The role of non-tariff measures in EU dairy trade with Russia

Abstract

This article investigates Russian non-tariff measures (NTMs) on dairy products and their implications for EU dairy exports. Based on survey results, numerous and detailed Russian standards on imported dairy products are considered by respondents as redundant and unnecessary from a food safety perspective. Conformity assessment procedures are identified as a major problem when exporting to Russia. They are non-transparent, time-consuming and expose exporters to significant risk that their products may be refused entry at the Russian border. Audits by Russian inspectors seem to be subject to arbitrary rules and exporting companies face great uncertainty because of unclear and often changing rules. Both fixed and variable costs may increase due to Russian non-tariff measures, adding an estimated 5–10% of export value to costs. The gravity model estimates indicate that, after controlling for other variables, non-tariff measures are more restrictive on US exports to Russia than on EU exports to Russia, while New Zealand's exports to Russia are least affected by NTMs. Overall, the estimates do not confirm that Russia's NTMs are significantly more restrictive than is the case with other countries' NTMs. Although Russian standards for dairy imports are inhibiting trade they are not more restrictive than those implemented by other countries.

Acknowledgements

We acknowledge financial support from the European Commission FP7 project ‘Assessment of the impact of non-tariff measures – NTM – on the competitiveness of the EU and selected trade partners (NTM IMPACT)’ as well as from the Slovak Research and Development Agency under contract No. APVV-0894-11. The views expressed in the article are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission.

Notes

1. The gravity approach to modelling bilateral trade has a long tradition. Tinbergen (1962) and Poyhonen (1963) are credited as the originators of the method. The theoretical foundation for the use of the gravity equation was provided by Bergstrand (1985, 1989) who associated it with the simple monopolistic competition model. Helpman and Krugman (1985) and Helpman (1987) derived a gravity equation from the model of increasing returns to scale and differentiated products in international trade. Deardorff and Stern (1997) show that the gravity model is also consistent with the Heckscher–Ohlin theory of international trade. The gravity approach has also been widely used to investigate the impact of NTMs on agricultural trade. Moenius (2004), Fontagné et al. (2005), De Frahan and Vancauteren (2006) and Disdier et al. (2008) study aggregate agricultural trade with a gravity model. Disaggregated agricultural trade is analysed by Wilson and Otsuki (2001) and Otsuki et al. (2001). NTMs vary significantly across goods (Anderson and Wincoop 2004). Dairy products are among the most protected, with high tariffs and tariff rate quotas as well as divergent standards posing difficulties to trade. These all are likely to be addressed within the current discussion on further trade and agricultural policy.

2. Bilateral distances have been calculated using the city distance tool available from: http://www.geobytes.com/citydistancetool.htm.

3. Empirical estimation of the gravity model suffers from the problem of heteroscedasticity and zero trade flows. Heteroscedastic errors lead to biased estimators (Santos Silva and Tenreyro 2006). Zero trade flows occur because of errors, omissions, rounding and due to real absence of trade. The extent of zero trade flows is especially large when disaggregated data are used. Zero values of dependent variables can lead to biased estimation. The reason is that the sample selection process is not independent from error terms and relevant explanatory variables are omitted. In the presence of heteroscedastic errors the Poisson pseudo-maximum-likelihood (PPML) estimator is the least biased. The PPML estimator, however, does not perform well when a large proportion of observations is censored. In such a case, the least biased estimator is that proposed by Eaton and Tamura (1994).