Renegotiation of Trade Agreements and Firm Exporting Decisions: Evidence from the Impact of Brexit on UK Exports Meredith A. Crowley Oliver Exton Lu Han University of Cambridge July 2018
Disclaimer This work contains statistical data from HMRC which is Crown Copyright. The research datasets used may not exactly reproduce HMRC aggregates. The use of HMRC statistical data in this work does not imply the endorsement of HMRC in relation to the interpretation or analysis of the information.
Renegotiation of trade agreements Countries regularly negotiate new, more liberal trade agreements: Canada-EU, CP-TPP, KORUS Recently, several trade agreements have announced renegotiation: UK in EU, US in NAFTA, KORUS with threats that collapse of negotiations will lead to tariff increases. Research question: What is the impact of the renegotiation of a trading relationship (and its implied uncertainty about future trade policy) on firm entry into and exit from exporting?
Summary of results UK firm entry into the EU fell dramatically for products that would face increases in tariffs if negotiations breakdown. A 10 percentage point increase in the threat point tariff decreases the entry rate by 10.1 ppt and increases exit rate by 4.9 ppt. Products facing extreme threat point tariffs experience a 25.3 ppt decline in relative entry rate. Products facing high threat point tariffs experience a 12.3 ppt decline in relative entry rate. Partial equilibrium aggregation: entry into (exit from) the EU by UK firms would have been 5.1% higher (4.3% lower) in 2016 if there had been a guarantee that EU import tariffs on UK exports would remain at zero post-brexit.
Theory and empirics on trade policy uncertainty Trade policy uncertainty Handley (2014), Handley and Limão (2015), Pierce and Schott (2016), Handley and Limão (2017), Crowley, Meng and Song (2018). Trade agreement design and renegotiation Bagwell and Staiger (1999), Maggi and Rodgriguez-Clare (2007), Horn, Maggi and Staiger (2010), Maggi and Staiger (2011), Maggi and Staiger (2015), Limão and Maggi (2015).
Outline Introduction Theoretical motivation The Brexit referendum Empirical model Data Results Conclusion
Firm exporting decisions under trade policy uncertainty Model from Handley and Limão (2017) Consider the problem of a firm with marginal production costs c that must decide whether or not to enter into exporting to a foreign market with its product h under tariff τ sh in policy state s with fixed cost of entry K h. The per-period operating profits are: π(τ sh ). A firm enters if its production costs are below the cost threshold c c certain sh, where c certain sh is given by: ( ) 1 c certain a sh σ 1 sh =. (1 β)k h For the marginal firm, with c = c certain sh, the discounted value of profits exactly equals the sunk cost to enter into exporting: π(τ sh, c certain sh ) = K h. 1 β
Trade policy uncertainty The probability that two countries choose to renegotiate the terms of their trading relationship is γ. Renegotiation (s=r): A state with uncertainty about the eventual outcome of renegotiations. After renegotiation, two outcome states are possible: Free Trade (s=ft): Outcome state with zero tariffs on all products. Probability: λ FT = 1 λ WTO Threat point tariffs (s=wto): Outcome state with threat point tariffs on all products. Probability: λ WTO
Firm decisions under trade policy uncertainty Renegotiation regime: Firm solves optimal stopping problem. Value of entering into exporting: Π e (τ R, c) = π(τ R, c) + β [ γ ( λ WTO Π e (τ WTO, c) + (1 λ WTO )Π e (τ FT, c) ) +(1 γ)π e (τ R, c) ] Value of waiting: Π w (τ R, c) = 0 + β [ γ ( λ WTO Π w (τ WTO, c) + (1 λ WTO ) max {Π e (τ FT, c) K, 0} ) +(1 γ)π w (τ R, c) ] A marginal firm with c = c U R,h just satisfies entry condition: Π w (τ R,h, c U R,h ) = Π e(τ R,h, c U R,h ) K h Uncertainty factor
Theoretical results and mechanisms Result: Entry rate is lower in the uncertain renegotiation regime in which tariff increases are possible, τ WTO,h > τ R,h and u(γ) > 0. Implications of trade policy uncertainty: 1. Products that face higher threat point tariffs τ WTO,h have lower cost cutoffs (so higher cost firms do not enter). [cross-sectional] 2. Increased probability of entering into renegotiation γ lowers the cost cutoff (so higher cost firms do not enter). [inter-temporal] 3. Increased probability that negotiations break down λ WTO lowers the cost cutoff (so higher cost firms do not enter). [inter-temporal] Endogenous exit
Brexit A switch to a less certain trade policy environment On 23rd June 2016 the British electorate voted to Leave the European Union: Government announced that UK would leave Customs Union. UK began negotiations with the EU with aims to ensure continued tariff free access to EU markets. Outside option or threat point (if no deal is reached) is for UK-EU bilateral trade to be conducted under WTO rules. Under WTO rules, UK exports to EU would be charged EU s external tariff rates (defined by EU s WTO tariff schedule). Brexit can be interpreted as an unexpected increase to the probability of a new set of tariffs (γ = 0.3 γ = 1).
