{ "title": "Forecasting Tail Risk via Neural Networks with Asymptotic Expansions", "publication": "IMF Working Papers, May 10, 2024", "sourceUrl": "https://www.imf.org/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841", "canonical": "https://www.imf.org/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841", "overlayPath": "/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/index.md", "summary": "We propose a new machine-learning-based approach for forecasting Value-at-Risk (VaR) named CoFiE-NN where a neural network (NN) is combined with Cornish-Fisher expansions (CoFiE).", "sections": [ { "heading": "Methodology and model", "content": "- Proposal: CoFiE-NN — a machine-learning-based approach for forecasting Value-at-Risk (VaR) that combines a neural network (NN) with Cornish-Fisher expansions (CoFiE).\n- NN specification: Long Short-Term Memory (LSTM) is employed as the main NN specification.\n- Rationale:\n - NN (LSTM) captures non-linear dynamics of high-order statistical moments due to its flexibility.\n - CoFiE preserves interpretability of outputs by using a well-known statistical formula (Cornish-Fisher expansions)." }, { "heading": "Evaluation strategy", "content": "- Comparative frameworks:\n - Monte Carlo simulation.\n - Real data applications across different asset classes, with a focus on foreign exchange markets.\n- Benchmark models for comparison:\n - EGARCH-t model.\n - Extreme Value Theory model.\n - Three conventional models in total (CoFiE-NN compared against these conventional approaches)." }, { "heading": "Key empirical findings", "content": "- Performance:\n - CoFiE-NN outperforms the conventional EGARCH-t model and the Extreme Value Theory model in several statistical criteria for both the simulated data and the real data.\n- Tail risk factor structure (under CoFiE-NN):\n - Only 20 percent of tail risk dynamics across 22 currencies is explained by one common factor.\n - In contrast, 60 percent of volatility dynamics across the same 22 currencies is explained by one common factor.\n- New empirical measure:\n - Introduction of a new empirical proxy for tail risk named \"tail risk ratio\" under CoFiE-NN." }, { "heading": "Contributions and implications", "content": "- Methodological contribution: Integration of asymptotic expansions (Cornish-Fisher) with neural-network-based forecasting to balance flexibility and interpretability.\n- Practical implication for risk monitoring:\n - CoFiE-NN provides improved VaR forecasts relative to standard EGARCH-t and Extreme Value Theory approaches in the reported evaluations.\n - The low common-factor share (20 percent) for tail risk across currencies implies substantial idiosyncratic tail dynamics that may necessitate currency-specific tail-risk monitoring and management, unlike volatility where a large common factor (60 percent) explains much of the variation." }, { "heading": "Keywords and technical terms", "content": "- Machine learning\n- Neural Network\n- Value-at-Risk\n- LSTM\n- Cornish-Fisher expansions (CoFiE)\n- EGARCH-t\n- Extreme Value Theory\n- Tail risk ratio\n\nForecasting Tail Risk via Neural Networks with Asymptotic Expansions — By Yuji Sakurai, Zhuohui Chen; May 10, 2024; 38 pages.\n\n---\n\n Content in this bundle\n\n- Working Paper\n - Working Paper (Markdown version){rel=\"alternate\" type=\"text/markdown\"}\n - Working Paper (PDF){rel=\"external\" type=\"application/pdf\"}\n\n---\n\nSource: https://www.imf.org/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841" } ], "bullets": [ "[Markdown version](/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/index.md)", "[Structured JSON version](/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/index.json)", "[Bundle manifest](/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/bundle-manifest.json)", "Authors: Yuji Sakurai, Zhuohui Chen", "Published: May 10, 2024", "Series: IMF Working Papers", "DOI: https://doi.org/10.5089/9798400276637.001", "Proposal: CoFiE-NN — a machine-learning-based approach for forecasting Value-at-Risk (VaR) that combines a neural network (NN) with Cornish-Fisher expansions (CoFiE).", "NN specification: Long Short-Term Memory (LSTM) is employed as the main NN specification.", "Rationale:", "Comparative frameworks:", "Benchmark models for comparison:", "Performance:", "Tail risk factor structure (under CoFiE-NN):", "New empirical measure:", "Methodological contribution: Integration of asymptotic expansions (Cornish-Fisher) with neural-network-based forecasting to balance flexibility and interpretability.", "Practical implication for risk monitoring:", "Machine learning", "Neural Network", "Value-at-Risk", "LSTM", "Cornish-Fisher expansions (CoFiE)", "EGARCH-t", "Extreme Value Theory", "Tail risk ratio", "**Working Paper**" ], "related": [ { "title": "Working Paper", "role": "paper", "sourceUrl": "https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024099-print-pdf.pdf", "summary": { "path": "/-/media/files/publications/wp/2024/english/wpiea2024099-print-pdf.pdf.md", "mime": "text/markdown" }, "binary": { "path": "/-/media/files/publications/wp/2024/english/wpiea2024099-print-pdf.pdf", "mime": "application/pdf" } } ], "alternates": { "markdown": "/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/index.md", "json": "/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/index.json", "bundleManifest": "/en/publications/wp/issues/2024/05/10/forecasting-tail-risk-via-neural-networks-with-asymptotic-expansions-548841/bundle-manifest.json" }, "generatedAtUtc": "2026-09-17T00:01:51.099Z" }