SERCAN DEMİRALAY FINANCIAL RISK MANAGEMENT

What is Risk? Risk is generally defined as ‘’uncertainity about future.’’ The term ‘’risk’’ in terms of banking sector is the probability of facing losses because of various unexpected reasons.

Knight’s Distinction between Risk and Uncertainty Risk is probabilistically knowable and therefore manageable while uncertainty is unknowable. It has been remarked that in Knight’s distinction, “Risk relates to objective probabilities, uncertainty relates to subjective probabilities. Risk management relies on knowledge/observations of past and present states of the world but must assume a view of the (probable) future.

Modern Portfolio Theory (Markowitz) “… the rule that the investor does (or should) consider expected return a desirable thing and variance of return an undesirable thing.” Markowitz suggests a portfolio investment decision rule based on expected return and variance of return: Investors should weigh individual equities in such a manner (by taking the covariances between individual stocks into account) that, for a given desired level of return, the portfolio variance is minimized. This is referred to as mean-variance optimization.

Modern Portfolio Theory (Markowitz)

CLASSIFICATION OF RISKS

CLASSIFICATION OF RISKS Market risk is the day to day potential for an investor to experience losses from flactuations in securities prices. Market risk cannot be diversified away and it also referred to as ‘’systematic risk’’. Market risk is divided into equity risk, interest rate risk, currency risk and commodity risk. Credit risk is risk of loss resulting from failure to obligors to honor their payments. Credit risk can also be described as potential defaults by borrowers, counterparties in transactions.

CLASSIFICATION OF RISKS Actual defaults are not the only sources of credit risk, there are other sources of credit risk which derives from perceived or actual changes in the credit quality. Types of credit risk are transaction risk and concentration risk. Liquidity risk stemming from the lack of market ability of an investment that cannot be bought or sold quickly enough to prevent or minimize a loss. Liquidity risk can also be described as financial risk due to uncertain liquidity. An institution might lose liquidity if its credit rating falls. Liquidity risk is sub-divided into funding liquidity risk and asset liquidity risk.

RISK MEASUREMENT METHODOLOGIES

VALUE AT RISK It is a A technique used to estimate the probability of portfolio losses based on the statistical analysis of historical price trends and volatilities. There are many ways to measure VaR depending on the methodological decision made, either non-parametric, parametric or Monte-Carlo VaR. The main example of a non-parametric Var method is historical simulation. Historical simulation is the simplest way of measuring VaR for many portfolios. In non-parametrics approach, the VaR for a portfolio is calculated by setting a hypothetical time series of returns on the portfolio.

VALUE AT RISK Advantages & Disadvantages of Historic Simulation: Advantages : It makes no assumption about distribution because it is non-parametric. It relies on volatility and correlation embedded in time series. It captures fat tails(events) in price change distribution. Disadvantages: VaR incorporates only historic changes in price. It is data insentive which means that it needs many time series.

VALUE AT RISK Parametric VaR approaches are to measure the parameters of the underlying distribution to the observed data. . Two of the most important parametric approaches are based on the normal and student-t distrubitions. The main advantage of parametric approaches is their efficient use of data and the great amount of information they can provide; the downside is that this information can be seriously wrong if based on mistaken parametric assumptions.

Variance-Covariance Method It is an an approach which has benefits of simplicity but it is limited by the difficulties associated with deriving probability distributions. Example; When there is only one asset; Position : 1,000 shares IBM stock ($180 per share) Dollar value per basis point change in rate: 1,000 x $180 = $180,000 One week volatility: 10% Risk: $180,000 x 10% = $18,000

Variance-Covariance Method When there are two assets; Position : 1,000 shares IBM stock ($180 per share) and $10mm 10 year US treasury note Dollar value per basis point change in rate: 1,000 x $180 = $180,000 and $760/bp x 10 = $7600 /bp. (since $1mm 10 year USD swaps are equal to $760 basis points.) One week volatility: 10% and 10bps. Risk: $180,000 x 10% = $18,000 and $7600/bp x 10 = 76,000 Hence portfolio risk is ;

Variance-Covariance Method When there are three assets; Position : 1,000 shares IBM stock ($180 per share), $10mm 10 year US treasury note and 45 3-month Eurodollar futures Dollar value per basis point change in rate: 1,000 x $180 = $180,000 , $760/bp x 10 = $7600 /bp and $25/bp x 45 = $1125/bp One week volatility: 10% , 10bps and 6bps Risk: $180,000 x 10% = $18,000 , $7600/bp x 10 = 76,000 and -$1,125/bp x 6 = -$6,750. Portfolio risk is $88,875. Thus, the fact that as the numbers of assets rises, portfolio risk also goes up can be seen.

Variance-Covariance Method Advantages of Variance-Covariance Method: It gives results fast. It is relatively easy to implement. It requires only portfolio level sensitivities. It can be modified to capture some measure of convexity. Data sets are readily available. Disadvantages of Variance-Covariance Method: It does not revalue positions. It cannot account for complex or discontinuous payoffs. It cannot incorporate multiple time horizons. It assumes nornal or normal-like distributions.

