Trade barriers in services and merchandise trade in the context of TTIP: Poland, EU and the United States

Transcription

1 Trade barriers in services and merchandise trade in the context of TTIP: Poland, EU and the United States Jan Hagemejer University of Warsaw, Faculty of Economic Sciences and National Bank of Poland 1 Katarzyna Śledziewska 2 University of Warsaw, Faculty of Economic Sciences Abstract We evaluate both the tariff and non-tariff barriers in the bilateral trade of the EU member and the United States. We differentiate between EU-15, the New Member States and Poland to account for their structural differences and different trading patterns. We use a standard gravity framework and attribute the differences in importer fixed effects to non-tariff barriers. We find that the level of non-tariff barriers trump in general trumps the one of the tariffs across different sectors and analyzed groups of countries. The overall tariff equivalent in merchandise trade is around 20%, at least twice the level of tariffs. However, there are sectoral differences and in many cases the NTBs exceed 40%. We also find the non-tariff barriers in services trade to be overall higher than those in merchandise trade. At the same time, Poland and the remaining NMS exhibit higher barriers to trade in services than in the EU15 and the US. Keywords: TTIP, international trade, non-tariff barriers JEL: F13 1 The views presented here are those of the authors and not necessarily of the National Bank of Poland. This project is financed by the National Science Centre, decision number: DEC- 2013/09/B/HS4/ Corresponding author: k.sledziewska@uw.edu.pl

2 Introduction After remarkable reductions in tariffs resulting from the General Agreement on Tariffs and Trade (GATT) and World Trade Organization (WTO) negotiating rounds, technical barriers to trade (TBTs) represent the hampering factors for trade relations. Recent studies identify that non-tariff measures (NTMs) or alternatively non-tariff barriers to trade (NTBs) represent the major sources of import protection and therefore provide considerable distortions for trade. The issue at hand gets the particular actuality for the case of the two main economic powers in the world, the EU and the US especially, since proposition of free trade agreement between the EU and the US under the Transatlantic Trade and Investment Partnership (TTIP) in The partnership implies alignment of the NTMs and regulatory divergences by cutting non-tariff costs imposed by bureaucracy as well as from liberalising trade in services and public procurement. Therefore, the analyses of the distortionary effects of NTMs on trade flows and economic performances of the two economic giants is highly considerable. Already in 2003, Bradford estimates that US NTMs add 12 % to the cost of trade with the United States, while European NTMs add between 48 and 55 % to the cost of traded consumer goods. In the context of the European Single Market, a study by Copenhagen Economics (2005) underlines that removal of NTMs for the EU services directive might yield remarkably positive economic impacts. Likewise, according to the more recent studies conducted by ECORYS (2009) and Francois, Manchin, Norberg, Pindyuk and Tomberger (2013) the removal of non-tariff barriers through the trans-atlantic partnership might bring remarkable trade expansions and considerable welfare gains. Based on the theoretical framework of the Gravity model we extend the analyses of the topic from the Polish perspective. We analyse the sectoral trade flows between the EU and the USA by employing the Global Trade Analysis Project (GTAP) dataset covering the period Based on the estimation results, finally we compute the non-tariff measures. The paper is organized in the following way: section 2 provides the literature review and the theoretical framework of the analyses, section 3.. followed by the data description and estimation results in the section 4, finally section 5 summarizes the findings of the paper.

3 Review of methods The most popular framework for empirical analyses of the trade remains the gravity model introduced by the crucial work of Jan Tinbergen (1962). Reflecting the relationship between the size of economies, their distance and the amount of their trade, the gravity equation can be expressed in the following form: X ij= GS j M j ij where Xij is the monetary value of exports from i to j, M j controls for all importer-specific factors that make up the total importer s demand (for example the importing country s GDP) and Sj comprises exporter-specific factors (for example the exporting country s GDP) that represent the total amount exporters are willing to supply. G is an independent variable from i and j, such as the level of world liberalization. Finally, ϕij represents the ease of exporter i to access of market j (trade costs). Typically, trade costs are measured by the bilateral distance suggesting that transport costs increase with the distance between countries. In addition, empirical studies also reveal the existence of the information costs. For example firms are more likely to search for the new market in countries where the business environment is familiar to them. In other words similar business practices, common cultural features or a common language lowers the information costs for trading. Finally, the very important role comes on tariff rates. However, after the repeated reductions in tariffs due to the WTO negotiations, the analyses of NTMs gets the considerable importance. According to the definition provided by the WTO, technical barriers to trade are regulations, standards, testing and certification procedures, which could obstruct trade. It consists of non-tariff barriers or alternatively non-tariff measures such as quotas, import licensing systems, sanitary regulations, prohibitions, etc. Computations of tariff equivalents reveal the existing protection which distort the trade flows. However, because NTMs are not directly computable, the measurement of their impact is not easy to derive.

