Chapter Ten Market Risk INTRODUCTION market risk: Market risk (or value at risk) can be defined as the risk related to the uncertainty Risk related tothe” of an FI’s earnings on its trading portfolio caused by changes, and particularly uncertainty of an IS “extreme changes, in market conditions such as the price of an asset, interest ‘gamnings omits trad” tes, market volatility, and market liquidity.'? Thus, risks such as interest rate iePanges nmantet tisk (discussed in the last two chapters) and forcign exchange risk (discussed in by changes ° . conditions, ‘Chapter Ld) alfect market risk. However, market risk emphasizes the risks to FIs that Actively trade assets and liabilities (and derivatives) rather than hold them for term investment, funding, or hedging purposes. ‘Conceptually, an FI's trading portfolio can be differentiated from its investment portfolio on the basis of time horizon and liquidity. The trading portiolio contains assets, liabilities, and derivative contracts that can be quickly bought or sold on “organized financial markets (such as long and short positions in bonds, commodi- ties, foreign exchange, equity securities, interest rate swaps, and options). Further, with the increasing securitization of bank loans (e.g, mortgages), more and more assets have become liquid and tradable (e.g., mortgage-backed securities). The investment portfolio (or, in the case of banks, the so-called banking, book) con- tains assets and liabilities that are relatively illiquid and held for longer holding periods (such as consumer and commercial loans, retail deposits, and branches). Table 10-1 shows a hypothetical breakdown between banking book and trading book assets and liabilities. Note that capital produces a cushion against losses on cither the banking or trading books—see Chapter 20. Income from trading activities is increasingly replacing income from traditional Fl activities of deposit taking and lending. The resulting earnings uncertainty, o market risk, can be measured over periods as short as a day or as long as a year. While bank regulators have normally viewed tradable assets as those being held for horizons of less than one year, private Fls take an even shorter-term viow. In particular, Fis are concerned about the fluctuation in value—or value at risk (VAR)—of their trading account assets and liabilities for periods as short as one day (so-called daily earnings at risk [DEAR))—especially if such fluctuations pose 1 threat to their solvency. Moreover, market risk can be defined in absolute terms 14.8 Maigan, Introduction to Rik Matis (Now York: 1 Morgan, Octaber 1964), 2. There i an ‘ongoing debate about whether spread sks a part of market rik or creat rst. J.P Morgan, ncluces [spread iat as et ck (and includes itn she CredtMetics measure [see Chapie’ 11) rather than at part of market risk Market ise usd by Fl manager and regulator ic not synonymous with systematic markt risk analyzed by investors in secunties marcets. systemavc (market) rik reflects the cowmoverent of asecunty with the market portfobo(e‘leced by the security beta) although betas used to measure the market rik of ‘eqites, 25 noted below.TABLE 10-1 ‘The Investment (Banking) Book and ‘Trading Book of a Commercial Bank ‘Assets cash oars Other liquid borrowed funds Banking Bock Premises and equipment Capital Otherilguid asets ‘Bonds (ong) Bonds Grom Commodives (long) Commodities (shor) Fxlang) Fx (short) Trading Book Equities Vong) Equities (hort) ‘Mortgage-backed securities (long) Derivatives ong) Devatves® (Short "Deivatie ae fblaneabeet as desl fn Chapter) a a dollar exposure amount oF as a relative amount against some benchmark. For example, KeyCorp’s 2005 Annual Report (p. 41) states, “Management uses a value at risk (“VAR”) simulation model to measure the potential adverse effect of changes in interest rates, foreign exchange rates, equity prices, and credit spreads on the fair value of Key’s trading portfolio.” In recent years, market risk of Fls has raised considerable concern among regu- lators as well. For example, in February 1995, Barings, the U.K. merchant bank, was forced into insolvency as a result of losses on its trading in Japanese stock index futures. The largest trading loss in recent history involving a “rogue trader” occurred in June 1996 when Sumitomo Corp. (a Japanese bank) lost §2.6 billion in commodity futures trading. More recently, a single trader's actions in the FX markets resulted in almost $700 million in losses to Allfirst, the US. subsidiary of Allied Irish Bank. So important is market risk in determining the viability of an Fl that since 1998, US. regulators have included market risk in determining the required level of capital an FI must hold? “Table 10-2 summarizes several benefits of measuring market risk, including providing management with information on the extent of market risk exposure, market risk limits, resource allocation, and performance evaluation, as well as providing regulators with information on how to protect banks and the finan- ial system against failure due to extreme market risk. The sections that follow concentrate on absolute dollar measures of market risk. We look at three major approaches that are being used to measure market risk: RiskMetrics, historic or back simulation, and Monte Carlo simulation. The link between market risk and required capital levels is also discussed in the chapter CALCULATING MARKET RISK EXPOSURE Large commercial banks, investment banks, insurance companies, and mutual funds have all developed market risk models. In the development of these ‘models—so-called internal models—three major approaches have been followed: + RiskMetrics (or the variance /covariance approach). + Historic or back simulation. + Monte Carlo simulation. 2 Ths requtomont was introduced eal (1966) in the ELL268 Part Two Meauiring Risk TABLE 10-2 Benefits of Market Risk Measurement (RM) information. MRM provides senior management with information on the ‘ek exposure taken by Fl traders. Management can then compare ths risk expoure te the Fis capital resources. 2. Setting limits. MBM considers the market rie of traders’ porttlias, which wil lead to the establishment of economically logical position limits per tader in each area of wading. 