Brexit vote Pre-referendum, the probability of renegotiating the UK-EU trade relationships was 30%. Post-referendum, certain. Market implied probability that the UK would leave the EU 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Brexit referendum 23rd June 2016
Empirical predictions for UK based firms Application of Brexit to Handley and Limão (2017) Entry Products facing higher threat point tariffs will experience decreased entry during renegotiation relative to products facing lower threat point tariffs. Exit Products facing higher threat point tariffs will experience increased exit during renegotiation relative to products facing lower threat point tariffs. Exporters Products facing higher threat point tariffs will experience fall in number of exporters during renegotiation relative to products facing lower threat point tariffs.
Entry Empirical model Exit Exporters threat point entrants h,t = b 0 + b 1 τh,t + η h,t threat point exiters h,t = b 0 + b 1 τh,t + η h,t threat point exporters h,t = b 0 + b 1 τh,t + η h,t Growth measure: Davis and Haltiwanger (1992) Y h,t = 2(Y h,t Y h,t 1 ) (Y h,t + Y h,t 1 ) where Y ht is the growth in the aggregate of outcome variable Y (exporters, entrants, exiters) for product h in time t.
UK customs data Information on the universe of firm exporting decisions is obtained from the UK HMRC Overseas Trade Statistics databases. Table 1: Total value and number of UK-based firms exporting to EU Value ( bn) Firms Firm-productss 2013 145 21,301 334,095 2014 141 20,918 348,872 2015 128 21,124 366,169 2016 140 21,103 384,044
Definition of entry into and exit from exporting Empirical analysis conducted at the firm-product level. Entry is defined as an export of product h by a firm f to the EU at time t after no observed export at time t 1 of that product h by firm f. Exit is defined equivalently as a firm-product export in year t 1, but not t. Table 2: UK exporters, entrants and exiters to the EU at the firm-product level Exporters Entrants (1 year) Exiters (1 year) 2013 334,095 96,510 85,996 2014 348,872 99,340 84,563 2015 366,169 102,792 85,495 2016 384,044 107,436 89,561
Tariff rates under EU WTO commitments Threat point tariffs that UK-based firms face on exports to the EU are defined by the EU s WTO commitments and WTO tariff schedule. Classify the 8,500 products exported from the UK to the EU into discrete tariff categories. Table 3: Discrete tariff categories Category Tariff Number of firm-products UK export value to EU Quota non ad-valorem 4.8 % 4.3% Extreme 15% 1.9% 1.0% High 10 15% 12.2% 12.0% Medium 5 10% 21.4% 13.7% Low 0 5% 39.3% 29.4% Zero 0 20.6% 39.6 %
Tariff exposure across product categories of UK-EU exports Number of Firm-product Exporters 0 20,000 40,000 60,000 80,000 ANIM VEGE FOOD MINE CHEM PLAS HIDE WOOD TEXT FOOT STON META MACH TRAN MISC Extreme High Medium Low Zero Figure 1: No. of UK firm-products in 2015 exporting to EU by HS industry and exposure to EU tariffs
Tariff exposure across product categories of UK-EU exports Value of exports ( bn) 0 5 10 15 20 25 ANIM VEGE FOOD MINE CHEM PLAS HIDE WOOD TEXT FOOT STON META MACH TRAN MISC Extreme High Medium Low Zero Figure 2: Value of exports in 2015 to EU by HS industry and exposure to EU tariffs
Main findings Products facing higher threat point tariff rates experience reduced entry increased exit relative to industries facing low or zero tariff increases. Results are robust: Control for industry sensitivity to exchange rate. Control for supply shocks/global demand shifts. Timing of the Referendum.
