance for panel B. (2) A two category classification - under and over 200 - shows the distributions are not independent at the 99 per cent level of significance, both for panels A and B. The same results are obtained for the two-category classification by income - under 833, and 996 and above. (3) A comparison of the two original classifications, panel A and panel B, without any condensation, shows that the two distributions are not significantly different, at the 95 per cent level of significance regardless of which panel is taken as the “expected” distribution and which is taken as the “observed”. These results show that the existence of a balance of payments deficit is not independent of the level of income per capita. Two attempts were made to quantify this relationship. For both attempts, the deficits and surplus figures are expressed as ratios of total imports of goods and services (exclusive of capital charges for the “goods and services” definition of deficit). First, rank correlation coefficients were computed, with countries ranked by income per capita and by balance of payments - from largest relative deficit to largest relative surplus with the following results: (1) For 79 countries (the total sample exclusive of South Vietnam and South Korea), the rank correlation coefficient is 0.259 for the “goods and services” definition, which is significant at the 99 per cent level. (2) For the same countries, the “current account” definition gives a
112
N.Halevi, The ‘balance of payments stages” hypothesis
rank correlation coefficient of 0.335, which is significant at the 99 per cent level. (3) Excluding an additional six countries (five low income countries with export surpluses and Israel) ’ ’ raises the rank correlation coefficient to 0.402 for the goods and services definition and 0.462 for the current account definition; again, both are significant at the 99 per cent level. For samples restricted to the 46 countries with deficits at least 3 per cent of total imports of goods and services exclusive of capital charges, and to 56 countries with deficits at least one per cent of total imports were estimated to test the of goods and services, ’ * regressions relationship between income per capita and relative size of deficit. The relationships estimated are linear in the logarithms. R* is 0.25 for the goods and services definitions, and 0.14 for the current account are significantly definition. Although quite low, these coefficients different from zero at the 99 per cent level. The regression equations are, for the goods and services deficit: log Def. = 2.00 - 0.373 log (Y/P), (0.097) for the current
account
log Def. = 1.852 -
deficit: 0.306 log (Y/P) . (0.103)
Since the regressions are log-linear, the coefficients are income elasticities; i.e. showing what percentage change in deficits is associated with a percentage change in income per capita. Thus, the level of income per capita is a factor explaining the relative size of balance of payments deficits; however, it certainly is not so important a factor as to justify exclusive reliance upon it for predictive purposes. ’ 3
r1 Their exclusion is desirable because they are not relevant for the thesis being tested. r* In these groups South Vietnam, South Korea and Israel are excluded; the reasoning is that in these cases the deficits clearly reflect foreign exchange supply conditions. l3 The income per capita data are often very rough estimates, and conversion to dollars problematic. For regressions, adjusted data such as in Hagen and Hawrylyshyn (1969) would give more exact results.
113
N.Halevi, The “balance of payments stages” hypothesis
4. Debtors
and creditors
The debtor-creditor classification appears to be where the data least fit the predicted pattern. Table 6 presents a frequency distribution of debtors, creditors, and “not clear-cut”; the latter includes all countries whose net investment income (plus or minus) was less than one per cent of total imports. There are only eight creditors (mostly rich), three of which are problematic; surprisingly, there are twenty which seem to be following Polonius’s sound advice to “neither a lender nor a borrower 200 with those in the be”. Comparing countries with income under 200-400 range shows a higher concentration of not clear-cut cases in the former; there is a higher concentration of such cases among the rich 996. Part of the departure from the in a comparison of over and under traditionally predicted pattern occurs in the distinction between immature and mature debtors. However, as table 3 makes clear the immature debtors include countries that are not yet debtors at all, as measured by net capital charges. An examination of the 59 immature and mature debtors in the first two columns of table’ 3 above, shows that at the extremes there appears to be some relationship between income class and stage of indebtedness; the poorer countries are mainly in the immature class, the relatively richer ones less so. But chi-square
Frequency
distribution
of countries,
Table 6 by income per capita
Income per capita class (in S)
Debtors
33-98 110-196 202-390 441-832 996-2562
9 13 17 9 5
0 1 1 1 5
Total
53
8
Creditors
and by debtor-creditor
Not clear-cut
5 6 0 4 5 20
position.
Total
14 20 18 14 15 81
a Classified as debtors (creditors) when the average of net payments (net receipts) of investment income exceeded one per cent of total imports of goods and services. Not clear-cut - all other countries. Source: Same as table 2.
