Section II Measuring Market Risk with VaR
Section III Banks’ Investment Portfolio in India—Valuation and Prudential Norms
We have seen in the chapter ‘Banks Financial Statements’ that ‘investments’ constitute a major asset on banks’ books and along with ‘loans and advances,’ take up a major share in banks’ total assets. We have also seen in an earlier chapter the rationale for bank credit and its key role in a country’s economic growth. If credit growth is a major driver of economic growth then what part do ‘investments’ play in banks’ larger role as financial intermediaries?
Typically, banks invest in securities to meet the following objectives:
Safety of Capital We have seen that substantial credit risk is attached to the loan portfolio of banks. To offset this risk, banks invest in securities with low default risk thus preserving the capital.
Liquidity Banks need adequate liquidity to pay off unanticipated demands from the depositors and other liability holders, as well as meet the loan demands. In case a bank does not have liquid funds at the time these demands are made, it has two options—it can borrow from the market or it can sell off near cash assets. Market borrowings carry an inherent risk—interest rates at the time of borrowing may be more than the cost of deposits to be repaid or the contracted yield on the loans. This is called ‘interest rate risk’ and can lead to earnings volatility. On the other hand, banks can invest surplus funds in marketable securities, which can be liquidated in the short term. However, if the market value of these investments at the time of liquidation is less than the book value, the bank will have to report a loss. If the banks decide to invest in low rate securities with fairly stable market values, there is an opportunity loss in the form of reduced interest income.
Yield From the above point, it follows that banks will have to make investments in securities that will also yield reasonable returns. However, paradoxically, higher returns will flow from investments with higher risk. Portfolio managers will, therefore, have to look at risk return trade offs while building the investment portfolio of the bank.
Diversification of Credit Risk Over time, banks develop expertise in lending to a specific sector or industry and find it difficult to diversify their loan portfolios. Hence, banks invest in securities spanning diverse geographic areas or industries to offset credit risk, as well as ensure safety of the capital invested.
Managing Interest Rate Risk Exposure Banks can easily and quickly adjust the maturity or duration (refer chapter on ‘risk management’) of their securities portfolio in times of interest rate volatility. This flexibility will not be typically available with loan portfolios.
Meeting Pledging Requirements Most bank borrowings can be collateralized with assets and marketable securities are accepted as qualifying collateral.
From the foregoing discussion, it is evident that a bank’s investment portfolio cannot satisfy all the objectives. There may be circumstances under which a bank looks for yield from security investments to bolster its dwindling spreads on loans made. There could be other times at which the bank’s liquidity needs would be overarching. How should then the investment portfolio of banks be balanced? Regulators in most countries deal with this dilemma by stipulating that banks hold their investments in three classes to satisfy most of the objectives given above.
Typically, regulators stipulate that banks divide their security holdings into three categories, based on the objectives of such investment as given below.
It is evident that the above classification based on the motive of investment determine the accounting impact on banks’ financial statements. For instance, a fixed rate bond (without options) will sell at par if the market rates and the coupon rate on the bond are the same. If market rates rise above the coupon rate, the market value of the bond falls below the par value. If market rates fall, the reverse would happen. Thus, the difference between the market value and par value would be the unrealized1 gain or loss on the security.
However, if the intention is to hold the bond until maturity, changes in interest rates and the unrealized gains or losses will not affect the value of the security in the bank’s financial statements.
Therefore, prudential treatment requires that adequate provisions are made in the income statements for unrealized gains or losses based on market values in the case of securities held by the trading categories.
A bank’s ‘treasury’ is a source of substantial profit to the bank and can therefore create substantial value to shareholders. In addition to managing liquidity, the treasury is responsible for managing assets and liabilities, trading in currencies and securities, and developing new products. High-performing treasuries systematically identify and mitigate profit from market risk—the risk arising from changes in interest rates, exchange rates and the value of securities and commodities. Banks’ treasuries typically perform the following basic functions within their investment and trading activities.
As markets develop, many credit products are being replaced by treasury products. An example is working capital credit being substituted by commercial paper. Loans are being converted into tradable securities through securitization. Since treasury products are more marketable, liquidity can be infused when required.
1. Foreign exchange business: Buying and selling foreign currency to customers is a source of non-interest income for banks. The difference between the ‘bid’ and ‘ask’ rates, called the spread, constitutes the banks’ profit. Banks buy foreign currency from exporter customers and sell the foreign currency in the inter-bank market. They can also sell the foreign currency to customers (importers), for which they can buy the currency in the inter-bank market or use the currency bought from exporter customers. Banks buy and sell in the inter-bank markets to square the foreign currency balances at the end of each day. Thus, banks do not maintain a stock of foreign currency at the end of each day. In case, a bank maintains an open position (overbought or oversold) at the end of any day, it exposes itself to exchange risk (adverse changes in the value of currency overnight).
New treasury products in the foreign exchange markets have considerably widened the range of services that the banks offer. These markets are the most liquid in respect of currencies that can be freely bought and sold, such as US dollars, Euros and Sterling pounds. These markets are also transparent markets, since information on currency movements are freely available. They are virtual markets with little or no physical boundaries and effective information dissemination. Hence, foreign exchange markets are likened to near-perfect markets with an efficient price discovery mechanism.
Some of the prevalent treasury products3 in these markets are listed below:
2. Money market products: These are securities with short maturities and durations, typically 1 year or less. They are held to meet liquidity and pledging requirements and also for a reasonable return.
Some of the prominent ‘money market’ products are as follows:
3. Securities market products: Banks’ investment portfolios are typically dominated by securities that can be bought and sold in the government securities and capital markets. Each of these securities exhibit varying risk and return features. In most countries, regulations restrict banks’ investments to ‘investment-grade’6 securities only.