3. Resource allocation. MRM involves the comparson of returns to market risks in different ‘areas of trading, which may allow for the identification of areas with the greatest potential ‘return per unit of risk into which more capital and resources can be dicted. 4 Performance evaluation. MRM, relatedly, considers the retur-risk ratio of traders, which ‘may allow a more rational bonus (compensation) system to be put in place. That is, those ‘traders with the highest returns may simply be the ones whio have teken the largest risks. Itis not dear that they should receive higher compensation than traders with lower turns ‘and lover tsk exposures. '5. Regulation. With the Bank for International Settlamants (BIS) and Federal Reserve ‘currently regulating market rick theough capital requicements (discussed later this ‘chapter, private sector benchmarks are important, since its possible that regulators ‘will overprice some risks. MRM conducted by the FI can be used to point to potential ‘misellocations of resources as a result of prudential regulation. As 2 result, in certain cases, teguators are allowing banks to use thei own (internal) models to cakulate thet capital requirements We consider RiskMetrics* first and then compare it with other internal model approaches, such as historic or back simulation. THE RISKMETRICS MODEL The ultimate objective of market risk measurement models can best be seen from the following quote by Dennis Weatherstone, former chairman of J. P. Morgan (PM), now J. P. Morgan Chase: “At close of business each day tell me what the market risks are across all businesses and locations.” In a nutshell, the chairman of J.P. Morgan wants a single dollar number at 4:15 pat New York time that tells him J.P. Morgan’s market risk exposure the next day—especially if that day turns out tobe an extremoly “bad” day. ‘This is nontrivial, given the extent of JPM’s trading business. When JPM devel- ‘oped its RiskMetrics Model in 1994 it had 14 active trading locations with 120 independent units trading fixed-income securities, foreign exchange, commodi- ties, derivatives, emerging-market securities, and proprietary assets, with a total daily volume exceeding $50 billion. This scale and variety of activities is ‘of the major money center banks, large overseas banks (e.g,, Deutsche Bank and Barclays), and major insurance companies and investment banks. “ince regulators are concerned with the soda costs ofa faite or insolvency, incuding contagion ef {fects and other exerraites, regulatory models wil normaly tend to be more consenatve than pivate “ector mode shat are cancorned only with the prvata costo! fare. 1. F Morgan UP) fist developed RiskMetrcsin 1998. In 1998 the development group formed a sepa rate company, pally Gnmned by PM. The material presented inthis chapter & an overview ofthe Aske Meties model. Te det, adeitonal decusion, and examples ae found in “Return to RiskNetres: The Fvoluton ofa Standard,” Apil 2001, avalanlo atthe | P Morgan Chase Wab sit, vnnwjpmocganchate com or wurusiskmetis.com,Chapter 10 Market Risk 269, Here, we will concentrate on measuring the market risk exposure of a major FI on a daily basis using the RiskMetrics approach. As will be discussed later, mea- suring the risk exposure for periods longer than a day (eg., five days) is under certain assumptions a simple transformation of the daily risk exposure number. Essentially, the FI is concerned with how to preserve equity if market condi- tions move adversely tomorrow; that is: ‘Market risk = Estimated potential loss under adverse circumstances More specifically, the market risk is measured in terms of the FI's daily earnings at risk (DEAR) and has three components: Daily earnings Dollar market Price (Potential atrisk — = valueof sensitivity of x adverse move the position) _the position) in yield) Since price sensitivity multiplied by adverse yield move measures the degree of price volatility of an asset, we can also write this equation as: Daily earnings (Dollar market atrisk "= valueof =X Price Oy the position) _volatilty) How price sensitivity and an adverse yield move will be measured depends on the Fland its choice of a price-sensitivity model as wellasits view of what exactly isa potentially adverse price (yield) move. We concentrate on how the RiskMetrics model calculates daily earnings at risk in three trading areas—fixed income, foreign exchange (FX), and equities—and then on how it estimates the aggregate risk of the entire trading portfolio to meet, Dennis Weatherstone’s objective of a single aggregate dollar exposure measure across the whole bank at 4:15 rat each day.* The Market Risk of Fixed-Income Securities Suppose an FI has a $1 million market value position in zero-coupon bonds of seven years to maturity with a face value of §1,631,483. Today’s yield on these bonds is 7243 percent per year? These bonds are held as part of the trading port folio. Thus, Dollar market value of position = $1 million “this dear from the above discussion that interest rate rk (s2e Chapters 8 and 3) s part of market sk. However, in market risk medal, we are concorned withthe intrest rat senstity ofthe fusdsinceee secures held as part ofan Fs acine trading portfolio. Many fiedencome secures ae held as part of an Fisinvestment porfolo, White the latter are subject to interest rate risk, they will nt be included in 2 market rik elution, "The ‘ace value ofthe Bonds is 51,631 483—that is, $1,631,483A1,.07243) = $1,000,000 mareet value In the otginal mode, prices were determined using a dscrete rate of return In te Apri 2001 docu ‘ment “Return to fisketrics: The Evolution ofa Standard,” prces are determined using a continuously ‘compounded turn, @ The change was mplorented beca.se continuous compounding has prope. ‘es that faiitate mathematical eaten. For example, the logarithmic reurn on @ ze@ The reasons tat the hisivic, or back simulation, approach ses actual exCharge aces on each day that cexplitl indlude corelations or comoiemants with other exchange rates and azat return on that ay. © The S percent number in RskMetic tell us that we willloze ore than this amount on 5 cays