«m.dimaio f.tamagni The Authors are very grateful to Alessandro An- ∗ timiani, Giulio Bottazzi, Luca De Benedictis, Giovanni ...»
The evolution of world export
sophistication and the Italian trade
Michele Di Maio
Universit´ di Macerata
a Scuola Superiore Sant’Anna, Pisa
email@example.com; firstname.lastname@example.org. The Authors are very grateful to Alessandro An-
timiani, Giulio Bottazzi, Luca De Benedictis, Giovanni Dosi, Fabio Sdogati, Lucia Tajoli and
two anonymous referees for insightful comments and suggestions. The work also beneﬁted from
discussions with participants to the 11th International Schumpeter Society Conference in Nice- Sophia Antipolis, 21-24 June, 2006; participants to the 47a Riunione della Societ´ Italiana degli a Economisti in Verona, 27-28 October, 2006, and the CNR meeting ’Internazionalizzazione e Sistema Produttivo Italiano’, 22-23 February, 2007. The usual disclaimers apply.
1 Abstract This work provides an empirical assessment of the sophistication level of the Italian international specialization pattern from 1980 to recent years. In particular we present an original analysis which explores the intertemporal co-evolution of the newly proposed PRODY index of export sophistication (Hausmann et al. 2005, 2007) with standard measures of revealed compar- ative advantage. We argue that the results of this exercise can shed light on the Italian ’trade specialization anomaly’ [JEL Codes: C14, F14 ].
Questo lavoro presenta un’analisi empirica del grado di sophistication delle esportazioni italiane per il periodo dal 1980 ﬁno ad anni recenti. Pro- poniamo, in particolare, un originale tentativo di comparazione fra l’evoluzione della distribuzione settoriale dei vantaggi comparati rivelati, e la dinam- ica degli indici PRODY, recentemente introdotti nella letteratura che lega crescita economica e caratteristiche delle esportazioni (Hausmann et al.
2005, 2007). I risultato di questo esercizio fornisce alcuni nuovi spunti in- terpretativi sulle anomalie della specializzazione internazionale italiana.
2 1 Introduction There is widespread consensus among economists and policy makers on the idea that economic growth of a country is to an important extent determined by its external performance. A number of theoretical models has also shown that the strength of this link heavily depends on the sophistication levels associated with the specialization pattern of a country (Kaldor, 1966; Thirwall, 1979; Pasinetti, 1981; Dosi et al., 1990; Grossman and Helpman, 1991). This theoretical apparatus has been largely employed in the debate on the performance of the Italian economy, and often invoked to support the claim that one major diﬃculty for the country lies in its mis-directed pattern of specialization.1 Indeed, it is very well known that Italy is characterized by a peculiar ’trade specialization anomaly’ (Onida, 1999), consisting of a) strong comparative advantages in low-skilled and labor intensive sectors, implying that Italy, in terms of specialization, is much more similar to an emerging economy than to countries with comparable levels of per-capita income;
and b) a remarkably high degree of persistence of such a peculiar structure of specialization (Epifani, 1999; De Benedictis, 2005; Monti, 2005).
The view that the sophistication content of the specialization pattern does matter for economic growth, despite theoretically well grounded, it is very diﬃcult to test empirically. The results of the few studies attempting to do that strongly indicate that openness is not growth enhancing per se, but, rather, it is important to consider the type and the characteristics of the sectors which a country is specialized into. Dalum et al. (1999) convincingly show that the technological characteristics of the specialization pattern are important to explain growth diﬀerentials. Similarly, Feenstra and Rose (2000) ﬁnd a strong relationship between what they call an ’advanced export structure’, on the one hand, and higher productivity levels and faster growth rates, on the other.
