review financial system - bank of canada · bank of canada 234 wellington street ottawa, ontario...
TRANSCRIPT
20072007
www.bankofcanada.ca
Financial SystemReviewDecember 2007
Finan
cial System R
eview—
Decem
ber R
evue d
u systèm
e finan
cier — D
écembre
www.banqueducanada.ca
Revue dusystème financierDécembre 2007
fsr_cover_december_07.qxd 11/29/2007 1:06 PM Page 1
FinancialSystem ReviewDecember 2007
Bank of Canada234 Wellington Street
Ottawa, Ontario K1A 0G9
ISSN 1705-1290
Printed in Canada on recycled paper
The Financial System Review and Financial Stability
The financial system makes an important contribution to the welfare of all Canadians. Theability of households and firms to confidently hold and transfer financial assets is one of thefundamental building blocks of the Canadian economy. As part of its commitment to pro-moting the economic and financial welfare of Canada, the Bank of Canada actively fosters asafe and efficient financial system. The Bank’s contribution complements the efforts of otherfederal and provincial agencies, each of which brings unique expertise to this challengingarea in the context of its own institutional responsibilities.
The financial system is large and increasingly complex. It includes financial institutions (e.g.,banks, insurance companies, and securities dealers); financial markets in which financial as-sets are priced and traded; and the clearing and settlement systems that underpin the flowof assets between firms and individuals. Past episodes around the world have shown thatserious disruptions to one or more of these three components (whether they originate fromdomestic or international sources) can create substantial problems for the entire financialsystem and, ultimately, for the economy as a whole. As well, inefficiencies in the financialsystem may lead to significant economic costs over time and contribute to a system that isless able to successfully cope with periods of financial stress. It is therefore important thatCanada’s public and private sector entities foster a financial system with solid underpin-nings, thereby promoting its smooth and efficient functioning.
The Financial System Review (FSR) is one avenue through which the Bank of Canada seeks tocontribute to the longer-term robustness of the Canadian financial system. It brings togetherthe Bank’s ongoing work in monitoring developments in the system and analyzing policydirections in the financial sector, as well as research designed to increase our knowledge. Thestrong linkages among the various components of the financial system are emphasized bytaking a broad, system-wide perspective that includes markets, institutions, and clearing andsettlement systems. It is in this context that the FSR aims to
• improve the understanding of current developments and trends in the Canadian andinternational financial systems and of the factors affecting them;
• summarize recent work by Bank of Canada staff on specific financial sector policies andon aspects of the financial system’s structure and functioning;
• promote informed public discussion on all aspects of the financial system, together withincreased interaction on these issues between public and private sector entities.
The FSR contributes to a safe and efficient financial system by highlighting relevant informa-tion that improves awareness and encourages discussion of issues concerning the financialsystem. The Bank of Canada welcomes comments on the material contained in the FSR.
Members of the Editorial Committee
Pierre Duguay and David Longworth, Chairs
Steve AmblerLloyd Barton
Allan CrawfordPaul Fenton
Clyde GoodletDonna HowardLouise Hyland
Jean MairJohn Murray
Graydon PaulinGeorge Pickering
Denis SchutheJack Selody
Robert TurnbullMark Zelmer
Jill MoxleyMadeleine Renaud
Lea-Anne Solomonian(Editors)
The significant contribution of the working group mandated with the preparation andorganization of the Review is gratefully acknowledged.
Bank of CanadaDecember 2007
The Bank of Canada’s Financial System Review is published semi-annually. Copies maybe obtained free of charge by contacting
Publications Distribution, Communications Department, Bank of Canada, Ottawa,Ontario, Canada K1A 0G9Telephone: 1 877 782-8248; email: [email protected]
Please forward any comments on the Financial System Review to
Public Information, Communications Department, Bank of Canada, Ottawa,Ontario, Canada K1A 0G9Telephone: 613 782-8111, 1 800 303-1282; email: [email protected]
Website: <http://www.bankofcanada.ca>
Contents
Developments and Trends ............................................................ 1
Financial System Risk Assessment......................................................... 3
Overview..................................................................................................... 3
The Financial System............................................................................. 7
Financial markets ........................................................................... 7
Highlighted Issue: The market for Canadian asset-backedcommercial paper, revisited .......................................................... 13
Payment and settlement systems ........................................................ 16
Financial institutions ....................................................................... 17
The Macrofinancial Environment............................................................ 19
The international environment.......................................................... 19
Canadian developments.................................................................... 20
Highlighted Issue: Stress testing the Canadian household sectorusing microdata .......................................................................... 26
Important Financial System Developments ............................................ 31
Financial Collateral for Eligible Financial Contracts ................................. 31
Elimination of Withholding Tax on Interest ............................................. 31
The Mortgage Insurance Market ............................................................. 31
Highlighted Issue: Free trade in securities ................................................ 32
Reports ...................................................................................... 35
Introduction .................................................................................................... 37
An Approach to Stress Testing the Canadian Mortgage Portfolio ........................... 39
The Changing Landscape of Securities Trading ................................................... 45
Reforming the Credit-Rating Process .................................................................. 51
Policy and Infrastructure Developments ........................................ 59
Introduction .................................................................................................... 61
Developments in Processing Over-the-Counter Derivatives ................................... 63
Management of Foreign Exchange Settlement Risk at Canadian Banks ................. 71
Contents (continued)
Research Summaries ................................................................... 79
Introduction .................................................................................................... 81
The Provision of Central Bank Liquidity under Asymmetric Information................ 83
Collateral Portfolios and Adverse Dependence ..................................................... 87
Housing Market Cycles and Duration Dependence in the United Statesand Canada ................................................................................................ 91
The following people contributed to theDevelopments and Trends section:
Jim ArmstrongLloyd Barton
Michael BonazzaLindsay Cheung
Daniel de MunnikRamdane Djoudad
Chris GrahamToni GravelleDylan Hogg
Jocelyn JacobMarianne Johnson
Nadja KamhiMichael King
Ilan KoletRobert LavigneDanny Leung
Guy MacKenzieJean Mair
Paul MassonKaren McGuinness
Graydon PaulinChris Reid
Gerald StuberVirginie Traclet
Eric Tuer
Developments
and
Trends
Notes
The material in this document is based on information available to 22 November 2007unless otherwise indicated.
The phrase “major banks” in Canada refers to the six largest Canadian commercialbanks by asset size: the Bank of Montreal, CIBC, National Bank, RBC Financial Group,Scotiabank, and TD Bank Financial Group.
Assessing Risks to the Stability of theCanadian Financial System
The Financial System Review (FSR) is one vehicle that the Bank of Canada uses tocontribute to the strength of the Canadian financial system. The Developmentsand Trends section of the Review aims to provide analysis and discussion ofcurrent developments and trends in the Canadian financial sector.
The first part of this section presents an assessment of the risks, originating from bothinternational and domestic sources, that could affect the stability of the Canadianfinancial system. Key risk factors and vulnerabilities are discussed in terms of anypotential implications for the system’s overall soundness. The second part of theDevelopments and Trends section examines structural developments affecting theCanadian financial system and its safety and efficiency; for example, developmentsin legislation, regulation, or practices affecting the financial system.
The current infrastructure, which includes financial legislation, the legal system,financial practices, the framework of regulation and supervision, and the macro-economic policy framework, significantly influences the way in which shocks aretransmitted in the financial system and in the macroeconomy, and thus affectsour assessment of risks.
Our risk assessment is focused on the vulnerabilities of the overall financial sys-tem, and not on those of individual institutions, firms, or households. We there-fore concentrate on risk factors and vulnerabilities that could have systemicrepercussions—those that may lead to substantial problems for the entire finan-cial system and, ultimately, for the economy. In examining these risk factors andvulnerabilities, we consider both the likelihood that they will occur and theirpotential impact.
Particular attention is paid to the deposit-taking institutions sector because of itskey role in facilitating financial transactions, including payments, and its interac-tion with so many other participants in the financial system. For instance, theseinstitutions assume credit risks with respect to borrowers such as households andnon-financial firms. Thus, from time to time, we assess the potential impact thatchanges to the macrofinancial environment may have on the ability of householdsand non-financial firms to service their debts.
Risk factors and vulnerabilities related to market risks are also examined. Thepotential for developments in financial markets to seriously affect the financialposition of various sectors of the economy and, ultimately, to disrupt the stabilityof the Canadian financial system is assessed.
Financial System Review – December 2007
Financial System Risk Assessment
his section of the Review presentsan assessment of the risks arisingfrom both international anddomestic sources bearing on the
stability of the Canadian financial system.The objective is to highlight key risk fac-tors and vulnerabilities in the financialsystem and to discuss any potential impli-cations for the system’s overall soundness.
T
Key Points• Sudden repricing of risk has been accom-
panied by considerable turbulence in themoney market.
• Liquidity evaporated in the market forstructured products.
• Increased demand for short-term liquidityby banks, together with concerns aboutthe creditworthiness of counterparties, putupward pressure on money market ratesinternationally and in Canada.
• There has been some tightening in creditconditions.
• The solid financial positions of the Cana-dian financial, non-financial corporate,and household sectors have helped themweather the turbulence.
• Fears of a much-greater-than-expected dete-rioration in the U.S. housing market and inthe quality of U.S. mortgage-related assets inthe future have further exacerbated liquid-ity problems in some financial markets,increasing counterparty concerns, andraising risk premiums.
• Financial market turbulence, together with asignificant slowdown in the U.S. economy,could lead to and be exacerbated by a disor-derly resolution of global current accountimbalances.
Overview
The Financial System Review has been highlightingthe possibility of a sudden repricing of risk forsome time, pointing to unusually narrow riskspreads that resulted from the search for yield inan environment of low interest rates, and thedifficulty of assessing the risks of the increasinglycomplex products being generated by new finan-cial engineering. Since the publication of theJune FSR, the sudden repricing has materialized.Risk spreads have widened, volatility in financialmarkets has increased, and liquidity in the mar-kets for some structured products has evaporated.The turbulence has primarily affected moneymarkets, although longer-term spreads havealso widened. The U.S. dollar has also fallensignificantly against all other major currencies,including the Canadian dollar.
The turbulence in global financial markets wastriggered by concerns about the value of structuredproducts based on U.S. subprime mortgages,reflecting growing delinquencies in these mort-gages. This disquiet subsequently broadened toinclude a wide range of structured products—because some of these products containedsubprime mortgages and because investors haddifficulty in valuing these securities, owing to theircomplex structures and a lack of information aboutthe assets backing them. Liquidity evaporatedin the secondary market for structured productssince investors feared that they would be unableto sell assets quickly at prices commensuratewith what they thought they should be worth—a fear that became a self-fulfilling prophecy.There was a flight to quality assets, and yieldson treasury bills and government bondsdropped significantly. Yield spreads widened,albeit less for long-term debt than for short-term debt, and equity markets fell significantly.
The effects have been most marked in short-termmoney markets. One market that has beenparticularly affected is that for asset-backedcommercial paper (ABCP). As concerns
3
Developments and Trends
increased about the quality of the underlying as-sets in these structures, issuers had trouble roll-ing over commercial paper. Rates on the paperrose sharply, and the maturity of the paper fell.In Canada, an additional difficulty was that somenon-bank-sponsored conduits were unable todraw on backup liquidity lines from banks,prompting a call for a standstill (the MontrealProposal)1 to effect an orderly workout.
An unexpected consequence of these develop-ments was their impact on interbank markets.Many banks were affected, both in Canada andabroad, because of commitments to providefunding to the ABCP conduits, and because itbecame much more difficult to securitize assets.As well, with commercial paper rates elevated,companies began to draw on their lines of creditwith banks. Banks were also concerned that theywould be called upon to provide financing tocompanies that had formerly made use of longer-term financial markets. Banks started to buildup liquidity, and interbank rates, especially forterm lending, rose well above their usual spreadover expected future overnight rates. In manycases, this was exacerbated by concerns aboutthe creditworthiness of counterparties. Centralbanks in many countries, including Canada,offered assistance to overnight markets to keepovernight rates near target levels when pressureson short-term funding emerged.
Although liquidity in some short-term marketsin Canada has improved since mid-August, andrates for bank funding and bank-sponsored ABCPhave come off their August peaks, these ratesremain elevated. Banks have taken back ontotheir balance sheets some of the asset-backedcommercial paper issued by conduits that theysponsored. The market for non-bank-sponsoredABCP remains frozen with the standstill periodunder the Montreal Proposal having been ex-tended to 14 December. But, at the time of writ-ing, it appears that progress is being made in thenegotiations to establish a framework to convertshort-term non-bank ABCP into medium-termtradable financial instruments. It was announcedthat the restructuring process should be com-pleted by the end of March 2008.2
1. See <http://documentcentre.eycan.com/pages/main.aspx?SID=35> for information concerningdevelopments in the Montreal Proposal.
2. Purdy Crawford, letter to Financial Post and Le Devoirnewspapers, 22 November 2007.
4
In most countries, including Canada, longer-termmarkets have been less affected by the turbulence.Although credit spreads have widened from thelow levels seen earlier this year, they remain farbelow the peaks seen in 1998 and 2002. More-over, the effect on funding costs for corporationshas been mitigated by a decline in governmentbond yields.
It is unlikely that financial markets will returnto their pre-turbulence state. First, as discussedin previous issues of the FSR, the narrow creditspreads observed then did not adequately reflectthe risks that were being taken. As well, onewould expect some changes in the functioningof the financial system because of the difficul-ties highlighted by the recent market events: theprincipal-agent problem associated with theway that the originate-and-distribute model hasbeen applied, and the difficulty that investorsface in evaluating opaque and complex financialassets. For example, sponsors of asset-backedcommercial paper conduits may have to give aclearer idea of the assets being funded beforethere is a ready market for the commercial paperthey issue.
Indeed, a number of international groups haverecognized the importance of carefully examiningthe sources of the recent turbulence and the les-sons that can be learned from it. In September2007, G-7 finance ministers and central bankgovernors asked the Financial Stability Forum(FSF) to establish a working group to identifyweaknesses that merit attention from policy-makers and to recommend actions needed toenhance market discipline and institutional resil-ience. The working group (which includes theCanadian Superintendent of Financial Institu-tions) is examining:
• risk-management practices (includingliquidity management, stress testing, andassessment of counterparty risk)
• valuation and risk disclosure
• the role of credit-rating agencies
• principles and practices of prudentialoversight, particularly with respect to off-balance-sheet exposure
• key issues related to authorities’ capacity torespond to episodes of market turbulence,including the tools and instruments avail-able to central banks and supervisors intimes of distress
Financial System Review – December 2007
The group’s preliminary report was presented tothe G-7 ministers and governors on 16 October2007. 3
The Bank of Canada is working closely with otherCanadian regulatory authorities in reviewing therecent market events and the issues that theyraise. The Department of Finance is coordinatingthese efforts.
In this context, the Bank of Canada will be exam-ining the principles and practices relating to itsliquidity facilities. In particular, it will examinewhether it might be useful to have a facility thatwould provide liquidity to banks at terms long-er than overnight, possibly collateralized with awider range of securities than the Bank currentlyaccepts. This examination will involve identifyingthe kinds of market failure any such facility wouldbe designed to deal with. It will also benefit fromparallel work going on in other central banks.
Market participants, including financial institu-tions, hedge funds, and rating agencies, are alsoassessing the need for changes in light of recentevents. The Institute of International Finance4
has established a committee to review risk-man-agement issues, the use of off-balance-sheetvehicles, the valuation of complex products,the interpretation and evaluation of credit ratings,and transparency. A group chaired by Sir AndrewLarge has published a consultation documentproposing best practice standards for hedge funds,focusing on valuation, risk management, disclo-sure, and fund governance. The practices wouldbe voluntary and would operate on a “complyor explain” basis.5
Canadian financial situation
While there may be some continued dislocationin some financial markets as they struggle to re-price risk, this does not appear to pose a systemicthreat to the Canadian financial system. Therewill be some impact on the Canadian economydirectly through credit spreads and availability,and indirectly through the effects on the U.S.economy. The effects on the Canadian financialsystem, however, should be mitigated bythe strong balance sheets of financial and
3. Available at <http://www.fsforum.org/publications/publication_24_88.html>.
4. “Regulators urged to take a back seat,” FinancialTimes, 22 October 2007.
5. Available at <http://www.hfwg.co.uk/?section=10365>.
non-financial corporations built up throughyears of strong growth and substantial profits.Moreover, Canadian domestic demand andhigh commodity prices provide support for theincomes of Canadian corporations and house-holds. In fact, the tightening of credit condi-tions resulting from the turbulence is notunwelcome, given the strength of the housingmarket and the continued increase in theindebtedness of the household sector.
Canadian banks appear to be well positionedto absorb the effects of the recent market turbu-lence, because of their initial sound capitalpositions and strong profits. Although bankfunding costs have increased, major Canadianbanks have been able to bolster their liquiditypositions by issuing medium-term paper. Theyhave also been able to issue some capital-eligiblesubordinated debt. Their loan-loss performancehas been good, and their exposure to the U.S.subprime-mortgage market and to leveragedloans appears to be small and manageable. Withthe Canadian household and non-financial cor-porate sectors also in good financial shape forthe most part, and given that banks’ exposuresto the more vulnerable sectors are not large, thedeterioration in the quality of bank loans shouldbe limited. Canadian banks have relied muchless than their U.S. counterparts on the originate-and-distribute model, which should also mitigatethe effect of the turbulence.
As noted above, the recent developments willaffect the Canadian non-financial corporatesector directly through their effects on the costand availability of credit. There will also be animpact through reduced exports because of aslowing U.S. economy, which will exacerbatethe effect of the sharp appreciation of the Cana-dian dollar. The non-financial corporate sectorappears reasonably well positioned, overall, towithstand the effects of recent events. For themost part, profits have remained high and balancesheets are healthy, with low debt-to-equity ratiosand high levels of corporate liquidity (althougha small portion of this liquidity is held in theform of, now-frozen, non-bank-sponsoredABCP). Thus, while there may be major challengesfor some companies, especially those in exportsectors, it is unlikely that there will be widespreadproblems.
The household sector also appears to be in goodfinancial shape. While the debt-to-income ratiohas continued to rise, arrears on loans and
5
Developments and Trends
bankruptcies all remain at relatively low levels,reflecting the buoyant labour market, solidgrowth in household disposable income, and arelatively low debt-service ratio. In contrast tothe United States, there is little likelihood of awidespread weakness in house prices that mightpose a threat to household net worth.6 As withthe corporate sector, a major source of stress tothe household sector is likely to be slowing U.S.demand for Canadian products, which mightlead to job losses. This risk should be mitigated,however, by the current strong domestic demandin Canada. Recent events appear to have had littleeffect on the availability of credit to the house-hold sector, although the cost of credit has in-creased. While this may cause a small increasein the number of households with high debt-service ratios, the effect on the banking system islikely to be limited.
Risks
Uncertainty remains in financial markets. Thefragility of some markets—especially those forstructured products—is continuing to hinderthe ability of market participants to value assetsnormally traded in those markets. The difficultyin valuing assets is creating uncertainty aboutthe accuracy and comparability of the lossesreported by participants.
Amid this uncertainty, a shock could lead to amarked increase in risk aversion, a further dete-rioration in liquidity in markets, and a wideningin risk premiums. The impact could extend tosome markets that, to date, have been littleaffected.
One such shock could be a much greater deteri-oration in the U.S. housing market than financialmarkets currently expect—a deterioration reflect-ing the oversupply of housing and the numberof mortgages that are likely to be subject to aresetting of interest rates over the next year. Asignificant decline in house prices could leadto a large unexpected increase in mortgagedelinquencies causing more foreclosures anda further tightening in credit conditions forhouseholds. This would exacerbate excess
6. In addition, the non-prime-mortgage marketaccounts for a much smaller proportion of origina-tions in Canada than in the United States, is not asdependent on securitization, and was not character-ized by the same relaxation of lending standards.
6
supply in the housing market and cause a fur-ther deterioration in the quality of mortgageloans, and of the financial assets that have beencreated using them. Deteriorating conditions inthe U.S. housing sector could also shake con-sumer confidence more widely, leading to amore pronounced slowing of consumer spend-ing and of the U.S. economy more generally.
Such a deterioration in the U.S. economy, andin the quality of mortgage assets that are presentin many structured products, could exacerbatethe liquidity problems in the markets for struc-tured products and the difficulty in valuing theseassets. A possible need to sell assets to meetmargin calls, together with other forced sales,could trigger a downward adjustment in struc-tured-asset markets; the turmoil might also spreadinto other asset markets. This could reignitecounterparty concerns, as institutions mighthave to mark down assets further. Increases inrisk premiums might spread to long-term creditmarkets, which have been relatively unaffectedby the recent market events.
A further increase in risk premiums, together withthe slowing in the U.S. economy, could also leadto a slowing of activity in the world outside theUnited States. This could reduce or reverse therecent improvement in global imbalances, againraising the prospect of a disorderly resolution ofthese imbalances. This could entail an abrupt andsizable decline in the value of the U.S. dollar,greater volatility in financial markets, a furtherrise in risk premiums, and an increase in protec-tionism. The resulting slowing in world growthwould lead to a decline in commodity prices.
The probability of this scenario is low. But if itwere to materialize, the greater-than-expectedslowing in the U.S., and possibly the global,economy, together with a decline in commodityprices and an unexpected rise in the Canadiandollar, would reduce the profitability of Canadianexporters and increase stress on Canadian busi-nesses, households, and financial institutions.This would have a significant effect on the Cana-dian financial system. The deterioration in thequality of structured-asset markets would againraise questions about the quality of assets heldby conduits. Canadian banks might also be af-fected by a worldwide ratcheting up of concernsabout counterparties, which could affect the costand availability of interbank credit.
Financial System Review – December 2007
At the same time, Canadian businesses couldfind it more expensive and difficult to obtaincredit in financial markets and would turn to thebanks for financing. However, some companies,especially those without a strong ongoing busi-ness relationship with a Canadian bank, andthose seeking funding for transactions per-ceived to be risky (e.g., leveraged mergers andacquisitions) could find it more difficult to ob-tain funding. This might threaten the viabilityof a number of firms.
The Financial System
Financial markets
Since August, credit markets in Canadaand in other major economies, particu-larly the market for the short-term debtof banks and corporations, have beencharacterized by a marked decline in mar-ket liquidity and a repricing of risk. Asidefrom particular issues related to the typeof liquidity facility used in Canada tobackstop ABCP conduits, many of the fac-tors affecting Canadian credit marketshave been those that generated turbulentcredit markets in Europe and the UnitedStates; namely, the transmission of prob-lems in U.S. subprime-mortgage debt viathe market for securitized instruments.
The global strains in credit markets can be tracedback to the past spring, when a repricing of cred-it was triggered by news that delinquency ratesand foreclosures associated with subprimemortgages in the United States had been risingquickly (Box 1). In June, credit-rating agenciesbegan to downgrade mortgage-backed securities(MBSs) and collateralized debt obligations(CDOs) that included U.S. residentialsubprime-mortgage debt. The extent of thedowngrades surprised market participants andled to a reappraisal of credit risk, as well as fearsthat more downgrades were forthcoming. Theunderlying backdrop, prior to the repricing, wasan environment of historically low risk-free in-terest rates, which boosted global demand forhigher-yielding and riskier financial products,including MBSs and CDOs. Indeed, centralbanks, including the Bank of Canada, had forsome time voiced concern that credit risk mightbe mispriced as a result of this “search for yield”phenomenon.7 The news about rising defaults on
7. For more on this, see the Highlighted Issue on p. 18of the June 2007 FSR.
U.S. subprime mortgages in the spring initiateda sustained widening of long-term creditspreads in Canada and elsewhere (Chart 1).
Spreads on longer-term credit continued to wid-en as the summer progressed, not only becauseof further news of downgrades, but also increas-ingly because of declining liquidity in the sec-ondary market for CDOs and MBSs backed bydebt related to subprime mortgages. This, inturn, negatively affected the valuation of theseinstruments (Box 2). The speed and extent ofthe decline in market liquidity created broaderproblems in the market for structured productsas the mark-to-market valuation of these, oftenleveraged, instruments fell. Declines in marketliquidity were partly driven by the “forced”selling of structured products by leveragedinvestors, such as hedge funds and structuredinvestment vehicles, to meet margin calls and/or (anticipated) redemptions.8 (See Box 2.)
The announcement by BNP Paribas on 9 Augustthat it had closed redemptions of three invest-ment funds because it could not value their as-sets in the prevailing illiquid market environ-ment for structured products, provided the cat-alyst for a sharp decrease in the global appetitefor risk. This triggered a broader repricing ofrisky financial assets in world financial markets,including in Canada. Around that time, thesecondary markets for leveraged loans related toleveraged buyouts also shut down.9 Yields ongovernment bonds and treasury bills plummet-ed as investors fled from risky asset holdings
8. The collapse of two Bear Stearns funds resulted pri-marily from these funds having to mark-to-markettheir positions in illiquid and declining CDO andMBS markets. This highlighted to market participantsthe degree of illiquidity in the market for CDOs andother similarly structured products, as well as the riskthat other market participants could be forced tomark-to-market at significantly lower prices, incur-ring significant losses in the process. This led toheightened concerns that this would generate moremargin calls, a further wave of selling, and exposeother leveraged investors to large losses.
9. This reflected both a broad-based decline in riskappetite and a sharp drop in the demand for the typeof CDOs that typically purchased leveraged loans.This, in turn, cut out an important source of fundingfor leveraged loans, closing the market to new issu-ance and to the sale of leveraged loans in secondarymarkets and, thus, forced banks to warehouse loansthat had already been committed.
7
Developments and Trends
Box 1
Recent Developments in Subprime-Mortgage Markets1
Developments in the United StatesThe continued deterioration in the U.S. subprime-mortgage market over the past few months has had awidespread impact on financial markets. After aboutthree years of sustained declines, the delinquency rateon U.S. subprime mortgages started to increase in ear-ly 2006 and is currently over 9 per cent (Chart A). Ris-ing delinquencies and the associated difficulties insecuring funding in financial markets have led to fi-nancial problems among the originators of subprimeand Alt-A mortgages,2 with a number of originatorsfiling for bankruptcy or stopping activity. The dete-rioration in the subprime market also led to tighterlending standards3 and to the removal of severalalternative mortgage products—notably the 2-28mortgages that are particularly prone to delinquency.4
Problems in the U.S. subprime-mortgage market arelikely to continue for some time, with further increas-es in delinquencies and losses. Thus, the pace of newnon-prime mortgage originations—which accountedfor nearly half of total mortgage originations in2006—will likely slow further. This, together with ageneral tightening in mortgage-lending standards,may contribute to further weakness in the housingsector.The U.S. government recently proposed measures toallow a small number of vulnerable borrowers to re-negotiate the terms of their mortgages. The impact ofthese measures on the subprime-mortgage market islikely to be limited.