cut of ‘very 106; foes not tll us the maximum amaunt we can lose. Ae nated in the tor, thecretical, with 2 normal dsuibution, this could bean infinite amount282 Part Two Meauiring Risk example). One possible solution to the problem is to go back in time more than 500 days and estimate the 5 percent VAR based on 1,000 past daily observations (the 50th worst case) or even 10,000 past observations (the 500th worst case). The problem is that as one goes back farther in time, past observations may become decreasingly relevant in predicting VAR in the future. For example, 10,000 obser- vations may require the FI to analyze FX data going back 40 years. Over this period we have moved through many very different FX regimes: from relatively fixed exchange rates in the 1950-70 period, to relatively floating exchange rates in the 1970s, to more managed floating rates in the 198¢s and 1990s, to the aboli- tion of exchange rates and the introduction of the euro in January 2002. Clearly, ‘exchange rate behavior and risk in a fixed-exchange rate regime will have little relevance to an FX trader or market risk manager operating and analyzing risk: a floating-exchange rate regime. ‘This seems to confront the market risk manager with a difficult modeling prob- Jem. There are, however, at least two approaches to this problem. The first is to ‘weight past observations in the back simulation unequally, giving a higher weight to the more recent past observations.” The second is to usea Monte Carlo simula- tion approach, which generates additional observations that are consistent with recent historic experience. The latter approach, in effect, amounts to simulating or ‘creating artificial trading days and FX rate changes. The Monte Carlo Simulation Approach’? To overcome the problems imposed by a limited number of actual observations, ‘we can generate additional observations (in our example, FX changes). Normally, the simulation or generation of these additional observations is structured using Monte Carlo simulation approach so that returns or rates generated reflect the probability with which they have occurred in recent historic time periods. The first step is to calculate the historic variance-covariance matrix (2) of FX changes. This matrix is then decomposed into two symmetric matrices, A and A'* This allows the FI to generate scenarios for the FX position by multiplying the A” matrix, which reflects the historic volatilities and correlations among FX rates, by a random number vector 2 10,000 random values of = are drawn for each FX ‘exchange rate.* This simulation approach results in realistic FX scenarios being ‘generated as historic volatilities and correlations among FX rates are multiplied by the randomly drawn values of z. The VAR of the current position is then cal- ‘culated as in Table 10-6, except that in the Monte Carlo approach, the VAR is the ‘500th worst simulated loss out of 10,000. 2 See Allen, J. Goudoukh, and A. Saunders, Undestancing Market, Credit and Operational Rsk: The Value at Fisk Agprosch (New Yor Blackwell 2004), Chapters 1-3. This section, whic conteins move tecnicel deta, may be induced incr dropped from the chapter reading depending on te igor of te course, 22 The only difference between ad A ie thatthe numbers inthe ows ofA become the numbers in he ‘columns of 4°. The technical term for this procedure fs he Cholescy decomposition, were + vwhere zs assumes to be normaly drive with 2 mean of zet0 and a standard devation of 1 or 2G, 1) * Techricaly, lety be an FX scerati; theny =A'z. for each FX rte, 10,000 values of z are tancoly “generated 10 produce 10,000 values of. They value ae shen used to revalve the FX postion and caus late gains and ses. see, for empl, J.P Moraan, Rsketncs Technical Document, shed. 1997Chapter 10 Market Risk 283 Questions 1. What te the advantages ofthe historic, 0” back simulation, approach over RiskMetics to measure market risk? 2. What are the steps involved with the historic, or back simulation, approach to mea suring market risk? 3. What is the Monte Carlo simulation approach to measuring market risk? REGULATORY MODELS: THE BIS STANDARDIZED FRAMEWORK specific risk charge the risk of a decline in the liquidity or credit risk quality of the trading portfolio. ‘The development of intemal market risk models by Fls such as J.P. Morgan Chase ‘was partly in response to proposals by the Bank for International Settlements (BIS) in 1998 to measure and regulate the market risk exposures of banks by imposing capital requirements on their trading portfolios.” As noted in Chapter 7, the BIS is 2 organization encompassing the largest central banks in the world. Aftor refining these proposals over a number of years, the BIS (including the Federal Reserve) decided on a final approach to measuring market risk and the capital reserves nec- essary foran Fl to hold to withstand and survive market risk losses. These required levels of capital held to protect against market risk exposure are in addition to the minimum level of capital banks are required to hold for credit risk purposes (see Chapter 20). Since January 1998 banks in the countries that are members of the BIS can calculate their market risk exposures in one of two ways. The first is to use a simple standardized framework (to be discussed below). The second, with regulatory approval, is to use their own internal models, which are similar to the models described above. However, ifn internal model is approved for use in cal- culating capital requirements for the FI, its subject to regulatory audit and certain constraints. Before looking at these constraints, we examine the BIS standardized framework for fixed-income securities, foreign exchange, and equities. Additional details of this model can be found at the BIS Web site, www.bis.org. Fixed Income We can examine the BIS standardized framework for measuring the market risk on the fixed-income (or debt security) trading portfolio by using the example for a typical FI provided by the BIS (see Table 10-7). Panel A in Table 10-8 lists the security holdings of an FI in its trading account. The FI holds long and short post- tions in {column (3)] various quality debt issues [column (2)] with maturities rang- ing from one month to over 20 years [column (1)]. Long positions have ratve values; short positions have negative values. To measure the risk of this tra Portillo, the BIS uses two eaplal charges: (1a specific risk change [columns () and (5)| and (2). general market risk charge [columns (6) and (7)]. Specific Risk Charge ‘The specific risk charge is meant to measure the risk of @ decline in the liquid- ity or credit risk quality of the trading portfolio over the F's holding period. As 231s, Basel Committee on Barking Supervsion, “The Supervisory Teatrent cf Market Rss,” Basel, Switzerland, Api 1983, Proposal to ssue a Supplerrent tothe Basel Accord to Cover Market fisks, aed, Switzerland, April 1995; and "The Now Basal Capital Accord: Tid Concutatve Papa,” Base, ‘switzerland, Api 2003. > The requirenen's were Invosuced earer in 1996 Inthe European Union.288 Part Two Meauring Risk ‘general market risk reflosting the modified dura- tion and interest rate shod for each vertical offsets ‘column (4) in panel A of Table 10-7 indicates, Treasuries have a zero risk weight, while junk bonds (e.g.,10- to 15-year nonqualifying “Non Qual” corporate debt) hhave a risk weight of 8 percent. As shown in Table 10-7, multiplying the absolute dollar values of all the long and short positions in these instruments {column (3)] by the specific risk weights [column (4)] produces a specific risk capital or require- ment charge for each position [column (5)]. Summing the individual charges for specific risk gives the total specific risk charge of $229.” General Market Risk Charge The general market risk charges or weigh's—column (6)—reflet the product of the modified durations and interest rate shocks expected for each maturi ‘The weights in Table 10-7 range from zero for the 0- to I-month Treasuries to 6 percent for the longterm (longer than 20 years to maturity) quality corporate debt securities. The positive or negative dollar values of the positions in each instru- meat (column @)] are multiplied by the general market risk weights [column (6)] to determine the general market risk charges for the individual holdings [column @)]. Summing these gives the total general market risk charge of $66 for the whole fixed-income portfolio. Vertical Offsets ‘The BIS model assumes that long and short positions, in the same maturity bucket but in different instruments, cannot perfectly offset each other. Thus, the $66 gen- ‘eral market risk charge tends to underestimate interest rate or price risk exposure. For example, the Fis short $1,500 in 10- to 15-year US. Treasuries producing a market risk charge of $67.50 and is long $1,000 in 10- to 15-year junk bonds (with a risk charge of $45). However, because of basis risk—that is, the fact that the rates on Treasuries and junk bonds do not fluctuate exactly together—we cannot ‘assume that a $45 short position in junk bonds is hedging an equivalent ($45) risk value of US. Treasuries of the same maturity. Similarly, the FI is long $2,500 in three- to four-year Treasuries (with a general market risk charge of $56.25) and ‘short $2,000 in three- to four-year quality corporate bonds (with a risk charge of $45). To account for this, the BIS requires additional capital charges for basis risk, called vertical offsets or disallowance factors. We show these calculations in part 2 of panel B in Table 10~ In panel B, column 1 lists the time bands for which the bank has both a long {and short position, Columns (2) and (3) list the general market risk cha column (7) of panel A—resulting from the positions, and column (4) lists the dif- ference (or residual) between the charges. Column (5) reports the smallest value ‘of the risk charges for each time band (or offset). As listed in column (6), the BIS disallows 10 percent! of the $45 position in corporate bonds in hedging $45 of 2 Note that the risk weight for specie sts are nt based on obvious theory, empirical research, or past ‘experience. Rather. the weights ar base on regulators’ percentions of what was oproprate wen the rode wes established, ®Ferexanpee, fot 15:10 20,eer Tessie in Tale 10-7, the modified durations assumed tobe 875, year, and the interest rate shocks 0.60 percart. Thus, 8.75 < 06 = 5.25, whichis the gen- ‘eral market rk weaht fr these secures shonin able 10-7. Aultiying 5.25 bythe $1,500 ona postion in these secstes ests in a general maiket rk charge of $78.75. Note that the shocks a= Sumed fo shor-tem secures, such st tes-month Tolls, ar larger (at 1 perent than those assumed ‘or longesmatunty secures. Tis eles the fact that short-term rates are more impacted by menetay policy Frally, rote that the sandedized model combires unequal rate shocks vith estimated medified ‘durations to calculate market rise weight. Techeicaly this violates the underhig assumptions ofthe du ration model which assumes parallel yield shits (ee Chapter 9) at each maturty. “Note again thatthe disallowance factors were set subjectnely by regulators.Chapter 10 Market Risk 285 TABLE 10-7 BIS Market Risk Calculation (Debt Securities, Sample Market Risk Calculation) Panel A: Fl Holdings and Risk Charges Specific Risk ‘General Market Risk o 2 @ @ ©. © a Time Band. Issuer Position(s) Weight (%) Charge Weight (%») Charge 0-1 month Treasury 5,000 0.00% 0.00 0.00% 0.00 1-3 months Treasury 5,000 0.00 0.00 020 10.00 3-6 months QualCop 4,000 025 10.00 0.40 16.00 6-12 months QualCorp (7/500) 1.00 75.00 070 (6250) 1-2years Treasury 2.500) 0.00 0.00 125 6125) 2-3 yests Treasury 2500 0.00 0.00 175 43.75 3-4 years Treasury 2,500 0.00 0.00 225 56.25 3-4years Qual Corp (2,000) 1.60 32.00 225 45.00) 15 years Treasury 4,500 or) 0.00 275 41.25 57 years Qual Corp (1,000) 1.60 16.00 325 @250) T=10 yeas easy (500) 0.00 0.00 375 66.25) 10-15 years Treasuy (1500) 0.00 0.00 450 (6750) 10-15 years ‘Non Qual 1000 8.00 30.00 450 45.00 15-20 years Treasury 1,500 0.00 0.00 525 78.75 > 20 years QualCorp 1/000 1.60 16.00 6.00 60.00 Specific risk 229.00 Resiual general market risk 66.00 Panel 8: Calculation of Capital Charge o @ @ @ © ©. @ charge 1. Spexttc Risk 22300 2. Vertical Offsets within Seme Time Bands ‘Time Band. Longs Shorts Residual* Offset Disallowance Charge 34years 