a
114
N.Halevi, The “balance of payments stages” hypothesis
tests both for a five-income class distribution and a three income class one show no relationship significant at the 95 per cent level. The traditional theory equates transfers of real resources with the capital account of the balance of payments. But in the post-World War II world, unilateral transfers have emerged as a major source of finance. In the period here discussed, sixteen countries received unilateral transfers averaging more than 10 per cent of their total imports of goods and services; nine of these received over 20 per cent. Of the 39 countries classified as immature creditors (including categories 1 and 2) fully one third received unilateral transfers equal to at least half of the current account deficit, and seven received more than 7.5 per cent. There is no discernible pattern relating the relative size of unilateral transfers and income per capita, though no rich countries (except Finland) are included in this group. These figures do not, of course, reveal much about past unilateral transfers, which, by substituting for capital flows reduced current interest charges. But the large size of past transfers, especially to countries which by the early 1960’s were in the middle or even higher income brackets, certainly have much to do with “distorting” the pattern of investment income and thus the traditional classification scheme. The rich developing countries such as Canada, Australia and New Zealand, are among those which attract large foreign investments. Clearly, if capital flows are attracted by the return on capital, the experience of these countries suggest that no inverse association can be assumed between present returns on capital and present levels of income per capita. This point is also reflected in the large flow of investments from developed countries to other developed countries, i.e., from the U.S.A. to the EEC. ’ 4 Table 7, which presents a classification of net long-term capital by income class, shows that government capital went to poorer countries, while private capital went mainly to the middle-income-class countries. But since this table shows net rather than gross capital flows, it does not show the intradeveloped-country flows. For this, more detailed figures would be necessary. I4 Kuznets (1964, pp. 71-77) has shown that a large share of true lending income countries and to those closely associated with large development units.
goes to high
115
N,Halevi, The “balance of payments stages” hypothesis
Net long-term Capital
Income per capita (in $)
Table I capital movements.
transfers
Private A. in millions
33-98 110-196 202-390 441-832 832+
Total
Government
282.7 141.7 870.9 1788.2 _-2763.3
$ 1319.6 391.1 396.0 -84.4 -1722.6
1602.3 539.4 1266.9 1703.8 -4485.9
78.6 13.7 31.3 -5.0 -38.4
100.0 100.0 100.0 100.0 -100.0
B. Per cent 33-98 110-196 202-390 441-832 832+ Source:
21.4 26.3 68.7 105.0 -61.6
Same as table 2.
Classification
Table 8 of richer countries.
-___ Balance Income per capita (in $)
of payments
category Transactions
All transactions
With USA only
With other countries
II II VI VII III VI VI VI II I
VIII I II VII VI IX I III VI VII
Netherlands Finland Norway Belgium and Luxembourg West Germany France Denmark New Zealand Canada Sweden Source:
Same as table 2.
996 1129 1189 1191 1254 1270 1335 1502 1602 1802
VII I III VII VI IX VII IV III VII
116
N.Halevi, The
balance of payments stages” hypothesis
An indication of the results one might expect from such a study is given by table 8. The balance of payments of ten richer countries have been divided into two sections: transactions with the United States, and transactions with the rest of the world; table 8 summarizes the classification by balance of payments categories. Three countries which are classified as debtors on the basis of transactions with the U.S. alone fall into the creditor class on the basis of other transactions only, and two more move to the last debtor class. However, the original classification is not significantly altered, in most cases, by exclusion of transactions with the United States. We would expect that a detailed analysis of capital flows would show that some developed countries have offset their lending to developing countries by borrowing from other developed countries.
Conclusions The neo-classical “balance of payments stages” thesis is not borne out by examination of multi-country cross-section data: there is no discernible orderly progression through balance of payments stages connected with rising income per capita level. Departures from the predicted pattern are most evident in the debtor-creditor relationship. There is some relationship between per capita income and balance of payments deficit, but it is not strong enough to support the contention that the former is the major explanatory variable. In the modern world, even the simplified neo-classical theory of capital movements need not lead to capital movements from the more developed to the least developed countries. But the empirical examination of capital flows and the testing of theories of capital movements, including the neo-classical thesis, deserves more attention. Two further approaches suggested by the above results are time series studies for developing countries, and cross-section studies based on more detailed breakdowns of capital movements of the more developed countries.
References Adelman, I. and C.T. Morris, 1961, Society politics and economic development (Baltimore). Avramovic, D. and associates, 1964, Economic growth and external debt, IBRD (Baltimore). Bach, G.L., 1960, Economics (3rd edition) (Englewood Cliffs). Caimes, J.E., 1874, Some leading principles of political economy (New York).
N.Halevi, The “balance of payments stages” hypothesis
117
Crowther, G., 1957, Balances and imbalances of payment (Boston). Domar, E., 1950, Effect of foreign investment on the balance of payments, American Economic Review, Vol. XL (December). Enke, S. and V. Salera, 1947, International economics (New York). Hagen, E.E. and 0. Hawrylyshyn, 1969, Analysis of world income and growth 1955-1965, Economic Development and Cultural Change, Vol. XVIII, No. 1, Part II (October). Kendall, M.G., and A. Stuart, 1967, The advanced theory of statistics, 2nd edition, Vol. II (London). Kuznets, S., 1964, Quantitative aspects of the economic growth of nations: IX. Level and structure of foreign trade: Comparisons for recent years, Economic Development and Cultural Change, Vol. XIII, No. 1, Part II (October). Meier, G.M. and R.E. Baldwin, 1957, Economic development: Theory, history, policy (New York). Nurkse, R., 1958, Problems of capital formation in underdeveloped countries (Oxford). Taussig, F.W., 1928, International trade (New York).