There are four major types of marketable securities issued by the Department of Treasury in the US namely, bills, notes, bonds and treasury inflation protected securities (TIPS). These are direct obligations of the government and are considered free from default risk.
Treasury bills are issued for very short maturities, not exceeding a year. They are issued at a discount. The interest earned by the bank is the difference between the face value paid at maturity and the discount price.
The US treasury issues treasury notes, which are interest bearing notes with original maturities of 1–10 years. The prices and yields on these notes are set through auctions and investors include pension funds, insurance companies, financial institutions, corporate bodies and some foreign institutions. The secondary market for these securities is extremely deep, due to the large volumes being traded, low default risk and wide range of investors. In addition, thirty year bills or STRIPS (Separate trading of registered interest and principal of securities) are permitted by the US Treasury. STRIPS are created out of standard T-bills, treasury notes and bonds and are issued as bearer instruments. These instruments are ‘stripped’ into their interest and principal components and traded as ‘zero coupon’7 securities. Each zero coupon security is priced by discounting the promised cash flow at the appropriate interest rate. Banks find these instruments attractive since they are assured of fixed interest payment and yield for the selected maturity. Further, as there are no interim cash flows, there is no reinvestment risk. For example, a 10 year USD 10 million par value treasury bond that pays 10 per cent coupon interest (5 per cent semi-annually) pays USD 500,000 semi-annually. Thus, this security can be ‘stripped’ into 20 disparate interest payments of USD 500,000 each and a single USD 10 million payment at the end of 10 years or 21 separate zero coupon securities. If the market rate on the 3 year zero coupon security is 9 per cent, the associated price of the zero related to the sixth semi-annual payment would be 500,000/(1.045)6 or USD 383,949.
The US Treasury created TIPS in 1997. TIPS pay a fixed rate of interest semi-annually on the inflation adjusted principal amount. The amount paid out as interest will be calculated as the annual interest rate multiplied by the adjusted principal. Since the principal is adjusted for inflation, its value may fluctuate. But, at maturity, the greater of the face value or the inflation adjusted principal is paid. These securities are also eligible for the treasury’s STRIPS.
In addition, banks in the US also invest in government agency securities, which exhibit characteristics similar to those of the US treasury securities. Agency securities are interest bearing and are issued at a discount to the face value. However, the agency securities are less liquid, since the issues are much smaller than treasury issues. Further, unlike treasury securities, agency securities may not be backed by federal government and may not be eligible for tax exemptions. Therefore, in order to compensate for these risks, the returns on agency securities are higher than treasury securities.
Other investments by banks in the US capital market include mortgage backed securities (MBS) that exhibit characteristics similar to corporate bonds. An MBS is a security that evidences an undivided interest in the ownership of mortgage loans. The most common form of MBS is the ‘pass-through security’,8 in which homogenous mortgages are pooled and investors buy an interest in the pool in the form of securities. Other popular forms of MBS are: (a) Government National Mortgage Association, known as Ginnie Mae or GNMA pass-through securities, (b) Federal Home Loan Mortgage Corporation, popularly termed as Freddie Mac or FHLMC participation certificates, guaranteed mortgage certificates and collateralized mortgage obligations, (c) Federal National Mortgage Association, popularly termed as Fannie Mae or FNMA securities, and (d) other privately issued pass-throughs.9
The primary objective of banks’ investment portfolio and treasury operations is to maximise earnings while mitigating risks that are involved.
There are three predominant methods by which bank investments contribute to earnings, which are as follows:
However, the earnings are susceptible to the following risks:
The increasing volatility of financial markets has necessitated design and development of more sophisticated risk management tools. Value at risk (VaR) has become one of the standard measures to quantify market risk.
The concept and use of VaR as a risk management tool gained prominence only about two decades ago. Major financial firms in the late 1980s were using the VaR to measure the risk of their trading portfolios. Since then, VaR has had a meteoric rise as a market risk management tool. Most derivative dealers around the world use the concept to measure and manage market risk. In 1994, J P Morgan released ‘Risk Metrics’TM as a market standard and this provided further fillip to VaR usage.
VaR is defined as the maximum potential loss in the value of a portfolio due to adverse market movements for a given probability. The conceptual simplicity of this measure has made it immensely popular. VaR reduces the (market) risk associated with any portfolio to just one number, the loss associated to a given probability. VaR is a single, summary, statistical measure of possible portfolio losses. Specifically, it is a measure of losses due to ‘normal’ market movements. Losses greater than the value at risk are suffered only with a specified small probability.
VaR measures are used both for risk management and for regulatory purposes. For instance, the Basel Committee on Banking Supervision13 at the Bank for International Settlements used to advocate VaR estimates as a basis for meeting capital requirements in banks. Box 10.2 describes the basic features of the VaR calculation.
How the VaR evolved
It is said that J P Morgan Chairman Dennis Weatherstone used to call for a simple report at the end of each day, showing how the firm’s position would be impacted by the market risk. Analysts of J.P. Morgan evolved the VaR concept and this number was included in the ‘4.15 report’, as the report to the Chairman was called. In 1993, the Washington ‘Group of thirty’, headed by Weatherstone, recommended in its study on ‘Derivatives: Practice and Principles’ that VaR was an appropriate measure for measuring a firm’s market risk. Since then, VaR has been a very popular measure adopted by almost all financial and other institutions for reporting on market risk.
Rationale of VaR methodologies
The statistical features of financial markets have been well documented14: (a) distribution of financial market returns are leptokurtotic (in other words, they have heavier tails and higher peak than the normal distribution), (b) equity returns are negative skewed, and (c) volatilities (as measured by squared returns or standard deviations) of market factors show a cluster tendency. One or more of these empirical features form the basis of the popular VaR models.