1 For an opposite view see Faini (2004).
3 An important contribution to this line of research comes from the recent work by Hausmann et al. (2005, 2007), also developed in Rodrik (2006), where a new quantitative measure of sophistication of exports is presented. Speciﬁcally, they introduce an index – called PRODY – which returns, for each traded sector (product), a weighted average of the per-capita incomes of the countries which are exporting in that particular sector (product). Sectors are therefore ranked in terms of their productivity/income content, whence the name of the index. Based on this, a measure of the overall sophistication associated with the export vector of a country – called EXPY – is also computed, and shown to be a good predictor of subsequent growth.
This recent evidence, supporting that the speciﬁc type of products exported do matter for the growth records of a country, with more sophisticated sectors displaying higher growth enhancing potentials, turns particularly relevant when considering the Italian economy. Indeed, the most remarkable feature of the Italian ’specialization anomaly’ is that Italy, notwithstanding its apparently misdirected specialization pattern, has for a long time enjoyed satisfactory growth records.
The problem is that such positive link does not seem to characterize the country anymore. Why ? Can the evolution of world trade sophistication help explaining the trend observed in recent years ?
Our paper is an attempt to answer these questions. To do that, we structure the study into three steps.
The ﬁrst contribution of the article is to pursue an analysis of the intertemporal evolution of the PRODY index. Extending the work by Hausmann et al. (2005,
2007) we highlight some crucial dynamic properties of this measure and then compute its value for three reference years, 1980, 1990, and 2000, over a sample of 90 countries and 777 traded products. We ﬁnd evidence of signiﬁcant changes occured in the sectoral ranking of sophistication levels during the last twenty years, and 4 we argue that the entry of many low-medium income countries into international trade is largely responsible for the observed dynamics. Then, as a second step, we turn to Italy and describe the characteristics and the evolution of the specialization pattern of the country in between 1977 and 2004. The exercise is conducted with a ﬁner level of sectoral disaggregation with respect to previous studies about Italy, but our results conﬁrm much of the existing evidence: the structure of comparative advantages has been highly persistent and characterized by a pronounced bimodality. Finally, we present two complementary methodologies which are useful to evaluate the sophistication levels associated with the specialization pattern of a country. The ﬁrst one is based on the EXPY index presented in Hausmann et al.(2005, 2007). The second represents our original contribution and consists of exploring the co-evolution of the sectoral PRODY indexes with a standard measure of sectoral specialization. Both the methods are then applied to Italy, allowing to provide a quantitative assessment of the evolution of export sophistication in this country over the period considered.
Overall, our ﬁndings suggest a possible answer to our initial questions. Ultimately, the evolution of world trade seems the very reason why the ’specialization anomaly’ of Italy, which did not prevent to achieve suﬃciently good growth performances in the past, has instead started to become a point of weakness for the economy in more recent years, since the late ’90s at least. Indeed, while the changes occured in the PRODY indexes reveal that, during the last two decades, the entry of new competitors (emerging countries in particular) as well as a vast world-wide redistribution of production have signiﬁcantly changed the relative gains associated to exporting in each speciﬁc sector, Italy has remained stuck to its initial specialization pattern. A problem then arises because most of the sectors wherein Italy has persistently been, and still is, highly specialized, are characterized by an intertemporal decrease in the associated PRODY index, entailing a reduction in
5their sophistication content as compared to the past.
The paper is organized as follows. In Section 2 we introduce the PRODY index and provide a theoretical discussion of its crucial dynamic properties. Section 3 presents a set of parametric and non-parametric exercises exploring the evolution of the index during the period 1980-2000. The evolution of the Italian specialization patterns is then analyzed in Section 4. Section 5 applies to Italy the two methodologies proposed to evaluate the sophistication associated with international specialization of a country, namely the EXPY index and our novel analysis of the co-evolution of the PRODY indexes with sectoral comparative advantages.