Analysis of UK based firms: Tariff rate Table 4: Entry into/ exit from exporting to the EU: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point tariff rate -0.00155** -0.0105*** 0.00490** (0.000668) (0.00225) (0.00219) Constant 0.0443*** 0.0858*** -0.0236 (0.00456) (0.0146) (0.0144) Observations 7,665 7,436 7,336 R-squared 0.001 0.004 0.001 Interpretation: Products facing a 10 ppt threat point tariff will have 10.5ppt lower growth of entry into exporting than the 8.6% growth of entry in the zero tariff group, implying an overall growth rate of -1.9%.
Analysis of UK based firms: Discrete tariff categories Including quotas Table 5: Entry into/ exit from exporting to the EU: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point quota -0.0376*** -0.234*** 0.0869** (0.0121) (0.0425) (0.0436) Extreme threat point tariffs -0.0509*** -0.253*** 0.0795 (0.0184) (0.0648) (0.0659) High threat point tariffs -0.00109-0.123*** 0.100*** (0.0107) (0.0335) (0.0338) Medium threat point tariffs 0.00905-0.0288 0.0393 (0.00893) (0.0288) (0.0294) Low threat point tariffs -0.00348-0.0616** 0.0606** (0.00854) (0.0268) (0.0265) Constant 0.0383*** 0.0871*** -0.0437** (0.00671) (0.0216) (0.0214) Observations 8,520 8,231 8,131 R-squared 0.003 0.007 0.001 Interpretation: Products facing the extreme threat point tariffs experienced a 25.3 ppt decline in relative entry rate, compared to 8.7% growth of products facing no increase in threat point tariffs.
Partial equilibrium aggregation exercise Firm-product entry in 2016 would have been 5.1% if UK firms were guaranteed zero tariffs on exports to EU. Use regression coefficients and firm-product exporter statistics. Aggregating over categories finds 5,221 firm-products did not enter into the EU in 2016 relative to counterfactual. Using the average value of entrants (exporters) in 2015, reduced entry accounts for a 226 million ( 1.4bn) loss of export value in 2016. [Aggregate exports to the EU were 140bn in 2016.] Number of exiters would have been 4.3% lower in counterfactual with a value of 182 million ( 1.1 bn).
