Global value chains at tariff line level

Transcription

1 Global value chains at tariff line level WORK IN PROGRESS Lars Nilsson September 2013 Abstract: The emergence and rapid growth of global value chains (GVCs) has become a main feature of the international economic integration process in recent years. The role played by GVCs is frequently assessed economy-wide and at broad sector level using international or global input-output tables containing data at a high level of aggregation. Much less is known about the interaction at lower levels of aggregation. Based on a unique and detailed dataset obtained from Iceland's customs authorities, this paper maps all tariff line level imports into Iceland originating in individual EU countries (the origin country) including also transactions via a (sometimes geographically distant) third country (the trading country). The paper compares the unit values of direct EU exports to Iceland with those via the most important trading countries and the value of these trade flows. Finally, the choice (or not) of trading country is tested empirically. Lars Nilsson is Deputy Head of the Chief Economist and Trade Analysis Unit in the European Commission's Directorate General for Trade (lars.nilsson@ec.europa.eu). The opinions expressed in this paper are the author's own and do not necessarily reflect any views of the European Commission.

2 1. Introduction With the launch of the World-Input-Output Database (WIOD) 1 in 2012 and the joint OECD WTO Trade in Value-Added (TiVA) 2 initiative (which made use of the WIOD), the concept of global value chains (GVCs) or trade in value-added has gained substantial traction over the last couple of years (though the concept of fragmented production goes back to the 1990s). As opposed to statistics on traditional (gross) trade, trade in value-added estimates the value of various intermediates (goods or services) by source. This is usually not reflected in the former statistical measure. Estimates on trade in value-added thus corrects for the double counting that is implicit in current gross flows of trade in those cases where the value of exported final products embody imported inputs used in the production process. The introduction of both WIOD and TiVA has thus provided a global view of the importance of GVCs and led to an improved understanding of several key trade policy issues. For example, the role played by imports as inputs for exports, the contribution of services trade for goods production and for providing more meaningful measures of bilateral trade balances. In addition, the benefits of comprehensive trade opening on a multilateral (and plurilateral) basis are also magnified. However, our understanding of GVCs at a more detailed level is still relatively scarce (WIOD and TiVa contain 35 and 18 product and service categories, respectively). This should be seen in contrast to the approximately 5000 products at 6-digit level of the Harmonised System (HS) or more at lower levels of aggregation. This paper examines GVCs at tariff line level. Based on a unique and detailed dataset obtained from Iceland's customs authorities, at the lowest level of aggregation, it maps all EU originating (the origin country) imports into Iceland as well as EU originating imports into Iceland via a third country (the trading country), see Section 2. Limited processing of the products must have taken place through the GVC logic in the trading country; otherwise the products would not be exported to Iceland via the trading country in the first place. The paper compares unit values and analyses the value of the trade flows of EU exports, via the most important trading countries, to Iceland (Section 3). Finally, the choice (or not) of trading country is tested empirically in Section 4, while Sections 5 summarises and concludes. 2. Overview of the data All 27 EU Member States (MS) export to Iceland, see Annex Table 1. 3 At 8-digit level, there are some 240,000 transactions. All EU countries are trading countries for other EU countries except for Romania and Malta. EU exports to Iceland from EU trading countries sum to close to 228,000 flows. About 60% (some 145,000 flows) of all export transactions are shipped directly from the origin country to Iceland without passing through a trading country The data is for 2011 and Croatia had thus not yet acceded to the EU.