Section 6 concludes and suggests directions for future research.
normalizes country i’s Balassa index of Revealed Comparative Advantage (RCA) with respect to those of all the countries exporting in the same sector (Rodrik, 6 2006).2 The PRODY index is thus a sectoral measure returning a weighted average of the levels of development (proxied by per-capita income) of all the countries producing and exporting in a given sector. By construction, sectors with high values of PRODY are those where high income countries play a major role in world exports. Therefore, under the reasonable assumption that high income/high wage countries display a strong presence in sectors wherein comparative advantages are determined by factors other than labor cost (such as know-how, technological content, intrinsic quality, and so on), then sectors with an high PRODY index are more sophisticated than sectors with a low value of the index. To the extent that such factors set the stage for non-perfectly competitive environments, an high value of PRODY in a given sector also signals that higher proﬁt margins and, thus, greater growth opportunities are associated with producing and exporting in that sector.3 In the study of Hausmann et al. (2005, 2007) a static analysis is provided, computing a single value of each sectoral PRODY index as a three-year average for 1999-2001. We pursue a complementary dynamic perspective, and examine the evolution of the PRODY indexes over time. The idea is that intertemporal changes in both the value and the ranking of the indexes can reveal if signiﬁcant mobility and transformations have taken place in world production and trade over 2
Recall that the Balassa-RCA index is deﬁned as:
Xlit /Xit RCAlit = (3) Xlw /Xw where X denotes the value of exports, l the sector, i the country, t the year considered, and w stands for the world economy. The index measures the importance (in terms of value) of sector l in the export bundle of country i with respect to the importance of sector l in worldwide export ﬂows. If the index is bigger (smaller) than 1, then country i is said to be specialized (de-specialized) in sector l.
3 On this interpretation see also Lall et al. (2006).
7 time. In addition, recalling the above mentioned relationship between sophistication and the growth enhancing opportunities stemming from exports, intertemporal changes in PRODY may also indicate whether the potential contribution to economic growth associated with trading speciﬁc goods has changed with time.
Therefore, following the dynamics of the PRODY index should provide us with a quantitative assessment of these events.
In order to understand and interpret the observed intertemporal movements of the index it is however necessary to describe how the index responds to changes in its main determinants. These are a) the income levels and the extent of specialization of the countries involved in world exports of each sector; and b) the number of exporting countries in each sector, resulting from entry and exit dynamics. From the above deﬁnition 1, one can make the following statements. First, an increase (decrease) recorded in the per-capita income of one of the countries exporting in sector l increases (decreases) the value of the PRODY index of the same sector l, proportionally to the share of world exports of sector l which pertains to that country. Second, the eﬀect on the PRODY index of sector l due to an intertemporal change recorded in the degree of specialization of one of the countries exporting in the same sector l depends in a non-trivial way on the simultaneous modiﬁcations occuring in the distribution of export shares across all the countries involved in world exports of sector l (see Di Maio and Tamagni (2007) for a derivation). Third, the way in which entry and exit of diﬀerent exporting countries aﬀect the value of each sectoral PRODY crucially depends on the relative income levels of the new as compared to the already exporting countries: the higher (lower) the per-capita income of the entrant country vis a vis the income level of the incumbents, and the stronger the increase (the decrease) which is expected in the index of a given sector.4 4 See Proposition 1 in the Appendix for a precise statement and a formal proof of this property 8 The last consideration is particularly important for two reasons which are both closely connected with the dynamic perspective which we pursue here. First, because the property provide a guidance to attach an economic interpretation to the empirical evidence about the evolution of the PRODY indexes which we are going to present in the next section. Second, the peculiar eﬀects of entry and exit of diﬀerent exporting countries makes explicit that our attempt to follow the intertemporal evolution of PRODY indexes requires speciﬁc care in the construction of the dataset. Indeed, in order to avoid mis-measurement of each sectoral index over time, it is not enough to gather information on the largest possible number of countries. It is also essential to maximize the number of countries for which data are available and reliable over the entire sample period considered.5 3 Evidence on the dynamics of PRODY In this section we present our empirical analysis of the evolution of the sectoral PRODY indexes, focusing on the dynamics observed across three reference years, 1980, 1990 and 2000. Exploring parametric and non-parametric exercises our main goal is to document the extent and the directions of the changes occured.