Trade Facilitation and Supply Chain Security: Evaluation of Integrated Cargo Containers Control (IC3) Program between Pakistan and the United States WCO-PICARD Conference 27-29 September 2016 Presenter: Salamat Ali Pakistan Customs/University of Nottingham, UK Structure Introduction of IC3 Program Research Question and Findings Related Literature Data and Methodology Main Results Summary and Policy Implications 1
Trade Facilitation and Supply Chain Security: Evaluation of Integrated Cargo Containers Control (IC3) Program between Pakistan and the United States
Motivation: Trade Facilitation and Supply Chain Security 3
Integrated Cargo Containers Controls (IC3) Pilot project at Port Qasim, Pakistan; Southampton Port, UK; Puerto Cortès, Honduras Intrusive scanning and monitoring via live video link even before loading on vessels Limited implementation at ports of Singapore, Hong Kong and Busan (Korea) Extension of the 100% scheme to all US-bound cargo from all origins 4
Research Question(s) What is the impact of IC3 on firm-level exports of Pakistan to the United States? Effect of subsequent policy adjustments in 2011 to facilitate the process Examine the heterogeneity of the trade effect across firms (incumbents and switchers) Mechanisms of adjustment along: Extensive margins of firms and products Prices and quantities Over time 5
Preview of Results Pakistan s exports to the US relative to the EU drop by 15% in the post- IC3 period. The drop with respect to India and China is around 40%. Adjustment comes along extensive margins of firms and products as well as along margins of price and quantities. Effect of IC3 is persistent for four years until subsequent policy interventions reverse this trend to some extent. The security policy caused a cumulative loss of US market access to the tune of US$ 8 billion during 2007-2014. 6
Related Literature Trade Costs: Arkolakis (2010); Feyrer (2009); Anderson and Van Wincoop (2003, 2004); Baier and Bergstrand (2001); Donaldson (2014) Technology and Trade: Bernhofen et al. (2015); Pascali (2014); Hummels (2007) Economic Sanctions: Afesorgbor &Mahadevan (2016); Yang et al. (2009); Caruso (2003) Trade Diversion: Yang et al. (2014); Liu et al. (2013); Carrere (2006) Supply Chain Security: Mirza and Verdier (2008); EC(2009); WCO(2008); GAO (2008) Uniqueness of this Work Exogenous nature of the shock Differential effect on trade costs across markets Unique data sets and first firm-level study to quantify the effect of typical trade restriction 7
Data Description Sources: Firm-level exports from Pakistan Customs and domestic sales from Inland Revenue Services (administrative datasets) Period: Jan 2000 to Dec 2015 Frequency: Transaction level Observations: 8.6 million; EU (Control)- 4.7m and US (Treatment)-3.9m Firms: 24,174 Variables: Products at HS-8 digit level, unit values, quantities, identity of exporter and importers, location of production facility of firms, location of their exporting station, mode of shipment, etc. Estimation level: Firm-product-market-year (463, 931 observations) 8
Influence of IC3 on US-bound Exports from Pakistan Beyond-the-border Behind-the-border Eliminating Transhipment requirements and allowing direct shipments to the US Diversion of USbound exports from all dry ports and other seaport to Qasim Port, Karachi 9
Effect of the IC3 on Trade Costs en Route 10
At- and Behind-the-Border Effect of IC3 11
Behind-the-Border Effect Percentage of US-bound exports handling at various stations Source: Pakistan Customs 12
2000 3000 4000 5000 6000 Exports Volume Influence of IC3 on US-bound Exports Index, 2000=1 December 2006 2000 2002 2004 2006 2008 2010 2012 2014 European Union United States 13
