Estimating the impacts of US tariffs on UK exports of single malt Scotch whisky

Footnotes

1. See the United States' Harmonized Tariff Schedule Search (usitc.gov)

2. HMRC's Overseas Trade Methodology, Section 8: The valuation of exports/dispatches is on a Free on Board (FOB) delivery terms basis, i.e. the cost of goods to the purchaser abroad, including packaging, Inland and coastal transport in the UK, dock dues, loading charges and other profits, charges, and expenses (e.g. insurance) accruing up to the point where the goods are deposited on board the exporting vessel or aircraft or at the land boundary of Northern Ireland.

3. Elyse R. Grossman, Sara E. Benjamin-Neelon, Susan Sonnenschein, 2020. "Alcohol Consumption during the COVID-19 Pandemic: A Cross-Sectional Survey of US Adults." International Journal of Environmental Research and Public Health, vol 17(24), p. 9189. doi:10.3390/ijerph17249189

4. Michael S. Pollard, Joan S. Tucker, Harold D. Green, 2020. "Changes in Adult Alcohol Use and Consequences During the COVID-19 Pandemic in the US." JAMA Network Open, vol 3(9): e2022942. doi:10.1001/jamanetworkopen.2020.22942

5. Samantha N. Sallie, Valentin Ritou, Henrietta Bowden-Jones, et al, 2020. "Assessing international alcohol consumption patterns during isolation from the COVID-19 pandemic using an online survey: highlighting negative emotionality mechanisms." British Medical Journal Open, vol 10: e044276. doi: 10.1136/bmjopen-2020-044276

6. Available at: https://pubs.niaaa.nih.gov/publications/surveillance-covid-19/COVSALES.htm#fig19

7. OECD, 2010. "Sensitivity of trade flows to price and income changes," Measuring Globalisation: OECD Economic Globalisation Indicators 2010, OECD Publishing, Paris. DOI: https://doi.org/10.1787/9789264084360-31-en

8. See, for example: Marie M. Stack, Martin Bliss, 2020. "EU economic integration agreements, Brexit and trade." Review of World Economics vol 156, pages 443–473. https://doi.org/10.1007/s10290-020-00379-x

9. Thomas Chaney, 2018. "The Gravity Equation in International Trade: An Explanation," Journal of Political Economy, vol 126(1), pages 150-177.

10. David Card and Alan B. Krueger, 2000. "Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Reply", The American Economic Review, Vol. 90(5), pp. 1397-1420

11. Bruce D. Meyer, W. Kip Viscusi, David L. Durbin, 1995. "Workers' Compensation and Injury Duration: Evidence from a Natural Experiment." The American Economic Review, vol. 85(3), pp. 322–340.

12. Moulton, B. (1990). An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Units. The Review of Economics and Statistics, 72(2), 334-338. doi:10.2307/2109724

13. Marianne Bertrand, Esther Duflo, Sendhil Mullainathan, 2004: "How Much Should We Trust Differences-In-Differences Estimates?". The Quarterly Journal of Economics, Oxford University Press, vol. 119(1), pages 249-275.

14. Alberto Abadie and Javier Gardeazabal, 2003. "The Economic Costs of Conflict: A Case Study of the Basque Country." American Economic Review, 93 (1): 113-132. Obtained from: https://economics.mit.edu/files/11870.

15. Alberto Abadie, Alexis Diamond, Jens Hainmueller, 2010. Journal of the American Statistical Association. June 1, 2010, 105(490): 493-505. doi:10.1198/jasa.2009.ap08746. Version: Author's final manuscript.

16. Note here the switch from 'group' to 'unit' showing that we are now dealing with aggregate observations.

17. John J. Donohue & Abhay Aneja & Kyle D. Weber, 2019. "Right‐to‐Carry Laws and Violent Crime: A Comprehensive Assessment Using Panel Data and a State‐Level Synthetic Control Analysis," Journal of Empirical Legal Studies, vol 16(2), pages 198-247.

