Risk Model Blocks

RISK MODEL BUILDING BLOCKS Risk-return is the centrepiece of risk management processes. All bank systems provide common measures of income such as accrual revenue, fees and interest income for the banking portfolio and profit and loss (P&L) for the market portfolio. Measuring risk is a more difficult challenge. Risk models provide a measure of risk. The models have a common basic structure consisting of main building blocks and modules.

BLOCK I : STANDALONE RISK Block I requires risk and transaction data to capture individual transaction risks. Modules 1. Risk Drivers Risk drivers are those parameters whose uncertainty alters values and revenues. For ALM, the risk drivers are the interest rates that drive the interest income. For market risk, risk drivers are all market parameters that trigger changes in the value of the transactions. They differ across the market compartments: fixed income, foreign exchange or equity. The credit risk drivers are all parameters that influence the credit standing of counterparties and loss for the bank if it deteriorates. They include exposure, recoveries under default, default probability (DP) and its changes( migration risk) when time passes. Risk factors alter risk drivers which then directly influence risk events such as default or change in the value of the transaction. In risk models, risk drivers directly relate to risk events while risk factors serve for modeling risk drivers.

• For risk drivers such as interest rates, instability over time is readily observable. Interest rate models can also be used to derive multiple scenarios of interest rates that are consistent with their observed behaviour. In case of credit risk, ratings are obtained from rating agencies or from internal rating systems. Rating models attempt to mimic ratings from agencies through some observable characteristics of firms such as financial data. Default probability models such as KMV’s Credit Monitor or Moody’s RiskCalc provide modeled estimates of default probabilities.

2. Exposures and Valuation • Exposure size is denoted by book value or notional amount (for derivatives). Economic valuation is the fair value- risk and revenue adjusted. It requires mapping exposures to relevant risk drivers and deriving values from them. Because it is risk adjusted, any change in the risk drivers materializes in a value change, favourable or unfavourable. • ALM uses both book values and economic values. ALM’s first stage is to consolidate all banking portfolio exposures by currency and type of interest rates, thereby capturing the entire balance sheet view from the very beginning.

• For market risk, exposures are mark to market and change continuously with market movements. Exposures for credit risk are at book value for commercial banking. Since book values do not change when default probabilities change, economic values serve for full valuation of credit risk. Exposure data include the expected size under a default event [exposure at default (EAD)] plus the recoveries that result in a loss given default (Lgd) lower than exposure.

3. Standalone Risk Valuation • Standalone risk valuation looks at future events and involves marking to future for events such as deviation of interest rates for the banking portfolio, all market parameters for market risk and all factors influencing the credit standing of borrowers for credit risk. Marking to future is the technical process for calculating potential values at a future date and deriving losses from this distribution (since future values are uncertain, they follow a distribution at this horizon). • Standalone risk does diversification effect.

not

take

into

account

the

BLOCK II : PORTFOLIO RISK The goal of Block II is to capture the risk profile of the portfolio rather than the risk of single transaction considered in isolation. 4. Correlations Diversification requires new inputs characterising the portfolio risk, mainly the sizes of individual exposures and correlations between risk events of individual transactions. The adverse deviations of interest income, a usual target variable for ALM policies, result from interest rate variations. The correlation between interest rates tends to reduce the diversification effect. All interest revenues and costs tend to move simultaneously in the same direction. Asset and liability exposures to interest rates are offset to the extent that they relate to the same interest rate reference.

• The adverse deviations of the market value of trading portfolio correlate because they depend on a common set of market parameters, equity indexes, interest rates and foreign exchange rates. While some of the risk drivers are highly correlated, the others are not and sometimes they vary in the opposite directions giving rise to diversification effect. Moreover positions in opposite directions offset.

• Credit defaults and migrations tend to correlate because they depend on the same general economic conditions although with different sensitivities. Portfolio risk is very sensitive to such correlation. The opposite of diversification is concentration. If all the firms in the same region tend to default together, there is a credit risk concentration in the portfolio because of very high default correlation. Granularity designates risk concentration due to size effect. A single exposure of Rs. 1,000 might be riskier than 10 exposures of Rs. 100 each, even if the borrowers’ defaults correlate. A small number of losses is much more frequent than a simultaneous failure of all of them. There is also no diversification for the large exposure whereas there is some across 10 small exposures.

