Journal of Risk

Farid AitSahlia
Warrington College of Business, University of Florida

The increasing frequency and severity of weather events attributable to climate change are affecting all corners of the economy. This issue of The Journal of Risk opens with a paper analyzing their impact on financial risk management specifically. This is followed by contributions regarding the accuracy of quasi-Monte Carlo simulation, the interplay between investor sentiment and downside liquidity risk, and the risk assessment of cryptocurrency returns.

In the issue’s first paper, “Estimating the impact of climate change on credit risk”, Stuart M. Turnbull uses meteorological data to infer probabilities of default via an analytical approach and, via Monte Carlo simulation, estimates of value-at-risk and expected shortfall for loan portfolios. While his study illustrates the marked impact of a changed climate, Turnbull further assesses the credit risk due to climate change based on scenarios capturing various transition strategies reflecting regulatory differences between sovereign entities on the path to a low-carbon economy.

Our second paper, “The importance of being scrambled: supercharged quasi- Monte Carlo” by Sergei Kucherenko and Julien Hok, develops randomized quasi- Monte Carlo methods with the goal of improving their convergence and providing associated practical error bounds. They show in particular that the combination of Sobol’s low-discrepancy sequences and Owen’s scrambling, together with an effective dimension-reduction technique such as a Brownian bridge or principal component analysis, significantly improves the efficiency of quasi-Monte Carlo simulation.

Next, in “Research on the premium for the joint lower-tail risk of liquidity and investor sentiment”, Yuting Hou, Xiu Jin and Weiqiang Huang conduct an empirical study to estimate the premium associated with the joint occurrence of downside liquidity and extreme negative sentiment. Using Chinese stock market data, they show that a joint measure of these dual risks through copulas captures information that would otherwise not be elicited based on their separate evaluations. The authors further suggest that, except for firm size, typical firm characteristics have no impact on this joint measure, but that the latter can accurately predict cross sections of returns.

The final paper in this issue is “Extremes of extremes: risk assessment for very small samples with an exemplary application for cryptocurrency returns” by Christoph J. Börner, Ingo Hoffmann, Jonas Krettek, Lars M. Kürzinger and Tim Schmitz, who propose a method that relies on scaling the statistical distribution of extreme values to assess risk based on very small data sets. They specifically illustrate how their approach can be applied to estimate cryptocurrency returns, for which stable distributions have been shown to capture fat tails and skewness.