Analysis of UK based firms: Tariff rate Broader definitions of entry Table 6: Entry into exporting to the EU: 2016 relative to 2015 (1) (2) (3) Entrants (1 year) Entrants (2 year) Entrants (3 year) Threat point tariff rate -0.0105*** -0.0116*** -0.0118*** (0.00225) (0.00240) (0.00246) Constant 0.0858*** 0.0848*** 0.0885*** (0.0146) (0.0156) (0.0160) Observations 7,436 7,362 7,310 R-squared 0.004 0.004 0.004 Increasing the strictness in definition of entry - removing possibility of re-entry - implies a slight increase in magnitude of baseline result. Discrete categories
Controlling for exchange rate sensitivity Immediately after the Brexit vote, sterling depreciated by 15% and reached its lowest value since 1985. Empirical strategy implemented in two steps: 1. Estimate exchange rate sensitivity at HS02 sectoral level k: z hfd uv k hfdt = αk e z hfd e dt + z hfd X dt αk x + z hfd ɛ k hfdt 2. Second, use estimated α k e as control in regression: Y EU ht = b 0 +b 1 τ threat point h,t +b 2 Sensitivity to exchange rate k +η ht
Analysis of UK based firms: Tariff rate Control for industry exchange rate sensitivity Table 7: Entry into/ exit from exporting to the EU: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point tariff rate -0.00173** -0.0103*** 0.00575*** (0.000678) (0.00230) (0.00223) Sensitivity to exchange rate 0.0241-0.0240-0.114* (0.0184) (0.0589) (0.0613) Constant 0.0404*** 0.0898*** -0.00484 (0.00561) (0.0178) (0.0180) Observations 7,665 7,436 7,336 R-squared 0.001 0.004 0.002 Controlling for exchange rate sensitivity has no substantive impact on entry. Industries more sensitive to exchange rate experienced decrease in exit following large depreciation. Discrete categories
Impact on firm exporting decisions to EU relative to non-eu Control for product-level demand and/or cost shocks Threat point tariffs to the non-eu were unchanged in June 2016, τ non EU h,t = τ non EU threat point,non EU h,t+1 τh,t = 0. Product-level supply shocks S ht could affect entry, but would affect EU and non-eu markets equally, Sht EU = S non EU ht. Y EU ht Y non EU ht threat point,eu = b 0 + b 1 (τh,t + b s (S EU ht S non EU ht ) + η ht threat point,non EU τh,t ) Y EU ht Y non EU ht threat point,eu = b 0 + b 1 τh,t + η ht
Analysis of UK based firms: Triple difference Control for product-level demand and/or cost shocks Table 8: Entry into /exit from EU relative to the non-eu: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point tariff rate -0.00447*** -0.0116*** 0.00741*** (0.00121) (0.00289) (0.00277) Constant 0.0422*** 0.0801*** -0.0369** (0.00741) (0.0174) (0.0172) Observations 7,198 7,006 6,890 R-squared 0.002 0.003 0.001 Increase in magnitude of coefficients: possibly suggests extensive margin trade deflection into more certain markets. Discrete categories
Analysis of UK based firms: Tariff rate Robustness check: Half year entry post-referendum Table 9: Entry/exit exporting to the EU: July-Dec 2016 relative to July-Dec 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point tariff rate -0.00145* -0.0108*** 0.000536 (0.000741) (0.00238) (0.00236) Constant 0.0494*** 0.111*** 0.0131 (0.00489) (0.0154) (0.0156) Observations 7,521 7,270 7,146 R-squared 0.001 0.004 0.000 Similar results found to whole year estimation. Discrete categories
Analysis of UK based firms: Tariff rate Placebo test: half year entry pre-referendum Table 10: Entry/exit exporting to the EU: Jan-June 2016 relative to Jan-June 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Threat point tariff rate 0.000246-0.00360 0.00302 (0.000701) (0.00231) (0.00234) Constant 0.0235*** 0.0212 0.00560 (0.00473) (0.0152) (0.0153) Observations 7,505 7,231 7,141 R-squared 0.000 0.000 0.000 Placebo test shows that there was no impact in first half of 2016 before the referendum and increase in trade policy uncertainty. Discrete categories
Conclusion This paper estimates the impact of the announcement of a trade agreement renegotiation on firm exporting decisions. We exploit the natural experiment of the leave vote of the Brexit referendum on 23rd June 2016. Firm entry into (exit from) the EU significantly declines (increases) for those products facing higher threat point tariffs under renegotiation regime, even with no change in applied tariffs. Magnitudes are economically significant indicating a rapid response - aggregate entry would have been 5.1% higher and exit 4.3% lower in 2016 if UK firms were guaranteed zero tariffs on exports to EU, despite only 6 months of post-referendum exposure.
Appendix
Model assumptions Model from Handley and Limão (2017): Consumers have CES preferences over differentiated goods. Continuum of monopolistically competitive firms producing varieties v in industry h with heterogeneous productivity 1/c. Consumer prices p v include an ad valorem tariff, τ sh = 1 + t sh, which differs across industries h in policy states s. Economic conditions summarized by a sh = f (τ sh, E h, P h ) Firms form rational expectations over future expected profits with a discount factor β = (1 δ)(1 + R). Firms make entry decision into exporting in state s to maximize expected profits net of sunk entry cost K h.
