A large scope of indicators is available, each measuring a specific risk, and some specific to an underlying exposure. Bespoke solutions are generally required to match the risk profile of the client.
When it comes to the risk management of investment assets, either individually or on a portfolio scale, there are a number of choices to be made relative to the investment appetite of the firm, and the risk profile of the underlying exposures. In light of the continually increasing regulatory requirements on both financial and peripheral industries, risk management practices and their evolution are firmly on the radar of senior management, board members, auditors and other stakeholders.
Implications to capital and risk management continue to evolve and at times converge under MiFID II, Solvency II, CRR/CRD-IV, PruVal and Basel III, but there is undoubtedly a greater focus on more granular risk management reporting, and the quality of the underlying analytics and models used to create the metrics.
The quantification of risk is a key step towards the management and mitigation of risk, whatever the industry, and there are many approaches to consider. Below we discuss some key metrics relating to market risk and credit risk. Other processes to quantify risk include Stress Testing, Scenario Analysis, assessment of Economic Capital etc. These concepts are introduced within the paper titled “Capital Risk Management Tools and Metrics.”
Market Risk Metrics
We consider the two most commonly used metrics:
Value at Risk (VaR)
From a regularity standpoint, financial service firms are required to use VaR as a core metric in setting minimum capital requirements, first introduced by the Securities and Exchange Commission (SEC) in the 1980s. Recently, VaR is increasingly accepted as a key tool in measuring risk-exposure for financial and non-financial firms.
VaR is a single, summary, statistical measure of possible portfolio losses, due to normal market movements. It is defined as the loss level that will not be exceeded with a certain confidence level during a particular period of time. This aims to capture the maximum expected loss on a portfolio.
As an example, if the 10-day, 95% VaR of a portfolio is £1 million, then it is considered that there’s a 5% chance that losses will exceed £1 million over the 10-day period.
VaR provides a single, easily-digestible metric that reflects the riskiness and diversification benefits for a portfolio given severe perturbations to underlying prices or curves.
There are, however, shortcomings of VaR. For instance, it is typically derived based on assumed (normal market) distribution conditions, and is not a very good predictor of extreme situations and outcomes with fat tails. It may also under estimate or overestimate diversification benefits, and may not be able to capture quickly and dynamically changing fundamentals that might alter the short term returns and correlations across asset classes.
Risks that are not quantified in a VaR methodology such as proxy risks, basis risks, risks from calibration parameter errors, and other higher order risks are usually scoped out separately using a “Risks not in VaR (RNIV)” framework.
Expected Tail Loss (ETL)
Since VaR is not without its failings, and provides a contextual starting point for risk assessments, other metrics and measures have been developed to sit alongside. One of them is the Expected Tail Loss (ETL), sometimes referred to as Conditional VaR (CVaR) or Expected Shortfall (ES), which aims to quantify extreme losses. Where VaR aims to capture how bad things can get, the ETL provides a quantification of what the expected loss will be in a tail event. ETL is the expected loss during a determined period of time, with a percentile probability that puts it further out on the tail.
For e.g. following on from the VaR example before of 95% confidence and a 10-day horizon, an ETL could entail the expected shortfall over the same 10-day period, however at a 99th% percentile of the loss distribution. In other words, the chances of the loss getting any worse would be 1%.
Credit Risk Metrics
For Illiquid investments such as debt, loans, and other contingent claims, it is necessary to ascertain the likelihood and costs of default. As such, there are a number credit risk metrics that are used to measure this. Below we review a few. Additional metrics are reviewed within the paper titled “Counterparty Risk Advisory Solutions”.
Loss Given Default (LGD)
In the event of default, the loss (or value recovered) is of key importance to the valuation. For synthetic credit products this is contractual, for anything else this depends on the market. Modelling this requires consideration of the seniority of the assets, the industry, the issuer, historical recoveries and trends, and an assessment of the current climate, as well as the assets and liabilities of the issuer.
Probability of Default (PD)
Probability of Default is the measure of the likelihood of an entity defaulting over a particular time horizon. Fundamentally this depends on the on the credit spread of the issuing entity, in conjunction with the LGD. For liquid issuers, this can be derived from credit default swap curves, but for less liquid issuers it is necessary to assess the contributing components of credit risk to establish a credit spread. This is therefore dependent on the region, industry and state of the issuer’s business, interest rate and economic climate.
Exposure at Default (EAD)
The EAD defines the value of the investment at risk in the event of default, specifically, the maximum exposure to loss at that time. It is quite possible that current exposure and future exposures will be different, and as such this is linked to the fair valuation of the assets (and liabilities) in question. The calculation of this requires the modelling of the investment value now, and over the duration of the investment life, which in turn depends on the evolution of the market factors upon which the value of the investment depends.
The Complex Asset Solutions practice at Kroll provides an expert team of quantitative analysts together with a library of cutting-edge valuation and risk models to assist our clients with enhancing their market and credit risk management practices.
Specifically, our areas of expertise include:
- Computation and valuation of VaR and ETL for client portfolios
- Estimation of PD, LGD and EAD for all categories of asset types
- Providing analytics and subsequent support for integration with in-house systems
- Assistance and advisory work in implementation of market and credit risk metrics
- Model validation and assurance for client’s market and credit risk management models, calibration parameters, input data
- Calculation, Reporting of “Greeks”
- Back testing of models
- Development of a RNIV framework
- Training on best practices