4 The recent literature reveals the two ways of the possible computation procedures. First, as suggested by Park (2002) the distribution of the residuals of an estimated gravity equation can be used for the computation of the equation of tariff equivalents. Alternatively, we can compute the average protection applied by each importer, from the importer fixed effects coefficients (Fontagne, Guillin and Mitaritonna, 2011). In the first case, following Anderson (2001) and Park (2002), some constraints have to be imposed on the sum of residuals for a given importing country in a given year, namely it should be equal to zero and the sum of all residuals for a given year must also be equal to zero: ε ij = 0 i ε ij = 0 i j According to Anderson and Wincoop (2003) and Park (2002) we may define the residual as the difference in log values of actual and predicted import values from country i to country j. Then the difference between the total actual and predicted value of country imports may indicate the level of distortion to trade caused by existence of trade barriers. However, the absolute differences should be normalized relative to a benchmark free-trade country case. Finally the relative trade barrier (tj) of country j can be measured by the following equation (Park, 2002): σlnt j = ln X j a p ln X a b p, X j X b where the sub-index a, p and b represent actual, predicted and benchmarks respectively, X j is the country j s simple average imports and σ stands for the constant elasticity of substitution. Following the assumption made by Park (2002) we take the value of σ equal to 5.6. The countries with the greatest positive differences between actual and predicted imports are considered to be the free trade benchmark countries in the regression. In the second, fixed effect, methodology, (Fontagne, Guillin and Mitaritonna, 2011) we can obtain tariff equivalent by the difference between the fixed effects of the country j and the benchmark country. Specifically,

5 σlnt j = Feγ j Feγ benchmark As in the first approach, σ stands for the constant elasticity of substitution assumed to be equal to 5.6. However, in this methodology the benchmark country is the one which yields the highest importer fixed effect coefficient. The above-mentioned theoretical approaches provide the main framework for recent empirical analyses of impacts of non-tariff barriers between the EU and the US. According to a study conducted by ECORYS (2009) in case of alignment of 50% of the NTMs and regulatory divergences, US exports are expected to go up by 6.1% and the EU exports by 2.1% yielding the trade balance improvement for both the EU and US. As a result of extended bilateral trade flows, the EU GDP could increase by 0.7% and the US GDP by 0.3% per year bringing additional 122 billion Euros to the EU and 41 billion Euros to the US as an annual gain. The study also reports the effects on sectoral levels. For the EU, trade flows in motor vehicles, chemicals, cosmetics & pharmaceuticals, food & beverages and electrical machinery are found to be mostly constrained by the NTMs. For the US case, the report outlines that together with the goods sectors trade in services are also hampered by nontariff barriers. Namely, together with electrical machinery, chemicals and cosmetics & pharmaceuticals, insurance and financial services are mainly constrained by NTMs. More recently, Francois, Manchin, Norberg, Pindyuk and Tomberger (2013) analyse the role of non-tariff barriers (NTBs) in the EU-US bilateral trade. As the study shows, 80% of the total potential gains are expected to come from removal of trade costs imposed by bureaucracy and regulations. Estimations of the expected increases in trade flows as a result of the removal of NTMs are close to those provided by ECORYS (2009). Namely, the study estimates that removal of NTBs could bring 119 billion for the EU and 95 billion for the US per year. Additionally, predictions are extended by outlining that the benefits for the EU and the US would not be at the expense of the rest of the world. On the contrary, according to the study liberalizing trade between the EU and the US would increase GDP in the rest of the world by almost 100 billion. The sectoral analyses on the GTAP level also reveals that the motor vehicles sector is characterized by an initial combination of high tariffs and high nontariff barriers, mainly such as different safety standards. In this sector, EU exports to the US are expected to increase by 149%. As authors explain, this partly reflects the importance of

