How trade restrictive are non tariff measures? New evidence

Transcription

1 Hw trade restrictive are nn tariff measures? New evidence Olivier Cadt Julien Gurdn Wrkshp CEPII-PSE 30 June,

2 NTMS: WHY SHOULD WE CARE? NTMs matter: Tariff reductins have shifted fcus n NTBs Reginal blcs which have successfully eliminated tariff barriers struggle with the eliminatin f NTBs (ex. EAC, COMESA, ASEAN) Harmnizatin/MRA issues may becme imprtant issues in trade negtiatins (ex. Transatlantic partnership) Plicy issues are cmplex Authrity ver NTM is fragmented (industry, health, ag. Ministries) Technical issues, easily captured by lbbies NTMs have cmplex effects n market structure (Asprilla et al. 2014) 2

3 NTMS: WHY SHOULD WE CARE? Sme unanswered questins Questin 1: Hw have NTMs evlved since last cmprehensive snapsht in Kee, Nicita and Olarreaga (2009, but based n 2001 TRAINS data)? Questin 2 (related): Regulatins (SPS & TBT) have prliferated. Hw restrictive are they? Hw respresentative is the aflatxin syndrme (Otsuki Wilsn Sewadeh 2001)? Questin 3: There is advcacy, at the WTO and in regins, fr harmnizatin and mutual recgnitin. Fragmentary evidence (Reyes 2011, Disdier et al. 2014, Chenn Matt 2008) suggests cmplex effects. D they make regulatins mre businessfriendly/less trade inhibiting? 3

4 PREVIEW OF OUR APPROACH/FINDINGS What we d 1. Use a new NTM database cvering 65 cuntries at HS6/MAST measure level 2. Estimate the impact f different three srts f NTM (SPS, TBT and Others) n trade unit values (i.e. estimate directly price-raising effects) 3. Interact NTMs with harmnizatin/mra dummies t estimate stringencyreductin effects What we d nt d 1. D nt seek cuntry-specific effects thrugh interactin terms as in Kee et al (using crss-cuntry estimates t prduce cuntry-specific numbers generates t much cnfusin) 2. D nt aggregate measures int an verall measure f prtectin (OTRI); assuming that all regulatins are trade barriers leads t wrng plicy advice What we find 1. Regulatins have lwer AVEs than Kee et al s cre NTBs (which included QRs): arund 10 percent n average 2. Significant stringency-reducing effect f Harmnizatin and MRA (AVEs cut by abut ne quarter) 4

5 NTMS DATA NTM data with ld classificatin circa 2001 used by Kee et al (2009) and publicly available n WITS fr 78 cuntries (cmes frm WTO ntificatin + ther surces) WTO members ntificatin (used by Disdier and Fntagne 2009) cllected fr several years and analyzed by UNCTAD (available fr 92 cuntries) Recent MAST data cllectin ( ) fllwing the 2009 classificatin fr 40 cuntries + EU (27 cuntries), available n WITS 5

6 WHAT ARE NTMS? The MAST nmenclature used in ur data 6

7 PREVALENCE OF NTMS Frequency Index by sectr (HS sectins): It makes sense Higher SPS cverage fr fd prducts Higher TBT index fr prcessed fd, textile chemical prducts and mt vehicles Imprtance f separating standards frm ther NTMs SPS TBT C: Pre- Shipment D: Cntingen t E: Quantity cntrl Live animals Vegetable prducts Fats and Oil Prcessed fd Minerals prducts Chemical prducts Rubber and Plastics Raw hide and skins Wd Paper Textile Ftwear Stne and Cement Base Metals Machinery Electrical Equipment Mtr Vehicles Optical & Medicals instruments Miscellaneus gds