What is VaR?
VaR summarizes the predicted maximum loss (or worst loss) over a target horizon within a given confidence interval.
How is VaR computed?
Assume a bank holds ₹100 crore in medium-term investments. How much could the bank lose in a month? As much as ₹100,000 or ₹10 lakh or ₹1 crore? An appropriate answer to this question would enable the banks to decide whether the return they receive is an appropriate compensation for risk.
In order to answer this question, we first have to analyse the characteristics of medium-term securities. Let us do this in the following steps:
Step 1: Obtain, say, monthly returns on medium-term bonds over the last 40 years. It is possible that these returns could range from a low of26 per cent to a high of +15.0 per cent.
Step 2: Construct regularly spaced ‘buckets’ going from the lowest to the highest number and count how many observations fall into each bucket (frequency distribution). Thus, we can construct a ‘probability distribution’ for the monthly returns, which counts how many occurrences have been observed in the past for a particular range.
Step 3:For each return, compute a probability of observing a lower return. This is done as follows. Pick a confidence level, say 95 per cent. For this confidence level, find on the graph a point such that there is a 5 per cent probability of finding a lower return. This number is, say,22 per cent, which implies that all occurrences of returns less than22 per cent add up to 5 per cent of the total number of months or 24 out of 480 months. (Note that this could also be obtained from the sample standard deviation, assuming the returns are close to normally distributed).
Step 4: Now, we can compute the VaR of a ₹100 crore portfolio. There is only a 5 per cent chance that the portfolio will fall by more than ₹100 crore times22 per cent or ₹2 crore. The VaR is therefore, ₹2 crore.
Step 5: In other words, the market risk of this portfolio can be communicated effectively to a non-technical audience with a statement such as: Under normal market conditions, the most the portfolio can lose over a month is ₹2 crore.
Choosing VaR parameters
In the above example, VaR was reported at the 95 per cent level over a 1 month horizon. How do we choose these two quantitative parameters?
Changing VaR parameters
If we assume that portfolio returns follow the normal distribution then we can easily convert one horizon or confidence level to another.
As returns across different periods are close to uncorrelated, the variance of returns for ‘n’ days should be n times the variance of a 1 day return. Hence, in terms of volatility (or standard deviation), it can be adjusted as: VaR(n days) = VaR ( 1 day) × n
In order to convert from one confidence level to another, use standard normal tables. From these tables, we know that the 95 per cent one-tailed VaR corresponds to 1,645 times the standard deviation, the 99 per cent VaR corresponds to 2.326 times standard deviation and so on. Therefore, in order to convert from 99 per cent VaR to 95 per cent VaR,
Use of VaR
This single number summarizes the portfolio’s exposure to market risk as well as the probability of an adverse move. It measures the risk using the same monetary units as the bottom line. Investors can, therefore, decide whether they feel comfortable with this level of risk.
If the answer is no, the same process that led to the computation of VaR can be used to decide where to trim risk. For instance, the riskiest securities can be sold or derivatives such as futures and options can be added to hedge the undesirable risk. VaR also allows users to measure incremental risk, that is, the contribution of each security to total portfolio risk.
Generalizing, using a probability of ‘x’ per cent and a holding period of n days an entity’s VaR is the loss that is expected to be exceeded with a probability of only x per cent during the next n day holding period. In other words, it is the loss that is expected to be exceeded during x per cent of the n-day holding period. Typical values for the probability x are 1, 2.5 and 5 per cent while common holding periods are 1, 2, and 10 (business) days and 1 month. Values of x are determined primarily by how the user of the risk management system wants to view the VaR number is an ‘abnormal’ loss that occurs with a probability of 1 per cent or 5 per cent. For example, J P Morgan’s Risk Metrics system uses 5 per cent, while there are others who use 0.3 per cent. The parameter n is determined by the entity’s horizon. Those which actively trade their portfolios such as, financial firms typically use 1 day, while institutional investors and non-financial firms may use longer holding periods.
A VaR number applies to the current portfolio, so an implicit assumption underlying the computation is that the current portfolio will remain unchanged throughout the holding period. This may not be reasonable, particularly for long holding periods.
References: Jorion, Philippe, 2005, accessed at http://www.gsm.uci.edu/~jorion/oc/case.html and ‘Value at Risk’, Harvard Business School, (9-297-069), rev 15 July 1997.
There are various approaches to VaR computation. Therefore, it is likely that firms calculating VaR for the same portfolio using different approaches may arrive at different VaR figures. Of course, each approach has its advantages, disadvantages and limitations and hence, this aspect should be borne in mind while choosing the appropriate approach and interpreting the results.
However, the approaches follow a common structure, summarized in three steps: (a) mark to market the portfolio, (b) estimate the distribution of portfolio returns and (c) calculate the portfolio VaR. Depending on the method used to arrive at (b), the models can be grouped in three categories.15
The more prevalent ones—historical simulation, the variance-covariance approach and Monte Carlo simulation—are briefly discussed below. The summary provided below is the approach for a typical, simple portfolio. The basic approach can be expanded to include more types of assets with diverse types of market risk.
The common features in all the approaches are: (a) they use historical (over long or short periods) data on the assets contained in the portfolio, (b) they value the portfolio in the next period and compare the future value with the current value and (c) they generate the distribution of the risk portfolios for the required period in future,
1. The Historical Simulation Approach
2. The Variance-Covariance Approach (also called the ‘analytic’ or ‘parametric’ approach) and J.P. Morgan’s Risk Metrics™
The name ‘variance-covariance’ simply signifies the covariance16 matrix of the distribution of changes in the values of the underlying market factors in the portfolio. It is based on the key assumption (there are also other assumptions) that the underlying market factors follow a multivariate normal distribution. With the normal distribution assumption, it is possible to determine the distribution of mark-to-market portfolio profits and losses, which is also normal (being a linear combination of normal variables). Return volatilities (measured by standard deviations) and correlations of risk factors are calculated using historical data. Once the distribution of expected portfolio profits and losses has been arrived at, standard mathematical properties of the normal distribution are used to determine the loss that will be equaled or exceeded x per cent of the time, i.e., the VaR.
