Marginal Intra-Industry Trade: The Case of Jamaica's Trade with CARICOM

Abstract

Of the dearth of empirical studies on marginal intra-industry trade, there is a relative paucity of empirical evidence involving developing countries. The major contribution of this article is two-fold: Firstly, it measures to our knowledge for the first time the extent of Jamaica's marginal intra-industry trade with CARICOM. This study is warranted bearing in mind that Jamaica's involvement in recent integration efforts with CARICOM will have adjustment costs implications. Secondly, we compare the performance of various Jamaican export sectors with their CARICOM counterparts. The article concludes by proposing some recommendations for policy makers to ensure smoother functioning of the Jamaican labour market.

ACKNOWLEDGMENTS

The author gratefully acknowledges suggestions and comments by participants at conferences and seminars at the Universities of Manchester and Nottingham. This article, which evolved from the author's Ph.D. dissertation, has benefited from the supervision of J. S. Metcalfe, Ken Clark, and Ronnie Ramlogan. The usual disclaimer applies.

Notes

¹ Shelburne (1993, p. 831) previously presented an ‘identical’ index.

²As a result, Brülhart (2000, p. 423) refers to it as his “trade performance index.”

³ Brülhart et al. (2006, p. 523) suggest that “any change in the domestic economy that can be traced to a change in trade volume or prices vis-à-vis the rest of the world, assuming that trade patterns are an unbiased measure of exogenous changes in trade conditions” is a working definition of “trade induced.” Brülhart (2002, pp. 110–111) alternatively defines a change to be trade induced if it can be tracked to a change in the trade-policy regime.

⁴“Sectoral performance” is measured by change in exports vis-à-vis the change in imports, with exports indicating good domestic performance and imports indicating weak domestic performance in a given sector (Brülhart, 1999, p. 48).

⁵In this case, the scaled C index is determined by dividing the basic C index by a scaling variable such as gross initial trade, production or employment (Brülhart, 1994, p. 608).

⁶This is when there is no marginal intra-industry trade and for example, the B index value is 1 or −1 (see Brülhart, 2002). Situations in which trade changes have the opposite signs, i.e., ΔX > 0, ΔM < 0 and ΔX < 0, ΔM > 0 specifically occur in quadrants II and IV of the “trade adjustment space” (see Azhar and Elliott, 2003, p. 423).

⁷Note that previously Azhar et al. (1998, p. 415) had proposed another measure of marginal intra-industry trade.

⁸Alternatively, we could have used export and import price indices to deflate the trade data but these were not available for Jamaica. Constant 1995 values were also ascertained using the GDP deflator to deflate the trade data into base year values and these values turned out to be similar to those obtained using the CPI. Brülhart (1994, p. 610), Brülhart and Hine (1999, pp. 158, 228) and Erlat and Erlat (2003, p. 9) also utilized a CPI to deflate trade data when calculating MIIT.

⁹The authors further pointed out that in calculating and interpreting MIIT indices, the choice of the level of aggregation is also important and a sensitivity analysis with varying levels of aggregation should be carried out where possible (Oliveras and Terra, 1997, pp. 175 and 176). However, due to the unavailability of trade data below the 3-digit SITC level of aggregation, we could not calculate MIIT indices at different levels of aggregation in order to perform such sensitivity analysis. Sensitivity analysis can form a basis for future work when the trade data allow.

¹⁰ Brülhart and Elliott (1998, p. 227) define adjustment costs as those arising “from temporary inefficiencies when markets fail to clear instantaneously in response to changes in demand or supply conditions.” Brülhart et al. (2006, p. 524) further point out that since adjustment costs are not directly measurable, most empirical work focusing on trade and adjustment utilize measures of factor reallocation that intuitively correlate with adjustment costs. This point however mainly applies to other empirical work, including for example, a formal test of the “smooth adjustment hypothesis” where variables such as absolute value of the change in sectoral employment, distance of worker moves or the rate of intra-industry job turnover ratio (as in Brülhart, 2000) proxy for adjustment costs in the labour market (see also Lovely and Nelson, 2002, p. 189). Alternatively, in Brülhart and Elliott's (2000) study, which involved UK manufacturing industries, three indicators of adjustment costs that were utilized are average employment duration, gross variability of industry-level wages and conditional variability of industry-level wages. Note that tests of the smooth adjustment hypothesis involve inter alia testing for correlation between MIIT (preferable to IIT) measures and proxies for adjustment costs as well as a regression analysis where a proxy for adjustment costs is regressed on an M(IIT) measure and other covariates including a variable representing the degree of “trade exposure” (i.e., total trade divided by GDP) (Brülhart and Hine, 1999, p. 125). Some of these studies, for example, check for a positive and significant relationship between the change in employment and Brülhart's (1994) A or B MIIT index (see for example, Brülhart and Hine (1999, pp. 184 and 197) and Cabral and Silva (2006, p. 501) for comprehensive reviews of research that tests the smooth adjustment hypothesis).

¹¹ Balassa (1966) was one of the first to suggest that intra-industry trade might result in lower adjustment costs than inter-industry trade. Lovely and Nelson (2002, p. 182) formally define the “smooth adjustment hypothesis” as the assertion that intra-industry trade generates smaller inter-industry factor movement than inter-industry trade (net trade), and that inter-industry mobility is higher cost than intra-industry mobility.

¹²As Brülhart and Hine (1999) point out, this ratio is useful since the B indices might not be meaningfully aggregated across industries. In our study, the B indices did not all have the same sign and consequently we also employed the b ratio.

¹³Due to space considerations, the full results for the two sub-periods were excluded. These results are available from the author upon request.

¹⁴Less aggregated employment data were available from STATIN's Large Establishment Survey (LES) which is a separate survey from the Labour Force Survey (LFS) and focuses on large establishments only. Nevertheless, it was advised by statisticians from STATIN that it would be very misleading to use inferences from the LES to further disaggregate total industrial employment in the LFS in order to create some concordance between the trade and employment data. Moreover, detailed production data by SITC industry group were also unavailable for Jamaica.

¹⁵ Lewis (2006) already examined Jamaica's trade patterns with individual CARICOM countries in detail. Consequently for the sake of brevity and to avoid unnecessary duplication, we'll not re-examine these issues here.

¹⁶Recall that manufacturing corresponds to the group of SITC sectors that theoretically should have a higher average intra-industry trade relative to non-manufacturing (primary) sectors (Lancaster, 1980).

¹⁷ Greenaway and Milner (1986, p. 162) argue that adjustment problems are perennially associated with disequilibria in labour markets. Brülhart and Thorpe (2001, p. 97) in a similar vein pointed out that analyses of intra-industry trade have been implicitly concerned with adjustment in labour markets.

¹⁸ Cabral and Silva (2006), for example utilized industry-level panel data on over 200,000 firms and 2 million workers for Portugal for the periods 1995–1997 and 1997–1999 in a direct test of the SAH using a newly proposed adjustment cost variable that accounts for reallocation between sectors and occupations, the authors applied pooled OLS and fixed-effect techniques and found strong and robust support for the SAH (Cabral and Silva, 2006, p. 511).