FRTB Internal Models Approach (IMA) - Part 2: Risk Factor Eligibility Test (RFET)
In my previous article, I had attempted to explain everything up to PLAT and Backtesting from below diagram. In this part I will illustrate step 3 in detail which is RFET i.e. Risk Factor Eligibility Test.
Once the firm/bank and desk(s) in question have passed PLAT and Backtesting: next step is to check if the risk factors (on which we should calculate Expected Shortfall) are 'modelable' or not i.e. it needs to make sure the risk factor is suitable for their model.
But what the hell is a 'risk factor'?
When a bank trades in financial markets, it faces risks that the value of its trading positions will decrease due to changes in market rates and prices. To manage these risks, the bank creates a model, that identifies and monitors the market rates and prices that affect its trading positions. These market rates and prices are called risk factors.
The bank's risk management model must include an appropriate set of risk factors, that are representative of the risks present in the bank's trading portfolio. This means that the model should capture all the important factors, that could cause changes in the value of the bank's trading positions. By monitoring these risk factors, the bank can identify potential risks and take measures to mitigate them.
When would the risk factor be considered 'modellable'?
Regulator is not ready to believe that bank's risk factors are modellable just because aa report says so. To do this, bank must perform a test called the Risk Factor Eligibility Test (RFET). The RFET requires that they have enough 'real prices' that accurately represent the risk factor. A real price is a price that meets at least one of the following conditions:
- Historical cost: Banks can value an asset based on the actual price they paid for it. This is a straightforward method.
- Market price: Banks can value an asset based on the price of a similar asset in an active and transparent market. This method is useful when a market exists for the asset and the price is publicly available.
- Model-based valuation: Banks can use models that estimate the value of an asset based on market data, such as interest rates or credit spreads. This method is commonly used for complex assets, such as derivatives.
- Vendor pricing: Banks can use prices provided by third-party vendors who specialise in valuing financial assets. These vendors collect data from various sources and use proprietary models to calculate asset values.
To be considered "real" and "usable" in the model, a risk factor (a variable that affects an investment's value) must have at least 24 actual prices per year, with at least four price changes every 90 days.
Why so much fuss? Is this not easy?
It's not that easy really. Banks must gather all relevant price data, organize it into categories that match the risk factors, and make sure it meets certain quality standards.
They need to test the data to ensure it is accurate and usable, and then use it to run the RFET (Risk Factor Eligibility Test). Banks need to keep monitoring the test results and watch out for risk factors that are not usable in the model. To prevent this, they can try to gather more information from other sources that provide committed quotes.
领英推è
Why should the bank bother? They have already passed PLAT and BackTesting?
This is important as not only do non-modellable risk factors (NMRF) result in a higher capital charge but also using more detailed risk factors can help improve the accuracy of the Profit Loss Attribution test, which is needed to check if the risk model is working correctly.
Bank will fail RFET; so what?
RFET aims to provide assurance that the risk factors included in IMA are sufficiently liquid and observable. If a bank fails RFET, the ES calculation will be done on 'stressed' data resulting into higher capital charge.
FRTB made a new rule called the NMRF framework to prevent banks from underestimating risk when trading in illiquid markets (where it's hard to buy or sell quickly). This rule says that banks must have enough evidence that the markets they trade in are liquid enough to use internal models to measure the risk. This applies to positions in their regulatory trading book that are capitalized using approved internal models.
Why NMRF should have a higher capital charge?
Banks use past data to build models that estimate the amount of risk they are taking in the markets. But if they are trading in markets where there isn't much activity, the data they use to build the model can become outdated, making it look like the risk is low, when really it's just because there hasn't been much activity. This can be a problem because it can lead banks to underestimate their actual risk. Hence, regulator provisions/slaps capital add-on.
In NMRF each risk factor has to be capitalised using a stress scenario calibrated to be at least as prudent as ES at 97.5% calibrated over an?extreme stress period. For each non-modellable risk factor, the liquidity horizon of the stress scenario must be the greater of the largest time interval between two consecutive price observations over the prior year and the liquidity horizon assigned to the risk factor as below. Meaning that we may end up with longer liquidity horizons than above for specific risk factors, resulting in a higher stress scenario charge.
In addition no correlation or diversification effect between other non-modellable risk factors is permitted.
Meaning that the aggregation of stress results will?lead to a much higher charge than aggregation for modellable risk factors under ES, where diversification is permitted within a risk class.
In next parts of the series, I will attempt to deep dive into below Mathematically and Logically
- Difference between VaR and ES.
- Reduced and Full Set of Risk Factors.
- Difference between ES (for MRF) and Stressed ES (for NMRF).
- DRC for IMA.
- Overall Aggregated Capital Charge
Feel free to connect/follow or revisit this space. I will put backlinks accordingly.