Dollar Exchange Rate and Commercial Balance
Previous e&e studies about exchange rate
In the 36th (i) issue of e&e it was shown the strong correlation between the annual dollar exchange rate, taking into consideration the American dollar inflation, the Brazilian Real (or other previous currencies) inflation, and the balance of goods and services (non-factors).
The availability of monthly values of these two variables makes it attractive their use for establishing medium term economical projections. In the previous issue and in the present one we try to apply monthly indexes in order to “update” the commercial and the parallel dollar exchange rates. For this purpose we have used the IGP-DI of the Brazilian Fundação Getúlio Vargas (FGV) and the CPI of the United States.
The IPEA (Institute of Applied Economical Research) already follows up the parity rate – R$ / US$ - consumers price – (December 1998 = 100) index “calculated by removing the American National Price Index (INPC) and the Consumers Price Index (CPI) from the real (R$) / American dollar (US$) exchange rate nominal series”.
By using here the referred exchange rate relative to a more recent date and in a larger time interval, we tried to offer a historical perspective of the past exchange rate variation within parameters that can be analyzed now. Furthermore, by trying to identify an “equilibrium” exchange rate, we are indicating to what extent the present exchange rate is above or below this reference exchange rate. Actually, it also became clear that this exchange rate depends on the results of the external accounts that the country wants (or needs) to generate.
In what follows, we will try to extract a correlation between the monthly exchange rates and the results of the commercial balance relative to the GDP. In this approach, the GDP is expressed in its real value, measured by the IBGE (Brazilian Institute of Geography and Statistics), relative to some arbitrary date. In order to exchange reais for dollar we have used the average nominal R$/US$ exchange rate of 1969 that corresponds to an “equilibrium” value of the balance of goods and services non-factor. The value of this exchange rate in currencies of January 2003 is 2.71 R$/US$. It is also close to the average value from January 1959 to January 2003, namely, 2.74 R$/US$.
The exports, imports and commercial balance result values are expressed in current dollars and converted to the reference date (in the present article, January 2003) using the monthly CPI of the United States. Therefore, we are using GDP values expressed in dollars that keep a close relation to its real variation and that are not “distorted” by the current exchange rate. It should be no surprise that the exports/GDP, imports/GDP and commercial balance result/GDP values are much different from those obtained when current dollars are used, or internal currency used in the IBGE’s National Accounts statistics.
External Trade and GDP
The values obtained for the GDP express the value of national production in a currency corrected by an “equilibrium” exchange rate (very close to the average value). In all cases, the dollar devaluation is corrected by the same index, the CPI of the United States.
The values obtained for exports, imports and commercial balance result are shown in Figure 1.
Figure 1: Exports, Imports and Commercial Balance (goods) of Brazil, expressed in US$, corrected monthly by the CPI of the United States relative to January 2003.
It can be observed that the exports values have been growing regularly. The commercial balance variations are fundamentally due to imports variations.
An analysis of the monthly data shows a clear seasonal characteristic of exports and a less clear one of imports whereas that of the commercial balance result is not easily identifiable. The monthly GDP values are estimated by IPEADATA(ii) from 1979 on. For the previous years, we have repeated the average seasonal characteristic taking into account the annual production. The same procedure was adopted for 2003 when the annual estimate is known. For the first months of 2003 a GDP growth of 1.6% is being considered.
The values considered here include data available from January 1959 on. In Figure 2 we show the GDP values relative to the current exchange rate and the GDP in terms of relative real value and of value converted to dollar using the 1969 “equilibrium” exchange rate. Both values were converted to January 2003 dollar using the CPI of the United States.
The GDP expressed in current dollar is highly influenced by the exchange rate variation while the values proportional to the GDP real value present a progressive growth without the variation of the former. Using this last set of values for calculating the behavior of the external trade variables it is natural that one finds values free from sudden exchange rate alterations, as will be seen in what follows
Figure 2: GDP in real values, expressed in dollars of January 2003, using commercial dollar of each month (red curve) or 1969 exchange rate corrected by the CPI of the USA (blue curve). 1969 is the year when the exchange rate allowed for equilibrium in the balance of goods and services. The blue curve is proportional to the real GDP, the red curve is highly influenced by the exchanged rate.
Figure 3 shows the exports, imports and commercial balance result values relative to the GDP.