56.25 (45.00) 1125 45.00 10.00% 450 40-15 years 45.00 6750) @250) 45.00 10.00 450 3. Horizontal Ofists within Same Time Zones Zone 1 ‘0-1 month 0.00 1-3montns 10.00 3-6montns 16.00, 6-12 months 6250) ‘otal zone! 26.00 6250) @65 2600 40.00% 10.40 Zone 2 4-2 years 21.25) 2-3 years 4375 2A years 11.5 Total 7ore2 $500 125) 2375-3125 30.00% 938 Zone 3 45 years 4125 5-7 years 2150) TAD years (66.25) 10-15 years 2250) 15-20 years 78.75 220 years 60.00 Total zore3__180.00_(111.25) 687511125 30.00% 3338 ‘ontinued)285 Part Two Meauring Risk TABLE 10-7 (continued) Time Band longs Shorts «Residual ——Offset_—_Disallowance Charge 4 Horizontal Offsets between Time Zones Zones 1 and 2 2375 26.50) 275) 23:75 40.00% 950 Zones 1 ad 2 ears 75) 6509, 275 15000% an 5. Total Capital Charge Specific risk 229.00 artical ieatowences 00 Fonizontal dsalowances Offsets within same time zones 53.16 Offsets betneen time zones 13.62 Residual general market rk after al offsets 66.00 Total a7 “Aetna aneun caret andor dnl ofating appropiate Nate GulCarpscnavston pdr OW ond son) No Ql bbw gd dt an ogre fiber ae patos the Treasury bond position. This results in an additional capital charge of $4.50 (G45 x 10 percent). The total charge forall vertical offsets is $9. Horizontal Offsets within Time Zones Tn addition, the debt trading portfolio is divided into three maturity zones: zone 1. month to 12 months), zone 2 (more than 1 year to 4 years), and zone 3 (mare than 4 years to 20 years plus). Again because of basis risk (Le. the imperfect cor relation of interest rates on securities of different maturities), short and long pos tions of different maturities in these zones will not perfectly hedge each other. This results in additional (horizontal) disallowance factors of 40 percent (zone 1), 30, percent (zone 2), and 30 percent (zone 3).® Part3 of the bottom panel in Table 10-7 horizontal offsets shows these calculations. The horizontal offsets are calculated using the sum of ‘Addiional capital the general market risk charges from the Tong and short positions in each time args required be zone—columns (2) and (3). As with the vertical offsets, the smallest of these totals eS faiterent 'Sthe offset value against which the disallowance is applied, For example, the total Raturtiesdonor — Z0Me I charges for long positions equal $26 and for short positions ($52.50). A dis Periecy hedge each allowance of 40 percent of the offset value (the smaller of these two values), $26, is charged, that is, $10.40 ($26 x 40 percent). Repeating this process for each of the three zones produces additional (horizontal offset) charges totaling $53.16. Horizontal Offsets between Time Zones Finally, because interest rates on short maturity debt and long maturity debt donot fluctuate exactly together, a residual long or short position in each zone can only partly hedge an offsetting position in another zone. This leads to a final set of off- sets, or disallowance factors, between time zones, part 4 of panel B of Table 10-7. Here the BIS model compares the residual charges from zones 1 ($26.50) and 2 (623.75). The difference, $2.75, is then compared with the residual from zone 3 ($68.75). The smaller of each zone comparison is again used as the offset value “© ntti, this impli that long-term US. Treasury fates and long-term jurk bone rates are approx- rately 90 percnt corelated. However in th final pla, was decided to cut vertical dsalowance fac. {arsin hal Thus, a 10 percent lsalowance factor becomes a5 percent dhllowance factor, end so 0°. The zones were abo se sujectively by requaiorsTABLE 10-8 Example of the BIS Standardized Framework “Measure of Foreign Exchange Risk (in millions of dollars) Sour B15, re, ison Chapter 10 Market Risk 287 ‘Once a bank has calculated iis net position in each foreign currency, it converts each post tion into its reporting currency and calculates the rick (capita) measure asin the following ‘exampla, in which the postion in the reporting currency (dollars) has been excluded: Yen" Euros GBE AS SF +50 +100 +150 ‘The capital charge would be 8 percent of the higher of the longs and shorts (Le., 300). TAlcurnde nS epi against which a disallowance of 40 percent for adjacent zones and 150 percent® for nonadjacent zones, respectively, is applied. The additional charges here total $1362. Summing the specific risk charges ($229), the general market risk charge (S66), and the basis risk or disallowance charges ($9.00 + $53.16 + $13.¢2) produces a total capital charge of $370.78 for ths fixed-income trading portfolio Foreign Exchange ‘The standardized model or framework requires the FI to calculate its net expo- sure in each foreign currency—yen, euros, and so on—and then convert this into dollars at the current spot exchange rate. As shown in Table 10-8, the FI is net ong (million-dollar equivalent) $50 yen, $100 euros, and $150 pounds while being short $20 Australian dollars and $180 Swiss francs. Its total currency long posi tion is $300, and its total short position is $200. The BIS standardized framework imposes a capital requirement equal to 8 percent times the maximum absolute value of the aggregate long or short positions. In this example, 8 percent limes $200 million = $24 million. This method of calculating FX exposure assumes some partial, but not complete, offsetling of currency risk by holding opposing long or short positions in different currencies. Equities As discussed in the context of the RiskMetrics market value model, the two sources of risk in holding equities are (1) a firm-specific, oF unsystematic, risk element and. ©) amarket, or systematic, risk clement. The BIS changes for uncystematic risk by adding the long and short positions in any given stock and applying a 4 percent charge against the gross position in the stock (called the x factor). Suppose stock number 3, in Table 10-9, is IBM. The FI has a long $100 million and short $25 mil- lion position in that stock. Its gross position that is exposed to unsystematic (Firm- specific) risk is $125, which is multiplied by 4 percent to give a capital charge of $8 million. “For oxample, ones 1 and are adjacant to each atherin tas af maturity By comparison, zones 1 and 3 are not ajacent 