Developments in theUnited Kingdom and CanadaOver the past few years, subprime-mortgage lendinghas grown rapidly in a number of other countries, no-tably the United Kingdom and Canada, although suchlending accounts for a much smaller share of theirmortgage markets than in the United States.5
In the United Kingdom, underwriting standards ap-pear to have remained tighter than in the United
8
1. For an extensive discussion of the U.S. and Cana-dian subprime-mortgage markets, see the June 2007FSR, pp. 6–9.
2. Alt-A customers are borrowers who have a good credithistory but lack income documentation.
3. The July 2007 Federal Reserve Senior Loan OfficerSurvey shows that of the 16 institutions that reportedhaving originated subprime residential mortgages,about 56 per cent had tightened standards on suchloans. Moreover, 14 per cent of the banks surveyedhave tightened lending standards for prime loans.
4. After the initial two-year period of fixed low interestrates, rates are reset at higher levels for the remaining28 years of the loan.
5. In the United Kingdom, subprime mortgages are esti-mated at 3 to 4 per cent of the total mortgage market(Bank of England, Financial Stability Report, October2007, p. 25). In Canada, subprime-mortgage origina-tions are estimated to account for only 5 per cent oftotal mortgage originations and less than 3 per centof total mortgage loans outstanding in 2006. In theUnited States, subprime mortgages accounted forapproximately 14 per cent of total mortgages outstand-ing and 22 per cent of new mortgage originations in2006.
States.6 The U.K. Financial Services Authority, howev-er, recently concluded that a number of intermediariesand lenders in the U.K. subprime-mortgage markethad failed to adequately assess customers’ ability to af-ford the mortgage or to check the plausibility of infor-mation provided by borrowers.7 Securitization—mainly in the form of Residential Mortgage Backed Se-curities (RMBS)—is the predominant technique usedto fund subprime mortgages in the United Kingdom.Because of problems in the U.S. subprime market andthe associated liquidity crunch, some U.K. subprimelenders have recently tightened their lending condi-tions as market funding becomes more expensive.Recent market developments suggest that growth inthe U.K. subprime-mortgage market should slow.In Canada, subprime lenders have been focusing onnear-prime and Alt-A customers. Moreover, subprimeproducts are more “conservative” than U.S. productssince they do not contain some of the features thathave contributed to the recent rise in delinquenciesamong U.S. subprime mortgages. Consequently, thequality of the Canadian subprime-mortgage marketremains good, as illustrated by low delinquency rates(Chart A). While market funding is used in the Cana-dian market, it is less predominant than in the UnitedStates and the United Kingdom, since some subprimelenders in Canada also rely on deposits. Because ofmore expensive market funding conditions, a numberof Canadian subprime lenders recently announcedthat they would not provide loans to new customers,at least in the near future, and others have tightenedtheir lending conditions. Thus, growth in subprime-mortgage lending in Canada is likely to slow, al-though this slowing may be less pronounced than inother countries, since the subprime lenders that relyon deposits do not encounter the same funding prob-lems as those that rely on market funding.
Chart A Delinquencies on Subprime-MortgagePayments
Per cent in arrears over 90 days or in foreclosure
%
0
2
4
6
8
10
12
14
0
2
4
6
8
10
12
14
1998 2000 2002 2004 2006
* Average of major subprime lendersSource: Mortgage Bankers Association and Bank of Canada
United StatesCanada*
6. See Bank of England Financial Stability Report, April2007, pp. 28–29.
7. For more details, see <http://www.fsa.gov.uk/pages/Library/Communication/PR/2007/081.shtml>.
Financial System Review – December 2007
Chart 2 Yields on 10-Year Government Bonds
%
3.5
4.0
4.5
5.0
5.5
6.0
3.5
4.0
4.5
5.0
5.5
6.0
2006 2007
United StatesCanadaEuro zone
Source: Bank of Canada
Chart 1 Corporate Bond Spreads
Basis points Basis points
50
100
150
200
250
0
300
600
900
1,200
2002 2003 2004 2005 2006 2007
Canada investment-grade (left scale)U.S. investment-grade (left scale)U.S. high-yield (right scale)EMBI+ (right scale)
Source: JPMorgan Chase & Co. and Merrill Lynch
Chart 3 Canadian Short-Term Interest Rates
%
Source: Bank of Canada and Reuters
3.0
3.5
4.0
4.5
5.0
5.5
6.0
2006 2007
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Treasury billsOvernight targetBankers’ acceptancesPrime commercial paperBank-sponsored ABCP
to the safety and liquidity of these securities(Charts 2 and 3). Yields on longer-term corpo-rate bonds rose only modestly or remained rel-atively stable as these events unfolded (Chart 4),owing to the decline in yields on governmentbenchmark bonds. In short, the broad-baseddecline in risk appetite reflected a cumulativebuildup of negative news about the U.S. hous-ing market, ratings downgrades, and the report-ing of losses at financial institutions andleveraged funds, which preceded the triggeringevent of 9 August. (See Table 1 for the chronol-ogy of events.)
The most pronounced impact of these events—aside from major ongoing concerns regardingthe market for longer-term structured prod-ucts—has been a widening of spreads in theshort-term credit markets of most industrializedcountries. In particular, these events led to asharp and, in many cases, unprecedented wid-ening in the spreads between rates in short-termcredit markets (such as ABCP, corporate paper,3-month LIBOR, and 3-month bankers’ accep-tances), and expected overnight rates (over thesame term) in Europe, the United States, andCanada (Charts 3 and 5).
This rise in interbank lending spreads, such asthose on LIBOR and bankers’ acceptances, re-flected an increase in the precautionary demandfor liquidity by banks, mainly because of uncer-tainty regarding funding requirements arisingfrom a potential expansion of bank balancesheets. Specifically, as money market investordemand for ABCP dissipated (HighlightedIssue, p. 13), it became more likely that the var-ious types of conduits would draw on backuplines of liquidity or that banks would effectivelymove these ABCP conduits onto their balancesheets. This occurred at a time when majorbanks already faced the prospect of warehousingon their balance sheets loans made in support ofleveraged buyouts, which would normally havebeen sold shortly after the loan commitmentswere made. This greater demand for liquiditywas amplified by heightened concerns aboutcounterparty risk, largely generated by thebroad-based uncertainty as to where the lossesrelated to U.S. subprime-mortgage debt and themark-to-market losses related to market illi-quidity in structured products resided. Banksfaced difficulties over the period in obtainingfunding with a maturity beyond a week or two.
9
Developments and Trends
The turbulence in global credit markets canbe linked to difficulties in the U.S. subprime-mortgage market and the transmission ofthese problems via the market for asset-backedsecurities and structured products. This partlyreflects the increased repackaging, or securiti-zation, of mortgage loans, particularly in theUnited States and, to a lesser extent, in otherindustrialized countries, into asset-backedsecurities (ABSs) such as mortgage-backedsecurities (MBSs). More recently, these MBSs,along with other ABSs, have been furtherrepackaged into complex structured products,such as collateralized debt obligations (CDOs),and CDOs have themselves been used as theunderlying assets in asset-backed commercialpaper (ABCP).1
When liquidity fell in the market for CDOsbacked by U.S. subprime-mortgage debt, sell-ing by leveraged investors spread to other moreliquid segments of the structured-productmarket, to the ABCP market (HighlightedIssue, p. 13) and, to some extent, to other“more standard” segments of credit markets.Since the valuations of CDOs are usually modelbased and assume that markets are relativelyliquid, forced sales of these assets into anilliquid market led to unanticipated mark-to-market losses for many CDO investors. Thisled to a widespread repricing of structuredand securitized products more generally,including higher-rated U.S. mortgage-backedsecurities (unrelated to subprime mortgages),mortgage-backed securities in other countries,and ABCP backed by more “standard” receiv-ables such as credit card payments, as wellas, to a lesser extent, corporate and emerging-market bond debt and leveraged loans.This widespread increase in credit spreadsreflected not only the growing uncertainty ofmarket participants about the valuations ofa broad range of structured credit instru-ments and their concern that forced “firesales” of these assets would engender moremark-to-market losses for financial institutions,
Box 2
The Loss of Confidence in Ratings
1. For definitions and a discussion of ABCP, MBSs, andCDOs in a Canadian context see Kiff and Morrow(2000), Kiff (2003), Toovey and Kiff (2003), Arm-strong and Kiff (2005), and Kamhi and Tuer (2007).
10
conduits, and leveraged funds, but was alsoamplified by a loss of confidence in thecredit-rating process for structured products(including, later on, ABCP). When the CDOmarket (and, to some extent, the MBS market)came under stress, investors questioned theappropriateness of ratings, given sharp, large-ly mark-to-market declines in the valuationsof structured instruments (partly because ofmarket illiquidity), which were not necessarilyaccompanied by declines in the ratings forthese instruments.The complexity of CDOs, as well as the re-quirement of many institutional investors tohave the level of their fixed-income holdingsdriven by credit ratings, may have led the ul-timate investors to place too great a relianceon the rating of the CDO tranches to guidetheir investment decisions. (See Zelmer,p. 51.) They may not always have paid suffi-cient attention to how the credit and liquiditycharacteristics of these CDOs differed fromthose of more “ordinary” debt instruments,such as corporate bonds. It appears that theseinvestors may have underestimated the mar-ket liquidity risks in these instruments.2
Since many of these structured products arevery complex and relatively opaque—partic-ularly those that are themselves backed byother structured or securitized products—itis often more difficult for investors to deter-mine their direct exposures to various risksthan it is for more traditional debt instru-ments.3 While securitization helps disperserisk to those more willing to bear it, it canalso make it difficult for market participantsto easily understand the full extent of theirexposures to poorly performing underlyingassets and, in turn, the full extent of theircounterparties’ exposure to these same assets.This greater uncertainty about risk exposuresand valuations contributed to a greater degreeof indiscriminate selling of structured productsand to lower market liquidity.
Assigned to Structured Products
2. See Gravelle (2007) for more on the mispricing ofliquidity.
3. See Barker (2007) for a discussion on the opacity andcomplexity of CDOs.
Financial System Review – December 2007
Table 1
Chronology of Events
Source: BIS Quarterly Review (September 2007) and Bank of Canada
Date Event
15 June Moody’s downgrades the ratings for 131 MBSs backed
20 June Two Bear Stearns funds that invested in structured pro
10–11 July S&P places US$7.3 billion of MBSs backed by residenbacked CDO tranches on a downgrade review and do
30 July–1 August Germany’s IKB warns of losses related to U.S. subprimfacilities that IKB provided to an IKB-backed ABCP cobailout of IKB is announced.
1 August Coventree, the largest third-party sponsor of ABCP inexposure of its ABCP conduits to U.S. subprime-mort
6 August American Home Mortgage Investment Corp files for Cone of its funding conduits.
9 August • BNP Paribas freezes redemptions for three funds,environment.
• Bank of Canada issues statement on its readiness
13 August Coventree announces that it could not roll over its ma
15 August Bank of Canada announces a temporary expansion o
16 August The Montreal Proposal is announced in which majorconverting their holdings of ABCP into floating-rate n
17 August The U.S. Federal Reserve cuts its discount rate by 50 b
6 September Bank of Canada restores standard terms for SPRA.
18 September The U.S. Federal Reserve cuts its discount and target fe
15 October Participants involved in the Montreal Proposal annou
22 November The chairman of the Pan-Canadian Committee of Invshould be completed by the end of March 2008.
Chart 4 Yield Levels on Corporate Bonds
%
2
4
6
8
10
12
14
16
2
4
6
8
10
12
14
16
2002 2003 2004 2005 2006 2007
Canada investment-gradeU.S. investment-gradeU.S. high-yield
Source: Merrill Lynch
Rates for overnight loans in the United States,Europe, and Canada also moved above the re-spective target policy rates. As a result, severalcentral banks, including the Bank of Canada,moved quickly to provide significant amountsof overnight liquidity to their financial systemsto keep the overnight rates close to policy targetrates (Box 3). Shortly after mid-August, condi-tions in the overnight markets began tostabilize.
In Canada, between mid-September and earlyNovember, there were an increasing numberof transactions at longer maturities in short-term credit markets, including the market forbankers’ acceptances, and spreads narrowed.More recently, credit conditions have again be-gun to deteriorate. Thus, spreads on short-termcredit market instruments remain elevated,both in Canada and abroad. Further news ofsignificant losses and writedowns amongfinancial institutions, funds, and some mono-line insurers and reinsurers, announced inOctober and November, has served to prolong
11
by U.S. subprime-mortgage debt.
ducts backed by subprime debt are reported to be about to fail.
tial mortgages on a negative ratings watch. Moody’s places 184 mortgage-wngrades US$5 billion.
e-mortgage debt, and its main shareholder (KfW) assumes the liquiditynduit highly exposed to U.S. subprime loans. On 1 August, a US$3.5 billion
Canada, issues a statement to its selling group regarding the extent of thegage debt.
hapter 11 bankruptcy, leading to a term extension on outstanding ABCP by
citing an inability to appropriately value them in the current market
to provide liquidity to support the efficient functioning of financial markets.
turing ABCP conduits, and third-party ABCP funding effectively stops.
f the list of securities eligible for SPRA transactions.
participants in third-party ABCP agree to a 60-day standstill to work onotes that match the maturity of the underlying assets.
asis points, to 50 basis points above the target federal funds rate.
deral funds rates by a further 50 basis points.
nce an extension of the standstill agreement until 14 December.
estors in third-party ABCP announces that the complex restructuring process
Developments and Trends
The financial market turbulence that began in Au-gust caused a sharp decrease in liquidity in marketsfor short-term funding, including the overnight in-terbank market (where banks provide unsecuredovernight loans to each other) and the overnight re-purchase market. This deterioration reflected percep-tions of increased counterparty risk, combined withprecautionary hoarding of funds by financial institu-tions. This box provides a brief summary of how theBank of Canada implements monetary policy by in-fluencing liquidity in the overnight market and howit has addressed the needs of Canadian short-termcredit markets during this period of market stress.
To meet the increased demand for overnight liquid-ity, the Bank of Canada used the standard tools itemploys for implementing monetary policy. TheBank’s monetary policy implementation frameworkcentres on keeping the overnight rate close to itstarget.1 The Bank’s primary influence on the over-night rate is through its 50-basis-point operatingband.2 To reinforce the target when there is devia-tion in the overnight rate, the Bank uses open marketbuyback operations. If the overnight rate is generallytrading above the target rate intraday, the Bank willintervene with special purchase and resale agreements(SPRAs). If the overnight rate is generally tradingbelow the target rate, the Bank will intervene withsale and repurchase agreements (SRAs). The Bank canalso adjust the targeted level of settlement balancesabove or below the typical $25 million setting. SPRAsare routinely conducted around month-, quarter-,and year-end periods, and when large payment flowsare going through the system.
In the initial stages of the market “dislocation,”starting on 9 August, the increased demand for li-quidity caused the rate on overnight collateralizedloans to move well above the target overnight rate(Chart A). This market response was not unique toCanada. In the United States and Europe, the over-night interbank rate also moved significantly abovethe respective target policy rates. In this circum-stance, the Bank of Canada engaged in multiplerounds of SPRAs, providing overnight funds to theprimary dealers at the target overnight rate in ex-change for Government of Canada securities. TheBank also increased the level of settlement balancesto $500 million beginning 15 August. Since thattime, the target level of settlement balances hasfluctuated (between $25 million and $500 million)in response to the occasional upward pressure on thecollateralized overnight lending rate.
After being well above target in the first two days ofthe market dislocation, the overnight rate fell below
1. See C. Reid, “The Canadian Overnight Market:Evolution and Structural Changes.” Bank of CanadaReview (Spring 2007): 15–29 and <http://www.bankofcanada.ca/en/lvts/lvts_primer_2007.pdf>.
2. The Bank of Canada pays 25 basis points below itstarget overnight rate on settlement balances left over-night and charges 25 basis points above the targetrate on overdrafts.
Box 3
The Bank of Canada’s Activity in
12
target. The broader money market remained understress, however. On 15 August, the Bank respondedby temporarily expanding the list of securities ac-ceptable for SPRA transactions to include those secu-rities that are already eligible as collateral for theStanding Liquidity Facility. Subsequently, on6 September, the Bank of Canada restored the stan-dard terms for SPRA, accepting only Government ofCanada securities, since it judged that the temporaryexpanded facility was no longer required.
In late September and early October, the rate ofcollateralized loans tended to set slightly higher thanthe target overnight rate. The Bank responded withhigher levels of settlement balances and the use ofmultiple rounds of SPRAs. After a period of relativestability in late October and early November, therehas more recently been some occasional upwardpressure on the overnight rate, requiring the use ofSPRAs.
Overall, the Bank relied on the standard tools it em-ploys for implementing monetary policy in provid-ing overnight liquidity during the credit marketdislocation. This framework has generally served theBank of Canada well in supporting the efficient func-tioning of capital markets. It has become evident,however, that the statutory framework within whichthe Bank conducts these operations needs to be re-viewed. The Bank of Canada Act defines the range ofsecurities that the Bank may buy and sell in its openmarket buyback operations. Some of the powers tobuy and sell securities are outdated and open to dif-fering interpretations as to how broad a range of se-curities may be accepted by the Bank for its buybackoperations in situations such as that which devel-oped in August. The Bank is therefore reviewingthese powers to determine whether amendments tothe Act may be needed to clarify the types of securi-ties it may buy and sell in various situations, so thatit can respond effectively to market conditions.
Chart A Overnight Market Conditions
$ millions %
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
4.25
4.30
4.35
4.40
4.45
4.50
4.55
4.60
4.65
2007
4.70
August September
* CORRA: Canadian overnight repo rate averageSource: Bank of Canada
October November
CORRA* (right scale)Target for the overnight rate (right scale)Target settlement balances (left scale)SPRAs (left scale)
the Overnight Funding Market
Financial System Review – December 2007
Chart 5 Spreads between 3-Month LIBOR andOvernight Index Swaps
2 January 2006–22 November 2007Basis points
Source: Bloomberg
-25
0
25
50
75
100
-25
0
25
50
75
100
CanadaUnited StatesEuro zone
2006 2007
stressed market conditions. These more recentannouncements of losses and writedowns haveled to declines in major equity markets.
References
Armstrong, J. and J. Kiff. 2005. “Understandingthe Benefits and Risks of Synthetic Collat-erallized Debt Obligations.” Bank ofCanada Financial System Review (June):53–61.
Barker, W. 2007. “The Global Foreign ExchangeMarket: Growth and Transformation.”Bank of Canada Review (Autumn): 3–12.
Gravelle, T. 2007. “Bank of Canada Workshopon Derivatives Markets in Canada andBeyond.” Bank of Canada Review(Autumn): 37–45.
Kamhi, N. and E. Tuer. 2007. “HighlightedIssue: Asset-backed commercial paper:Recent trends and developments.” Bankof Canada Financial System Review (June):24–27.
Kiff, J. 2003. “Recent Developments in Marketsfor Credit-Risk Transfer.” Bank of CanadaFinancial System Review (June): 33–41.
Kiff, J. and R. Morrow. 2000. “Credit Deriva-tives.” Bank of Canada Review (Autumn):3–11.
Toovey, P. and J. Kiff. 2003. “Developments andIssues in the Canadian Market for Asset-Backed Commercial Paper.” Bank ofCanada Financial System Review (June):43–49.
Highlighted Issue
The market for Canadian asset-backed commercial paper,revisited
Prepared by Nadja Kamhi and Eric Tuer
Since the publication of the June 2007 FSR,which featured a discussion of the recent devel-opments in the market for Canadian asset-backed commercial paper (ABCP) by Kamhiand Tuer, global credit markets, including thiscommercial paper market, have come understress. This Highlighted Issue updates that
13
Developments and Trends
Chart 6 Composition of Canadian Money Market(July 2007)
Municipal + provincialtreasury bills$15 billion
Asset-backedcommercial
paper (ABCP)$116 billion
Bankers’acceptances$56 billion
Corporatecommercial
paper$54 billion
Federaltreasury bills$119 billion
Non-bank-sponsored
ABCP$35 billion
Bank-sponsored
ABCP$81 billion
Source: Bank of Canada and DBRS
discussion, outlining the series of events andfactors that led to the disruption in the non-bank-sponsored segment of the Canadian ABCPmarket.
BackgroundAt the end of July 2007, $116 billion of ABCPwas outstanding in Canada, which is approxi-mately one-third of the total money market(Chart 6). Of this, about $81 billion was ABCPsponsored by the major Canadian commercialbanks who also provide the backstop liquidity.The remaining $35 billion was non-bank-spon-sored or third-party ABCP. Non-bank or third-party sponsors are independent, non-bank-affiliated firms that acquire assets in order tostructure conduits backed by liquidity facilitiesprovided by large commercial banks, usuallyforeign based. The majority of assets in theseconduits have been CDOs (Kamhi and Tuer2007). In fact, off-balance-sheet vehicles, suchas ABCP conduits, have been a dominant sourceof demand for CDOs globally. ABCP conduitsthat purchase longer-term CDOs typically havelower funding costs (through the issuance ofshort-term debt in the commercial paper mar-ket) than the returns on the underlying assets.Funding a portfolio of longer-duration assetswith short-term paper, however, creates theneed for programs that continually “roll over”the notes and exposes ABCP programs to a con-siderable “funding mismatch” risk.
Global conditionsOver the past several years, a search for higheryields and some complacency with respect torisk on the part of Canadian and global fundmanagers became prevalent across credit mar-kets, partly because of the long time span sinceany class of assets was subject to a significant ad-verse credit event. This complacency was mostevident in the market for ABCP backed by com-plex structured products. Reassured by the highcredit ratings assigned to the paper, investorswere willing to finance these conduits despitethe lack of transparency with respect to the as-sets that the conduits were funding and otherrisks associated with structured financial instru-ments.10 As concerns mounted regarding theunderperformance of U.S. subprime-mortgagesecurities, market participants found it difficult
10. These risks include illiquidity, opaqueness, and valu-ation difficulties (Barker 2007).
14
Financial System Review – December 2007
to assess their direct and indirect exposure tothis market or the exposure of their counter-parties. Although concerns initially centred onmortgage-backed securities and ABCP backedby U.S. subprime mortgages and CDOs, inves-tors quickly stopped discriminating betweenthe different types of ABCP programs. Thisstemmed from concerns that the relatively illiq-uid structured products and their related valua-tion problems would generate further mark-to-market losses; from a loss of investor confidencein the credit ratings given to these instruments;and from a worldwide retrenchment from risktaking that came to the forefront with the freez-ing of three BNP Paribas investment funds on9 August.
As maturities on ABCP came due and were notrenewed (i.e., rolled), designated liquidity pro-viders had to be called upon to extend fundingto ensure timely repayment of maturing paper.In Canada, the non-bank-sponsored segment ofthe ABCP market was especially affected, sincesome liquidity providers refused to providefunding, arguing that the conditions of a gener-al market disruption (GMD) that would havetriggered their involvement had not been met.11
Non-bank-sponsored ABCPAmid deteriorating conditions in the globalmarket for ABCP, Coventree, the largest sponsorand administrator of non-bank ABCP conduitsin Canada, issued a public announcement on13 August stating that it could not roll its out-standing ABCP and was thus unable to continueto finance its conduits. Many providers ofliquidity to non-bank-sponsored ABCP con-duits refused to extend the liquidity supportrequested, since their obligation was limited toconditions of a general market disruption.Those banks, largely foreign based, argued thatthe conditions in the market did not constitutea GMD because commercial paper could still beissued in spite of severe liquidity problems. Atthis point, trading in the market for non-bank-sponsored ABCP came to a halt. In an effort torestore confidence and liquidity, major finan-cial market players involved in the market fornon-bank-sponsored ABCP reached a standstill
11. The limitations of a GMD condition as a trigger forliquidity were discussed in the June 2003 (p. 45)and June 2007 (p. 25) issues of the FSR. These limi-tations caused U.S. rating agencies to refuse to rateABCP backed by GMD liquidity provisions.
agreement that became known as the MontrealProposal.12 The agreement incorporated a com-mitment to hold the non-bank-sponsoredABCP, while conduit sponsors agreed not todraw down any liquidity facilities until a restruc-turing solution was reached. The long-termsolution would involve restructuring the financ-ing into floating-rate notes (FRNs) that alignwith the various maturities of the underlyingassets in the conduits. This would effectivelyeliminate the need for liquidity facilities. Workon the agreement is ongoing. The target dateto present a restructuring plan to noteholders,originally set for 15 October, has since beenextended to 14 December 2007.13 It has beenannounced that the restructuring process shouldbe completed by the end of March 2008.
Bank-sponsored ABCPProblems were not restricted to non-bank-spon-sored ABCP programs. Bank-sponsored ABCPhas also been affected by concerns regarding ex-posure to U.S. subprime mortgages and the ro-bustness of its liquidity facilities. Starting inearly August, the yields on bank-sponsoredABCP increased dramatically. Nevertheless, un-like non-bank-sponsored ABCP, this marketcontinued to function as major banks pledgedliquidity support for their programs.14 How-ever, ABCP that had previously traded at yieldsjust above CDOR,15 was being offered at 50 to60 basis points above that benchmark near theend of the month. Despite these attractiveyields, investor demand for ABCP remainedhesitant, particularly for maturities greater thanone month, forcing the sponsoring banks to in-ventory the unsold paper. CDOR has declinedfrom its late August peak by around 35 basispoints. However, yield spreads on ABCP remainat about 50 to 60 basis points above CDOR.
12. See <http://documentcentre.eycan.com/pages/main.aspx?SID=35> for information concerningMontreal Proposal developments.
13. See <http://documentcentre.eycan.com/eycm_library/Canadian%20Commercial%20Paper/English/Media%20Releases/ABCPPressRelease15Oct07.pdf>for the press release.
14. See Bank of Canada press release dated 21 August2007.
15. CDOR, the Canadian Dollar Offer Rate, is the interestrate for Canadian bankers’ acceptances and is deter-mined daily.
15
Developments and Trends
After a sharp run-up in August and September,bank inventories of ABCP have come downmoderately. Total bank-sponsored ABCP out-standing declined to roughly $84 billion inOctober from $86 billion in August.
Changes to liquidity facilities standardsIn reaction to these events, all major Canadianbanks have agreed to replace GMD-style facili-ties with global-style liquidity facilities. As dis-cussed in Kamhi and Tuer (2007) and in Tooveyand Kiff (2003), global-style liquidity facilities,which are the standard in the United States andEurope, are available under a much broader setof conditions than a GMD. The use of global-style liquidity facilities will likely lead U.S. rat-ing agencies to start rating these CanadianABCP programs in the future.16
In January 2007, Dominion Bond RatingService (DBRS), the only rating agency that hasbeen providing ABCP ratings in Canada, madeglobal-style liquidity facilities a condition forproviding the highest rating to new issues ofABCP backed by structured financial assets(i.e., CDOs) (Kamhi and Tuer 2007). This ledto a substantial reduction in new issuance ofthis type of program. In September 2007, DBRSfurther announced that it would be adopting anew rating framework under which all conduitswill have to have global liquidity facilitystandards if they wish to be assigned the highestrating. Moreover, DBRS stated that ratings onABCP programs issued prior to the Septemberannouncement might be revised unless theyadopt the global liquidity facility standard byDecember 2007 (Loke, Feehely, and Wong2007). Since the announcement was made,virtually all bank-sponsored ABCP has metglobal liquidity facility standards.
References
Barker, W. 2007. “Highlighted Issue: Structuredfinance: The changing nature of creditmarkets.” Bank of Canada Financial SystemReview (June): 20–22.
Kamhi, N. and E. Tuer. 2007. “Highlighted Issue:Asset-backed commercial paper: Recenttrends and developments.” Bank of CanadaFinancial System Review (June): 24–27.
16. Earlier this year, a few Canadian ABCP programssatisfied U.S. rating-agency criteria and were thusrated by them.
16
Loke, H., J. Feehely, and J. Wong. 2007. “DBRSUpdates Criteria for Rating CanadianABCP Programs and Outlines GlobalLiquidity Standard ABCP.” 12 September.Available on DBRS website.
Toovey, P. and J. Kiff. 2003. “Developments andIssues in the Canadian Market for Asset-Backed Commercial Paper.” Bank of CanadaFinancial System Review (June): 43–49.
Payment and settlement systems
During the financial market turmoil in August,the designated Canadian payment, clearing, andsettlement systems, which include the LVTS, CDSX,and CLS Bank, managed their operations well.While a marked increase in transaction volumesin all three designated systems put extra pres-sure on their processing capacity, there was onlyone settlement delay related to capacity pres-sure in August. However, the capacity issue wasdealt with quickly and effectively, and no delaysoccurred on subsequent record-volume days.
The drying up of liquidity in the market forABCP resulted in a number of defaults andextensions of entitlement payments related tosecurities held in CDSX.17 On 13 August 2007,approximately $2 billion in maturities of ABCPheld at CDS was not paid. Issuers of this ABCPthen had to either extend the maturities of thispaper, where this right existed, or leave the ma-turities unpaid. In the event, about $1.6 billionof ABCP was left unpaid, and another $500 mil-lion was extended. CDS was not exposed to anyfinancial risks arising from the unpaid maturi-ties, because CDS does not execute an entitle-ment payment in CDSX unless it is pre-fundedby the issuer’s paying agent. On 14 August, CDStook steps to assist issuers and participantsholding the defaulted paper, including facilitat-ing direct interaction between issuers and par-ticipants, to reach mutually agreeable solutions,and enabled procedures to process partial pay-ments on maturing ABCP. As well, CDS issueddaily bulletins to system participants to provideinformation on unpaid and extended maturi-ties, and on the evolving value of unpaid ABCPheld at CDS.
17. Entitlements include dividends, interest, paymentupon redemption or maturity, and other payments ordistributions to holders of securities. Entitlementsmay be distributed in the form of a money paymentor as a distribution of securities or other property.
Financial System Review – December 2007
Chart 7 Profits of Major Banks
% Can$ billions
Source: Banks’ quarterly financial statements
0
5
10
15
20
25
30
0
1
2
3
4
5
6
1996 1998 2000 2002 2004 2006
Return on equity (left scale)Net income (right scale)
Financial institutions
The major Canadian banks were profitable andwell capitalized before the recent market turbu-lence. In the third quarter of fiscal 2007 (ending31 July), the after-tax profits of the major bankswere at a near-record level of $5.3 billion, withreturns on equity (ROEs) averaging about 22 percent (Chart 7). This trend continued to be drivenby strong loan growth, as well as strength on thewealth-management side. Operations in capitalmarkets continued to make a sizable contribu-tion to earnings, although two banks reportedsignificant trading losses on specific transactions.Loan-loss provisions are at very low levels, al-though some recent data (mainly for consumerand credit card loans) suggest a modest weaken-ing in credit quality.
Prospects for bank earnings in the immediatefuture may be less robust if market-based activ-ity slows. In addition, funding costs, both short-and long-term, have risen and become volatile,which, at least initially, could result in narrowerspreads, as banks’ funding costs are not auto-matically passed on to borrowers.
Given the market turbulence, the major banksused the occasion of their third-quarter profitannouncements to provide an update oncertain exposures in areas highlighted by therecent market events. The banks generally indi-cated minimal or manageable exposure to“pipeline risk”; that is, bridge financing com-mitments that backed loan syndications relatedto merger/acquisition transactions.18 They reit-erated that, for the most part, their exposures tothe U.S. subprime-mortgage market (eitherthrough direct lending or structured transac-tions) and to the third-party-sponsored ABCPmarket were minimal. The major banks also re-ported that their exposures to hedge funds are
18. A study by BMO Capital Markets (“Bridge Anyone?”1 August 2007) concluded that overall bank exposureto this type of activity, particularly in the context ofthe size of their overall balance sheets, was manage-able. The study noted that if every dollar of the esti-mated $14 billion of bridge loan commitments hadbeen drawn at that date, the overall Tier 1 capitalratio for the major banks would have dropped by15 basis points.
17
Developments and Trends
modest, collateralized, and typically well diver-sified by counterparty and investment style.19
In early November, the major banks as a grouppre-announced roughly $2 billion of losses on apre-tax basis for the fourth quarter. These werelargely related to their holdings of ABCP andU.S. subprime securities.
The banks sponsor and provide liquidity supportfor about $130 billion in securitization conduitsissuing ABCP, both in Canada and in other coun-tries. In general, bank-sponsored ABCP has beenmuch less affected by the turmoil than third-party ABCP. Nevertheless, the banks have takenback onto their balance sheets at least some ofthe assets in conduits that they have sponsored.Moreover, they may see a strong demand forcredit from corporations previously able to fundthemselves in capital markets.
With their healthy capital, profitability, and liquid-ity positions, banks had at least some capacity toadd to their assets at the onset of the turbulence.Since mid-August, the banks have been successfulin funding through a number of instruments,such as medium-term notes, subordinated debt,CMHC-insured mortgage-backed securities andmortgage bonds, and covered bonds. Balancesheet data suggest that, as a group, they were ableto raise wholesale deposit funding from both fi-nancial and non-financial sectors.
It should be noted that Canadian banks rely onsecuritization for a relatively small proportionof their funding. Securitized credit accounts forabout 13 per cent of total household and busi-ness credit in Canada. Specifically, it accountsfor about 21 per cent of residential mortgagecredit (of which 83 per cent is through theCMHC-sponsored MBS program, which contin-ues to run smoothly), 18 per cent of consumercredit, and 5 per cent of business credit.
Even though bank share prices have under-performed the market, the Bank of Canada’sindicator for distance to default suggests thatmarkets continue to view banks as financiallyhealthy in the aftermath of the crisis (Chart 8).For example, the average distance to default for
19. Early in 2007, OSFI conducted a review of Canadianbank exposures to hedge funds and concluded thatbanks’ exposures are relatively small and that risk-management practices are adequate (J. Dickson,Remarks to Senate Standing Committee on Banking,Trade and Commerce, 31 January 2007).
18
the major banks has declined only slightly fromits recent high. Simulations suggest that if therecent volatility were to persist for a full year, allelse remaining the same, the distance to defaultfor the major banks would decline to a levelslightly below its historical average, but wouldremain above the trough reached during thetechnology-sector adjustment in the early2000s.20
The major Canadian life and health insurancecompanies reported firm growth in profits inthe third quarter, ending 30 September 2007,with ROEs in the 15 to 19 per cent range. The di-versity of their business lines rewarded the firmswith strong sales posted in both wealth-man-agement and protection products. Both domes-tic and international operations contributed tothe strong profits. Credit quality remains firm infixed-income portfolios. Two of the major lifeand health companies had reported modest ex-posures to the U.S. subprime-mortgage market.
There has been a marked increase in the expo-sure of Canadian financial institutions to Cana-da’s non-residential commercial real estatesector in the past three years.
The exposure of Canadian banks has increasedby 30 per cent to $40 billion over this period,following 10 years in which exposure remainedrelatively stable.21 Despite the large increase,this exposure has remained relatively stable asa proportion of assets (at around 2.2 per cent)and as a percentage of Tier 1 capital (below40 per cent). The exposure of banks to commer-cial real estate outside Canada22 has also beenincreasing—from $8 billion in 2004Q2 to$22 billion in 2007Q2.
20. The high-volatility scenario takes the average of dailydeviations in equity market capitalization (from theone-year mean) as observed over the period from24 July 2007 to 21 November 2007. This average isthen assumed to be constant one year into the future,ceteris paribus.
21. This exposure is calculated as the sum of residentnon-mortgage loans (in Canadian dollars and for-eign currency) to commercial builders and develop-ers, land developers, and real estate operators, plusnon-farm, non-residential mortgage loans for proper-ties in Canada.
22. This includes non-residential mortgages secured byproperty located outside Canada and the sum of non-resident non-mortgage loans to commercial buildersand developers, land developers, and real estateoperators.
Financial System Review – December 2007
Chart 8 Distance to Default for Major Banks
Chart 9 Evolution of Consensus Estimates forAnnual Global Economic Growth*
%
Note: See text for a description of the simulation.Source: Bank of Canada calculation based on data from OSFI and
Thomson Financial Datastream
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
2006
0
10
12
1983 1987 1991 1995 1999 2003 2007
2
4
6
8
Maximum: (Actual) (Simulation)Average: (Actual) (Simulation)Minimum: (Actual) (Simulation)
0
10
12
2
4
6
8
2007
20072008
* This estimate covers over 46 countries. Country weights are determinedusing country GDPs converted at 2006 market exchange rates.
Source: Consensus Economics Inc.
Long-runaverage
Chart 10 U.S. House Prices and Inventory:Existing Homes
Year-over-year percentage change Months’ supply
0
2
4
6
8
10
12
14
16
3
4
5
6
7
8
9
10
11
Source: Office of Federal Housing Enterprise Oversight andNational Association of Realtors
House price (left scale)Inventory (right scale)
2000 2001 2003 2005 2006 20072002 2004
The data on other financial institutions coveronly non-residential mortgages. There has beena steady increase from close to $37 billion inAugust 2002 to over $51 billion in August 2007.Much of the increase is accounted for by thecredit unions. Life insurance companies, whichaccount for over half of the exposure of non-bank financial institutions, have had onlymarginal increases.
The MacrofinancialEnvironment
The international environment
The outlook for global economic growth in2008 has been revised up slightly since June2007 (Chart 9), with an acceleration in emerg-ing Asia offsetting an expected slowing in theadvanced economies.
In the United States, recent developments in themortgage market are expected to prolong theadjustments in the housing sector. Sales of newand existing homes continue to decline, the in-ventory of unsold homes remains elevated, andprice declines are expected to continue (Chart 10).Consumption growth slowed appreciably in thesecond quarter of 2007 and is expected to beweak for the remainder of this year. U.S. GDPgrowth is likely to remain modest in 2007 andthrough 2008.
The turbulence in financial and credit markets,coupled with a further deterioration in thehousing sector, has heightened the risk of anabrupt slowdown in the U.S. economy. The re-cent monetary loosening by the Federal Reserveshould help to mitigate this risk.
Growth in the overseas advanced economies isexpected to slow gradually in response to previ-ous monetary policy tightening and the recentfinancial market turmoil. In the latest consen-sus forecasts, expected growth in 2008 has beenlowered marginally for the euro area and theUnited Kingdom, and is unchanged for Japan.Meanwhile, growth in emerging Asia hasshown stronger momentum, and effects fromthe financial market turbulence are expected tobe modest.
We continue to expect that global imbalanceswill unwind in a smooth and gradual manner.The U.S. current account deficit has narrowedmodestly, and the continued depreciation ofthe U.S. dollar and expectations of more
19
Developments and Trends
Chart 11 Emerging-Markets Sovereign Bond Spread(EMBI+)
Basis points
0
200
400
600
800
1,000
1,200
0
200
400
600
800
1,000
1,200
2000 2002 2004 2006
Source: JPMorgan Chase & Co.
Chart 13 Financial Position of the CanadianNon-Financial Corporate Sector
% %
Source: Statistics Canada
2000 2001 2002 2003 2004 2005 2006 2007-5
0
5
10
15
20
25
0.85
0.90
0.95
1.00
1.05
1.10
1.15
Return on equity (left scale)Debt-to-equity ratio (right scale)
Chart 12 Real GDP Growth: Canada
%
-1
0
1
2
3
4
5
6
7
-1
0
1
2
3
4
5
6
7
2000 2001 2002 2003 2004 2005 2006 2007
Source: Statistics Canada
Year-over-year percentage changeQuarterly growth at annual rates
balanced growth internationally (slower growthin the United States and continued stronggrowth elsewhere) are consistent with a furtherreduction in the deficit. That said, the tradesurpluses of Asian and oil-exporting countriescontinue to expand, underpinned by high oilprices and official intervention to support fixedexchange rate regimes. One development thatbears further monitoring is the willingness offoreigners to hold U.S. assets. The U.S. Trea-sury’s International Capital (TIC) data showednet foreign outflows in August and September,the first such outflows since 1998. Markets willbe looking for any signs of a decline in foreignwillingness to finance the U.S. current accountdeficit, including an increase in the cost of U.S.borrowing.
In comparison with previous market disrup-tions, emerging markets have held up well, ow-ing to stronger macroeconomic fundamentals,ample foreign exchange reserves, and reducedexternal debt (Chart 11). The main risk foremerging markets remains the potential fora global slowdown with reduced demand forexports and commodities.
Canadian developments
Canadian economyEconomic growth in Canada picked up marked-ly in the first half of 2007 (Chart 12). In theOctober Monetary Policy Report, momentumin domestic demand was projected to remainstrong, despite tighter credit conditions. How-ever, net exports were expected to exert a moresignificant drag on the economy in 2008 and2009 than previously expected. Major downsiderisks to the outlook would materialize if theCanadian dollar were to persist above the levelof 98 cents U.S. assumed over the projectionhorizon for reasons not associated withstronger-than-projected demand for Canadianproducts, or if the effect of the weakness in theU.S. housing sector turned out to be greaterthan anticipated.
Corporate sectorThe financial position of the aggregate non-fi-nancial corporate sector was still relatively solidin the third quarter of 2007. Indeed, profitabil-ity picked up, partly owing to higher crude oilprices, and the ratio of debt to equity fell stillfurther (Chart 13).
20
Financial System Review – December 2007
Chart 14 Liquidity Indicators
1.14
1.16
1.18
1.20
1.22
1.24
1.26
1.28
1.30
0.68
0.70
0.72
0.74
0.76
0.78
0.80
0.82
0.84
2000 2001 2002 2003 2004 2005 2006 2007
Current ratio* (left scale)Quick ratio** (right scale)
* The current ratio is calculated as current assets dividedby current liabilities.
** The quick ratio is calculated as current assets minusinventory divided by current liabilities.
Source: Statistics Canada
Chart 15 Indicators of Credit Quality
Standard deviation %
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0
4
8
12
16
20
24
1994 1996 1998 2000 2002 2004 2006
Source: Bank of Canada calculation based on data from ThomsonFinancial Datastream, the Globe and Mail, the Financial Post, andMoody’s KMV. For CCA, all datapoints denoted January 2005 andthereafter are provided by Moody’s KMV; datapoints prior to January2005 continue to be provided by the Globe and Mail and ThomsonFinancial Datastream.
Volatility of return on Canadiancorporate portfolio (left scale)Share of companies withweak financial ratios(right scale)
By and large, firms have continued to haveaccess to credit. Markets for long-term debt re-main open, at least for investment-grade firms;enough bonds were issued in August and Sep-tember to roughly offset maturing issues, andmodest growth in bonds outstanding resumedin October. Non-financial firms have retainedready access to bank lending since the turbu-lence began, and bank credit to the sector grewat a strong pace in August and September. How-ever, there has been some tightening of creditconditions, particularly for large firms. Theweighted average cost of credit (taking intoaccount both market and bank financing)has increased modestly since the end of July.
In general, the non-financial sector seems wellplaced to deal with the tightening of credit condi-tions, given high levels of retained earnings, lowleverage, and high levels of liquidity (Chart 14and Box 4). Although reported liquid assetswould include investments in ABCP, overall,it would appear that the third-party ABCP hold-ings by non-financial firms represent a relativelymodest proportion of the total liquid assets ofthe non-financial corporate sector (which amountto $255 billion). Bank of Canada indicators ofcorporate credit quality—namely, the volatilityof returns on the Canadian corporate portfolio(CCA indicator) and the share of companies withweak financial ratios (microdata indicator)—continue to suggest that overall corporate creditquality remains robust (Chart 15).23 While thelatter indicator remains unchanged from theJune FSR, the former has risen only slightly inrecent months, driven mainly by modestlyhigher volatility of returns in the industrial andutilities sectors.
IndustryWeakness in U.S. demand for housing contin-ued to exert significant adverse effects on the
23. The CCA indicator represents the volatility of market-valued assets in a portfolio consisting of nine broadnon-financial corporate industries. The monthly CCAindicator is based on data up to, and including, Octo-ber 2007. The microdata indicator represents theshare of total assets attributable to companies with acomparatively weak leverage ratio, current ratio, andnet profit margin. The annual microdata indicator iscurrently based on data up to the end of 2006.Detailed descriptions of the CCA and microdata indi-cators can be found in the June 2006 (pp. 43–51) andDecember 2005 (pp. 37–42) issues of the FSR,respectively.
21
Developments and Trends
Since 2003, Canadian firms have been holding asignificantly higher level of liquidity than usual(Chart A). This increase has been attributed to(i) the amount of uncertainty facing some busi-nesses and/or (ii) preparation for future invest-ment. Because these two factors have differentimplications for the economy, it is important tounderstand more about which businesses areholding excess liquidity and why.Consequently, a special question was added tothe Bank’s Business Outlook Survey. From Sep-tember 2006 through June 2007, firms wereasked, “If you are currently holding a level ofliquid assets (cash, deposits, and short-termfinancial assets) that is above normal, what isthe main reason for doing so?” The results arereported below.
Who reported holding excessliquidity?
Of the 392 firms surveyed, 125 (32 per cent)were holding above-normal levels of liquidity,and the phenomenon was relatively wide-spread. Larger firms (37 per cent) were some-what more likely to be holding above-normallevels of liquidity than smaller or medium-sized firms. The manufacturing and trade sec-tors had the lowest incidence of excess cash(29 per cent), while the highest was among pri-mary firms (38 per cent). Ontario (21 per cent)was well below other regions (Table 1).
Why were firms holding excessliquidity?
Reasons most often cited (Table 2) were: plan-ning to undertake a capital expenditure project(32 per cent); planning to undertake a merger oracquisition/financial investment (24 per cent);and currently evaluating possible investmentopportunities (23 per cent).1 In comparison,only 9 per cent of firms holding excess cash saidthey were doing so because of uncertainty aboutthe economic outlook and/or their financial
1. These percentages are based on the results of thelast three surveys, owing to a change in the list ofpossible responses. Since December 2006, 97 of292 firms surveyed (33 per cent) reported holdingexcess liquidity.
Box 4
Special Survey Question on Exc
22
situation, and only 4 per cent said that theywere using liquidity to improve their balancesheets. Although the reasons for holding excessliquidity were fairly evenly spread across regionsand sectors, several interesting differences areworth noting:
• Forty per cent of Western Canadian firmsholding excess liquidity were planning toundertake more capital spending, comparedwith 26 per cent in the rest of the country.
• A large proportion of the 31 firms that citedincreased capital spending as the primary reasonfor holding excess cash were small or medium-sized firms. Only 4 were manufacturers.
• The reasons most often cited by the 31 manu-facturing firms holding above-normal amountsof liquidity were merger/acquisition-relatedactivity and distributions to shareholders.
• Companies planning to undertake mergers/acquisitions were almost exclusively large ormedium-sized.
• Companies holding excess liquidity becauseof economic and financial uncertainty tendedto be small and/or located in Eastern Canada.They were mainly in the manufacturing ortransportation sectors.
• Service sector firms were more likely to beevaluating all possible investment opportunities.
Chart A Liquid Assets of the Non-FinancialCorporate Sector
$ billions %
0
50
100
150
200
250
300
350
400
6.0
6.5
7.0
7.5
8.0
8.5
9.0
9.5
10.0
1988 1992 1996 2000 2004
Level (left scale)As a per cent of total assets(right scale)
Source: Statistics Canada
ess Liquidity
23
Financial System Review – December 2007
Box 4
Special Survey Question on Excess Liquidity (cont’d)
Table 1
Firms Holding Excess Liquidity
Region Per centholdingexcess
liquidity
Industry Per centholdingexcess
liquidity
Firmsize*
Per centholdingexcess
liquidity
National 32 Primary 38 Small 30
Atlantic 41Manufac-
turing 29 Medium 28
Quebec 32 Trade 29 Large 37
Ontario 21 FIRE 33
Prairies 38 CITU 34
BritishColumbia 32 CPBS 33
Note: FIRE represents finance, insurance, real estate and leasing; CITUstands for construction, information, transportation, and warehousing,and utilities; CPBS represents commercial, personal, and businessservices.
* Firm size: Small <100 full-time employees (FTE); medium, 100 to499 FTE; and large >500 FTE
Table 2
Reasons for Holding Excess Liquidity and Firm Distribution
Percentage of firms*
Reason** National East West Goods Services Small firms Mediumfirms
Large firms
Uncertainty about the economicoutlook and/or financial situation 9 13 5 11 8 24 0 7
To improve their balance sheet 4 6 2 4 4 4 0 7
Currently unable to find suitable capitalproject 9 9 9 6 12 8 7 12
Mergers and acquisitions/financinginvestment 24 22 26 26 22 12 28 28
Currently evaluating all possibleinvestment opportunities 23 28 16 15 30 16 28 23
Planning more capital expenditures 32 26 40 34 30 24 34 12
Planning to distribute dividend orequity buyback to shareholders 19 20 16 19 18 20 7 26
Other/firm-specific reason 12 15 9 13 6 20 14 7
* Reflecting the learning that took place in the September 2006 survey, the list of possible responses was adjusted for the December 2006 round. Thesepercentages are based on the final three quarterly surveys. Since December 2006, 97 out of 292 surveyed firms reported holding excess liquidity.Firm size: Small <100 full-time employees (FTE); medium,100 to 499 FTE; and large >500 FTE
** Those conducting the survey were asked not to read this list to the firms and to record all applicable responses. Twenty per cent of firms cited morethan one reason for holding excess liquidity.
Developments and Trends
Chart 16 Real Prices for Housing in Canada*
Year-over-year percentage change
%
-6
-4
-2
0
2
4
6
8
10
12
-6
-4
-2
0
2
4
6
8
10
12
1994 1996 1998 2000 2002 2004 2006
New housingExisting housing
* Deflated by CPISource: Royal LePage, Statistics Canada, and Bank of Canada
calculations
Chart 17 Real Prices for Existing Houses by City*
Year-over-year percentage change
%
-20
-10
0
10
20
30
40
50
60
-20
-10
0
10
20
30
40
50
60
1994 1996 1998 2000 2002 2004 2006
* Deflated by CPISource: Royal LePage, Statistics Canada, and Bank of Canada
calculations
MontréalTorontoEdmontonCalgaryVancouver
profitability of Canada’s forest products indus-try in the first three quarters of 2007. On theother hand, profitability in Canada’s auto man-ufacturing industry did recover somewhat overthis period from a weak financial performancein 2006. However, a slowdown in U.S. growth,combined with the rapid appreciation of theCanadian dollar, would contribute to a furtherdeterioration in the financial positions of theseparticular industries and could possibly resultin a further major restructuring of their opera-tions. But many firms in the automobile manu-facturing industry had significantly bolsteredtheir liquidity prior to the onset of the turbu-lence in global financial markets by securinglong-term funding.
Profitability in many Canadian manufacturingindustries other than autos and forest productshas also been lower than normal in recent years,chiefly owing to the appreciation of the Canadi-an dollar and strong competition from overseasproducers. A prolonged slowdown in thegrowth of U.S. consumer spending (togetherwith adverse effects from the ongoing rise in theCanadian dollar) would likely have a majornegative impact on the financial positions of abroader range of Canadian manufacturers ofconsumer products (such as food and beverages,furniture and household appliances, and printingand publishing). In addition, many Canadianlivestock producers have been experiencingsignificant losses in recent months as a result ofthe sharp rise in the Canadian dollar and theappreciable increase in feed costs.
Although a number of manufacturing compa-nies continue to face serious financial risks, itremains unlikely that their problems wouldhave significant adverse effects on the Canadianfinancial system, since the exposure of Canadi-an banks to these industries remains limited.
House pricesThe Canadian housing market is much healthierthan that in the United States.
House prices across Canada have continued toincrease, albeit at a slower pace, fuelled by sus-tained income growth, low unemploymentrates, robust consumer confidence, and stillrelatively low interest rates (Chart 16). TheCanadian housing market remains character-ized by regional differences, with stronger priceincreases in Western Canada—where stronglabour markets continue to support housing
24
Financial System Review – December 2007
Chart 19 Housing Supply
Thousands
100
120
140
160
180
200
220
240
260
100
120
140
160
180
200
220
240
260
1994 1996 1998 2000 2002 2004 2006
Housing starts (12-monthmoving average)New listings of existing homes
Source: CMHC and MLS
Chart 18 Recently Completed Unoccupied Dwellings
As a percentage of population
%
0.0
0.5
1.0
1.5
2.0
2.5
0.0
0.5
1.0
1.5
2.0
2.5
1994 1996 1998 2000 2002 2004 2006
Source: Statistics Canada and Bank of Canada calculations
CanadaMontréalTorontoCalgary
Chart 20 Household Sector: Indebtedness Indicators
% %
90
95
100
105
110
115
120
125
130
135
6.0
6.5
7.0
7.5
8.0
8.5
9.0
9.5
10.0
10.5
1994 1996 1998 2000 2002 2004 2006
Source: Statistics Canada, Ipsos Reid, and Bank of Canada calculations
Debt-to-disposable-income ratio(left scale)Debt-service ratio (right scale)
demand—than in Central Canada (Chart 17).The growing use of mortgages with longer am-ortization periods and, to a lesser extent, otherinnovations such as near-prime mortgages mayhave added to housing demand. New mortgageinsurance products have also likely contributedto rising housing demand, since they allow newborrowers into the mortgage market.24Anecdot-al evidence suggests that there has been somespeculative buying and construction in certainlocal housing markets or segments of markets.
The proportion of unoccupied dwellings inmost cities remains below historical averages—and well below the peaks of the early to mid-1990s—suggesting that a major widespreadreversal in house prices is most unlikely(Chart 18).25 However, there should be someslowdown in the pace of increase in house pric-es as a result of the continued high level of sup-ply in both the market for new houses and theresale market (Chart 19) and a recent deteriora-tion in the affordability of home ownership.
Overall, recent indicators support the view thatthe Canadian housing market does not pose amajor threat to the stability of the Canadianfinancial system, although there may be risksof a decline in house prices in particular localmarkets.
Household sectorDisposable income continued to increase ata solid pace over the first half of 2007. Butsince households continued to accumulate debtat a faster pace, the debt-to-income ratio hasrisen further, reaching 128 per cent in 2007Q2(Chart 20). This increase in indebtedness hasbeen accompanied by higher interest rates. Asa result, the household aggregate debt-service ratio(DSR) rose further, to 7.3 per cent in 2007Q2, upfrom 7.15 per cent in 2007Q1 (Chart 20).26 Theaggregate DSR will likely continue to increase,since Canadian households continued to accu-mulate debt at a solid pace in the third quarter,
24. See the June 2007 FSR, pp. 23–24, for a list of prod-uct innovations in the mortgage insurance market.
25. There could, however, be imbalances in certain seg-ments of local housing markets. The increase in theproportion of unoccupied multiple dwellings inMontréal, for instance, suggests some possible down-ward pressure on condominium prices there.
26. The aggregate debt-service ratio includes only interestpayments on debt. For details about the estimation ofthe aggregate DSR, see Box 2 in the December 2006 FSR.
25
Developments and Trends
Chart 21 Household Sector: Financial StressIndicators
% %
0.24
0.26
0.28
0.30
0.32
0.34
0.36
0.38
0.40
0.42
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1994 1996 1998 2000 2002 2004 2006
0.44
Personal bankruptcies* (left scale)Residential mortgage loans in arrears3 months or more** (right scale)
* As a percentage of population aged 20 and over** As a percentage of total mortgage loans outstandingSource: Statistics Canada and Bank of Canada
including during the period of financial marketturbulence, and effective borrowing costs haveincreased over the same period.27
Despite rising indebtedness and a rising debt-service ratio, the household sector appearssound, as illustrated by aggregate indicators ofhousehold financial stress. Mortgage loans inarrears have remained at historically low levels,and the personal bankruptcy rate increasedmodestly in the third quarter of 2007 (from an11-year low) (Chart 21).
Nevertheless, the proportion of vulnerablehouseholds —defined as households with a DSRabove specific thresholds—and the proportionof debt they hold have recently increased, andare currently above their 1999–2006 average.28
(See Table 2 on p. 29.) However, they remainbelow the peaks observed in 2001.
Overall, the financial position of the Canadianhousehold sector does not seem to pose a threatto the stability of the Canadian financial systemat present, although some households may be-come more vulnerable to negative shocks overtime. The following Highlighted Issue suggeststhat the vulnerability of the household sectorcould rise if indebtedness and/or interest ratescontinue to increase.
Highlighted Issue
Stress testing the Canadianhousehold sector using microdata
Prepared by Ramdane Djoudad andVirginie Traclet
Household credit accounts for 62 per cent of thetotal loan exposure of the Canadian bankingsector, 30 per cent of its total assets, and 710 percent of Tier 1 capital. Consequently, assessing thefinancial health of Canadian households is animportant part of our assessment of risks in thefinancial system. Past analysis, based on micro-data indicators, concluded that Canadian house-holds are currently in relatively good financial
27. See Technical Box 3 in the October 2007 MonetaryPolicy Report, p. 20.
28. We use two vulnerability thresholds for the DSR:23 per cent and 40 per cent. For information on howthese thresholds were chosen, see the December 2006FSR, pp. 15–16.
26
Financial System Review – December 2007
health (Faruqui, Lai, and Traclet 2006). Forward-looking simulations of the aggregate householddebt-service ratio (DSR), however, suggestedthat the Canadian household sector will be-come more vulnerable to shocks over time. (SeeJune 2007 FSR, p. 14.) Projections of the aggre-gate DSR can illustrate what might happen tothe average household under hypothetical sce-narios; however, they cannot indicate whatmight happen to the proportion of vulnerablehouseholds, since aggregate data mask informa-tion about the underlying distribution of debt.This Highlighted Issue supplements our pastanalysis by using a stress-testing exercise to as-sess the effect that various hypothetical scenari-os might have on the distribution of the DSRand on the proportion of vulnerable house-holds. This analysis supports the conclusion inthe June 2007 FSR: The household sector is be-coming more vulnerable to shocks as indebted-ness continues to increase. Moreover, a higherproportion of households could become vul-nerable to negative shocks if interest rates wereto rise significantly.
Data and methodologyOur focus here is the distribution of the DSR,which we calculate as total debt payments (in-terest and principal payments on debt) dividedby household gross income for the householdsthat have debt.29 We simulate the impact thatrising indebtedness and/or rising interest rateswould have on the distribution of the DSR overtime under different stress-test scenarios. We arealso interested in the proportion of householdsthat have a DSR above some critical level indic-ative of vulnerability and in the proportion ofdebt held by these households. At this time, it isnot clear what this critical level is; research isongoing to identify this threshold. We use twothresholds that have been used in past FSRs:23 per cent and 40 per cent. The 40 per centthreshold is a rule of thumb used by financialinstitutions in Canada to assess whether or nota loan should be granted. (Note that empirical
29. This measure of the DSR differs from the aggregateDSR, which focuses on interest payments only, is cal-culated for all households—whether or not they havedebt—and is based on personal disposable income.Thus, the results of the simulations presented hereare not directly comparable to those done with theaggregate DSR in the June 2007 FSR. Data used herecome from the Canadian Financial Monitor (CFM).For more details, see December 2006 FSR, p. 14.
evidence indicates that this is not a hard-and-fast rule.)
A number of assumptions are required to un-dertake this stress-testing exercise.30
First, it is assumed that all components ofnon-mortgage consumer debt, except creditcard debt, are at a variable interest rate, which isconsistent with stylized facts for consumer cred-it.31,32 Consumer debt payments include aninterest rate component and a principal compo-nent. For each component of consumer debt,the historical interest rate is taken directly fromthe dataset as reported by households. For thesimulation period, we assume that the interestrate on each component of consumer debtmoves by the same amount as the prime rate(which has moved closely with movements inthe target overnight rate).
Second, the interest rate on fixed-rate mortgagesis calculated as the sum of the overnight rate, aterm premium, and a risk premium. For eachmaturity, the risk premium is proxied by the2000Q1–2007Q2 average of the difference be-tween the mortgage rate and the yield on gov-ernment bonds of the same maturity.33 The riskpremiums remain unchanged during the simu-lations.34 In the simulations, the term premi-ums rise from their current level to their averagehistorical yield spread for each maturity withinthe period over which the overnight rate in-creases under the scenario of rising interestrates, while they remain unchanged at theircurrent level under the scenario of unchangedinterest rates.
30. The methodology used here will be discussed ingreater detail in an upcoming Bank of Canada Reviewarticle.
31. Consumer debt in the CFM includes personal loans,personal lines of credit, vehicle loans, and credit carddebt.
32. Because of data constraints, it is assumed that allhouseholds make payments on credit card debt equalto 2 per cent of their current balance each month.This is the minimum payment required by mostcredit card companies.
33. Mortgage maturities in the CFM are 6 months, 1 year,2 years, 3 to 4 years, 5 years, 7 years, and 10 years andmore.
34. This may create a downward bias in the simulatedDSR: As indebtedness increases further, financialinstitutions might increase individual risk premiumsto compensate for the potentially higher riskiness ofsome individual borrowers.
27
Developments and Trends
Third, it is assumed that, for households thathave variable-rate mortgages, mortgage paymentsare not affected by changes in the interest rate.35
Finally, a certain proportion of households areassumed to renew their mortgages each year.For each mortgage term, this proportion isequal to one divided by the term of the mort-gage; e.g., 20 per cent of households that have a5-year mortgage renew each year.36 For simplic-ity, the distributions of mortgages by type (vari-able, versus fixed-rate) and by term are assumedto remain similar to the distributions in 2006.37
The simulation period is 2007Q3–2010Q2.
Impact of rising indebtedness on thedistribution of the debt-service ratioTo assess the impact of a rising debt-to-incomeratio on the distribution of the DSR, we consid-er a debt-to-income scenario similar to thatused in the simulations of the aggregate DSR inthe June 2007 FSR. This is done in two steps.First, we develop an aggregate scenario that setsthe assumptions for total growth in debt and in-come. Second, we model what is happening toindividual households in order to take hetero-geneity among households into consideration;this is implemented in a simple way.38
35. In practice, when interest rates increase, payments onmost variable-rate mortgages do not increase, buttheir composition changes, with a decrease in theproportion allocated to the principal and an increasein that allocated to interest. Payments would increasewhen mortgages are renewed at higher rates. Becauseof data constraints, it is not possible to determinewhen variable-rate mortgages are due for renewal.Consequently, we keep payments on these mortgagesunchanged over the simulation period. This createsa downward bias in the simulated DSR, but othercalculations suggest that this bias is relatively small.
36. One hundred per cent of households with a 6-monthor 1-year mortgage renew each year.
37. In practice, if interest rates were higher at the time ofrenewal, some households would likely switch fromvariable- to fixed-rate mortgages to limit the increasein their DSR. It is difficult to include this in our simu-lations, since it would require arbitrary assumptionsabout which households would switch (and to whichterm) and which would not.
38. Heterogeneity among households could be takeninto account with a model that endogenizes house-hold borrowing decisions. But this would requirenumerous underlying assumptions and behaviouralresponses, for which empirical evidence is quitelimited.
28
The aggregate scenario assumes that total con-sumer debt and mortgage debt continue to in-crease at their average annual growth rates overthe 2000Q1–2007Q2 period.39 But to allow forthe different preferences and initial debt levelsof households, the magnitude of the increase indebt is not the same for all households. Forthose with relatively high DSRs, debt increasesat a slower pace than for those that have lowerDSRs. This is consistent with the rule of thumbused by financial institutions to assess whetheror not a loan should be granted. The aggregatescenario also assumes that aggregate incomecontinues to increase at a trend rate of 5 percent.40 Here again, to account for heterogeneity,it is assumed that income increases at differentrates for different households. Based on thesetwo assumptions, the increase in the aggregatedebt-to-income ratio over the simulation periodis similar to its recent trend.
To study the impact of such an increase in thedebt-to-income ratio on the DSR distribution,we use a scenario in which the overnight inter-est rate remains unchanged at its current level(4.5 per cent). Under this scenario, the meanDSR would increase by 0.94 percentagepoints—to 16.6 per cent—over the simulationperiod.41
The distribution of the DSR would also changeover time. As illustrated in Chart 22, the propor-tion of households with a high DSR would in-crease. There would be an increase in theproportion of households above both of thethresholds typically used to assess vulnerabilityand in the proportion of debt owed by thosehouseholds (Table 2). For example, using the40 per cent threshold, rising indebtednesswould lead to an increase in the proportion of“vulnerable” households and the debt they oweto levels higher than their 1999–2006 averagesand higher than in 2001 (Table 2).
39. It is also assumed that all components of consumerdebt increase at the same pace as total consumer debt(8 per cent annually) and that all components ofmortgage debt increase at the same pace as totalmortgage debt (6 per cent annually).
40. Household gross income—the income measure inthe CFM—is assumed to increase at the same trendpace as household disposable income.
41. This compares with a simulated 0.70 percentage pointincrease in the aggregate DSR over the same period(June 2007 FSR, p. 14–15).
Financial System Review – December 2007
Chart 22 Impact of Rising Debt-to-Income Ratio onthe Distribution of the Debt-Service Ratio
Density
60-10 0 10 20 30 40 500.00
0.01
0.02
0.03
0.04
0.05
0.06
0.00
0.01
0.02
0.03
0.04
0.05
0.06
Source: Ipsos Reid and Bank of Canada calculations
DSR, 2006H2–2007H1DSR, 2010Q2
DSR
Table 2
Impact of Rising Indebtedness on VulnerableHouseholdsa
Proportionof
householdswith
DSR>23%b
Share oftotal debtowed by
householdswith
DSR>23%
Proportionof
householdswith
DSR>40%b
Share oftotal debtowed by
householdswith
DSR>40%
1999–2006average 26.13 43.87 3.33 6.28
2001c 28.47 47.84 4.04 7.83
2006H2-2007H1 23.35 43.96 3.45 7.40
Stress-test results
2008Q2 27.38 48.58 4.17 8.46
2009Q2 30.14 52.20 5.46 9.96
2010Q2 32.07 52.12 6.46 11.46
a. Historical numbers for vulnerable households and vulnerable debt havebeen revised compared with the numbers reported in the December2006 FSR for two reasons. First, credit card debt is now taken into ac-count in the calculations of the DSR, which was not the case previously.Second, the population weights used to calculate the distributions havebeen modified to make the sample more representative of the popula-tion.
b. As a percentage of total households with debtc. We report data for 2001 because the share of debt held by vulnerable
households was at its maximum during the sample period (1999–2006)in that year. This provides a reference point to assess the results of oursimulations.
Impact of rising interest rates on thedistribution of the debt-service ratioTo assess the impact of interest rate changes onthe distribution of the DSR, we consider a stress-test scenario in which interest rates increasesharply. Specifically, the overnight rate risesfrom 4.5 per cent to 6 per cent within fourquarters and remains at this level for the rest ofthe simulation period. This increase is transmit-ted to consumer and mortgage interest rates, asdescribed above. As in the previous scenario,the aggregate debt-to-income ratio continues toincrease at a pace similar to that observed overthe recent past. This combination of rising in-debtedness and rising interest rates can beviewed as unlikely, since it is assumed that debtcontinues to increase at the same pace over thesimulation period, despite significantly higherinterest rates (by comparison, the average over-night rate over the 2000–2007Q1 period was3.50 per cent), whereas higher rates would like-ly be accompanied by some slowing in debtaccumulation.
With such an increase in interest rates andindebtedness, the mean DSR would rise by3.26 percentage points—to 18.89 per cent—over the simulation period. As illustrated inChart 23, the distribution of the DSR wouldshift more than under the previous scenario,with a stronger increase in the proportion ofhouseholds with a high DSR. The proportionof “vulnerable” households and the proportionof debt owed by those households would alsorise more markedly than under the previousscenario (Table 3).
While these results must be viewed with cau-tion, given the biases in the simulated DSRthat are due to some of our assumptions,42 theyremain qualitatively informative: They suggestthat more households would see their ability toweather negative shocks significantly reduced ifboth indebtedness and interest rates were to risemarkedly.
To put into perspective what such an increase inthe proportion of debt held by “vulnerable”households means for the financial system, weconsider a stress-testing exercise using two ex-treme assumptions. First, we assume that 25 percent of the change in the proportion of debt
42. It is difficult to measure the extent of these biases andtheir impact on the simulated DSR.
29
Developments and Trends
Chart 23 Impact of Rising Debt-to-Income Ratio andRising Interest Rates on the Distributionof the DSR
Density
-10 0 10 20 30 40 60
0.00
0.01
0.02
0.03
0.04
0.05
0.06
Source: Ipsos Reid and Bank of Canada calculations
0.00
0.01
0.02
0.03
0.04
0.05
0.06
50
DSR, 2006H2–2007H1DSR, 2010Q2
DSR
Table 3
Impact of Rising Indebtedness and Rising InterestRates on Vulnerable Households
Proportionof
householdswith
DSR>23%
Share oftotal debtowed by
householdswith
DSR>23%
Proportionof
householdswith
DSR>40%
Share oftotal debtowed by
householdswith
DSR>40%
2006H2–2007H1 23.35 43.96 3.45 7.40
Stress-test results
2008Q2 28.56 50.36 4.57 9.05
2009Q2 31.53 54.46 6.36 12.05
2010Q2 34.28 55.49 7.37 13.57
owed by vulnerable households (as definedwith the 40 per cent DSR threshold) during thesimulation period under the second scenariopresented above goes into default. Second, weconsider that the loss-given-default is 100 percent of the defaults.43 Under these extreme as-sumptions, the associated losses for the Canadi-an banking sector would represent 0.5 per centof the total assets in that sector or 11 per centof Tier 1 capital. This suggests that the bankingsector would not be significantly affected byrising vulnerability in the household sector.
References
Faruqui, U., S. Lai, and V. Traclet. 2006. “High-lighted Issue: An analysis of the financialposition of the household sector usingmicrodata.” Bank of Canada FinancialSystem Review (December): 14–17.
43. Actual losses would be much lower, because a 100 percent loss-given-default certainly overestimates theactual loss-given-default on household debt, particu-larly on mortgage debt. In addition, since a largeproportion of mortgages are insured, the lossesassociated with those would be borne by the mort-gage insurers and not the banking sector.
30
Financial System Review – December 2007
Important Financial System Developments
his section of Developments andTrends examines the structural de-velopments affecting the Canadianfinancial system and its safety andefficiency.
Financial Collateral forEligible Financial Contracts
The March 2007 federal budget set out plans toamend Canadian bankruptcy and insolvencylegislation to ensure full protection for financialcollateral arrangements supporting eligible fi-nancial contracts (EFCs) in Canada. Bill C-52,the bill containing the provisions in the federalbudget, was granted royal assent on 22 June2007. The majority of the amendments cameinto force at that time.44
Prior to the amendments, uncertainty existed inCanada as to whether and how quickly creditorscould access financial collateral pledged againstEFCs in the event of the insolvency of a counter-party. The amendments seek to address thesedeficiencies by clarifying collateral-protectionrights and the priority pertaining to them. In ad-dition, they seek to modernize the definition ofEFCs and financial collateral to keep pace withmarket developments.
The use of financial collateral to reduce counter-party risk in the event of default on obligations
44. The section pertaining to EFCs and financial collateralis contained in Part 9 of the bill and provides foramendments to the following acts: Bankruptcy andInsolvency Act, Companies’ Creditors ArrangementAct, the Winding-up and Restructuring Act, theCanada Deposit Insurance Corporation Act, and thePayment Clearing and Settlement Act. Each of theseacts would be amended to provide specific protectionagainst stays on financial collateral for obligationsunder EFCs so that counterparties can realize on theircollateral in a timely manner.
T
has become an increasingly important risk-management tool. Financial collateral is unique,differing from other types of collateral, such asthat used to secure loans for commercial prop-erty or plant and equipment. The legislationrecognizes that it is important that financialcollateral be appropriated quickly, without thedelay associated with court proceedings, shouldthe provider default and go into bankruptcy.Liquidation permits the proceeds to be appliedagainst losses to the creditor quickly, thus min-imizing the risk of potential damage to collateralvalue. The changes should ensure both the moreefficient functioning and the continued develop-ment of the use of collateral in financial trans-actions in Canada, resulting in reduced riskfor market participants and, in the end, a moreefficient financial system.Elimination of WithholdingTax on Interest
The elimination of the withholding tax an-nounced in the 2007 budget takes effect on1 January 2008 for interest paid on or after thisdate between arm’s-length parties. For non-arm’s-length parties, the maximum withhold-ing rate on interest payments between Canadaand the United States will be eliminated in threestages. This initiative will facilitate cross-borderinvestments and is expected to reduce the costsof Canada’s multinational enterprises, increasetheir access to cross-border capital, and contrib-ute to the evolution of the Canadian market-place.
The Mortgage InsuranceMarket
A third private mortgage insurer started opera-tions in the autumn, thus bringing the totalnumber of players in the Canadian mortgage
31
Developments and Trends
insurance market to four. Another private insur-er (that received federal regulatory approval tocommence and carry on business earlier thisyear) is still in the process of receiving provinciallicences to begin to actively market its products.
There has recently been further product innova-tion in the mortgage insurance market. InSeptember, CMHC introduced mortgage loaninsurance for mortgages with no down paymentfor investment properties. This may add pres-sure to housing demand at a time when it isalready high and may also contribute to afurther increase in household indebtedness.
Highlighted Issue
Free trade in securities
Prepared by Karen McGuinness
Global financial markets are changing rapidlyand are becoming more integrated. One impor-tant area in which securities regulation has notkept pace with changes in global markets is thetreatment of direct cross-border securities trans-actions, which most domestic securities regula-tors prohibit in their jurisdictions. An investorwishing to trade a foreign security thus needs togo through a broker from that jurisdiction. Forexample, a Canadian broker must engage a U.S.broker to purchase a U.S. security on behalf of aCanadian client. Free trade in securities is a pol-icy initiative aimed at simplifying cross-bordersecurities transactions and reducing the barriersand costs that investors face when trading for-eign financial instruments. This HighlightedIssue examines some of the major issues relatedto free trade in securities.
Free trade in securities would put in place a pro-cess where nations participating in an accordwould broadly agree on the equivalence of oneanother’s regimes as they apply to brokers andsecurities exchanges. The approach that is likelyto be adopted is based on the concept of mutualrecognition of regulatory regimes. For instance,the regulator of an exchange’s or a broker’s homemarket would be responsible for overseeingtheir activities and for protecting the interestof all investors, regardless of their geographiclocation. Such a framework could permit tradingscreens for a Canadian exchange, such as the TSX,to be placed on a U.S.-based market participant’sdesk, and the U.S. market participant could
32
trade directly on the exchange without the needfor a Canadian-based intermediary. Under thisnew mode of operation, the TSX would likelybe subject to Canadian regulations and wouldbypass the U.S. Security and Exchange Commis-sion’s (SEC) exchange registration regulation(assuming that an agreement on Canada’s regu-latory comparability with the United States is inplace). Such an agreement could also allow aU.S. broker-dealer to directly access Canadianinvestors without going through a Canadianintermediary broker-dealer, which is currentlyrequired. In this new set-up, the U.S. broker-dealer would be registered with the SEC andwould bypass the Canadian registration regula-tion (assuming that an agreement on U.S. regu-latory comparability with Canada is in place).
Efficiency and investor protectionFree trade in securities should improve marketefficiency by making it easier and less costly totrade in non-domestic securities. Simplifying theprocess and, in particular, reducing the numberof intermediaries necessary for conducting across-border securities transaction, shouldreduce the cost of that transaction. Furthermore,the removal of regulatory requirements prohib-iting foreign broker-dealers from entering thedomestic market would expose the Canadiansecurities industry to increased broker-dealercompetition, promoting greater efficiency andreducing the costs that Canadian investors face.The process of reaching an agreement on freetrade in securities and achieving consensus onregulatory comparability would likely result ingreater alignment of disclosure requirements inthe two jurisdictions. From an issuer standpoint,free trade in securities should give Canadianissuers more access to foreign investors, possiblyreducing their capital costs.
In Canada, financial market participants arebroadly supportive of the initiative to pursuefree trade in securities with the United Statesand other G-7 countries. For example, the In-vestment Industry Association of Canada (IIAC)has called for regulatory reform in trade poli-cies, which they see as necessary for Canada tocompete in an increasingly global environ-ment.45 The TSX has argued that free trade insecurities would reduce costs and improve theefficiency of capital markets and allow Canadianmarket participants to grow internationallyby leveraging their strengths (Cowan 2007).
Financial System Review – December 2007
Overall, the more open process to access cross-border securities should benefit both investorsand issuers, since transactions would be lesscumbersome and thus likely less costly.
The principal concern about free trade in securi-ties is the risk that investor protection might beweakened. Domestic regulators would relyupon the supervision and investor-protectionstandards of the home regulator of the foreignbroker or exchange. There could be concernsabout the comparability of regulations, thepractical implementation and enforcement ofthem, and the effectiveness of the regulatory en-vironment. For example, not all countries em-ploy equal resources for enforcement, and theeffectiveness of enforcement may differ amongcountries. Regulators are aware of these issues,however, and the process of negotiating bilater-al agreements may lead to greater convergenceand the adoption of higher regulatorystandards.46
How might free trade in securities beexpected to work?The SEC is expected to propose a framework(for comment) for mutual recognition some-time in the second half of 2007. The SEC heldits first round table discussion in June 2007 todevelop a set of criteria against which foreignregulatory regimes would be judged to deter-mine comparability of standards.47 This waspreceded by a theoretical framework proposedby staff from the SEC in early 2007 to facilitatefree trade in securities (Tafara and Peterson
45. The IIAC has formed a national committee on freetrade in securities, the initial mandate of which is togain better access to the U.S. institutional market-place for Canadian investment dealers. In September2007, the IIAC submitted a proposal to the SEC relat-ing to free trade in securities, focusing on the institu-tional market.
46. The European Union has a history of mutual recogni-tion. The E.U. Financial Services Action Plan began aprocess of breaking down barriers and enhancingfinancial integration. See European Commission(2005).
47. U.S. Treasury Secretary Henry Paulson has led an ini-tiative to examine the global competitiveness of U.S.capital markets. In the June 2007 update, he high-lights international investment opportunities withrecognition of comparable regulatory regimes as astrategy to enhance competitiveness, specifically cit-ing the SEC consideration of mutual recognition asan effective example.
2007). They suggest an arrangement with for-eign regulators in which each side would relyon the other’s regulatory systems, an approachknown as mutual recognition. The arrangementwould allow foreign exchanges to offer theirlisted securities and foreign broker-dealers to of-fer only foreign investment products to U.S. in-vestors directly without having to submit to SECregulation (if approved by a regulator abroadwith equivalent rules). The proposal outlined afour-step process applicable to both foreign ex-changes and broker-dealers: i) a foreign entitypetitions the SEC to seek an exemption fromSEC registration; ii) the home market regulatorof an entity and the SEC assess comparability ofregimes, including enforcement, and negotiate aco-operative and information-sharing arrange-ment; iii) the SEC and the petitioning entityestablish an agreement as to jurisdiction andservice of process; and iv) the SEC assesses thepetition, following public notice and publiccomment.
Canadian contextIn Canada, the 2007 federal budget discussespursuing free trade in securities with the UnitedStates and other G-7 countries. In February2007, the G-7 countries agreed to explore freetrade in securities based on the concept of mu-tual recognition of regulatory regimes. High-level discussions exploring the potential bene-fits of free trade in securities have occurred inthe G-7 and between Canada and the UnitedStates; the European Union has also been ex-ploring this issue with the United States.48
Canada faces potential obstacles, however, inpursuing free trade in securities agreements. TheMulti-Jurisdictional Disclosure System (MJDS)offers a precedent that can be built on, anddemonstrates that mutual recognition betweenCanada and the United States in the field of se-curities regulation is possible. But free trade insecurities may represent a significantly morecomplex undertaking than the disclosure com-ponent already achieved in MJDS.49 A greaterdegree of comfort in the regulatory framework
48. European Commissioner for the Internal Market andServices, Charlie McCreevy, proposed the building ofa transatlantic marketplace (E.U.-U.S.) and identifiedprinciples underpinning it (McCreevy 2007).
49. The reciprocal agreement called MJDS recognizeseach nation’s disclosure requirements for securitiesofferings in each market (Wolburgh Jenah 2007).
33
Developments and Trends
will be necessary for free trade in securities, mostnotably in the area of enforcement. Canada’senforcement has been criticized and could pose arisk to the process when other countries assess thecomparability of standards (Cory and Pilkington2006). Canada might also be at a disadvantagein implementing a nationwide agreement on freetrade in securities, since it has multiple securitiesregulators and thus greater jurisdictional complex-ity for implementing mutual recognition with aforeign entity.
ConclusionWhile many questions remain unanswered as tohow such a plan would be implemented, thepotential exists for a clear net benefit to finan-cial markets. Free trade in securities could createbroader investor and issuer access, improvemarket efficiency, and increase competition, allof which could reduce transactions costs. Con-cern about a possible weakening of investorprotection has been identified as a top priorityin discussions between Canadian and other se-curities regulators. It is likely that the resolutionof this issue will be found in a co-operative ap-proach to securities regulation that is focusedon regulatory outcomes rather than on detailedprocesses.
References
Cory, P. and M. Pilkington. 2006. “CriticalIssues in Enforcement.” In Canada StepsUp, Vol. 6, 165–252. Final Report to theTask Force to Modernize Securities Legis-lation in Canada.
Cowan, K. 2007. “Free Trade in Securities.”Remarks to National Centre for BusinessLaw at the University of British Columbia(18 October).
European Commission. 2005. “FinancialServices Policy Priorities 2005–2010.”Available at <http://ec.europa.eu/internal_market/finances/docs/white_paper/white_paper_en.pdf>.
McCreevy, C. 2007. “Building the TransatlanticMarketplace.” Speech to the HeymanCenter on Corporate Governance andthe Securities Industry and FinancialMarkets Association (SIFMA), New York,New York, 7 March. Available at<http://www.heyman-center.org/programs/mccreevy%20speech.pdf>.
34
Tafara, E. and R. J. Peterson. 2007. “A Blueprintfor Cross-Border Access to U.S. Investors:A New International Framework.” HarvardInternational Law Journal 48 (Winter):32–68.
Wolburgh Jenah, S. 2007. “Commentary on ‘ABlueprint for Cross-Border Access to U.S.Investors: A New International Frame-work.’” Harvard International Law Journal48 (Winter): 69–83.
Reports
Financial System Review – December 2007
Introduction
eports address specific issues ofrelevance to the financial system(whether institutions, markets, orclearing and settlement systems)in greater depth.
In An Approach to Stress Testing the CanadianMortgage Portfolio, Moez Souissi presents oneapproach for assessing the vulnerabilities of theCanadian mortgage market. To illustrate thisapproach, he evaluates the overall risk of defaulton Canadian mortgages under a hypotheticalscenario in which house prices are falling. Hemeasures the overall default rate using two-stepdefault analysis. First, the probabilities of defaultfor different loan-to-value ratios are estimatedusing an option-pricing model. Then, the over-all default rate is estimated by applying theseprobabilities to the loan-to-value distributionfound in the Canadian Financial Monitor, asurvey conducted by Ipsos Reid Canada. Thesimulation results appear reasonable in the lightof historical experience.
The competitive landscape for securities tradingis being affected by two opposing trends. Secu-rities exchanges are consolidating, both withinand across borders, and alliances between spe-cialized marketplaces are being formed. At thesame time, new trading venues offering the pos-sibility of trading outside an exchange are beingcreated. In The Changing Landscapeof SecuritiesTrading, Éric Chouinard looks at how competi-tion among securities marketplaces is changing.Overall, the level of competition appears to berising as the competitive power of traditionalexchanges is being challenged by the emergenceof alternative trading systems (ATSs). Whilethe rise of ATSs has the potential to fragmentmarkets, thereby reducing liquidity and hinderingprice discovery, this is mitigated by regulationsrequiring that marketplaces be linked and by
R
technological tools that consolidate prices acrossthe multiple marketplaces in which a securitytrades.The downgrading of an unprecedented amountof asset-backed securities and collateralized debtobligations has raised concerns about the abilityof credit-rating agencies to provide opinions onthe credit quality of these instruments. In thereport, Reforming the Credit-Rating Process,Mark Zelmer reviews the various proposals thathave been put forward to enhance the processused to produce credit ratings, especially in thearea of structured products. The report notes thatthere are some natural self-correcting marketforces at work that should ensure that ratingagencies continue to improve their processes. Inthat regard, the recent market turbulence shouldserve as a useful stress test to help calibrate theiranalytic tools in the future. That said, the reportproposes some further steps that could be takento reinforce market discipline in the rating industry.
37
Financial System Review – December 2007
An Approach to Stress Testing theCanadian Mortgage PortfolioMoez Souissi
n Canada, residential mortgage loans ac-count for close to 47 per cent of the totalloan portfolio of commercial banks.Despite this large exposure to the housing
and mortgage markets, most of the risk of de-fault rests with the mortgage insurers rather thanwith the banks.1 Currently, the default rate onmortgage loans is near its historical low. Never-theless, it is important to improve our ability toassess the risks to the financial system associatedwith the housing and mortgage markets.
For example, it would be helpful to have tools toassess how a slowdown in the Canadian housingmarket of a magnitude similar to that observedin Canada during the early 1990s, or to that cur-rently under way in the United States,2 wouldincrease the overall risk of default on mortgageloans, particularly for highly leveraged loans.
The objective of this work is to present an option-pricing approach to assessing the vulnerabilitiesof the Canadian mortgage market. This approachis confined to analyzing only financially moti-vated defaults. It is based on the microeconomicprinciple that default can be a rational responseto adverse changes in the housing market. Itdoes not take into account involuntary defaults
1. In Canada, mortgages with a down payment of lessthan 20 per cent must be insured. In 2006Q4,insured mortgages represented more than 45 per centof total residential mortgage balances outstanding atchartered banks. Uninsured mortgages are associatedwith a low risk of default, because they are backed bya relatively large amount of collateral.
2. In the United States, the largest recent decrease inhouse prices was the cumulative decline in the nomi-nal median selling price of existing houses of 8.1 percent between June 2005 and October 2006 (NSA-National Association of Realtors). Note that thismeasure of house prices has increased since October2006.
I
caused by income constraints, such as thosecaused by job loss.To illustrate how this approach could be ap-plied, we evaluate the overall risk of default onCanadian mortgages under a scenario in whichhouse prices are falling. This is done using theempirical distribution of loan-to-value (LTV)ratios in 2006, obtained from the CanadianFinancial Monitor (CFM), a survey conductedby Ipsos Reid Canada.3
The Model
There is a growing body of literature on mort-gage default risk and how it relates to houseprices and interest rates. One strand of this liter-ature, which is motivated by option theory,maintains that, in a perfectly competitive market,mortgage borrowers can increase their wealth bydefaulting when the market value of the mort-gage equals or exceeds the market value of thecollateral, which depends on the price of thehouse.
Indeed, under conditions of limited liability,negligible transactions costs, and no exogenousreasons for residential mobility, default can beseen as a financial decision that can be separatedfrom the real (housing) decision, and the Mertontheory of the pricing of corporate debt can beapplied.4
Here, we analyze the homeowner’s decision todefault based on this criterion. We use a stan-dard two-factor theoretical contingent-claimspricing model. This model, which was initially
3. For details on CFM data, see Faruqui, Lai, and Traclet(2006).
4. For a more detailed discussion, see Deng, Quigley,and Van Order (2000) and Kau et al. (1995).
39
Reports
developed to evaluate mortgage contracts,5
generates all the information needed to com-pute the probability of default on any fixed-ratemortgage contract.6
The first factor in the model is the price of hous-ing, which is assumed to follow the standardgeometric Brownian motion, the equivalent incontinuous time of a random walk with a drift.The return from owning a house consists ofprice appreciation and the flow of services thatthe owner incurs by living in the house.
The second factor is the short-term interest rate.We assume that it follows a mean-reverting pro-cess. This process assumes that the interest ratereverts to its long-term value at a certain speed,but that this pattern is constantly disturbed bystochastic events.
In our model, for every possible outcome forhouse prices and interest rates over the length ofthe contract, the borrower faces three options:making the required payment, defaulting, orprepaying the mortgage.
The opportunity to default is treated as a putoption, since it enables the borrower to sell theproperty to the mortgagee at a price equal to theloan’s outstanding balance. This opportunityhas value if the expected present value of theremaining payments becomes higher than themarket price of the house.
The mortgagor also has the opportunity to pre-pay the mortgage loan.7 Prepayment can beviewed as refinancing. We treat the opportunityto prepay as a call option, in that it allows theborrower to buy all future obligations remainingunder the mortgage at a price equal to the loan'soutstanding balance. Prepayment has value ifinterest rates fall below the fixed rate of themortgage to the extent that the expected presentvalue of the remaining payments becomes higherthan the unpaid balance of the mortgage.
5. As was pointed out in Chatterjee, Edmister, andFlatfield (1998), the two-factor model is efficient inpredicting market mortgage values.
6. In this work, we focus exclusively on fixed-rate mort-gage loans, which account for about 75 per cent oftotal mortgage loans outstanding in Canada.
7. As suggested in Deng and Gabriel (2006) and Deng,Quigley, and Van Order (2000), one cannot accu-rately calculate the economic value of the defaultoption without simultaneously considering thefinancial incentive for prepayment.
40
Note that closed mortgages generally cannot bepaid off before maturity without a penalty. Pre-payment penalties in Canada are frequently cal-culated as the greater of three months’ interestor the interest differential applied to the out-standing balance. For simplicity, we use theformer.
These options are “embedded” in the sense thatthey give the mortgagor not only the opportunityto default or prepay now, but also the opportu-nity to postpone the default or the prepaymentby at least one period to see if it will provideadditional value.
Hence, at every period, the borrower solves adynamic problem wherein today's options areconsidered, as well as the potential options overthe rest of the contract. At any time, the borrowerobserves the current values of the house priceand the interest rate. Given these values and theassumed processes for how these variables evolveover time, the homeowner evaluates ex ante thepossible values of the house price, the interestrate in the next period, and their respective pro-babilities. Based on these values, the borrowerassesses whether it is less costly to default, toprepay, or to make the scheduled payment.
Caveats
Several caveats apply to our approach:
• Limited liability is assumed. This assump-tion may lead to an exaggerated measure ofthe risk of default on mortgages because, inCanada, borrowers remain liable for theunpaid balance of the mortgage loan overand above the current value of the house.
• As noted earlier, income constraints are nottaken into account within this methodology.
• Costs associated with the loss of reputationfor a defaulting borrower are not consideredhere. These costs can be significant (Kau,Keenan, and Kim 1994). The decision todefault can make it more difficult for theindividual to obtain credit in the future. Thiscreates an upward bias in our estimatedprobability of default. These costs could beincorporated into the default decision byadding a cost term to the outstanding bal-ance at the time of default.
• As mentioned above, prepayment can beviewed as refinancing. Although refinancing,like prepayment, implies termination of the
Financial System Review – December 2007
Table 1
Base-Case Parameters for Numerical Modelling
Note: Values of other parameters related to the stochastic behaviour ofhouse prices and the interest rate are chosen as follows. The standarddeviation of stochastic disturbances to change in house prices isestimated (over the 2001–06 period) at 4 per cent per year. Thestandard deviation of stochastic disturbances to interest rates andthe reversion parameter, which measures the speed of return to themean interest rate, are set equal to 10 basis points and 25 per centper year, respectively. These values are within the range of thosereported in previous works by McManus and Watt (1999) andBolder (2001).
Parameters Base case
Mortgage term 5 years
Amortization period 25 years
Contract mortgage rate at origination rc = 5.70%
Expected rate of appreciation of nominal houseprice
α = 6.50%
Original 1-month interest rate r0 = 3.00%
Transaction cost of prepayment(three months’ interest, dollar amount)
1% of the mortgagebalance
current mortgage contract, it also impliesthe origination of a new mortgage loan onwhich the borrower may default. This is notmodelled in this study because of its com-plexity. Consequently, the probability ofdefault that we compute at a given time isspecific to the original mortgage contract.This leads to a downward bias in our esti-mated probability of default, since refi-nanced mortgages are assumed not todefault.
The Simulations
These simulations illustrate how this modelcould be used to analyze the impact of decreasinghouse prices on mortgage defaults. 8
We measure the overall default rate using a two-step default analysis. First, the probabilitiesof default for different loan-to-value ratios areestimated using an option-pricing model asdescribed above. The overall default rate is thenestimated by applying these probabilities to theempirical LTV distribution, which we constructfrom the CFM database.
Parameters of the simulations
We consider a representative homeowner whohas taken out a 5-year mortgage contract with a25-year amortization period.
To illustrate how the model works, we calibratedthe parameters of our model so that they reflectas closely as possible the economic situation inCanada over the 2001Q1–2006Q1 period. Thisis our base case. In fact, we used the averagevalues over the period of the 5-year discountedmortgage rate, the rate of nominal appreciationin house prices, and the 1-month treasury billrate. The latter is used for both the originaldiscounting rate and the rate to which it revertsover the given 5-year period. We also assumethat some transactions costs are charged in thecase of a prepayment. The chosen parametersare summarized in Table 1.
After valuing the probability of default for dif-ferent LTV ratios at origination in the base case,we repeat the exercise, assuming other scenariosfor the evolution of house prices.
8. The same method could be used to examine thepotential impact of a change in interest rates.
41
Reports
Table 2
Probability of Default at Maturity Date
Per cent
LTV ratios Overalldefault rate
40% 75% 80% 90% 95% 100%
Base case (α = 6.50%)
0.00 0.05 0.36 1.39 2.62 3.80 0.31
Moderate case (α = 2.5%)
0.00 0.19 1.08 2.51 5.10 6.98 0.63
Extreme case (α = -2%)
0.00 0.77 2.89 5.53 9.11 12.10 1.35
Very extreme case (α = -5%)
0.00 2.01 5.96 8.13 12.47 16.22 2.25
In the first of three further scenarios consideredin this illustration, the moderate case, we assumethat house prices are expected to increase mode-rately at an annual rate of 2.5 per cent. The secondscenario is the extreme case in which nominalhouse prices decline at an annual rate of -2 percent (the rate of decline observed over the1990Q1–1995Q1 period). In the third scenario,the very extreme case, nominal house prices de-crease at an annual rate of -5 per cent. This valuereflects a real decrease in house prices equiva-lent to that observed in the early 1990s. All otherparameters are equal to those in the base case.9
Results
The results of our simulations are summarizedin Table 2. The first six columns indicate the cu-mulative probabilities of default over the fiveyears of the loan for mortgages with differentLTV ratios.
As expected, the probability of default is greaterthe higher the LTV ratio10 and the lower the rateof increase in house prices. For example, as shownin Table 2, under the base-case scenario (houseprice increase), a loan with a 75 per cent LTVhas a 0.05 per cent chance of reaching a pointwhere it is optimal to default. This probability ishigher, 0.77 per cent, in the extreme scenario. Inthe case of a 100 per cent LTV ratio, these prob-abilities increase to 3.8 per cent (base case) and12.1 per cent (extreme case).
For a given LTV ratio, the cumulative probabili-ties of default over the five years of the contractcan be interpreted as the proportion of defaultin the pool of current mortgages that share thesame LTV ratio and were signed five years earlier.
The overall default rate is a weighted averagecalculated by multiplying these cumulativeprobabilities by the weights given by the empir-ical distribution of LTV ratios. For simplicity,
9. To better reflect the current interest rate environment,we also simulated the outcomes of these scenariosusing 4.5 per cent as the value of the original dis-counting rate and the rate to which it reverts overthe coming 5-year period. Our results did not changesignificantly.
10. The insurance premium paid by a mortgagor whosedown payment is less than 20 per cent increases withthe LTV ratio. This is consistent with our resultsshowing that probabilities of default increase withLTV ratios (at origination).
42
Financial System Review – December 2007
Table 3
Distribution of Mortgages in 2006 by LTV Ratio
As a percentage of asset values
LTV ratios Frequency
Less than 75 79.45
75 to 80 5.34
80 to 90 8.81
90 to 95 1.53
95 to 100 0.00
100 and > 100 4.87
we used the 2006 distribution, as shown inTable 3, in our examples.
In what follows, we compare our estimatedoverall default rate with actual default rates. Thesimulated default rates differ from observedrates, because we consider only fixed-rate mort-gages in our model, while actual default ratesreflect defaults on both fixed-rate and variable-rate mortgages. Defaults on variable-rate mort-gages may be more sensitive to changes in interestrates than defaults on fixed-rate mortgages. Thiscomparison is intended to provide only a roughtest of whether our estimates are in the generalrange of historical experience.
Our estimated rates of default appear reason-able and broadly within the range of historicallyobserved default rates. The overall rate of de-fault estimated for the base case (0.31 per cent)is slightly higher than the actual rate of defaultobserved in 2006 (0.23 per cent).
Also, our results suggest that the rate of defaultwould reach 1.35 per cent following a persistentdecrease in house prices similar to the one ob-served over the 1990Q1–1995Q1 period. Thisrate is higher than the peak observed in Canadain 1992Q1 (0.62 per cent).11 This is because, asmentioned in the caveats, the assumption oflimited liability may lead to an exaggeratedmeasure of the risk of default, particularly inscenarios where defaults are more likely to hap-pen (i.e., decreasing house prices). The rate ofdefault is still much higher in the very extremescenario (2.25 per cent).
These rates do not reflect actual losses to banksand mortgage insurers, because the loss-given-default on mortgages is considerably less than100 per cent of the balance of the mortgage.Anecdotal evidence suggests that the loss-given-default on mortgages may be around 10 percent.
These comparisons should, however, be inter-preted with caution, given the caveats men-tioned above. Nevertheless, they suggest thatthe methodology applied here can be useful forstress testing the portfolio of Canadian mort-gage loans.
11. The 0.62 per cent rate is measured as a percentage ofthe number of mortgage loans in arrears three monthsor more. Data on default rates as a percentage of assetvalues are not available before 1997.
43
Reports
Conclusions
This work applies a contingent-claims modelof mortgage default to analyze the impact ofchanges in house prices on the decision todefault.
This approach has limitations. In particular, it istechnically very difficult to introduce additionalfactors into this framework to take into accountother important aspects of the default decision,such as the risk of income loss. Also, we do notexplicitly model the fact that, besides the optionsto default and to prepay, the mortgagor canchoose to refinance his loan at a new mortgagerate. This would require the introduction of athird stochastic variable, which would make thesolution of the model extremely complex.
On the whole, however, this work appears help-ful in gauging the risk of default on mortgageloans under different scenarios and assump-tions regarding the evolution of the distributionof LTV ratios.
References
Bolder, D. J. 2001. “Affine Term-Structure Models:Theory and Implementation.” Bank ofCanada Working Paper No. 2001-15.
Chatterjee, A., R. O. Edmister, and G. B. Flat-field. 1998. “An Empirical Investigation ofAlternative Contingent Claims Models forPricing Residential Mortgages.” Journal ofReal Estate Finance and Economics 17 (2):139–62.
Deng, Y. and S. Gabriel. 2006. “Risk-BasedPricing and the Enhancement of MortgageCredit Availability among Underservedand Higher Credit-Risk Populations.”Journal of Money, Credit and Banking38 (6): 1431–60.
Deng, Y., J. M. Quigley, and R. Van Order. 2000.“Mortgage Terminations, Heterogeneityand the Exercise of Mortgage Options.”Econometrica 68 (2): 275–307.
Faruqui, U., S. Lai, and V. Traclet. 2006. “AnAnalysis of the Financial Position of theHousehold Sector using Microdata.”Bank of Canada Financial System Review(December): 14–17.
44
Kau, J. B., D. C. Keenan, and T. Kim. 1994.“Default Probabilities for Mortgages.”Journal of Urban Economics 35 (3): 278–96.
Kau, J. B., D. C. Keenan, W. J. Muller III, andJ. F. Epperson. 1995. “The Valuation atOrigination of Fixed-Rate Mortgages withDefault and Prepayment.” Journal of RealEstate Finance and Economics, 11 (1): 5–36.
McManus, D. and D. Watt. 1999. “EstimatingOne-Factor Models of Short-Term InterestRates.” Bank of Canada Working PaperNo. 99-18.
Financial System Review – December 2007
The Changing Landscape of SecuritiesTradingÉric Chouinard
he competitive landscape for securitiestrading is being affected by two opposingtrends: on the one hand, the consolida-tion of exchanges, both within and
across borders, and the formation of alliancesbetween marketplaces specializing in differentasset classes; on the other, the creation of newtrading venues that offer the possibility of tradingoutside an exchange. This article examines howthese changes are affecting competition amongsecurities marketplaces. The focus here is onequity and derivatives markets that are builtaround a central system to match orders. Fixed-income markets (and over-the-counter marketstructures more generally) are not examined.
As the competitive power of traditional exchangesis challenged by the emergence of alternativetrading systems (ATSs), the level of competitionappears to be increasing. While the rise in ATSshas the potential to fragment markets, therebyreducing liquidity and hindering price discov-ery, this is mitigated by regulation and techno-logical tools that consolidate prices across themultiple marketplaces where a security is trading.
The first section discusses the consolidation ofsecurities exchanges, its causes, and factors thatare acting as barriers to further consolidation.The second section focuses on the emergence ofalternative trading systems, including dark li-quidity pools, and examines the potential forfragmentation. Implications of these trends forcompetition are discussed in the third section.While the bulk of the discussion in this articleis focused on global trends, the Canadian situa-tion is presented briefly in the fourth section.
The Consolidation ofSecurities Exchanges
In recent years, the securities industry has expe-rienced a rapid transformation as marketplaceoperators formed various alliances. These have
T
ranged from full-blown mergers to looser formsof co-operation: for example, the creation of in-formal networks for cross-listing securities orfor sharing technology. Marketplace consolida-tion has the potential to result in deeper, moreliquid markets if the structures merging are wellintegrated and their order flow is, therefore,aggregated.Marketplace consolidation is not new. A waveof exchange consolidation occurred in the UnitedStates as long ago as the 1930s, when severalregional exchanges merged to better competeagainst the New York Stock Exchange (NYSE).What distinguishes the current environment isthat, increasingly, alliances cross geographicalborders. Stock exchanges are also looking to enterinto derivatives trading by forming partnershipswith derivatives exchanges. Table 1 presents anon-exhaustive list of the most important alli-ances formed since the beginning of 2006, whenthe pace of consolidation increased. Some ofthese mergers have not been finalized, pendingapproval either from regulators or from theexchanges’ shareholders.
Developments in communication and informa-tion-processing technology play an importantrole in the consolidation of marketplaces. Almostall of the major exchanges around the worldhave adopted electronic systems.1 In electronicmarkets, orders are routed to a central systemusing an electronic interface, and the process ofmatching prospective buyers with prospectivesellers is largely automated.
The shift towards electronic trading has encour-aged consolidation in several ways. First, it has in-creased the incentives for growth, since electronic
1. The New York Stock Exchange, the last importantexchange that matches orders on a trading floor, iscurrently moving towards a hybrid model, wheretraders will be given the choice of trading on an elec-tronic platform or on the trading floor.
45
Reports
Table 1
Significant Exchange Mergers, 2006 and 2007
Mergers that combined trading in cash and derivatives
Sydney Futures Exchange and Australian Securi-ties Exchange
Completed inJuly 2006
International Securities Exchange (U.S.) andDeutsche Börse (Germany)
Announced inApril 2007
Mergers of derivatives exchanges
Chicago Mercantile Exchange and Chicago Boardof Trade
Completed inJuly 2007
Mergers of stock exchanges
New York Stock Exchange (U.S.) and Euronext(pan-European)
Completed inFebruary 2007
NASDAQ (U.S.), OMX (Scandinavia), and BorseDubai (U.A.E.)
Announced inSeptember 2007
London Stock Exchange (U.K.) and Borsa Italiana(Italy)
Completed inOctober 2007
systems can execute more trades than traditionalvenues, where traders physically interact witheach other on a trading floor. Second, with a coststructure more heavily geared towards fixed costs,electronic trading increases the potential econo-mies of scale resulting from a merger. Third, elec-tronic trading permits linkages between exchangesthat floor-based trading systems preclude becauseof geographical or space constraints. Fourth,since the cost of developing trading technologyremains high, some alliances may be motivatedby a desire to gain access to better technology.
Consolidation is also supported by changes tothe governance of exchanges. In the past decadeor so, most exchanges have evolved from mem-ber-owned mutual entities to profit-seeking cor-porations. Demutualization has increased theincentives for exchanges to gain a competitiveedge and enhance value for their shareholders.It has also provided easier access to the capitalneeded to achieve their business plans.
As mentioned previously, the trend towardscross-border and cross-asset alliances suggeststhat exchanges want to diversify their opera-tions geographically and to increase the scopeof their services. In principle, the alliance of amarket operator with a marketplace in a foreignjurisdiction could enhance the liquidity of thesecurities they trade, because they would haveaccess to a larger investor base. Moreover, withmultinational exchanges, investors could diver-sify away from country-specific risks or imple-ment investment strategies involving multiplesecurities listed in different countries moreeasily and with less market risk than by tradingin a number of different exchanges.2 Alliancescombining a stock and a derivatives exchangecould have the same type of benefits if theyfacilitated the simultaneous trading of relatedcash securities and derivatives instruments.
Currently, however, regulatory constraints limitthe benefits of cross-border consolidation. Reg-ulators have oversight responsibilities for boththe operation of exchanges and the securitieslisted on them, and most regulators restrict theaccess of marketplaces that they do not overseeto investors from their jurisdiction. This limits
2. An example of such an investment strategy is a “long-short” trade, which involves buying a security whilesimultaneously selling short another, in the hope ofprofiting from changes in the price difference betweenthe two securities.
46
Financial System Review – December 2007
the integration of market structures when ex-changes from different jurisdictions merge. Inmost cases, the newly formed entity continues tooperate two distinct marketplaces, each offeringtrading for a separate group of securities. There-fore, cross-border consolidation does not neces-sarily facilitate trading at present, and economiesof scale can be limited. Such mergers may involvesharing technology. They may also increase rev-enues—for example, from listings—and diversifythese revenue streams geographically.
This situation could change if the G-7 countriesmake significant progress in achieving free tradefor financial securities. (See Highlighted Issueon p. 32.) Countries reaching such an agree-ment would allow investors within their bor-ders direct access to foreign marketplaces. Thiswould be made possible by a mutual recogni-tion regime of rules and enforcement decisionsof foreign regulatory authorities from partici-pating jurisdictions. Free trade would facilitatethe integration of marketplaces involved in across-border merger.
Another constraint to consolidation is the frag-mentation of clearing and settlement systems.Clearing involves the confirmation of the termsof a trade by the buyer and the seller after thetrade has been executed and the calculation ofeach party’s obligations. Settlement entails thetransfer of funds and assets between the buyerand the seller. Clearing and settlement process-es are a key component of any securities trans-action.
Many exchanges (for example, those trading de-rivatives) are vertically integrated, using theirown subsidiary to perform this service.3 Owner-ship of clearing and settlement systems can beprofitable for exchanges, because it reduces theirpost-trading costs. It can also be a source ofrevenue, if clearing and settlement of tradesconducted over-the-counter or in another mar-ketplace is offered.
Differences in clearing and settlement systemscomplicate the integration of exchanges becauseof the lack of fungibility or interoperability be-tween systems. Efforts are under way to enhancethe interoperability of post-trading systems, both
3. European stock exchanges are also largely verticallyintegrated. This contrasts with stock trading inCanada and the United States, where there is acentral—and independent—clearing and settlementagency.
within and across borders, which will reducethe difficulties of integrating two marketplaces.These efforts involve agreeing to common tech-nical standards for messaging and communica-tions, eliminating paper, and strengthening risk-management standards (Group of Thirty 2006).
It is difficult to anticipate how far convergencewill go. Many believe that the industry will reachan equilibrium, where a small number of largeexchanges with a global reach and offering tradingin various types of assets may coexist with smallerexchanges specializing in the trading of particu-lar securities: for example, those issued by firmsfrom a given industry or country.
Most stock exchanges are actively looking to ex-pand into derivatives markets—the most profit-able and fastest-growing segment of the industry.Both NASDAQ and NYSE Euronext have ex-pressed a desire to continue to expand geogra-phically by merging with an Asian marketplace.
The Emergence of AlternativeTrading Systems
Many ATSs are simple order books in which buyand sell orders are electronically matched. Theydiffer from traditional exchanges in two impor-tant ways. First, ATS operators can—and oftendo—grant direct access to their system to insti-tutional investors, allowing them to trade with-out a securities dealer acting as an intermediary.Second, ATSs do not restrict trading to securitiesthat meet certain admission requirements. Anysecurity can, in principle, be traded on any ATS,provided its issuer is registered with regulators.
By allowing securities to trade on marketplacesother than those where they are officially listed,ATSs represent perhaps the most significant de-velopment for the competitive structure of mar-kets. Traditionally, exchanges have enjoyed anatural monopoly in the trading of the securi-ties listed on them, except when the issuer madethe decision to list on multiple exchanges.
A particular type of ATS has recently been re-ceiving considerable attention: internal crossingnetworks. These are also known as “dark liquiditypools” because they do not display standingorders to the public. Such systems without pre-trade transparency are ideally suited for con-ducting large trades that might move marketprices if the order was disseminated publicly. Insome ways, dark liquidity pools are a substitutefor the “upstairs market” of a traditional
47
Reports
exchange, where large trades are matched out-side of the central order book.
Most dark pools are operated by securities dealersthat internally match order flows originatingfrom their various business lines, such as theirretail, institutional, or proprietary trading desks.Securities dealers have followed the practice ofinternalizing orders for decades. What is changingis that regulation in many jurisdictions is nowrequiring that internal trading be automated.Dark liquidity pools must also be linked to publicmarkets in two ways. First, transactions mustoccur at prices that are, at worst, consistent withthe best bid or offer posted across all publicmarkets. In practice, most dark pools conducttrades within the bid/ask spread, thereby im-proving on market prices. Second, regulationrequires that information on completed trans-actions be disseminated publicly. Dark poolsare therefore not entirely opaque.
Many large institutional investors value darkpools, mainly because orders can be kept pri-vate until after they are executed. As for dealers,they can save on transaction fees by matchingorders from various sources internally.
The emergence of ATSs is supported in many ju-risdictions by rules to improve competition andincrease the efficiency of markets. In Europe, forexample, the recent Markets in Financial Instru-ments Directive (MiFID) allows investment firmsto route orders to all types of marketplaces, notonly their national exchanges, as was previouslythe case. This greater use of alternative systemsis expected to increase trading speeds and cuttrading costs. It will also take away business fromtraditional exchanges. MiFID is being creditedwith triggering the recent emergence of ATSs inEurope. Several projects are in the planningstage, the most important being Project Tur-quoise, which is a system owned by seven largesecurities dealers that accounts for about halfthe trading on European exchanges.
In the United States, the Securities and ExchangeCommission adopted Reg NMS. This regulationconnects marketplaces and contains a provisionpreventing standing orders on an automatedmarket from being bypassed in favour of inferiororders submitted elsewhere. This protectionexisted before, but did not cover orders fromATSs. The Canadian Securities Administrators(CSA) released a proposal in the spring to ex-tend similar “trade-through” protection to ATSs.It has yet to be implemented, since comments
48
received during a public consultation are stillbeing reviewed.
The emergence of ATSs has raised concerns thatthey may fragment markets. Fragmentation arisesin the context of securities markets when all or-ders do not interact with each other via a singleorder-execution mechanism. Fragmentation re-duces market liquidity and hinders the price-discovery mechanism. Fragmentation is not anew concept. It occurs, for example, when a firmlists its shares on multiple exchanges. But theemergence of ATSs has brought this issue to theforefront, particularly the rise of dark liquiditypools, where orders are internalized. More tra-ditional ATSs can also lead to fragmentation,unless quotes and trades from various marketswhere a security is traded are brought togetherto provide a consolidated overview of pricesacross all marketplaces.
Fragmentation is being offset by regulation re-quiring marketplaces to be linked together andby technological tools. These tools allow tradersto connect to multiple marketplaces rapidly andinexpensively, to scan prices across them, and todirect orders to the marketplace in which theprice is the most advantageous.
The development of ATSs can be seen as part ofa broader response to changes in market struc-ture, with dealers and investors attempting tocounteract a possible rise in the competitivepower of exchanges as they consolidate.
What Does This Mean forCompetition?
The net impact on competition of these twotrends is difficult to assess. Economic theorysuggests that consolidation increases the marketpower of the firms left in the industry and thatthey may raise their prices. But the threat ofcompetition from new entrants, such as ATSs,may limit their market power.
The limited data available indicate that tradingcosts are trending downwards, which suggeststhat the emergence of new marketplaces is in-creasing competition in the industry. Accordingto data from Elkins/McSherry, average tradingcosts for stock transactions during the periodsJuly 2004 to June 2005 and July 2005 to June2006 decreased by about 6 basis points fromthe first period to the second. These averagecosts declined by about 29 basis points over thepast 10 years (Paulden 2006; Willoughby 1998).
Financial System Review – December 2007
While this is an average over the 42 countriesthat Elkins/McSherry track, trading costs haveconsistently decreased in most countries. InCanada, for example, the TSX Group reducedtrading fees up to 20 per cent in August 2007.It should be noted that the Elkins/McSherryestimates of trading costs are not entirely deter-mined by marketplaces. They combine market-place fees, dealer commissions, and the marketimpact of trades. All three components havebeen trending downwards.
Marketplaces collect two types of fees from trad-ers: one for the right to access the market (theseare akin to periodic membership fees) and onefor conducting a trade. With electronic tradingplatforms, the cost of executing a trade for amarketplace is generally very small (anecdotalevidence suggests that it is close to zero in manyinstances). The Elkins/McSherry data suggestthat most of these savings have been passed onto customers. They indicate that in developedcountries, marketplace fees account for about2 to 5 per cent of the total cost of conducting atrade.
Reduced trading costs cannot be directly attrib-uted to increased competition—or the threat ofcompetition—from ATSs, at least not entirely.The increasing use of electronic trading and itsenhanced efficiency over traditional floor-basedsystems do play a role. It appears, however, thatincreased competition from emerging tradingvenues, such as dark liquidity pools and otherATSs, may be curbing the enhanced pricing powerthat might otherwise arise from consolidation.
Trading facilities compete for securities orderson the basis of factors other than costs. As thenumber of trading venues increases, operatorsare using issues of market design (for example,the degree of transparency or the speed of exe-cution) to distinguish themselves from theircompetitors and attract order flow. Since differ-ent types of traders value these factors different-ly, trading venues are being shaped according towhich type of trader they wish to attract.
It should be noted that the amount of tradingfor a given security in a given marketplace im-proves the competitive position of the market-place for trading in that security. This is becauseliquidity is self-reinforcing. Simply put, a liquidmarket will, everything else being equal, attractmore orders than an illiquid one, and, as thesenew orders are placed, liquidity will continue toimprove. This partly explains the advantage that
incumbent marketplaces have had when com-peting with less-established venues.
Canadian Developments
Canada currently has two stock exchanges: theTSX Group, which operates a “senior” market-place for companies with a large capitalization,as well as a venture marketplace; and the Cana-dian Trading and Quotation System Inc., whichoperates a marketplace for micro-cap stocks.Five ATSs have been launched in the past twoyears, and other facilities are at an advancedplanning stage. These facilities will offer tradingin all stocks listed on the “senior” TSX market-place.
Three types of ATSs are emerging in Canada.The first group consists of transparent centrallimit-order books. The first two Canadian ATSsin that category, Pure Trading and Omega, werelaunched this autumn. Two more, ICX andAlpha, are expected to be launched by the endof 2008. Alpha will be owned by CanaccordCapital, the Canada Pension Plan InvestmentBoard, and the securities dealers linked to thesix main Canadian banks. Its shareholders ac-count for about 65 per cent of trading volumeon the TSX.
The second group of ATSs consists of dark li-quidity pools. The first Canadian dark pool,Match NOW, has been operational since July2007 and guarantees that any trade executed onthe system will be within the best bid and askquotes available across transparent markets.Another dark pool, ATX, will be operated by theTSX and is awaiting regulatory approval.
The third group of ATSs is for trading largeblocks of securities. Two facilities, BlockBookand Liquidnet, operate systems that are accessi-ble to either dealers or institutions and allowtraders with opposite needs to negotiate priceselectronically while preserving their anonymity.The ability to negotiate prices is an importantdifference between these facilities and the darkpools. On the latter, prices are established by amechanism according to prevailing conditionsacross all public markets.
The Montréal Exchange (MX) is currently the onlyexchange for derivatives trading in Canada. Butthe TSX Group has announced plans to offerderivatives trading in 2009 upon the expirationof a non-competition agreement signedby Canadian exchanges in 1999, when they
49
Reports
restructured markets. The TSX Group has formedan alliance with the International SecuritiesExchange to create DEX, a new exchange that,based on an agreement signed with Standard& Poor’s, will have exclusive rights for tradingderivatives on the TSX-S&P equity indexes. TheMontréal Exchange is the majority shareholderin the Boston Options Exchange.
Concluding Remarks
Improvements in information technology, theglobalization of financial markets, and regula-tory changes are altering the competitive land-scape for market providers. The capabilities oftraditional exchanges to compete in each other’smarkets have increased, and exchanges aremerging and reaching strategic alliances withinand across borders. They are also facing increas-ing competition from alternative trading sys-tems, which raises the potential for market frag-mentation. But, to date, fragmentation appearsto be more than offset by regulation and bytechnological tools that allow greater connectivityacross marketplaces.
The structure of the industry is changing rapidly,and its future will likely be determined to a largeextent by the ability of new ATSs to gain marketshare and by the success of policy-makers andmarket participants in removing the remainingbarriers to consolidation.
References
Group of Thirty. 2006. Global Clearing andSettlement: Final Monitoring Report.Washington, D.C.: Group of Thirty.
Paulden, P. 2006. “Keep the Change.” Institu-tional Investor 40 (12): 55–59.
Willoughby, J. 1998. “Execution’s Song.”Institutional Investor 32 (11): 51–54.
50
Financial System Review – December 2007
Reforming the Credit-Rating ProcessMark Zelmer
he downgrading (and warnings of poten-tial downgradings) of an unprecedentedamount of asset-backed securities andcollateralized debt obligations (CDOs) in
July have raised concerns among investors aboutthe ability of rating agencies to assess the creditquality of these instruments. Even though theaffected securities represented a small share ofthese markets, the action surprised investorsand led them to wonder why ratings had notbeen cut earlier, since the problems in the U.S.subprime-mortgage market had been appar-ent for some time. Rating agencies have alsobeen criticized for putting themselves in con-flict-of-interest situations. These concerns havebeen accompanied by calls in a number of coun-tries for greater public scrutiny of rating agen-cies and for more transparency in the ratingprocess. In October, the G-7 indicated its sup-port for the Financial Stability Forum’s plans tostudy the role, methodologies, and use of ratingagencies in structured-finance markets.
Various proposals have been put forward interna-tionally to enhance the ratings process, especiallyin the area of structured products. This reportprovides a brief overview and discusses the meritsof each one.
Increased Regulation ofRating Agencies
Citing past high-profile rating mistakes, conflictsof interest inherent in the rating business, andthe oligopolistic nature of the industry, someobservers have suggested that rating agenciesshould be regulated more closely. The majorrating agencies operating in the United Statesare currently regulated by the U.S. Securities andExchange Commission (SEC), which introduceda new regulatory framework for rating agencies
T
in June.1 To make the case for increased regula-tion, it must be demonstrated that there has beena market failure that is not likely to be correctedby market forces alone, and that more governmentregulation represents a viable, cost-effectivesolution.Rating agencies have been criticized for the con-flicts of interest inherent in their business: Theyreceive a significant portion of their revenuesfrom the issuers that they rate, even though thesame ratings are provided as a service to investorspurchasing the securities in question; they provideadvice to issuers before securities are issued thathelps them qualify in advance for desired ratings;and they publish unsolicited ratings that could,potentially at least, pressure issuers to pay themfees.
Rating agencies acknowledge these inherent con-flicts of interest and argue that they have a wide
1. In September 2006, the U.S. Congress passed theCredit Rating Agency Reform Act of 2006, which pro-vides new recognition standards and introduces moreformal oversight of rating agencies. However, the SECis prohibited under the Act from regulating the sub-stance of credit ratings or the process by which rat-ings are determined. The SEC introduced a new set ofrules to implement the Act in June 2007 and is cur-rently in the process of redesignating rating agenciesunder the new rules. It is therefore too soon to assessthe effectiveness of the new regulatory regime. Out-side of the United States, there is little oversight ofthe major rating agencies. The European Commis-sion has so far taken a “wait and see” attitude towardsthe rating industry. However, the recent market tur-bulence has resulted in calls from some Europeangovernments for more formal oversight of ratingagencies, and the European Commission has askedthe Committee of European Securities Regulators(CESR) to review the rating process surroundingstructured products.
51
Reports
range of mechanisms in place to manage them.For example, they have a highly diversified clientbase, which does not leave them overly dependenton any one client or sector. In addition, they maketheir rating criteria and opinions public, whichpromotes a better understanding among investorsof the rationale behind published ratings. Agenciesalso maintain a separation between the analyticaland commercial activities associated with anygiven rating to foster the independence of theirratings. The compensation of their analysts is notdependent on the fees related to the ratings theyassign, plus committees review and approvethe ratings proposed by the analysts. Moregenerally, they claim that the need to preservetheir reputation is an incentive for them to tryand provide fair, objective, and independentratings—a claim supported by Covitz andHarrison (2003).
While rating agencies have experienced somecontroversial failures, they have also shownan ability and willingness to learn from theirmistakes, and they are regularly refining theirrating processes. It is not clear that regulatorshave any comparative advantage in overseeingthe rating process, since they are further removedfrom the entities being rated than the agencies.Regulators may not be in the best position toevaluate the methods used by rating agencies toassess financial instruments. Venturing into thisterritory would expose them to the risk of beingheld publicly accountable for the mistakes ofrating agencies. It could also stifle innovationin the rating industry, as well as the developmentof financial markets more generally, since regu-lators would have difficulty keeping abreast ofthe flow of new products that are regularly beingdeveloped in financial markets. It is also notclear who would be best placed to regulate therating process, since rating agencies have sig-nificant cross-border activities and rate productsthat trade in more than one jurisdiction
Where public authorities may have a role toplay is in continuing to press rating agencies toadhere to the provisions of the IOSCO Code ofConduct Fundamentals for Credit Rating Agencies(IOSCO Code) so that investors have confidencethat conflicts of interest are well managed. Fourrating agencies have provided the Committeeof European Securities Regulators (CESR) withself-assessments of their adherence to the IOSCOCode. In general, the CESR concluded that the
52
rating agencies’ own codes of conduct comply,to a large extent, with the Code.2
Making Investors, rather thanIssuers, Pay for Credit Ratings
For most of their history, rating agencies mademoney by charging investors subscription fees;they did not charge the issuers of securities. Thischanged in the 1970s, when agencies begancharging fees to debt issuers to rate the creditquality of their securities. As credit ratings becamemore widely used, they were leaking into thepublic domain and becoming public goodswith all of the attendant free-rider issues. Today,ratings are available free of charge on rating-agency websites. Furthermore, issuers haveincentives to be rated because credit ratingsfacilitate their access to markets. Rating structuredproducts now accounts for as much as 45 per centof the revenues earned by some rating agencies.
Given the apparent conflicts of interest associatedwith rating agencies being paid by the issuersof securities they rate, some commentators havesuggested that the agencies should revert to thepractice of having investors, instead of issuers, payfor credit ratings. This may no longer be possibleor practical, however, now that credit ratings arepublic goods. The quality of ratings could declineover time if rating agencies were not able to fundan appropriate level of supporting research.There would also be less public scrutiny of ratingsif fewer investors had direct access to them, whichmight weaken the pressure on rating agencies toproduce high-quality ratings.3
Improving Competition inthe Credit-Ratings Industry
Moody’s and Standard & Poor’s have dominatedthe credit-ratings industry for many decades. Theyenjoy healthy profit margins, because there are
2. IOSCO (2007) reviewed the implementation of itsCode by a broader set of rating agencies. It found thatthose recognized by the U.S Securities and ExchangeCommission (SEC) have adopted codes of conductthat largely follow the provisions of the IOSCO Codewith few variations, but that additional efforts areneeded to promote adherence among some otherrating agencies. IOSCO is also reviewing the Codeto see if it needs to be revised.
Financial System Review – December 2007
significant barriers to entering the industry. Akey barrier is reputation, since investor faith inthe quality of credit ratings can be earned onlyover time through a proven track record of ratingsthat are reliable indicators of credit risk. Economiesof scale in the ratings business reinforce reputationas a natural barrier to entry. Rating a wide rangeof instruments and entities helps rating agenciessignal their reputations to investors. Referencesto specific rating agencies in a large number oflaws and government regulations may haveinadvertently added to the barriers to enteringthe industry.
Increased competition in the rating industrywould give investors access to a broader set ofviews on the credit quality of their investmentsand would help keep agency fees at an appropriatelevel. There are, however, few practical suggestionsabout how to do this. Some, like Pollock (2005),have suggested eliminating SEC designation ofrating agencies as a way to ensure that publicauthorities do not inadvertently contribute toentrance barriers. But the new regulatory frame-work introduced by the SEC in June now providesclearer criteria for rating agencies to achieve SECdesignation status, which may foster more com-petition.4 In addition, designation/recognitionby public authorities can help rating agencies re-duce the high cost of building a reputation, thuspromoting more competition in the industry.
Of broader interest is the reference to ratingagencies and ratings in other laws and regulations,and the risk that this may encourage marketparticipants to rely on ratings as a summarystatistic of risk. The Bank of England commentedin its October 2007 Financial Stability Report thatthere is a risk that banks may come to rely heavily
3. Public scrutiny of rating agencies is also reinforcedby academic research into the usefulness of creditratings. Numerous academic studies over severaldecades have generally found that rating actionslagged movements in market prices. Ammer andClinton (2004) found that rating downgrades havea larger impact on the prices of structured productsthan on those of traditional “plain-vanilla” instru-ments. By contrast, the market impact of ratingupgrades is insignificant for both types of instruments.
4. For example, the new criteria require a rating agencyto be in business as a rating agency for at least threeyears, and it would need to provide certificationsfrom at least ten institutional investors that they usethe agency’s ratings.
on ratings, particularly for structured products.Under Basel II, banks are required to use the ratingspublished by the rating agencies to determinecapital charges for structured products, wheresuch ratings exist. The Bank of England expressedsome concern that this regulatory requirementmay result in some banks using external ratingsas their only input when assessing structuredproducts, and recommended that this potentialoverreliance be addressed by banks and theirregulators.
Injecting More Transparencyinto the Rating Process
While rating agencies have a strong track recordin rating financial instruments, history is repletewith examples of high-profile debt restructuringsand defaults that the agencies failed to anticipatein a timely manner. Some recent examples includeIndonesia and Thailand in the mid-to-late 1990sand Enron and WorldCom a few years ago. TheCommittee on the Global Financial System(CGFS 2005) suggests that credit ratings ofstructured products are more fragile than thoseof other securities. This is consistent with datapresented in a recent Moody’s report, which issummarized in Charts 1 through 4. They indicatethat credit ratings of structured products aregenerally more stable than those of corporatesecurities, mainly because they have a lowerprobability of being upgraded. But when a down-grade occurs, credit ratings of structured productsare more likely to fall several notches. The lattermay reflect the greater uncertainty in assessing thecredit risk of structured products and the higherleverage embedded in some of those products.
The different behaviour of credit ratings forstructured products has led to calls for moretransparency in the process for rating thosesecurities. These include the IMF’s proposal invarious Global Financial Stability Reports that aseparate rating scale be used for structured productsand the CESR’s suggestion that rating agenciesbe encouraged to provide more meaningfulinformation on the analysis underpinning ratingdecisions. The Bank of England has also putforward some ideas on how the informationcontent of ratings could be improved. (See Box 6in its October 2007 Financial Stability Report.)They include thoughts on how rating agenciescould provide more information on the risksassociated with structured products and a
53
Reports
Chart 1 Probability of No Change
%
50
60
70
80
90
100
50
60
70
80
90
100
Aaa Aa2 A1 A3 Baa2 Ba1 Ba3 B2 Caa/C
CDOsCorporate bonds
Chart 2 Probability of an Upgrade
%
0
10
20
30
40
50
0
10
20
30
40
50
Aaa Aa2 A1 A3 Baa2 Ba1 Ba3 B2 Caa/C
CDOsCorporate bonds
Chart 3 Probability of a Downgrade
%
0
10
20
30
40
50
0
10
20
30
40
50
Aaa Aa2 A1 A3 Baa2 Ba1 Ba3 B2 Caa/C
CDOsCorporate bonds
Chart 4 Probability of More than a One-NotchDowngrade
%
0
10
20
30
40
50
0
10
20
30
40
50
Aaa Aa2 A1 A3 Baa2 Ba1 Ba3 B2 Caa/C
CDOsCorporate bonds
Source: Moody’s “Credit Migration of CDO Notes, 1996-2006, for US and European Transactions,” 28 February 2007.Note: Charts 1 through 3 sum to 100 per cent for each rating category. Chart 4 does not show a value for the two lowest rating categories because it is not possible
for these categories to be downgraded more than one notch.
suggestion that rating agencies adopt the samedefinitions for scoring credit risk.
A separate rating scale forstructured products
A separate rating scale for structured productswould emphasize that the ratings of those productsbehave differently from those of other financialinstruments. This might make investors thinktwice before purchasing them. Of course, thereis always a risk that investors may simply spendtheir time trying to map any new rating scale back
54
to theconventionaloneusedforother instruments.So, they would still need information about thecharacteristics of ratings for structured products.
Recent attempts by rating agencies to developmetrics to highlight the fragility of ratings forstructured products (such as Fitch’s stability scoresfor structured products) are an encouraging signthat market forces are at work, and that the privatesector will find an appropriate solution on itsown. A possible role for governments and regu-lators could be to speed up the process by bringingstakeholders together to discuss possible solutions.
Financial System Review – December 2007
More meaningful disclosure ofsupporting information
A recent assessment of rating agencies conductedby the CESR (2006) noted that they have a ten-dency to provide investors with descriptionsof rating methods that, for proprietary reasons,are quite general and not very precise or ex-haustive. In particular, CESR noted that it is oftendifficult to understand how a rating agency ar-rived at a particular rating. This makes it hard forinvestors to compare the opinions of various rat-ing agencies and to draw their own conclusions,especially for complex structured products.
While market forces should ultimately determinethe kind of information (and the format) thatwould best suit investors, there is again a possiblerole for governments and regulators, who couldhelp to facilitate an improvement in disclosureby arranging stakeholder discussions of this topicor by setting principles to guide the formulationof disclosure standards.5
Holding Rating AgenciesLegally Liable for the RatingsThey Publish
Partnoy (2006) notes that in the United Statesrating agencies have generally not been legallyliable for the credit ratings they publish, becausethe latter are considered to be “opinions,” andare thus treated as free speech under the law. Someobservers have suggested that rating agenciesshould be legally liable for the quality of theirratings, just as auditors are liable for the opinionsthey provide to investors and other stakeholders.It could be argued, however, that credit ratingsrequire more judgment than auditor opinions.
5. In its new rules for rating agencies, the SEC requiresthem to publish performance-measurement statisticson ratings for short-, medium-, and long-term peri-ods. An example of such statistics is the rating-transi-tion matrices published by the rating agencies thatwere used to prepare Charts 1 through 4. However,the SEC indicated that it is not prepared to prescribestandard metrics at this time. It is also studyingwhether it would be appropriate to require ratingagencies to furnish additional types of performancestatistics to be disclosed as an alternative, or in addi-tion, to rates on historical defaults and downgrades.Examples given included comparing a given creditrating to the market value of the rated security or toextreme declines in its market value after the rating.
Clearly, rating agencies must be held accountablefor the quality of their credit ratings. While rep-utation is a barrier to entering the industry, it isalso an inducement to continuous improvementin rating methods. Recent experience offersencouraging evidence that this is taking place.For example, rating agencies regularly revisetheir credit-assessment models in light of newinformation and as new analytic techniquesare developed so that they can be seen by investorsas being at the leading edge of credit-risk assess-ment. The crisis in the subprime-mortgage marketwill likely serve as a useful stress test that maywell help to inform future rating decisions.6
Banning the Rating ofProducts for Which anAgency Has Provided Advice
Rating agencies are actively engaged in advisingthe issuers of securities about the credit qualityof their securities before they are issued. This hasled to concern about the conflict of interest ofagencies that publish ratings on products for whichthey have provided advisory services. While ratingagencies have always been ready to adviseprospective issuers of securities, the concernsare magnified for structured products. Issuers ofthose products may have an opportunity to“game” rating agencies’ credit-assessment models.This is especially true for the newest products,where lack of data may limit the usefulness ofthose models.7
Banning rating agencies from all involvementin the development of structured products orrequiring them to split their rating business from
6. It may also result in more official scrutiny of ratingagencies. For example, the SEC has launched aninquiry into the behaviour of rating agencies in themarket for subprime residential mortgage-backedsecurities, and the President’s Working Group onFinancial Markets is examining the role of ratingagencies in lending practices, how their ratings areused, and how securitization has changed the mort-gage industry and related business practices.
7. While rating agencies have always interacted withissuers before a new instrument is rated and issued,issuers of structured products have more scope toadjust the terms and conditions of their products toachieve a desired credit rating in advance; for exam-ple, by varying the degree of over-collateralization foreach tranche.
55
Reports
their consulting activities seems excessive. Ratingagencies have useful expertise and insights tooffer, which should result in better products forinvestors. In addition, since the transparency ofthe rating process continues to improve, theremay not be any further tangible benefits fromimposing such a separation.
Instead, one could envision a regime in which arating agency is either: allowed to rate the productsfor which it has provided advisory services, butonly under the condition that it fully discloses itsinvolvement in the development of the product;or barred from rating products for which it hasprovided advice prior to issuance (while beingfree to rate other products for which its compet-itors provided advice). Here in Canada, recentmarket developments may lead investors toincreasingly follow the international practice ofrequiring structured products to be rated by twoor more rating agencies.
Market-Based Credit Ratings
There is also an issue of whether credit ratingsshould be replaced by market-based measuresof credit risk. Moody’s has already done muchof the work needed to generate market-basedratings. It publishes “Market Implied Ratings,”or MIRs, designed to reflect the credit-ratingequivalent of the market price of credit for variousinstruments over time. The main advantage ofsuch a metric is that it would incorporate allavailable information into a rating, includingthe ratings of other rating agencies. Moreover,it could be designed to be very timely. The maindisadvantage is that these measures may bemisleading, since market prices can be distortedby fads or bubbles. For example, many asset-backed securities traded at spreads that did notreflect their inherent liquidity risks prior to therecent market turbulence. Thus, prudence wouldsuggest that MIRs not be exclusively relied uponin managing credit risk.
Conclusion
Investors have had a long-standing need forspecialists who can advise them on the creditquality of their investments. Most find it cost-effective to delegate this task to rating agenciesthat can benefit from economies of scale inassessing the credit risk of a wide range of issuersand financial instruments. The advantages of this
56
approach have become increasingly pronouncedwith the emergence of structured products, sinceproper assessment of the credit risk of theseproducts requires sophisticated tools andmodelling skills.
The difficulties that structured products have facedin the recent market turbulence have renewedconcerns about rating agencies and their role infinancial markets. Rating agencies have benefitedhandsomely from the rapid growth in thosemarkets. While their ratings are generally con-sistent with actual default experience, ratingagencies have had some notable failures. Theyalso continue to have some well-recognizedconflicts of interest that need to be managed.Moreover, ratings are playing an increasinglyimportant role in financial markets as financialinstruments become more complex and difficultfor investors to understand. And ratings playimportant roles in public sector activities moregenerally.
Some natural self-correcting market forces are atwork, which should ensure that rating agenciescontinue to improve their processes. Althoughreputation is a natural barrier to entering therating industry, it is also an inducement tocontinuous improvement in rating methods.The recent turbulence in structured-financemarkets will, no doubt, be a useful stress test tohelp calibrate analytic tools in the future andmay lead investors to demand increased trans-parency in those markets so that they can bettermanage their exposure to credit risk. Moreover,the role of rating agencies in developing newstructured products may increase their exposureto legal risk,making themmore legallyaccountablefor the quality of the ratings they produce.
Nevertheless, there are some further steps thatcould be taken to reinforce market discipline inthe rating industry. For example, the rating processwould benefit from greater transparency, so thatinvestors are better able to critically assess ratings.The best solutions are likely to emerge throughan active dialogue between rating agencies andtheir stakeholders; regulators are unlikely tohave any comparative advantage in imposingsolutions. They could possibly play a usefulrole, however, in setting or supporting someminimum principles of disclosure, which couldbe the basis of a dialogue between investors andissuers, and could possibly lead to an agreementon industry-led best practices or a code of conduct.Initiatives such as the CESR’s assessment of rating
Financial System Review – December 2007
agency compliance with the IOSCO Code couldbe a useful means of maintaining public pressureon the rating agencies to continue to enhance themanagement of conflicts of interest in theirindustry. The role that rating agencies and ratingsplay in various laws and regulations and publicsector activities could possibly be re-examined.Formally recognizing specific rating agenciesmay have inadvertently reinforced barriers toentering the rating industry. It may also haveencouraged some investors to rely heavily oncredit ratings inappropriately as a summarystatistic of risk.
In the end though, investors need to acceptresponsibility for managing credit risk in theirportfolios. While complex instruments such asstructured products enhance the benefits to begained from relying on credit ratings, investorsshould not lose sight of the fact that one candelegate tasks but not accountability. Suggestionssuch as rating structured products on a differentrating scale could be helpful, in that this mayencourage investors to think twice before investingin such complex instruments. Nevertheless,investors still need to understand the productsthey invest in, so that they can critically reviewthe credit opinions provided by the rating agencies.
References
Ammer, J. and N. Clinton. 2004. “Good NewsIs No News? The Impact of Credit RatingChanges on the Pricing of Asset-BackedSecurities.” Board of Governors of theFederal Reserve System, InternationalFinance Discussion Paper No. 809.
Bank of England. 2007. “Mitigating Risks to theUK Financial System: Weaknesses in CreditRisk Assessment.” Financial Stability Report(October): 54–55.
Committee of European Securities Regulators(CESR). 2006. “CESR’s Report to the Euro-pean Commission on the Compliance ofCredit Rating Agencies with the IOSCOCode.” No. CESR/06-545. Available at<http://www.cesr-eu.org/data/document/06_545.pdf>.
Committee on the Global Financial System(CGFS). 2005. The Role of Ratings in Struc-tured Finance: Issues and Implications. Basel:Bank for International Settlements.
Covitz, D. M. and P. Harrison. 2003. “TestingConflicts of Interest at Bond Ratings Agen-cies with Market Anticipation: Evidencethat Reputation Incentives Dominate.”Federal Reserve Board Finance and Eco-nomic Discussion Paper No. 2003–68.
Moody’s Investors Service. 2007. “Credit Migra-tion of CDO Notes, 1996–2006, for USand European Transactions.” Structured-Finance Special Reports (28 February).
Partnoy, F. 2006. “How and Why Credit RatingAgencies Are Not Like Other Gatekeep-ers.” In Financial Gatekeepers: Can TheyProtect Investors? Y. Fuchita and R. E. Lian,editors. Brookings Institution Press andthe Nomura Institute of Capital MarketsResearch.
Pollock, A. J. 2005. “End the Government-Sponsored Cartel in Credit Ratings.”Financial Services Outlook (January). Wash-ington, D.C.: American Enterprise Insti-tute for Public Policy Research (January).
Technical Committee of the InternationalOrganization of Securities Commissions(IOSCO). 2007. “Review of Implementa-tion of the IOSCO Fundamentals of aCode of Conduct for Credit Rating Agencies:Consultation Report.” (February). Availa-ble at <http://www.iosco.org/library/pubdocs/pdf/IOSCOPD233.pdf>.
57
Policy and
Infrastructure
Developments
Financial System Review – December 2007
Introduction
he financial system and all of itsvarious components (institutions,markets, and clearing and settle-ment systems) are supported by a
set of arrangements, including governmentpolicies, that influence its structure andfacilitate its operation. Taken together,these arrangements form the financialsystem’s infrastructure. Experience hasdemonstrated that a key determinant ofa robust financial system is the extent towhich it is underpinned by a solid, well-developed infrastructure. This section ofthe Review highlights work in this area,including that related to relevant policydevelopments.
The market for over-the-counter (OTC) deriv-atives has grown exponentially in recent years; thetotal size, measured by the notional amount out-standing, increased at an annual average rate of20 per cent between 1998 and 2005, reachingUS$370 trillion worldwide in June 2006. Hedgefunds have become far more active, and, in somesegments of the market, they are among the mostactive participants. They have also contributedto the emergence of new practices, such as primebrokerage arrangements for OTC derivatives.There have also been significant developmentsin infrastructure, with vendors providing newservices for processing OTC derivatives. InDevelopments in Processing Over-the-CounterDerivatives, Natasha Khan discusses the mainfindings of a report by the Bank for Interna-tional Settlements that assesses these devel-opments and their implications for theefficiency and stability of the OTC derivativesmarket.
In a standard foreign exchange transaction, partiesto a trade (referred to as “banks” for simplicity)agree to exchange a value denominated in onecurrency for a value denominated in another
T
currency. A bank that irrevocably pays out thecurrency sold to its transaction counterpartyunconditional on the final receipt of the currencyit has purchased is exposed to financial loss upto the principal value of the trade if its counter-party fails to deliver the purchased currency. Inthe current context, this risk of financial lossis referred to as “foreign exchange settlement risk.”In the report,Management of Foreign ExchangeSettlement Risk at Canadian Banks,NevilleArjanihighlights key aspects related to foreign exchangesettlement and discusses how major Canadianbanks manage their exposure to foreign exchangesettlement risk. The author suggests that someof the major Canadian banks have significantlyreduced their exposure to foreign exchangesettlement risk during the past decade throughtheir participation in CLS Bank, which providespayment versus payment settlement of foreignexchange transactions, thus eliminating foreignexchange settlement risk. In addition, all of themajor Canadian banks continue to employ acomprehensive framework for managing thisrisk, involving governance, measurement, andcontrol. For some institutions, however, there isstill room for improvement in managing thistype of risk.61
Financial System Review – December 2007
Developments in ProcessingOver-the-Counter DerivativesNatasha Khan*
his article discusses the main findingsof the report New Developments inClearing and Settlement Arrangementsfor OTC Derivatives (CPSS 2007) in a
Canadian context. The complete report, pub-lished by the Committee on Payment and Settle-ment Systems (CPSS),1 is available on the websiteof the Bank for International Settlements (BIS).
The market for over-the-counter (OTC) deriva-tives has continued to grow exponentially inCanada and abroad.2 The size of the global OTCderivatives market, measured by notional amountoutstanding, increased at an average annual rateof 20 per cent between 1998 and 2005. As ofJune 2006, the total notional amount outstandinghad reached US$370 trillion worldwide.
By 2005, this rapid growth, coupled with limiteduse of automation for processing these transac-tions, had caused significant backlogs in tradeconfirmations. The backlog created uncertaintiesregarding counterparty risk and credit exposurefor major derivatives participants, thereby raising
1. The CPSS was established in 1990 as a permanentBIS committee reporting to the G-10 governors. TheCommittee contributes to strengthening the financialmarket infrastructure by promoting sound and effi-cient payment and settlement systems.
2. In a broad sense, an over-the-counter derivativescontract is a bilaterally negotiated transaction whosevalue depends on the value of one or more underlyingreference assets, rates, or indexes.
* The author was a member of the working groupestablished by the Bank for International Settlements(BIS) Committee on Payment and Settlement Sys-tems (CPSS) that published the report discussed inthis article on 16 March 2007. Members of the workinggroup included representatives from the G-10 centralbanks, the Hong Kong Monetary Authority, the U.K.Financial Services Authority, the U.S. Securities andExchange Commission, and the German Bundesan-stalt für Finanzdienstleistungsaufsicht (BaFin). Secre-tariat services were provided by the BIS.
T
issues concerning financial system efficiency andstability. Prudential supervisors began to expressconcern in early 2005. The situation, whichwas particularly serious in the market for creditderivatives, was highlighted in an industry-sponsored report published in July 2005.3InSeptember2005,prudential supervisorsbrought14 major derivatives dealers together at theFederal Reserve Bank of New York to encouragean industry solution. This prompted the 14 firmsto make a public commitment to decrease thebacklog in the credit derivatives market.
At the same time, central banks and prudentialsupervisors recognized that several recent devel-opments in the broader OTC derivatives marketwarranted further analysis. Thus, in February2006, the CPSS set up a working group, com-posed of member central banks and the pruden-tial supervisors of major derivatives dealers, tofollow up on issues identified in an earlier re-port (CPSS 1998) and to identify and analyzeany new issues raised by more recent develop-ments.
The working group conducted interviews withderivatives dealers in each jurisdiction and metwith industry groups, trade organizations, andinfrastructure service providers. The resultingreport, which complements other supervisoryinitiatives, provides a comprehensive view ofexisting arrangements and risk-managementpractices in the broader OTC derivatives market.
The report concludes that, although the infra-structure for processing OTC derivatives hasstrengthened since 1998, further action is neededto ensure that all OTC derivatives trades areconfirmed in a timely fashion, to identify steps
3. The report (Counterparty Risk Management PolicyGroup II 2005) called for an industry round table toaddress the serious and growing backlog of uncon-firmed trades in the credit derivatives market.
63
Policy and Infrastructure Developments
to mitigate the potential market impact followingthe close-out (default) of one or more majormarket participants, and to achieve open accessto essential post-trade services and efficient con-nectivity between service providers.
The study analyzes six main issues: documenta-tion backlog, use of collateral, use of centralcounterparties, prime brokerage, novations, andclose-out. The first three issues were raised inthe CPSS report published in 1998. The lastthree have been identified as relevant because ofdevelopments in the marketplace since thattime.
Major Issues
Documentation backlog
Unsigned master agreementsA master agreement sets the terms and condi-tions for all, or for a defined subset of, transac-tions into which two parties, such as a dealerand an institutional investor, may enter. Thepractice of executing trades before signing amaster agreement may create legal risk by jeop-ardizing a firm’s ability to close out and nettransactions in the event of a counterparty’sdefault.4
In Canada, a master agreement must be in placeto achieve the benefits of netting in the event ofcounterparty default.5 Canadian insolvencystatutes protect close-out netting for eligiblefinancial contracts (EFCs).6 (See “ImportantFinancial System Developments,” on p. 31.)
In contrast to 1998, virtually all the internationaldealers surveyed for this report have signed masteragreements with each other. Dealers report thatthe majority of existing unsigned master agree-
4. Netting essentially means offsetting positions or obli-gations with a particular counterparty, so that lossesincurred on one contract can be offset by gains onother contracts in the event of counterparty default.Data from U.S. commercial banks show that nettingdecreased counterparty exposures by 85 per cent as ofJune 2006.
5. In the United States and the United Kingdom, legisla-tion provides a strong case for the non-defaultingparty to close out and net swap agreements in theevent of a counterparty default, even in the absenceof a signed master agreement.
6. In March 2007, the Canadian federal governmentintroduced amendments to various acts that modernizeCanadian insolvency laws with respect to EFCs.
64
ments are with clients who have executed onlyone trade and, hence, would not benefit fromnetting.
Moreover, most G-10 dealers surveyed havepolicies in place to reduce the risks associatedwith unsigned master agreements. The most im-portant policy limits the number of trades thatcan be executed with a particular counterpartyin the absence of a signed master agreement.Most dealers require a master agreement to besigned before the first trade with non-investment-grade counterparties and before the secondtransaction with others. Where a master agree-ment has not been signed, dealers typically usea “long-form confirmation,” which incorpo-rates the industry standard form of masteragreement in the confirmation. In addition,dealers routinely monitor backlogs of unsignedagreements and prioritize efforts to completedocumentation based on risk of, and exposureto, a particular counterparty.7
Outstanding confirmationsOral contracts are legally enforceable in mostjurisdictions, including Canada. Thus, althougha written confirmation is best practice, failure toconfirm a trade in writing does not make thetrade unenforceable. Recordings of phone con-versations, emails, information from brokers(for brokered trades), exchange of payments, ormargin (collateral) can serve as evidence to provethe existence of a trade. But even if the existenceof a transaction is not in question, the details ofa trade may later be disputed between counter-parties. In addition, unconfirmed trades mayallow errors in the books and records of a firmto go undetected, leading to an incorrect mea-surement of counterparty credit risk. This mayresult in payment and margin breaks.8 There-fore, a written confirmation detailing the termsof the trade is the best practice.
In 1998, dealers reported hundreds of outstandingconfirmations, with a significant portion out-standing for 90 days or more. Survey data col-lected by the International Swaps and DerivativesAssociation (ISDA 2006) show that the number
7. Dealers also have the option of suspending tradingwith a counterparty that has not signed a masteragreement.
8. A “payment break” refers to the failure to receive anexpected payment or the receipt of an unexpectedpayment. A “margin break” refers to disagreementsabout the amount of collateral required.
Financial System Review – December 2007
of outstanding confirmations continued to riseuntil late 2005, when the issue began to receiveincreased attention from the industry. Data forthe 2006 calendar year suggest that outstandingconfirmations have decreased across all assetclasses of OTC derivatives at large firms, butoutstanding confirmations for interest rate de-rivatives have increased at small and medium-sized firms.9
In Canada, anecdotal evidence suggests that,because of lower trading volumes, Canadiandealers have not experienced the huge backlogsin credit derivatives encountered by their U.S.counterparts. However, the number of out-standing confirmations for interest rate swapshas increased over the past year across the bigsix Canadian banks.
Interviews with dealers across the G-10 coun-tries indicate that, in the short run, firms usevarious procedures to mitigate the risks arisingfrom unconfirmed trades. Many dealers verifythe key economic terms of a trade shortly afterexecution while the confirmation is outstanding.Some firms believe that this step, known as“economic affirmation,”10 is important, butothers feel that completing a full confirmationas soon as possible is more beneficial becausenon-economic terms such as “business-day con-ventions,” “holidays,” etc., can lead to problemsat other stages of the trade cycle. The CPSS studystates that, despite the divergent views on themerits of economic affirmation, this is an im-portant risk-mitigation tool if full confirmationis not expected to occur promptly, especially forcomplex products where full confirmation cantake 30 days or more.
Most of the international dealers surveyed rou-tinely monitor backlogs of outstanding confir-mations and report progress to senior manage-ment. They have policies in place to prioritizeand escalate efforts to complete confirmationsbased on metrics such as days outstanding (age)and the value of the transaction.
9. The survey defines large firms as those with morethan 1,500 deals per week, medium firms as thosewith fewer than 1,500 but more than 300 deals perweek, and small firms as those with fewer than300 deals per week.
10. “Economic affirmation” (also known as trade verifi-cation) is the process through which counterpartiesverify approximately a dozen key economic details ofa trade. This additional step is taken before the twoparties begin to review the full terms of a trade.
The industry recognizes that, in the long run,manual procedures for obtaining confirmationsare not feasible for more standardized products,given the large volume of trades, and that auto-mation is the key to managing confirmations.Electronic confirmation platforms currently op-erating include Deriv/SERV, which appears tobe the dominant platform for credit derivatives;SwapsWire, which is seen as the preferred plat-form for interest rate swaps; SWIFTNet Accord,which is being used for foreign exchange and in-terest rate derivatives; and eConfirm, which of-fers confirmation services for OTC commodityderivatives.
Most Canadian dealers are using Deriv/SERV toconfirm credit derivative trades with their inter-dealer counterparties. However, adoption of au-tomated confirmation services for interest rateswaps has been slower in Canada than in someother G-10 countries. When the CPSS study waspublished earlier this year, Canadian dealerswere not using an automated service for confir-mations of interest rate swaps; confirmationswere being communicated by fax. Non-Canadiandealers were using SwapsWire to confirmCanadian-dollar swaps, however. While Cana-dian dealers recognize the operational efficiencyprovided by automated confirmation services,they note that the benefits of joining such a ser-vice are limited unless counterparties are alsousing the service. Since the publication of thereport, the first Canadian dealer has joinedSwapsWire.
Use of collateral
The use of collateral to mitigate counterpartycredit risk has increased dramatically since the1998 report. Collateralization has been adoptedin all major jurisdictions worldwide. At the endof 2005, in excess of US$1.3 trillion was postedin collateral to support exposure to OTC deriva-tives versus US$200 billion in 2000. The numberof collateral agreements11 has increased evenmore dramatically, from 12,000 to 110,000,over the same time period.
Collateral decreases credit risk, but it does noteliminate it. Credit risk is the risk that a
11. Collateral agreements are legal agreements that gov-ern the use of collateral in OTC derivatives transac-tions. Most collateral agreements are documentedusing the master agreement’s credit support annex(CSA).
65
Policy and Infrastructure Developments
counterparty will not settle an obligation forfull value when it is due or at any time thereaf-ter. Using collateral decreases regulatory capitaland frees up bilateral counterparty credit lines,making it possible to continue trading activity.But market movements and delays in mark-to-market valuations or margin calls can lead touncollateralized exposures.
While the use of collateral reduces credit risk, itcan increase legal, custody, operational, market-liquidity, and funding-liquidity risks. Legal riskis the risk of loss because a contract cannot beenforced or because a law or regulation is beingapplied in an unexpected manner. Custody riskis the risk of losing securities held in custody be-cause of the insolvency, negligence, or fraudu-lent action of the custodian. Operational risk isthe risk of unexpected loss caused by deficien-cies in information systems or internal controls.Market-liquidity risk is the risk of loss due to adecline in the market value of the collateral,while funding-liquidity risk is the risk that acounterparty will experience demands for col-lateral that are too large to meet when due.
Dealer interviews suggest that significant progresshas been made since 1998 in reducing legal,custody, operational, and market-liquidity risksassociated with the use of collateral. There is ahigh degree of confidence in the legal enforce-ability of collateral agreements. Enhancementsin collateral-management systems have decreasedcustody and operational risks. Market-liquidityrisk is typically addressed by adequate haircutsand frequent mark-to-market valuations. Theeffectiveness of market participants’ efforts tomanage funding-liquidity risk is, however, moredifficult to assess, partly because it tends to crys-tallize during stressed market conditions.
The CPSS report notes that the issue of incorpo-rating collateral demands into a firm’s overallrisk-management procedures needs continuedattention from market participants.
Central counterparty
A central counterparty (CCP), such as a clearinghouse, is counterparty to both sides of a trade;that is, a seller to every buyer and a buyer toevery seller.
Central clearing of OTC derivatives was quitelimited at the time of the 1998 report. InSeptember 1999, SwapClear was launched as aclearing house for interdealer single-currency
66
interest rate swaps. As of December 2006, Swap-Clear had cleared US$35.5 trillion in swaps.This was nearly 40 per cent of the global inter-dealer market for interest rate swaps in 2006.12
Canadian dealers are not currently members ofSwapClear.13 One of the key benefits of a CCPis multilateral netting,14 which has the poten-tial to reduce members’ credit exposures relativeto those that exist in bilateral deals. It can beargued, however, that these benefits are reducedbecause a CCP is unlikely to clear the full rangeof OTC derivatives products. This could poten-tially increase the credit exposures of the re-maining, more complex, bilateral deals. Recentinterviews with dealers suggest that this concernhas decreased since 1998, and that most dealersdo not view the limited coverage of SwapClearas materially affecting their decision to use theservice.
Some market participants, including some Ca-nadian dealers, believe that the primary benefitsof a CCP are purely operational rather than creditrelated and that many of the operational benefitscan be realized through other services. For in-stance, TriOptima’s triReduce service, which isbeing used by Canadian dealers, has been citedas providing large operational gains by elimi-nating trades through a multilateral, voluntarytermination service.15 Deals that are removedfrom the portfolio do not have to be collateralized,and they do not require further payments,
12. In the autumn of 2006, the Canadian DerivativesClearing Corporation (CDCC), a wholly owned sub-sidiary of the Montréal Exchange, launched Con-verge, a clearing service for combining exchange-traded and OTC equity derivatives.
13. SwapClear currently has 20 members, includingsome of the largest international derivatives dealers.
14. Arithmetically, netting on a multilateral basis isachieved by summing each participant’s bilateral netpositions with the other participants to arrive at amultilateral net position, which represents the bilat-eral net position between each participant and thecentral counterparty. This allows a reduction in coun-terparty risk.
15. triReduce provides a service through which partici-pants identify trades that they wish to remove fromtheir balance sheets, subject to a set of constraints(tolerances) relating to changes in counterparty creditexposure, market risk, and cash payments. triReducematches the identified trades with those of other par-ticipants and terminates offsetting positions, whilemaintaining the participant’s predefined tolerances.
Financial System Review – December 2007
which reduces both margin and paymentbreaks.
From a systemic perspective, a CCP concen-trates risk and risk management. Thus, its po-tential to reduce systemic risk depends on theeffectiveness of its risk-management procedures.A CCP for OTC derivatives faces two particularrisk-management issues. First, more complexOTC derivatives products require the use ofcomplex pricing models that involve modelrisk. Second, default procedures must accom-modate the inherent illiquidity of OTC deriva-tives instruments.
In recognition of these issues, SwapClear haslimited its service to less complex, single-curren-cy interest rate swaps and has adapted its defaultprocedures accordingly. Market participantsmust recognize the differences between thedefault procedures adopted by a CCP for OTCderivatives and traditional procedures used byCCPs for exchange-traded derivatives.16 Mem-bers of a CCP should also be comfortable withthe valuation models used by the CCP to pricepositions, since margin requirements will bebased on prices derived from these models. Thiswill affect the cost and risk of participation inthe CCP.
Prime brokerage
In a prime brokerage arrangement, a primebroker agrees to intermediate specified eligibletransactions between a client, such as a hedgefund, and a list of approved executing dealers.
Prime brokerage services have been offered forcash equity, fixed-income securities, and foreignexchange products for some years, but primebrokerage for OTC derivatives is a very recentphenomenon. At present, the service is offeredby only a very small number of large interna-tional dealers and is geared specifically to thehedge fund community.
A prime brokerage service for derivatives allowsa hedge fund to enter into trades with multipleexecuting dealers while using the back-officesystems of a single prime broker to clear and set-tle those trades, thus providing operational effi-ciency. The service can also decrease the hedgefund’s margin requirements, because all eligibletrades are subject to bilateral netting.
16. See CPSS (2007) for a detailed discussion of thedefault-management process adopted by SwapClear.
In a typical prime brokerage arrangement forderivatives, once the executing dealer and thehedge fund have agreed to a trade, each mustnotify the prime broker of the terms. If theprime broker accepts the trade, it becomes acounterparty to two back-to-back trades, onewith the hedge fund and the other with theexecuting dealer.
Canadian dealers are not currently offeringprime brokerage services for derivatives,17 butthey do serve as executing dealers in prime bro-kerage arrangements.
The 2007 CPSS report states that all partiesinvolved in a prime brokerage arrangementshould carefully assess the legal documentationand understand their rights and responsibilities.
Novations
A novation is the replacement of a contract be-tween two initial counterparties to an OTC de-rivatives trade (the transferor and the remainingparty) with a new contract between the remainingparty and a third party (the transferee).
Novations were rare in 1998, but the practicehas increased with the growth of the hedge fundsector. When a hedge fund seeks to exit an OTCderivatives position, it often does so through anovation rather than by negotiating a termina-tion of the contract (which may require thefund to accept the price offered by the originalcounterparty) or by entering into an offsettingcontract (which is likely to create additionalcounterparty exposure).
Standard master agreements allow novations aslong as the transferor obtains written consentfrom the original counterparty prior to the trans-fer. Without written consent, the remaining partyhas full discretion to reject the proposed nova-tion. Dealers, however, were frequently acceptingnovations of credit derivatives without priorconsent. This was causing errors in measuringcounterparty credit risk, as well as causing pay-ment and margin breaks. The practice was oneof the major factors contributing to the huge
17. Derivatives prime brokerage places very large demandson the prime broker’s back-office systems, and, asstated earlier, this service is currently offered only bysome of the largest international dealers. Canadiandealers offer prime brokerage services for foreignexchange products, cash equity, and fixed-incomesecurities.
67
Policy and Infrastructure Developments
backlog of outstanding confirmations in thecredit derivatives market.
