To gauge their risk, banks as well as investors such as pension funds and insurance companies, need to report to regulators on their solvency. To this they need to report on the risk of the positions they hold, typically in terms of Value at Risk (Var) or, in the case of the new bank solvency reporting framework FRTB (Fundamental Review of the Trading Book), in terms of Expected Shortfall (ES).
Creating Accurate Risk Factors Out of Market Data
Value at risk is defined as how much a set of investments might lose (with a given probability), given normal market conditions, in a set time period such as a day. It provides an upper bound on losses for say 95% or 99% of the trading days but doesn’t say how much will be lost in the remaining 5% or 1% of occasions.
Expected shortfall is defined as the expected return on the portfolio book of record in the worst 1% or 5% of cases. It is an alternative to VaR that is more sensitive to the shape of the tail of the loss distribution and gives a better view on what could happen when it is not ‘business as usual’.
Risk factors are the value drivers for portfolios and can be individual instrument prices (e.g. stock prices), indices or summary stats driving a certain market such as interest rate curves, credit spread curves or volatility smiles. To be able to do this statistical analysis and get to metrics such as VaR or ES, firms need historical data on their risk factors. Risk data management can be seen as a special case of market data management.
Time Series Data Management
To produce the time series the market risk department needs, data can be collected from data sources including exchanges, brokers, market makers and data aggregators. In case of missing data, this data can be extended using evaluated prices or proxy prices using comparable instruments.
Sometimes time series need to be backfilled to get sufficient history. This can occur when a company starts trading a newly IPOed company that lacks historical data. Some risk factors are not directly observed market data but need to be inferred. For example, a 10 year bond yield can be taken from a bond curve fitted through bond yields from a number of currently traded bonds. A curve is created with a number of tenor points. Similarly, volatilities can be inferred from option prices to create a volatility smile or surface. Good reference data management is needed as well: terms and conditions of products are required for valuation and to classify instruments for different risk factors.
Market Data Needs for Risk Regulation
On top of the market data, several reference data finance fields need to be populated. This includes instrument classification to properly identify the risk class, for example general interest rate risk (GIRR), different types of credit spread risk (CSR), equity risk, commodity risk and foreign exchange (FX) risk. Also, when risk factors such as curves and smiles are calculated, it is important to track the method or model used as well as the parameters such as interpolation method or proxy rule. Other than the risk factors themselves, modified risk factors are needed to. With this we mean the application of shifts and shocks to risk factors to do impact analysis given certain stress-test scenarios. In addition, banks that use their own models under FRTB have extra market data needs. They require “real prices”, that is to say actual observed prices in transactions for their risk factors. All in all, the market data needs for risk management are complex.
State of the art market data management solutions will provide firms with integrated data sourcing, cross-referencing, financial data quality management and risk factor completion as well as with integration with leading risk platforms such as Murex.