6 two-way trade in parts and components and the further integration of the two industries across the Atlantic. Putting the emphasis on the trade in services, Ghodsi, Hagemejer and Kwiecinska (2015) assess the degree of the trade liberalization and its impact on bilateral trade of rail transportation services during for 27 European countries. Together with the standard gravity variables to explain the bilateral trade, authors use four indices measuring the market liberalization in rail transport services, elaborated by the IBM Consulting Services. Estimation results of Fixed and Random effects reveal that among all the measures, only the liberalization of the legal framework has a significant impact on the volume of imports. Additionally, estimations of tariff equivalents did not reveal a clear downward trend in the levels of non-tariff barriers between EU countries and OECD countries over the analysed period. As the literature outlines the cost of NTMs in the trade relations between the EU and the US is considerable. Based on the actuality and importance of the issue at hand, we extend the analyses of the topic from the Polish perspective. We employ the GTAP dataset to provide the calculations of tariff equivalents on the sectoral level for the period Tariff Profiles: the EU and the US We can analyze trade costs between the EU and the US by considering the tariffs applied on the imports of goods in the different sectors based on the GTAP classification. Table 1 shows the tariff rates applied on the imports from the EU in the US and in all other trade partners in 1999 and in Here the tariff rates represent the effectively applied simple average tariffs, averaged over all the partners. First of all, we may observe, that the tariff rates applied by the US are remarkably lower than those applied by the other trade partners in 1999 as well as in Second, the table shows that compared to 1999, in 2013 tariff rates are reduced by all the trade partners of the EU, however the tariffs applied by other than US countries still remain at much higher level than tariffs applied by the US. The reduction in tariff rates is also observed in the case of the US, however with few exceptions. Namely, compared to

7 1999, in 2013 tariff rates applied by the US slightly increase for fuels, non ferrous metals, transport equipment nec and manufacturing nec. Table 1. Tariffs applied on the imports from the EU in 1999 and GTAP sectors USA Other Trade Other Trade USA Partners Partners Agriculture 0,7 8,2 0,5 6,2 Mining 8,6 26,4 5,5 25,0 Food 8,5 15,8 7,5 13,5 Textiles 11,2 19,1 10,7 17,6 Wearing apparel 8,6 16,5 8,9 15,3 Leather 1,3 15,0 1,2 11,5 Wood 1,1 12,5 0,0 9,4 Paper, publishing 3,0 7,8 3,5 6,1 Fuels 2,8 12,3 2,6 9,7 Chemicals 4,1 13,7 4,1 11,4 Minerals nec 1,5 10,4 0,9 8,1 Steel 2,3 9,4 2,6 7,7 Non ferrous metals 2,4 13,7 2,2 11,4 Metal products 2,3 14,0 1,3 11,5 Motor vehicles 1,0 9,5 1,3 8,5 Transport equipment nec 0,9 8,0 0,6 6,0 Electrical appliances 1,4 9,9 1,3 8,1 Machinery and equipment 0,7 8,2 0,5 6,2 Manufacturing nec Source: World Integrated Trade Solution (WITS), own calculations. Table 2 shows the tariff rates applied by the EU and by the all other trade partners on the imports from the US in 1999 and in Likewise in the previous case, tariff rates represent the effectively applied simple average tariffs, averaged over all the trade partners. Again, we may observe, that the tariff rates applied by the EU are remarkably lower than those applied by the other trade partners in both, 1999 and Second, the table shows that compared to 1999, in 2013 tariff rates are reduced by all the trade partners of the US, however the tariffs applied by other than the EU countries still remain at much higher level compared to the EU tariffs. The reduction in tariff rates is also observed in the case of the EU, however again there are few exceptions. Specifically, compared to 1999, in 2013 tariff rates applied by the EU slightly increase on agricultural products, mining, fuels, metal products, motor vehicles. The increase in tariffs on leather and paper, publishing is negligible (0,03 for the former and 0,04 for the latter) so that, we can consider them unaffected.

