Skip to main content
SpringerLink
Log in

Economic determinants of regional trade agreements revisited using machine learning

  • Published:
Empirical Economics Aims and scope Submit manuscript

Abstract

While traditional empirical models using determinants like size and trade costs can predict RTA formation reasonably well, we demonstrate that allowing for machine-detected nonlinear patterns helps to improve the predictive power of RTA formation substantially. We find that the fitted tree-based methods and neural networks deliver sharper and more accurate predictions than the probit model. For the majority of models, the allowance of fixed effects increases the predictive performance considerably. We apply our models to predict the likelihood of RTA formation of the EU and the USA with their trading partners, respectively.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16

Similar content being viewed by others

Notes

  1. We provide a graph showing the development of the number of trade relationships with different types of RTAs over the years 1960–2019 in the Online Appendix.

  2. Since Baier and Bergstrand (2004) many papers refined the specification, including for example neighboring effects like Egger and Larch (2008), Chen and Joshi (2010), Baldwin and Jaimovich (2012), and Baier et al. (2014) or political economy motives like Facchini et al. (2013), Maggi and Rodríguez-Clare (2007), Liu (2008), and Liu and Ornelas (2014). See Maggi (2014) for an excellent survey.

  3. The data are freely available at https://www.ewf.uni-bayreuth.de/de/forschung/RTA-daten/index.html.

  4. The data can be downloaded at https://databank.worldbank.org/source/world-development-indicators.

  5. The data can be downloaded at https://worldmap.harvard.edu/data/geonode:country_centroids_az8.

  6. Note that \(\text{ DROWKL }\) is significant but positive for all RTAs in the cross section in Egger and Larch (2008). Hence, using more recent data and data for more countries seems to bring the probit estimates closer to the theory.

  7. Available for download at https://CRAN.R-project.org/package=fixest.

  8. We provide formal demonstrations of these multi-collinearities in the Online Appendix.

  9. A regression of \(\text{ REMOTE }\) on \(\text{ NATURAL }\) and importer and exporter fixed effects delivers and \(R^2\) from basically 1 and a residual standard error of 0.0008558.

  10. Available for download at https://cran.r-project.org/package=rpart.

  11. Available for download at https://cran.r-project.org/web/packages/randomForest/.

  12. \(\lceil \cdot \rceil \) denotes the ceiling function, which always rounds to the next largest integer.

  13. Available for download at https://cran.r-project.org/web/packages/xgboost/.

  14. Available for download at https://cran.r-project.org/web/packages/GA/index.html. Alternative ways to do automatic hyperparametrization optimization are Bayesian optimization or simple random search (see Chollet and Allaire 2018 for example).

  15. Note that these are only possible explanations. It would be interesting in future research to dig deeper into the specific events and mechanisms.

  16. Available for download at https://cran.r-project.org/package=tensorflow.

  17. Available for download at https://cran.r-project.org/package=keras.

  18. While in the 1990s a big focus was on investigating the effect of different activation functions, the consensus nowadays is that a simple, computationally efficient nonlinear transformation is sufficient if one uses enough nodes and layers (see Taddy 2019).

  19. The corresponding figure without fixed effects is in the Online Appendix as Figure A.8.

  20. https://trade.ec.europa.eu/doclib/docs/2006/december/tradoc_118238.pdf.

  21. See https://trade.ec.europa.eu/doclib/docs/2006/september/tradoc_122530.pdf.

  22. https://ustr.gov/trade-agreements/free-trade-agreements/bahrain-fta, https://ustr.gov/trade-agreements/free-trade-agreements/morocco-fta.