Note: As an example, the expected volatility (standard deviation) of a two-asset portfolio can be calculated using σp2 = w12σ12 + w22σ22 + 2w1w2Cov (r1, r2), where w1 and w2 are the respective asset weights in the portfolio and r and σ indicate the return and the standard deviation of the two assets, respectively. This formula can be extended to any number of assets in the portfolio.
The above approach forms the basis for the widely used Risk MetricsTM package17 popularized by J P Morgan. Box 10.3
In 1994, J.P. Morgan developed a VaR model primarily to support the internal reporting system called the ‘4:15 report’ that measured end of day portfolio risk. The Risk Metrics methodology was then standardized and published and developed into a software product in 1996. Two years later, Risk Metrics was spun out of J.P. Morgan as a separate company.
Risk Metrics group acquired institutional shareholder services (ISS) and Centre for Financial Research & Analysis (CFRA) in 2007. ISS was founded in 1985 to promote good corporate governance in the private sector and raise the level of responsible proxy voting among institutional investors and pension fund fiduciaries and in 1986, ISS launched its Proxy Advisory Service to assist institutional investors in fulfilling their fiduciary obligations with comprehensive proxy analysis. ISS gave Risk Metrics access to substantial data on pay, governance and other corporate practices at 41,000 companies, as well as 1,200 new clients who collectively manage $20 trillion in assets. CFRA was born in 1994 to provide institutional investors with early warning signs of business deterioration within portfolio companies. CFRA built a rigorous and proprietary forensic accounting research process for assessing the quality and sustainability of companies’ reported financial results and expanded into specialty legal, regulatory and due diligence research.
Risk Metrics is widely recognized in VaR measurement. However, the industry has yet to settle on a single industry standard for VaR calculation.
In the variance-covariance methodology outlined above, we make a critical assumption that volatility is constant. However, in practice, volatility is not constant and varies from day to day. This problem has been recognized by researchers and a widely used solution was proposed in 1986 by Tim Bollerslev that generalized the pioneering and Nobel prize winning paper of Robert Engle in 1982.19 The time varying volatility approach is commonly called the GARCH method,20 and uses heavier weights for recent returns (and their variances), than for those more distant in time (whereas the constant volatility method outlined above assumes equal weights for all squared returns—variances—in the past). However, this process of weighting calls for a complex, computer intensive procedure.
Risk Metrics is a risk management system that includes techniques to approximate GARCH volatilities. Risk Metrics uses a similar method–Exponentially Weighted Moving Average (EWMA) estimates of daily volatility that represent the weighted average of past squared returns, with the more recent returns receiving heavier weights. However, the set of weights used by Risk Metrics are easier to compute than in the GARCH methodology, and the same set of weights can be used for any asset in the portfolio–say, bonds, stocks or currencies. Similarly, for calculating covariances in the portfolio, the Risk Metrics GARCH approximation can be used to estimate time varying covariances.
Since Risk Metrics is able to yield both volatility and covariance estimates, it can also handle Monte Carlo simulation of derivative portfolios. The general view is that Risk Metrics (as well as the GARCH methodology) tend to underestimate VaR, since the normality assumption made by the model is not consistent with financial market return characteristics.
Risk Metrics has periodically been updating its computation versions—known by the year they were updated in. For example, RM 1994 indicates the methodology proposed in the year 1994. RM 2006 is the latest updated version of the methodology.
RM 1994 methodology relies on the measure of volatilities and correlations given by the historical data, using the EWMA. The appeal of this methodology is its simplicity, conceptually and computationally. However, with more understanding of and experience with volatilities, RM 2006 has introduced a process leading to simple volatility forecasts, while retaining the appeal and advantages of RM 1994. The updated version can also deal with long term horizons.
In 2010, Risk Metrics group became part of MSCI, a leading provider of investment support tools, that include indices, portfolio risk and performance analytics and governance tools. The technical document of RM 2006 is now available on www.msci.com
The appeal of the approach lies in the speed with which computations can be done and the ability to examine alternative assumptions about correlations and standard deviations.
However, the approach is limited by the normal distribution assumption, since movements in market returns do not always follow a normal distribution. The tendency of the market returns to exhibit ‘fat tails’ (extreme values or extraordinary events) can result in misleading conclusions due to the normal distribution assumption. Further, the approach has limited ability to capture the risks of portfolios containing derivatives such as options.
3. The Monte Carlo Simulation Approach
This simulation is much more rigorous than the historical simulation approach described above. It uses mathematical models to forecast future market shocks.
A comparative picture of the three most popular approaches is shown below:
Approach | Advantages | Disadvantages |
---|---|---|
Variance covariance (also called parametric or analytic approach) | The least complex, hence easy to understand. | Limited ability to capture the risks of portfolios which include options, hence may misstate nonlinear risks. |
Easy to implement for portfolios covered by available ‘offthe- shelf’ software. Ease of implementation depends upon the complexity of the instruments and availability of data. | Unable to examine alternative assumptions about the distribution of the market factors, i.e., distributions other than the normal. | |
The least intensive computation and hence, can be done fast. | ||
Historical simulation | Can capture risks of portfolios which include options. | May produce misleading value of risk estimates when the recent past is atypical. |
Easy to implement for portfolios for which data on past values of market factors is available. | Difficult to perform scenario analysis under alternative assumptions. | |
Performs well when back tested. | Can be computationally intensive. | |
Monte Carlo simulation | Easy to implement for portfolios covered by available ‘off-the-shelf’ software. Ease of implementation depends upon the complexity of the instruments and availability of data.