3 The most common EU trading countries are Denmark, Germany, the UK and Sweden with between 30,000 and 50,000 transaction. Denmark accounts for somewhat more than 20% of all transactions and Germany for somewhat less, while shipments through the UK represents circa 16%. In terms of value of the flows, the percentage figure remains by and large the same. 4 In terms of non-eu trading countries, Norway with some 9,000 transactions, Switzerland with 1,617, the USA with 738 and the Other group of countries (including 33 countries spread across the world) at 177 transactions dominate. *** Examining the broad product category breakdown (see Annex Table 2), for the EU countries taken together, about 17% or 40,000 transactions take place in TDC11 (Textiles and textile articles) followed by TDC6 (Chemical products), TDC20 (Miscellaneous manufactures) TDC7 (Plastics) & rubber and TDC17 (Transportation equipment) all with shares of the total number of shipments above 10%. With some variations, these categories are also the most important for shipments via the trading countries. Germany, the UK and the USA show fairly high numbers of transactions in TDC17 (Transportation equipment) and Germany, Norway and the Other group of countries in TDC16 (Machinery & mechanical appliances). TDC13 (Articles of stone, plaster, cement, asbestos) is also of importance for export flows via Norway. At a lower level of disaggregation (see Annex Table 3), for the EU countries taken together, partly reflecting the above, most transactions take place in HS87 (Motor vehicles), HS39 (Plastics and rubber), HS94 (Furniture) and in HS61 (Women's apparel) and HS62 (Men's apparel), see the first column. Except for the UK, which displays some 10% of the number of transaction in HS33 (Essential oils and resinoids; perfumery, cosmetic or toilet preparations), these HS chapters figure in the top of those most frequently exported via EU trading countries. The pattern is slightly different for exports via non-eu countries as trading countries. Goods shipped to Iceland via Norway tend to fall into the following categories: HS 70 (Glass and glassware) at 13%, HS33 (Essential oils and resinoids; perfumery, cosmetic or toilet preparations) and HS34 (Soap, waxes etc.) at 12% and HS 84 (Machinery) with close to 11%. Further, about 30% of the number of all exports shipped to Iceland via Switzerland took place in HS33 (Essential oils and resinoids; perfumery, cosmetic or toilet preparations). Fewer transactions, about 140, took place in HS48 (Paper and paperboard; articles of paper pulp, of paper or of paperboard) but represented close to 35% of the value exported (not displayed in the table). The opposite holds for HS87 (Motor vehicles), which accounts for 144 transactions (about 9% of the total), but it only represents 0.2% of the value (not displayed in the table). 4 Germany, the UK and Denmark also rank highest in terms of the value of exports, though the German share of the value of EU exports is somewhat lower than its share of the number of transactions.

4 For the USA, the picture is relatively similar in terms of the most important products exported. Still, HS19 represent more than 36% of the value of EU exports via the USA to Iceland (but only 14 transactions are registered). In the case of HS48, relatively few transactions (22) make up for close to 15% of the value of goods exported (not displayed in the table). For HS87, 108 transactions or close to 20% of all transactions correspond to no more than 3.4% of the value of EU exports to Iceland via the USA (not displayed in the table), similar to the exports of this category of products via Switzerland. EU exports to Iceland via the group of Other countries distinguishes itself from EU exports through Switzerland and the USA by having most transactions (27 out of 208) and 25% of the value exported (not displayed in the table) in HS39 (Plastics) and 17 transactions in HS85 (Electronics). *** Annex Table 4 shows that the overall EU preference utilisation rate (PUR) for exports to Iceland is high at above 91%. It also shows that the PUR is higher for EU exports via Denmark, Sweden, Norway and Switzerland as trading countries, where the figure for exports via Switzerland stands out at 98.7%, compared to the UK, the USA and the group of Other trading countries. Interestingly, the overall EU PUR for direct exports to Iceland (89.9%) is lower than the PUR for total exports (91.2%). By originating EU country, one may note that the overall PUR of exports from the UK and Latvia are lower than for the other countries at about 80% and 70%, respectively, and that the PUR for Cyprus stands out at 26% only. This is explained by Cyprus exports to Iceland via the UK, for which preferences are hardly used at all. 5 On the other hand, direct Cyprus exports to Iceland have a PUR of more than 90% and goods shipped via Denmark has a PUR of 100%. Similarly, most countries display 100% or close to full preference utilisation rates for goods shipped via Switzerland. The only exceptions are for the Netherlands with as low as PUR of 16.8% and Spain with a PUR of 69.8%. The simple preferential margin for EU exports to Iceland is on average 10% (not displayed) varying between 9% in several countries such as Luxembourg, Czech Republic, Germany and the Slovak Republic to 14% for Cyprus (Women's and Men's apparel). 3. Unit values of EU exports to Iceland and the total value of the flows Product upgrading, that is a shift to the production of more sophisticated and higher-priced goods may, command a higher unit value and price and unit values are commonly used as proxies or surrogates for product quality (see e.g. IMF et al. (2009)). The shipment of exports to Iceland via trading countries could include such elements, even though the origin of the products remain the same that of the origin country, and therefore warrants a closer look. 5 The underlying goods exported via the UK are found in HS62, primarily in HS6202 (13 transactions) and in HS 6204 (31 transactions).

