Manufacture--manufacture terms of trade deterioration: A reply

WorldDevelopment, Vol. 21, No. 10, pp. 1617-1620, Printed in Great Britain. 0305-750X/93 $6.00 + 0.00 0 1993 Pergamon Press Ltd 1993. Manufacture-Manufacture Deterioration: Terms of Trade A Reply PRABIRJIT SARKAR Centre for Studies in Social Sciences, Calcutta and H. W. SINGER Institute of Development Studies, University of Sussex log Y = a, + b,d The essence of the Sarkar-Singer (S-S) article was that increasing reliance on manufactured exports does not offer the developing countries (DCs) any escape from unequal exchange with the industrially developed countries (ICs). Bleaney and Athukorala challenge the empirical basis of our article. Bleaney seeks to undermine the S-S results by pointing out that the endpoint of the period of the study is marked by the debt crisis and real devaluation of the currencies of the DCs. The merit of this point has been judged by undertaking a dummy variable analysis. The period of our earlier study was 1970437. Now it can be updated to include 1988 and 1989. The debt crisis erupted in the middle of 1982 when Mexico announced its default. Hence the period of our study has been divided into two subperiods, 1970-82 and 1983-89. Consider two separate regressions for the two subperiods: The unit value of manufactured exports of the DC in relation to that of the IC declined at an annual average rate of 1% over the whole period, 1970-89, irrespective of whether the unit values are measured in terms of US$ or SDR. The same trend rate of decline applies to 1970-82 (see the estimates of b, in Table 1). While the trend decline however, in the series of unit value ratio measured in terms of US$ shows a statistically significant acceleration during 1983-89, the trend decline in the series measured in SDR shows no (statistically significant) sign of acceleration or deceleration during the same period (see the estimate of a2 a, and b, - b,). Perhaps this discrepancy is the result of a sharp rise in the exchange rate of US dollar in the middle of the 1980s. Hence it does not appear that our result was due simply to our choice of an endpoint marked by the debt crisis,“export desperation” and real devaluation of the currencies of the DCs. It is, however, true that I. log Y = a, + b,4; the negative trends in both the series become more II. log Y = a2+ bft pronounced if we include 1983-89 in our period of where Y = terms of trade index, r = time variable study (compare the estimates of simple and multiple (taken as natural numbers 0, 1,2 .). Using intercept regressions). and slope dummies, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA D and D,r (respectively), these Athukorala has raised a number of important two simple regressions can be combined into a single points. First, he mentions the limitations of unit value multiple regression: indices as proxy measure of the “true” price changes of manufactured goods. In particular, he has argued log Y = a, + b, t + (a2 - a,) D + (b, - b,)Dt that changes in product-mix can generate spurious where D = 0 for 1970-82 and D = 1 for 1983-89 and price movements. As he himself admits, however, it accordingly D.t varies. does not follow automatically that this limitation of This multiple regression has been fitted to two unit value indices will create a systematic bias in any series of the terms of trade index (the ratio between the particular direction. unit values of manufactured exports of the DC and IC Second, he points out the fact that the bulk (80%) groups) - one in terms of US$ and the other in terms of manufactured exports of the KS constitutes intraof SDR. Besides this multiple regression, a simple IC trade whereas only a small part (25%) of the manuregression has been fitted to the two series for the sake factured exports of the DCs constitutes intra-DC of comparison: trade. Hence in the Sarkar-Singer terms of trade esti- 1617 1618 WORLD DEVELOPMENT Table I. Trends and structurcrl shifts in the unit values ufmanufuctured Eq. Type* 01 b, Q2--n1 bz-6, exports ofthe DCc vis-d-vis the IC.5, 1970-89 R’ F D-W A. Unit values measured in US$ 0.52 21.30 I zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 4.781 -0.01 t 0.59 9.99 II 4.7lf -O.Olj 0.39$ -0.03$ 1so 2.06 B. Unit values measured in SDR 1 4.72t -0.01 t II 4.711 -O.Ol$ 1.89 2.18 0.27 -0.02 0.46 0.46 17.08 6.40 *The fitted equations are: I. log Y = a,+ b, t where Y is the terms of trade index and t is the time variable. II. log Y = a, + b, t + (a>- b,) D + (b2 - b,) D t where the intercent dummv D = 0 for 1970-82 and = 1 for the rest of the period; accordingly iSignificantly different frim zero at I’% level. D,t = 0 and t. $Significantly different from zero at 5% level. mate, the numerator (unit value of manufactured exports of the DC) is likely to be a closer approximation of the unit value of exports from the DC to the IC whereas the denominator is likely to reflect the unit value of intra-IC exports of manufactures rather than the unit values of manufactured exports from the IC to DC. As the commodity composition of IC exports to the DC is likely to be significantly different from that of intra-IC exports, he argues that relative price movements faced by the DC-buyers of IC exports are