8 EXPANSION OF HARMONIZATION AND RECOGNITION IN PTAS Budetta and Piermartini (2009): 5 majr integratin appraches: Mutual Recgnitin f Technical regulatins Harmnizatin f Technical regulatins Transparency requirements Mutual recgnitin f Cnfrmity assessment Harmnizatin f Cnfrmity assessment List 58 RTAs by prvisin, include prvisins n standards In ur sample, cncerns cuntry-pairs ut f Technical Regulatin Cnfrmity Assessment Mutual Recgnitin Harmnizatin Transparency requirements 21 Technical Regulatin Cnfrmity Assessment Mutual Recgnitin Harmnizatin Transparency requirements 926 8

9 EMPIRICAL APPROACHES SO FAR: QUANTITY-BASED Quantity-based apprach: Cmbine variatin in imprt flws with price elasticity f demand t retrieve ad-valrem equivalent (AVE) 1. Kee Nicita Olarreaga 2009 Use TRAINS 2001 Use crss-cuntry variatin in imprts (all rigins) + price elasticity f demand t retrieve AVEs NTMs aggregated int ne brad categry Use expnential transfrmatin t eliminate negative AVEs 2. Disdier Fntagné Mimuni (2008) Use WTO ntificatins Distinguish between TBT and SPS regulatins AVEs as in Kee et al (2009) 3. Beghin Disdier Marette van Tngeren (2011) Use TRAINS 2001 Use crss-cuntry variatin in imprts (all rigins) + ε t retrieve AVEs like KNO Cnsider trade-enhancing NTMs with mdel 9

10 EMPIRICAL APPROACHES SO FAR: PRICE-BASED Price-based apprach : Generalizatin f price-gap methd t retrieve AVEs directly Ferrantin (2006) Cmparisn f averages n a case-by-case basis Andriamananjara et al. (2004) Treatment-effect (prices treated by NTMs) Use dmestic prices Cadt and Gurdn (2014) Treatment-effect (prices treated by NTMs) Use Wrld Bank ICP Prblem: dmestic prices generally nt available EIU prices: reflect expatriate basket + expensive Wrld Bank ICP: aggregated categries + dd chice f prducts 10

11 USING CEPII TRADE UNIT VALUES Only prices bservable in abslute frm (nt index) + disaggregated: trade unit values, btained by dividing trade values by quantities. Prblems 1. Data is nisy, as custms typically mnitr imperfectly imprt quantities Data cleaned by CEPII team Nt a prblem as lng as measurement errrs are uncrrelated with ther regressrs n the right-hand side 2. Trade unit values d nt include intermediatin margins : prblematic in the presence f quantitative restrictins : if licenses are given t dmestic distributrs; TUV will nt reflect the shadw value f the licenses. Nt a majr prblem as mst quantitative restrictins have been eliminated But shuld keep it in mind when interpreting results fr NTMs ther than SPS and TBT 3. May nt reflect true cmpliance csts if n pass-thrugh Empirical evidence n pricing t market (Berman et al 2012) shw that prducers absrb 10 percent f trade cst variatins s pass n abut 90 percent 11

12 OUR APPROACH Sample structure Prduct-by-prduct estimatin at HS6 (5 000 regressins, as in KNO 2009) Unit f bservatin: rigin-destinatin (like a disaggregated crss-sectin gravity) Data Dep. variable: Trade unit values frm CEPII fr prduct p, between imprter d and exp Regressrs: NTMs A (SPS) and B (TBT) and ther frm new database (binary) Tariffs, bilateral cntrls & imprter-exprter fixed effects Estimatin equatin Estimatr OLS with fixed effects by exprter, imprter characteristics z, dyad characteristics x & tariff Instrumentatin (e.g. using NTMs f adjacent cuntries and KNO AVE) will cme later Btstrapping standard errrs f AVEs will cme later 12

13 RESULTS: CAVEAT 1. In calculating averages, we use cnventin that AVE = 0 fr jk pairs with NTM = 0; alternative is t cde thse as missing values 2. Negative AVEs in quantity estimatin can mean facilitating NTM shuld nt be eliminated; negative AVEs in price estimatin difficult t justify, we replace with zer 3. Sme AVEs suspiciusly large; squash them using hyperblic tangent functin (keep values largely unchanged up t 100%, squashes them t 100% tangent thereafter) 13

