THE SOUTH AMERICAN CRAWL: EXCHANGE RATE
VARIATION IN DEVELOPING COUNTRIES
One of the first economists to propose balance-of-payments adjustment through regular, small changes in exchange rates was James E. Meade. Meade's September 1964 proposal, which has come to be known as the "crawling peg", was intended as a means by which industrialized countries could adjust their exchange rates in an orderly manner. Although the crawl has been widely discussed, the industrialized countries have opted for fluctuating rates and wider bands as alternative approaches to adjustment. The crawl has however been implemented in a context for which it was not originally intended: in the less-developed countries. At least four countries in South America have made recourse to the crawl: Argentina (1965),Brazil (1968 - ), Chile (1965-70 and 1973- ) and Uruguay (1972- ). Colombia has since 1968 had a managed flexible rate, which has been continuously depreciating in a manner similar to the operation of the crawl in the other South American countries. Our objective in this p aper will be to evaluate the experience of Brazil, Chile, and Colombia;* the three countries of this group which have either had the crawl or managed flexibility for the longest period.
The theoretical literature dealing with the crawling peg assumes the existence of integrated capital markets and highly mobile short-term funds. The rate adjustments envisioned are as low as two per cent per year in increments of 1/26th of one per cent every week or perhaps one-sixth of one per cent per month. Advocates of the crawl believed that by adjusting its exchange rate over an extended period of time, a country could prevent itself from being overwhelmed by massive movements of speculative funds. The crawl would offer the advantage of removing the cloak of secrecy from the rate change process, thus allowing the private sector to adjust its expectations. The crawl was intended by its proponents to be an extended interlude between fixed pegs, a means of eliminating a "fundamental disequilibrium" in external payments.
While the crawling peg appears to have generated little enthusiasm in official policy circles in the North Atlantic community, it has been seized almost as a panacea by several South American countries. Chronic inflation, recurrent payments crises, and sporadic devaluations have been a virtually permanent feature of economic life since World War II in all the South American countries except Ecuador, Peru, and Venezuela. To deal with these difficulties, virtually every imaginable control over trade and payments has been employed, ranging from multiple exchange rates and exchange auctions to advance deposits on imports. Economic ills were thought to stem from the instability and slow growth of markets for exports, together with excessive dependence on single export commodities (Brazilian and Colombian coffee and Chilean copper). An inward-looking policy of import substitution was pursued by combining tax incentives and prohibitive tariffs on consumer goods with overvalued exchange rates t o favour imported investment goods. By the mid-1960's, many observers spoke of import substitution as having reached an impasse. High-cost, low-output industries had been established behind tariff walls. Not only were the new industries incapable of capturing foreign markets because of their cost disadvantages, but some traditional exports declined as a consequence of the adverse effects of exchange rate overvaluation. The time was ripe in South America for a new approach and the crawling peg seemed to be an appropriate cure-all. Chile was the first to introduce the crawling peg in April 1965, barely seven months after Meade's seminal article and before that proposal had even been debated in the literature. Colombia and Brazil followed suit not long after in 1968.
In the South American context the crawl is not seen as a short-term transitional device, but rather as a permanent scheme to be abandoned only in the event of altered circumstances or of substantial negative feedback. At the time that Chile abandoned the crawl in August 1970, it did so as a political expedient: presidential elections were less than one month away and the opposition was highly critical of what it regarded as the scandalous devaluations. Chilean Central Bank officials regarded the experience with what they termed a "programmed rate" to have been highly successful. Colombia and Brazil, on the other hand, are still using the crawl, after eight and seven years, respectively.
The benefits of the crawl to less developed countries are alleged to be far more numerous than those which might be gained from it by an industrialized country. In either case, the crawl would tend to minimize speculative capital movements. In the case of a less developed country (LDC) capital outflows would be triggered by the anticipation of a large devaluation, rather than by slight interest rate differentials. The crawl represents the next stage in the import substitution process, a stage in which a push is being made to export non-traditional products, especially manufactured items. Adoption of the crawl marks the abandonment of the turbulent seas of overvaluation and deficits for the calmer waters of undervaluation and surpluses. For the policy-makers, the crawl has permitted a shift of focus from almost daily external crises to the achievement of internal goals. For the businessman, the crawl has eliminated major uncertainties surrounding the profitability of new export ventures. In Chile, the crawl was seen as enabling the country to reduce its dependence on external indebtedness, while in Brazil it was seen as a device for attracting foreign funds (a goal in which Brazil had unparalleled success). In producing a surplus in the payments' balance, the crawl provides countries with the opportunity to reduce or eliminate a portion of the array of controls over foreign transactions which had accumulated over the years on an ad hoc basis.