Figure 3: Commercial Balance, Exports and Imports relative to the GDP. The values are proportional to their real variation when an “equilibrium” exchange rate close to the average value is adopted.
The first interesting conclusion is that the exports values are extraordinarily stable from 1973 on. In the sixties exports were about 6% of the GDP. From 1973 on they reached 9.5% of the GDP and slowly decreased to 8.5% of the GDP in the last years. The influence of the large variations that occurred in the world economy in these last thirty years is not visible at all.
None of the dramatic changes in the exchange rates that occurred in this period have significantly altered the values of our exports relative to the GDP, shown in Figure 3. When the participation of exports in the GDP is expressed in the internal currency, it follows the exchange rate (variation of the numerator). The exports/GDP ratio expressed in dollars varies as well due to alteration of the denominator. When the exports (Figure 1) and GDP (Figure 2) values are expressed in dollar, correcting the GDP by the equilibrium (or average value) exchange rate means that we are dividing two “well-behaved” functions and the quotient shows the regularity observed in Figure 3.
Variations in the internal currency mean that the part of the economy dedicated to exports is better remunerated. It can also occur variation in the quantity that would be compensated (in the opposite sense) by prices variation. The country would export more but it would not get the corresponding benefits. This is a phenomenon that surely should be analyzed.
In Figure 3 it can be seen that alterations in the commercial balance result are mainly due to variations of imports. The only important exceptions are verified in 1986 and 1987 (years of the Cruzado Plan and of the “technical moratorium”, respectively). The exports amounts were not significantly altered by Mercosul or by the various modifications of the international conjuncture in these thirty years.
Influence of the Exchange Rate On the External Trade
There are important alterations regarding imports and the Brazilian commercial balance that are correlated with the exchange rate. In a previous article (i) (Feu in e&e 36), the correlation between the commercial exchange rate and the balance of goods and services (non-factors) was studied. We are assuming in this approach that the external trade suffers the influence of both the commercial and the parallel exchange rates. The latter can indicate with some anticipation the changes of the commercial exchange rate. Furthermore, it can be supposed that part of the official transactions were made – whenever there was a significant difference between them – using the parallel exchange rate. In this case, even though the amount has no been registered, the parallel exchange rate influences the external trade. The approach here is merely statistical. A “compound” exchange rate was created that contains a fixed share of the parallel and that was compared to the adjustment obtained with the commercial exchange rate. A substantial improvement of the adjustment – higher than the one expected by the simple introduction of an additional constant of the process – would justify considering the influence of the parallel exchange rate.
Monthly deflators are available such as the IDP-DI and the CPI of the United States that permit a monthly correction of the values expressed in real or dollar. The commercial balance value is also available, as estimate, in the month that follows the calculated period. That permits the short-term projection.
The commercial exchange rate (sale at the end of the month) and the parallel one (sale at the end of the month) corrected by the American (CPI of USA) and Brazilian (IGP-DI of FGV) inflations are shown in Figure 4. The average value of the parallel exchange rate in the period was 24% higher than the commercial one. The exchange rate composition indicated (45% of the parallel one and 55% of the commercial one) corresponds to the linear adjustment of the composed exchange rate versus commercial balance. Temporal delays regarding the exchange rate relative to the commercial balance were also tested and the best adjustment obtained was 14 months. In order to prevent corrections due to seasonal factors and because this is a preliminary approach, we have considered a delay of 12 months.
Figure 4:Commercial and parallel exchange rates, corrected by the American and Brazilian inflations and the “compound” one, obtained by their weighted average (0,45 weight for the parallel and 0,55 for the commercial)
In Figure 5, the average annual values of the “compound” exchange rate of the previous year and the annual commercial balance result are shown. The adjustment permits projecting the commercial balance of the current year.
Figure 5: Adjustment of annual data (Commercial Balance as % of GDP) in the period
With a similar adjustment it is also possible to project imports and exports and the latter is independent of the exchange rate. It was imposed on the adjustment that the coefficients would be coherent with the relation Commercial Balance = Exports – Imports.. In the new adjustment the 1974/1976 period was excluded because it was strongly influenced by the first petroleum prices shock and by changes in the economical model. In Figure 6 the new adjustment is shown and the excluded points are pointed out.