8 Table 2. Tariffs applied on the imports from the US in 1999 and GTAP sectors EU Other Trade Other Trade EU Partners Partners Agriculture Mining Food Textiles Wearing apparel Leather Wood Paper, publishing Fuels Chemicals Minerals nec Steel Non ferrous metals Metal products Motor vehicles Transport equipment nec Electrical appliances Machinery and equipment Manufacturing nec Source: World Integrated Trade Solution (WITS), own calculations. To conclude, the tariff data on the sectoral level demonstrates that the tariffs applied by the EU on the imports from the US as well as tariffs applied by the US on the imports from the EU are considerably lower than those applied by the other biggest trade partners. Even though the dynamics from 1999 till 2013 shows the apparent overall decrease in tariffs rates, in the case of the EU and the US the reduction is not remarkable since initially the rates were already low in This trend suggests that tariffs do not represent the main trade barriers between the EU and the US. And therefore, they cannot be considered as the main tool for the further trade liberalization between these two partners. Estimation method and data We follow the Park methodology using the importer fixed effects in order to obtain the average trade levels for countries under investigation as well as the reference country. In our estimation equations both for merchandise trade and services we use the standard gravity variables such as (logs of) reporter and partner GDP, population, distances between group of countries, contiguity, common language and colonial ties. Besides reporter fixed effects we also include partner dummies as well as year dummies to capture variation in trade over

9 time. In the merchandise trade equation we also include a bilateral level of effectively applied tariffs. The merchandise trade data comes from Comtrade through the WITS database. The tariff levels come from WITS. Services data are taken from Francois and Pindyuk (2013). We match the trade data with the macroeconomic characteristics of partners and reporters that are taken from World Development Indicators Database. The geopolitical gravity variables are provided by CEPII in their Mayer and Zignago (2011) paper. We limit the year time span of the database to start from We do it in order to provide a fairly recent estimate of trade barriers (including the effects of the EU enlargement) while at the same time maximize the number of observations. The resulting merchandise trade database is much larger than the one for services: it includes over 350 thousand of observations as opposed to only 78 thousand observations for services. We initially planned to focus on extra-eu trade data only, but this was only possible in the case of merchandise trade as in trade for services the number of observations dropped sharply. Therefore, the services data estimations include both intra- and extra-eu trade in servicese, but we add an additional dummy variable corresponding to both importer and partner being a member of the EU in order not to inflate the importer fixed effect in the case of EU members. The merchandise trade database covers and the services trade data All estimations are performed in GTAP sectors to facilitate the use of the obtained tariff equivalents in trade policy simulations. While WITS database offers GTAP aggregation as one of the standard choices, Francois & Pindyuk services dataset is also aggregated into that classification. The obtained importer fixed effects are then sorted for each sector to identify the reference (most liberal country). The tariff equivalent is computed as the difference the a country s fixed effect and the reference country fixed effect multiplied by the elasticity of substitution. While the estimation equation enables us to identify the elasticity of substitution from the tariff coefficient, it does not prove to be very reliable. Instead, we use GTAP sectoral elasticity of substitution in order to increase compatibility with CGE simulations. Table 3. Gravity estimations for merchandise trade part 1

