Abstract:Extremal quantile regression, i.e. quantile regression applied to the tails of the conditional distribution, counts with an increasing number of economic and financial applications such as value-at-risk, production frontiers, determinants of low infant birth weights, and auction models. This chapter provides an overview of recent developments in the theory and empirics of extremal quantile regression. The advances in the theory have relied on the use of extreme value approximations to the law of the Koenker and Bassett (1978) quantile regression estimator. Extreme value laws not only have been shown to provide more accurate approximations than Gaussian laws at the tails, but also have served as the basis to develop bias corrected estimators and inference methods using simulation and suitable variations of bootstrap and subsampling. The applicability of these methods is illustrated with two empirical examples on conditional value-at-risk and financial contagion.
What problem does this paper attempt to address?
This paper aims to explore the theory and application of Extremal Quantile Regression (EQR). Specifically, the paper focuses on the quantile regression method applied to the tails of conditional distributions, that is, quantile regression when the conditional probability index \(\tau\) is close to 0 or 1. This research area has wide applications in economics and finance, such as Value - at - Risk (VaR), production frontier analysis, factors influencing low infant birth weight, and auction models.
### Main issues
1. **Theoretical development**:
- The paper reviews the theoretical progress of Extremal Quantile Regression in recent years, especially how to use Extreme Value Theory (EVT) to improve the performance of high - tail quantile regression estimators.
- By using extreme value approximation, these theories not only provide more accurate tail estimates than Gaussian approximation, but also lay the foundation for developing bias - corrected estimators and inference methods, such as simulation, bootstrapping and subsampling techniques.
2. **Practical applications**:
- The paper demonstrates the application of these methods through two empirical examples, namely Conditional Value - at - Risk (CVaR) and Financial Contagion.
- Among them, CVaR is a commonly used risk measurement method for predicting or explaining the low quantiles of future portfolio returns. Financial Contagion studies how extreme market events spread among different financial markets.
### Theoretical framework
1. **Extremal Quantile Regression model**:
- It is assumed that the tail distribution of the response variable \(Y\) has Pareto - type behavior, that is, the tail decay is close to the power function form.
- The model assumes that the conditional quantile function \(Q_Y(\tau|x)\) has a linear parametric form, that is, \(Q_Y(\tau|x)=B(x)'\beta(\tau)\), where \(B(x)\) can include polynomials of \(x\), splines, indicator variables or interaction terms.
2. **Estimation and inference methods**:
- The paper discusses the estimation methods based on the Koenker and Bassett (1978) quantile regression estimator, including analytical methods and resampling methods.
- It introduces methods such as bootstrapping and subsampling to calculate the quantiles of the limiting extremal distribution for statistical inference and bias correction.
### Application examples
1. **Conditional Value - at - Risk (CVaR)**:
- Predict the low quantiles of future portfolio returns \(Y\) through current information \(X\), usually focusing on the extreme quantile regression functions with \(\tau = 0.01\) and \(\tau = 0.05\).
- This method is widely used in financial risk management, insurance and actuarial science.
2. **Financial Contagion**:
- Study how extreme market events spread among different financial markets, and analyze the spread paths and influence degrees of these events through the Extremal Quantile Regression method.
In conclusion, through theoretical analysis and empirical research, this paper systematically explores the application of Extremal Quantile Regression in dealing with extreme events, and provides important tools and methods for risk management and policy - making in economics and finance.