5 The overall unit value (UV) 6 of EU exports to Iceland equals 3.7 (see Annex Table 5), ranging from 1.8 in the case of Lithuania to close to 15 for Greece, Malta and Slovak Republic and above 50 for Cyprus. The UV for direct exports at 3.4 is the lowest in the sample indicating higher UVs for exports shipped through a trading country. Furthermore, UVs for exports to Iceland via non-eu trading countries are higher than the UVs for EU trading countries e.g.: Norway ( 5.2), Switzerland ( 4.1), the USA ( 4.9) and the Other group of countries ( 5.9), with the exception of Germany ( 4.4) which is above the Swiss figure. There are significant differences in origin country unit value of exports depending on the trading country. For example the UV of Lithuanian exports to Iceland via Switzerland is more than 220 (due to a high UV of HS62 (Men's apparel), not displayed) while the UVs of the same goods via Germany or the UK are lower at 16 and 45, respectively. The same by and large holds for Bulgaria, Hungary, Italy and Romania which also exhibit high UVs for exports of Mens's apparel to Iceland via Switzerland. At TDC section level TDC 18 (Instruments - measuring, musical) and TDC 14 (Pearls, (semi- ) precious stones & metals) stand out. The UV of the former totals 260 and the latter 124 but with variations depending on trading country reaching close to 680 for TDC 18 through Germany and above 1150 for exports of TDC 14 via the Other group of countries. In both cases Germany is the origin country (not displayed). 7 Several product categories show relatively low UVs across the trading countries, e.g. goods in TDC4 (Prepared foodstuffs), TDC6 (Chemical products), TDC7 (Plastics and rubber), TDC9 (Wood and wood products) and TDC10 (Wood pulp products). *** To assess the actual value added in the trading countries more information is needed. One source of potential information is data on the customs valuation (CV) of the goods included in the dataset. The CV includes items which cannot be computed in the FOB value of the goods e.g. the cost of packaging whether for labour or materials. 8 By comparing the difference between CV and the FOB value of goods exported directly with goods shipped via trading countries, one may obtain an indication of value added to the products if the CV for exports via third countries is relatively higher than corresponding CVs for direct exports. Bold and underlined figures in Annex Table 4 indicate where this is the case. In total, as trading countries only the UK and the USA are concerned. In particularly in case of the latter, to which all but a handful origin countries display figures indicating that more value has been added to the goods compared to direct exports. From the origin country perspective, CVs for 6 The unit values are defined as the FOB value divided by the quantity measure. Supplementary quantity measures are been ignored and shipments for which the net weight is listed as zero have been excluded from the sample. 7 The latter concerns shipment of one product only, (Jewellery, goldsmiths' and silversmiths' wares and other articles of silver, whether or not plated or clad with other precious metal). 8 See WTO (1994), Article 8.

