REGIONAL TRADE AND ECONOMIC INTEGRATION IN SOUTH AMERICA 
41
ducing different varieties of the same good and trade among each other if tastes are identical and homothetic. This is what has been suggested by Helpman (1987), and later by Debaere (2002), who applied this version of the gravity model to a group of OECD countries, according to the fact that the similarity in the level of development (measured in terms of per capita income) should be a condition promoting intra-industrial (ex. manufacturing products) rather than inter-industrial trade (ex. manufacturing vs agricultural products). This effect, considered not at a specific regional level but for a more general pair of countries, is known as the Linder hypothesis (Linder, 1961), according to which countries with similar levels of economie development (measured through per capita income) will have more similar tastes and then will exchange more between themselves. Therefore, the theory suggests that the gravity equation succeeds in explaining multilateral trade if the countries in the sample are spe-cialized in the production of differentiated goods, which is more likely in the case of manufacturing rather than agriculture: this is a reason why the gravity model would be more suitable for OECD countries. Nevertheless, the equa-dons seems working well also for developing countries. Feenstra, Markusen and Rose (2000) showed that the gravity equation can arise from different types of models. Some in fact imply a 'home market' effect, in the sense that an increase in the exporter's income has a more than proportionate effect on ex-ports, while others imply the opposite, called 'reverse' home market effect. The first effect seems holding for differentiated goods, while the latter is more pronounced for homogeneous goods (Feenstra, Markusen and Rose, 2000).
Attempts were also made in order to assess the impact of the so-called «multilateral resistance terms» (Anderson and Van Wincoop, 2003), which represent a deeper contribution to catch the effect of trading costs and other variables affecting bilateral trade in a negative way. Many empirical analyses were applied to measure the impact of these variables to bilateral trade flows. The South American region, object of this paper, has been already investi-gated in previous empirical works. Carrillo and Li (2004) applied the gravity model to a sample of South American countries to examine the effects on in-tra-regional and intra-industrial trade in the period 1980-1997. They also looked for the impact of regional trade agreements like Mercosur and the Andean Community (see previous paragraph). The researchers found that the AC had a significant effect on the reference products, while Mercosur had a stronger impact on capital intensive goods (Carrillo and Li, 2004). Another work by Larson, Bittencourt and Thompson (2005), applied on Mercosur countries, shed more light on the issue of trading barriers. After taking into account the macroeconomic internai problems of the main Mercosur members (Argentina and Brazil), they showed that an increase in exchange rate volatility contributes to affect regional trade in a negative way.