The few studies which deal directly or indirectly with the use of the crawl in Brazil, Chile, and Colombia have pronounced it an unqualified success. That success is measured by changes in reserves, changes in capital flows, increases in manufactured exports, higher growth rates of income, lower rates of inflation, and abolition of controls. Our purpose is to question whether too much of a good thing is not without costs, albeit costs of an unfamiliar nature in South America. On the whole we agree with Gottfried Haberler's observation that "in a country with a highly inflationary economy, a fixed exchange rate cannot be maintained for any length of time without intolerable consequences." The central question is the manner in which the crawling peg is managed. When should the crawl be slowed in rate and/or frequency? What are the costs to a developing country of prolonged currency undervaluation? Does acceptance of the crawl lead one to moderate efforts to reduce double-digit inflation? What are the effects of undervaluation on income distribution and on ownership patterns? These are but some of the, questions which must be answered in order to evaluate the South American experience with the crawl and to guide its future use.
In the discussion which follows my consideration will largely be limited to the behavior of foreign exchange reserves and merchandise trade. I hope that I shall be able on some future occasion to share with you further results of this continuing inquiry.
Foreign Exchange Reserves
The ability of an exchange authority to defend a fixed rate hinges largely on its holdings of exchange reserves. The question of optimum reserve levels or ranges has generated an extensive literature. However, in the absence of econometric studies of reserve behavior in the three countries, I shall rely on old-fashioned rule-of-thumb notions relating reserves to imports. Let us hypothesize that an acceptable range for reserves (R) in relation to imports (M) runs from 0.25 < R/M < 0.40. That is, if reserves are between the value of 90 and 144 days supply of imports, a Central Bank will probably not feel it necessary to either add to or remove existing payments' restrictions. (Between 1961 and 1970, the value of R/M for the non-Communist world fell from about 0.5 to 0.3, while the Latin American average varied from about 0.29 to 0.35). As Table 2 indicates, for all three countries R/M was well below the 0.25 lower limit in the year preceding the introduction of the crawl and at the end of the year in which the crawl commenced. For all three countries R/M grew to exceed 0.40 and all three countries increased their share of total Latin American and World reserves. Brazil's experience has clearly been the most spectacular: for 1973, R/MR20.936 and an increase in its share of rapidly increasing World reserves from 0.27 per cent in 1967 to 3.5 per cent at the end of 1073. It is our contention that the reserve buildup in the Brazilian case is clear evidence of marked undervaluation which has not been adequately offset by liberalization of trade and payments. Moreover, the holding of substantial idle reserves represents growth forgone, in that a portion could have been used for imported capital goods for development projects.
Another cost of reserve accumulation is in the loss of control over expansion of the money supply. Unless the monetary expansion accompanying reserve buildup is sterilized through the sale of debt instruments or by other policy offsets, prices may escape from the at best tenuous grasp of the planners. Large capital inflows led the Brazilians in late 1972 to require that banks provide a twenty-five per cent interest-free deposit with the Central Bank against new foreign borrowings in order to limit monetary expansion.
Crawling: How Often and By How Much
The frequency of devaluation in the three countries has ranged from almost daily in Colombia to as few as five times in 1973 in Brazil. There appears to be little discussion of how often to devalue, although there is a consensus that it should be often enough both to discourage speculative capital outflows and to avoid dampening the incentives to exporters.
Consideration of the magnitude of rate adjustments, on the other hand, has generated alternative formulations. Jay Levin has suggested that the pace of the crawl should be paired with target changes in exchange reserves, an approach which appears to conform to Colombian rhetoric. The Colombian Central Bank has spoken of managing the peso's float to the extent necessary to achieve the goal of sustained growth of exchange earnings. While this broad goal has been translated into explicit exchange reserve targets, the enunciation of policy guidelines and adherence to them are different matters. The Colombian peso was devalued by approximately the same percentage in 1972 as in 1971, although reserve targets were exceeded in every quarter of 1972. At the end of 1972 Colombian exchange reserves exceeded the stated desired level by at least 35 per cent.
In Chile, the rate was alleged to have been regulated in accordance with projections of the medium term current account balance adjusted for changes in domestic savings and foreign indebtedness. These projections could have been at best only tentative given both the importance of copper in Chile's exports and the volatility of copper prices. Optimistic projections made on the basis of high copper prices in the first half of 1966 had to be drastically revised as prices tumbled in August 1966. One may well ask whether these projections of the value of external transactions, which undoubtedly formed an important input in Chilean economic planning, actually influenced in any appreciable way the pace of the crawl of the Chilean escudo, or whether internal and external price movements may not have been more the focus of attention.
