Empirical Methods in Trade:Analyzing Trade Costs and Trade FacilitationJune 2015Bangkok, ThailandCosimo Beverelli(ERSD/WTO)Simon Neumueller(ERSD/WTO)1

Content (I)a.Measuring trade costsb.Descriptivesc.Deriving trade costs - Novy Method2

a. Measuring trade costs Trade costs include all costs include all costs from producer until finalconsumer transportation costs are for example tariffs and non-tariff measuresinformation costscustoms fees and chargesthe cost of time Some are easier to measure, think fees vs the cost of time Direct measures: Data on fees or transport charges Indirect measures: Inferred data from price differences3

b. Descriptives Arvis et al. (2013) find that trade costs are equivalent to 297 per cent advalorem tariff on international trade4

Descriptives – Components Font size indicates importance of trade cost components, the bigger themore important (based on Arvis et al. (2013))Production value of goodTrade costs297% ofproductionvalueDistance andbordersTariffsCultureCurrencyOther policycostsTrade facilitation(logistics) andconnectivity5

Descriptives – Sectoral differences Trade costs are not the same across different sectors, think perishable,agricultural goods vs natural resources such as crude oil In 2012, trade costs were 68% higher for agricultural goods compared tomanufacturing Also variance differs, 11 times higher in manufacturing than inagricultural goods, with respect to trade facilitation6

Descriptives – Global value chains 4th global review of Aid for Trade (WTO 2013) finds that biggest obstaclesfor developing countries to enter global value chains are: Customs proceduresTransportation costsTransportation delaysSource: Nordås et al. (2006)7

Descriptives – Sectoral differences (Zaki (2015))8

Descriptives – Geographical patterns9

Descriptives – Geographical patterns10

b. Deriving trade costs - Novy method Trade has increased considerably in the last decade One reason is the reduction in transport costs, but which trade costsdecreased the most? We know very little about them. Trade costs are generally difficult to measure, hence: Derive trade costs from comparing the levels of trade flows Comprehensive measure of trade costs, well grounded in theory Measure captures transport costs, tariffs, language barriers, red tape, 11

Novy method ct’d The gravity equation can be used in reverse to measure bilateral tradecosts and to decompose trade costs into a tariff and non-tariffcomponent (see for instance Jacks et al., 2011 and ESCAP/WB database)a. Theoretical approachTake expression (10) from the gravity slides and write it down for Xij, Xji,Xii and Xjj, where Xii and Xjj are the expressions for intra-national tradeRearrange trade costs1 X ii X jj 1 tii t jj X ij X ji tij t ji (1)Tariff equivalent of bilateral trade costs relative to domestic trade costscan then be expressed as geometric average of trade barriers in both1directions X ii X jj 2 1 tij t ji(2) 1 1 X X tii t jj ij ji 12

a.Theoretical approach (ct’d) Expression (1) is a derivation of overall trade costs from the gravityequation without imposing a cost function It is neither assumed that domestic trade costs are zero, nor that theyare the same across countries (tii may differ from tjj) nor that bilateraltrade costs are symmetric (tji may differ from tij) Under the specific assumption that domestic trade costs are zero andbilateral trade costs symmetric (as implied by taking the geometricaverage as a measure of bilateral trade costs), it is also possible todecompose overall trade costs in their various cost components byassuming an arbitrary trade cost function13

a.Theoretical approach (ct’d) For example, it is possible to decompose overall trade costs in their tariffand non-tariff component by simply estimatingln( ij ) 1 ln( distij ) 2 ln( tariff ij ) 3 ln( NTBij ) ijwhere NTB is a dummy To compute the tariff equivalent of a quota, we only need to calculatewhat is the percentage change in tariff that has the same impact ontrade costs as a quota Tariff equivalent exp(δ3/δ2) - 114

b.Empirical implementation The difficulty in calculating (1) is to get figures for intra-national tradeOne approach is to estimate these figures as the difference betweenproduction and exports (see for instance Novy 2012) The use of GDP instead of production data tends to overstate intra-nationaltrade, and therefore trade costs, because a growing share of trade is services(largely non-tradable) One can use the TPP datasets (CEPII)See Exercise 2 of Chapter 3 of the Practical Guide to Trade Policy Analysis15

the cost of time Some are easier to measure, think fees vs the cost of time Direct measures: Data on fees or transport charges . See Exercise 2 of Chapter 3 of the Practical Guide to Trade Policy Analysis 15. Title: Slide 1 Author: Owner Created Date: