Turbulence in Mature Financial Markets · Until July 1998, the mature financial markets in the...

Until July 1998, the mature financial markets in theUnited States and Europe remained buoyant,

largely avoiding significant negative spillovers fromthe Asian crisis despite some episodes of increasedvolatility, most notably in October 1997.1 Governmentbond yields continued to decline, while equity pricesrecorded further strong gains, especially in continentalEurope, where markets surged in a number of coun-tries by 45–65 percent over end-1997 levels. Favorableeconomic developments, including very subdued in-flation, solid domestic demand growth in most coun-tries, and increased confidence in a successful launchof EMU contributed to this market buoyancy.2 In ad-dition, the mature financial markets were boosted by a“flight to quality” as investors shifted funds awayfrom Asia and other emerging markets. Despite thesegenerally favorable developments, however, therewere some signs of a weakening in sentiment in themonths leading up to July 1998. Major stock marketindices in the United States and the United Kingdomcontinued to advance, but the gains were increasinglynarrowly based, and market indices for “small cap”stocks began to weaken. Also, yield spreads on below-investment-grade bonds in the United States widenedby about 90 basis points from their historic lowsreached in mid-1997 prior to the Asian crisis.3 Else-where, equity markets and the exchange rate weak-ened further in Japan, where domestic economic con-ditions continued to worsen, and also came underdownward pressure in countries with strong trade linksto Asia or heavy reliance on commodity exports (no-tably, Canada, Australia, New Zealand, and Norway).

Equity markets in the United States and Europe gen-erally peaked in mid-July. While it is difficult to iden-tify a particular event that triggered the initial down-turn, several factors may have led investors to reassess

35

IIITurbulence in Mature Financial Markets

1For details, see the September 1998 International CapitalMarkets report, Chapter IV.

2Initial positive effects of the Asian crisis on economic activity inthe United States and Europe, including declines in inflation and in-terest rates and an associated boost to real incomes, may have con-tributed to a perception that negative spillovers from the crisiswould be relatively limited. In addition, market participants gener-ally viewed the prospect of a moderate slowdown in the UnitedStates as beneficial in terms of reducing overheating risks.

3This widening coincided with a general weakening in U.S. cor-porate earnings growth and an increase in the number of corporatecredit rating downgrades relative to upgrades.

Figure 3.1. Major Industrial Countries:Stock Market Price Indices1

(National currency; week ending January 2, 1997 = 100)

United States (S&P 500)

Germany(DAX 100)

Canada (TSE 300)

Japan (TOPIX)

France (SBF 250)

United Kingdom (ASX)Italy(MIBTEL)

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

Stock market average index Bank/financial sector index

250

200

150

100

50

0

250

200

150

100

50

0

250

200

150

100

50

0

250

200

150

100

50

0

250

200

150

100

50

0

250

200

150

100

50

0

250

200

150

100

50

0

Equity markets mostly peaked in mid-July before a sizablecorrection led by bank stocks; markets have since rebounded quitestrongly, particularly in the United States.

Source: Bloomberg Financial Markets, LP.1For United States, Standard & Poor’s 500 Index; for Japan, Price

Index of Tokyo Stock Exchange; for Germany, DAX 100 Index; forFrance, Société des Bourses Françaises 250 Index; for Italy, MilanStock Exchange MIB Telematico Index; for United Kingdom,Financial Times Stock Exchange All-Share Index; and for Canada,Toronto Stock Exchange 300 Composite Index.

©1998 International Monetary Fund

III TURBULENCE IN MATURE FINANCIAL MARKETS

the sustainability of historically high equity marketvaluations reached after a period of sustained andrapid price increases. First, the negative effects of theAsian crisis on output growth and corporate earningswere becoming more visible, particularly in the UnitedStates. In addition, it was increasingly apparent thatthe contraction in the Asian emerging market econo-mies was much deeper than initially expected, and thatprospects for early recovery in Japan had diminished.The deteriorating situation in Russia also contributedto concerns that the emerging market crisis mightspread beyond Asia. Bank stocks were hit particularlyhard, in part unwinding earlier sharp gains but also re-flecting concerns about bank exposures to emergingmarkets (Figure 3.1, preceding page). In credit mar-kets, spreads on lower-quality U.S. corporate bondswidened by a further 50 basis points in the first half ofAugust, but there was only a modest widening inspreads on investment grade bonds.

Recent Developments

The situation deteriorated rapidly in the second halfof August as the devaluation and unilateral debt re-structuring by Russia sparked a period of turmoil inmature markets that is virtually without precedent inthe absence of a major inflationary or economic shock.Neither Russia’s relative importance in the worldeconomy nor the size of bank exposures to Russia4 canfully explain the magnitude of the market movementsthat followed, including a broad-based reassessmentof the risks associated with emerging market invest-ments and a large-scale—partly involuntary—portfoliorebalancing across a range of global financial markets.In subsequent weeks, conditions in the mature finan-cial markets deteriorated sharply. The equity marketsell-off intensified, largely wiping out the gainsrecorded earlier in the year. In the United States, eq-uity markets bottomed out in late August, roughly 20percent below their highs, while European marketscontinued to decline through the first half of October,falling on average by about 35 percent. At the sametime, the decline in government bond yields acceler-ated, taking yields to their lowest levels since at leastthe mid-1960s and in some cases since World War II,as investors increasingly sought to shift funds into thesafest and most liquid assets (Figure 3.2). In the six-week period between mid-August and early October,for example, government bond yields fell by about 70basis points in Germany, 110 basis points in the UnitedKingdom, and 120 basis points in the United States,implying price gains in the range of 6–11 percent forthe benchmark seven- to ten-year bonds. Elsewhere in

36

4In 1997, Russia accounted for roughly 1!/2 percent of world GDPand 1.2 percent of world trade; BIS bank claims on Russia ac-counted for less than 1 percent of BIS total claims.

Figure 3.2. Major Industrial Countries:Nominal Interest Rates(Percent)

Germany

United States

Canada

France

Japan

United KingdomItaly

Long-term1 Short-term2

47

6

5

4

3

7

6

5

4

3

2

8

7

6

5

4

3

2

9

8

7

6

5

4

3

3

2

1

0

2

3

3

4

5

6

7

8

9

4

5

6

7

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

1997 Dec. 15,98

The downward trend in bond yields accelerated in the wake of theRussian crisis, while short-term rates have also declined, reflectingcuts in official interest rates.

Sources: WEFA, Inc.; and Bloomberg Financial Markets, LP.1Yields on government bonds with residual maturities of ten years or

nearest.2Three-month maturities: treasury bill rates for United States and

United Kingdom; interbank rate for Germany, France, Italy, andCanada; and deposit rate for Japan.

Europe, yield spreads over German rates widened totheir highest levels of the year within the euro area,and even more dramatically outside the euro area, withspreads for Denmark and Sweden widening by 30–40basis points in less than a month.

Corporate bond spreads also widened sharply, bothrelative to government bond yields and in terms ofthe spreads between high- and low-quality corporatebonds. Comprehensive data are most readily availablefor the United States, where the corporate bond mar-ket is relatively large and well developed (Figure 3.3).Yield spreads over U.S. treasury bonds for below-investment-grade bonds widened from about 375 basispoints immediately before the Russian debt restructur-ing to almost 600 basis points by mid-October, thehighest level since the collapse of the U.S. junk bondmarket at the beginning of the 1990s. For an averagehigh-yield bond, this spread widening was equivalentto a loss of about 8 percent on the value of the bond,more than half the average loss recorded over a longer4!/2 month period at the height of the market turmoil in1990–91. Spreads for the highest rated (Aaa) invest-ment-grade bonds also widened from about 90 basispoints in early August to about 150 basis points inmid-October, while spreads for lower-rated (Baa)bonds rose from about 150 to 230 basis points, in bothcases reaching levels that typically have been ob-served only during periods of recession. For the mostpart, the rise in spreads on higher-grade credits re-flected the fall in treasury bond yields rather than arise in actual borrowing costs. However, below invest-ment grade, the spread widening was also associatedwith a sharp increase in nominal yields. The U.S.credit market may have been particularly vulnerable toa setback, given that prolonged periods of economicexpansion such as that achieved by the United Statesin the 1990s often culminate in excessive borrowingand underpricing of risk.5 However, corporate bondspreads also appear to have widened in some Euro-pean markets, though time-series data on these spreadsare much more limited. For example, spreads on AAeuro sterling bonds over U.K. gilts widened fromabout 90 basis points to 130 basis points during thesame period,6 and spreads also widened for bonds is-sued by financial institutions in France and Germany.New debt issuance activity dropped off markedly,most notably in the high-yield market in the UnitedStates, where the volume of bonds issued in Octoberfell to about $2 billion, compared with a monthly av-erage of roughly $15 billion in the second quarter(Figure 3.4). A substantial though less pronounced

Recent Developments

37

5U.S. credit spreads had become unusually compressed in thepast two years amid a marked acceleration in private indebtedness,much of it securitized and held off banks’ balance sheets.

6Individual U.K. corporate bond spreads also widened signifi-cantly during the third quarter. See Bank of England, InflationReport (London: November 1998), p. 6.

Figure 3.3 United States: Yields on Corporateand Treasury Bonds1

(Percent)

1978 80 82 84 86 88 90 92 94 96 Dec. 15,98

1978 80 82 84 86 88 90 92 94 96 Dec. 15,98

Merrill Lynchhigh-yield bonds


Thirty-yeartreasury bond

Baa

Baa

Aaa

Aaa

20

18

16

14

12

12

10

10

8

8

6

6

4

4

2

0

–2

Yield

Yield Differential with U.S. Treasury Bond

In the wake of the Russian crisis, yield spreads on corporate bondswidened sharply to levels not seen since the early 1990s.

Sources: Board of Governors of the Federal Reserve System;Bloomberg Financial Markets, LP; and Merrill Lynch.

1Weekly data; the Moody’s ratings of the corporate bonds are shownin the panels. Yields on thirty-year treasury bonds of constant maturi-ties are used for the U.S. treasury bond. The shaded regions indicaterecession periods.


drop-off was observed in the issuance of U.S. invest-ment-grade bonds, and there are reports that high-yield corporate bond issuance also slowed sharply incontinental Europe.

In September and early October, indications ofheightened concern about liquidity and counterpartyrisk emerged in some of the world’s deepest financialmarkets. As discussed further below, a key develop-ment was the news of difficulties in, and ultimately thenear-failure of, a major U.S. hedge fund—Long-TermCapital Management (LTCM)—which had largehighly leveraged positions across a broad range ofmarkets, and substantial links with a range of U.S. andEuropean financial institutions. Although a privaterescue of LTCM, organized with the help of the NewYork Federal Reserve Bank, was announced on Sep-tember 23, the market reverberations intensified in theensuing weeks as previous positions were unwoundand as concerns increased about the extent to whichother financial institutions might be in trouble or facea need to unload assets into illiquid markets at dis-tressed prices. In response to these developments,market volatility increased sharply, and there weresome significant departures from normal pricing rela-tionships among different asset classes.7 In the U.S.treasury market, for example, the spread between theyield on “on-the-run” and “off-the-run” treasurieswidened from less than 10 basis points to about 15basis points in the wake of the Russian debt restruc-turing, and to a peak of over 35 basis points in mid-October, suggesting that investors were placing an un-usually large premium on liquidity (Figure 3.5).8 Interms of the value of the bonds, this spread wideningwas equivalent to a relative price movement of about4 percentage points—a relatively large differential forbonds of similar duration and the same underlyingcredit risk. Spreads between yields in the Eurodollarmarket and on U.S. treasury bills for similar maturitiesalso widened to historically high levels, as did spreadson fixed-for-floating interest rate swaps, pointing toheightened concerns about counterparty risk.

In exchange markets, the U.S. dollar continued tostrengthen on a multilateral basis through mid-August,remaining relatively stable against major Europeancurrencies but rising further against the Japanese yenand currencies of the major commodity-exportingcountries (Figure 3.6). As the emerging market crisistook on global dimensions, however, the dollar beganto weaken amid increased concerns about the down-

38

7While the observed movements in market prices suggest prob-lems of reduced liquidity and perhaps broader disruption of normalmarket functioning, reports of such problems remain largely anec-dotal (see the next section).

8This particular comparison refers to the spread between the 25-year and the 30-year benchmark treasury, but a similar pattern wasobserved for other maturities. On-the-run securities are the latestissue of a particular maturity. Off-the-run securities are the previousissues of the same maturity.

Figure 3.4. United States: Corporate Bond Market

Jan. Mar. May Jul.1998

Sep. Nov.

Jan. Mar. May Jul.1998

Sep. Nov.

Investment-grade AAA spread1

(right scale; percent)

High-yield grade B2 spread1

(right scale; percent)

Investment-grade bond issuance(left scale; billions of U.S. dollars)

High-yield bond issuance(left scale; billions of U.S. dollars)

0.8

6

5

4

3

2

1

0

80

60

40

40

30

20

10

0

20

0

0.6

0.4

0.2

0.0

Investment Grade

High Yield

The widening in yield spreads was associated with a marked drop-offin new issuance activity, particularly for lower-quality bonds.

Source: Bloomberg Financial Markets, LP.1Spread between yields on corporate bonds and ten-year U.S. gov-

ernment bonds. Monthly average of daily observations.

side risks to U.S. growth and a shift in market expec-tations about the direction of U.S. monetary policyfrom modest tightening to significant easing.9 Thesedevelopments, combined with signs in Japan ofgreater progress with long-awaited bank reform10 andadditional moves there toward fiscal and monetarystimulus, significantly altered the balance of risks fac-ing investors with yen-denominated exposures. Theinitial weakening of the dollar was relatively orderly;it fell by less than 10 percent against both the yen andthe deutsche mark between mid-August and earlyOctober. However, the situation changed in the weekbeginning October 5 when the dollar fell by almost 15percent against the yen in the space of 3 days, includ-ing the largest one-day movement in the yen/dollarrate since the collapse of the Bretton Woods system.This latter adjustment mainly reflected a sharp generalappreciation of the yen: the dollar fell less than 2 per-cent against the deutsche mark over the same period(Figure 3.7). It also coincided with an unusuallyabrupt steepening of mature market yield curves out-side Japan, as bond yields rose from their historic lowswhile short rates continued to fall. Over the sameweek, for example, the gap between three-month andten-year rates widened by about 85 basis points in theUnited States, 50 basis points in Germany, and 60basis points in the United Kingdom. The coincidenceof such dramatic moves in the yen/dollar rate and inmajor credit markets is difficult to explain in terms ofchanging economic fundamentals alone, and appearsto have reflected a large-scale unwinding of yen-denominated exposures—the “yen carry trade”—am-plified by technical factors linked to stop-loss ordersand dynamic hedging strategies (Box 3.1, page 43).These developments were a particularly visible mani-festation of a global move by investors to close outopen positions and reduce leverage in the wake of theheightened market turmoil.

In response to these developments, the U.S. FederalReserve Board moved to cut interest rates on three oc-casions beginning in late September. An initial cut of!/4 of 1 percentage point in the target federal funds ratewas announced following the Federal Open MarketCommittee (FOMC) meeting on September 29 butfailed to have any significant effect in calming mar-kets; spreads continued to widen, equity markets fellfurther, and volatility continued to increase. Againstthis background, the Federal Reserve followed up onOctober 15 with !/4 of 1 percentage point cuts in boththe federal funds target and the discount rate, a movethat proved to be the key policy action that stemmed

Recent Developments

39

9For example, the implied yield on the eight-month federal fundsfutures contract fell from about 5.6 percent in May and June, to 4.25percent by mid-October, suggesting that market participants ex-pected a sizable easing over the subsequent months.

10Recent developments in the Japanese financial system are dis-cussed in Chapter I (Box 1.2).

Figure 3.5. United States: Developmentsin Fixed-Income Securities Markets(Basis points)

140

120

100

80

60

40

20

0

Repo Spread1 Swap Spread2

CorporateSpread3

LiquiditySpread4

Jan. Apr. Jul. Oct. Dec.151998

Jan. Apr. Jul. Oct. Dec. 151998



Russia F1 F2 Russia F1 F2

Russia F1 F2 Russia F1 F2

700

600

500

400

300

200

100

0

140

40

35

30

25

20

15

10

5

0

120

100

80

60

40

20

0


Baa

Aaa

Growing concerns about liquidity and counterparty risk werepartially alleviated after the second cut in the Fed funds rate targetin mid-October.

Source: Bloomberg Financial Markets, LP.Note: The vertical lines represent the following: Russia = Russian

debt moratorium (August 17); F1 = Federal Reserve interest rate cut(September 29); and F2 = Federal Reserve interest rate cut (October 15).

1Rate on three-month U.S. treasury repos minus yield on three-monthU.S. treasury bill.

2Spread of fixed-rate leg of 10-year U.S. dollar interest rate swapsover yield on 10-year U.S. treasury bond.

3Spread over 30-year U.S. treasury bond.4Spread of 25-year U.S. treasury bond over a 30-year on-the-run U.S.

treasury bond.


and ultimately reversed the deteriorating trend in mar-ket sentiment. The second easing was not particularlylarge, but the fact that it came so soon after the firstrate cut and outside a regular FOMC meeting—thefirst such move since April 1994—sent a clear signalthat the U.S. monetary authorities were prepared tomove aggressively if needed to ensure normal marketfunctioning. The Federal Reserve subsequently cutboth the federal funds target and the discount rate by afurther !/4 of 1 percentage point at the next FOMCmeeting on November 17, noting that although finan-cial market conditions had settled down materiallysince mid-October, unusual strains remained. Else-where, the Bank of Japan reduced the guideline for theuncollateralized call rate by 25 basis points to !/4 per-cent on September 9, and official interest rates havebeen reduced since late September in Australia,Canada, and Europe. While these moves have beenmotivated primarily by domestic considerations, theyhave also played a helpful role from a global perspec-tive by contributing to the broad easing of monetarypolicy in the industrial countries.

Since mid-October, a significant degree of calm re-turned to financial markets. Indicators of reduced li-quidity and heightened counterparty risk were sub-stantially, though not completely, reversed (see Figure3.5), and exchange rate volatility declined somewhat.Mature equity markets also rebounded, most dramati-cally in the United States, where the main indices morethan regained their earlier losses by late November,before a moderate downward correction. Equity mar-kets in Europe also strengthened, although they re-main significantly below their earlier peaks. Sovereignyield spreads over German rates also generally nar-rowed within continental Europe, particularly in theeuro area. In addition to the recent interest rate movesin the industrial countries, further steps in Japan to ad-dress the problems in the banking sector and to pro-vide additional fiscal stimulus, and agreement on aneconomic program in Brazil, played important roles inthe restoration of market confidence. In credit mar-kets, spreads on U.S. corporate bonds narrowed byabout 90 basis points in the high-yield market and byabout 25 basis points on investment-grade bonds; cor-porate bond spreads also narrowed somewhat in theUnited Kingdom.

Significance for Economic Activity

The significance for real economic activity of the re-cent turmoil in mature financial markets remains some-what uncertain. In credit markets, although spreadshave narrowed since mid-October, they remain wellabove pre-August levels and near levels that in the pasthave generally been associated with periods ofmarkedly slower growth if not actual recession. ByNovember, there were signs of a significant pickup in

40

Figure 3.6. Major Industrial Countries:Effective Exchange Rates(Logarithmic scale; 1990 = 100)

130

130

140

150

160170180

120

120

120

120

110

110

110

120

120

110

100

90

130

100

100

100

90

90

90

80

80

80

70

60

130

110

110

110

100

100

100

90

90

90

80

70

60

80

United States Germany

France

Italy

United Kingdom

Japan

Canada

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

1991 93 95 97 Nov.98

Real effective exchange rate1 Nominal effective exchange rate2

The U.S. dollar continued to strengthen until mid-August, before amoderate decline associated in part with a sharp rebound in the yen.

1Defined in terms of relative normalized unit labor costs in manufac-turing, as estimated by the IMF’s Competitiveness Indicators System,using 1989–91 trade weights.

2Constructed using 1989–91 trade weights.

new debt issuance in the U.S. high-yield bond market,though volumes remained below their earlier levels.Some further recovery in volumes and narrowing ofspreads can reasonably be expected as long as financialmarket conditions continue to stabilize and economicgrowth remains well sustained, but a return to thehighly compressed credit spreads applying before theRussian crisis is probably neither likely nor desirable. Itis worth noting that, except for low-grade credits, actualborrowing costs in mature markets do not appear tohave increased significantly during the recent episodeand may even have declined for many borrowers.

The risks for real activity will also depend in part onthe impact of the turbulence in emerging and maturemarkets on mature banking systems. Available, but in-complete, balance-sheet data indicate that, as of mid-1998, banking system loan exposures to emergingmarkets amounted to about $536 billion in the euroarea (9 percent of 1997 GDP),11 $211 billion in Japan(5 percent of GDP), and roughly $120 billion in boththe United Kingdom and the United States (9 percentand 1!/2 percent of GDP, respectively; Table 3.1).12

These exposures already reflect a substantial pullbackin net bank credit outstanding mainly to the Asianemerging market economies, since the beginning ofthe year. So far, a number of major financial institu-tions have announced significant profit declines as aresult of their losses on emerging market investmentsand the recent turbulence more generally. For exam-ple, in the third quarter of 1998, profits of U.S. money-center banks declined to about half the level recordeda year earlier. However, bank rating agencies presentlyestimate that the turbulence has had a manageable im-pact on the mature banking systems, in part becausethe hardest-hit financial institutions were generallywell capitalized going into the turbulence, were rea-sonably well provisioned (or had state guarantees) onmuch of their emerging market portfolios (and have inmany cases increased provisioning further), and werefairly profitable in the early part of 1998.

The paucity of off-balance-sheet data makes it diffi-cult to assess overall exposures and vulnerabilities, al-though in some instances these exposures may be rel-atively large. For example, one estimate suggests thattotal credit exposure (including off-balance-sheet po-sitions) of foreign banks to Russia may have been 40to 65 percent higher than on-balance-sheet exposure.13

In addition, as the recent period of turbulence amplydemonstrated, when emerging market financing andleveraged derivatives positions are unwound, the ma-


41

11Exposures for all European Union banking systems totaledabout $676 billion (9 percent of GDP).

12These data include securities holdings.13Off-balance-sheet exposures may increase or decrease total ex-

posure. For example, counterparty risks increase exposure, sincecounterparties may default on positions; hedges decrease it, sincethey decrease the exposure to market risk.

Figure 3.7. Selected Countries: BilateralU.S. Dollar Exchange Rates(Currency units per U.S. dollar)

150

140

130

120

110

0.70

0.65

0.60

0.55

0.50

Japan

United Kingdom

Germany

Switzerland

2.10

2.00

1.90

1.80

1.70

1.70

1.60

1.60

1.50

1.50

1.40

1.30

1.20





In early October, the dollar weakened sharply against the yenbut more modestly against other major currencies.

Source: Bloomberg Financial Markets, LP.


ture markets that finance these positions are alsoaffected. Even if the off-balance-sheet exposures ofmature banking systems to emerging markets are rela-tively limited, therefore, they may still have signifi-cant consequences for the mature derivatives markets.

Looking ahead, the mature banking systems and theinstitutions within them face three risks that are par-ticularly relevant to the outlook. First, in addition tothe direct exposures described above, financial institu-tions have significant indirect exposures to emergingmarket risks, including to counterparties that take onemerging market risks, and to the mature securitiesmarkets themselves. As recent events have shown, ma-ture markets can experience sizable turbulence di-rectly related to developments in emerging markets.Second, there are concerns, as discussed furtherbelow, that risk management practices continue to lagdevelopments in financial markets, increasing poten-tial vulnerability to turbulence. Third, the fact that thecredit cycles in some major countries are in a maturephase, with economies close to potential after severalyears of strong growth, suggests that banks could facea deterioration in credit quality on domestic expo-sures, including from a slowdown in economic activ-ity.14 Available data on bank lending volumes do notpoint to any broad-based curtailment of credit avail-ability to date, outside Japan, in part because compa-nies have been able to draw down existing bank credit

lines as a substitute for reduced access to the corporatebond and commercial paper markets.15 However, aFederal Reserve Board survey in November showedthat U.S. bank lending practices had tightened signifi-cantly, reflecting increased concern about the eco-nomic outlook, and with the number of domestic re-spondents reporting tightened loan standards reachingthe highest level since 1990. Some tightening of loanstandards is probably a welcome development—in-deed, in July 1998 the Federal Reserve expressed con-cern that standards had become too lax—but a sub-stantial cutback in credit availability clearly wouldhave negative implications for economic growth.

Equity market developments also are likely to havea significant bearing on near-term growth prospects—particularly in the United States, where equity pricegains have been a major driving force behind the rapidgrowth in private consumption in recent years. For ex-ample, U.S. consumer confidence fell significantlyfrom the record high reached in June, no doubt at leastpartly reflecting the downward correction in the stockmarket, before a partial recovery in November as thestock market rebounded; there have been indications ofsimilar effects in some other countries. Given that eq-uity prices have recovered rapidly since early October,it seems unlikely that the recent relatively short-lived

42

14One credit rating agency takes the view that the U.S. creditcycle has already peaked.

15Data for the United States indicate that bank lending growth ac-celerated in the period after late August, in part reflecting growth incommercial and industrial loans. Similar trends are evident in theUnited Kingdom.

Table 3.1. Claims of Banks in BIS-Reporting Countries on Selected Emerging Markets as of June 19981

(Billions of U.S. dollars)

All BIS-Reporting United United Euro Countries Japan Kingdom States Area2 France Germany

Asia 639.4 186.7 84.6 31.7 237.3 55.3 92.6China 59.3 17.5 7.8 2.1 23.5 8.0 7.4Hong Kong SAR 174.6 54.6 32.8 6.1 59.4 12.6 24.1Asian-5 210.3 74.3 15.1 16.6 98.9 20.0 26.9

Latin America 295.7 14.8 23.1 64.2 140.7 25.1 39.5Argentina 60.2 1.7 5.2 10.2 34.3 5.2 7.5Brazil 84.6 5.2 5.8 16.8 37.6 7.9 12.8Mexico 62.9 4.4 5.7 16.7 24.9 6.1 6.1

Transition countries 133.4 4.1 3.9 12.4 92.4 11.1 52.5Russia 75.9 1.0 1.8 7.8 51.5 6.7 31.3

Middle East 57.3 3.0 6.5 5.3 25.7 7.0 11.6

Africa 58.3 2.3 3.9 4.8 39.4 18.7 9.4

All emerging markets 1,184.0 210.9 122.0 118.4 535.5 117.2 205.6

Sources: BIS; and IMF staff calculations.1On-balance-sheet claims, excluding claims on offshore centers (with the exception of Hong Kong SAR

and Singapore, which are included in Asia).2Because data are not reported for Greece and Portugal, data are for Austria, Belgium, Finland, France,

Germany, Ireland, Italy, Luxembourg, the Netherlands, and Spain.

stock market correction alone will be sufficient to havea significant dampening effect on consumer spending.Indeed, with equity markets back in some cases to neartheir all-time highs, previous concerns about the sus-

tainability of current market valuations have resurfaced(Box 3.2, page 48; and Figure 3.8), and the possibilityof a more pronounced downward correction in equityprices remains an important risk.


43

One outstanding feature of foreign exchange market de-velopments in the past few months was the sharp and un-precedentedly sudden appreciation of the yen vis-à-vismost major currencies in early October that ended thetrend of yen depreciation since mid-1995. While the recentyen appreciation to some extent may have been warrantedby fundamental forces (among them a reassessment of rel-ative monetary policy stances), the timing and the speed ofthe exchange rate changes strongly suggest that short-termtrading conditions (such as the large-scale unwinding of“yen-carry trades”) and technical market factors (includingrepercussions from the expiration of barrier options) con-tributed significantly to the sharp dynamic adjustments inthe yen/dollar market. This box describes these adjust-ments and the technical features that drove them.

With a brief interruption in mid-1997, the yen depreci-ated vis-à-vis the dollar by some 40 percent during the pastthree years and reached an eight-year low at ¥147.26 perdollar on August 11, 1998 (see figure). This long-runningappreciation of the dollar was abruptly and sharply reversedin the wake of renewed turbulence in emerging markets. Ofparticular interest are the developments surrounding the un-precedented, sharp appreciation of the yen during October6–9, 1998, when the yen appreciated 15 percent vis-à-visthe dollar. This episode was driven by a confluence of fac-tors—some of a fundamental nature, others largely techni-cal but generating positive feedback dynamics.

Various catalysts may have sparked an initial rally inthe yen and the 6.2 percent surge in the Nikkei stockindex on October 7, 1998. The new draft banking bill wassubmitted to parliament on that day; there was talk of anadditional fiscal stimulus package; and the relative mon-etary policy stance in Japan, the United States, andEurope was reassessed in part based on the Bank of Japan(BOJ) balance sheet for September, which was inter-preted by some market participants as casting doubts onprevious hints of extensive monetary easing by the BOJ.

The initial spate of dollar selling, in the wake of someturbulence in U.S. markets and the cut in interest rates bythe Federal Reserve, may have induced a change in senti-ment that the dollar’s long-standing strengthening vis-à-vis the yen had run its course. The impression of a turningpoint was reinforced by indications of a cascade of dollarselling by institutional investors, including hedge funds.Large financial institutions were reportedly unwindingtheir yen-carry trade positions,1 as part of the process ofinternational deleveraging. According to market partici-pants, technical factors stemming from standard hedging

procedures may have contributed to the sudden surge inthe yen. In particular, the cancellation of complex optionsas the yen surged through several trigger levels and deal-ers’ unwinding of hedges against these options, as well asthe bunching of limit orders, created additional momen-tum that boosted demand for yen. Some temporary de-mand for yen appears to have originated from foreign in-vestors who had short positions in Japanese stocks anddecided to cover them ahead of new rules (effectiveOctober 23) that bar investors from selling borrowedstock in declining markets. Foreign exchange trading vol-ume surged initially, but liquidity tightened up quickly(see figure) and contributed to large price discontinuities.In these circumstances, market participants took a cau-tious view, reflecting, among other factors, the very sharpincrease in implied foreign exchange volatility.

The dollar eventually stabilized on October 9, 1998, re-portedly after market participants began to think that theFederal Reserve was prepared to intervene in support of thedollar. But unlike in June, when a concerted intervention ofthe Federal Reserve and the BOJ was aimed at boosting thevalue of the yen, no central bank interventions are reportedto have taken place between August and November 1998. Inthe following weeks, the yen/dollar rate weakened slightly,and liquidity returned to the foreign exchange market.Implied volatility remained high, however, as uncertaintyabout prospects for Japan and in particular its financial sys-tem lingered (see Chapter I, Box 1.2).

Yen-Carry Trade

Tempted by low borrowing costs in Japan, proprietarytrading desks of major financial institutions and hedgefunds, and even some corporations, had borrowed in yen toinvest in U.S., European, and emerging market assets,thereby shorting the yen. The borrowing took variousforms. Funds were either raised in the interbank market,through term repo agreements, or by issuing money marketpaper. Subsequently the funds were swapped for foreigncurrency or exchanged in the spot market. Owing to the in-terest rate differentials between Japan and, say, dollar assetsand the appreciation of the dollar vis-à-vis the yen, these po-sitions had been highly profitable during the past few years.

Japanese banks also exploited the yen-carry trade byaccumulating open foreign asset positions. In the firstthree quarters of 1998, the net holdings of assets denom-inated in foreign currencies increased by about $44 bil-lion, while the net holdings of yen-denominated assetsabroad declined by $103 billion (see figure). Against thebackground of the yen depreciation, a shift toward a“long” position in foreign currencies became increas-ingly attractive to Japanese banks.

Box 3.1. Recent Dollar/Yen Exchange Rate Movements

1The yen-carry trade was discussed in detail in the September1998 International Capital Markets report, p. 44; it is also ex-plained below. (Box continues on next page.)


In exchange markets, the most notable by-product ofthe recent turmoil has been a substantial strengtheningof the yen, which by late November was up roughly20 percent in nominal effective terms from its August

low. The multilateral value of the U.S. dollar weakenedby about 6 percent over the same period, while that ofthe deutsche mark was little changed on balance,though it strengthened significantly against the dollar.

44

Box 3.1 (concluded)

Currency Markets and Net Foreign Position of Japanese Banks

Sources: Bank of Japan; Bloomberg Financial Markets, LP; and Nikkei.1Average daily turnover in January-June 1998 in the spot and swap markets was $14 billion and $17 billion, respectively.2The underlying data are in yen (left scale); the U.S. dollar values (right scale) are computed at an exchange rate of 120 yen per U.S. dollar.

160 30 1.85

1.80

1.75

1.70

1.65

1.60

1.55

25

20

15

10

5

0

150

140

130

120

110

100

60 40 333

250

167

83

0

–83

30

20

10

0

–10

48

36

24

12

0

Feb. Apr. Jun.1998

Aug. Nov. 16 Feb.

1994Sep. 21 Oct.1998

Nov. 2 95 96 97 Sep.98

Apr. Jun.1998

Aug. Nov. 19

Japanese yen/U.S. dollar(left scale)

Implied volatility(right scale)

Total Yen-denominated

Foreign-currency-denominated

Swap

Spot

Spot Exchange Rate and Implied Volatility

Daily Trading Volume in the U.S. Dollar/Yen Market1

(billions ofU.S. dollars)

Net Foreign Position of Japanese Banks(left scale, trillions of yen;

right scale, billions ofU.S. dollars)2

Spot Exchange Rate:Deutsche Mark/

U.S. Dollar

As discussed above, the sharp rise in the yen appears tohave been at least partly attributable to the sudden un-winding of large short yen positions, amplified by tech-nical factors, and it remains to be seen to what extent

the current yen/dollar alignment will persist as the im-pact of these essentially one-off factors begins to fade.

At a more fundamental level, however, the yen’srise also appears to have reflected expectations of ad-


45

Worsening investment opportunities following the tur-bulence in some emerging markets, and corrections in ma-ture markets during the summer of 1998, prompted a re-versal of yen-carry trade positions. In addition to profittaking, the unwinding was also triggered by institutionalinvestors, including hedge funds, that were confrontedwith margin calls on positions in other markets. Many in-stitutions closed out carry trades because they faced ashortage of liquidity as banks increasingly cut credit linesto leveraged investors—as part of a global reevaluation ofthe willingness to take risks. Some margin calls on hedgefunds might have dampened the yen appreciation as theytriggered the sale of liquid yen assets to raise cash, whichwas subsequently converted into foreign currency.

Japanese institutions began to repatriate funds in lightof rising funding costs—including an increasing “Japanpremium” and higher rates on basis swaps (currencyswaps)2—and reportedly deteriorating access to interna-tional interbank markets. In this situation, long dollar po-sitions represented a potential source of vulnerability andcontributed to unwinding pressures. These pressures wereintensified by the need to overcome capital shortfalls on

their half-year balance sheets (September 30) and as an in-creasing number of Japanese banks reconsidered the via-bility of their overseas business. Selling pressures on thedollar reportedly originated from Japanese investorspulling funds out of U.S. securities as those markets soft-ened and the higher exchange rate volatility increased for-eign-currency risk and made hedging more expensive.

Technical Factors

Large dollar-yen movements were likely exacerbated bya series of technical factors, such as stop-loss orders and,more important, by the cancellations of barrier options3

and the unwinding of associated hedging positions bydealers. Leverage and the hedging of yen positions canoften be achieved more cheaply through derivatives. Thevolume of outstanding yen foreign exchange contracts hadgrown at end-June 1998 to $3.4 trillion, of which about$400 billion corresponded to options (see table). Althoughno official breakdown by type of options is available,market participants consider barrier options a popularinstrument.

Knockout options (a special form of barrier options)4 arewidely used as a hedge of currency risk because they areless expensive than standard options. But knockout optionsprovide protection only against moderate exchange ratechanges and leave the investor unhedged against large cur-rency movements, since as soon as the exchange ratebreaches a certain level, the knockout option is canceled. AJapanese exporter, for example, might buy dollar knockoutput options, which expire prematurely if the dollar ex-change rate drops below a certain level, to protect againsta (moderate) depreciation of the dollar. A drop of the dol-lar large enough to trigger the cancellation of the optionwould, however, expose the exporter to losses. In response,the exporter might be inclined to sell dollars into a fallingmarket to cut his expected losses.

Additional feedback in a falling market may have origi-nated from the dynamic hedging strategies commonly em-ployed by dealers who sell knockout options. As the dollarexchange rate fell and knockout call options were canceled,dealers immediately sold the long dollar positions they heldas a dynamic hedge for these options. The hedging ofknockout put options typically involves more complex buy-ing and selling of standard options as exchange rates vary.In a down market, standard put option values are bid upeven higher by the dynamic hedging and expose dealers tosignificant losses. These reactions can contribute to an over-shooting in the price (and implied volatility) of options.

2In a basis swap, a bank swaps principal and interest obliga-tions on a yen liability for dollar principal and interest pay-ments. This arrangement converts yen-denominated debt heldby a Japanese bank into dollars and thus helps fund the bank’soverseas activities. Basis swaps had offered Japanese banks anattractive source of foreign currency at a time when their creditratings made it difficult to borrow abroad.

3Barrier options are options that either come into effect or arecanceled if the price of the underlying asset crosses a stated level.

4Knockout options are barrier options that are canceled as theunderlying asset price breaks through a specified level.

Japan: Notional Amounts of Over-the-Counter(OTC) Derivatives Outstanding, End-June 1998(Trillions of U.S. dollars)

Of Which: Of Which:Grand 1 Year OTC 1 YearTotal or Less Options or Less

Foreign exchange contracts 3.37 2.87 0.40 0.26

With reporting dealers 2.44 2.18 0.31 0.20With other financial

institutions 0.58 0.47 0.07 0.05With nonfinancial

customers 0.35 0.22 0.02 0.02

Single-currency interest rate contracts 9.54 5.27 0.89 0.25

With reporting dealers 7.54 4.42 0.53 0.14With other financial

institutions 1.41 0.72 0.23 0.08With nonfinancial

customers 0.59 0.13 0.13 0.03

Source: Bank of Japan, Regular Derivatives Market Statisticsin Japan (Tokyo).


ditional policy steps to boost growth in Japan and ofslower growth in the United States, which wouldimply a narrowing of the current large divergences inrelative cyclical positions. So far, there is little evi-dence of such narrowing in the economic data, how-ever, and the recent abrupt strengthening of the yencould be damaging to prospects for recovery in Japanat a time when domestic private sector demand re-mains very weak. Indeed, as discussed in Chapter IV,the appreciation of the yen is one of the factors behindthe further downward revision in the 1999 staff fore-cast for Japan. From a global perspective, however, amoderate realignment of the yen/dollar rate also car-ries several potential benefits. First, by introducing agreater element of uncertainty about the future direc-tion of exchange rates, it has probably reduced the in-centives for investors to take on large short-yen expo-sures that may have contributed to the emergence ofexchange market strains and asset price bubbles else-where in the global financial system over the past twoyears. Second, it has helped to alleviate downwardpressures on the currencies of other emerging marketeconomies in Asia and elsewhere—particularly thosewith explicit or implicit links to the U.S. dollar—andthus has provided scope for these economies to pursuesomewhat easier monetary policies. Particularly in thecurrent environment where global deflationary pres-sures remain relatively strong, such cautious monetaryeasing is generally helpful in terms of reducing therisks of a more pronounced slowdown in worldgrowth. Third, it has provided additional scope for theJapanese authorities to pursue an aggressively expan-sionary monetary policy aimed at boosting domesticdemand and easing strains in the financial sector.Nonetheless, a further significant strengthening of theyen or weakening of the dollar in the short term wouldalso carry risks, in terms of both the prospects for eco-nomic recovery in Japan and the scope for further in-terest rate cuts in the United States.

Systemic Aspects of Mature MarketTurbulence

Recapping some of what has already been dis-cussed, the turbulent dynamics in mature markets hadbeen preceded by a steady buildup of prices in the ma-ture equity and bond markets during the years andmonths preceding the Russian crisis in mid-August1998. The long-standing rise in asset values was sup-ported by several important developments in the mid-to-late 1990s—including the widespread reduction ininflation rates in the world economy; the continuednoninflationary expansion in the United States, andthe belief by some that the U.S. economy had entereda new, high-productivity growth age; continued flowsof funds into U.S. and other mature equity and bondmarkets; and the relatively successful convergence

46

Figure 3.8. United States: Equity MarketPerformance

35

14

8

6

4

2

0

–2

–4

–6–8

6

4

2

0

12

10

8

6

4

2

0

30

25

20

15

10

5

0

S&P 500 Price-Earnings Ratio

S&P 500 Dividend Yield (percent)

S&P 500 Price-Book Value Ratio

Equity Yield Gap (percent)1

1926 36 46 56 66 76 86 98

1926 36 46 56 66 76 86 98

1926 36 46 56 66 76 86 98

1970 75 80 85 90 95 Nov.98

With the latest rebound in U.S. equity prices, several indicators ofmarket valuation have moved further away from their long-termaverages; however the yield gap vis-à-vis treasury bonds remainsbelow recent peaks, reflecting the decline in government bond yields.

Sources: Board of Governors of the Federal Reserve System;Bloomberg Financial Markets, LP; WEFA, Inc.; and Standard and Poor’s.

1Difference between the yield on long-term government bonds andthe inverse of the price-earnings ratio.

process toward EMU. All of these developments con-tinued through the early summer, amid earlier warningsigns that many advanced country equity markets, notjust in the United States, were reaching record andperhaps unsustainable levels. In addition, as early asmid-1997, differences in the cost of borrowing be-tween high- and low-risk borrowers began to narrowto the point where several advanced country centralbanks sounded warnings that credit spreads werereaching relatively low levels and that lending stan-dards had been relaxed in some countries beyond areasonable level.

In mid-July 1998, equity markets in the advancedcountries began to decline somewhat on reports of poorcorporate earnings and concerns about a slowdown inU.S. economic growth. At the same time, some maturemarkets—notably the U.S. fixed-income markets—began experiencing a widening of interest rate spreadsbetween low- and high-quality borrowers, in part theresult of concerns over growth prospects and what thismight imply for the ability of higher-risk debtors to ser-vice obligations in the future and in part the result ofthe continued flight to quality and liquidity (and awayfrom Asian emerging markets). This relatively smallwidening of interest rate spreads between low- andhigh-quality fixed-income securities was then followedby a more dramatic widening of spreads, and subse-quently by a period of severe turbulence in mature andinternational financial markets, triggered, and driven,by several related events—including the Russian uni-lateral restructuring of GKOs (ruble-denominated dis-count instruments); the immediate partial closing out,and deleveraging, of positions in other emerging mar-kets; and later in mid-September the near collapse of ahighly leveraged hedge fund, LTCM.

Dynamic adjustments in other emerging markets re-lated to the Russian crisis necessarily entailed someadjustments in the mature markets as well, reflectingthe important role of these markets in financing andleveraging investments in Russia and in other emerg-ing markets. But it would normally be expected thatsuch adjustments would occur relatively smoothly andwithout the kind of severe financial turbulence that oc-curred in September and October 1998 in some of thedeepest and most liquid markets in the world. By def-inition, deep and liquid markets, such as those in theUnited States, might have been expected to be able toabsorb the after-effects of what were relatively moder-ate shocks with relatively limited price and liquidityeffects.

However, the financial turbulence in the maturemarkets appeared out of proportion to the events thattriggered it. The events surrounding the Russian uni-lateral debt restructuring led to large investment andtrading losses and changed market perceptions of de-fault and convertibility risk, which together affectedthe balance of risks and returns in international portfo-lios. Because of the new financial calculus that re-

sulted, the internationally active financial institutionsand other asset managers appear to have engaged in awholesale reassessment and repricing of financial risk,which was accompanied by a rebalancing and delever-aging of international portfolios in a short period oftime, accented by risk avoidance, market illiquidity,and extreme price movements.

As a result, and contrary to what would normally beexpected, the mature markets subsequently experi-enced dramatic, and in some cases unprecedented,price and liquidity adjustments that cut across matureequity, fixed-income, currency, and derivative marketsand caused some of them to become illiquid, and attimes to seize up temporarily; liquidity spreadsreached record highs.16 Despite the apparent concen-tration of mature market turbulence in U.S. financialmarkets and the focus of attention on some newswor-thy U.S. financial institutions, internationally activeEuropean and Japanese financial institutions were in-volved in similar leveraged risk taking, in some caseson a very large scale, and, as of end-November 1998,appear to be undergoing a similar process of risk re-assessment and rebalancing. The negative impact onasset values during the most turbulent subperiod—between mid-September and mid-October—was suffi-ciently severe that it triggered fears of significant neg-ative spillover effects on world economic growth.

The severe nature of the mature market turbulenceraises several issues about private risk and portfoliomanagement, banking supervision, financial marketsurveillance, the management of systemic risk, and theoperation of the international financial system. The re-mainder of this section examines several features ofinternational financial markets that provide an under-standing of why there was a reassessment of risks andrebalancing of mature market portfolios in the periodfrom mid-August through mid-October 1998. Thesefeatures include the following: (1) the unilateral debtrestructuring in Russia challenged underlying assump-tions of investors in mature markets about sovereignrisk and potential international financial support;(2) mature markets financed a significant share ofemerging market exposures; (3) a large number of di-verse financial institutions, not just hedge funds, had

Systemic Aspects of Mature Market Turbulence

47

16The turbulence was also affected by a number of features of thestructural transformations that have occurred in financial marketsduring the past 10–15 years, some of which contributed to, andmagnified the effects of, the turbulence, and some of which moder-ated the turbulence and its impact. These features include the ex-panded opportunities for unbundling and repackaging componentsof financial risk, and advances in trading and portfolio managementtechniques (stop-loss orders, portfolio insurance, dynamic hedging)afforded by advances in information and computer technologies;the evolution of commercial and investment banks into financialconglomerates with global reach; and the growing importance of in-stitutional investors (insurance companies, pension funds, hedgefunds). For analyses of some of these changes see “Globalization ofFinance and Financial Risk,” Annex 5 of the September 1998International Capital Markets report.


similar risk exposures and became vulnerable to a con-tinued widening of interest rate spreads; (4) risk man-agement models did not prevent the buildup, and mod-

ern portfolio management exacerbated the unwinding,of the preponderance of credit risk convergence plays;and (5) a disorderly unwinding and deleveraging, if it

48

Since its peak in July 1998, the U.S. stock market hasexperienced a significant correction and then a rebound.The Standard & Poor’s 500 index fell some 19 percentfrom peak (July 17) to trough (October 8). The S&P 500index has since rebounded by about 24 percent (havingreached a new high on November 27). The correctionwas triggered in part by reports of lower corporate earn-ings and reevaluations of earnings expectations. At thesame time, long-term bond yields declined markedly.A key question is: At present levels of interest rates,what is the implied growth rate of future nominal earn-ings that would justify equity prices in early Decemberas the present value of discounted future earnings? Theanswer provided by the analysis below is that, assum-ing an unchanged risk premium, the implied nominalearnings growth rate (7!/2 percent) is about the same as inlate 1997, when the S&P 500 index was significantlylower than in early December 1998. The decline in long-term interest rates thus makes the 20!/2 percent higherS&P 500 consistent with unchanged expected earningsgrowth. The analysis supports the view that current eq-uity valuations are unsustainably high, especially in lightof the relatively late stage of the U.S. business cycle, byshowing that the implied earnings growth rate adjustedfor inflationary expectations is at a post–World War IIpeak.

Calculation of Implied Earnings Growth Rates

The relationship between equity prices and the im-plied growth of future nominal earnings can be derivedfrom the hypothesis that the current equity price, Pt, isequal to the discounted present value of future earnings,Et+j (j ≥ 1), with discount factor ρt.

1 Pt = ∑∞

j=1 (–––––) j E t+j. (1)

1 + ρt

Assuming that future earnings grow at a constant rate, gt,such that Et+j+1 = (1 + gt)Et+j, equation (1) becomes:

1 + gtPt = Et ∑∞

j=1 (–––––) j. (2)

1 + ρt

This implies the following relationship between the cur-rent price earnings ratio, (Pt/Et), the discount factor, andthe future earnings growth rate:

Pt 1 + gt–– = (–––––) . (3)Et ρt – gt

Given a suitable discount factor, equation (3) can besolved for the implied earnings growth rate, gt. Thediscount factor, ρt, is equal to (rt + e), where rt denotesthe 30-year U.S. treasury bond yield and e is a con-

stant equity risk premium (assumed to be 6 percentagepoints).1

Assessment

The S&P 500 price index on December 2, 1998, al-though 20!/2 percent higher than at the beginning of theyear, is underpinned by about the same earnings expecta-tions as at the end of 1997 (see figure). The drop in long-term interest rates (by about 90 basis points) can, accord-ing to this approach, almost fully explain the gains in theS&P 500. Had interest rates not fallen but remained at theirend-1997 levels, the implied earnings growth rate as ofearly December would be consistent with the S&P 500being 235 points (20!/4 percent) lower. Alternatively, tosupport current equity prices at end-1997 interest rates, theimplied earnings growth would need to be about 90 basispoints higher indefinitely into the future (see figure).2

From a historical perspective, implied earnings growthrates3—at about the same level as in the 1980s—can beconsidered high (see the bottom panel of the figure). Theimplied earnings growth rates, gt, may not provide the bestcomparison of equity valuations over a long time horizonbecause they depend in part on the discount factor, ρt, andthus on the bond yield, rt , (see equation (3)). Bond yieldsare influenced by many factors that may distort intertem-poral comparisons and are sensitive to market participants’inflationary expectations. A measure of implied earningsgrowth independent of expected inflation is the ratio of the(gross) nominal earnings growth rate to the (gross) nominalbond yield, which can be approximated by the spread be-tween the earnings growth rate and the bond yield. Thisspread was at an all-time high in November 1998 (see bot-tom panel of the figure), raising questions about the sus-tainability of current high equity valuations.

Box 3.2. What Is the Implied Future Earnings Growth Rate that Would Justify Current Equity Prices in the United States?

1R. Mehra and E. C. Prescott, “The Equity Premium: APuzzle,” Journal of Monetary Economics, Vol. 15, (1985), pp.145–61, and J. Y. Campbell, A. W. Lo, and A. C. MacKinlay, TheEconometrics of Financial Markets (Princeton, New Jersey:Princeton University Press, 1997), find a risk premium of about6 percentage points. An 8 percent risk premium (instead of 6 per-cent) would shift the level of the earnings growth path plotted inthe figure up by approximately 2 percentage points. Thus, thechoice of the risk premium would not affect the comparisons ofearnings growth between end-1997 and November 1998 so longas the risk premium did not change over this interval of time.

2Consider the following scenario that telescopes the neededchanges in earnings into 1999. Assume that in 2000 and there-after the paths for interest rates and earnings revert back to thepaths implied at the end of 1997; in this case earnings wouldneed to be about 5!/2 times larger in 1999 than in 1998 in orderto justify the S&P 500 level in early December.

3Owing to data limitations, the historical implied earningsgrowth rates for 1954–98 were derived based on the 10-yearU.S. treasury bond yield, rather than the 30-year bond yield.

had been allowed to continue to build momentum,would have posed systemic risks in international fi-nancial markets.

Although these features provide some understand-ing of why the mature market adjustments occurred,taken together they do not provide an understanding ofwhy the recent mature market turbulence was so ex-treme. The turbulence was extreme enough that it wasjudged to have posed threats to internationally activefinancial institutions, and potential systemic problemsin the mature and international financial markets. Theapparent under-estimation of the extent to which thesevulnerabilities and risks had accumulated raises anumber of systemic concerns, which are raised brieflyin the third and final subsection.

Why Did the Russian Crisis Create More MatureMarket Turbulence than the Asian Crises?

The depth and scale of the recent financial marketturbulence certainly cannot be explained by the po-tential direct impact of the unilateral debt restructur-ing in Russia. First, the value that could be lost froman outright default was relatively small, and only one-third of it was held by nonresident investors. Second,interest rates on Russian GKOs in the period leadingup to the unilateral restructuring were high relative tothe cost of borrowing by other emerging market sov-ereign credits. This suggests that investors wereaware of the risks of lending to Russia, as reflected inthe significantly higher returns that were necessary tocompensate for the higher perceived risks. Indeed, fi-nancial markets should have realized that the eco-nomic and financial problems faced by Russia weresignificantly more protracted than in many otheremerging markets. Finally, Russia was perceived asunique among emerging markets in one significant re-gard: long-standing political and foreign policy con-siderations implied to many investors that Russiamight continue to receive the funds it required fromthe international community to finance the requiredadjustments. In effect, Russia was perceived as “toobig to fail.” Overall, these factors suggest that it is notobvious why the Russian unilateral restructuringwould trigger a wholesale exit from emerging mar-kets and a period of unprecedented turbulence in ma-ture and international markets.

Nevertheless, for market participants the Russianunilateral restructuring seems to have been a definingevent. The restructuring appears to have challengedfundamental assumptions about emerging market fi-nance—widely held by the major financial institutionsand priced into all but the safest investments—perhapsincluding a presumption that countries would not uni-laterally restructure sovereign debt obligations. In theevent, and regardless of the subjective reasons, theRussian crisis drove risk managers and investors toquestion the validity of their assumptions and the bal-ance of financial risks in their international portfoliosmade up of investments in both emerging and maturemarkets.


49

United States: Earnings Growthand Share Prices

Sources: Bloomberg Financial Markets, LP; HaverAnalytics; and IMF staff calculations.

9.0

8.5

8.0

7.5

7.0

6.5

1250

1200

1150

1100

1050

1000

950

900

168

4

0

–4

–8

–12

12

8

4

0

–4

6.0Dec.1997

Dec. 2Feb. Apr. Aug. Oct.Jun.1998

Dec.1997

Dec. 2Feb. Apr. Aug. Oct.Jun.1998

Implied earnings growth based onend-1997 bond yield

Implied earnings growth

Spread betweenexpected earnings

growth andten-year treasury bonds

(left scale)

Implied expectedearnings growth

(right scale)

Implied Earnings Growth(percent)

S&P 500 Price Index

1954 59 64 69 74 79 84 89 94 Nov.98

Implied Expected Earnings Growth and Spreadwith Ten-Year Treasury Bonds(percent)


Accordingly, risk reassessments implied that manyportfolios might be riskier than risk managementmodels had previously indicated, and that even ma-ture market portfolios with even relatively low-riskcombinations of emerging and mature market invest-ments might be riskier than perceived before August17. In addition to affecting emerging markets, devel-opments in mid-to-late August, including the unwind-ing of mature market financing, led to a furtherwidening of interest rate spreads between relativelylow-quality advanced country fixed-income securities(such as asset-backed securities and low-grade corpo-rate bonds) and very high-quality government (U.S.and German) debt securities, as the general flight toquality proceeded.

Unlike the Asian crises, the Russian financial prob-lems triggered mature market turbulence because ofdifferences in the nature of the shock. The Russiancrisis was a unilateral restructuring of sovereigndebt—a traded financial market instrument—that in amark-to-market environment would immediately trig-ger the unwinding of leveraged positions by large, in-ternationally active, financial institutions. In theAsian crisis, by contrast, the bulk of the financial con-tracts that immediately became at risk consisted of(nontradable) interbank loans. In addition, Russia’sunilateral restructuring was a sudden and definingevent, whereas the Asian crises developed moreslowly in several stages. But the Asian crises, ofcourse, contributed to the recent turbulence by al-ready reducing appetites for risk. Another differenceis that the Russian crisis occurred amid greater con-cerns about the health of the U.S. economy and thesustainability of the valuation of U.S., and other ma-ture, equity markets.

Ultimately, the Russian unilateral debt restructuringtriggered an abrupt, post-Asian-crisis flight from awide range of emerging financial markets, a sharpwidening of emerging market interest rate spreads to1,700 basis points, and a drying up of liquidity in in-ternational capital markets. The related flights to qual-ity and liquidity in international capital markets set offa process of deleveraging of financial transactions andvirulent turbulence in mature financial markets thatrapidly and sharply affected many investors and awide range of mature financial markets.

Mature Markets Financed and Leveraged Emerging Market Investments

At least some of the immediate impact on maturemarkets of the unilateral debt restructuring in Russiareflected the fact that a significant share of financ-ing for Russian and other emerging market invest-ments had been arranged and leveraged in the maturemarkets, in particular U.S. financial markets. Forexample, some investors had purchased RussianGKOs, on margin, through investment banks that had

funded the purchases with short-term repurchaseagreements and commercial paper in U.S. markets.Other Russian and emerging market securities pur-chases had been funded in Japan and swapped intolocal currencies or dollars. Accordingly, the initialunwinding of financing for emerging market posi-tions, hedges, and leverage meant that mature marketpositions related to these investments also had to beunwound or hedged, because the Russian restruc-turing triggered margin calls and led to a widespreadincrease in margin requirements. Because many ofthe investments that needed to be unwound werehighly leveraged, the downward price adjustmentswere unusually sharp in a wide range of markets. Theleveraging of investments magnifies returns whenasset prices are appreciating, but it also magnifieslosses, and requires the expenditure of scarce capitalto meet margin calls, when adverse price movementsoccur, thereby forcing market participants to liquidatepositions as rapidly as possible. Thus, the simultane-ous presence of a high degree of leverage in a widerange of interconnected markets forced a large num-ber of investors simultaneously to sell assets into de-clining markets. This contributed to the rapid speedand heightened intensity of the downward price pres-sures and adjustments in September and October1998 (Box 3.3).

The widening of spreads and liquidity pressuresthat immediately followed the Russian crisis and theflight from emerging markets destroyed value infixed-income positions that had been predicated onthe perception that “credit risk” spreads between low-quality (mortgage-backed securities and corporatedebt, for example) and high-quality (sovereign) bor-rowers in the advanced countries had widened beyondsustainable levels because of the “Asian contagion.”These so-called credit-risk “convergence plays” (notto be confused with EMU currency and interest rateconvergence plays) were widely held. They weremade up of financial positions in advanced countryfixed-income markets in which the investor wouldsimultaneously be “long” in relatively high-risk debtsecurities (such as U.S., German, and Danish asset-backed and corporate securities)—expecting theirvalue to appreciate—and “short” in sovereign debtinstruments of similar currency denomination andmaturity (such as U.S., German, and Danish govern-ment bonds)—expecting their value to depreciate.The plays were calculated gambles that, once Asiancontagion dissipated, credit spreads would narrow tomore normal (historically consistent) levels andwould be associated with the expected price move-ments. Instead of narrowing, however, credit riskspreads widened further. Depending on how the trans-actions were financed, these adverse price movementswere associated with further margin calls, liquida-tions, and hedging, leading to further significantdemands on the shrinking pool of liquidity. This led

50

to even wider spreads in some markets as more posi-tions were liquidated to make margin calls in othermarkets.

This adjustment process ultimately posed potentialsystemic risks because of its impact on market liquid-ity and dynamics. Market liquidity dried up temporar-ily even in the deepest and most liquid markets as risks were repriced and positions deleveraged. Anecdotalevidence suggests that this occurred in the U.S. trea-sury securities markets (in both August and Septem-ber), the U.S. repo market (in October), and the yen/dollar market (during October 7–9). The drying up ofliquidity had a visible impact on prices and flows, andthere were repeated instances when concerns about liq-uidity were heightened, with markets becoming one-sided until prices declined enough to bring buyers backinto the fray. There also appears to have been unusual

concern about counterparty risk in collateralized mar-kets, although this can be explained by the uncertaintysurrounding the turbulence, most notably around thetime of the near collapse of LTCM and in the earlyweeks of the unwinding of LTCM’s positions. It wasnot clear how far asset prices needed to fall, or howbadly counterparties’ balance sheets had been damagedby the turbulence.

The potential for these dynamics to have systemicconsequences was especially visible in U.S. dollarmarkets, the center of gravity for much of the port-folio rebalancing and deleveraging. Moreover, thedollar is the principal financial vehicle for inter-national financial transactions and for managing li-quidity, engineering leverage, and speculation. Asnoted in previous International Capital Markets re-ports, financial institutions and markets in the United


51

Leverage is the magnification of the rate of return (pos-itive or negative) on a position or investment beyond therate obtained by a direct investment of own funds in thecash market. It is defined as the ratio of assets to equity.Leverage is achieved by increasing the investment througheither outright borrowing or derivative instruments. In theformer case, a loan (including repurchase agreements) isused to supplement the equity investment, which is ex-pected to have a rate of return higher than the interest rateon the loan. Instead of cash, the loan could consist of a se-curity (as in short-selling operations). In the latter case, de-rivative positions (such as futures and options) allow theinvestor to earn the return on the notional amount underly-ing the contract by committing a small portion of equity inthe form of initial margin or option premium payments.1To measure precisely the use of leverage by a firm, oneneeds to know all of the firm’s positions. This is frequentlynot possible because activities such as repurchase agree-ments and derivatives take place off the balance sheet andare therefore not observable to an outsider.

Leverage is of concern because of two effects. By defin-ition it creates and enhances the risk of default by marketparticipants; furthermore, rapid deleveraging—the unwind-ing of leveraged positions—can cause major disruptionsin financial markets by exaggerating market movements.2

If the rate of return on an investment to which borrowedfunds have been committed turns out to be less than ex-pected, the investor’s equity may very quickly diminishand become insufficient to cover the loans. In response toan adverse price movement, a leveraged position will beclosed faster by an investor (with a given loss tolerance)than if it were not leveraged. The larger the leverage, thesmaller is the needed price change to trigger an unwindingof the position. The need to quickly unwind large positionsin response to margin calls following exogenous pricemovements can magnify these movements in a destabiliz-ing manner. That is, a “long” leveraged position will besold as a result of an exogenous price decline, thus con-tributing to the price movement even further. Conversely,a “short” position needs to be covered in a rising market bybuying the security, therefore contributing to upward pricepressure. While any (unleveraged) position would requiresimilar actions, leveraged positions may increase volatilitymore rapidly.

If there are many similar leveraged positions, if there isa single large position, or if the underlying market is notvery liquid, rapid deleveraging can create price discon-nects (large price moves resulting from temporarily one-sided markets). These price movements in a mark-to-mar-ket environment will trigger margin calls or cause otherinvestors to reevaluate their positions. This, in turn, willforce the liquidation of more leveraged positions, result-ing in a knock-on effect, which can send ripples throughdiverse financial markets spawned by leveraged positions.

Some institutional investors, such as hedge funds, em-ploy leverage in various ways, aimed at designing strate-gies to bet on developments in many different markets bycommitting as little of their own equity as possible. Thefollowing hypothetical scenario illustrates how an institu-tion can lever up its equity several times. It also shows howseemingly exogenous events can cause this strategy to un-ravel, magnifying the turbulence in financial markets (seefigure). The example is inspired by, but does not necessar-

Box 3.3. Leverage

1The degree of leverage for a futures contract can be definedas the ratio of the notional value (assets) to the margin posted(equity). The degree of leverage for an option contract can be de-fined as the delta times the underlying price, times the notionalvalue, divided by the option premium.

2Leverage also has benefits not only to participants but to thesystem as a whole. It can be usefully employed to hedge an ex-isting commitment in a cost-saving manner. It also facilitatesspeculation, which is necessary for the efficient functioning ofmarkets and enhances liquidity. Furthermore, commercial banksare by their nature highly leveraged without necessarily build-ing up leveraged positions as described in this box.

(Box continues on next page.)


States have played the critical role of international fi-nancial intermediary. Some have interpreted this roleas bestowing benefit on the U.S. economy: develop-ments between August and October suggest that thereare also significant costs and risks. However, al-though U.S. markets were most visibly affected by theturbulence, other mature financial systems were alsoat risk if the disorderly unwinding had continued toescalate.

The resulting dynamics experienced in the maturemarkets—in particular, dollar financial markets andyen/dollar currency markets—reflected the very highdegree of leverage that accumulated in these marketsthrough the late summer 1998. This high degree ofleverage itself reflected the relatively low margin re-quirements on over-the-counter derivative transac-tions and the increasingly accepted practice of very

low, or zero, “haircuts” on repo transactions. Theseand other features of modern finance may have in-creased the vulnerability of the involved markets, andthe investors trading in them, to the kind of sellingpressures, liquidity needs, rapid price adjustments,and illiquidity that ultimately occurred. These fea-tures and market dynamics ultimately exposed the in-ternational financial system to unexpected and un-warranted risks.

A Large Number of Diverse Financial Institutions,Not Just Hedge Funds, Drove the Turbulence: AnAsian Lesson Revisited

Partly because of the near collapse of LTCM andthe publicity it attracted, there has been a tendency toexaggerate the role that hedge funds played in the ma-

52

Box 3.3 (concluded)

ily accurately reflect, positions alleged to have been takenby Long-Term Capital Management (LTCM). It is by nomeans unique; similar leveraged positions might also betaken by many hedge funds and investment banks alike.

The first layer involves an outright loan, using a smallamount of equity to secure a yen-denominated loan inorder to take advantage of the interest differential be-tween Japan and the United States. The proceeds fromthe loan are exchanged into U.S. dollars and used as col-lateral to short-sell on-the-run government bonds.3 The

proceeds from this sale finance the purchase of off-the-run government bonds in the expectation that the yieldspread between the two bond types would narrow. In thethird layer of leverage, the fund would then use its longposition in off-the-run government bonds as collateral toborrow funds under a repurchase agreement.4 The pro-

4The lender of cash in a repo may also demand a “haircut” (mar-gin payment) to limit his credit exposure resulting from a declinein the price of the collateral. This margin payment would reduceleverage. While stock margins are 50 percent and exchange-tradedfutures margins are between 2 and 8 percent, haircuts on repos arebetween 1 and 2 percent. Hedge funds are often able to negotiatea zero margin. Without any cushion to accommodate fluctuations,a 4 percent price movement (see the text) on a few trillion dollarsof assets serving as collateral in a repo would cause massive mar-gin calls and result in a major market movement.

3On-the-run securities are the latest issue of a particular matu-rity. Usually they are the most actively traded issues for a particu-lar maturity. Off-the-run securities are the previous issues of thesame maturity. For example, in October 1998 the on-the-run 30-year treasury bond matures in August 2028; the most recent off-the-run 30-year treasury bond matures in November 2027.

Hypothetical Example of Leverage

$1,000 notional value of call option on equity in firms targeted for takeover

$125 of U.S. investment bank floating-rate notes (FRN)

$25 worth of off-the-run bonds

Cash $5

$1 equity base

Repo FRN into cash,(repo is initially

use cash to pay option premiumcollateralized) Borrow $100 for

margin purchase(secured by $25 equityin FRNs)

Borrow on-the-runbonds worth $20 (securedby $5 equity in bonds)

¥400 Japanesebank loan

Sell (short)

Exchange into $4

Repo off-the-run bond(repo is initially

into cash and post margincollateralized)

ture market turbulence. Within the large universe of3,000 hedge funds, LTCM was a unique institution(Box 3.4). Like other bond arbitrage operations, itsstrategy was to profit from small price discrepanciesin the safest securities markets in the world, a strategypursued by other hedge funds and other financial in-stitutions. LTCM was unique, however, in the waythat it tried to magnify the value of seemingly low-risk, low-profit gambles by taking very high-volumeand highly leveraged positions to a greater extent thanmost other hedge funds. It was viewed in the marketsas having a large appetite for risk. Single LTCMtrades would involve large positions (for example,one trade of $10 million worth of U.S. treasury secu-rities) financed on 10 percent (or smaller) margin, inorder to profit from interest rate spread changes of 1or 2 basis points. By employing this strategy—high

leveraging, with supposedly low-risk and low-returninvestments—a small profit from each trade poten-tially would be magnified, and the same trade wouldbe repeated over and over again. Few hedge fundsemployed the scale of leverage that LTCM assem-bled.17 In the end, both the leverage and the positionsultimately destroyed LTCM.

Although LTCM was the best-known loser from thekind of low-profit-margin, fixed-income position tak-ing that became exposed to a widening of credit riskand liquidity spreads, other, much larger institutionstook similar positions, in some cases with consider-


53

17See Barry Eichengreen, Donald Mathieson, and others, HedgeFunds and Financial Market Dynamics, Occasional Paper 166(Washington: IMF, 1998).

ceeds of the repo could be invested in floating-rate notes(FRNs) issued by U.S. investment banks, earning ahigher return than would be paid under the repo. The in-vestor could lend these FRN securities back to the in-vestment bank from which it had bought them throughanother repo agreement. As is common under repoagreements, the investor would continue to earn thefloating-rate coupon on the FRN, which by assumptionis higher than the rate the investor has to pay for therepo. Furthermore, the last repo frees up cash that can inturn be used for another investment, facilitating the in-crease in leverage through a derivative instrument. Forexample, the fund could buy a call option on equity offirms targeted for a takeover, which represents the fourthlayer of leverage.5

At each stage of this strategy the assets of the insti-tution are increased without committing further equity,leveraging up its equity base through a series of in-vestments on margins, short sales, repurchase agree-ments, and derivative securities. It allows the firm to beton yen depreciation, narrowing U.S. treasury yieldspreads, rising U.S. investment bank FRN prices, and ris-ing equity prices of takeover targets. Furthermore, someof these activities are not recorded on the firm’s balancesheet. The initial loans and U.S. treasury positions will bebooked on the balance sheet. However, the repos and thederivative transaction will facilitate off-balance-sheetactivity.

The investment strategy illustrated in this examplewould have unraveled after the Russian debt restructur-ing when the flight to quality and liquidity widened theyield spread between on-the-run and off-the-run U.S.treasury bonds, and the turbulence and losses dampenedprospects for the U.S. investment banking industry. Thewidening of the liquidity spread not only implied a losson the second leg of the transaction, owing to relativeprice movements, but also triggered a margin call on thefirst repo as the value of off-the-run bonds serving as col-lateral was reduced. This meant the fund would not onlyhave to buy on-the-run bonds when their price was risingto cover its short position, but also to sell off-the-runbonds in a falling market to meet the margin call.Similarly, the FRN securities in the second repurchaseagreement dropped in price and would therefore also trig-ger a margin call. To raise the cash needed to meet themargin calls, the investor would likely sell the FRNs inthe rapidly declining market. As rumors spread that a fi-nancial institution was in a liquidity crisis, counterpartieswould raise the “maintenance” margin to the level of theinitial margin to ensure that the loss in value of collateralwould not expose counterparties to credit risk. This inturn would accelerate the unwinding of leveraged posi-tions, causing even sharper price movements. The FRNmarket seized up completely at the beginning of October.Similarly, the bond market experienced significant turbu-lence in the period of October 7–9, around the time thatmuch of this deleveraging may have been going on.Furthermore, the continuous appreciation of the yen priorto October would have squeezed the fund on the first legof the transaction, triggering a rush into yen to repay theinitial loan, which is consistent with the observed sharpyen appreciation in the first week of October. This illus-trative scenario shows how leveraged positions may haveamplified the exogenous price movement triggered bythe Russian crisis and spread the turbulance to othermarkets.

5Balance sheet leverage is limited by two factors: underlyingequity and requirements to hold capital against the assets createdfrom the equity, which limits the number of times equity can belent out. Leverage accumulated through off-balance-sheet deriv-ative contracts is limited by the amount of margin paymentscounterparties require. If there is no margin payment, leveragecan be unlimited.


able leverage.18 These institutions included interna-tionally active commercial and investment banks, bro-

kers and dealers, and other institutional investors.While LTCM was alleged to have had $80 billion dol-

54

Long-Term Capital Management (LTCM) manages aninvestment fund1—a hedge fund—that attempts to profitfrom (often small) discrepancies in the relative value ofgovernment bonds, fixed-income derivatives, equitiesand equity derivatives primarily in the U.S., Japanese,and European markets. The fund also invests in a fewmarkets outside the G-7 countries. LTCM’s traders arelegendary for being the best and brightest technicians inthe hedge fund community, and the firm is reported tohave recorded total returns, after fees, of 43 percent in1995, 41 percent in 1996, and 17 percent in 1997.2

In the week of September 21—amid market rumorsabout LTCM and some of its major creditors and coun-terparties, and concerns over potential liquidity problemsin financial markets—the Federal Reserve Bank of NewYork (FRBNY) helped to organize and coordinate a $3.6billion private rescue of LTCM by a consortium of 14major international financial institutions. All of these in-stitutions are either counterparties, creditors, or investorsof the hedge fund. According to press reports andFRBNY press statements at the time, the rescue was seenas necessary for two reasons: LTCM’s financial conditionhad deteriorated to the point where it might not be able tomake either loan repayments or margin calls on its highlyleveraged positions in U.S., Japanese, and Europeanbond markets, and might require either recapitalization orliquidation; and immediate closure of LTCM would haveworsened the financial condition of some already weak-ened international financial institutions and could havetriggered a massive simultaneous sale of LTCM’s collat-eral (securities) by creditor institutions. This would havefurther strained the stability of the world’s major bondmarkets, creating the potential for debilitating and wide-spread spillovers and contagion, including in emergingmarkets.

What Brought LTCM to the Brink of Collapse?

LTCM specializes in fixed-income and equity conver-gence strategies, taking complex and leveraged positionsin order to profit from (often small) discrepancies in therelative price of bonds, swaps, options, and similarly inthe relative price of equities and their derivative instru-ments. The bulk of LTCM’s investments are convergencetrades in U.S., Japanese, and European bond markets—essentially gambles that interest rate spreads had widenedbeyond a sustainable level and would narrow and return

to more normal spreads. These transactions typicallyhave horizons of 6 to 24 months. These so-called conver-gence trades are based on the judgment, probably sup-ported by asset-pricing models, that markets have beenundervaluing relatively risky bonds (mortgage-backedsecurities, for example) and overvaluing low-risk bonds(G-7 government securities).3 LTCM purchased or bor-rowed large volumes of relatively risky bonds and soldshort G-7 government bonds. These positions were lever-aged using borrowed funds from internationally activecommercial and investment banks. At the beginning of1998, on capital of just $4.8 billion, LTCM managed bal-ance sheet positions totaling about $120 billion, implyinga leverage ratio of 25-times-capital. At the same time,LTCM was managing total gross notional off-balance-sheet derivative contracts amounting to about $1.3 tril-lion. These leverage ratios and off-balance-sheet transac-tions did not change significantly in the months leadingup to the August 1998 crisis.

LTCM’s trading book was “long” in relatively illiquid,low-quality securities (mortgage-backed securities, ad-vanced country junk bonds) and “short” in liquid, high-quality securities (U.S. treasuries and other G-7 sover-eign credits). It was also short volatility in the mainequity markets. Contrary to LTCM’s judgment, interest-rate spreads widened throughout most of 1998 in most ofthe theaters of its operations—earlier in the year, as theintensification of the financial crises in Asia encourageda flight to quality in the G-7 government securities mar-kets and, later in the year, as the Russian involuntary re-structuring of GKOs (ruble-denominated discount in-struments) led to an even greater widening of creditspreads.4 Similarly, implied volatilities reached all-time

Box 3.4. The Near Collapse and Rescue of Long-Term Capital Management

1The fund is Long-Term Capital Portfolio.2By comparison, the return to the S&P 500 was 34 percent in

1995, 20 percent in 1996, and 34 percent in 1997.

3Bankers familiar with LTCM’s portfolio suggested thatLTCM engaged in a lot of transactions involving total returnswaps, which allow investors to profit or to lose from pricemovements on securities without actually purchasing them. Forexample, for an obligatory financing charge, LTCM would bor-row asset-backed securities and receive both interest and capitalgains or losses. The securities would have to be returned and thefinance charge paid regardless of the value of the securities.LTCM’s expectation was that they would lock in relatively highinterest payments and at the same time receive capital gains asthe value of these securities rose as the markets’ assessment re-turned to normal.

4In 1997, LTCM is reported to have borrowed aggressively,and on relatively favorable terms, to increase its exposure andleverage as spreads widened, presumably on the strong belief inits technical judgment that spreads would ultimately narrow andthat it would be able to manage its growing exposure.

18According to an analysis by Salomon Smith Barney, balancesheet data indicate that investment banks are highly leveraged insti-tutions. Gross leverage ratios (ratio of gross assets to equity) for the

top firms range between 25 and 35, while net leverage ratios (ratio ofgross assets excluding matched-book financing to equity) range be-tween 10 and 20. These ratios exclude off-balance-sheet activities.

lars in balance-sheet arbitrage positions in U.S. trea-sury security markets, commercial banks alone wereestimated to have had $3 trillion in similar exposures.The widespread position taking, the complexity of thelayers of derivative and leveraged spot market trans-actions, and the relatively closed circle of counterpar-

ties created a potentially unsustainable balance anddistribution of financial risks. The confluence of theseinstitutions’ positions led to a situation in which arapid unwinding of LTCM’s portfolios in fixed-income and equity markets might have meant not onlythat direct creditors and counterparts could have had


55

highs. During these flights to quality, liquidity dried upin the high-yield (high-risk) sectors of U.S., Japanese,and European bond markets, driving spreads evenhigher. Because of the illiquidity of its positions, and dif-ficult market conditions, LTCM was unable to reduce thesize of its positions and strategies. The losses associatedwith this divergence of spreads and increased volatilitiesreduced the hedge fund’s equity (net asset value) from$4.8 billion in January 1998 to $2.3 billion in August.This resulted in an increase of leverage to 50-times-cap-ital on its balance sheet positions alone.

In a September 2 letter to investors, LTCM informedinvestors that the value of the fund was down 44 percentin August and 52 percent for the year. LTCM also re-ported that losses occurred in a wide variety of strategies,distributed approximately 82 percent to relative valuetrades and 18 percent to directional trades; only 16 per-cent of losses were attributed to emerging market invest-ments. On average, spreads continued to diverge inSeptember. Through Friday, September 18, LTCM wasmeeting margin calls, in part by drawing on a long-standing $580 million credit facility headed by ChaseManhattan Corporation.

After it was determined that LTCM’s financial condi-tion had deteriorated to the point where it might not beable to service its debt obligations and make margin callsduring the week, the FRBNY organized a meeting ofLTCM creditors and counterparties and began helping tocoordinate a private consortium to rescue LTCM. As ofTuesday, September 23, LTCM’s equity position stood atjust $600 million and was supporting balance sheet posi-tions in excess of $100 billion, implying balance sheetleverage of 167 times capital; the hedge fund’s losses onits highly leveraged positions (but not necessarily on thesecurities that it was holding) had wiped out 90 percentof its equity.

Why Was LTCM Rescued?

At some point—probably culminating on Sunday,September 20, when staff at the FRBNY and the U.S.Treasury visited LTCM headquarters in Greenwich,Connecticut—the assessment was made that the poten-tial for market disruption supported facilitating a pri-vate-sector solution, given the similarity of positiontaking by other large internationally active financialinstitutions and the large number of other institutions(including brokers and dealers) that held large inven-tories of securities across the credit spectrum. It wasknown by the banks and U.S. authorities that LTCM’sexposure was large enough, and cut across a suffi-

cient number of important markets, that if the fund wereforced into a sudden and disorderly liquidation, mar-kets around the globe could be disrupted as LTCM’silliquid securities were dumped at prices well belowface value. The further widening of spreads and dra-matic price dynamics that might have accompanied animmediate closing out and deleveraging of LTCM’s po-sitions would have entailed a simultaneous and massivetrading of a large volume of securities by the large inter-national financial institutions in New York, Japan, andEurope, posing risks of systemic proportions. Theprocess could have led to many technical insolvencies.Given the number of institutions involved in these trans-actions, even a small probability of this occurring posedthe serious systemic risk broadly encompassing financialmarkets.

The threat of a massive “fire sale” of LTCM’s collat-eral did exist. Repurchase and reverse repurchase agree-ments are governed by a provision of the bankruptcy lawthat would have allowed LTCM’s creditors to sell imme-diately the collateral that secured repos and swaps usedextensively by LTCM if it were allowed to fail. WithLTCM’s large balance sheet exposures and additionallarge off-balance-sheet positions, a bankruptcy filingcould have touched off a simultaneous and potentiallydestabilizing effort by all creditors to buy and sell the se-curities that were backing these huge repo and swap po-sitions, at a time when markets were already strained andjittery. In the days before the rescue was arranged, the $1trillion U.S. repo market was showing strains on marketrumors and concerns about the credit quality of someleading investment banks. A liquidation of collateralwould have had repercussions in the underlying repo andswap markets themselves—in New York, Japan, andEurope. There was growing evidence that liquidity inthese important funding markets was drying up, and themounting nervousness might have encouraged theFRBNY to arrange the rescue in order to avoid a panicand the potential for systemic problems. The private res-cue by creditors was orchestrated, in part, to allow for amore orderly unwinding of positions and to remove thepotential for a rapid draining of liquidity in the world’smajor securities markets, and the systemic risks thatsuch an event might entail.

Most if not all of the credit supplied to LTCM bythe major financial institutions was fully collateralizedwith high-grade paper, in most cases U.S. treasury secu-rities (probably some of it off-the-run issues). Never-

(Box continues on next page.)


difficulties collecting on their contracts, but also thatthe wide array of other institutions holding similar po-sitions could have been adversely affected by sharpprice movements.

Some of the proprietary trading operations of com-mercial and investment banks rival in scale the opera-

tions of large hedge funds and at times may have morerisky proprietary trading books than many hedgefunds. In addition to taking positions on their own ac-count, the globally active commercial and investmentbanks may encourage clients to take similar positions,and so the volumes of trades and capital these institu-

56

theless, potential losses on bank loans to LTCM mighthave resulted from adverse movements in the marketvalues of this collateral were LTCM to default. Therewas also concern that many of the world’s largest inter-nationally active institutions stood to lose from a mas-sive sell-off of positions and the liquidity problems itmight create. But the predominant consideration seemsto have been that a very large number and variety offinancial institutions could have been affected by sucha wave of selling and repricing. This could have dis-rupted financial markets around the world. One way tocontain the potential systemic risk was to organize aprivate consortium of LTCM’s main creditors andcounterparties, in part to engineer a more orderlyprocess of unwinding and deleveraging, but also tointernalize many of the risks. The pooling and inter-nalization of risk was possible because the consor-tium included many, if not all, of the financial institu-tions that would necessarily have been involved in theclosing out and deleveraging of positions in the mar-kets, because they are the major market makers, dealers,brokers, and counterparties. An additional benefit mightbe that some share of the transactions could be nettedwithin the consortium, and others could be closed outand deleveraged within the consortium rather than inthe markets, which would help to reduce selling pres-sures in the major markets. Alternative solutions couldhave been found with similar benefits. For example, asingle buyer might have been found for LTCM, and therewas at least one such interested party. In the short timeavailable for finding solutions, however, such alternativeswere not agreed by LTCM’s partners and potential buyers.

The rescue of LTCM can be seen as an out-of-courtbankruptcy-type reorganization of LTCM in which itsmajor creditors have become its new owners in charge ofevery aspect of the business, with the objective of sal-vaging as much value from the wreckage as they can.With the benefit of hindsight, it is safe to conclude thatthe outright failure of LTCM would have posed signi-ficant risks of systemic problems in international finan-cial markets, and that it was necessary to restructureLTCM. A more rapid and disorderly unwinding ofLTCM’s very large and highly leveraged fixed-incomepositions and related positions of other institutions couldhave triggered an even more destructive forced delever-aging in U.S., German, and Japanese fixed-income mar-kets and in the major currency markets. This would havenecessarily included equally disruptive selling pressuresin the associated derivative markets, where the volumeand notional value of transactions are several multiples

of the volume and face value of the underlying securi-ties. One can only speculate how much worse the marketturbulence would have been had LTCM been allowed tocollapse.

Two counter arguments against the Federal ReserveSystem’s involvement in facilitating the private rescueof LTCM have been suggested. First, it has been argued,because LTCM was not subject to the Federal Reserve’ssupervisory and regulatory jurisdiction, it was inap-propriate for the Federal Reserve to risk its reputationand goodwill. Second, the involvement of the centralbank in facilitating the private rescue might entail moralhazard for institutions not ordinarily regulated or super-vised by the central bank, or for institutions that ordi-narily take on high leverage in their activities. Moralhazard clearly is a concern with any central bank or otherofficial interventions in individual financial institu-tions. Because no public funds were necessary in rescu-ing LTCM, the moral hazard implications of this par-ticular intervention would be limited to the signalsimplicit in the Federal Reserve’s involvement. It is notpossible to evaluate objectively the potential costs ofthese signals against the benefits of the Federal Reserve’sinvolvement.

Who Are LTCM’s New Owners?

A total of $3.625 billion was injected as equity intoLTCM by 14 international financial institutions, many ofthem creditors: 11 institutions took equity stakes of $300million (Banker’s Trust, Barclays, Chase Manhattan,Credit Suisse First Boston, Deutsche Bank, J.P. Morgan,Goldman Sachs, Merrill Lynch, Morgan Stanley DeanWitter, Travelers, and Union Bank of Switzerland); 2 in-stitutions took a $125 million stake (Société Générale,Lehman Brothers); and 1 institution took a stake of $100million (Paribas).5

The new owners together own 90 percent of LTCM’sequity for a period of three years (the remaining 10 per-cent is held by the original partners) and have the optionto obtain 50 percent of the management company for anominal fee of $1. Most new investors are reported to behoping to be bought out before then. The terms of theagreement provide the private consortium with full au-thority over the investment strategy, capitalization struc-ture, credit and market risk management, compensation,and all other significant decisions.

Box 3.4 (concluded)

5Among the new owners, Union Bank of Switzerland had al-ready announced the writing down of its original equity stake inLTCM of $685 million.

tions place are larger than their proprietary positions,and than the positions taken by hedge funds.19 Muchof this investment and trading activity takes place inthe relatively safe fixed-income markets in the ad-vanced countries.

Nevertheless, the hedge funds together have enoughcapital and are sufficiently highly leveraged to have anoticeable impact on market liquidity when they with-draw from markets during turbulence: they often arethe first to exit a market, but they also often are thefirst to reenter.20 However, it is doubtful that any sin-gle hedge fund, or small group of hedge funds, alonecould pose a risk of systemic problems. Rather it wasthe simultaneous and interrelated involvement of alarge number of players, including both hedge fundsand many internationally active commercial and in-vestment banks, that recently raised the potential forsuch risks.

Why Did Risk Management Models Not Preventthe Buildup of Positions and Leverage?

The ultimate impact of the turbulence on bank bal-ance sheets and profitability, and on the financial con-dition of other financiers and investors, raises ques-tions about why so many sophisticated institutions mayhave engaged in similar positions. A key issue is: Whydid risk management technologies (models, stress tests,and scenario analyses) and internal operational controlmechanisms apparently not warn risk managers andtop management about the growing vulnerabilities?

Modern risk management models are designed (andare used along with stress testing) to measure and as-sess the riskiness of portfolios on the basis of assump-tions about the likelihood of outcomes that might put afirm’s capital at risk and ultimately risk its solvency.Value-at-risk (VaR) models are one way of achievingthis objective.21 One problem with relying on thesetechniques is that they may provide a false sense ofprecision, in part because the output of models dependson inputs that depend on human judgment. Faulty as-sumptions—about the probability of adverse events,the structure of markets, cross-market price correla-tions, and within stress tests and scenario analyses—might have impaired the usefulness of models. Themodels also assume that market liquidity will be suffi-cient to allow positions to be closed out without ex-

tremely large price changes or market disorder.Moreover, because they usually rely on historical rela-tionships between price movements in many markets,the models can break down during times of unusualstress and turbulence, particularly when structuralbreaks occur in cross-market relationships. Such struc-tural shifts, even if temporary, may imply dramaticshifts in risk in portfolios, and call for a rapid portfoliorebalancing. Even asset managers who only employednormal principles of portfolio diversification mighthave called for reductions in emerging market and sim-ilar risk exposures and for increases in low-risk or risk-free assets (Box 3.5). If many investors adjust largeportfolios simultaneously in the same direction, marketliquidity would tend to be adversely affected.

High-tech computer-driven portfolio managementtechniques, such as portfolio insurance22 and dynamichedging,23 are supposed to have helped firms to mini-mize their losses and maximize their gains when thesetechniques worked, but they probably also exacerbatedlarge volumes of sell orders flowing into decliningmarkets where buyers and sellers were uncertain aboutwho owned what risks and how they should be priced.Risk components that in normal circumstances wouldhave remained isolated became blurred, and liquidityand counterparty risk considerations came to the foreas key concerns. At a time when there already mayhave been heightened concern about credit, market,and settlement risks, market participants appear to haveintensified their focus on whether securities could beliquidated quickly if necessary (liquidity risk) regard-less of the quality of the paper. It is this kind of blur-ring of risks, the intermittent, absolute focus on marketliquidity, and the almost automatic nature of some po-sition-unwinding and liquidity-seeking trades thatcould explain unusual developments in some marketsegments. One such example is the sharp rise in thepremium on off-the-run over on-the-run U.S. treasurysecurities, which increased to more than 35 basis pointsin at least one segment of the market (possibly whenLTCM was liquidating its long position in off-the-runU.S. treasuries).

Threat of Systemic Problems: AcceleratedDeleveraging and Widening of Spreads Surrounding the Near Collapse of LTCM

By late August 1998 the initial widening of creditrisk spreads likely contributed to pressures on the port-


57

19Hedge funds obtain most of their operational financing from,and place their trades with, the large commercial and investmentbanks. The banks are seen by hedge funds as “front-running” thehedge funds’ own position taking. Because of this front-running be-havior, many hedge funds see commercial and investment banks asa threat to their returns. This is one of the reasons why hedge fundsare so secretive about their positions and use complex tradingstrategies and tactics.

20See the 1997 International Capital Markets report.21VaR models measure how much of the firm’s capital could be

lost because of swings in the value of its portfolio.

22Portfolio insurance is a technique that changes a portfolio’smarket exposure systematically in reaction to prior market move-ments, with the objective of avoiding large losses and securing asmuch participation as possible in favorable market movements.

23Dynamic hedging is a position-risk management technique inwhich option-like return patterns are replicated by adjusting portfoliopositions to offset the impact of a price change in the underlying mar-ket on the value of an options position (the “delta”). Dynamic hedg-ing relies on liquid, continuous markets with low transaction costs.


folios of the major financial institutions and LTCM,and the rebalancing and deleveraging of mature mar-ket portfolios accelerated. Although the Russian uni-lateral restructuring might have been perceived as asignificant event for traders in New York, London, andFrankfurt, an important wake-up call for mature mar-ket position takers occurred in early September, whenLTCM announced that 52 percent of its capital hadbeen spent on margin calls, only 16 percent of whichwere related to emerging market investments (see Box3.4). This apparently triggered rounds of speculation,selling, and counterparty concerns in the internationalmarkets. At that time, LTCM also had engaged in fundraising to try to increase capital to sustain its conver-gence plays long enough for interest rate spreads toconverge.

Fears in the markets that many other institutionsmight be holding similar positions created substantialuncertainty about counterparty risk and generated ru-mors, which probably contributed to heightened mar-ket turbulence even after LTCM’s rescue by a privateconsortium of its creditors and counterparties had been

agreed. In the week beginning October 5, while theconsortium was beginning to unwind and sell offLTCM’s complex positions, there were several wavesof turbulence (especially during October 7–9). Theyen/dollar market, in particular, experienced extremeturbulence as the financing of global investments fromJapan was abruptly unwound and the dollar fellsharply against the yen. As discussed earlier (see Box3.1), this unprecedented yen/dollar adjustment waspartly the result of the unwinding of the “carry-trade,”in which investors around the world borrowed yencheaply, swapped into other currencies (probablymostly dollars), and then purchased assets with higherreturns in a wide range of countries and markets, in-cluding U.S. government securities. Much of the li-quidity-driven (and large volumes of) trading that ac-companied the further widening of spreads, thedramatic depreciation of the dollar, rumors of failingfinancial institutions, and other uncertainty-creatingevents was probably partly driven by the actual un-winding and deleveraging of positions by the consor-tium on behalf of LTCM and by consortium members.

58

Box 3.5. Risk Management: Progress and Problems

The most familiar risk management model is the value-at-risk (VaR) model.1 The VaR model measures howmuch of the firm’s capital could be lost owing to swingsin the value of its portfolio, given a host of assumptionsabout correlations among security prices, the way that se-curity prices move over time, and so forth. More techni-cally, the VaR model estimates the loss on the firm’s port-folio that should be exceeded with no more than a certainprobability, given a model for changes in the prices of allthe assets in the portfolio. For example, a firm’s VaRmight indicate that its losses over the coming weekshould exceed $10 million with no more than 5 percentprobability.2 Financial institutions base VaR calculationson historical data, and some also use stress tests, in whichscenarios are simulated. Regardless of the methodology,if the potential loss is too large, the firm might rebalance,or hedge, part of its portfolio to reduce the value at risk.Some financial institutions also allocate capital on thebasis of results of VaR and similar models.

Prevailing risk management practices focus on marketrisk, for which modeling is advanced and has reached a rea-sonable level of performance. Modeling of market risk en-tails formal models of the behavior of the prices of the as-sets making up the portfolio. A simple VaR model mightassume that changes in the value of each asset in the port-folio are normally distributed and independent across time,with correlations and variances that change slowly. Such

models would also tend to assume that the structure of mar-kets is relatively simple and stable; for example, that the re-lationship between stock and bond prices is simple and doesnot change sign or magnitude, or that the complex relation-ship of a derivative security to its underlying security can bewell approximated by a simpler relationship.

However, models of market risk have, over time, re-vealed weaknesses, owing primarily to their heavy relianceon historical data and relationships. Such analysis tends tounderstate the likelihood of extreme events and often in-volves the assumption that the processes generating marketprices are stable. Recent events have underscored that rare,“fat-tailed” events can occur more frequently than ex-pected, that correlations can increase and change sign sud-denly, and that volatility can increase sharply withoutwarning. Moreover, it is also presumed that market liquid-ity will be maintained so that positions can be closed outwhen necessary without adverse price declines.

Significant gaps also remain in the modeling of othertypes of risk, particularly credit and liquidity risks.3Credit risk modeling is still in its infancy, in part becausecredit risk is difficult to model: defaults are rare, and therisk of default evolves over time in a complicated fash-ion. However, the bulk of risk on balance sheets consistsof credit risk; indeed, credit losses have been the keysource of financial distress.

Even less advanced is the modeling of liquidity risk—the risk that transactions cannot be executed without

1For further discussion of recent advances in risk manage-ment, see the September 1998 International Capital Marketsreport, Annex V.

2Philippe Jorion, Value at Risk (New York: McGraw-Hill,1997) provides technical details of risk management models.

3Other types of risk—which for the most part remain to bequantitatively addressed—include legal and operational risks;see Jorion, Value at Risk.

Some trading was also likely to have been generatedby the uncertainty over whether the emerging marketcontagion would spread to Latin America, in particu-lar Brazil. It was not until after the U.S. FederalReserve’s second interest rate cut on October 15 thatmarket pressures began to ease.

It is uncertain how much more deleveraging willoccur in the period ahead. The manner in which mar-kets adjusted; the apparent price disconnects in usuallydeep and liquid, but temporarily thin, markets; thepanic-driven rumors; and other aspects of the turbu-lence strongly suggest that the unwinding that had tobe accomplished to reach more comfortable risk levelswas unusually large relative to underlying balancesheet positions. However, given the limited availabil-ity of transaction data, and the infrequency and in-completeness of the reporting of derivatives markettransactions, it is not possible to assess the degree ofleverage remaining in the system that might create fur-ther turbulent dynamics in the future. Considering thatthe process of deleveraging that followed the bondmarket turbulence in early 1994 took some eight

months to complete, and that the price and flow ad-justments at their nadir in the more recent episodewere in many respects unprecedented—in particularthe massive shifts out of relatively risky and illiquidassets—it would seem that the process of deleveragingmay still have some distance to go, although presum-ably not with the extreme tensions that developed inSeptember and early October.

While the preceding analysis provides some under-standing of why the Russian crisis and some featurespresent in mature markets might lead to a reassess-ment of risks and rebalancing of emerging and maturemarket portfolios, this understanding is not entirelyreassuring. In particular, the associated surprisinglylarge flights to safety and liquidity, the rapid dryingup of liquidity in international capital markets, andturbulence in a wide range of mature markets—de-fined at times by price disconnects and near seizuresin some markets—all appear to have been out of pro-portion to the factors that triggered them. Thus, theoverriding concern is not that the reassessments andportfolio adjustments occurred; instead, it is that they


59

unsettling markets. Liquidity risk may have been themost significant type of risk in the recent turbulence,since a lack of liquidity may have been responsiblefor some of the price disconnects that occurred.4Liquidity risk can make risk management extraordinar-ily difficult. For example, firms that take positions underthe assumption that the positions can be unwoundsmoothly if events turn against them sustain larger-than-expected losses when they attempt to unwind in illiquidmarkets.

There are likewise significant gaps in the modeling ofthe nexus of market, credit, and liquidity risks, gaps thatcame into sharp focus during the Asian crises and in theaftermath of the Russian devaluation and moratorium.5For example, in the recent turbulence, market risk itselfgave rise to credit and liquidity risks. In some cases,firms that had significant paper gains vis-à-vis emergingmarket counterparties were unable to collect these gainsas counterparties went bankrupt: while market risk hadmoved in banks’ favor, counterparty risk had moved justas strongly against them. Similar problems arose whenlarge market moves sharply reduced the value of collat-eral. In other instances, large market moves created sell-ing pressures, which in turn impaired market liquidityand gave rise to liquidity risk.

These problems—the unstable nature of market risk,the difficulty of modeling credit and liquidity risk, and thehighly complex relationship between them—also under-cut attempts to hedge. Banks that hedged the ruble expo-sure in GKOs (ruble-denominated discount instruments)through forward contracts with Russian banks facedlosses on both the GKOs and hedges. Also, during the re-cent market stresses, assets that usually were uncorrelatedor negatively correlated, and hence offered diversificationor hedging, declined together as liquidity evaporated.

Successful risk management requires human judgment,including a balanced (lack of) respect for statistical mod-els built on historical data. Another human factor is the in-teraction between the risk management and business func-tions in banks; sometimes, the dictates of risk models areoverridden by business units in pursuit of, for example, astrategic relationship.6 Indeed, within financial institu-tions there is a tension between taking risk during marketupturns and managing the risk from market downturns.

There are two lessons for risk management from themarket turbulence of the past summer and fall. First, riskmanagement should promote a conservative and compre-hensive approach to risk; a piecemeal approach clearlycan miss significant risks and the interactions amongthem. Second, in risk management as in other areas, com-plex, formal models can complement but cannot fullysubstitute for judgment and experience.

4There is also a different type of liquidity risk (the ability toroll over funding). See Jorion, Value at Risk.

5For an approach to modeling both market and credit risk, seeTheodore M. Barnhill, Jr., and William F. Maxwell, “ModelingCorrelated Interest Rate and Credit Risk” (unpublished;Washington: George Washington University and GeorgetownUniversity, October 1998).

6See Bank for International Settlements, “On the Use ofInformation and Risk Management by International Banks:Report of a Working Group Established by the Euro-CurrencyStanding Committee of the Central Banks of the Group of TenCountries” (Basle, October 1998).


were so violent and widespread that they might haveposed systemic risks for world financial markets andsignificant downside risks to the world economicoutlook.

Shortcomings in Risk Management and Implications for Prudential Regulation and Supervision

As has been suggested, deficiencies in both privateand systemic risk management probably contributed tothe recent financial market turbulence. In private mar-kets, a large number of diverse market participantswere apparently surprised by sharp adverse pricemovements in asset markets, and this suggests thatthey engaged in excessive risk taking, excessive lever-age, and ultimately an unsustainable structure of fi-nancial positions.24 In addition, they may have paid in-sufficient attention to the interplay of market andcredit risk. This confluence of mistakes might havecreated an accident waiting to happen. On the publicside, although public systemic risk management in theperiod September–November 1998 alleviated thethreat of a systemic problem in international markets,at least two lines of defense—banking supervision andmarket surveillance—that would ordinarily safeguardagainst the buildup of such a threat did not appear toprovide sufficient warnings.

Weaknesses in Private Risk Management

Financial systems have several lines of defenseagainst systemic problems. The first are the risk man-agement systems and internal management controlmechanisms of private financial institutions that are de-signed to prevent them from taking excessive risks thatcould ultimately threaten their capital position and via-bility. In view of the extent of losses suffered by a num-ber of large institutions, the degree of surprise associ-ated with those losses, and the reaction of equity pricesfor these institutions, risk management systems appearto have not worked very well for a large number of di-verse and systemically important financial institutions,including many of the internationally active commer-cial and investment banks, proprietary trading desks,market makers, broker/dealers, and foreign-exchangetraders/dealers. Bank loan books had already becomesomewhat weakened by the Asian crises and wereeroded further by the Russian crisis in August 1998.The profitability of trading books was also affected bythe emerging market turbulence through the Asian andRussian crises, which induced a shift to advanced

country fixed-income markets and the currency mar-kets, and most notably the yen/dollar market. Manyplayers appear to have been involved in convergenceplays, in part because yield curve plays were not lucra-tive given the compression of the term structure.

The evidence suggests that many market partici-pants did not adequately anticipate or understand therisks that were realized in the period mid-Augustthrough mid-October 1998. Several systemically im-portant internationally active financial institutions ap-pear to have made similar, if not the same, misjudg-ments in their risk assessments, risk management, andinvestment strategies. This suggests that managementcommand and control systems now used by these fi-nancial institutions may be flawed, and raises con-cerns about the adequacy of risk and portfolio man-agement systems and operational controls within someinternational financial institutions. The systems nowin use apparently have not adequately incorporatedsome of the lessons of the 1994–95 Mexican crisis—some mistakes appear to have been repeated in the1997 Asian crises—and some deeper-seated problemssurfaced in the recent management of advanced coun-try portfolios during the most recent turbulence.During the Mexican crisis, some investors made sig-nificant paper gains on their trading books by shortingthe peso with Mexican entities, but they could not col-lect their sizable gains because the Mexican counter-party did not survive the peso devaluation. In the mostrecent turbulence, similar mistakes appear to havebeen made. Internationally active banks extendedcredit to LTCM and also were counterparties to someof LTCM’s transactions without adequately under-standing the size, complexity, and riskiness of LTCM’sbalance sheet and off-balance-sheet positions. As isnow known, some creditors held many of the same po-sitions as LTCM. At least some of the turbulence in1998 probably could have been avoided if a more in-tegrated approach to market and credit risk manage-ment and position taking had been the normal practicewhen the positions were being taken weeks, months,and years ago. This also was a lesson apparently notlearned during the Asian crises. Greater diligence incredit risk assessment and oversight of credits toLTCM and other hedge funds also might have dimin-ished the extent to which their exposures made credi-tor and counterparty institutions vulnerable to thiskind of turbulence.

Risk management models and modern techniquesof portfolio management typically presume that posi-tions can be closed out in orderly, liquid, continuousmarkets, with low transaction costs. These conditionsare unlikely to be present during times of stress andturbulence. Similarly, although market risk assess-ment and management appear to have reached a so-phisticated level, and have met with some success incontaining private risk in normal markets, such sys-tems can break down in times of stress and turbu-

60

24During the past two years, central banks have voiced concernsabout credit standards, the relaxation of loan covenants, and com-pressed interest rate spreads.

lence, when market correlations suddenly shift. In ad-dition, the near collapse of LTCM, and the potentialcounterparty risk its failure posed, can be seen as awarning sign that some banks may not be devotingsufficient attention to credit risk assessment andmanagement. Moreover, previous crises, and now themature market turbulence, strongly suggest that so-phisticated financial institutions with diverse rangesof semi-independent operations have not yet found afully satisfactory way to assess the level of consoli-dated risks at the institutional level, and in particulardo not yet fully incorporate (into their position tak-ing, financing, and leverage) an adequate understand-ing of the impact of market risk events on credit riskassessments. Finally, the recent turbulence in themature markets indicates that liquidity risk is animportant area that also may not be sufficiently wellunderstood.

It is tempting to blame the shortcomings in the ap-plication of modern quantitative approaches to risk as-sessment, risk management, and portfolio manage-ment to failures in the risk management technologiesand systems themselves. The technical details of mod-els, the sensitivities to assumptions, including the as-sumed probabilities of adverse events in stress testing,and the excessive risk tolerance limits may all be partof the problem. But an equally important shortcomingmay be the element of human judgment required toimplement these technologies and systems and to as-sess the economic and financial environment. Also rel-evant are the incentives within these organizations tomaximize short-term gains and individual bonuses, attimes at the expense of the firm’s overall risk exposureand longer-term profit. In this regard, greater dili-gence, especially surrounding creditor and counter-party relationships between the major financial insti-tutions and the hedge fund LTCM, was probablycalled for.

The recent turbulence also appears to have extendedbeyond the aggregation of the individual responses tomargin calls and the need to unwind, deleverage, andhedge positions as protection against the adverse con-sequences of owning assets in declining markets. Inparticular, the impact of the simultaneous rush ofmany investors to close out positions and deleverageseemed to have been magnified by the inability of in-ternational financial markets to absorb the first roundof the Russian unilateral restructuring. The process ofadjustment seems to have taken on a life of its ownthrough the structure of linkages between the rela-tively small circle of counterparties; the high, and per-haps excessive, degree of short-term competitive pres-sures; the complex manner in which the positions wereoriginally financed, leveraged, and hedged; and the di-versity of mature markets, currency denominations,and maturity structures that facilitated these transac-tions. In short, there also seemed to be systemic com-ponents that contributed to the virulence of the mature

market turbulence that few, if any, participants fullyanticipated. Accordingly, the market turbulence, andthe issues raised by it, also need to be examined at thesystemic level.

Public Systemic Risk Management

An important line of defense against systemic prob-lems is financial supervision and regulation. Modernfinancial institutions are complex organizations, andthe risks taken by them may not always be fully un-derstood by those who manage them. How, then, cansupervisors be expected to make informed judgmentsabout whether institutions are financially sound?Finding operational answers to this difficult questionfor the industrial countries of North America, Europe,and Japan, of a form that can be applied with reason-able consistency for a wide range of internationally ac-tive commercial banks, is the continuing task of theGroup of Ten (G-10) banking supervisors under theauspices of the Basle Committee on Banking Super-vision. The established approach, which relies onquantitative rules, such as the Basle capital ratio, hasproved useful in providing enhanced discipline for risktaking by covered institutions—although there havebeen instances where the rules have apparently notbeen rigorously applied. However, it is also widelyrecognized (Box 3.6) that the established Basle ap-proach has significant deficiencies, especially in thearea of off-balance sheet activities, which are of sub-stantial and growing importance for most, large, inter-nationally active commercial banks.25 This has led theBasle Committee to consider alternative approachesthat would supplement present rules with supervisoryassessments of the adequacy of risk management sys-tems and systems of operational controls (see Box3.6). In this regard, the recent experience suggestingsignificant deficiencies in the performance of riskmanagement and operational control systems in atleast some systemically important institutions clearlyraises important concerns upon which the BasleCommittee and others interested in effective bank su-pervision will need to reflect.

It is not evident that any system of supervision orsurveillance could have identified these problems asthey were actually developing. Nevertheless, it seemsplausible that some of the excessive risk taking andleveraging could have been avoided if home nationalsupervisors, and those responsible for market surveil-lance, knew more about the buildup of both balancesheet and off-balance-sheet positions, leverage, andboth the aggregate amount and distribution of risk tak-ing in national and international markets. For example,LTCM was known, and even advertised, to have a

Shortcomings in Risk Management, and Prudential Implications

61

25These deficiencies have been discussed in previous IMFInternational Capital Markets reports.


large appetite for risk.26 If the institutions that pro-vided credit to facilitate LTCM’s position taking andleverage had been encouraged to come to the judg-ment that LTCM was an excessively risky counter-

party, then LTCM might not have gone as far as it didin taking risk. Likewise, had LTCM’s counterpartiesnot been so exposed to high-risk positions with asmany high-risk players, the buildup would have beenmore limited.

Another line of defense against systemic problemsis financial market surveillance. For example, the U.S.Federal Reserve’s involvement in the markets meansthat it has continuous access to market intelligence andinformation. Moreover, to fulfill its mandate to ensureU.S. financial system stability, it has over the years de-

62

Structural changes in financial markets have encour-aged international bank supervisors and the BasleCommittee on Banking Supervision to shift graduallyfrom “rules-based” to “risk-focused” methods of super-vision, particularly in setting capital requirements formarket and credit risks. Despite some progress in thetreatment of market risk, recent events have exposedshortcomings in the treatment of credit risk, as laid outin the 1988 Basle Accord. These shortcomings are pri-marily in three areas: the distortions arising from riskweights, including on interbank claims and sovereigndebt of OECD countries; the disregard of the effects ofthe business and credit cycles on credit risk; and the ne-glect of the broader operating environment for banks inemerging markets. Measures to address such shortcom-ings are under discussion by members of the BasleCommittee and other international groupings of finan-cial supervisors.

Encouraged by the rapid changes in the structure offinancial markets and perceived shortcomings of pastapproaches, supervisors are gradually changing em-phasis, focusing less on static concepts of risk and moreon the systems and procedures that firms use to measureand manage risk. Leading the way within this newparadigm have been supervisory changes to capitalrequirements relating to market risk. The Basle Com-mittee’s 1996 guidelines on capital requirements formarket risk, which became applicable and mandatoryfor internationally active banks on January 1, 1998,represent a watershed in the regulatory treatment ofcapital. The new requirements allow banks to use theirown internal models for the determination of market riskcapital. While only a few countries have banks withmodels that pass muster, their sanctioned use has raisedissues about the role of regulatory capital moregenerally.

Notwithstanding this recent progress, three shortcom-ings remain in the regulatory treatment of credit risk.First, distortionary effects of the capital accord arise fromthe arbitrary manner in which the risk weights are as-signed. For instance, under the 1988 Basle Accord, short-

term claims on banks from any country carry a relativelylow (20 percent) risk weight, leading to a lower cost ofborrowing in the interbank market and a heavier relianceon interbank funding.2 Also, the accord assigns a zerorisk weight to instruments issued or guaranteed by OECDgovernments. It has been suggested by some BasleCommittee members that the OECD designation hasserved as a “stamp of approval” and has encouragedbanks to steer funds to OECD emerging markets ratherthan to non-OECD countries with equivalent or lowersovereign risks.

The second shortcoming is the arbitrary and unchang-ing 8 percent minimum capital assigned to risk-weightedassets. The constancy of this capital requirement over thebusiness and credit cycles is viewed by some as unneces-sary and undesirable. An alternative is to require banks tohold a higher ratio of capital to risk-weighted assets dur-ing the cycle’s upswing, for two reasons. First, somecushion above the 8 percent minimum would be in placewhen the business cycle turns down. Although capitalmay be above the required minimum at the peak of thecycle, the additional buffer may not be sufficient in lightof the increased risk. Second, at the peak of the cycle, theriskiness of banks’ assets may be well above the averagefor the cycle, but the 8 percent minimum does not adjustupward as the cycle matures to take account of the in-crease in risk. Indeed, the risk of a sharp downturn inasset quality is probably highest at the peak of the busi-ness cycle, but current capital standards do not accountfor this dynamic. There is also a well-documented “creditcycle” in most countries, with banks moving down thecredit curve as the cycle expands, taking on more andmore risk. To the extent that the credit cycle tends to co-incide with the business cycle, as it likely does, thiswould tend to augment the risk of a sharp downturn inasset quality during the mature phase of the businesscycle. A minimum capital requirement that acknowl-edged these dynamics in credit risk and varied with riskover the business cycle would help to accommodate theserisks.

Box 3.6. Supervisory Reforms Relating to Risk Management1

1This box draws heavily on the September 1998 InternationalCapital Markets report, Chapter V.

2See Box 5.8 in International Capital Markets, September1998, for the accord’s current risk-weighting scheme for on-balance-sheet assets.

26LTCM’s original prospectus warned investors that “it is ex-pected that the Portfolio Company will generally be very highlyleveraged, and that such leverage will generally be higher than thatof other typically leveraged investment funds” and that “returns areanticipated to be volatile, especially on a monthly and quarterlybasis.”

veloped sophisticated research and surveillance func-tions, which include aspects of banking supervision aswell. Given the turbulence that occurred, and its statu-tory mandate for ensuring financial stability, theFederal Reserve reached the judgment that it needed tofacilitate a private rescue of LTCM as well as reducethe cost of liquidity. Together these measures fenced inan important aspect of the ongoing turbulence—by in-ternalizing some of the unwinding of the positions ofLTCM and the other financial institutions that hadsimilar positions—and eased liquidity pressures in the

markets when such action was most needed. The finalcut in mid-November probably was taken, in part, asinsurance against a relapse.

Probably no system of market surveillance, in par-ticular of the U.S. financial system, could have accu-rately foreseen what unfolded during the turbulence inSeptember and October. However, the bouts of turbu-lence, illiquidity, price disconnects, and other featuresof the sharp dynamics strongly suggest that the turbu-lence that erupted in the aftermath of the flight fromemerging markets in mid-1998 may have been partly


63

A third shortcoming is the neglect of the banking sys-tem’s larger operating environment. The 8 percent mini-mum was set with the industrial countries’ banking sys-tems in mind. The accord’s adoption by many developingcountries, where economic business cycles have largerswings and the operating environment for banks is muchriskier, means that these banking systems are less pro-tected than those in industrial countries. These problemsargue for a more flexible approach toward capital re-quirements for credit risk in which a broader view aboutrisk is incorporated.

Members of the Basle Committee recognize such defi-ciencies and are discussing the merits of a possible revi-sion to risk weights. Members’ suggested revisions to thecapital accord include promoting better implementation,altering risk weights to better reflect actual risk, and in-corporating portfolio-based risk models along the linesof capital requirements for market risk. U.S. FederalReserve Board Chairman Alan Greenspan has advocatedan increase in the risk weight on short-term interbankclaims, which would raise the cost of borrowing, dis-courage excessive use of interbank funding, and encour-age securitization of short-term claims. Other suggestionshave included additional requirements that would need tobe met before a zero risk weight could be applied toOECD sovereign debt, such as transparency and disclo-sure about the financial sector and implementation of theBasle Core Principles. The International Swaps andDerivatives Association (ISDA) has promoted a mixedapproach involving the current accord, a modified ver-sion of the current accord, and a portfolio-modeling ap-proach to capital requirements. Despite the pressures tomove on the topic, the Basle Committee is likely to main-tain its consensus-oriented deliberateness.

A reevaluation of the role of capital is also under waywithin the International Organization of SecuritiesCommissions (IOSCO) and many of the securities com-missions it represents. In the former regime, capital pro-tected securities firms against unexpected liquidityshortages, allowing them to meet daily settlement flowsand initiate an orderly windup if necessary. As banksand securities firms become increasingly involved insimilar products and business activities, it has becomeless clear whether the different motives for capital re-quirements for the two types of firms still make sense.

Level playing fields and regulatory arbitrage mean thatcapital requirements for banks and securities firms areunlikely to be far different for long.

Since market risk is the dominant risk faced by securi-ties firms, capital requirements for market risk are likelyto be a significant part of any unified approach. For ex-ample, the U.S. Securities and Exchange Commission(SEC) is already trying to determine how best to gain ex-perience with the use of VaR models in the determinationof capital requirements. In the United Kingdom, theSecurities and Futures Authority (SFA), to be merged intothe Financial Services Authority, has released a consulta-tive paper outlining the impact of the introduction of theEuropean single currency on its regulatory capitalregime. The SFA is using this opportunity to revisit anumber of issues.

While credit risk capital requirements are beeing de-bated, supervisory guidance dealing with operational andother risks is also being considered by both the BasleCommittee and IOSCO. Recognizing that operationalfailures are the most common cause of financial institu-tion failures, the two organizations are promoting opera-tional controls and guidelines. Previous guidance issuedby the Basle Committee has covered internal controls as-sociated with specific areas of banks’ activities, while therecent document, “Framework for the Evaluation ofInternal Control Systems,”3 provides a framework for acomplete evaluation of internal controls for all on- andoff-balance-sheet activities. The IOSCO initiative, “RiskManagement and Control Guidance for Securities Firmsand their Supervisors,” combines risk management andoperational controls as part of a larger goal of managingall types of risk—market, credit, legal, operational, andliquidity—noting that risks can come from both internal(for example, insufficient internal controls) and externalsources (for example, sharp price changes). The princi-ples of good risk management and control systems are in-tended as benchmarks against which firms and supervi-sors in each jurisdiction can judge the adequacy of theircontrol systems.

3Basle Committee on Banking Supervision, “Framework forthe Evaluation of Internal Control Systems” (Basle, January1998).


the result of pressures that accumulated over a longperiod of time—in particular, during the long “bull”runs in fixed-income markets. The potential risks as-sociated with these developments—which could havebeen triggered by some event that threatened the posi-tions held by the large and diverse group of financialinstitutions—clearly should have received greater at-tention. Moreover, disclosure requirements for the un-regulated hedge funds are limited, and so their activi-ties are not transparent to all stakeholders. Indeed,market behavior during the summer and fall suggestthat there may not be sufficient disclosure and trans-parency for even the most sophisticated players toknow enough about the credit and counterparty risksthey are taking.

Another observation is that public systemic riskmanagement did not become proactive about the po-tential accumulation of financial vulnerabilities anddisturbances until some of their adverse consequenceswere becoming painfully apparent. Warning signswere present almost two years ago when some centralbanks suggested that equity valuations were beginningto look unsustainable (irrational exuberance), creditrisk spreads were unusually narrow and compressed,and loan covenants and nonfinancial terms were beingrelaxed. With the benefit of hindsight, it is possible tosuggest that absent from these concerns were warningsthat the degree of (off-balance-sheet) leverage was po-tentially becoming excessive, that credit extension tohigh-risk enterprises (hedge funds) was widespreadand also possibly excessive, and that there was an ex-cessive amount of position taking on the faith that ma-ture market credit risk spreads would narrow once the“Asian contagion” dissipated. This suggests that thereshould be a more heightened awareness about poten-tial financial vulnerabilities and disruptions when theyare least expected—when economic and financial con-ditions are favorable and, in particular, when expan-sions in economies are reaching a mature stage. It is atthe top of business and credit cycles as credit riskspreads narrow, when there are strong incentives toimprove asset returns through leverage, and whenbank capital appears ample.

* * *

The difficulties encountered in September andOctober 1998 suggest that financial markets can be ad-versely affected by the manner in which individual fi-nancial institutions react to market pressures, stress,and turbulence, particularly when many of them holdsimilar highly leveraged positions. Mistakes appear tohave been made by systemically important, interna-tionally active financial institutions, and it appears thatmanagement command and control systems should bereassessed to better cope with the risks inherent inmodern financial markets. It is a necessary first step toenhance disclosure of the financial activities of finan-cial institutions, particularly the least regulated among

them. This can enhance the ability of both private andpublic sector stakeholders to assess financial risks andto understand where the risks reside. The widespreadapplication of private risk management systems,which could benefit from greater disclosure, alsoshould be reassessed in light of recent experience.These systems appear to have not incorporated somelessons from the Mexican and Asian crises, whichcould have aided institutions in avoiding vulnerabili-ties exposed by the most recent bout of turbulence inmature markets. Because many of the involved insti-tutions are part of their national financial safety nets,supervisors and regulators need to be closely involvedin this reassessment.

In addition to the adverse developments related di-rectly to behavior of individual financial institutions,the mature market turbulence also seemed to have re-flected features of the international financial system,including the highly integrated and complex nature offinancial position taking, institutions, and markets,and in particular the linkages of financial positionsacross national and international markets. These link-ages derive in part from basic elements of financialtransactions, such as the need or desire to finance,leverage, hedge, and risk-manage diversified portfo-lios of claims. The linkages themselves, the dramaticnature of the position unwinding and deleveraging,and the severe price dynamics and illiquidity in theinterconnected markets together revealed the possi-bility of significant problems in how the internatio-nal financial system—composed of the intercon-nected national and international financial markets—performs in the presence of heightened financial vul-nerability, stress, and ultimately excessive turbulence.Particularly disturbing is the revelation that otherwisedeep and liquid mature (dollar-based) and interna-tional (yen/dollar) markets experienced difficulties inabsorbing what appeared initially to be relativelymoderate shocks.

The feedback effects of the growing market ten-sions, illiquidity, and rapid market dynamics intensi-fied pressures on market participants to shed risk andacquire liquidity rapidly, which apparently impairedthe ability of the mature financial markets to smoothlyand efficiently facilitate the closing out and deleverag-ing of positions and exposures. Distinctions betweencredit, market, liquidity, and operational risks appar-ently became blurred as replenishing liquidity becamethe driving force behind risk assessment, asset pricing,and portfolio rebalancing. Ex ante, few would haveexpected this to occur in deep and liquid financialmarkets. This suggests that private risk models andtheir underlying principles of risk management andportfolio management have a tendency during periodsof stress and turbulence to accelerate dynamics in de-clining markets, which in this most recent episode ledto a temporary drying up of liquidity in even the mostliquid markets.

64

The role of banking supervisors and those responsi-ble for market surveillance in warning about the accu-mulation of increasing levels of risk and leverage inthe mature markets is also an issue of concern. The ar-gument often heard in the aftermath of the Asian crisiswas that no one could see through the opaque financialstructures and markets. Yet the markets and institu-tions that experienced the turbulence this summer andfall are the most open and transparent in the world.Why then were potential dangers not more accuratelyperceived at an earlier stage?

The features of the international financial system re-vealed by the turbulence in the period mid-Augustthrough mid-October 1998 suggest that neither privatemarket participants nor the institutions in charge ofprudential supervision and market surveillance have afull understanding of the ever-changing nature, struc-ture, and dynamics of the rapidly changing interna-

tional financial institutions and markets.27 This, ofcourse, is not an entirely new problem, and it is not aproblem that possesses a complete and final solution.Rather, the difficulties revealed by recent financialmarket turbulence testify to the urgency of continuingefforts, in the private and public sectors, to improvethe performance and enhance the stability of the inter-national financial system.


65

27U.S. Federal Reserve Board Chairman Alan Greenspan madethis observation in his remarks, “Risk Management in the GlobalFinancial System,” before the Annual Financial Markets Con-ference of the Federal Reserve Bank of Atlanta, in Miami, Florida,on February 27, 1998. In referring to the dramatic changes that haveoccurred in what was characterized as “the global financial sys-tem,” Chairman Greenspan observed, “We do not as yet fully un-derstand the new system’s dynamics. We are learning fast, and needto update and modify our institutions and practices to reduce therisks inherent in the new regime.”

Turbulence in Mature Financial Markets · Until July 1998, the mature financial markets in the...

Documents

Transcript of Turbulence in Mature Financial Markets · Until July 1998, the mature financial markets in the...