References

  • Akman E, Karaman AS, Kuzey C (2020) Visa trial of international trade: evidence from Support vector machines and neural networks. J Manag Anal 7:231–252

    Google Scholar 

  • Alimi OA, Ouahada K, Abu-Mahfouz AM (2020) A review of machine learning approaches to power system security and stability. IEEE Access 8:113512–113531

    Article  Google Scholar 

  • Alschner W, Seiermann J, Skougarevskiy D (2018) Text of trade agreements (ToTA)–a structured corpus for the text-as-data analysis of preferential trade agreements. J Empir Legal Stud 15:648–666

    Article  Google Scholar 

  • Athey S, Imbens G (2017) The state of applied econometrics: causality and policy evaluation. J Econ Perspect 31:3–32

    Article  Google Scholar 

  • Baier S, Bergstrand J (2004) The economic determinants of free trade agreements. J Int Econ 64:29–63

    Article  Google Scholar 

  • Baier SL, Bergstrand JH, Mariutto R (2014) Economic determinants of free trade agreements revisited: distinguishing sources of interdependence. Rev Int Econ 22:31–58

    Article  Google Scholar 

  • Baldwin R, Jaimovich D (2012) Are free trade agreements contagious? J Int Econ 88:1–16

    Article  Google Scholar 

  • Batarseh F, Gopinath M, Nalluru G, Beckman J (2019) Application of machine learning in forecasting international trade trends. arXiv preprint arXiv:1910.03112

  • Bergé L (2018) Efficient estimation of maximum likelihood models with multiple fixed-effects: the R Package FENmlm. CREA Discussion Paper Series available for download at https://EconPapers.repec.org/RePEc:luc:wpaper:18-13

  • Bernhofen DM, El-Sahli Z, Kneller R (2016) Estimating the effects of the container revolution on world trade. J Int Econ 98:36–50

    Article  Google Scholar 

  • Breeden JL (2020) Survey of machine learning in credit risk. Available at SSRN 3616342

  • Breinlich H, Corradi V, Rocha N, Ruta M, Santos Silva JAMC, Zylkin T (2021) Machine learning in international trade research—evaluating the impact of trade agreements. University of Surrey Discussion Papers in Economics DP 05/21

  • Cameron A, Trivedi P (2005) Microeconometrics–methods and applications. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Chen M, Joshi S (2010) Third-country effects on the formation of free trade agreements. J Int Econ 82:238–248

    Article  Google Scholar 

  • Chollet F, Allaire J (2018) Deep Learning with R. Manning

  • Circlaeys S, Kanitkar C, Kumazawa D (2017) Bilateral trade flow prediction. Unpublished manuscript, available for download at http://cs229.stanford.edu/proj2017/final-reports/5240224.pdf

  • De Veaux RD, Ungar LH (1994) Multicollinearity: a tale of two nonparametric regressions. Selecting models from data. Springer, Berlin, pp 393–402

    Chapter  Google Scholar 

  • Efron B, Hastie T (2016) Computer age statistical inference–algorithms, evidence, and data science. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Egger P, Larch M (2008) Interdependent preferential trade agreement memberships: an empirical analysis. J Int Econ 76:384–399

    Article  Google Scholar 

  • Facchini G, Silva P, Willmann G (2013) The customs union issue: why do we observe so few of them? J Int Econ 90:136–147

    Article  Google Scholar 

  • Ghoddusi H, Creamer GG, Rafizadeh N (2019) Machine learning in energy economics and finance: a review. Energy Econ 81:709–727

    Article  Google Scholar 

  • Goodfellow I, Bengio J, Courville A, Bach F (2016) Deep learning. MIT Press, Massachusetts

    Google Scholar 

  • Gopinath M, Batarseh F, Beckman J (2020) Machine learning in gravity models: an application to agricultural trade. NBER Working Paper No. 27151

  • Hastie T, Tibshirani R, Friedman J (2009) The elements of statistical learning–data mining, inference, and prediction, 2nd edn. Springer, New York

    Google Scholar 

  • Herbrich R, Keilbach M, Graepel T, Bollmann-Sdorra P, Obermayer K (2001) Neural networks in economics: background. Technical University of Berlin, Applications and New Developments, Department of Computer Science

    Google Scholar 

  • Hummels DL, Schaur G (2013) Time as a trade barrier. Am Econ Rev 103:2935–2959

    Article  Google Scholar 

  • Jahn M (2020) Artificial neural network regression models in a panel setting: predicting economic growth. Econ Model 91:148–154