Can capture the risks of portfolios which include options. Can handle statistical assumptions about risk factors Various scenarios can be tested with the model. |
The model can produce misleading values of risk estimates when recent past is atypical. However, alternative estimates of parameters may be used. |
Although VaR is being used for multiple purposes—risk reporting, risk limits, regulatory capital, internal capital allocation and performance measurement—experts opine that VaR is not the answer for all risk management challenges. The perceived shortcomings of VaR are as follows:
However, it is a very popular and promising tool with wide use by practitioners, regulators and academicians. Annexure I to this chapter provides a case study of the sensational collapse of LTCM and its link with VaR.
A hybrid model, which is believed to address some of the shortcomings of the popular models is also used by some practitioners. This model, developed by Boudoukh et al21 combines Risk Metrics and historical simulation methodologies. According to proponents of the approach, the results are more precise than those obtained with the other methods. The approach is implemented in three steps, which are as follows:
Some simple numerical examples of VaR applications are given in Illustration 10.1–10.3. However, before we understand the applications, we have to understand the basics of the normal distribution.
We can see from the diagram above that the normal distribution is a symmetric, bell shaped distribution, defined by its mean (µ) and standard deviation (σ). For simplicity, we assume that the mean is zero and the standard deviation is one—this is called the standard normal distribution. We also note that the distribution is symmetric around the mean—that is, the two halves of the curves are mirror images and the total area under the curve is 1 (or 100 per cent). Typically, the distribution exhibits the following characteristics:
A bank has recently added to its portfolio equity shares of firm ABC Ltd. The shares were bought at ₹500 per share. If the volatility of the share price is 20 per cent per annum and there are 250 trading days in a year, what is the 1 day VaR at 95 per cent confidence?
Stock price | ₹500 |
Volatility | 20% |
Calculate daily volatility as 20/√250 = 1.26% = σ
At 95% confidence, VaR would be
500 (1 × 2*1.26) or 500 (122*1.26)
That is, ₹512.65 or ₹487.35, which translates into a potential gain or loss of ₹12.65 per day.
A bank in India has bought USD 100 million in the spot market. It would like to assess the risk of holding this position for 1 day given the volatility of the currency to be at 10 per cent per year, at the exchange rate of USD1 = ₹50 (that is, ₹1 = 1/50 USD =.02 USD). Assume 250 trading days in a year and the required confidence level at 99 per cent.
The bank’s position in ₹ is 50 × 100 million = ₹500 crore. The daily volatility is 10/√250 = 10/15.81 =.632%.
Hence, the value of the Re. to USD is likely to fluctuate between 0.02(1 + 3 ×.00632) and 0.02(1-3 ×.00632), that is between 0.0203 and 0.0196.
The bank’s position may therefore result in a loss at 99 per cent confidence level as follows:
The volatility of fixed income securities (such as bonds) is measured in terms of volatility in yield.
A bank has invested in fixed income bond with a price of ₹100 and coupon of 9 per cent. The market yield of a similar government security is also 9 per cent. with an annual volatility of 2 per cent. The holding period of the bank’s security is 4 months. Assume the confidence interval to be 95 per cent.
Standard deviation for 4 months = 2/√3 (since 4 months is 1/3rd of a year] = 1.15%
Hence, yields for 4 months at 95% confidence level would be
For simplicity, it can be assumed that 95 per cent of the values fall within +/-2σ and that 99 per cent of the values fall within +/-3σ. As we have seen earlier, 95 per cent, 99 per cent, etc. represent confidence intervals in the VaR calculation.
Calculations of VaR for instruments such as forwards and options are somewhat more complex.
Supplementing VaR—Stress Testing and Scenario Analysis When the VaR is exceeded, how large can the losses be?
Stress testing attempts to answer this question by performing a set of scenario analyses to assess the effects of extreme market conditions. These extreme scenarios may be hypothesized using unexpected events and their impact on the prices of instruments in the portfolio is determined If the effects are unacceptable, the portfolio or risk management strategy needs to be revised or contingency plans prepared. However, the process is more intuitive and depends substantially on the judgment and experience of the risk manager.
Scenario analyses are also used to assess the impact of assumptions underlying VaR calculations being violated.
Alternative Measures to VaR: VaR may not be appropriate for all situations or types of firms. The alternatives are sensitivity analysis, cash flow at risk and expected shortfall methodologies. Sensitivity analysis is considered less sophisticated than, and cash flow at risk and Expected Shortfall (ES) more sophisticated than VaR.
Sensitivity Analysis: Sensitivity analysis are a reasonable alternative for simple portfolios. The approach in this ubiquitous methodology is to imagine hypothetical changes in the value of each market factor, compute the value of the portfolio given the new value of the market factor and determine the change in portfolio value resulting from the change in the market factor. The methodology works well when the magnitudes of likely changes in market factors can be realistically predicted.
Cash Flow at Risk: Risks inherent in operating cash flows are captured reasonably well by this measure and hence is preferred by non-financial firms
Cash flow at risk measures are typically estimated using Monte Carlo simulation but with differences from the use of Monte Carlo simulation to estimate VaR. First, the time horizon is much longer in cash flow at risk simulations. Second, the focus is on cash flows, not changes in mark-to-market values. Third, operating cash flows are the focus of the calculation. Hence, the factors included in the simulation are not the basic financial market factors included in VaR, but those factors affecting operating cash flows such as changes in customer demand or the outcomes of advertising programs. Finally, the primary emphasis is on planning (as opposed to control, oversight and reporting).