The projected value for the commercial balance in the present year (2003) would be (1.4±0.8) % of the GDP or 10 ± 6 billion dollars. In the way it was carried out (one year of delay) the value of the present year would be determined by the average exchange rate of the previous year. The projected values for exports and imports would be, respectively, 57 and 47 billion dollars.
6: Adjustment for the annual data where values of the period 1974/1976 are
The same type of adjustment can be made with the monthly data. Naturally we will be dealing with larger data dispersion and consequently a smaller correlation in the adjustment.
In Figure 7 the adjustment corresponding to monthly values is shown. Two types of adjustments were considered. The first one tries to correlate the exchange rate of the month of the previous year with the monthly commercial balance relative to the GDP. The second one, corresponding to the best adjustment, tries to correlate the average exchange rate in subsequent months with the commercial balance relative to the GDP. In this preliminary calculation it was chosen the average between the 5th and 23rd months before the month whose commercial balance one wants to estimate. As in the previous adjustment the years from 1974 to 1976 were not considered.
7: Adjustments for the commercial balance and exchange rate for the period
The corresponding projections are shown in Figure 8. Taking into account that the projection along time was not adjusted and that we are trying to describe a complex phenomenon such as the commercial balance as a function of a linear adjustment with only one variable (or a fixed composition of two variables), the result is even surprising.
Figure 8: Projected and verified values of the commercial balance as a function of the monthly exchange rate. The final values correspond to projection based on linear adjustment of the exchange rate relative to commercial balance/GDP
Some oscillations verified in the commercial balance in the last months were anticipated in the projection of adjustment with the exchange rate of one month. These projections indicate a decrease of the commercial balance at the end of the present year due to the present decrease of the dollar
An interesting indication, to be used in the future improvement of the adjustment, could come from the identification and analysis of the periods when the adjustment did not produce an adequate description. Besides the periods corresponding to the effects of the 1973 petroleum prices shock, the periods close to the second petroleum prices shock in 1979 and the periods of assets retention by the Collor Plan in 1990 stand out. In what concerns the petroleum prices shocks, it was difficult for the economy to respond quickly to the considerable increase of imports values. It can be observed in Figure 3 that during the first shock it occurred a large increase in the exported GDP fraction. In the case of the second shock the response was either substitution of imports by local petroleum production or the use of surrogates (alcohol, firewood, hydraulic energy and national steam coal).
Short-term projections are always the most interesting and also the most dangerous (short-term profits have a short life). Nevertheless, we show the results in Figure 9.
A preliminary analysis points out to very high exports values relative to the expected historical behavior. Applying the seasonal factor observed in the previous years and assuming the behavior of the present exports, there would be an increase of 26% in the exported value relative to the last year totaling 75 billion US$ and import equal to that of last year (47 billion US$). Imports coincide with our projection based on the annual adjustment but exports exceed much beyond our expectation. Furthermore, the commercial balance result based on the first trimester (28 billion US$) is much above our projected value of 10±6 billion US$ (annual projection) or 13 billion US$ for the monthly projection (the two adjustments have the same annual value). We have not made monthly adjustments for exports and imports.
That is, even though the values of the last months follow qualitatively the projected monthly adjustment, they were much above the expected absolute value. This indicates that one can foresee a worse performance of the commercial balance for the rest of the year. The form of the curve suggests a decrease in the next months and some recovery at the end of the year.
Figure 9: Projections based on the monthly exchange rate and on the exchange rate in the interval of 5 to 23 previous months.
In what concerns the values of Figure 9, it should be observed that the result of the exports value in 2002 could be due to the superposition of two effects. If we examine Figure 4, we can verify that the dollar presented exchange rate peaks in October 2001 and in October 2002. It is natural that due to the high values of the exchange rate verified in October 2003 the exporters had some income anticipation. This has coincided with an exports peak whose origin would rest on the exchange rate peak of the previous year (October 2002).
In principle one can dream of a real increase in exports as it has occurred in 1973. However, the behavior in the last thirty years recommends prudence.
Is it possible to project the exchange rate?
In the (preliminary) adjustment carried out here we have found an indication that changes in the exchange rate precede those of the commercial balance. Evidently, macroeconomic reasons can turn the country into a net exporter or importer of goods and services and of capitals. In this case, the exchange rate would become a function of the needs or possibilities of the external trade or of the political determination.