• Obtaining correlations requires modeling. Correlation between risk events of each individual transaction results from the correlation between their risk and value drivers. Correlations for market risk drivers and interest rates for ALM are directly available. Correlations between credit risk events such as defaults suffer from a scarcity of data. This difficulty is overcome by inferring correlations between credit risk events from the correlations of observable risk factors that influence them.

5.Economic Capital VaR is also the economic capital at a preset confidence level. After obtaining the correlations, the next step is to characterise the downside risk of the portfolio. Since the portfolio is subject to various levels of losses, its risk profile can be characterised with the help of distribution of losses, assigning probability to each level of portfolio loss. From the loss distributions, the main risk characteristics extracted are the expected loss, the loss volatility i.e. the dispersion around the mean and the loss percentiles at various confidence levels for measuring downside risk and VaR.

BLOCK III : REVENUE & RISK ALLOCATION Calculation of overall portfolio risk and required global return are not sufficient to develop risk management practices within business lines and for decision making purposes. Moving downwards to risk processes requires other tools. They are top-down links sending signals to business lines and the bottom up links consolidating and reporting. Two basic devices serve for linking global targets and business policy:

• The Funds Transfer Pricing (FTP) System It allocates income to individual transactions, business lines, product families or market segments. Transfer prices are internal references used to price financial resources across business units. Transfer pricing applies essentially to the banking book. The interest income allocated to any transaction is the difference between the customer rate and internal reference rate or the transfer price applicable to that transaction. Transfer prices also serve as a reference for setting customer rates. They should represent the economic cost of making the funds available.

• The Capital Allocation System It allocates risk to any portfolio subset and down to individual transactions. Portfolio risk models help determine the overall portfolio risk taking into account the diversification effects. But the problem is to divide it into risk retained by each transaction, after diversification effects. This is because pre-diversification risks of individual transactions do not arithmetically add up to diversified portfolio risk. For this purpose Capital Allocation Model is required.

• The Capital Allocation Model assigns to any subset of a portfolio, a fraction of the overall risk called the ‘risk contribution’. Risk contributions are much smaller than ‘Standalone’ risk measures, or measures of the intrinsic risk of transactions considered in isolation before diversification effects because they measure only a fraction of risk retained by transactions postdiversification. The Capital Allocation System allocates capital in proportion to the risk contribution.

• Example: • The standalone risks of three transactions are 20,30 and 30 adding up to 80. However the portfolio model shows the VaR at 40. The capital will be allocated in the ratio of 2:3:3 i.e. 10,15,15 to the three transactions. • • The FTP and Capital Allocation Systems provide the necessary link between global portfolio management and the business decisions. They complement each other in obtaining risk adjusted performance and lead to the ultimate stage of defining risk and return profiles, risk based performance and pricing

BLOCK IV : RISK AND RETURN MANAGEMENT Risk Adjusted Performance Measurement (RAPM) designates the risk adjustment of individual transactions that allows comparison of their profitability on the same basis. It is an ex-post measure of performance. RAPM simply combines the FTP and the Capital Allocation Systems. For example: Transa Capital Reve Risk adjusted ction Allocati nue return on on capital (RaRoc) A 5 0.5 0.5/5 = 10% B 3 0.4 0.4/3= 13.33% The less risky transaction B has a better risk adjusted profitability than the riskier transaction A even though B’s revenue is lower.

• Risk based Pricing (RBP) relates to ex-ante decision making, when we need to define the target price ensuring that the transaction profitability is in line with both risk and the overall target profitability of the bank.

• Risk Modeling and Risk Decisions All model blocks lead ultimately to risk and return profiles for transactions, sub-portfolios of business units, product families or market segments plus the overall bank portfolio. The purpose of characterising this risk-return profile is to provide essential inputs for decision making. The spectrum of decisions include: New transactions Hedging both individual transactions (micro-hedges) and sub-portfolios (macro-hedges) Risk return enhancement of transactions through restructuring etc. Portfolio management consisting of enhancing the portfolio risk-return profile. • The spectrum includes both on-balance sheet and off-balance sheet decisions. On balance sheet actions refer to any decisions altering the volume, pricing and risk of transactions while off- balance sheet decisions refer more to hedging risk.