Ahuja, A., K. Wiseman, M. Syed. 2017. “Assessing country risk — selected approaches.” IMF Technical Notes and Manuals 17/08 (Washington: International Monetary Fund).
Alain, G., Y. Bengio. 2014. “What regularized auto-encoders learn from the data-generating distribution.” J. of Machine Learning Research 15: 3743—3773.
Alessi, L. et al. 2014. “Comparing different early warning systems: results from a horse race competition among members of the Macro-prudential Research Network.” Mimeo.
Alexander, M., M. Harding, C. Lamarche. 2008. “The human cost of economic crises.” SIEPR Working Paper 08–029. Stanford Institute for Economic Policy Research.
Arora, R., A. Basu, P. Mianjy, A. Mukherjee. 2018. “Understanding Deep Neural Networks with Rectified Linear Units.” International Conference on Learning Representations.
Basu, S., M. Chamon, C. Crowe. 2017. “A model to assess the probabilities of growth, fiscal, and financial crises.” IMF Working Paper WP/17/282.
Babecky, J. T. Havranek, J. Mateju, M. Rusnak, KI. Smidkova, B. Vacisek. 2012. “Banking, debt, and currency crises: early warning indicators for developed countries.” ECB Working Paper 1485 (Frankfurt, European Central Bank).
Beutel, J., S. List, G. von Schweinitz. 2018. “An evaluation of early warning models for systemic banking crises: does machine learning improve predictions?” Discussion Paper No. 48/2018 (Frankfurt, Deutsche Bundesbank).
van Buuren, S. 2007. “Multiple imputation of discrete and continuous data by fully conditional specification.” Statistical Methods in Medical Research 16(3): 219—242.
Cerra, V., S. Saxena. 2008. “Growth dynamics: the myth of economic recovery.” American Economic Review 98(1): 439–57.
Chamon, M., C. Crowe. 2013. “Predictive indicators of financial crises.” In Caprio, G., T. Beck, S. Claessens, S. Schmukler (eds.). The Evidence and Impact of Financial Globalization (Academic Press).
Chawla, N. V. N. Japkowicz, A. Kotcz. 2004. “Special issue on learning from imbalanced data sets.” ACM Sigkdd Explorations Newsletter 6(1): 1–117.
Chawla, N. V., K.W. Bowyer, L.O. Hall, W.P. Kegelmeyer. 2002. “SMOTE: synthetic minority over-sampling technique.” J. of Artificial Intelligence Research 16: 321–357.
Christofides, C. T., T.S. Eicher, C. Papageorgiou. 2016. “Did established early warning signals predict the 2008 crisis?” European Economic Review 81(1): 103–114.
Clevert, D.-A., T. Unterthiner, S. Hochreiter. 2016. “Fast and accurate deep network learning by exponential linear units (ELUs).” Fourth International Conference on Learning Representations.
Dempster, A., N. Laird, D. Rubin. 1977. “Maximum likelihood from incomplete data via the EM algorithm.” J. of the Royal Statistical Society Series B (Methodological) 39 (1): 1–38.
van Dijk, M. 2013. “The social cost of financial crises.” Mimeo. Rotterdam School of Management, Erasmus University.
Frankel, J. A., A.K. Rose. 1996. “Currency crashes in emerging markets: and empirical treatment.” J. of International Economics 41(3–4), 351–366.
Holpainen, M., P. Sarlin. 2017. “Toward robust early-warning models: a horse race, ensembles and model uncertainty.” Quantitative Finance 17(12): 1933–1963.
Kaminsky, G. L., S. Lizondo, C.M. Reinhart. 1998. “Leading indicators of currency crises.” IMF Staff Papers 45(1): 1–48.
Klambauer, G., T. Unterthiner, A. Mayr. 2017. “Self-normalizing neural networks.” 31st Conference on Neural Information Processing Systems (NIPS 2017). Long Beach, California.
Kotsiantis, S., D. Kanellopoulos, P. Pintelas. 2006. “Handling imbalanced datasets: a Review.” GESTS International Transactions on Computer Science and Engineering 30: 25–36.