10 VARIABLES Total Agriculture Mining Food Textiles Apparel Leather Wood Paper CoalPetrol log(gdp Reporter) 1.066*** 0.886*** 0.508*** 0.474*** 1.098*** 1.910*** 1.303*** 1.145*** 0.999*** 0.599*** (0.0220) (0.0615) (0.110) (0.0578) (0.0617) (0.0647) (0.0732) (0.0600) (0.0632) (0.128) log(gdp Partner) *** * *** 0.250** *** (0.0370) (0.108) (0.187) (0.102) (0.0982) (0.0976) (0.119) (0.102) (0.104) (0.192) log(population Rep.) 0.161*** 0.272*** 1.235*** 0.563*** 0.178*** *** *** 0.571*** (0.0200) (0.0585) (0.107) (0.0567) (0.0571) (0.0605) (0.0672) (0.0567) (0.0587) (0.125) log(population Part.) *** ** *** *** ** ** *** (0.110) (0.418) (0.853) (0.391) (0.287) (0.310) (0.373) (0.341) (0.322) (0.763) log(distance) *** *** *** *** *** *** *** *** *** *** ( ) (0.0263) (0.0399) (0.0234) (0.0239) (0.0253) (0.0281) (0.0264) (0.0277) (0.0483) contiguity 0.663*** 0.662*** 0.621*** 0.423*** 0.703*** 0.668*** 1.066*** 0.614*** 0.651*** 0.691*** (0.0256) (0.0870) (0.125) (0.0760) (0.0755) (0.0864) (0.0958) (0.0817) (0.0963) (0.117) common language 0.546*** 0.434*** 0.568*** 0.761*** 0.478*** 0.702*** 0.351*** 0.461*** 0.960*** (0.0193) (0.0567) (0.0947) (0.0523) (0.0528) (0.0579) (0.0592) (0.0573) (0.0617) (0.109) colony(1945) 0.878*** 1.346*** 0.331** 1.098*** 1.072*** 1.147*** 1.141*** 0.892*** 0.966*** (0.0370) (0.107) (0.141) (0.121) (0.129) (0.129) (0.124) (0.124) (0.126) (0.206) log(tariff_weighted) *** *** *** *** *** *** *** *** (0.0698) (0.161) (0.420) (0.124) (0.220) (0.227) (0.292) (0.242) (0.293) (0.449) Constant ** *** *** ** (1.853) (6.632) (13.24) (6.262) (4.921) (5.152) (6.303) (5.627) (5.299) (12.23) Observations 357,139 19,186 16,654 19,821 19,695 19,035 18,345 19,025 19,738 14,044 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Estimation results Our gravity estimations are focused on the flows of imports. In the merchandise trade estimations the results seem to be fairly plausible. The standard gravity variables have in general expected signs. The GDP of reporter is positively related to its imports. The overall elasticity across all sectors is close to 1, however, there are visible differences, ie. agriculture, food, petrol, mining as well as chemicals have a coefficient that is significantly smaller than 1, in line with the idea of the non-homothetic demand for basic goods and at the same time the fuel intensity of GDP decreasing with income. The partner size is not as important in explaining trade flows and also the signs vary across sectors. However, we have to keep in mind that due to the inclusion of reporter and partner fixed effects the cross-country size variation is already embedded in those effects and therefore the macro coefficients are related to within-pair changes. Therefore as there is a one-to-one reporter size-imports relationship, we do not find anything like that for the exporter-size link. The distance elasticity is negative across estimations and in most cases it is much larger than one indicating that trade dies out fairly quickly with increasing distance. The contiguity, common language and colony are positive, significant and rather large in most estimations.

11 The calculated tariff equivalents of the trade barriers are given in table 5. Column two of that table reveals the reference country (a country with the highest importer fixed effects). In most cases, the reference countries are South-East Asian countries such as Malaysia, Vietnam and Thailand but also Australia and Chile. We base our calculation on GTAP Armington elasticities as we believe that the estimated elasticities of substitution as identified by the coefficient on the tariff yield values that are too low compared to common sense. Table 4. Gravity estimations for merchandise trade part 2 VARIABLES Chemicals Minerals Steel Metals Met. Motor Tr. Eq. Electric Machinery MnfcsNec Prod. Veh. Nec log(gdp Reporter) 0.789*** 1.141*** 0.761*** 1.399*** 1.322*** 1.276*** 1.293*** 1.861*** 1.461*** 1.477*** (0.0513) (0.0594) (0.0805) (0.0884) (0.0568) (0.0675) (0.0862) (0.0600) (0.0492) (0.0540) log(gdp Partner) 0.235** *** 0.285*** ** 0.825*** * (0.101) (0.0936) (0.129) (0.136) (0.0973) (0.111) (0.136) (0.0988) (0.0891) (0.0973) log(population Rep.) 0.468*** *** 0.157* *** *** *** (0.0489) (0.0557) (0.0757) (0.0842) (0.0520) (0.0644) (0.0819) (0.0556) (0.0453) (0.0518) log(population Part.) * ** *** ** (0.255) (0.277) (0.459) (0.470) (0.296) (0.352) (0.478) (0.384) (0.261) (0.409) log(distance) *** *** *** *** *** *** *** *** *** *** (0.0196) (0.0235) (0.0288) (0.0324) (0.0234) (0.0290) (0.0325) (0.0243) (0.0199) (0.0234) contiguity 0.585*** 1.188*** 0.378*** 0.396*** 0.529*** 0.383*** 0.710*** 0.500*** 0.495*** 0.617*** (0.0693) (0.0859) (0.0859) (0.0922) (0.0749) (0.0887) (0.0954) (0.0924) (0.0674) (0.0741) common language 0.698*** 0.530*** 0.502*** 0.755*** 0.692*** 0.458*** 0.651*** 0.544*** 0.540*** 0.905*** (0.0496) (0.0560) (0.0669) (0.0748) (0.0548) (0.0643) (0.0713) (0.0558) (0.0463) (0.0535) colony(1945) 0.596*** 1.177*** 0.641*** 0.426*** 1.136*** 1.151*** 1.559*** 0.901*** 0.989*** 1.141*** (0.106) (0.134) (0.119) (0.142) (0.126) (0.158) (0.139) (0.138) (0.135) (0.122) log(tariff_weighted) *** *** *** *** *** 1.213*** *** *** *** (0.190) (0.226) (0.311) (0.335) (0.262) (0.251) (0.327) (0.306) (0.213) (0.240) Constant *** *** ** *** ** ** *** (4.534) (4.695) (7.528) (7.636) (4.979) (5.828) (7.717) (6.110) (4.390) (6.352) Observations 20,881 19,122 17,764 17,241 19,944 18,958 17,142 20,132 20,981 19,431 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 The estimated tariff equivalents in general are higher than the corresponding tariffs. The overall tariff equivalent of NTBs for the EU-15 is at the level of 21% as compared to 23% in the US. This overall tariff equivalent is visibly higher in Poland (at 26%) and lower for the NMS (18%). However, the differences across countries and regions are in general not systematic, ie. one cannot clearly the same ordering of tariff equivalents across sectors. We also believe that there are at least two sectors that deserve extra comments.