1 eath other “ This ausunent oF 150 pescert was later duced 0 100 percent “Tis number can abo be recalculated in dskeausied asst ters wo compare with skeadjused assets fen the banking book. Thus, capital s meant to be a minimum of & percent of rsk-adusted assets, than {$370.78 « (1.08), or $370.78 « 125 = $4,534.75 i the equivalnt amount of wading bodk rk dusted assets supported by this aotal equrement.288 Part Two Meauring Risk TABLE 10-9 BIS Capital Requirement for Equities (Ilustration of x plus y Methodology soem S, 5 eboone ‘Under the propased two-part calevlation, there would be separate requirements for the position in each individual equity (12, the gross position) and for the net position in the market as a whole. Here we show how the system ‘would work for a range of hypothetical portfolins, assuming a capital charge of 4 percent for the gross positions and & percent forthe net positions. x rector ae y Factor o @ 8 @ © © @ ® Not Position Gross Position (difference Capital Sum oftong sumofshort (sum of 4% between 8% =Required Stock Positions Positions cols. 2and3) of Gross cols.2and3) of Net (gross + net) 1 100 ° 100 4 100 8 2 2 100 5 125 5 5 6 1 3 100 50 150 6 50 4 10 4 190 B 175 7 3 2 9 5 100 100 200 8 0 o 3 6 5 100 175 7 3 2 9 7 50 100 150 6 50 4 10 8 8 100 125 5 5 6 1 9 0 100 10 4 100 8 2 Market, or systematic, risk is reflected in the net long or short position (the ‘so-called y factor). In the case of IBM, this risk is $75 million ($100 long minus $25 shor’). The capital charge would be 8 percent against the $75 million, or $6 million. The total capital charge (x factor + y factor) is $11 million for this stock. ‘This approach is very crude, basically assuming the same systematic risk factor () for every stock It also does not fully consider the benefits from portiolio diver- sification (i.e,, that unsystematic risk can be diversified away). Concept 1. What isthe difference between the BS speci rsk and general market risk in measuring Questions trading portfolo risk? 2. What methods did the BIS model propose for calculating FX trading exposure? 3. How are uneystematic and systematic rcks in equity holdings by Fl reflected in charges assessed under the BIS model? THE BIS REGULATIONS AND LARGE-BANK INTERNAL MODELS ‘As discussed above, the BIS capital requirement for market risk exposure intro- duced in January 1998 allows large banks (subject to regulatory permission) to use their own internal models to calculate market risk instead of the standardized framework. (We examine the initiatives taken by the BIS and the major central banks, eg,, the Federal Reserve, in contzolling bank risk exposure through capital requirements in greater detail in Chapter 20.) However, the required capital cal- ‘culation has to be relatively conservative compared with that produced internally.Chapter 10 Market Risk 289 Acomparison of the BIS requirement for large banks using their internal models \with RiskMetries indicates the following, in particular: 1. In calculating DEAR, the FI must define an adverse change in rates as being in the 99th percentile rather than in the 95th percentile (multiply o by 2.33 rather than by 1.65 as under RiskMetrics) 2. The Fl must assume the minimum holding period to be 10 days (this means that RiskMetrics’ daily DEAR would have to be multiplied by J10).” ‘The FI must consider its proposed capital charge or requirement as the higher of: 1. The previous day’s VAR (value at risk or DEAR 10). 2. The average daily VAR over the previous 60 days times a multiplication fac- tor with a minimum value of 3, ie, capital charge = (DEAR) x (V0) x @). In general, the multiplication factor makes required capital signilicantly higher than VAR produced from private models However to reduce the burden of capital needs, an additional type of capital can, ‘be raised by Fis to meet the capital charge (or requirement). Suppose the portfolio DEAR was $10 million using the 1 percent worst case (or 99th percentile).!*The minimum capital charge would be” Capital charge = ($10 million) x (V0) x @) = $94.86 million As explained in Chapters 7 and 20, capital provides an internal insurance fund to protect an FI, its depositors and other liability holders, and the insur- ance fund (c.g., the FDIC fund) against losses. The BIS permits three types of capital to be held to meet this capital requirement: Tier 1, Tier 2, and Tier 3. Tier 1 capital is essentially retained earnings and common stock, Tier 2is essentially Iong-term subordinated debt (over five years), and Tier 3 is short-term subor- dinated debt with an original maturity of at least two years. Thus, the $94.86 zillion in the example above can be raised by any of the three capital types sub- ject to the two following limitations: (1) Tier 3 capital is limited to 250 percent of ‘Tier 1 capital, and (2) Tier 2 capital can be substituted for Tier 3 capital up to the same 250 percent limit. For example, suppose Tier 1 capital was §27.10 million and the Fl issued short-term Tier 3 debt of $67.76 million. Then the 250 percent limit would mean that no more Tier 3 (or Tier 2) debt could be issued to meet a target above $94.86 (627.1 x 2.5 = $67.76) without additional Ticr 1 capital being added. This capital charge for market risk would be added to the capital charge for credit risk and operational risk to get the F's total capital requirement. The different types of capital and capital requirements are discussed in more detail in Chapter 20, ts proposed that this ll be changed to a minimum hong period of ive days under Gaz! (at the end of 2005). Seo "Tho Now Basol Captal Acre: Third Coneultve Papo,” Base, Saitzecand, Apri 2003. Nace that this wil educe market rk capil equrements. © Using 2336 rater than 1.656. “© The dea of a mnimun multiplication factor of 3 isto create a scheme that “ncenve compat.” Spacifaly, if is using internal models constantly uncerestmate the arount of capital they need to meet ‘heir market rik exposure, regulators can punish those is by rising the multiplication factor to as high 254, Such a response may effectively put the Mout ofthe acirg business. The degree to which the ‘ubiplcation factor crazed above 3 depends on the numberof days