Measurement of comparative advantage and meaning of RCA indices

In the seminal paper that gave birth to the empirical literature on the Revealed Comparative Advantage (RCA) Balassa (1965) observed that the most direct way of ascertaining the pattern of comparative advantage would be through inter-country cost comparisons based on internationally comparable industrial censuses or through comparison of autarchic and free trade relative prices (Balassa, 1965). The only such cost comparison that existed at the time was the Organisation for European Economic Cooperation study (OEEC, 1958) that covered the United States and the United Kingdom in 1950 and, according to Balassa, suffered from methodological problems. Comparison of autarchic and free trade prices was yet more elusive as neither of these can be directly observed.⁷ Moreover, Balassa (1965) argued, neither the comparative costs, nor the autarchy price approach would be capable of capturing the non-price determinants of comparative advantage such as quality differences, goodwill or the availability and quality of servicing and repair.

In this context Balassa (1965) proposed that comparative advantage would be expected to determine the structure of exports⁸ and that export flows can be used to evaluate export performance of individual industries with the use of export-performance indices. Later these indices were dubbed in the literature as ‘revealed comparative advantage’ (RCA) indices⁹ and have been commonly used to measuring comparative advantage. Balassa’s approach was to use normalised export shares to evaluate export performance of individual industries. This normalisation was executed by dividing the export share of country i in world exports of individual commodity j by the country’s share in the combined world exports of manufactured goods ((a) in Equation 1). Alternatively, the same index could be expressed as the ratio of commodity j’s share in country i’s total exports and j’s share in world total exports [(b) in Equation 1].

As evident in formulation (a) of Equation 1 an RCA index greater than 1 (RCa>reflects the situation where the weight (or concentration) of a given product in the country’s exports is higher than the weight (concentration) of this product in world exports. Such a situation is interpreted as an indication of ‘revealed comparative advantage’. Thus, for example, it would be expected of a country to have a larger share in the export of those goods which are produced with factors it holds in relative abundance.

Yet, this interpretation is associated with several caveats. First, as can be seen in equation 1, the upper bound of the RCA index either varies by country and is fixed across sectors (a) or varies by sector but is fixed across countries (b). The implication of this is that the index is suitable for either cross-sector comparison within the same country or for a cross-country comparison within the same sector. A simultaneous cross-sector crosscountry comparison needs to be approached with caution. We can say an RCA of 2 for Germany’s product A compared to an RCA of 3 for Germany’s product B indicates a stronger export performance, higher degree of export specialisation, and a higher degree of revealed comparative advantage of Germany’s product B. We can also say that an RCA of 4 for France’s product B would indicate a higher degree of export specialisation and better export performance of France in the same product B. What we could not say is that an RCA of 4 for France’s product B compared with an RCA of 2 for Germany’s product A indicates a better export performance of the former as compared to the latter.

The latter restriction is related to the fact that in a simultaneous cross-country, crosssector and cross-time comparison both nominator and denominator vary, calling for a consideration of the joint effect. Hence, for example, an increase in an RCA index value between two time periods could entail both a decrease in sectoral export share and an increase in country’s share in world exports. In an empirical analysis this problem can be overcome with some of the statistical tools that consider the whole distribution of the RCA index and its evolution over time, such as the transition probability matrices or ergodic distributions. This is the approach taken in the remainder of this chapter.

As long as we remember about these caveats the index can be used as a cardinal, ordinal or a dichotomic measure of relative export specialisation. As an ordinal measure it allows ranking the index on the base of its value and asserting which sectors’ (of an individual country) had a stronger export performance. As a cardinal measure we can use RCA indices to directly compare two values of the index and measure their difference.¹⁰ Hence, for example, twice as large an RCA for product A indicates twice as high a country’s normalised export share in world market. Finally, the dichotomic interpretation of the index allows us distinguishing sectors with revealed comparative advantage (RCa>to sectors with revealed comparative disadvantage (RCA<1).

The RCA index is itself a ratio of export shares, which have attracted attention on their own. Indeed, export shares do often get analysed in studies of trade performance. A rise in a country’s share of the world market for a particular product j (Xi/X_wj) - the numerator of the RCA index - is sometimes interpreted as an indication of country’s commercial success or competitive edge in this product category. A rise in a country’s overall share in world exports (X_;/X_w) - the denominator of the RCA index - has also been interpreted as an indication of a country’s general commercial success.¹¹ An increase in the RCA index in this context reflects a situation where a commercial success of a particular product category exceeds the commercial success of a country as a whole.¹²

It has to be emphasised, however, that these interpretations are only loosely related to what trade theory tells us about the benefits of trade. Comparative advantage-based trade theories do not give any interpretation of product export shares -they only posit that countries will gain from specialisation according to comparative advantage so that welfare gains are associated equally with products that country exports (large relative export shares) as with those that country imports (small relative export shares). Similarly, theory does not provide any indication with respect to what a country’s overall share in world trade should be, except that in most analytical frameworks this share is related to country’s size.

Further investigation of some properties of the index presented in Annex 3.A reveals a number of interesting implications concerning the relationship between relative product export shares and country’s overall world export share. First, whether an increase in country’s share in world trade occurs depends on whether individual product shares increase on average, where average is weighted by product shares in world trade (Xwj/Xw). Secondly, a given percentage point increase in the world market share for a particular product of a given country results, ceteris paribus, in a larger impact on our measure of revealed comparative advantage the less traded worldwide the given product is. At the same time, a given percentage point increase in the world market for a particular product results, ceteris paribus, in a larger impact on country’s share in world trade the more traded the given product is worldwide. Finally, for any country the weighted average of RCA indices across the product space equals one, weights being again shares of particular products in world trade.¹³ Overall, however, any pattern of changes in RCAs across the product space consistent with the above-stated conditions could lead to an increase in country’s share in world trade. This property of the index is another indication of its approximate relation to theory.¹⁴

In conclusion, the RCA index is not a measure of comparative advantage that is strictly derived from any particular trade theory but this does not disqualify its usefulness as an analytical tool that can guide evidence-based policy discussions. Admittedly, the index is influenced by a multitude of factors, including country’s natural characteristics

(e.g. the amount of land available per head of population, based on current demographics) as well as economic policies which may benefit certain sectors more than others (e.g. infrastructure or R&D expenditures), irrespective of whether such policies are welfare enhancing or sustainable in the long term. Yet, it would be rather demanding, if at all desirable, to expect a general index to filter out all welfare reducing factors without filtering out the welfare enhancing factors, or vice versa. After all, as discussed above, the policy implications of the comparative advantage hypothesis remain somewhat vague. The uncomplicated and transparent nature of the index is its advantage that undoubtedly contributed to its popularity in applied trade analysis. It can help policy makers identify important trends that can be studied further using additional information on specific sources of comparative advantage or country or industry-specific information. This is the approach of the empirical investigation presented in the remainder of this chapter.