Rhetoric to the contrary notwithstanding, the purchasing power parity principle for exchange rate determination appears to have been given new life in Brazil and Chile. References are made in each case to the "real exchange rate", which is the nominal rate adjusted both for changes in domestic prices and prices in the United States, an important trading partner of all three countries. It would appear that the intent of policy has been to limit the variation in the real exchange rate or, to put it another way, to devalue often enough and by a large enough degree to prevent external and internal costs and prices from diverging relative to some base period. The present military government in Chile, for example, is attempting with its crawl to reestablish parity based on the 1970 relationship between international and domestic prices.
As with the purchasing power parity concept, the exchange rate is to be determined in relation to prices rather than to underlying changes in the balance of payments. The presumption is that the chosen base period nominal exchange rate was an "equilibrium" rate and that the adjustment of the nominal rates of succeeding periods for domestic and foreign price changes will be sufficient to maintain the balance of trade and/or payments in eqxilibrium. The Brazilian experience illustrates well the divergence which may arise between price considerations and the behavior of the payments balance: Brazil's real exchange rate was virtually unchanged in 1973 relative to 1970, yet reserves experienced a 5.4 fold increase.
As with purchasing power parity, there is the seemingly insoluble problem of choosing appropriate price indices. There is as well the question of the appropriate foreign currency to use in determining the real exchange rate. Latin America continues to be part of the dollar area and all three countries have for the periods we are examining continuously devalued their currencies relative to the dollar, with the single exception of a three per cent appreciation of the Brazilian cruzeiro relative to the dollar on February 14, 1973. With respect to the German mark and the Japanese yen, that adjustment of the cruzeiro represented a devaluation of eight per cent. Should dollar comparisons serve as the sole guide to calculations of the real exchange rate given that in 1969 the United States purchased 39 per cent of Colombia's exports and only 26 per cent of Brazil's? The more rapid crawl of the cruzeiro relative to the European currencies has been associated with a marked shift of Brazil's expo rts to the countries of Industrial Europe: Brazil's exports to Europe in 1973 were almost double in value those to the United States; four years earlier in 1969 the difference had only been 37 per cent.
The crawling peg is seen as a device for increasing exports by changing the profit expectations of exporters. The usual patterns of sporadic devaluation in the context of chronic inflation offers at best only temporary prospects of increased export profitability as generally brief periods of exchange rate undervaluation are followed by extended periods of overvaluation. As the three countries under consideration are (or have been) principally exporters of a single commodity, the crawl was intended, along with other policies, to stimulate diversification of exports. Through export promotion and diversification these countries hoped to extract themselves from the supposed dead-end to which import substitution had led them. It has been argued that high cost domestic manufactured goods could only compete in export markets if a country's exchange rate were to be systematically undervalued.
All three countries established programs to foster nontraditional exports by providing incentives and assistance beyond the assured profitability provided by either an undervalued exchange rate and/or a constant real exchange rate. In Brazil and Colombia tax incentives have been combined with an undervalued rate, thus providing export firms a competitive cost advantage in both home and foreign markets. In Colombia firms were able until 1967 to deduct notional profits of forty per cent of the value of non-traditional exports from their global profits, thus reducing the taxes on the domestic operations of export firms, an arrangement which even attracted some commercial establishments to involve themselves in export activity. The level of notional profits was reduced to 15 per cent in 1967 and then in 1974, following a change in elected governments, to a maximum of five per cent of the Value of exports net of the cost of imported raw materials.
A similar arrangement existed in Brazil: firms which export could deduct expenses associated with exports from the taxes on their domestic operations. The implications of this arrangement for the domestic market have been delineated by the Brazilian Finance Minister, Antonio Delfim. Neto, who is
reported to have said, "If Volkswagen of Brazil exports cars, Ford must export or die."
Chile had the least developed export promotion schemes and the least success; Brazil had the most elaborate schemes and the most positive results. Chile's efforts seem to have been limited principally to the granting of drawbacks and had achieved limited overall impact by 1970 when the crawl was abandoned: the dollar value of industrial exports was no higher in 1970 than it had been in 1966.