an fl: mode! underesimates te ‘market rik aver the procecig year For example, an undorestenation eror that occurs on mate than 10) ays ou ofthe past 250 days wil result inthe multiplcaion fac’ being raised wo 4.290 Part Two Measuring Risk TABLE 10-10 ‘Market Risk Capital Requirement Kato of Market Name to Total Capital Requirement (%) Requid Tot ——-Bankof America 0.814% Bank of New York 0503 Suntrust ost Wels Fargo 0.964 Kecop 0978 PNC Financial 1.708 faire Fedele Cigroup 2240 SEPSIS HSE North America 2320 Wachovia 2540 |LP-Mergan Chase 5.20 “Table 10-10 lists the market risk capital requirement to the total capital require- ment for several large U.S. bank holding companies in June 2006. Notice how small the market risk capital requirement is relative to the total eapital equire- ment for these banks. Only JP. Morgan Chase has a ratio greater than 4 percent, ‘The average ratio of market risk capital requited to total capital required for the 10 bank holding companies is only 1.95 percent Moreover, very few banks, other than the very largest (above), report market risk exposures at all. Concept 1. What isthe 2s standarcized framewak for measuring market risk? Questions 2. Whatistheetfectof usng tne 99m percenble (1 percent worst case) rather than the 95th percentile (5 percent worst case) on the measured sizeof an FS market rik exposures? Summary In this chapter we analyzed the importance of measuring an FI's market risk ‘exposure. This risk is likely to continue to grow in importance as more and more loans and previously illiquid assets become marketable and es the traditional fran- chises of commercial banks, insurance companies, and investment banks shrink. Given the risks involved, both private FI management and regulators are invest- ing increasing resources in models to measure and track market risk exposures. We analyzed in detail three approaches Fls have used to measure market risk: RiskMetrics, the historic (or back simulation) approech, and the Monte Carlo sim- ulation approach. The three approaches were also compared in terms of simplicity | D. Hendricks and 2 Mile in “Bark Capital Requirements for Market Rick The Internal Model p= proach,” Federal Rosen Bani of New York Economic Poy Rviow, Dacersbar 1937, pp. 1=12, abo find ‘hat the impact of the market ns capta charges on required capital rats using internal mode is smal. “Thay calculate an increase in th level of requred capital from the general market rst component to range betwen 1 5 and 7.5 percent fr the banks they examined. B. Mile, in “What Market Rsk Capital Report Ing Wels Us about Bank Risk, Federal Reserve Bank of New York, EcoromicPolcy Review, September 2003, pp. 27-54, finds that ance the plementation othe market ek capital standards at the begining ‘of 1988, the bark holding comparies that were subpct tothe market capital requirements accounted for ‘more than 98 percent of the racing positons held by al US. banking organization. Fr these bank, ate bet ice captal reprecented it 1.9 percent of overall capa requirements of the medion bank.Chapter 10 Market Risk 291 and accuracy. Market risk is also of concern to regulators. Beginning in January 1998, banks in the United States have had to hold a capital requirement against the risk of their trading positions. The novel feature of the regulation of market risk is, that the Federal Reserve and other central banks (subject to regulatory approval) have given large Fis the option to calculate capital requirements based on their ‘own internal models rather than the regulatory model. Questions and Problems 1. What is meant by market risk? 2. Why is the measurement of market risk important to the manager of a finan- ial institution? 3, What is meant by daily earnings at risk (DEAR)? What are the three measurable ‘components? What is the price volatility component? 4, Follow Bank has a $1 million position in a five-year, zero-coupon bond with a face value of $1,402,552. The bond is trading at a yield to maturity of 7.00 percent. The historical mean change in daily yields is 0.0 percent, and the stan- dard deviation is 12 basis points. a. What is the modified duration of the bond? . What is the maximum adverse daily yield move given that we desire no more than a5 percent chance that yield changes will be greater than this ‘maximum? ‘c. What is the price volatility of this bond? 4. What is the daily earnings at risk for this bond? 5. What is meant by value at risk (VAR)? How is VAR related to DEAR in J. P. ‘Morgan’s RiskMetrics model? What would be the VAR for the bond in prob- Jem 4 for a 10-day period? What statistical assumption is needed for this cal- culation? Could this treatment be critical? 6. The DEAR fora bank is $3,500. What is the VAR fora 10-day period? A 20-day period? Why is the VAR for a 20-day period not twice as much as that for a Weday period? 7. The mean change in the daily yields of a 15-year, zero-coupon bond has been five basis points (bp) over the past year with a standard deviation of 15 bp. Use these data and assume that the yield changes are normally distributed. a. Whats the highest yield change expected ifa 90 percent confidence limit is required; that is, adverse moves will not occur more than 1 day in 20? b. What is the highest yield change expected if a 95 percent confidence limit isrequired? 8. In what sense is duration a measure of market risk? 9, Bank Alpha has an inventory of AAA-rated, 15-year zero-coupon bonds with a face value of $400 million. The bonds currently are yielding 9.5 percent in the over-the-counter market. a. Whats the modified duration of these bonds? b. What is the price volatility if the potential adverse move in yields is 25 basis points? c 5 z H rs =292 Part Two Meauring Rsk 4. If the price volatility is based on a $0 percent confidence limit and a mean historical change in daily yields of 0.0 percent, what is the implied standard deviation of daily yield changes? 10. Bank Two has a portfolio of bonds with a market value of $200 million. The bonds have an estimated price volatility of 0.95 percent. What are the DEAR and the 10-day VAR for these bonds? 