different from those faced by IC-buyers. The question is - what will be the direction of bias? Based on a study by Lipsey. Molineri and Kravis (1990), Athukorala believes that our series has a bias in favor of our main thesis. But he has not explained why this should be so. It is accepted that commodity composition of intraIC trade is different from IC-to-DC trade. But it is not intuitively clear that the difference in commodity composition will generate different price trends which favor our result. In fact, one can make a case for an opposite bias. Rapid technological progress in the IC leading to an increasing technological gap between IC and DC coupled with the fact that the consumption pattern of the DC is coming closer to that of the IC due to the international demonstration effect leads to a rising monopoly power of the IC-manufacturers in the DC market.’ The product cycle pattern of division of labor between the IC and DC and technology transfer from IC to DC which follows a time path of obsolescence in the IC, bilateral tied aid etc. - all these factors add to this monopoly power. Hence the denominator of our terms of trade series is likely to underestimate the relative upward movements of the unit values of manufactured exports from the IC to DC. Thus the bias may be against our result. Third, Athukorala questions the UN definition of manufactures which covers all the commodities in SITC sections 5-8. This definition includes nonferrous metal (NFM) products (SITC 68), covering all mineral products, “mostly unprocessed,” such as tin, copper, zinc, lead and aluminium. Therefore, following Balassa (1990), Athukorala has removed the NFM prices from the UN manufacturing unit value indices for the DC but not from those of the IC. As a result, he questions the negative trends observed in our article. To illustrate the sensitivity of the results to the price movements in NFM exports, Athukorala compares our trend estimates (TR) for individual countries with the shares of SITC 68 exports in their total manufactured exports (NFMS). We have carried this analysis further forward and fitted a simple regression between the NFMS (given in Athukorala, 1993, Table 2, column 3, ibid.) and the TRs (given in Table 2, column 1 or column 2): TR = a + b (NFMS). The results are reported in Table 2. It is evident that for the DCs considered here the crosscountry variations in the rate of change in the terms of trade in manufactures of the individual DCs cannot be explained by crosscountry variations in the share of nonferrous metals in the total manufactured exports, NFMS. A similar conclusion can be drawn if the rates of changes in the terms of trade of the individual DCs vis-c-vis the United States, TR (b), are considered. Moreover, this conclusion will not change if the statistically insignificant rates of changes are ignored in each case. 1619 A REPLY Table 2. Regression analysis of the trend rates of changes in the terms zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPON of trade of the developing countries, 1965-85: A crosscountq study* Dependent variable I. a RZ F D-W Rate of changes in the terms of trade of the DCs zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONM vis-ir-vis the rest of the world, TR (a) A. Whole of the sample (n = 29) -0.27 B. Subset of the sample (n = 14)t 0.24 II. b -0.02 0.00 I .03 2.09 -0.03 0.10 1.26 2.61 Rate of changes in the terms of trade of the DCs vis-d-vis the USA, TR (b) A. Whole of the sample (n = 29) -2.92$ B. Subset of the sample (n = 7)1 4.84$ -0.00 -0.04 0.00 2.14 -0.00 0.00 0.02 1.86 *The fitted regression is: TR = a + b (NFMS) where TR is the trend rate of changes in the terms of trade of the individual developing contries (DCs) and NFMS is the share of non-ferrous metals (SITC 68) in the total manufactured exports (SITC 5-8) of the DCs. tOnly those countries are considered which experienced statistically significant trend rate of changes (+ or -). *Significant at I% level. There remains the question: what is the definition of processing? Does smelting represent processing or not? Here one can quote Teitel (1989), pp. 3 16, 337-338): It seems quite arbitrary that when metalworking processes are applied to iron and steel inputs the results should be considered manufactured products, but when similar processes are effected on non-ferrous metal inputs they should not. The non-ferrous metals case is, in principle, similar to iron and steel and its products; that is to say, various metalworking operations are involved to produce sheet and other flat products, bars and other shapes, tubes and pipes, wires and so on. Obviously these metal products constitute manufactured products, equivalent, from the point of view of the industrial operations involved, to similar iron and steel products. Fourth, Athukorala finds that the country-level results do not seem to support the results of the aggregative study. So he jumps to the conclusion that country- level estimates are “more appropriate on grounds of lower aggregation bias.” How does he know that the disaggregated results are more appropriate? The aggregation bias (if any) may go in either direction. We note that a more recent study by Liicke (1993) which confirms our results, is in fact