    Article  Google Scholar 

  • James G, Witten D, Hastie T, Tibshirani R (2021) An introduction to statistical learning–with applications in R, 2nd edn. Springer, New York

    Book  Google Scholar 

  • Lahann J, Scheid M, Fettke P (2020) Towards optimal free trade agreement utilization through deep learning techniques. In: Bui T, Sprague R (eds) Proceedings of the 53th Hawaii International Conference on System Sciences. Hawaii International Conference on System Sciences (HICSS-2020), January 7–10, Maui, Hawaii, United States. IEEE Computer Society

  • Liu X (2008) The political economy of free trade agreements: an empirical investigation. J Econ Integr 23:237–271

    Article  Google Scholar 

  • Liu X, Ornelas E (2014) Free trade agreements and the consolidation of democracy. Am Econ J Macroecon 6:29–70

    Article  Google Scholar 

  • Maggi G (2014) International trade agreements. In: Gopinath G, Helpman E, Rogoff KS (eds) Chapter 6 in the handbook of international economics, vol 4, Elsevier Ltd., Oxford, pp 317–390

  • Maggi G, Rodríguez-Clare A (2007) A political-economy theory of trade agreements. Am Econ Rev 97:1374–1406

    Article  Google Scholar 

  • Masini RP, Medeiros MC, Mendes EF (2020) Machine learning advances for time series forecasting. arXiv preprint arXiv:2012.12802

  • Mullainathan S, Spiess J (2017) Machine learning: an applied econometric approach. J Econ Perspect 31:87–106

    Article  Google Scholar 

  • Murphy K (2012) Machine learning–a probabilistic perspective. MIT Press, Cambridge

    Google Scholar 

  • Ni D, Xiao Z, Lim MK (2020) A systematic review of the research trends of machine learning in supply chain management. Int J Mach Learn Cybern 11:1463–1482

    Article  Google Scholar 

  • Quimba F, Barral M (2018) Exploring neural network models in understanding bilateral trade in APEC: a review of history and concepts. PIDS Discussion Paper No. 2018-33

  • Scrucca L (2017) On some extensions to GA package: hybrid optimisation. Parallelisation and islands evolution. The R Journal 9:187–206

    Article  Google Scholar 

  • Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 15:1929–1958

    Google Scholar 

  • Stammann A (2017) Fast and feasible estimation of generalized linear models with high-dimensional k-way fixed effects. Unpublished manuscript, available at arXiv:1707.01815

  • Stock J, Watson M (2017) Twenty years of time series econometrics in ten pictures. J Econ Perspect 31:59–86

    Article  Google Scholar 

  • Storm H, Heckelei T, Baylis K, Mittenzwei K (2019) Identifying effects of farm subsidies on structural change using neural networks. Agricultural and Resource Economics, Discussion Paper 2019:1. Available for download at http://ageconsearch.umn.edu/record/287343

  • Taddy M (2019) Business data science. McGraw-Hill Education Ltd

  • Varian H (2014) Big data: new tricks for econometrics. J Econ Perspect 28:3–28

    Article  Google Scholar 

  • Wohl I, Kennedy J (2018) Neural network analysis of international trade. Office of Industries Working Paper ID-049. Available for download at https://www.usitc.gov/publications/332/working_papers/neural_networks_and_international_trade_-_compiled_draft_06.pdf

Download references

Acknowledgements

We thank the handling editor Robert M. Kunst, the anonymous reviewers, and participants of the 2021 European Trade Study Group for useful comments.

Author information

Authors and Affiliations

  1. University of Bayreuth, Bayreuth, Germany

    Simon Blöthner

  2. Department of Law and Economics, University of Bayreuth, CEPII, Ifo, CESifo, GEP, Bayreuth, Germany

    Mario Larch

Authors
  1. Simon Blöthner
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Mario Larch
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Mario Larch.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file 1 (pdf 4200 KB)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Blöthner, S., Larch, M. Economic determinants of regional trade agreements revisited using machine learning. Empir Econ 63, 1771–1807 (2022). https://doi.org/10.1007/s00181-022-02203-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00181-022-02203-x

Keywords

JEL Classification

Search

Navigation