Marginal Intra-industry Trade: towards a Measure of Non-disruptive Trade Expansion

Economists conventionally assume that intraindustry trade (IIT) entails relatively smoothfactor-market adjustment to trade liberalisation. However, the appropriate definition of IIT inthe adjustment context continues to be a matter of debate. A consensus is emerging that, in thecontext of adjustment, one should use measures that are based on marginal IIT, and a range ofmeasures have been developed. These measures capture the degree of symmetry of changes inexports and imports at the sector level. In this paper I give a critical overview of thesemeasures. 1 Lovely and Nelson (2000) have explored changes in IIT in an Ethier (1982) trade model. In that model, tradeliberalisation can yield both changes in countries’ relative specialisation and changes in the size of industries atthe world level. In the conventional understanding of the SAH, the latter effect is subsumed into the ceterisparibus assumption, but the Lovely-Nelson (2000) analysis highlights the importance of controlling for world -wide structural change be it induced by trade liberalisation, technology or taste changes – in empirical analysesof the relationship between (M)IIT and factor-market adjustment.2 In the list of five similarity criteria used by the experts in charge of the third revision of the SITC code, forinstance, the first principle was “the nature of the merchandise and the materials used in its production”, while“the uses of the product” only ranked third (United Nations, 1986, p. viii). Evidence in favour of reasonablehomogeneity of statistical sectors in terms of factor requirements has been found by Elliott, Greenaway and Hine(2000). Some researchers, including Aquino (1978), Balassa (1985), Balassa and Bauwens (1987) andChristodoulou (1992), have re-arranged trade data into groups that would seem more appropriate in the IITcontext.3 A note on terminology. I refer to the GL index as a “static” measure, and to MIIT as a “dynamic” concept. TheGL index is calculated on the basis of cross-border flows of goods and is thus not a static measure in the strictestsense. Yet, “static” IIT in the sense of the GL index contrasts with “dynamic” measures of MIIT since the latterrelate to the change in these flows between two different years.4 Industry subscripts are implied. This will also be the case for all subsequent equations, unless stated otherwise.5 This graphical representation is originally due to Shelburne (1993) and has been developed as the “trade box”by Azhar, Milner and Elliott (1998).If the (constant) slope of a ray is defined as S = Mi / Xi, then the GL index on any point along a particular ray isgiven by:{ }()SSGL+−−=1/11.7 These considerations are supported by empirical evidence in Little (1996, p. 16), who observed that “regionswith rising IIT tended to experience a relatively large shift in the composition of their exports, imports, or both.By contrast, regions with declines in IIT faced somewhat less structural change. These results suggest the needto re-examine the conventional wisdom that increasing IIT automatically smoothes adjustment to tradeliberalisation”.8 Greenaway et al. (1994) also pointed out the importance of using deflated trade values for the calculation of∆M and ∆X. This is true for all measures which use first differences of trade flows, hence it applies to all theMIIT measures discussed below.9 Note also that the first measure is undefined if base-year IIT is zero, and the second measure is undefinedwhere base-year net trade is zero.10 Greenaway, Lloyd and Milner (1998) have applied this definition of “extended trade” to compute GL indicesfor the US and five of its major trading partners. They found that two-way foreign-owned production issignificantly larger than arms -length IIT.11 Another implication is that the magnitude of adjustment costs is unaffected by whether net trade changes arepositive or negative. This assumption is relaxed in Section 5.12 For a list of references to the SAH in the recent literature, see Brülhart (1999).13 The exercise of Brülhart (2000), which is conducted on Irish data and where adjustment costs are proxied byplant-level job turnover rates, includes among the independent variables the MIIT index (B) and a measure oftrade intensity ((X+M)/output). It does not consider measures of sectoral trade performance.