12 Table 5. Calculation of tariff equivalents in merchandise trade Sector Reference Importer Effects (log) Tariff Equivalent country Ref. EU-15 NMS Poland USA Elasticity EU-15 NMS Poland USA Total MYS 1,3-0,2 0,1-0,5-0,3 7,1 21% 18% 26% 23% Agriculture VNM 3,1 1,6 1,4 0,1 1,0 4,8 31% 36% 33% 44% Mining AUS 2,7 0,7 0,0-0,9-0,2 11,9 17% 22% 30% 24% Food AUS 3,6 1,8 1,3 0,0 2,3 5,0 36% 46% 46% 25% Textiles CHL 1,7-0,3 0,1 0,1-0,6 7,5 27% 22% 21% 32% Apparel VNM 3,2 0,8 1,2 0,1 0,1 7,4 33% 27% 26% 42% Leather VNM 3,5 1,1 1,7-0,4 0,1 8,1 30% 23% 28% 43% Wood AUS 2,0 0,3 0,2-0,4 0,6 6,8 25% 27% 33% 21% Paper MYS 1,8-0,4-0,1-0,9-0,9 5,9 37% 32% 47% 46% Coal Petrol USA 1,5-1,6-2,3-1,8 1,5 4,2 73% 91% 133% 0% Chemicals CHL 0,5-0,8-0,6-0,2-1,0 6,6 19% 17% 19% 23% Minerals ZAF 1,2-0,8-0,3-0,3-0,4 5,8 35% 26% 32% 28% Steel MYS 1,9-0,5-0,8-0,5-0,3 5,9 40% 45% 54% 37% Metals THA 2,7-0,6-0,2-0,8-1,1 8,4 40% 35% 44% 46% Met. Prod. CHL 1,3-0,8 0,1-0,8-1,6 7,5 29% 16% 27% 39% Motor Veh. VNM 1,2-1,0-0,3-0,5-1,3 5,6 38% 26% 35% 44% Tr. Eq. Nec MYS 2,1 0,3 0,1 0,0-0,4 8,6 21% 23% 23% 29% Electric MYS 2,7-0,4 1,1-1,0-1,8 8,8 35% 18% 29% 51% Machinery VNM 1,3-1,0 0,0-0,8-1,8 8,1 27% 15% 25% 38% MnfcsNec THA 1,9 0,0 0,6-0,7 0,7 7,5 26% 17% 26% 15% Source: own calculations using Comtrade data. GTAP elasticities. Poland's fixed effect is the difference between NMS and Poland fixed effects First of all, we believe that the tariff equivalents for agricultural goods may be underestimated, i.e. trade in agricultural products is overall low across the world and it may be the case that even the reference country is trading relatively little. We also observer that in the estimated regression for that sector, the coefficient on tariff is not significantly different from zero. While the overall weighted tariff level in agriculture does not reflect the possibility of tariff peaks and therefore underestimates the effective tariffs, the same sort of argument can be made on the estimates of NTBs tariff equivalents. We also believe that flows in processed fuels (CoalPetrol) are largely driven by the allocation of natural resources and we believe that the ability of the gravity model in explaining these trade flows is limited. In that sector the US is the reference country but imports of processed fuels in the European countries are caeteris paribus low, leading to somewhat large tariff equivalents. Turning to trade in services, the coefficients on standard gravity variables are also in line with the intuition. It is quite surprising though that the coefficient on reporter GDP in the aggregate flow of services trade significantly exceeds one while in sectoral regressions it is only the business services where that happens. One has to note, however, that unlike the merchandise trade regressions, the overall regression is done on aggregate trade and not on