Firm decisions under trade policy uncertainty Entry cost cutoff for uncertain state, c U R,h, is proportional to the deterministic cutoff, c certain R,h, by an uncertainty factor U(ω h, γ): c U R,h /ccertain R,h = U(ω h, γ) = where γ is the probability of a new set of tariffs. ( ) 1 1 + u(γ)ωh σ 1 1 + u(γ) ω h = (τ WTO,h /τ R,h ) σ is ratio of operating profits in high tariff state relative to uncertain renegotiation state. u(γ) = γλ WTO β/(1 β) is expected spell in high tariff state. TPU
Endogenous exit Endogenous exit is not explicitly modelled as no per period fixed cost to export, only sunk entry cost K h. Firms that exogenously exit due to death shock have opportunity to re-enter and not appear as Exiters. Following a trade policy regime shift that increases uncertainty, the entry cutoff falls c1h U / γ < 0. Therefore firms with ct,1h U < c cu t+1,1h will enter before the trade policy regime switch and not re-enter if hit by exogenous death shock after the trade policy regime switch.
Analysis of UK based firms: Discrete tariff categories Table 11: Entry into/ exit from exporting to the EU: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Extreme threat point tariffs -0.0509*** -0.253*** 0.0795 (0.0184) (0.0648) (0.0659) High threat point tariffs -0.00109-0.123*** 0.100*** (0.0107) (0.0335) (0.0338) Medium threat point tariffs 0.00905-0.0288 0.0393 (0.00893) (0.0288) (0.0294) Low threat point tariffs -0.00348-0.0616** 0.0606** (0.00854) (0.0268) (0.0265) Constant 0.0383*** 0.0871*** -0.0437** (0.00671) (0.0216) (0.0214) Observations 7,665 7,436 7,336 R-squared 0.002 0.004 0.001
Analysis of UK based firms: Discrete tariff categories Broader definitions of entry Table 12: Entry into exporting to the EU: 2016 relative to 2015 (1) (2) (3) Entrants (1 year) Entrants (2 year) Entrants (3 year) Extreme threat point tariffs -0.253*** -0.296*** -0.303*** (0.0648) (0.0696) (0.0712) High threat point tariffs -0.123*** -0.126*** -0.124*** (0.0335) (0.0360) (0.0373) Medium threat point tariffs -0.0288-0.0183-0.0196 (0.0288) (0.0309) (0.0320) Low threat point tariffs -0.0616** -0.0452-0.0500* (0.0268) (0.0288) (0.0295) Constant 0.0871*** 0.0748*** 0.0797*** (0.0216) (0.0231) (0.0238) Observations 7,436 7,362 7,310 R-squared 0.004 0.005 0.005 Main results
Analysis of UK based firms: Discrete tariff categories With control for industry exchange rate sensitivity Table 13: Entry into/ exit from exporting to the EU: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Extreme threat point tariffs -0.0526*** -0.249*** 0.0918 (0.0184) (0.0650) (0.0659) High threat point tariffs -0.00389-0.117*** 0.121*** (0.0110) (0.0346) (0.0351) Medium threat point tariffs 0.00699-0.0247 0.0542* (0.00906) (0.0293) (0.0298) Low threat point tariffs -0.00477-0.0589** 0.0702*** (0.00862) (0.0270) (0.0266) Sensitivity to exchange rate 0.0172-0.0349-0.126** (0.0187) (0.0596) (0.0620) Constant 0.0362*** 0.0914*** -0.0280 (0.00722) (0.0232) (0.0234) Observations 7,665 7,436 7,336 R-squared 0.002 0.004 0.002 Tariff rate
