Portal de Periódicos da CAPES

Whose trade statistics are correct? Multiple mirror comparison techniques: a test case of Cambodia

2012; Taylor & Francis; Volume: 15; Issue: 1 Linguagem: Inglês

10.1080/17487870.2012.657827

1748-7889

Shintaro Hamanaka,

World Trade Organization Law

Click to increase image sizeClick to decrease image size Notes 1. Conceptually, it is difficult to distinguish misclassification mainly committed by customs and misinvoicing mainly committed by traders. Thus, this paper consistently uses the term misclassification, but this term also includes the notion of misinvoicing. 2. The value of exports can also be manipulated (in this case through over-invoicing) as a way to misuse duty drawbacks. 3. While traders are considered the immediate party that commits errors (e.g., misreporting traded goods), we regard this as a misclassification committed on the part of the customs office since it is responsible for ensuring that data submitted by traders are accurate. 4. Exports are sometimes reported on a free alongside ship (FAS) basis, which equals FOB minus the cost of loading the ship. 5. To estimate a mirror export, the partner country's import value is divided by 1.1. The detailed estimation procedure is discussed on the IMF's Direction of Trade Statistics website at http://www2.imfstatistics.org/DOT/DOTEstim.htm. 6. Re-export takes place when goods enter a customs territory from one country and are shipped to another country without being modified. 7. Transshipment, or 'goods in transit,' includes merchandise that passes through ports but is not unloaded from the ship or aircraft. 8. The general trade system is in use when the statistical territory of a country coincides with its economic territory. The special trade system (strict definition) is in use when the statistical territory comprises only the free circulation area, that is, the area within which goods 'may be disposed of without customs restriction.' The special trade system (relaxed definition) is in use when (i) goods that enter a country for, or leave it after, inward processing and (ii) goods that enter or leave an industrial free zone are also recorded and included in international merchandise trade statistics based on the International Merchandise Trade Statistics Concepts and Definitions. United Nations. Series M, No. 52, Revision 2. See Eurostat (2009). 9. Using mirror trade data, Mahmood and Azhar (2001) hypothesized the presence of over-invoicing of exports in Pakistan due to the duty drawback incentive scheme. The study found that there is a strong presence of over-invoicing across various trade partners and products. 10. Javorcik and Narciso (2007 Javorcik, B. and Narciso, G., 2007. Differentiated products and evasion of import tariffs. World Bank Policy Research Working Paper Series, 4123. [Google Scholar]) found that the discrepancy between the value of exports reported by Germany and the value of imports reported by Germany's trade partners is positively related to the level of tariffs in eight out of ten countries surveyed. 11. So long as we compare only CIF import values and FOB export values compiled by different customs offices, separating the two (CIF–FOB factor and other factors) will remain extremely difficult. There are attempts to separate price factors from other discrepancies, using the FOB import data and CIF import data published by several customs offices (the US, Australia, New Zealand, Brazil, and Chile), because the difference between CIF and FOB import values from one customs office approximates transport and insurance costs. Yeats (1978 Yeats, A. 1978. On the accuracy of partner country trade statistics. Oxford Bulletin of Economics and Statistics, 40(4): 341–361. [Crossref], [Web of Science ®] , [Google Scholar]) attempts to separate the observed variations in FOB export–CIF import statistics (gap between mirror statistics) into freight costs and discrepancies caused by other factors (residual) using FOB import data, employing matched US and partner country trade data. He finds that the residual becomes larger and the discrepancy of the mirror becomes significant when lower digit commodity-level data are used. This implies that commodity misclassification is a serious problem in developing countries. Hummels and Lugovskyy (2006) compare two kinds of discrepancies: (i) the discrepancy between CIF-based bilateral imports and FOB-based bilateral imports, and (ii) the discrepancy between CIF-based bilateral imports collected by the importing side and FOB-based bilateral exports collected by the exporting side. They find that the discrepancy between matched partner trade statistics (a certain country's CIF-based imports and its partner's FOB-based exports) is significant and includes not only trade or transport costs but also various noise factors associated with structural factors as well as human factors. 12. For example, when Country A's imports from Country B are US$45 trillion (Country A's data) and Country B's exports to Country A are US$55 trillion (Country B's data), the average value of the aggregate bilateral trade becomes US$50 trillion. 13. 1% is used because there are roughly 100 two-digit commodity groups. While there are actually 97 groups and one group that includes goods not classified elsewhere (No. 99), 1% is used for ease of calculation. In the above example, the theoretically reasonable amount of each two-digit trade volume is US$500 million. 14. 10% is used because the discrepancy within 10% should be considered normal given the CIF–FOB factor. In the above-mentioned example, the two-digit tariff line with a discrepancy of over US$50 million is caught by this threshold.