based on country-level rather than aggregative analysis. Next, Athukorala questions our use of the UN productivity index for the total manufacturing sector of the DC as an indicator of labor productivity in the field of their export-oriented manufacturing sector. Our basic objective was to get a rough idea of the behavior of the factorial terms of trade, given the observed behavior of the barter terms of trade. Due to lack of readily available appropriate data, we used a proxy. Our conclusion that the factorial terms of trade have a tendency to turn against the DC does not rest “crucially on the implicit assumption that the productivity gap . . between export-oriented and import-substituting industries [of the DC] remained unchanged over the sample period.” Our crucial assumption is that the differences in the rate of growth of labor productivity in the total manufacturing sectors of the DC and the IC also indicate the actual difference between the rates of growth of labor productivity in the export-oriented manufacturing sectors of the two regions. It was beyond the scope of our article to examine whether export-orientation improves factor productivity and “low factor productivity and policy emphasis on ‘forced’ import-substitution go hand in hand.” Nevertheless we are provoked to point out that exportorientation is not always a “natural” phenomenon (governed by “free” market forces) - it is often “forced” on the DCs by domestic and/or international policy makers. Different forms of export subsidy, real devaluation of currencies etc. are often used for export promotion. Just as import-substituting output expansion may be “limited by the growth of the domestic market.“, so output expansion in the export-oriented sector of the DC may also be limited by the protectionism, anti-dumping laws and increased regionalism in IC markets. Hence it is a conjectural, though currently fashionable, argument that the rate of growth of productivity in the total manufacturing sector of the DC, and specifically in their import-substituting sector, is less than that in their export-oriented sector. Finally Athukorala alleges that “surprisingly,” on the pretext that it is “beyond the scope of the paper,” we “placed little emphasis” on our finding that the I620 WORLD DEVELOPMENT income terms of trade of the DC for their manufactured exports have shown a strong upward movement. This is incorrect. What we say is “beyond the scope of this paper” is not the income terms of trade but rather that “the increased capacity to import has served to finance debt payments rather than imports.” In other words. we say that dealing with the debt problem is beyond the scope of our paper.? We can hardly be criticized for that. To sum up, we maintain that the terms of trade of the South deteriorated not only in their exchange of primary products for Northern manufactures but also in their exchange of manufactures for Northern manufacturers.’ NOTES 1. See Bloch and Sapsford (1990) for empirical evi- us elsewhere, see Sarkar (1991) and Sarkar and dence. Singer (1992). 2. The debt problem of the DC has been discussed by 3. For a theoretical explanation of the Sarkar-Singer result, see Sarkar (1992). REFERENCES Athukorala. P., “Manufactured exports from developing manufactures, countries and their tertis of trade: A re-examination of Sarkar-Singer Results,” WorldDevelopment Vol. 21, No. IO (this issue). Balassa, Bela. “Trends in developing country exports, 1963-88” Paper presented at the Conference of the Atlantic Economic Association. (Williamsburg: October I l-13, 19YO). Bleaney, Michael F.. “Manufactured exports of developing countries and their terms of trade since 1965 ~ A Comment.” World De~~elopment Vol. 21, No. 10 (this issue). Bloch, H., and D. Sapsford, “Real commodity prices: Imperfect competition. relative wage growth and the Preb&h-Singer Terms of Trade Hypothesis” (Mimeograph: University of Tasmania, 1990). Lipsey, Robert E., Linda Molineri and Irvin B. Kravis. “Measurement of prices and price competitiveness in international trade in manufactured goods” Working Paper No. 3442 (Cambridge, MA: National Bureau of Economic Research, 1990). Liicke, Matthias, ” Developing countries’ terms of trade in Development 1967-87: A note,” Jourr~ctl of Studies, Vol. 29, No. 3 (April 1993). Sarkar, Prabirjit, “Debt crisis of the less developed countries and the transfer debate once again,” Journal elf Development Studies, Vol. 27, No. 4 (July 1991). pp. 84-101. Sarkar, Prabirjit, “The political economy of terms of trade between the North and the South,” in Helena Lindholm (Ed.), Approaches to the Study ofInternational Political Economy (Ciothenberg: PADRIGU, 1992). Sarkar, Prabirjit, and Hans W. Singer, “Manufactured exports of developing countries and their terms of trade since 1965,” World Development, Vol. 19, No. 4 (April 1991), pp. 333-340. Sarkar, Prabirjit, and Hans W. Singer, “Debt crisis, commodity prices, transfer burden and debt relief,” Dixussinn Paper (Brighton: IDS, University of Sussex, February, 1992). Teitel, Simon, “Industrialisation, primary commodities and exports of manufactures.” in Nurul Islam (Ed.). The Balance between Industry and Agriculture Dwelopmeru (London: Macmillan, 1989). in Economic