Empirical Setting: Diff-in-Diff Estimation Approach First Treatment (2007) ln(x ijkt ) = β 0 + β 1 (Treat) j + β 2 (After) t + β 3 (Treat x After) jt + α i + γ k + λ t + ε ijkt...(1) X ijkt denotes the value of exports of a firm i to market j of product k at a time t (intensive margins). The export volume is measured in PKR millions. Treat is a dummy variable equal to 1 if an observation pertains to the US and 0 for the EU. After is a dummy variable equal to 1 for the period 2007-2014 and 0 otherwise. α, γ and λ are firms, products and time fixed effects. ε ijkt is an idiosyncratic error term. Second Treatment (2011) ln(x ijkt ) = β 0 + β 1 (Treat) j + β 2 (After 1 ) t + β 3 (Treat x After 1 ) jt + β 4 (After 2 ) t+ β 5 (Treat x After 2 ) jt + α i + γ k + λ t + ε ijkt..(2) After 1 is a dummy variable equal to 1 for the period 2007-2014 and 0 otherwise. After 2 is a dummy variable equal to 1 for the period 2011-2014 and 0 otherwise. 14
Main Estimation Results The dependent variable is a log of exports per firm by destination (1) (2) (3) (4) Interaction (treat x after) 1st Treatment_ 2007-0.200 *** (0.022) -0.134 *** (0.019) -0.157 *** (0.018) -0.151 *** (0.018) 2nd Treatment_ 2011-0.218 *** (0.018) -0.014 (0.016) 0.036 ** (0.015) 0.030 ** (0.015) Firm FE y y y Product FE y y Time FE y R 2 0.02 0.40 0.51 0.51 Observations 463,931 463,931 463,931 463,931 Robust standard errors are in parentheses. These coefficients were obtained using stata 13SE * p < 0.10, ** p < 0.05, *** p < 0.01. The coefficients on other regressors and fixed effects are not reported. 15
Heterogeneity of the Effect across Firms The dependent variable is a log of exports per firm by destination (1) 1st Treatment_2007 x Incumbents at PQ -0.089 *** (0.029) Switcher to PQ -0.164 *** (0.018) Continuers at KP -0.149 ** (0.075) 2nd Treatment_2011 x Incumbents at PQ -0.018 (0.026) Switcher to PQ 0.041 ** (0.017) Continuers at KP 0.038 (0.095) R 2 0.505 Observations 463,931 Incumbents export from Qasim port before and after IC3 while Switchers export from Karachi port and dry ports before IC3 and switch to Qasim port after IC3 16
Mechanisms of Adjustment 17
Speed of Adjustment Coeff. SE (1) (2) Interaction (treat x after) x int_2007-0.478 *** (0.022) int_2008-0.495 *** (0.024) int_2009-0.497 *** (0.024) int_2010-0.349 *** (0.024) int_2011-0.083 *** (0.024) int_2012 0.245 *** (0.024) int_2013 0.367 *** (0.024) int_2014 0.398 *** (0.025) R 2 0.50 Observations 463,931 18
Summary and Conclusion We investigate the trade effect of IC3 in the wake of 9/11. The exogenous nature of this shock and its specificity to one export market allows us to use a diff-in-diff estimation approach. We find that: In the post treatment period, Pakistan s exports to the US relative to the EU drop by 15%, on average. The cumulative loss of US market access during 2007-2014 amounts to $8 billion. Switchers bear the main loss and the subsequent adjustments reverse this declining trend to some extent. These findings have policy implications to use similar technologies to secure supply chain and facilitate trade flows in the wake of changing security situation in different parts of the world. 19
Thanks for your attention Q&A 20
Integrated Cargo Containers Controls (IC3) 21 Source: European Commission (2009)