6 exports via third countries are relatively higher than corresponding CVs for direct exports in case of Estonia, Finland, Germany and Spain. At TDC section level the same pattern holds in particular for TDC 14 (Pearls, (semi-) precious stones & metals) and practically all TDC sections sent via the UK (see Annex Table 6). This means that origin countries send goods in many TDC sections via the UK where value seems to be added but that this trade only comes from a handful of origin countries, c.f. the Annex Table 4. Going down to HS2 chapter level, the UK pattern from above is repeated, most HS2 chapters show ratios of customs valuation UVs over FOB UVs which are higher for the trading countries compared to the numbers for direct exports (see Annex Table 7). The same holds true to some extent also for the USA and for the Other group countries. Across HS2 chapters, HS71 (Precious stones and pearls) show indications of value added through most trading countries. *** Previous work has highlighted that many highly disaggregated trade flows are of a small value. Nilsson (2011a) showed that the preference utilisation rates for small trade flows are markedly lower than average utilisation rates. Nilsson (2011b) examined the issue closer and demonstrated empirically that preference utilisation rates on the EU market decrease with lower values of preferential imports. Nilsson and Dotter (2012) further argued that these results indicate that there are fixed costs associated with using preferences and as the relative importance of these decline, PURs generally increase. Similarly, the exporting origin countries in this sample will only send their goods to Iceland via trading countries if the gains from doing so are higher than the costs. It seems reasonable to assume that there are fixed costs associated with sending goods to Iceland for further (minor) processing via a trading country and that this cost ought to decrease in relative terms with the size of the transaction. At the same time, it should be recalled that the many transactions in HS87 shipped via Switzerland and the USA were of low values. Annex Table 7 examines the value of the direct export shipments to Iceland and compares the values with those of the exports through the trading countries. It shows that at aggregate level the average value of EU exports to Iceland is as high or higher for direct exports ( 1503) or exports via EU trading countries as for exports via non-eu trading countries. The average values of exports to Iceland via the USA are strikingly low at 472 which is half the value or less than for most other trading countries. This is explained by more than 60% of transactions to a value of 100 or less (not displayed) as opposed to 20% for the sample as a whole. There is one exception; the overall average value of EU exports to Iceland via Switzerland is much higher at 3745 than for any other trading country. The opposite explanation holds here; 43% of all transactions through Switzerland take place in products with an average value of more than 1,000.

7 At country level, the highest average value, 11,864, is found for Estonian direct exports to Iceland, which is explained by exports of HS48 (Paper and paperboard; articles of paper pulp, of paper or of paperboard) to an average value of 33,329 (not displayed). At the same time, several small transactions are present, particularly via the USA (the Netherlands, Slovenia, Lithuania and Romania) to value of some 25 or less. In the case of the Netherlands this is due to exports of goods (not displayed) in the category HS87 (Motor vehicles). 4. Empirical relevance of unit values and export value for the choice of trading country It is assumed that exporters ship goods via the trading countries if the value of doing so exceeds the associated unobservable cost (C). Thus the potential value of transhipment (PVOT) must be larger than the unobservable cost. Hence, we can model exporters' decision to either export directly to Iceland or to ship via a trading country a discrete choice model: 9 y = 1,2,3.. if PVOT C > 0 (exports via trading country) y = 0 if PVOT C <= 0 (direct exports) (1) With a number of different trading countries, estimating a multinomial logit model which determines the discrete outcome of exports via different sets of trading countries against direct exports seems appropriate: P(y = 1,2,3...) = Pr(y 1,2,3.. > 0 X) = F(Xb) (2) The dependent variables are categorical outcomes, that is, EU originating exports to Iceland via another EU country, Norway, Switzerland, the USA and the Other group of countries. The explanatory variables X consist of the value of the trade flows (FOBVALUE), FOB unit values (FOBUV), compared to FOB UVs, relatively higher CVs for exports via third countries than corresponding CVs for direct exports (CVUV/FOBUV) and the sum of distance, i.e. the distance from the origin country to the trading country and from the trading country to Iceland: The sub-indices ijk refer to EU originating exports from EU country i to Iceland (country k) via trading country j. Finally, α, β 1, β 2, β 3 and β 4 are parameters to be estimated. P(y = 1,2,3...) = α+ 1 ln(fobvalue ijk ) + 2 ln(fobuv ijk ) + 3 ln(cvuv/fobuv ij )+ 4 ln(totdist ijk ) + (5) The trade data comes from Iceland's customs authorities and is for 2011 and distances are from Cepii, see Mayer and Zignago (2011). *** Annex Table 9 presents the set of five regression results. Each regression predicts the probability of the set of trading countries being chosen, given the explanatory variables. 9 See Train, K (2009) for an overview of discrete choice methods.