18. Scott Cunningham & Manisha Shah, 2018. "Decriminalizing Indoor Prostitution: Implications for Sexual Violence and Public Health," The Review of Economic Studies, vol 85(3), pages 1683-1715.

19. Kleven, Henrik Jacobsen, Camille Landais, and Emmanuel Saez, 2013. "Taxation and International Migration of Superstars: Evidence from the European Football Market," American Economic Review, vol 103(5), pages 1892-1924.

20. Swarnali A. Hannan, 2016. "The Impact of Trade Agreements: New Approach, New Insights," IMF Working Paper No. 16/117.

21. Swarnali A. Hannan, 2017. "The Impact of Trade Agreements in Latin America using the Synthetic Control Method," IMF Working Paper No. 17/45.

22. Matthew J. Slaughter, 2001. "Trade liberalization and per capita income convergence: a difference-in-differences analysis," Journal of International Economics, Vol 55(1), pages 203-228.

23. Georgias Fotopoulus and Dionysios Psallidas, 2009. "Investigating the Effects of Euro on Bilateral Trade: a Kernel Matching Approach," Journal of Economic Integration vol 24(4), pages 661-684.

24. In short: a relationship between two stochastic variables which individually may behave as a random walk, but a linear combination of them will not – i.e., there is a long-run relationship between them.

25. Ariel Linden, 2018. "Combining synthetic controls and interrupted time series analysis to improve causal inference in program evaluation", Journal of Evaluation in Clinical Practice, vol 24(2), pages 447-453. doi: 10.1111/jep.12882.

26. Bruno Ferman, Cristine Campos de Xavier Punto, and Vítor Augusto Possebom, 2020. "Cherry picking with synthetic controls," Journal of Policy Analysis and Management, John Wiley & Sons, Ltd., vol. 39(2), pages 510-532. Draft version obtained from https://mpra.ub.uni-muenchen.de/85138/.

27. Sergio Firpo and Vitor Possebom, 2018. "Synthetic Control Method: Inference, Sensitivity Analysis and Confidence Sets," Journal of Causal Inference, vol 6(2). https://doi.org/10.1515/jci-2016-0026

28. Abadie, 2021. "Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects". Journal of Economic Literature (forthcoming issue). Preview (2020) available at: https://economics.mit.edu/files/17847.

29. See Becker and Klößner (2017): https://cran.r-project.org/package=MSCMT

30. Martin Becker and Stefan Klößner, 2018. "Fast and reliable computation of generalized synthetic controls". Econometrics and Statistics, vol 5, pages 1-19. Preliminary version (2017) available at: http://www.oekonometrie.uni-saarland.de/papers/FastReliable.pdf

31. A ratio of mean squared error predictions can also be used. This would alter the test statistic for each country and simply shift the scale. This is used, for example, in Firpo and Possebom, 2018.

32. Sensitivity analysis could be performed by varying and .

33. A one-sided null hypothesis may be given by where a mean prediction error test statistic could be used where the p-value 𝑝𝑐 can be calculated as

34. Accessed 12 March 2021, available at: https://www.uktradeinfo.com/trade-data/

35. Accessed 12 March 2021, available at: https://www.uktradeinfo.com/trade-data/

36. Accessed throughout March 2021, available at: https://data.oecd.org/

37. This affected roughly 3% of observations in our donor pool. The simple imputation method chosen here is not the only option, see for example Rob Hyndman’s blog post here: https://robjhyndman.com/hyndsight/transformations/. Given the small frequency of these observations, it was deemed a simple imputation approach was sufficient for this analysis.

38. Christoph Sax and Peter Steiner, 2013. "Temporal Disaggregation of Time Series," The R Journal vol 5(2), pages 80-87. Available at: https://journal.r-project.org/archive/2013/RJ-2013-028/index.html

39. In short, the Denton-Cholette method was used to disaggregate quarterly series to monthly series subject to the constraint that each quarter's disaggregated values sum to the actual quarter's values.