Main Estimation Results ln(x ijkt )= β 0 + β 1 (Treat) j + β 2 (After) t + β 3 (Treat x After) jt + α i + γ k + λ t + ε ijkt The dependent variable is log of exports per firm by destination Interaction Term (Treat x After) Treatment 0.844 *** (1) (2) (3) (4) (5) -0.371 *** -0.133 *** -0.131 *** -0.106 *** -0.106 *** (0.025) (0.022) (0.021) (0.021) (0.021) (0.023) After -0.107 *** (0.017) 0.262 *** (0.021) 0.178 *** (0.014) 0.248 *** (0.020) 0.136 *** (0.014) 0.233 *** (0.020) 3.591 *** (0.128) 0.233 *** (0.020) 2.451 *** (0.202) Time trend 0.095 *** Firm fixed effects y y y y Prod. fixed effects y y y Time fixed effects y y (0.014) R2 0.011 0.394 0.422 0.437 0.437 Observations 472,258 472,258 472,258 472,258 472,258 Note: Robust standard errors are in parentheses. These coefficients were obtained using Stata 13 SE; * p < 0.10, ** p < 0.05, *** p < 0.01. 22
Robustness: Including time-varying Fixed Effects The dependent variable is the log of exports per firm by destination (1) (2) (3) (4) Interaction (Treat x After) -0.144 *** (0.030) -0.148 *** (0.029) -0.253 *** (0.018) -0.150 *** (0.030) Firm-year FE Y Y Y Product-year FE Y Y Prod.-market Y Y R2 0.504 0.534 0.159 0.535 N 472,258 472,258 472,258 472,258 Robust standard errors are in parentheses. These coefficients were obtained using stata 13SE * p < 0.10, ** p < 0.05, *** p < 0.01 23
Empirical Setting Composition of Exports to Control (EU) & Treatment (US) Groups US EU 90-99_Misc.s 86-89_Transport 84-85_Mach.Elect 72-83_Metals 68-71_StoneGlas 64-67_Footwear 50-63_TextCloth 44-49_Wood 41-43_HidesSkin 39-40_PlastiRub 28-38_Chemicals 27-27_Fuels 25-26_Minerals 16-24_FoodProd 06-15_Vegetable 01-05_Animal - 10 20 30 40 50 60 70 80 90 100 Source: Pakistan Customs Share of various products groups in the exports baskets (%), 2013 24
A: Effect of IC3 on Single and Multiple Market Firms Multi-Market Firms (EU &US) Single Market Firms (EU/US) (1) (2) Treat x After -0.176 *** (0.021) -0.099 *** (0.031) R 2 0.486 0.551 Observations 235, 830 225, 416 B: Effect of IC3 on the Continuing Cohort of Firms All Firms Multi-Market Firms (EU &US) Single Market Firms (EU/US) (1) (2) (3) Treat x After -0.172 *** (0.026) -0.155 *** (0.031) -0.206 *** (0.041) R 2 0.471 0.444 0.506 Observations 280,881 148, 494 130, 936 25
Collapsing Data to Single Pre- and Post-IC3 Period (1) (2) Treat x after 1 st Treatment_2007-0.173 *** (0.021) 2 nd Treatment_2011-0.030 (0.018) FE (firms, products) y y R 2 0.49 0.50 Observations 295, 668 295,668 26
Beyond-the-Border effect on US-bound Exports Maritime Distances A: Maritime Distances in Kilometres Destination Direct Via Sri Lanka Via Hong Kong Via Salalah (Oman) New York 14,812 18,424-19.60% 28,591 NA 14,852-0.27% Los Angles 19,564 19,756-0.97% 19,828-1.33% 21,754-10.07% B: Vessel Sailing Time in Number of Days Destination Direct Via Sri Lanka Via Hong Kong Via Salalah (Oman) New York 24 30-20.00% 45 NA 25-4.00% Los Angles 31 32-3.13% 32-3.13% 35-11.43% Source: http://www.searates.com/reference/portdistance/ Note: Firms do not appear to ship to New York via Hong Kong 27
Billion US$ Loss of Market Access 7 6 5 Actual Projected 4 3 2 1 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 28
Pakistan's major Export Markets, 2013 Trading Partners Trade (US$ M) Share % United States 3,746 14.91 China 2,652 10.56 Afghanistan 1,998 7.95 United Arab Emirates 1,775 7.07 European Union 5,932 23.01 29
Drop in Freight Rates (US$/MT) European Union United Kingdom United States Year Freight % Change Freight % Change Freight % Change (1) (2) (1) (2) (1) (2) 2003 194 251 285 2004 196 1 226-10 258-9 2005 187-5 241 7 267 3 2006 172-8 220-9 274 3 2007 162-6 209-5 263-4 2008 149-8 194-7 219-17 2009 101-32 159-18 191-13 2010 110 9 125-22 211 10 2011 113 2 134 7 227 7 Source: Computed using Pakistan Custom s Dataset Note: The freight values are in US dollars per metric ton and percentage changes are year-on-year basis 30
Global Financial Crisis and US Imports from Pakistan and the World Exports in 2002==1 31