Export Product Composition Indexes in Developing Countries: The Case of Peru, 1993–2004

Abstract

Liberal structural reforms associated to changes in the export product composition may affect economic growth, and at the same time, may yield biases in the official standard trade index numbers from developing countries and error measures in the real rate of economic and exports growth. This paper proposes a set of index numbers which incorporates the export product composition in a standard export index in such a way that changes in the current export value can be decomposed into: price changes; quantity changes, and product composition changes. In the applications of those indexes for the Peruvian case, it is found that the estimated overvaluations in the official annual average rate of growth of the real exports value and the GDP, were, respectively 3% and 0.6%, for the period 1993–2004.

ACKNOWLEDGMENTS

The study was financed by the academic research office of the university (DAI-PUCP). Cynthia Ortiz and Paolo Pinedo were the computational assistants and I owe them my gratitude. The errors as always are from the author.

Notes

This paper was written when the author was the CEPAL-UPR Celso Furtado Chair, Universidad de Puerto Rico, Rio Piedras

¹A recent survey on index prices and trade indexes is found in Diewert (2001) and Dridi and Zieschang (2004) respectively.

²The sources of these changes, among other, may be: the introduction of new products, the entry and exit of established products, market discontinuities, quality changes and seasonal products. In the index price literature ‘solutions’ of these sources varies and depend upon the type of the source (e.g., Greenless, 2000; Haussman, 2002; Schultze, 2003; Schultze and Mackie, 2002; Dridi and Zieschang, 2004; and Baldwin, Nakamura, Nakamura, 1996). The focus of the proposed index number is not on the sources but rather in the continuous change of the product composition.

³An earlier application of the same set of index numbers is provided by Tello (1997) for Nicaraguan trade data.

⁴The results in (15) are held regardless of the selected base period.

⁵An alternative family of index numbers is given by the following definition of kp and kq:

The notation is similar to the adjusted ideal index numbers proposed in this paper. Both set of index numbers satisfy practically the same properties except that the one proposed satisfy the chain property. When the number of products is equal for all the time periods, this family set of index numbers is transformed to the ideal index numbers given that kp = kq = kn = 1. Tello (1996) applies this set of index numbers to the analysis of export subsidies in Nicaragua.

⁶In the alternative set of index numbers proposed, kp = kq = 1.

⁷Note in the alternative set of index numbers, for any given product “i,” pio < pmi < pit (or pit < pmi < pio).

⁸Two well known properties, from the economic approach to index numbers theory are those of the exact and superlative properties (Diewert, 1976). Our conjecture is that the proposed set of index numbers does not satisfy these properties since it will be needed to find a production (o utility) function that takes account changes in product composition.

⁹The main traditional export products in Peru in this period were: Fishing flour; Fishing oil; Cotton, Sugar, Copper; Zinc; Gold; Refined Silver; Tin; Coffee; Iron; Lead. In average, the account for 63.2% of the total Peruvian export value.

¹⁰The large difference between the numbers of tariff lines included in the indexes computations and the total number of tariff lines are explained among other things by: i) missing data in the tariff line; ii) tariff lines with only one year of information; and iii) errors in the values.

¹¹A sample of “new” export products in Peru is provided in Table A4.

¹²The sources of the biases come on the one hand, due to omission of the changes of product export composition. On the other hand, due to that the estimated value of the rate of change of these traditional indexes, in the case of Peru, are dominated by the changes in prices and quantities of the 12 main traditional exports products that account for 63,5% of the total export value. The weighted average of the rate of change of prices and quantities of these 12 products are very similar to the respective rate of change of the official quantity and price indexes. The former are 8,9% and 3.0% respectively, whereas the official are 9.4% and 3.6%.

¹³ The average rate of change of the Peruvian real GDP growth in this period was 4.4%.