Paper: arXiv 2404.14136
Authors: Tobias Fissler, Fangda Liu, Ruodu Wang, Linxiao Wei
Abstract
Tail risk measures are fully determined by the distribution of the underlying loss beyond its quantile at a certain level, with Value-at-Risk, Expected Shortfall and Range Value-at-Risk being prime examples. They are induced by law-based risk measures, called their generators, evaluated on the tail distribution. This paper establishes joint identifiability and elicitability results of tail risk measures together with the corresponding quantile, provided that their generators are identifiable and elicitable, respectively. As an example, we establish the joint identifiability and elicitability of the tail expectile together with the quantile. The corresponding consistent scores constitute a novel class of weighted scores, nesting the known class of scores of Fissler and Ziegel for the Expected Shortfall together with the quantile. For statistical purposes, our results pave the way to easier model fitting for tail risk measures via regression and the generalized method of moments, but also model comparison and model validation in terms of established backtesting procedures.
Complexity vs Empirical Score
- Math Complexity: 9.2/10
- Empirical Rigor: 2.5/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper is highly theoretical, establishing abstract results on joint identifiability and elicitability of tail risk measures using advanced measure theory and rigorous mathematical proofs, with no implementation details, datasets, or backtests provided in the excerpt.
Research Flowchart
flowchart TD A["Research Goal:<br>Establish joint identifiability and<br>elicitability of tail risk measures<br>with the quantile"] --> B["Methodology:<br>Law-based risk measure generators<br>on tail distribution"] B --> C["Data/Inputs:<br>Loss distribution beyond<br>quantile at level α"] C --> D["Computation:<br>Construct weighted scores<br>for tail expectile & quantile"] D --> E["Outcomes:<br>1. Novel weighted score class<br>2. Easier model fitting (regression, GMM)<br>3. Model validation (backtesting)"] style A fill:#e1f5fe style E fill:#f1f8e9
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper establishes joint identifiability and elicitability results for tail risk measures together with the corresponding quantile, and provides the tail expectile as an example.
Source sentence
This paper establishes joint identifiability and elicitability results of tail risk measures together with the corresponding quantile, provided that their generators are identifiable and elicitable, respectively. As an example, we establish the joint identifiability and elicitability of the tail expectile together with the quantile.
Assumes 4
- The set A = {(VaR‑p(F), ρ(F)) : F ∈ M} has a simply connected interior. statistical
Source sentence
(i) Let A = {(VaR− p (F), ρ(F)) : F ∈M} ⊆R2 and suppose that the interior of A, int(A) is simply connected.
- The generator ρ* is identifiable with a strict M‑identification function that is locally bounded. statistical
Source sentence
(ii) The generator ρ∗is identifiable with a strict M-identification function Vρ∗: R2 →R, which is locally bounded.
- The expectation of the identification function V(v,x,·) with respect to any F ∈ M is continuously differentiable. statistical
Source sentence
(iii) The identification function V : R3 →R2 defined by V (v, x, y) = 1{y⩽v} −p 1{y>v}Vρ∗(x, y) is such that its expectation with respect to any F ∈M is continuously differentiable.
- The expectation of the score S with respect to any F ∈ M is twice continuously differentiable. statistical
Source sentence
(iv) The expectation of the score S with respect to any F ∈M is twice continuously differen-tiable.
Data
No dataset named in the text.
Costs
Not applicable
Finds 4
- Joint identifiability and elicitability of tail risk measures are established, provided their generators are identifiable and elicitable. theoretical
Source sentence
This paper establishes joint identifiability and elicitability results of tail risk measures together with the corresponding quantile, provided that their generators are identifiable and elicitable, respectively.
- The tail expectile together with the quantile is shown to be jointly identifiable and elicitable. theoretical
Source sentence
As an example, we establish the joint identifiability and elicitability of the tail expectile together with the quantile.
- The resulting consistent scores form a novel class of weighted scores that nest the known class of scores for Expected Shortfall and the quantile. theoretical
Source sentence
The corresponding consistent scores constitute a novel class of weighted scores, nesting the known class of scores of Fissler and Ziegel for the Expected Shortfall together with the quantile.
- The main results connect identifiability or elicitability of a tail risk measure ρ with that of its generator ρ*, and provide explicit conversions between their identification and score functions. theoretical
Source sentence
7 Concluding remarks Our main results in Sections 4 and 5 establish an intimate connection between identifiability or elicitability of a tail risk measure ρ and that of its generator ρ∗, as well as their identification functions and score functions.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 21 related papers). Coverage is partial, so this is neither replication nor refutation.
2 papers in the archive : Risk-sensitive Reinforcement Learning Based on Convex …
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (42)
- Put Option Premiums and Coherent Risk Measures (2002) — link
- Quantiles as optimal point forecasts (2010) — link
- Coherent Measures of Risk (1999) — link
- Strictly Proper Scoring Rules, Prediction, and Estimation (2007) — link
- Asymmetric Least Squares Estimation and Testing (1987) — link
- Elicitable distortion risk measures: A concise proof (2015) — link
- Comparing Density Forecasts Using Threshold- and Quantile-Weighted Scoring Rules (2010) — link
- Making and Evaluating Point Forecasts (2011) — link
and 34 more
- On elicitable risk measures (2015) — link
- Quantile Regression (2005) — link
- COHERENCE AND ELICITABILITY (2014) — link
- Beyond Value‐at‐Risk: GlueVaR Distortion Risk Measures (2013) — link
- Eliciting properties of probability distributions (2008) — link
- DISTRIBUTION‐INVARIANT RISK MEASURES, INFORMATION, AND DYNAMIC CONSISTENCY (2006) — link
- Should Scoring Rules be “Effective”? (1985) — link
- Conditional Extremes in Asymmetric Financial Markets (2018) — link
- Risk measures with the CxLS property (2015) — link
- Gini-Type Measures of Risk and Variability: Gini Shortfall, Capital Allocations, and Heavy-Tailed Risks (2016) — link
- Order-sensitivity and equivariance of scoring functions (2019) — link
- Focusing on regions of interest in forecast evaluation (2017) — link
- What is the best risk measure in practice? A comparison of standard measures (2015) — link
- On the Measurement of Economic Tail Risk (2016) — link
- Loss-based risk measures (2013) — link
- Risk Functionals With Convex Level Sets (2018) — link
- Forecast evaluation of quantiles, prediction intervals, and other set-valued functionals (2021) — link
- Correction note: Higher order elicitability and Osband’s principle (2021) — link
- Coherence and elicitability — link
- Higher order elicitability and Osband’s principle (2016) — link
- Systemic Risk Measurement: Multivariate GARCH Estimation of CoVaR (2012) — link
- Quantile-Based Risk Sharing (2018) — link
- Robustness and sensitivity analysis of risk measurement procedures (2010) — link
- Conditional value-at-risk for general loss distributions (2002) — link
- A Theory for Measures of Tail Risk (2021) — link
- Backtesting Systemic Risk Forecasts Using Multi-Objective Elicitability (2023) — link
- Bayes Risk, Elicitability, and the Expected Shortfall (2020) — link
- Robust Statistics (2009) — link
- Inf-Convolution, Optimal Allocations, and Model Uncertainty for Tail Risk Measures (2022) — link
- Tail Variance Premium with Applications for Elliptical Portfolio of Risks (2006) — link
- CoVaR (2016) — link
- Stochastic Finance (2016) — link
- Osband’s principle for identification functions (2023) — link
- Characterizing M-estimators (2023) — link
0 of 42 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.