8 Overall, the goodness of fit appears reasonable with pseudo R2 of about 0.30 across the regressions. The first regression puts all trading countries into one group and estimates the probability that EU countries will export to Iceland via these countries instead of exporting directly. The coefficient estimates for the value of the export transaction, unit value and the ratio of customs value UVs and FOB UVs between direct exports and indirect exports (proxy for value added in the trading countries) are all positive and significant at the 1%. However, the predicted coefficient of total distance is positive and statistically significant at the 1% level. Regression 2 splits all trading countries into EU27 (excluding direct exports) and the Rest. It shows that parameter estimate for the value of the transaction is positive and statistically significant for EU27, while it is insignificant for the Rest. The other estimates remain unchanged. In the 3 rd regression, the Rest is further split by putting Norway and Switzerland into a separate group of trading countries. The parameter estimate for transaction UV remains positive and statistically significant at the 1% level for EU27 but is insignificant for Norway and Switzerland and the Rest. In addition, the estimated coefficient for total distance remains positive and statistically significant at the 1% level for EU27 and for Norway and Switzerland but it is negative and statistically significant at the 1% level for the Rest indicating that distance matters for countries further away from the exporting origin countries. Regression 4 equals regression 3 but separates the USA from the Rest. Compared to the results in regression 3, one may note that the coefficient estimate of the value of the transaction is insignificant for the US, but that transaction UV is positive and statistically significant at the 10% level. In addition, the coefficient estimate for total distance is negative and statistically significant at the 1% level for the USA The last regression splits Norway from Switzerland so that there are five potential outcomes. What is noteworthy with these results is that the neither the transaction value nor the transaction unit value seem to be an important predictors for using Switzerland as trading country. The coefficient estimates for distance are positive and statistically significant at the 1% level for both Norway and Switzerland. Distance does not see to play a role for countries close to the exporting origin countries. 5. Summary and conclusions This paper examines a detailed dataset on Iceland imports from the EU which provides for trade data on so called origin country exports which may be shipped either directly to Iceland or through a trading country (which may be the same as the origin country). Most EU exports take place either directly or through other EU countries as trading countries. In terms of non- EU trading countries, Norway, Switzerland and the USA dominate but there are some additional 30 countries or so through which exports to Iceland also are routed. In an attempt to assess whether the use of trading countries reflects product upgrading, unit values of direct and indirect exports are examined. An ocular inspection points to that exports

9 via trading countries indeed are associated with higher unit values. On the other hand, assuming that there are fixed costs associated with sending goods to Iceland for further (minor) processing via a trading country and that this cost decreases in relative terms with the size of the transaction so that indirect exports ought to consist of higher value shipments are not directly supported by the descriptive data at the fairly high level of aggregation analysed. Employing a multinomial logit mode trying to determine the choice of trading country shows that some variables do indeed rather consistently turn out to be important predictors for the choice of trading country but that this depends on the trading country group. For example, the value of the transaction is a good predictor for EU exports to Iceland only via other EU countries, but not for the others, while the transaction unit value seems to positively influence the choice of EU27, Norway and the USA as trading countries. The parameter estimate for the proxy for value added in the trading countries (diff. in CV UV and FOB UV) is positive and statistically significant at the 1% level for all trading countries. However, distance seems to influence the choice of trading countries negatively only in the case of the USA and the Rest. For the countries close to the origin countries, other EU countries, Switzerland and Norway, the opposite holds. To conclude, this paper has shown that the determinants of global value chains seem to vary between countries in the value chain and hard generalised facts are difficult to establish. However, as expected, further processing in the trading countries is positively related to value added. Higher transaction unit values, to more limited extent, also seem to be related to further processing in trading countries before final exports to Iceland. Further analysis of trade flows at transaction level would most likely greatly benefit from industry/firm level knowledge. This could shed some light on trading structures which may be well known to industry practitioners but which are more difficult for desk researchers to access and learn about.