There is high degree of subjectivity involved in this approach since successful design and implementation of a cash flow at risk measurement system presupposes substantial knowledge and judgment in respect of the firm’s operating cash flows and the important factors impacting these cash flows and then fitting these into an appropriate and workable model.
Expected Shortfall (ES): This method has been proposed as an alternative to VaR and is designed to measure the expected value of portfolio returns, given that the VaR (or some threshold) has been exceeded. The distinction between VaR and ES has been found to be not very important if the loss distribution is normal. However, for nonnormal distributions, VaR and ES can yield quite different results.
As an example, assume that two firms A and B have invested in portfolios with 1 day VaR of ₹1 crore at 95 per cent confidence level. The ES is concerned with what happens on 5 per cent of the days when the loss exceeds ₹1 crore. For firm A, the loss during this period ranges between ₹1 crore and ₹3 crore with an average of ₹2 crore. For firm B, the loss may range from ₹1 crore to ₹5 crore with an average of ₹3 crore. This implies that firm B’s portfolio is riskier even though the two firms have the same VaR. The ES has brought out the comparative risks of both firms in absolute terms.22
Expected Shortfall has found favour with the Basel Committee as a better measure of market risk in its recently published standards on minimum capital requirements for market risk. Annexure II provides a synopsis of this and other international standards and regulations.
ES will be replacing VaR when the current market risk regulations come into force in 2019 (see Annexure II). The following are cited as the limitations of VaR as a risk measure:
ES scores over VaR since it is considered to have better theoretical properties than VAR. If two portfolios are combined, the total ES usually decreases, reflecting the benefits of diversification. In contrast, the total VaR could increase after combining two portfolios. Experts consider ES to be a “coherent” measure due to this property
A drawback of the ES measure is that it is difficult to back test. For example, when a one day 99% VaR model based on recent historical data is back tested, the number of exceptions can be observed, and can be tested for significant variations from what was expected. However, back testing a one day ES model poses challenges since the average size of the losses have to be computed when exceptions are observed.
According to some researchers, estimates of ES measure may not be as accurate as estimates of VaR.23
But the most severe limitation would be the requirement of data for calculating the ES. Data would be required over a long period to get a reasonable estimate of ES.
Did VaR and Other Such Measures Fail During the 2007–2008 Global Financial Turmoil? It has been pointed out that banks, which used VaR as a primary tool of market risk management, failed, while derivative exchanges, which deal with more complex products, did not. Derivative exchanges have moved away from VaR and used the standard portfolio analysis of risk (SPAN) system. SPAN was developed by the Chicago Mercantile Exchange in 1988 and is used to calculate the portfolio loss under several price and volatility scenarios.
Similar to the loan policy that sets the direction for the credit portfolio of banks, an investment policy needs to be in place with the Board’s approval. According to the RBI, the investment policy should be ‘implemented to ensure that operations in securities are conducted in accordance with sound and acceptable business practices’.25 Further, the central bank advocates that banks wanting to invest in the equity/bond markets should not only have a transparent policy for such investment but all direct investment decisions should be taken by the investment committee of the Board, which will be held accountable for the investment decisions. The central bank would like to see the banks build up adequate internal equity research capabilities.
In short, the RBI has prescribed that the investment policy of bank should clearly lay down the broad investment objectives to be followed while undertaking transactions in securities on their own investment account and on behalf of clients, clearly define the authority to put through deals, procedure to be followed for obtaining the sanction of the appropriate authority, procedure to be followed while putting through deals, various prudential exposure limits and the reporting system. Further, the RBI has spelt out the procedures to be followed by banks while transacting government securities.
The entire investment portfolio of banks (securities held to satisfy Statutory Liquidity Ratio (SLR)26 requirements and those held outside the purview of SLR) are classified as HTM, AFS or HFT.). Banks’ investments in non-SLR securities cover those issued by corporates, banks, FIs and State and Central Government sponsored institutions, Special Purpose Vehicles (SPVs) etc., including capital gains bonds, bonds eligible for priority sector status. The guidelines will apply to investments both in the primary market as well as the secondary market.
However, in the balance sheet, the entire investment portfolio (including SLR and non SLR securities) will continue to be disclosed with the following six classifications: (a) government securities, (b) other approved securities, (c) shares, (d) debentures and bonds, (e) subsidiaries/joint ventures and (f ) others (CP, mutual fund units).27
The objective of the investment (e.g., capital gains, trading profits) and the category–HTM, HFT or AFS should be determined and recorded by banks even at the time of acquisition.
Profit on sale of investments in this category should be first taken to the profit and loss account and thereafter be appropriated to the ‘capital reserve account’ net of taxes and transfer to statutory Reserve. Loss on sale will be recognized in the profit and loss account.
Held for Trading Individual securities in the HFT will be marked to market at monthly or more frequent intervals and provided for (as in the case of those in the AFS category). However, the book value of the individual securities in this category would not undergo any change after marking to market.
In case, the provision for depreciation on AFS and HFT categories is in excess of the required amount in a specific year, the excess should be credited to the P&L account of the bank and an equivalent amount (after tax) shown under ‘Reserves and Surplus’, which can be included as Tier 2 capital of the bank. Detailed instructions on the usage of the investment reserve (IRA) can be found in the quoted RBI circular.