The following figures for each of the three countries which show the principal exports (Xi) as a percentage of total exports (XT) provide one illustration of the relative success of Brazil:
Year 1967 1973
Brazilian coffee 41.1 20.1
Colombian coffee 62.9 55.0
Year 1965 1970
Chilean copper 73.1 73.0
Coffee's decline in Brazil has been relative, not absolute; indeed, the value of coffee exports attained a record high in 1973. Manufactured exports have grown vertiginously from $50 million in 1964 to almost $2 billion ($1.941 billion) or 31.3 per cent of Brazil's exports in 1973. Moreover, Brazil's manufactured exports in 1973 actually exceeded the total dollar value of all Brazilian exports just five years earlier in 1968, while the 1968 export level marked the end of a period of export stagnation extending back to 1951. Brazil's export drive has been so successful that when increased world prices in late 1972 and early 1973 provided yet an additional incentive to exporters, quantitative restrictions had to be imposed on the export of a number of raw materials to assure the flow of supplies to the domestic market. Among these commodities whose export was restricted were soya beans and oil, groundnuts, maize, rice, meat, cotton, timber, and iron ore.
Production and investment in all three countries are intimately tied to imported inputs; if imports fall so does the expansion of production, investment, and employment. The key function of both the crawling peg and of export expansion is clearly to facilitate the acquisition of necessary imports. Prior to the introduction of the crawling peg all three countries experienced frequent falls in import value during the recurrent cycles of overvaluation followed by exchange crisis-followed by devaluation. As long as currency devaluation proceeds at a relatively moderate rate as under the crawling peg, cet.par., import levels do not appear to be adversely effected. When, however, devaluation occurs in large steps, the impact on financial requirements is such as to depress imports, occasioning either cutbacks in production and/or postponement of new investment.
As Brazil's exports stagnated so did its imports. The peak dollar value of imports (cif) reached in 1951 was not surpassed until 1968 and in 1965 imports fell to 55 per cent of the current value of 1951 imports. The massive devaluation of 1953 was accompanied by a 33.5 per cent fall in import value, while those of 1964 and 1965 were accompanied by falls in import value of 15.1 and 13.2 per cent respectively. It seems evident that at least part of the "Brazilian miracle" is attributable to increased utilization of capacity made possible by the increased expenditures on imports, which have climbed from $2.1 billion in 1968 to $7.0 billion in 1973.
The Colombian experience is somewhat similar, with exports and imports peaking in 1954 and imports not attaining the same value again until 1966 (exports surpassed the previous peak only in 1970). Massive devaluation in Colombia in 1957, 1958, 1965 and 1967 were all accompanied by substantial falls in imports, of which the declines in 1957 and 1967 were both over 26 per cent. The Colombians give credit to the increase in imports of raw materials, intermediate goods, and capital goods for the reported reduction of fifty per cent in unemployment between 1967 and 1971.
William Tyler has estimated that the production of Brazil's exports of manufactured goods directly or indirectly accounts for only 2.3 per cent of Brazil's total industrial employment, which leads him to conclude that manufactured exports are not likely to be the solution to Brazil's employment problem. More important than the employment associated with the export of manufactures is the employment that is facilitated by the increased imports which are now made possible by increments to exports earnings.
My conclusions are both tentative and impressionistic. Some flow from the preceding discussion, others from my continuing work in this area. As with any exchange rate system the crawling peg exists in a total framework of economic policies. It cannot create or perform miracles (such as the Brazilian) unless the prerequisites are present and complementary policies in other areas are undertaken. It may under the proper conditions help break an export bottleneck.
In the cases which we have considered one senses a complete about face in economic policy: from import substitution at any cost to export promotion at any cost: from dogged defense of overvalued exchange rates to .continuous devaluation as a near-virtuous exercise. The appeal of the crawling peg is in creating the illusion that external policy may be operated automatically, thus freeing policymakers to focus on problems of internal balance. It is an approach which has found favor with reformist governments in Chile in the 1960's and Colombia and with rightist authoritarian governments in Brazil and in present day Chile. In the Brazilian case the stability brought by a strong, repressive military government would no doubt have been sufficient to increase profit expectations; the crawling peg, export tax incentives, and indexation (monetary correction) represent a rather considerable frosting on the cake.
The crawling peg appears to have helped each of these countries in bringing inflation down to rates which in their respective contexts are regarded as moderate. The likelihood remains that certain rates of inflation will be treated as a floor, as an immovable minimum. If constant rates of devaluation are pursued, the feedback of external prices on internal prices will certainly help maintain domestic inflation rates at levels higher than might otherwise obtain. However, the monetary impact of expanding reserves will quickly dispel the illusion of the automatic nature of exchange policy, as it did in Brazil in 1973 and in Colombia in 1972 and 1973.
In South America the crawling peg has not been a corrective measure of short duration; but then perhaps a lengthier period of "correction" has been needed to reduce the distortion created by decades of crisis policies. If pursued beyond a certain point, however, the crawling peg offers the prospect of a new round of distortions insofar as it transfers resources away from domestic activities into the export sector. The necessity for Brazil to limit the rapid growth of certain exports is indicative of the kind of distortions which are arising.