10 REFERENCES IMF, ILO, OECD, Eurostat, UNECE and the World Bank (2009), Export and import price index manual. Theory and practice, Washington, DC: IMF. Mayer, T and S. Zignago, Notes on CEPII's distances measures: The GeoDist database, CEPII Working Paper , December 2011, CEPII. Nilsson, L. (2011a). European Union Preferential Trading Arrangements: Evolution, Content and Use. In Salvatici, L and L. De Benedictis (eds.), The Trade Impact of European Union Preferential Policies: An Analysis through Gravity Models, Springer Publishers. Nilsson, L. (2011b). Small Trade Flows and Preference Utilisation: The Case of the European Union. South African Journal of Economics Vol. 79: 4 December. Nilsson, L and C. Dotter (2012). Small Flows, Compliance Costs and Trade Preferences: The Case of EU Imports from African LDCs. Economics: The Open-Access, Open- Assessment E-Journal, Vol. 6, Train, K (2009). Discrete Choice Methods with Simulation. Cambridge University Press, Second edition. WTO (1994), Agreement on Implementation of Article VII of The General Agreement on Tariffs and Trade,

11 ANNEX Table 1: Overview of the EU number of preference eligible export shipments destined for Iceland, by trading country (count) Most common trading countries EU origin country Total EU27 Direct exports Denmark Germany UK Sweden Norway Switzerland USA Other* Austria Belgium Bulgaria Cyprus Czech Republic Denmark dir. exp Estonia Finland France Germany dir. exp UK dir. exp Greece Netherlands Hungary Ireland Italy Latvia Lithuania Luxembourg Malta Poland Portugal Romania Slovak Republic Slovenia Spain Sweden dir. exp Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion, the Faraoe Islands but excluding Iceland itself.

12 Table 2: Overview of the EU number of export shipments destined for Iceland, by TDC section (count) TDC Section Total EU27 Direct exports Most common trading countries Denmark Germany UK Sweden Norway Switzerland USA Other* 04 Prepared foodstuffs Mineral products Chemical products Plastics & rubber Hides & skins, leather Wood & wood products Wood pulp products Textiles & textile articles Footwear, headgear Articles of stone, plaster, cement, asbestos Pearls, (semi-)precious stones & metals Base metals & articles thereof Machinery & mechanical applicances Transportation equipment Instruments - measuring, musical Arms & ammunition Miscellaneous Works of art Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion, the Faraoe Islands but excluding Iceland itself.

13 Table 3 Overview of the EU number of export shipments destined for Iceland, by HS2 Chapter (count) HS2 Chapter Total EU27 Direct exports Most common trading countries Denmark Germany UK Sweden Norway Switzerland USA Other* Sub-total Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself.

14 Table 4: Preference eligibility and preference utilisation rates of EU exports to Iceland by trading country (%) EU origin country Total EU27 Direct exports Preference utilisation in total and by trading country Denmark Germany UK Sweden Norway Switzerland USA Other* Share of preference eligibility Austria Belgium Bulgaria Cyprus Czech Republic Denmark dir.exp Estonia Finland France Germany dir.exp UK dir.exp Greece Netherlands Hungary Ireland Italy Latvia Lithuania Luxembourg Malta Poland Portugal Romania Slovak Republic Slovenia Spain Sweden dir.exp Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself.