MONTE CARLO SİMULATION It is a hybrid between analytic and simulation approaches. Instead of taking historical data as a starting point, Monte Carlo begins by estimating parametric distrubitions for the individual risk factors relevant to the simulation. Next, Monte Carlo generates simulated price paths for them. The resulting scenarios are used to fully valuate the to-be-simulated portfolio and to generate a returns distribution. From this distribution VaR can be derived at the desired confidence level.

MONTE CARLO SİMULATION Advantages of Monte Carlo Method: It produces a distrubition of price level changes. It allows for multiple periods with rehedging and maturation. It provides greatest level of control over price volatility. Disadvantages of Monte Carlo Method: It is mathematically intensive (scenario generation). It ıs parametric VaR which requires distrubition and correlation assumptions. It is less transparent.

VaR can be misleading There is no definite measurement of Value at Risk and all measurements come with its own limitations. The ending result is VaR computed for an asset, firm or a portfolio can be wrong and occasionally the errors can be large enough to make VaR a misleading measure of risk exposure. The causes of the errors can differ across firms and for different measurements and involve the followings: Return Distributions: Every VaR measurements makes assumptions about return distributions, if violated, work out incorrect calculations of the Value at Risk.

VaR can be misleading History may not be a good predictor: All measures of Value at Risk use historical data to some degree or the other. Non-stationary Correlations: Measures of Value at Risk are conditioned on explicit estimates of correlation across risk sources (in the variance-covariance and Monte Carlo simulations) or implicit assumptions about correlation (in historical simulations).

RISKMETRICS RiskMetrics was a free service which was offered by JP Morgan in 1994 to promote value at risk (VaR) as a risk management tool. In order to simplify the calculation, RiskMetrics matches positions to selected representative instruments such as various fixed income buckets, equity indexes, and commodity volatility series. Volatilities and correlations are then expressed and estimated respectively with reference to exponentially weighted daily historical observation adopting a decay factor of 0.94 in the case of trading and 0.97 in the case of investing.

CREDITMETRICS Launched by J. P. Morgan in 1997, CreditMetrics is a so-called credit migration approach to portfolio credit risk modelling; and it models the full forward distribution of the values of any bond or loan portfolio. To compute diversification benefits and concentration risks at the portfolio level, CreditMetrics models the correlations in obligors’ credit quality migrations on the joint probability of equity returns. CreditMetrics consists of three main components: Historical data sets A methodology for measuring portfolio Value at Risk (VAR) A software package known as CreditManager.

CREDITRISK+ CreditRisk+ is a methodology to calculate the distribution of possible credit losses from a portfolio and developed by Credit Suisse and launched in 1997. The CreditRisk+ methodology has attracted much attention from practitioners, academics and the regulatory community. The model is very fast: it employs an analytic method (not a simulation) to derive the distribution of losses, so calculations take seconds, not minutes or hours. Both portfolio level risk and contributions to risk by asset are calculated. 

COMPARING THE METHODS The variance-covariance approach with its delta normal and delta gamma variations, requires us to make strong assumptions about the return distributions of standardized assets, but is simple to compute, once those assumptions have been made. The historical simulation approach requires no assumptions about the nature of return distributions but implicitly assumes that the data used in the simulation is a representative sample of the risks looking forward.

COMPARING THE METHODS The Monte Carlo simulation approach allows for the most flexibility in terms of choosing distributions for returns and bringing in subjective judgments and external data, but is the most demanding from a computational standpoint. Since the end product of all three approaches is the Value at Risk, it is worth asking two questions. 1. How different are the estimates of Value at Risk that emerge from the three approaches? 2. If they are different, which approach yields the most reliable estimate of VaR?

CASE STUDY: THE 2007/2008 SUBPRIME MORTGAGE CRISIS After initial price decreases for subprime-related securities in June and July 2007, the 2007/8 subprime mortgage crisis arose with full force in early August 2007. In the following months it affected many financial intermediaries negatively. In response, financial institutions have tried to decline leverage with the result that credit markets have become much tighter and liquidity in many markets reduced. In the years 2001-2006, there was a huge growth in the US subprime mortgage market. Subprime mortgages conveys a comparatively higher risk of default. This growth was simplified by new figures of securitization which enabled banks and mortgage companies to adopt so-called “originate and distribute“ strategies.

CASE STUDY: THE 2007/2008 SUBPRIME MORTGAGE CRISIS With these strategies, the loan originator securitizes the loans of some mortgage portfolio, oftentimes adds enhancement features, and then sells most of the securities on to investors. For example, mortgage providers securitized mortgage pools as mortgage-backed securities (MBS). Such securities are then either sold directly to end investors, to structured investment vehicles (SIVs), or to the managers of special purpose vehicles/entities providing structured finance products such as collateralized debt obligations (CDOs). It was argued - as the subprime crisis showed somewhat short-sightedly - that securitization increased market efficiency and declines liquidity risk.

CASE STUDY: THE 2007/2008 SUBPRIME MORTGAGE CRISIS The subprime crisis affords a good example for the purposes of illustrating various risks of financial risk management. Model risk formalized in various forms during the subprime mortgage crisis. With Daníelsson, one can thus brief the situation: “… the current crisis took everybody by surprise in spite of all the sophisticated models, all the stress testing, and all the numbers”. The models, and risk management in general, did not implement as expected. Merrill Lynch quote mentions two of them – the confidence on ratings/rating agencies and the role of historical simulation in evaluating the risks of new products.