Lang, J., T. Peltonen, P. Sarlin. 2018. “A framework for early-warning modeling with an application to banks.” ECB Working Paper 2182 (Frankfurt, European Central Bank).
Medas, P. T. Poghosyan, Y. Xu, J. Farah-Yacoub, K. Gerling. 2018. “Fiscal crises.” J. of International Money and Finance 88: 191–207.
de O?a, J., C. Garrido. 2014. “Extracting the contribution of independent variables in neural network models: a new approach to handle instability.” Neural Computation & Applicactions 25: 859.
Pascanu, R., G. Montufar, Y. Bengio. 2014. “On the number of response regions of deep feed forward networks with piece-wise linear activations.” International Conference on Learning Representations.
Raghunathan, T. E., J.M. Lepkowski, J. vanHoewyk, P.A. Solenberg. 2001. “A Multivariate technique for multiply imputing missing values using a sequence of regression models.” Survey Methodology 27(1): 85–95.
Reinhart, C. M., K.S. Rogoff. 2014. “Recovery from financial crises: evidence from 100 episodes.” American Economic Review 104(5): 50–55.
Rezende, D. J., Mohamed, S., Wierstra, D., 2014. Stochastic Backpropagation and Approximate Inference in Deep Generative Models. arXiv Prepr. arXiv1401.4082.
Stiglitz, J. 2017. “Where modern macroeconomics went wrong.” NBER Working Paper No. 23795, National Bureau for Economic Research (Cambridge, Massachusetts).
Vincent, P., H. Larochelle, Y. Bengio, P.-A. Manzagol. 2008. “Extracting and composing robust features with denoising autoencoders.” Proceeding ICML ’08 Proceedings of the 25th international Conference on Machine learning.
Vincent, P., H. Larochelle, Y. Bengio, P.-A. Manzagol. 2010. “Stacked Denoising Autoencoders: Learning Useful Representations in a Deep Network with a Local Denoising Criterion.” The Journal of Machine Learning Research 11: 3371–3408.
Xie, J., R. Girshick, A. Farhadi. 2016. “Unsupervised deep embedding for clustering analysis.” Proceeding ICML’16 Proceedings of the 33rd International Conference on International Conference on Machine Learning 48: 478–487
The authors thanks seminar participants at the IMF and AMRO-3 for useful discussions; Ziya Gorpe, Chengyu Huang, Tae-Hwy Lee, James Mongardini, Chris Redl, Ruoyao Shi, Aman Ullah, and Harry Peng Zhao for their comments; and Chuqiao Bi and Lamya Kejji for preparing the data sets. All omissions and errors are the authors’ sole responsibility. Please address correspondence to both authors.
The biased label problem in crisis prediction is somewhat similar to the problem of label bias and fairness: data points are falsely attributed to a certain class even if the features may not justify it. See for instance, Jiang and Nachum (2019).
A related method is the Markov Chain Monte Carlo variational autoencoder-based of Rezende et al. (2015).
The activation function of this autoencoder, as well as the others unFEAR uses, is an exponential linear unit (ELUs) (Clevert et al., 2016). The convergence speed of ELUs outperforms that of rectified linear units (ReLUs) (Klambauer et al., 2017).
A more complex alternative to the use of a regularized loss function, as done here, is to use a denoising autoencoder incorporating the cluster requirement into the reconstruction error. On denoising autoencoders, see Alain and Bengio (2014).
A detailed description of the attributes is available upon request from the authors. Most variables are available from public IMF databases and/or private data providers. Probabilities of default are from the Credit Research Initiative at the Asian Institute of Digital Finance, National University of Singapore (https://rmicri.org). Researchers can access PD data upon registration.
Class 0 corresponds to the no crisis label, class 1 to financial crisis, class 2 to a sudden stop crisis, class 3 to an exchange rate market pressure index event, class 4 to a real sector crisis, and class 5 to a fiscal crisis.