40. Preferential or MFN applied tariffs, for which the data was obtained from the World Trade Organization (see Annex A - Donor Pool and Model Specifications).

41. In June 2017, Estonia increased the duties charged on many alcoholic beverages, potentially giving rise to increased cross-border alcohol trade with Latvia and Finland, for example (which would at least partially explain the increase in exports to Latvia following 2017). Although estimates of this effect are scarce, anecdotal accounts are given in various news reports. See for example BBC, 2017. "Estonian tax threat to Finns' booze cruises" and the excise duty page on the Ministry of Finance of the Republic of Estonia, 2019.

42. See Table A1 in the Annex a full list of countries in the final donor pool, and Table A2 for a breakdown of countries not selected.

43. T. Hale et al, 2021. "A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker)," Nature Human Behaviour.

44. Since we have 28 countries in the donor pool, the smallest 'p-value' obtainable is . To obtain this p-value, we would require that the US has the largest test statistic amongst all 29 countries – see equation 5. For this reason, we settle on a significance level (the closest we can get to a 10% significance level based on the low power) and a confidence set.

45. This is done using a series of linear approximations in the form of: , where is the difference in logs between the post-tariff and pre-tariff gap. See Annex C - Inference Confidence Sets for more informaiton for more information.

46. This single specification was selected to save computing time.

47. The synthetic control obtained when using value as a dependent variable and a first-lag specification.

48. T. Hale, N. Angrist, R. Goldszmidt, B. Kira, A. Petherick, T. Phillips, S. Webster, E. Cameron-Blake, L. Hallas, S. Majumdar, and H. Tatlow, 2021. "A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker)," Nature Human Behaviour.

49. Further differences could be due to the exclusion of duty free sales in these figures, the different comparison we are making (2019 vs. 2020 as opposed to pre-tariff and post-tariff), and the different quantity unit used in this data (volume of whisky as opposed to volume of pure alcohol). Note also that the volume and value of US retail sales are not mimicked by the synthetic control particularly well. This is because the US had one of the highest per-capita sales of whisky when compared to countries in the donor pool (and the countries with higher sales did not have much weight put on them).

50. It is clear from the data that a 100% reduction is impossible. The actual lower bound of the range is given by a number somewhere between 95% to 100%. Calculating the exact number was deemed too computationally intensive.

51. One example not mentioned in this section is the option to model multiple dependent variables simultaneously in the MSCMT R package by Becker and Klößner (2019).

52. See for example https://whiskycast.com/u-s-tariff-trouble-hits-scotch-whisky-industry/.

53. See for example https://www.thespiritsbusiness.com/2021/06/american-whiskey-tariffs-is-relief-in-sight/.

54. See the joint statements by the UK and US governments here: https://ustr.gov/about-us/policy-offices/press-office/press-releases/2021/march/joint-us-uk-statement-suspension-large-civilian-aircraft-tariffs and https://ustr.gov/about-us/policy-offices/press-office/press-releases/2021/june/joint-us-uk-statement-cooperative-framework-large-civil-aircraft.

55. This information was obtained using the WTO'S Tariff Analysis Online facility.

56. See here: https://www.wto.org/english/tratop_e/dispu_e/find_dispu_cases_e.htm

57. There were 88 relevant disputes for the EU, where DS316 (European Communities and Certain member States — Measures Affecting Trade in Large Civil Aircraft) was the only dispute for which retaliation (by the United States) was granted. Authorisation to retaliate was requested by the United States regarding DS291 (European Communities — Measures Affecting the Approval and Marketing of Biotech Products) with no further updates since 17 January 2008. Similarly, 3 relevant disputes existed for the United Kingdom, with DS316 being the only dispute where retaliation was granted.

58. Interpolation of annual population estimates did slightly increase the impact estimate across the board, e.g., a larger negative tariff impact on value, but only by c.a. 0.4% (this also resulted in the lower bound of the 90% confidence set being roughly 1% lower).