13 the pooled sample across sectors, therefore it also includes trade in services that is not necessarily included in the sectoral estimations. Moreover, services trade estimations are performed on intra and extra EU countries. It is also interesting, that in trade in services, distance has a visibly lower elasticity that in merchandise trade and the size of contiguity and language coefficient is also visibly lower. The EU dummy variable is significant only in some cases and it shows that EU membership increases bilateral services trade between partners by roughly 10% overall. However, in sectoral trade this coefficient is not always positive and significant. It is so in the case of trade (82% higher imports), transport (over 30%), financial services and non-market services. It is quite surprising to find a negative and significant EU dummy variable in the case of construction. Table 5. Gravity estimations for services trade VARIABLES Overall Construction Trade Transport FinServ Insurance BusNec RecOth NmktSvc log_gdp_rep 1.303*** 0.540*** 0.485*** 0.619*** 0.732*** 0.450*** 1.341*** 0.542*** 0.497*** (0.0422) (0.0900) (0.0776) (0.0505) (0.0847) (0.0772) (0.0597) (0.0737) (0.0632) log_gdp_par ** 1.720*** *** 0.679* 1.221*** (0.215) (0.459) (0.426) (0.306) (0.525) (0.468) (0.350) (0.396) (0.360) log_population_rep *** 0.297*** 0.213*** 0.172*** ** *** 0.264*** 0.313*** (0.0414) (0.0887) (0.0763) (0.0494) (0.0825) (0.0745) (0.0585) (0.0726) (0.0623) log_population_par *** *** * (0.539) (1.331) (1.228) (0.850) (1.347) (1.270) (0.944) (1.172) (1.019) log_dist *** *** *** *** *** *** *** *** *** (0.0196) (0.0374) (0.0353) (0.0276) (0.0384) (0.0357) (0.0294) (0.0304) (0.0297) contig 0.550*** 0.750*** 0.217** *** *** 0.535*** (0.0556) (0.0878) (0.0919) (0.0579) (0.0837) (0.0776) (0.0708) (0.0783) (0.0663) comlang_off 0.515*** *** 0.210*** 0.696*** 0.877*** 0.170*** 0.602*** 0.305*** (0.0401) (0.0923) (0.0807) (0.0670) (0.0815) (0.0794) (0.0646) (0.0698) (0.0673) col *** *** 0.712*** 0.466*** 0.822*** 0.775*** 0.890*** (0.0652) (0.166) (0.179) (0.0993) (0.172) (0.162) (0.137) (0.144) (0.104) eupair ** *** 0.821*** 0.345*** 0.251** ** (0.0476) (0.0947) (0.0865) (0.0808) (0.102) (0.0931) (0.0713) (0.0758) (0.0757) Constant *** (6.612) (19.88) (17.55) (11.87) (20.29) (19.72) (15.33) (17.48) (14.33) Observations 18,337 7,253 9,811 21,649 6,657 6,652 10,082 8,668 8,011 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 The estimated tariff equivalents are calculated with respect to the reference country. This time the reference countries are Thailand and Singapore, which is in line with previous estimates of trade barriers. The overall level of trade barriers in services is roughly twice as high as in the case of merchandise trade. Moreover, we also observer that in most cases imports of services are the most liberalized in the US (except construction) and the trade barriers are usually lower in the EU-15 than in NMS and Poland.