It was previously seen (i) that in order to maintain stable the external liability/GDP ratio it would be necessary a surplus of 14.2 billion US$ in dollars of 2003. In our estimate it would be necessary to have in 2003 an average “compound” exchange rate of 3.8R$/US$ or3.7 (±0.6) R$/US$ of the commercial exchange rate in order to get the same surplus in 2004. An exchange rate of 3.5 R$/US$ (currency of January 2003) seems a good indication for an exchange rate policy that does not increase the already elevated external dependence. Presently, our external liability (external debt + foreign investment in the country) is already about 2/3 of the GDP. This liability produces payment obligations of interests and dividends
It remains the question of how a variable (the exchange rate) that anticipates (another one) the commercial balance result can depend on the second one. The answer can be found in the “saw-shaped” type curve of the exports. The sudden exchange rate change is generally forced by pressures of the net external debt relative to the GDP. This includes the loss of reserves and the alteration of the external interest rate. It can also be the result of political pressures or be delayed (through concession of credits) due to external support for the government.
Normally the government tries to restrain the exchange rate through several mechanisms such as imports restrictions, exports subsidies, exchange rate control, reserves sales, external loans and lately high interests rates in order to attract speculators. When tension is unbearable, a government change, a change of the Ministry of Finance and lately a change of the President of the Central Bank makes politically viable the sudden increase of the exchange rate. Of course, there is always the extreme possibility of moratorium if the necessary direction is not corrected in due time.
The historical correlation projects a commercial balance result of 10±6 billion US$ (annual) or 13 billion US$ (monthly) for the present year (2003). However, the exceptional exports performance in the 1º trimester point out to the double of this value. The monthly adjustment indicates a reduction of the pace of exports at the end of the year.
In the past the parallel exchange rate alleviated pressures on the commercial one. In many occasions it was an anticipation of the commercial exchange rate moves. A “compound” (weighted average of the commercial and the parallel) exchange rate improves the correlation used to project the commercial balance result starting from the exchange rate of the previous period.
There is a good correlation between the commercial balance result and the exchange rate. This correlation also exists for imports. The value of the Brazilian exports have been surprisingly stable in the last thirty years vis-à-vis the alterations of the exchange rate and the variations of the world economy.
An exchange rate of 3.5±0.6 R$/US$ is the one indicated to maintain stable the Brazilian external liability. The historical average value of the exchange rate is 2.7 R$/US and coincides approximately with the value verified in 1969 that would maintain (i) the flux of goods and services and, in the long term, the capital flux.
(i) Aumara Feu, in e&e No 36, “Equilibrium Exchange Rate”
(ii) http://ipeadata.gov.br/ .
 In this case, even though the amount was not registered, the parallel exchange rate influences the volume of the external trade. It is well known the practice of paying part of the exports amount abroad or sub-invoicing imports in order to reduce imports taxes or over-invoicing them for the purpose of sending currency abroad.
 Independent adjustments and determination of angular and linear coefficients and requirement of coherence of the three adjusted lines and maintaining the coefficients of the straight line corresponding to the commercial balance. Practically, we have considered the three straight lines obtained by the least square adjustment: Yi = ai x+ bi , where Y1 is the commercial balance, Y2 is exports and Y3 is imports. The coefficients a1 and b1 corresponding to the commercial balance were maintained.
Since ∆b = b1- (b2 - b3), new values of b2’ and b3 were determined so that b2’ = b2 - ∆b/2 and b3’ = b3 + ∆b/2. Therefore b1 = b2’ - b3’. An analogous procedure was adopted for the ai angular coefficients. This corresponds to reducing the degree of freedom of the adjustment and four adjustment constants are determined instead of six.
 In the first trimester it had already reached 3.8 US$ bi.
 The more dynamic sectors of the economy in the eighties and nineties were exactly that of petroleum and its substitutes. This slightly denies the idea that this potential had been exhausted in the seventies.
 The services referred to here are those of non-factors of the debt.
 Net exports of goods and services means, in the medium term, net capital delivery. By analogy net imports means net capital influx.
 It assumes parallel/commercial equal to the average historical value and inflation in Brazil relative to that of the USA as 15.6% and GDP growth of 1.5% in 2003 and 2% in 2004.
Graphic Edition/Edição Gráfica:
Tuesday, 10 May 2011.