14 RESULTS: AVERAGES BY SECTION: IMPORTER CHARACTERISTICS Raw AVEs: Estimated AVEs ranging between 2% and 30% fr each type f NTMs Sum f AVEs ranging between 5% and 50% Tangent: Estimated AVEs ranging between 1% and 10% fr each type f NTMs Sum f AVEs ranging between 5% and 25% With Imprter characteristics Tangent Transfrmatin Cefficient SPS (A) TBT (B) Other (CDE) Ttal SPS (A) TBT (B) Other (CDE) Ttal Animals Vegetables Fats & ils Beverages & tbacc Minerals Chemicals Plastics Leather Wd prducts Paper Textile and clthing Ftwear Stne & glass Pearls Metals Machinery Vehicles Optical & med. Instr Arms Miscellaneus Wrk f Arts Average

15 NTM AVES SUBSTANTIAL COMPARED TO TARIFFS, BUT NOT PROHIBITIVE Plausible Relatively cnsistent acrss studies and methds In line with what private-sectr says in interviews Other NTMs (C, D,E) TBT (B) SPS (A)

16 RESULTS: INTERACTING WITH PTAS PROVISIONS Differentiate the effect f NTMs depending n integratin appraches n standards by interacting trade agreement characteristics with the presence f NTMs (SPS and TBT) where PTA d dentes thse trade agreement characteristics between cuntry and d. Alternative with mutual recgnitin r harmnizatin, f technical regulatin r f cnfrmity assessment, r transparency requirement 16

17 RESULTS : INTERACTING WITH PTAS PROVISIONS AVEs are significantly lwer when there exist a PTA n standard issues between tw partners PTA dealing with thse issues f technical regulatin cut the AVE by 24.8 percent AVE withut RTA AVE with RTA Change in AVE Cut in AVE (%) SPS (A) TBT (B) Ttal SPS (A) TBT (B) Ttal SPS (A) TBT (B) SPS (A) TBT (B) Ttal Animals Vegetables Fats & ils Beverages & tbacc Minerals Chemicals Plastics Leather Wd prducts Paper Textile and clthing Ftwear Stne & glass Pearls Metals Machinery Vehicles Optical & med. Instr Arms Miscellaneus Wrk f Arts Average

18 COMPARING RESULTS: DIFFERENT INTEGRATION APPROACHES Mutual recgnitin f technical regulatin appears t be the less efficient : surprising Fr cnfrmity assessment the mutual recgnitin apprach is the mst efficient Mutual Recgnitin f Technical regulatin Harmnizatin f Technical regulatin Cut in AVE (%) Cut in AVE (%) SPS (A) TBT (B) Ttal SPS (A) TBT (B) Ttal Mutual Recgnitin f Cnfrmity assessment Harmnizatin f Cnfrmity assessment Cut in AVE (%) Cut in AVE (%) SPS (A) TBT (B) Ttal SPS (A) TBT (B) Ttal Transparency requirement Cut in AVE (%) SPS (A) TBT (B) Ttal

19 PRELIMINARY CONCLUSION AND QUESTIONS REMAINING OPEN 1. Estimated AVEs frm price estimatin seem by and large plausible: NTMs increase ften trade unit values and NTBs shuld be a pririty fr trade negtiatrs 2. Shuld be used very cautiusly in plicy advice, since trade-restricting NTM can be gd plicy (human health/envirnment trumps trade, desn t it?). Rather, ur results suggest that NTMs measures deserve plicy attentin in terms f imprving design and implementatin 3. Implementing PTA with prvisins related t standards decrease cst f cmpliance t thse nn-tariff measures. 4. Mutual recgnitin f cnfrmity assessment (the easiest step tward crdinatin f standard related measures) appears t be the mst efficient integratin apprach t cut cmpliance cst t standards. 19