Neural Network Model for Classifying the Economic Recession and Construction of Financial Stress Index

In this paper, a C5.0 decision tree and neural network models are proposed to classify recessions in the US with 12 common financial indices and new financial stress indices inferred from the neural network models are created. A detailed experiment is presented and demonstrates that the neural network models with proper regularization and dropout achieve 98% accuracy in the training set, 97% accuracy in validation set and 100% accuracy in test accuracy. The financial stress indices outperform other existing financial stress indices in many scenes and can accurately locate crisis events even the most recent 2018 US Bear Market. With these models and new indices, contraction can be detected before NBER’s announcement and action could be taken as early as the situation get worse.

[1]  Kuldeep Kumar,et al.  Predicting financial distress: A comparison of survival analysis and decision tree techniques , 2015 .

[2]  Regime-Dependent Recession Forecasts and the 2001 Recession , 2002 .

[3]  Pankaj M. Madhani.  Rebalancing Fixed and Variable Pay in a Sales Organization: A Business Cycle Perspective , 2010 .

[4]  Ali Uyar,et al.  Measuring firm performance using financial ratios: A decision tree approach , 2013, Expert Syst. Appl..

[5]  Budi Warsito,et al.  Classification of Company Performance using Weighted Probabilistic Neural Network , 2018 .

[6]  A. Elsalamony,et al.  Bank Direct Marketing Based on Neural Network and C5.0 Models , 2013 .

[7]  M. Aziakpono,et al.  FORECASTING RECESSION IN SOUTH AFRICA: A COMPARISON OF THE YIELD CURVE AND OTHER ECONOMIC INDICATORS , 2007 .

[8]  Mu-Yen Chen,et al.  Predicting corporate financial distress based on integration of decision tree classification and logistic regression , 2011, Expert Syst. Appl..

[9]  Henri Nyberg,et al.  Dynamic Probit Models and Financial Variables in Recession Forecasting , 2010 .

[10]  Craig S. Hakkio,et al.  Financial stress: what is it, how can it be measured, and why does it matter? , 2009 .

[11]  Monica Lam,et al.  Neural network techniques for financial performance prediction: integrating fundamental and technical analysis , 2004, Decis. Support Syst..

[12]  Teruyoshi Kobayashi,et al.  Network models of financial systemic risk: a review , 2017, Journal of Computational Social Science.

[13]  Nitesh V. Chawla,et al.  SMOTE: Synthetic Minority Over-sampling Technique , 2002, J. Artif. Intell. Res..

[14]  María Isabel Cambón,et al.  A Spanish Financial Market Stress Index (FMSI) , 2016 .