![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
Economic theory indicates some ambiguity in the relationship between intellectual property rights (IPRs) and trade. Here, we extend the empirical literature that attempts to resolve this ambiguity by examining how IPRs affect trade along both the intensive – increasing volume of existing goods – and extensive – increasing variety of goods – margins oftrade. Our main results suggest that IPRs have a positive impact on imports, which is driven by a positive effect on the extensive margin and a negative impact on the intensive margin. Splitting countries according to their level of development, market size and imitative ability, we find that the positive impact of IPRs is strongest in less-developed countries, as well as larger countries and those with a higher degree of imitative ability.
Notes
1. A further complication concerns a firm's decision on its mode of serving a foreign market. In general it faces three possibilities: it may export the good, undertake FDI or license its intellectual asset to a foreign firm. The level of IPR protection may affect the firm's choice, and thus strong IPR protection might diminish trade if it induces firms to choose to serve a foreign market by FDI or licensing rather than exporting (Ferrantino 1993).
2. This number includes both West Germany and the reunified Germany. This means that in any particular year there are only 20 OECD included as exporters.
3. Hummels and Klenow (2005) for example decompose the exports of 126 countries into the contribution of the intensive (volume of goods traded) and extensive (variety of goods traded) margin and then relate each margin to country size (GDP) and its components (workers and GDP per worker). They find that the extensive margin accounts for about 60% of the greater exports of larger economies. Other examples include Schott (2004) who finds that richer countries export higher quality goods, and Funke and Ruhwedel (2002) who find a positive association between the variety of exports and total export volumes. Klenow and Rodriguez-Clare (1997), Hilberry and McDaniel (2002), Feenstra and Kee (2007), Debaere and Mostahari (2010) and Frensch (2010) all consider the impact of trade liberalisation on the margins of trade, generally finding that liberalisation tends to affect trade largely along the extensive margin.
4. Rafiquzzaman (2002) carries out a similar analysis on Canadian manufactured exports. Market expansion effects are found for countries with the strongest threat of imitation, and some evidence of market power effects is found where the threat of imitation is weakest. While the outcomes are broadly similar to those that Smith found for the US, the indications of market power effects are generally weaker for Canadian exports.
5. Liu and Lin (2005) consider exports by Taiwan in three knowledge-intensive industries (semi-conductor, information and communications equipment). For importing countries with a lower imitative (R&D) ability than Taiwan market power effects are found in countries with relatively low imitative ability and market expansion effects in the others. For importing countries whose imitative ability exceeds Taiwan's there are market expansion effects but no market power effects. In an interesting recent contribution, Ivus (2010) uses developing countries' previous colonial status and industry IPR-sensitivity to argue that the IPR strengthening under the TRIPs agreement has led to increased high-tech imports by developing countries.
6. Recent developments in the theory of the gravity equation suggest including importer-time and exporter-time fixed effects to account for the multilateral resistance term of Anderson and van Wincoop (2003). Were we to include importer-time effects in our model it would not be possible to estimate the coefficient on the IPR variable however, since this variable is also importer and time specific. In unreported results we do include exporter-time fixed effects alongside country-pair and time dummies. These results are consistent with those reported below and are available upon request.
7. The 21 countries are Australia, Austria, Belgium/Luxembourg, Canada, Denmark, Finland, Germany, West Germany, France, Greece, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom and the United States of America.
8. UNIDO (2002) notes that the share of R&D financed by enterprises in advanced countries was 98% in the 1980s and 94% in the 1990s.
9. www.cepii.fr/anglaisgraph/bdd/distances.htm.
10. Frensch (2010) defines a reference country that does not vary over time. This has the advantage that all variation across time in the extensive margin is due solely to variations in trade in the exporting country and not to variations in the reference country, but suffers from the drawback that the extensive margin can increase over time solely because of inflation.
11. Rather than estimating the threshold as in Falvey et al. (2009), we impose the thresholds as in Smith (1999). The main reason for this is that if we were to estimate the thresholds we may find different threshold values (as well as a different number of thresholds) for the different indicators of trade, which would then make the results for the different trade measures difficult to compare.
12. The four groups are: (i) less than the 25th percentile; (ii) between the 25th and 50th percentile; (iii) between the 50th and 75th percentile and (iv) above the 75th percentile.
13. Regardless of the set of dummies included the F-statistic from the first-stage regression is always large and significant. The coefficients on the excluded exogenous instruments all tend to be highly significant.
14. Other approaches are discussed in Frankel (1997) and include the use of Tobit estimation or using (Tij + 1) as the dependent variable. Both of these approaches are likely to lead to inconsistent estimates of the parameters however.
15. While both approaches have come in for some criticism, the Santos and Tenreyro (2006) approach is slightly easier to implement. The approach of Helpman et al. (2008) has been criticised due to its heavy reliance on assumptions of normality and homoscedasticity (see Santos-Silva and Tenreyro 2006), while Martinez Zarzoso et al. (2007) find that the Poisson method proposed by Santos-Silva and Tenreyro (2006) suffers from a similar bias as other methods and is not always the ‘best’ model, though it is less affected by heteroscedasticity. Santos-Silva and Tenreyro (2008) point out that,of course, no single estimation technique is always likely to be the ‘best’, but maintain their Poisson estimator is likely to work well in a variety of circumstances. They are also highly critical of the approaches adopted by Martinez Zarzoso et al. (2007) to compare models.