Quoted Securities The ‘market value’ for the purpose of periodical valuation of investments included in the AFS arid HFT categories would be the market price of the security available from the trades/quotes on the stock exchanges, price list of the RBI or prices periodically declared by the Primary Dealers Association of India (PDAI)29 jointly with the Fixed Income Money Market and Derivatives Association of India (FIMMDA).30
Central government securities: Banks should value the unquoted central government securities on the basis of the prices/YTM31 rates published periodically by the PDAI/ FIMMDA. Treasury Bills are to be valued at carrying cost.
State government securities: State government securities will be valued through the YTM method, marked up by 25 basic points above the yields of the central government securities of equivalent maturity (advised periodically by PDAI/FIMMDA).
Other ‘approved’ securities: These will be valued applying the YTM method by marking it up by 25 basic points above the yields of the central government securities of equivalent maturity advised by PDAI/FIMMDA periodically. The valuation of unquoted securities, not included under securities approved for investment under the SLR will be done as stipulated in Box 10.4.
Those securities, which have not been approved as SLR securities are called non-SLR securities.
A ‘rated’ security would be assigned a rating after a detailed exercise by an external rating agency in India which is registered with the SEBI and is carrying a current or valid rating. This implies, inter alia, that the rating should not be more than a month old on the date of issue of the securities. ‘Investment grade’ ratings would be reviewed by the Indian Banks’ Association (IBA) or the FIMMDA annually. Unrated securities are those which do not possess a valid rating.
Listed securities are those listed on approved stock exchanges. Those securities which are not listed on approved stock exchanges are called ‘unlisted’.
(i) the redemption value of the security receipts/pass-through certificates, or (ii) the net book value of the financial asset.
As in the case of non-performing loans described in the chapter on credit risk, if interest or principal is not paid in respect of securities in any of the three categories, the banks should not recognise income from the securities. Appropriate provisions for the depreciation in the value of the investment should also be made. Banks cannot set off the depreciation requirement in respect of these non-performing securities against the appreciation in respect of other performing securities.
A non-performing investment (NPI) (similar to a non-performing advance (NPA)) is one, where:
The RBI has also prescribed uniform accounting treatment for repo/reverse repo transactions. These instructions and numerical examples can be accessed on the website www.rbi.org.in as part of the Master Circular on ‘Prudential norms for classification, valuation and operation of investment portfolio by banks, July 1, 2015.
In the case of non performing loans, we learnt in the previous chapters that income is recognized on cash basis, implying that interest has to be paid to be recognized as income. However, in the case of non performing investments, income recognition is permitted on an accrual basis as shown in the following paragraphs.
Given below is the investment portfolio of Bank A at the end of March 2016.
Note the following:
The bank classifies the entire government and approved securities into current investment category. The prices of government securities on the RBI list for sale are as follows:
Security | Maturity | Sale price (`) |
---|---|---|
5.25% | 2020 | 98.90 |
6.10% | 2022 | 102.25 |
For all other government securities (central and state), the following YTMs are applicable on 31 March, 2016.
number of Years | YTM (%) |
---|---|
Less than 1 | 4.30 |
1 | 4.50 |
2 | 4.54 |
3 | 4.60 |
4 | 4.66 |
5 | 4.83 |
6 | 4.96 |
7 | 5.05 |
8 | 5.05 |
9 | 5.07 |
10 | 5.17 |
The shares of Bank G are traded in the market at ₹27. Since, AB Financial Services Ltd. is not a listed firm, the investment is valued, based on the latest audited accounts at ₹2.40 lakh.
For valuation of taxable bonds, 1 per cent above the applicable YTM rate is to be applied.
The net asset value (NAV) of the mutual fund is ₹10.75.
Provision made during the previous year stands at ₹18 1akh.
What would be the provision required to be made in respect of depreciation in investments during the current year?
Solution
The term to maturity of 9.28% GOI 2022 is 6 years and hence the corresponding YTM is 4.96 per cent from the table. We can use YTM formula33 to calculate the sale price of the security.
Similarly, YTM for 8.51 per cent state government loan for 9 years term to maturity is 5.07 per cent (the term to maturity of 8.9 years is rounded off to 9 years).
Similarly, the price for 7.55 per cent security maturing February 2019 is
Since AB Financial Services Ltd. is not a listed company, no market quote is available for it. The break up value of ₹2.40 lakh can, therefore, be assumed as the appropriate value. In the case of Bank G, the value would be 27.00 × 23700 = ₹6,39,900
The valuation of 7 per cent bonds maturing March 2019 (taxable) is as follows:
Term to maturity = 3 years
YTM = 4.60 + l.00% = 5.60%
Investments in the subsidiary, being permanent in nature, will be shown at book value.
Mutual funds are valued at current NAV.
Now, we can recast the investment schedule as follows:
Within an investment basket, appreciation is netted off against depreciation. However, the net depreciation is only considered and net appreciation is ignored. Thus, for the year the bank has to make a provision on the depreciated amount of `4,89,600 alone. (Baskets III + VI)
Answer ‘True’ or ‘False”
Check your score in Rapid fire questions
Substantiate your answer.
Substantiate your answer.
In 1998, LTCM, the world’s largest hedge fund, failed and sent shock waves through the world’s financial system. LTCM’s size made the Federal Reserve Bank of New York step in to facilitate a bailout to the fund, fearing the negative impact of the liquidation on global financial markets.
It has been widely reported that the primary contributory factor to LTCM’s failure has been its poor risk management.