15 Table 5: FOB unit values of EU exports to Iceland by origin country and trading country ( per quantity measure)** Most common trading countries EU origin country Total EU27 Direct exports Denmark Germany UK Sweden Norway Switzerland USA Other* Austria Belgium Bulgaria Cyprus Czech Republic Denmark Estonia Finland France Germany UK Greece Netherlands Hungary Ireland Italy Latvia Lithuania Luxembourg Malta Poland Portugal Romania Slovak Republic Slovenia Spain Sweden Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself. ** Supplementary quantities have been ignored and shipments for which the net weight is listed as zero have been excluded from the sample. Bold and underlined figures indicate where the ratio of customs valuation UVs over FOB UVs are higher compared to the numbers in the Direct exports column.

16 Table 6: FOB unit values of EU exports to Iceland by TDC section and trading country ( per quantity measure)** Most common trading countries TDC Section Total EU27 Direct exports Denmark Germany UK Sweden Norway Switzerland USA Other* 04 Prepared foodstuffs Mineral products Chemical products Plastics & rubber Hides & skins, leather Wood & wood products Wood pulp products Textiles & textile articles Footwear, headgear Articles of stone, plaster, cement, asbestos Pearls, (semi-)precious stones & metals Base metals & articles thereof Machinery & mechanical applicances Transportation equipment Instruments - measuring, musical Arms & ammunition Miscellaneous Works of art Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself. ** Supplementary quantities have been ignored and shipments for which the net weight is listed as zero have been excluded from the sample. Bold and underlined figures indicate where the ratio of customs valuation UVs over FOB UVs are higher compared to the numbers in the Direct exports column.

17 Table 7: FOB unit values of EU exports to Iceland by HS2 Chapter and trading country ( per quantity measure)** Most common trading countries HS2 Chapter Total EU27 Direct exports Denmark Germany UK Sweden Norway Switzerland USA Other* Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself. ** Supplementary quantities have been ignored and shipments for which the net weight is listed as zero have been excluded from the sample.

18 Table 8: Average FOB values of EU exports to Iceland by trading country ( ) EU origin country Total EU27 Direct exports Most common trading countries Denmark Germany UK Sweden Norway Switzerland USA Other* Austria Belgium Bulgaria Cyprus Czech Republic Denmark Estonia Finland France Germany UK Greece Netherlands Hungary Ireland Italy Latvia Lithuania Luxembourg Malta Poland Portugal Romania Slovak Republic Slovenia Spain Sweden Sum Source: Own calculations based on data from Iceland s Customs Authorities. Note: * Includes China, The Faroe Islands, Japan, Canada, Australia, Turkey, Liechtenstein, Chile, Mexico, Montenegro, Greenland, Pakistan, Sierra Leone, Iran, Ecuador, United Arab Emirates, New Zealand, Georgia, India, Ethiopia, Gabon, Thailand, Fiji, Russia, Nicaragua, South Africa, Niger, Brazil, Israel, Mauritius, Philippines, Reunion and the Faraoe Islands but excluding Iceland itself.

19 Table 9: Regression result, choice of trading country Regr. No. Trading country outcome: (compared to direct exports) Value of transaction Transaction UV Diff. in CV UV and FOB UV Total distance Constant Pseudo R2 No of obs 1 All 0.031*** 0.114*** 0.142*** 3.253*** *** EU *** 0.121*** 0.156*** 3.472*** *** Rest *** 0.068*** 2.205*** *** 0.28 EU *** 0.122*** 0.151*** 3.604*** *** 3 Norway and Switzerland *** 0.053*** 2.776*** *** 0.30 Rest ** 0.265*** *** 7.430*** EU *** 0.122*** 0.151*** 3.604*** *** 4 Norway and Switzerland *** 0.053*** 2.776*** *** USA * 0.254*** *** 7.029*** Rest *** *** 5.904*** EU *** 0.120*** 0.151*** 3.615*** *** Norway 0.026* 0.103*** 0.056*** 2.539*** *** 5 Switzerland *** 4.355*** *** 0.30 USA * 0.254*** *** 7.027*** Source: Own calculations Rest *** *** 5.902***

20