11. Bank of Southern Vermont has determined that its inventory of 20 million eu- 10s (€) and 25 million British pounds (€) is subject to market risk. The spot exchange rates are $0.40/€ and $1.28/£, respectively. The a’s of the spot ex- change rates of the € and €, based on the daily changes of spot rates over the ppast six months, ate 65 bp and 45 bp, respectively: Determine the bank's 10-day ‘VAR for both currencies. Use adverse rate changes in the 9th percentile. 12. Bank of Alaska’s stack portfolio has a market value of $10 million. The beta of the portfolio approximates the market portfolio, whose standard deviation (r,) has been estimated at 15 percent. Whats the five-day VAR of this portio- lio using adverse rate changes in the 99th percentile? 13, Jeff Resnick, vice president of operations of Choice Bank, is estimating the ag- ‘gregate DEAR of the bank's postiolio of assets consisting of loans (L), foreign currencies (FX), and common stock (EQ). The individual DEARs are $300,700; $274,000; and $126,700, respectively. Ifthe correlation coefficients (p)) between Land FX, Land EQ, and FX and EQ are 0.3, 0.7, and 0.0, respectively, what is the DEAR of the aggregate portfolio? 14, Calculate the DEAR for the following portfolio with the correlation coeffi- Gents and then with perfect positive correlation between various asset groups. Estimated Assots, DEAR (esr) Moss) (rca) ‘stocks (5) $300,000 0.10 075 020 Foreign Exchange (>) 200,000 Bonds (6) 250,000 ‘What is the amount of risk reduction resulting from the lack of perfect positive correlation between the various asset groups? 15. What are the advantages of using the back simulation approach to estimate market risk? Explain how this approach would be implemented. 16. Export Bank has a trading position in Japanese yen and Swiss francs. At the close of business on February 4, the bankhad ¥300 million and SF10 million.The exchange rates for the most recent six days are given below: Exchange Rates per U.S. Dollar at the Close of Business 7 23 2 2A 1218 Japanese yen 112.13 11284 1121411505 1163511632 Swiss francs 14140 1.4175 1.4133 1.4217—14157__1.4123 a. What is the foreign exchange (FX) position in dollar equivalents using the FX rates on February 4? b. Whatis the definition of delta as it relates to the FX position? ¢. Whatiis the sensitivity of each FX position; that is, what is the value of delta for each currency on February 4? c 4 2 H g 14 Ey es F & 5 3 3Vm 18. 19, a. Chapter 10 Market Risk 293 4. What is the daily percentage change in exchange rates for each currency over the five-day period? ‘e. Whats the total risk faced by the bank on each day? What is the worst-case day? What is the best-case day? {, Assume that you have data for the 500 trading days preceding February 4. Explain how you would identify the worst-case scenario with a 95 percent degree of confidence. g. Explain how the 5 percent value at risk (VAR) position would be inter- preted for business on February 5. ‘h. How would the simulation change at the end of the day on February 5? ‘What variables and /or processes in the analysis may change? What vari- ables and /or processes will not change? Whats the primary disadvantage of the back simulation a in measur. {ng market risk? What effect does the inclusion of more observation days have as a remedy for this disadvantage? What other remedies can be used to deal with the disadvantage? ‘How is Monte Carlo simulation useful in addressing the disadvantages of back simulation? What is the primary statistical assumption underlying its use? In the BIS standardized framework for regulating risk exposure for the fixed income portfolios of banks, what do the terms specific risk and general market risk mean? Why does the capital charge for general market risk tend to under- estimate the true interest rate or price risk exposure? What additional offsets, or disallowance factors, are included in the analysis? An Flhas the following bonds in its portfolio: long I-year US. Treasury bills, short 3-year Treasury bonds, long 3-year AAA-rated corporate bonds, and ong 12-year B-rated (nonqualifying) bonds worth $40, $10, $25, and $10 mil- lion, respectively (market values). Using Table 10-7, determine the following: ‘a. Charges for specific risk. 1. Charges for general market risk. ‘c. Charges for basis risk: vertical offsets within same time bands only (ie, ignoring horizon effect). 4. The total capital charge, using the information from parts (a) through (©). Explain how the capital charge for foreign exchange risk is calculated in the BIS standardized model. If an FI has an $80 million long, position in euros, a ‘$40 million short position in British pounds, and a $20 million long position in. ‘Swiss francs, what will be the capital charge required against FX market risk? Explain the BIS capital charge calculation for unsystematic and systematic risk for an FI that holds various amounts of equities in its portfolio. What would be the total capital charge required for an FI that holds the following portfolio of stocks? What criticisms can be levied against this treatment of measuring the risk in the equity portfolio? Company. Long ‘Short Texaco $45 milion $25 milion Microsoft $55 milion $12 millon Robeco 320 milion Cita 515 milion c 5 z E i € g F F298 Part Two Measuring Risk 23. What conditions were introduced by BIS in 1998 to allow lange banks to use internally generated models for the measurement of market risk? What types of capital can be held to meet the capital charge requirements? 24. Dark Star Bank has estimated its average VAR for the previous 60 days to be $355 million. DEAR for the previous day was $30.2 million. 4. Under the latest BIS standards, what is the amount of capital required to be held for market risk? b. Dark Star has $15 million of Tier 1 capital, $375 million of Tier 2 capital, and $55 million of Tier 3 capital. Is this amount of capital sufficient? If not, ‘what minimum amount of new capital should be raised? Of what type? EE) Bank for International Settlements www.bis.org Board of Governors of the Federal Reserve www.federalreserve.gov J.P. Morgan Chase ‘www. jpmorganchase.com RiskMetrics www.riskmetrics.com EE View Chapter Notation at the Web site for the textbook (www.mbhe.com/ saundersée). cs i i A r Le Fa = Es