Computation of AVEs of NTMs. We next analyze the impact on trade of the NTMs notified by TTIP and TPP countries. We estimate AVEs, i.e. the level of ad-valorem tariffs that would have an equally trade-restricting effect as the focal NTMs. These AVEs allow comparison between the trade effects of different NTMs and are used in the CGE simulations (see section 2). To compute the AVEs, we apply the method suggested by Xxx et al. (2009). First, we estimate the quantity impacts of NTMs on trade flows, and then convert these effects into AVEs using import demand elasticities computed by Kee et al. (2008). We use the information on NTMs at the HS6 product level, and perform the estimations sector by sector. We define 25 sectors (17 agri-food and 8 non agri-food sectors) which are used also in the CGE simulations. Since the CGE model covers the whole world economy, our sample is extended to include non agri-food activities. Furthermore, third countries are included in addition to TTIP and TPP countries.11 We select all countries notifying NTMs to the WTO, and collect all measures adopted up to 2012. Our final sample includes 125 countries, and covers 92.5% of world trade flows in 2012. Lastly because of the strong correlation between SPS and TBT – especially for agri-food products where countries often notify both types of measures (see Table 4) – we estimate the global effect of NTMs rather than the respective effects of SPS and TBT. Our dependent variable, Mrsi, is the dollar value of imports of good i by country s from country r. We consider only trade flows that are strictly positive in 2012.12 The estimated 11The CGE model also includes services sector, and AVEs for services come from Fontagn´e et al. (2011). 12The absence of bilateral trade between some countries and for some products may be determined by cross-section trade equation is as follows: ln Mrsi = a0 + a1tariffrsi + a2NTMsi + a3distancers + a4cbordrs + a5clangrs + FEr + FEs + FEi + εrsi (1)