Analysis of UK based firms: Discrete tariff categories Controlling for product specific shocks Table 14: Entry into/ exit from exporting to the EU relative to the non-eu: 2016 relative to 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Extreme threat point tariffs -0.0943** -0.221** 0.205** (0.0396) (0.0966) (0.0966) High threat point tariffs -0.0632*** -0.188*** 0.0796* (0.0192) (0.0422) (0.0431) Medium threat point tariffs -0.00987-0.0629* 0.0490 (0.0149) (0.0350) (0.0358) Low threat point tariffs -0.0207-0.0982*** 0.0705** (0.0135) (0.0312) (0.0315) Constant 0.0406*** 0.103*** -0.0547** (0.0108) (0.0251) (0.0257) Observations 7,198 7,006 6,890 R-squared 0.002 0.004 0.002 Tariff rate
Analysis of UK based firms: Discrete tariff categories Half year entry post-referendum Table 15: Entry/exit exporting to the EU: July-Dec 2016 relative to July-Dec 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Extreme threat point tariffs -0.0424** -0.264*** -0.0911 (0.0202) (0.0663) (0.0686) High threat point tariffs -0.00450-0.130*** 0.152*** (0.0118) (0.0356) (0.0379) Medium threat point tariffs 0.00230-0.0324 0.0913*** (0.00970) (0.0306) (0.0316) Low threat point tariffs -0.00855-0.0653** 0.117*** (0.00907) (0.0282) (0.0290) Constant 0.0472*** 0.114*** -0.0577** (0.00726) (0.0227) (0.0234) Observations 7,521 7,270 7,146 R-squared 0.001 0.004 0.005 Tariff rate
Analysis of UK based firms: Discrete tariff categories Placebo test - half year entry pre-referendum Table 16: Entry/exit exporting to the EU: Jan-June 2016 relative to Jan-June 2015 (1) (2) (3) Stock of exporters Entrants (1 year) Exiters (1 year) Extreme threat point tariffs -0.0177-0.0901 0.0700 (0.0199) (0.0671) (0.0704) High threat point tariffs 0.0293** -0.0398 0.0355 (0.0119) (0.0357) (0.0361) Medium threat point tariffs 0.0288*** 0.0202 0.0278 (0.00947) (0.0305) (0.0309) Low threat point tariffs 0.0109 0.000366 0.0473* (0.00884) (0.0279) (0.0281) Constant 0.0116* 0.00635-0.00966 (0.00704) (0.0225) (0.0225) Observations 7,505 7,231 7,141 R-squared 0.002 0.001 0.001 Tariff rate
Analysis of UK based firms: Tariff rate Total value of exports Table 17: Intensive margin exporting to the EU: 2016 relative to 2015 Panel A (1) (2) (3) (4) Value Transaction Units Mass Threat point tariff rate 0.00398** -0.000829-0.00628* 0.00185 (0.00160) (0.000795) (0.00378) (0.00179) Constant 0.0556*** 0.0549*** 0.120*** 0.0147 (0.0107) (0.00529) (0.0270) (0.0121) Observations 7,665 7,665 2,214 7,664 R-squared 0.001 0.000 0.001 0.000 Discrete categories
Analysis of UK based firms: Discrete tariff categories Total value of exports Table 18: Intensive margin exporting to the EU: 2016 relative to 2015 (1) (2) (3) (4) Value Transaction Units Mass Extreme threat point tariffs 0.0680-0.0438** -0.0950 0.0596 (0.0419) (0.0211) (0.125) (0.0474) High threat point tariffs 0.121*** 0.0192-0.0295 0.0450 (0.0266) (0.0125) (0.0552) (0.0330) Medium threat point tariffs 0.0418** 0.00107-0.00343 0.0217 (0.0211) (0.0101) (0.0536) (0.0243) Low threat point tariffs 0.0333* -0.00747 0.0312 0.0151 (0.0199) (0.00978) (0.0501) (0.0228) Constant 0.0384** 0.0536*** 0.0816* 0.00587 (0.0160) (0.00785) (0.0424) (0.0178) Observations 7,665 7,665 2,214 7,664 R-squared 0.003 0.001 0.001 0.000 Tariff rate