For the South American policy-maker the absence of almost daily crises arising from balance of payments deficits must indeed be a heady experience, so that one should perhaps be solicitous of some of the excesses that occur. The experience with the crawling peg is still a very new one and merits our continued scrutiny for the lessons which it has to offer.
TABLE 1: BRAZIL, CHILE, & COLOMBIA, EXCHANGE RATES, 1965-73
Exchange rate Change No. of Avg. Avg. No.
Year r0 r1 9;dr/r0 changes chng. o f days
hc/US$ % No. % Days
1968a 3.650 3.830 4.9
1969 3.830 4.350 13.6 8 1.7 46
1970 4.350 4.950 13.8 8 1.7 46
1971 4.950 5.635 13.8 9 1.5 41
1972 5.635 6.215 10.3 8 1.3 46
1973 6.215 6.220 0.1c 5 1.2c 73
a Crawl introduced on August 27
b As an annual rate: 14.3%
cOne 3.0 per cent appreciation (as an offset to US dollar depreciation) and four devaluations; average of latter, 0.78 per cent.
1965a 3.256 3.550 9.0b 7 1.3 35
1966 3.550 4.370 23.0 12 1.9 31
1967 4.370 5.790 32.5 16 2.0 23
1968 5 .790 7.650 32.1 24 1.3 16
1969 7.650 9.960 30.2 20 1.5 19
1970c 9.960 12.210 22.6d SUP> 12 1.9 18
a Crawl introduced on May 5
b As an annual rate: 13.5%
c Last adjustment on July 31
d As an annual rate: 38.9%
COLOMBIA Average monthly change
1968 16.32a 16.95 3.9b 0.78
1969 16.95 17.93 5.8 0.48
1970 17.93 19.17 6.9 0.58
1971 19.17 21.00 9.5 0.79
1972 21.00 22.88 9.0 0.75
1973 22.88 24.89 8.8 0.73
a Float commenced in July 1968
b As an annual rate: 9.4%
Sources: Bank of London and South America, Review, various issues. IMF, Annual Report on Exchange Restrictions, various issues. Dolores Libano, "El tipo de cambio en Chile desde 1948," Banco Central de Chile, Boletin Mensual, XLIII (June 1970), 743-46.
TABLE 2: BRAZIL, CHILE, & COLOMBIA, EXCHANGE RESERVES &
BRAZIL CHILE COLOMBIA
Imports Imports Impor ts
Reserves (cif) Reserves (cif) Reserves& #9;(cif)
Year R M R/M R M R/M R M R/M
million US$ % million US$ % million US$ %
1964 88.7 607.2 14.6a
1965 137.5 603.7 22.8b
1966 172.1 775.3 22.8
1967 199 1667 11.9a 126.4 722 .4 17.5 83 496.9 16.7a
1968 257 2132 12.1b 208.4 742 .7 28.1 173 643.3 26.9b
1969 656 2265 29.0 343.5 907.1 37.9 9;221 686.0 32.2
1970 1187 2849 41.7 388.5 930.8 41.7< SUP>c 206 754.6 27.3
1971 1746 3701 47.2 221.2 980.0 22.6< SUP>d 203 857.5 23.7
1972 4183 4783 87.5 325 836.5 38.9
1973 6417 6855 93.6 534 875.6 61.0
Notes: a Year preceding introduction of crawling peg.
b Year in which crawl introduced
c Year in which crawl discontinued
d Year following discontinuation of crawl.
Source: IMF, International Financial Statistics, various issues.
TABLE 3: BRAZIL, CHILE, COLOMBIA, CHANGES IN EXCHANGE RATE, MONEY SUPPLY, AND COST OF LIVING, 1965-1973
Year Exchange rate Money supply Cost of living
1968 41.1 42.3 21.8
1969 13.6 32.8 22.4
1970 13.8 26.7 22.0
1971 13.8 31.3 20.0
1972 10.3 41.4 16.7
1973 0.1 40.6 12.9
1965 28.5 65.4 28.8
1966 23.0 38.9 22.9
1967 32.5 24.9 18.2
1968 32.1 38.3 26.4
1969 30.2 35.4 30.7
1970 22.6 65.6 32.6
1968 7.1 15.9 5.9
1969 5.8 22.3 10.1
1970 6.9 15.5 6.8
1971 9.5 11.9 9.0
1972 9.0 27.1 14.3
1973 8.8 30.7 22.8
Source: IMF, International Financial Statistics, various issued.