With a capital base of USD 3 million, LTCM possessed derivatives whose value exceeded USD 5 trillion and controlled over USD 100 billion in assets, globally – making it one of the most highly leveraged funds in the history. To predict and mitigate its risk exposures, LTCM used a combination of different VaR techniques and claimed that its VaR analysis showed that investors might experience a loss of 5 per cent or more in about 1 month in 5 and a loss of 10 per cent or more in about 1 month in 10. Only 1 year in 50 should it lose at least 20 per cent of its portfolio. LTCM also estimated that a 45 per cent drop in its equity value over the course of a month was a 10 standard deviation event. In other words, this scenario had no likelihood of occurrence (Prabhu, 2001). However, this ‘most unlikely’ event happened in August 1998. In spite of sophisticated hedging strategies, LTCM went wrong in one basic assumption – that historical trends in securities movements were an accurate predictor of future movements. The VaR models used by LTCM relied on historical data to predict future price movements. Unfortunately, the future holds too many uncertainties that the past cannot predict was the lesson learned by the management and investors at a very great cost.
For example, on 18 October 1987, 2 month S&P futures contracts fell by 29 per cent, which under the hypothesis of a lognormal distribution with annualized volatility of 20 per cent (which was also the approximate historical volatility of this security) would have denoted a 27 standard deviation event. Experts opined that the possibility of such an event happening would have been virtually impossible. Again, 13 October 1989 saw the S&P 500 fall about 6 per cent, a 5 standard deviation event under similar assumptions as above. Again, experts opined that such an event would have the likelihood of occurring only once in 14,756 years!
In August 1998, some unexpected events occurred, which were beyond the ‘predictive abilities’ of the VaR models used by LTCM. The Russian crisis triggered drying up of liquidity in the global financial markets and derivative positions were quickly slackened. Yet, LTCM’s VaR models continued to estimate that the daily loss would be no more than USD 50 million of capital. But the fund started losing around USD 100 million every day. On the fourth day in the wake of the Russian debacle, the fund lost USD 500 million in a single trading day.
While LTCM prepared to declare bankruptcy, the US Federal Reserve extended a USD 3.6 billion bailout to the fund, which was questioned by the experts who felt that such an action could create a moral hazard problem.
Hence, though perceived to be a highly effective risk measurement tool, VaR has to be used judiciously. At the end of August 1998, LTCM had USD 2.3 billion of equity capital and USD 1 billion excess liquidity. However, the firm was caught in a dilemma between reducing risk and raising additional capital. The magnitude of its positions rendered it unable to reduce its risk exposures immediately or attract new investors and it appeared that it had underestimated its capital needs. In retrospect, the fund was maximizing return while based on the simplistic assumption that volatility would remain constant, while in reality it could easily double in turbulent times.
The fund was not able to liquidate quickly enough in case of adverse market conditions, leading to a risk of liquidity and insolvency. Liquidity risk is not factored into VaR models since they assume that normal market conditions will prevail.
Therefore, it appears that LTCM’s actual exposure to liquidity and solvency risks was not adequately exposed by its VaR model.
While VaR is seen as an invaluable tool to measure risks, complementary measures such as, stress testing could have probably offset the limitations and disadvantages of VaR While VaR could show fund managers what they could lose with a predetermined maximum probability, stress tests enable them to assess their possible loss if the worst case scenario materialized.
The case of LTCM serves to illustrate the practical dimensions of a popular risk measurement instrument like the VaR.
A series of bank failures in the 1970s prompted the setting up of the Basel Committee on Banking Supervision, with the intention to regulate financial institutions, particularly banks. A detailed discussion on the Basel Committee has been carried out in another chapter in this book.
Since its inception, the Committee has focused on regulation of bank capital globally. The measure of capital is based on the asset portfolio risks individual banks take, as well as operational risks in banking operations. An important part of the asset portfolio risk is ‘market risk’ caused by adverse fluctuations in market factors. The risk measure agreed upon for market risk measurement was the VaR.
Significant weaknesses were identified in the Basel II framework for trading activities, and the framework has taken its share of criticism for the global financial crisis of 2007-08. To address the most conspicuous deficiencies in the framework, the Basel committee introduced a set of revisions to the Market risk framework in 2009 – termed the Basel 2.5 package of reforms. A further process to fine tune and redesign the capital standard for market risk was undertaken.
Basel 2.5 required banks to hold additional capital for default and ratings migration risk, that is, the risk that a rating change in the market instrument triggers significant mark-to-market losses. Banks were also asked to calculate an additional VaR capital charge, called the “Stressed VaR”. The new framework also brought in changes in the treatment of securitization exposures. However, Basel 2.5 was seen to not address all the structural shortcomings with the market risk framework.
The Basel Committee has since published three consultative documents on the trading book review: Fundamental review of the trading book, May 2012, www.bis.org/publ/bcbs219.htm; Fundamental review of the trading book: A revised market risk framework, October 2013, www.bis.org/publ/bcbs265.htm; and Fundamental review of the trading book: Outstanding issues, December 2014, www.bis.org/bcbs/publ/d305.htm
The final rule on the Fundamental Review of the Trading Book (FRTB) was published in January, 2016, called the “Minimum Capital requirements for market risk” (http://www.bis.org/bcbs/publ/d352.pdf), after five years of discussion, four Quantitative Impact Studies (QIS), and three consultative papers.
The final rule is being commonly referred to as ‘Basel 3.5’ or ‘Basel 4’, since it is considered the first of many rules related to Basel 4 that are anticipated to be finalized soon, and is expected to have significant impact on trading book definition, risk measurement, assessment and reporting at financial institutions all over the world.
Figure 10.1 summarizes the key aspects of the FRTB
FIGURE 10.1 FRTB AND THE MARKET RISK FRAMEWORK – A BIRD’S EYE VIEW
Banks are required to report under the new standards by end of the year 2019.
Under the rules, Stressed Expected Shortfall is to be calculated at the 97.5th percentile for each trading desk (as specified in the Basel committee document quoted above). Appropriate “Liquidity horizons” are to be used for scaling up the ES from the base horizon. This will pose a challenge to banks, especially in the areas of data gathering and analytics and model building.