14 Table 6. Calculation of tariff equivalents in services trade Sector Reference Importer Effects (log) Tariff Equivalent country Ref. EU-15 NMS Poland USA Elasticity EU-15 NMS Poland USA Total THA 1,4-0,1-0,3-0,1 0,1 3,8 38% 43% 46% 33% Construction SGP -1,2-2,8-3,2 0,0-3,5 3,8 41% 53% 53% 62% Trade THA 2,4 0,5-0,6-0,5 1,1 3,8 48% 78% 91% 33% Transport SGP 0,3-1,6-2,6-0,1-1,0 3,8 50% 77% 80% 35% Financial Svcs SGP 1,9-0,2-0,7-0,1 1,5 3,8 55% 67% 69% 9% Insurance SGP 0,3-1,2-2,0-0,2-0,9 3,8 40% 58% 63% 30% Business Svcs SGP 1,3 0,1-0,2 0,2 0,3 3,8 32% 39% 34% 26% Nec Recreation & Other SGP 0,8-0,5-0,7-0,6 0,0 3,8 33% 39% 54% 21% Source: own calculations using Francois & Pindyuk data. GTAP elasticities. Poland's fixed effect is the difference between NMS and Poland fixed effects Conclusions The Transatlantic Trade and Investment Partnership is aimed inter alia at facilitating trade between the signing parties. However, due to overall significant progress in liberalizing trade in many manufacturing industries during in the framework of GATT and the WTO, especially in advanced economies, the further tariff liberalization is not expected to bring sizeable gains. Indeed we show that in most sectors overall tariff levels are below 10%. At the same time there is some anecdotic and empirical evidence that non-tariff barriers still matter a lot in international trade between advanced economies and are even more important in the services trade, where no tariffs per se exist. In order to assess the possible gains from trade liberalization through TTIP, we evaluate both the tariff and non-tariff barriers in the bilateral trade of the EU member and the United States. We differentiate between EU-15, the New Member States and Poland to account for their structural differences and different trading patterns. We use a standard gravity framework and attribute the differences in importer fixed effects to non-tariff barriers. We find that the level of non-tariff barriers trump in general trumps the one of the tariffs across different sectors and analyzed groups of countries. The overall tariff equivalent in merchandise trade is around 20%, at least twice the level of tariffs. However, there are sectoral differences and in many cases the NTBs exceed 40%. We also find the non-tariff barriers in services trade to be overall higher than those in merchandise trade. At the same

15 time, Poland and the remaining NMS exhibit higher barriers to trade in services than in the EU15 and the US. One has to keep in mind that these estimates are based on actual trade flows. Therefore a country that imports less, everything else equal, is going to have a higher estimated NTB tariff equivalent. However, when one thinks of the effects of NTB removal, countries with overall high NTBs have more potential for trade liberalization within TTIP. Keeping everything else equal, reaching the same target level of NTBs will give a higher trade boost in an initially more restrictive importer country. However, this is only true in percentage terms when one analyzes the volume of trade, the initially restrictive countries traded less initially and therefore from an aggregate point of view, the size of the response cannot be easily evaluated. This has to be taken into consideration when employing trade-based tariff equivalents in trade liberalization simulations non-linearities in response may matter (eg. in switching from prohibitory to non-prohibitory NTBs) and they are difficult to be implemented in simulation frameworks. References Anderson, James and van Wincoop, Eric, (2003), Gravity with Gravitas: A Solution to the Border Puzzle, American Economic Review, 93, issue 1, p Copenhagen Economics (2009), Assessment of barriers to trade and investment between the EU and Japan, report prepared for the European Commission, DG Trade Francois J., Pindyuk, O., "Consolidated Data on International Trade in Services," IIDE Discussion Papers , Institue for International and Development Economics. Francois, Joseph, Manchin, Miriam, Norberg, Hanna, Pindyuk, Olga and Tomberger, Patrick, (2013), Reducing Transatlantic Barriers to Trade and Investment: An Economic Assessment, No , IIDE Discussion Papers, Institue for International and Development Economics, Ghodsi, M., Hagemejer, J., Mach-Kwiecińska, A., 2015,The Implications of the Liberalization of Rail Services in Europe, in: Cieślik, A, Michałek, J.J. (eds) Liberalization of Transportation

16 Services in the EU: The Polish Perspective, Polish Studies in Economics, Volume 6, Peter Lang, Frankfurt Lionel Fontagné & Amélie Guillin & Cristina Mitaritonna, "Estimations of Tariff Equivalents for the Services Sectors," Working Papers , CEPII research center. Mayer, T., Zignago, S. (2011) Notes on CEPII s distances measures : the GeoDist Database CEPII Working Paper Park, S.-C. (2002). Measuring Tariff Equivalents in Cross-Border Trade in Services. Trade Working Papers 353, East Asian Bureau of Economic Research.