In the autumn of 2005, a group of dealers an-nounced their support for a novation protocolcrafted by ISDA for credit and interest rate deriv-atives. The protocol requires the transferor toobtain written consent from the original coun-terparty before 18:00 (in the location of thetransferee) on the day that the novation is ini-tiated. If consent is not obtained, the transferoris deemed to have two contracts, one with theoriginal counterparty and one with the transferee.
All the dealers interviewed, including Canadiandealers, have adopted the protocol, which hasbeen effective in achieving prompt notificationand consent. The 2007 CPSS report notes thatif novations become common for instrumentsother than credit and interest rate derivatives,the protocol will need to be extended to includethese products.
Close-out
Close-out netting is an arrangement to settle allcontracted, but not yet due, obligations to andclaims on a counterparty by a single payment,immediately upon the occurrence of one of thedefault events defined in the relevant documen-tation. Close-out netting provisions in masteragreements have been identified as a powerfultool for mitigating counterparty risk. At the timeof the 1998 report, some dealers had expressedconcerns about the enforceability of netting pro-visions. Recent discussions with dealers, howev-er, suggest that these concerns have diminishedconsiderably, because many jurisdictions havepassed legislation supporting close-out net-ting.18
Since 1998, however, concerns have arisen aboutthe potential for significant market disruptionsin the event of the close-out of a major marketparticipant, especially if it occurs at a time whenmarkets are already under stress.19
18. As stated earlier, close-out netting is supported byCanadian insolvency statutes for eligible financialcontracts.
19. Fear of major market disruptions caused by the closingout and replacement of positions with Long-TermCapital Management (LTCM) prompted a consor-tium of LTCM’s counterparties to recapitalize thefund, thereby preventing a close-out.
68
Market participants have identified two stepsthat can help mitigate the potential impact of amajor close-out. The first is to ensure timely andaccurate information on counterparty credit ex-posures to major participants. Regular portfolioreconciliation20 can facilitate this step. The secondstep is the routine identification of trades thatcan be voluntarily terminated in order to reducepositions that would need to be replaced fol-lowing a default. This can be accomplishedby using services, such as triReduce, that offermultilateral voluntary termination of trades.
Overall Evaluation
The clearing and settlement infrastructure forthe OTC derivatives market has been signifi-cantly strengthened since 1998.
Nevertheless, more progress is needed in someareas. Firms need to extend the successful effortsto decrease confirmation backlogs in credit de-rivatives to other OTC derivatives products sothat, over time, all standardized OTC derivativestrades are confirmed within five days of the tradedate (T+5), and complex, non-standardized tradesare confirmed within 30 days of the trade date(T+30). The use of automated systems to con-firm trades, whenever possible, will help ac-complish this goal and help prevent a futurebuildup of confirmation backlogs. Risks of ex-isting unconfirmed trades can be mitigatedby broader use of economic affirmations, asdiscussed earlier.
Anecdotal evidence suggests that Canadiandealers have not experienced significant confir-mation backlogs in the credit derivatives market,but the number of outstanding confirmationshas increased for interest rate swaps across thebig six Canadian banks over the past year. Cana-dian dealers have not moved quickly to adoptautomated services for confirming interest rateswaps, compared with dealers in some otherG-10 countries. Increased use of automation inconfirming interest rate swaps will help Canadiandealers confirm these trades in a timely fashionand will prevent a future backlog.
The 2007 CPSS report notes that daily portfolioreconciliation with active counterparties is ap-propriate for firms that are frequently involved
20. Portfolio reconciliation involves verifying the exist-ence of all outstanding trades and comparing theirprincipal economic terms.
Financial System Review – December 2007
in novations, terminations, or amendments ofcontracts. This will help ensure that firms haveaccurate records on their counterparty credit ex-posures. The report also concludes that marketparticipants should work together to identifyfurther steps that can mitigate the potential mar-ket impact of the close-out of one or more majormarket participants.
Over time, market infrastructure will continueto evolve. With increased centralization, openaccess to essential post-trade services and conven-ient connectivity to their systems will assumegreater importance.
Centralized processing of trades and post-tradeevents may leave the infrastructure more sus-ceptible to disruptions at single points of failure.Supervisors and central banks will need to de-termine whether existing standards for opera-tional reliability of securities settlement systemsand CCPs (CPSS-IOSCO 2001 and 2004) needto be applied to providers of clearing and settle-ment services for OTC derivatives that are notalready subject to these standards.
In addition, if an entity other than a CCP startssettling payments associated with OTC deriva-tives on a multilateral net basis, central banksand supervisors will need to consider whetherprinciples for systemically important paymentsystems (CPSS 2001) should be applied to themoney settlement arrangements.
References
Committee on Payment and SettlementSystems (CPSS) and the Euro-CurrencyStanding Committee of the Central Banksof the Group of Ten Countries. 1998.OTC Derivatives: Settlement Procedures andCounterparty Risk Management. Basel: Bankfor International Settlements. Available at<http://www.bis.org/publ/cpss27.pdf>.
Committee on Payment and Settlement Sys-tems (CPSS). 2001. Core Principles for Sys-temically Important Payment Systems. Basel:Bank for International Settlements.
———. 2007. New Developments in Clearing andSettlement Arrangements for OTC Deriva-tives. Basel: Bank for International Settle-ments. Available at <http://www.bis.org/publ/cpss77.pdf>.
Committee on Payment and Settlement Systems:Technical Committee of the InternationalOrganization of Securities Commissions(CPSS-IOSCO). 2001. “Recommenda-tions for Securities Settlement Systems.”Bank for International Settlements andInternational Organization of SecuritiesCommissions.
———. 2004. “Recommendations for CentralCounterparties.” Bank for InternationalSettlements and International Organiza-tion of Securities Commissions.
Counterparty Risk Management Policy GroupII. 2005. “Toward Greater Financial Stabil-ity: A Private Sector Perspective.” 27 July.
International Swaps and Derivatives Associa-tion (ISDA). 2006. “ISDA 2006 Opera-tions Benchmarking Survey.”
69
Financial System Review – December 2007
Management of Foreign ExchangeSettlement Risk at Canadian BanksNeville Arjani
n a standard foreign exchange transaction,parties to a trade agree to exchange valuedenominated in one currency for valuedenominated in another currency. The
transfer of funds to settle each party’s paymentobligation typically takes place within the rele-vant payments systems of the currencies involvedin the trade. Settlement of foreign exchange (FX)trades across national payments systems andlegal jurisdictions can expose banks to differenttypes of risk. The principal types of risk includecredit, liquidity, operational, and legal risk.1
Together, these risks comprise foreign exchangesettlement risk.
The focus here is on the credit risk dimension offoreign exchange settlement risk, henceforth re-ferred to as “FXSR.” A bank that irrevocably paysout the sold currency to its counterparty uncon-ditional upon final receipt of the currency it haspurchased is exposed to financial loss up to theprincipal value of the trade if its counterpartyfails to deliver the purchased currency. That is, abank is exposed to FXSR if settlement does nottake place on a payment-versus-payment (PvP)basis.
Given the global scope of the FX marketplace,trades often settle across international timezones. Differences in time zones could exacer-bate a bank’s exposure to FXSR, since it may berequired to pay out the sold currency before thebusiness day begins in the country of the currencyit has purchased.2 Thus, exposure to FXSR couldlast up to two business days, and possibly longer,when settlement is interrupted by weekends andholidays. At any given time, therefore, the valueof a bank’s exposure to a single counterparty
1. For a description of these and other risks, see Aaron,Armstrong, and Zelmer (2007).
2. Settlement of each currency leg must take place in thecountry or region where the currency is issued.
I
could equal two or more days' worth of trades,potentially exceeding the value of its capital(CPSS 1996). With almost US$4 trillion settlingdaily in the FX marketplace, large counterpartyFXSR exposures are likely to exist.This report highlights the key aspects of FX set-tlement and banks’ management of FXSR. Avail-able methods for settling FX trades and the riskcharacteristics of each are discussed. The neces-sary components of an effective risk-managementstrategy for individual banks are outlined. Thereport goes on to discuss how Canada’s majorbanks use these settlement methods and risk-management strategies.
Methods of Foreign ExchangeSettlement and AssociatedRisk
FX trades are usually settled using one of fourmethods, each of which is characterized by adifferent degree of risk.
Gross non-PvP settlement
Under this settlement arrangement, paymentobligations relating to each currency leg of anFX trade are transferred individually throughnational payments systems. Where a bank doesnot participate directly in the national paymentssystem for currencies that it actively trades in,it must rely on a correspondent (or nostro)bank to settle its payment obligations in thosecurrencies.
When settling trades using gross non-PvP settle-ment, a bank’s delivery of the sold currency isgenerally not made conditional on final receiptof the purchased currency. This exposes the bankto financial loss up to the principal value of thetrade until final settlement.
71
Policy and Infrastructure Developments
On-us settlement
On-us settlement takes place where both cur-rency legs of a trade are settled across the booksof the same bank. This could be a scenario inwhich a bank is settling an FX trade across itsbooks on behalf of two of its clients, or wherethe settlement bank is a counterparty to the trade.That is, one party to the trade is a client of theother party. This report deals with the latterarrangement only, where the settlement bank isa counterparty to the trade.3
When settling an FX trade on-us, a transfer offunds through national payments systems is notnecessary. Nonetheless, on-us settlement canalso expose a bank to FXSR, especially where thetrade is booked by the settlement bank acrosstime zones in separate subsidiaries or branches.If the bank credits the sold currency to its client’saccount prior to debiting the bought currencyfrom the client’s account, the bank is exposed toFXSR up to the principal value of the trade. Thisis because there is a possibility that the clientwould not have sufficient funds available to meetits obligation and that the bank would be unableto retrieve the sold currency.
Bilateral netting agreement
This method involves the netting of individualpayment obligations in the same currency stem-ming from two or more underlying FX tradesthat are due to settle on the same date. Bilateralnetting of payment obligations between a par-ticular pair of banks typically involves one banksending a single net payment in the respectivecurrency to the other, rather than settling eachtrade between them individually. Net paymentobligations are settled on a non-PvP basis.
To better understand bilateral netting, considerthe following example. Suppose that Bank A owesBank B individual amounts of 50 and 100 incurrency X, stemming from two trades betweenthem. In addition, Bank B owes Bank A 125 incurrency X to settle a third trade between them.All three trades mature on the same date and areeligible to settle under the bilateral netting agree-ment established between the banks. Bilaterallynetting these trades results in Bank A owing a
3. The larger a bank is, and the more extensive its clientbase and FX operations are, the greater the scale of itson-us settlement activity will likely be.
72
single amount of 25 in currency X to Bank B,while Bank B’s payment obligation to Bank Ain currency X is eliminated altogether.
Banks typically maintain bilateral netting agree-ments with certain counterparties. Provided thatan agreement is legally enforceable in all rele-vant jurisdictions, bilateral netting has the po-tential to reduce counterparty credit risk, butmay not eliminate it completely, as demonstratedin the example above. That is, under a legallyvalid bilateral netting agreement, a bank is ex-posed to FXSR vis-à-vis its counterparty for anamount equal to the net value owing from alltrades in the purchased currency.
Continuous linked settlement(CLS Bank)
CLS Bank owns and operates an electronic infra-structure linking together fifteen national pay-ments systems, including Canada’s Large ValueTransfer System, in real time.4 This arrange-ment—Continuous Linked Settlement (CLS)—facilitates the simultaneous (PvP) settlement ofeach currency leg for accepted FX transactionson a trade-by-trade basis. By employing specificrisk controls to limit participants’ exposure inthe system, CLS virtually eliminates credit riskassociated with settling foreign exchange trans-actions. Further, since participants’ settlementobligations to the system are calculated on amultilateral net basis, CLS also economizes onsettlement funding.
Management of FXSR
Banks exposed to FXSR are encouraged to havein place an appropriate risk-management frame-work to contend with this exposure. However,previous surveys conducted by the Bank forInternational Settlements (CPSS 1996 and 1998)found that some banks did not recognize theirexposure to FXSR as being similar to other creditexposures, and thus were not taking appropriateaction to manage it.
4. The CLS Bank began operations in September 2002.The Canadian-dollar leg of CLS is subject to Bank ofCanada oversight under the Payment Clearing andSettlement Act. For more information on CLS Bankand the Bank of Canada’s oversight of CLS, see Millerand Northcott (2002). For a more recent update onCLS oversight, see Goodlet (2007).
Financial System Review – December 2007
Citing the large scale of the FX trading and set-tlement activity by banks, and the resulting sizeand systemic implications of their exposures toFXSR, these studies set out a strategic frameworkfor action on the part of individual private banks,central banks, and industry groups. Indeed, theintroduction of the CLS Bank was one responseat the industry-group level to this call for action.
For an individual bank, a framework for managingFXSR should incorporate the following elements:a corporate governance structure that acknowl-edges exposure to FXSR; accurate measurementof the exposure associated with each settlementmethod; and the use of appropriate tools tocontrol this exposure where it exists.
Acknowledgement of exposure
Exposure to FXSR should be recognized as ashort-term credit exposure for a bank. To thatend, clear lines of responsibility should be es-tablished for managing this exposure through-out the organization, including the involvementof senior management.
Measurement of exposure
Banks should acknowledge the degree of exposureassociated with each settlement method. That is,they should recognize exposure to FXSR whensettling trades using gross non-PvP settlement(includingsettlementofbilaterallynettedamountsowing) and on-us arrangements involving non-PvP settlement. At the same time, they shouldalso recognize that certain settlement methods,such as CLS and on-us arrangements providingPvP settlement, can virtually eliminate FXSR.
For settlement methods that expose them toFXSR, banks should employ a measurement mech-anism that accurately gauges the extent of thisexposure, where exposure has both a value andduration element. For example, with gross non-PvP settlement, the value of a bank’s exposureto FXSR should be measured as the principalamount of the trade.
With regard to the duration of exposure, a bankshould be able to identify its minimum andmaximum exposure associated with settlinggross non-PvP. A bank’s minimum exposure isdefined as the period of time between whenpayment of the sold currency becomes unilater-ally irrevocable to when the purchased currency
is expected to be received with finality.5 Ofcourse, it may not be possible for a bank to verifyfinal receipt of the purchased currency immedi-ately, especially where a correspondent bank isreceiving these funds on its behalf. Until receipthas been confirmed, there is a possibility that acounterparty could default on its obligation.Thus, a bank’s maximum exposure is defined asthe length of time between when delivery of thesold currency becomes unilaterally irrevocableto when the bank is able to verify its final orfailed receipt of the purchased currency. Onlywhen non-receipt of payment within the allottedtime frame has been verified can a bank takeaction to recover settlement losses.6
Control of exposure
Once identified and measured, procedures shouldbe put in place to limit exposure to FXSR withinparameters that are acceptable to the bank. Forexample, this could include the use of daily set-tlement limits for FX counterparties. A daily set-tlement limit (DSL) granted to a counterpartyrepresents the maximum receivable (i.e., pur-chased) currency settlement position vis-à-visthat counterparty that the granting bank is willingto incur on a given day. DSLs are more effectivein limiting exposure when they are binding(e.g., pre-trade authorization by the credit-riskdepartment is necessary for anticipated limitbreaches before a trade can be confirmed). Fur-ther, counterparty exposures against these limitsshould be monitored and updated in real timeon a global basis (i.e., limits should be enforce-able across all of a bank's trading centres).
An institution should also employ a reporting andfollow-up procedure to deal with a counterparty’sfailure to deliver the purchased currency as ex-pected. For instance, a counterparty may experi-ence an internal operational problem that tempo-rarily prohibits it from transferring funds throughthe payments system. Alternatively, a counter-party may suffer from a more serious liquidityproblem that prevents it from meeting some or
5. Finality refers to the unconditional and irrevocablereceipt of funds.
6. As alluded to earlier, with gross non-PvP settlementacross international time zones and/or where corre-spondent banks are involved, there is a possibilitythat the banks’ exposure to FXSR could be greatlyincreased.
73
Policy and Infrastructure Developments
all of its payment obligations over a longer timeframe. Regardless of the cause, a failed traderepresents continued exposure to the counter-party for the principal value of the trade. Hence,banks should account for failed trades in theirmeasurement and control of FXSR.
The Canadian Environment:Stylized Facts
In 2006, the Bank of Canada, in conjunctionwith several other central banks, organized andconducted a survey of financial institutionsregarding their use of various FX settlementmethods and their FXSR management strate-gies.7 Canada’s major banks participated in thesurvey.8 The survey is intended to identifychanges in the use of available FX settlementmethods and to assess progress in managingFXSR exposure since the CPSS-BIS survey pub-lished in 1998. The FX settlement landscape haschanged considerably since then, particularlywith the introduction of the CLS Bank.
The survey consists of two sections. The firstasks respondents to report on average daily FXsettlement values according to currency, coun-terparty type, and settlement method for April2006. The second section consists of questionspertaining to the measurement and control ofFXSR. The survey covers settlement of FX spot,forward, and swap transactions.
Some stylized facts from the survey of majorCanadian banks are as follows.9
• The average daily FX settlement value(in terms of currency sold) reported bythe Canadian banks in April 2006 was
7. The survey was administered by member centralbanks of the BIS Committee on Payment and Settle-ment Systems (CPSS) Sub-Group on FXSR. The sub-group released a consultative report based on thesurvey findings in July 2007. The report is availableat <http://www.bis.org/publ/cpss81.htm>.
8. Surveyed banks include the Bank of Montreal, Scotia-bank, the Canadian Imperial Bank of Commerce(CIBC), National Bank of Canada, Royal Bank ofCanada, and TD Bank Financial Group.
9. Where appropriate, comparisons have been providedbetween the current survey findings and the surveyfindings from 1998. In some cases, certain factorspreclude an accurate comparison of these findings.
74
US$98.3 billion.10 The settlement value ofCanadian banks represents close to 3 percent of the total daily FX settlement value forall institutions participating in the survey.
• Settlement value of the Canadian banks isheavily concentrated in the U.S. dollar(US$), the Canadian dollar (Can$), and theeuro (EUR) (Table 1). Approximately 85 percent of daily settlement value involves thesecurrencies. Although the survey did not col-lect information on trade volumes for spe-cific currency pairings, these results suggestthat the majority of FX trades by Canadianbanks are US$/Can$ and US$/EUR. In1998, trades involving the U.S. dollar andthe Canadian dollar accounted for a slightlygreater combined proportion of the banks’daily settlement value.
• Overall, gross non-PvP settlement continuesto represent the largest source of exposure toFXSR for Canadian banks; however, its promi-nence as a settlement method has declinedsince the introduction of the CLS Bank.Gross non-PvP settlement currently accountsfor 55 per cent of daily settlement value(Table 2), compared with over 80 per centin 1998.
• Close to 23 per cent of the aggregate dailyFX value settled by Canadian banks wentthrough CLS. This accounted for about50 per cent of the daily FX value for thethree Canadian banks participating in thissystem in April 2006.
• Roughly 30 per cent of banks’ daily FX set-tlement value was bilaterally netted. Thispercentage was greater for non-CLS partici-pants (54 per cent) than for CLS participants(15 per cent). The proportion of total creditexposure eliminated by bilateral netting was17 per cent, which is similar to the percentagereported in the 1998 survey.
• On-us settlement was equal to 5 per cent ofdaily FX settlement value. This value is heavily
10. This does not necessarily include all FX tradesbooked by each bank, since the survey focused pri-marily on trades booked within Canada. However,some banks did provide figures for trades bookedoutside of Canada as well.
Financial System Review – December 2007
Table 1
Daily FX Settlement Value by Currencya
April 2006, percentage
a. In terms of currency sold. Similar figures emerge for currency purchased.
Total 100
U.S. dollar 47
Canadian dollar 31
Euro 7
Japanese yen 4
U.K. pound 3
Australian dollar 3
All others 5
Table 2
Canadian Banks’ Use of Available Settlement Methods
April 2006, percentage
Daily FX Settlement Value: 100(US$98.3 billion)
Proportion of which was:
- settled gross non-PvP 55
- extinguished by bilateral netting 17
- settled on-us 5
- settled in CLS 23
concentrated in the Canadian dollar and theU.S. dollar, with a limited amount in the euro.
The Canadian Environment:Management of FXSR
The survey also shed light on Canadian banks’management of exposure to FXSR.11
Acknowledgement of exposure
All of the Canadian banks surveyed view theirexposure to FXSR as a short-term credit expo-sure and have established a comprehensiveframework for managing this risk. Clear linesof responsibility have been established withineach bank, including the involvement of seniormanagement.
Measurement of exposure
All of the Canadian banks surveyed recognizethat they are exposed to financial loss up to theprincipal value of each FX trade settling grossnon-PvP and also for on-us trades settling on anon-PvP basis. With respect to bilateral netting,all banks maintain master bilateral nettingagreements with certain of their counterpartiesand view these agreements to be legally binding.12
Accordingly, five of the six banks measure theamount of their credit exposure stemming frombilaterally netted trades as the net amount owingfrom the counterparty in the purchased currency.One bank measures its exposure as the grossvalue owing, solely for administrative reasons.
11. Views in this section of the article are based onspecific criteria identified by the CPSS subgroup—acknowledgement, measurement, and control ofexposure. A comprehensive judgment about the man-agement of FXSR at each institution would need tofactor in the broader framework within which riskmanagement takes place (e.g., an assessment of con-tingency planning and stress-testing procedures). Formore on this issue, see BCBS (2000), which discussessupervisory guidance for managing FXSR exposure. Adescription of Canadian banks’ broader risk-manage-ment practices can be found in Aaron, Armstrong,and Zelmer (2007).
12. Under these arrangements, the necessary legal docu-mentation, including an ISDA agreement, must besigned with each counterparty and acceptable legalopinions for each respective currency jurisdictionmust be received.
75
Policy and Infrastructure Developments
When measuring their FX settlement exposure,Canadian banks participating in CLS recognizethe benefit of this system in eliminating creditrisk.
Banks measure the duration of their exposure toFXSR when settling trades gross non-PvP (thelargest source of the banks’ exposure) as lastingbetween one and three calendar days, dependingon the institution. With data provided by thebanks on their timelines for gross non-PvPsettlement, each bank’s actual minimum andmaximum exposure to FXSR is calculated forits major currency pairings and is comparedwith its measured duration of exposure.13
This comparison reveals that two of the sixCanadian banks measure their FX settlement ex-posure in a way that covers both their minimumand maximum exposures for all major currencypairings settling gross non-PvP. Two banksmeasure their exposure in a way that coverstheir minimum but not their maximum expo-sure for some or all of the major currency pairings.And two banks measure their exposure in a waythat covers neither their minimum nor theirmaximum exposure for some or all of the majorcurrency pairings. A discussion of these findingsis presented in the next section.
Control of exposure
All of the major Canadian banks use daily set-tlement limits and apply them in a manner sim-ilar to that described earlier. Limits are bindingand are usually programmed directly into thebanks' internal credit-control systems so that allpotential FX contracts are automatically appliedagainst the respective DSL at the time of thetrade. DSLs are usually established within thebroad guidelines for granting counterparty creditset by senior management. That is, DSLs may beone of several corporate credit lines that a bankchooses to grant to its counterparty. DSL valuesare based on factors such as counterparty type,historical trading patterns, and projected busi-ness requirements. Limits are typically reviewedon an annual basis, but they may be reviewedmore frequently if necessary.
13. Major currency pairings are defined as those involvingCanadian dollars, U.S. dollars, or euros against eachother.
76
All banks have procedures in place to deal withfailed trades. These typically include generatinga formal report and distributing it to seniormanagement. For all but one bank, the failedcounterparty’s DSL may be reduced until thepurchased currency is received. All banks usediscretion in dealing with failed trades. For ex-ample, if the value of the failed trade(s) is largeenough, then the counterparty's DSL may beshut down completely, rather than just reduced.Banks typically encounter only a few failed FXtrades per week. Temporary operational issuesare the primary cause of these failures, and failedtrades are generally resolved quickly.
It should be pointed out that, given currenttimelines for gross non-PvP settlement, by thetime that banks are able to identify that a tradehas failed (usually on the day after the settle-ment date), it may already be too late to canceldelivery of the sold currency to the counterpartyfor trades settling on that day. It might also notbe possible to cancel delivery of the sold currencyfor trades settling on the following day. But thisdoes not apply to trades involving the U. S. dollar,the Canadian dollar, or the euro, which makeup the bulk of the settlement activity of majorCanadian banks. Of course, this is a “worst-case” scenario, because it assumes, among otherthings, that a bank becomes aware of a counter-party problem only upon identification of thefailed receipt, which is not likely to be the casein practice.
The Canadian Environment:Discussion
The introduction of CLS since the CPSS-BISsurvey in 1998 has led to a significant reductionin the degree of exposure to FXSR for participatingCanadian banks. Nevertheless, the prominentuse of gross non-PvP as a settlement methodmeans that all banks continue to be exposed toa substantial amount of FXSR. That said, themanagement of this exposure by Canadian banksappears to have improved since 1998, althoughfurther improvement by some banks is still pos-sible.
As observed in the 1998 CPSS-BIS report, Cana-dian banks continue to view their exposure toFXSR as a short-term credit exposure, and haveestablished a comprehensive framework formanaging this risk. Currently, the measurement
Financial System Review – December 2007
method used by two banks covers their maxi-mum exposure. This is a slight improvementfrom 1998, when only one of these banks mea-sured its exposure in this way. Other bankscould improve by tightening their timelines forgross non-PvP settlement where possible, there-by reducing the duration of their minimum andmaximum exposure.14
Improvements are also observed in banks’ ap-plication of DSLs. In 1998, all but one bankmonitored their exposures against these limitsin real time. Moreover, DSLs were enforced on aglobal basis by only four of the six banks. Cur-rently, all banks monitor their exposures in real-time and enforce counterparty DSLs on a globalbasis. Nonetheless, the procedures used by cer-tain banks to deal with failed trades could beimproved, as discussed earlier.
Participation in CLS
Since CLS virtually eliminates the credit risk as-sociated with FX settlement, central banks andsupervisory authorities, including the Bank ofCanada and the Office of the Superintendent ofFinancial Institutions, encourage banks to par-ticipate in and use this system (Goodlet 2006).
Three of the six major Canadian banks partici-pated in CLS at the time of the 2006 survey.Royal Bank was the only Canadian settlementmember of the CLS Bank, while National Bankof Canada and Bank of Montreal participated asthird parties. CIBC was in the process of becom-ing a settlement member at the time of the survey.
As noted earlier, 23 per cent of the total daily FXvalue at Canadian banks was settled throughCLS Bank in April 2006. Participating banksnoted that they settle as many trades as possiblethrough this system. There are, however, obsta-cles to greater use of CLS in Canada, largely re-lated to the settlement of same-day Canadiandollar/U.S. dollar trades. Typically, these tradesare agreed upon, settled, and reconciled allwithin the same business day, whereas CLS set-tlement, which occurs overnight in North America,is completed on the following day.
14. For example, a bank could extend the cancellationdeadline for paying out the sold currency with itscorrespondent bank, or perhaps identify its finaland failed receipts earlier.
Same-day settlement is estimated at between 35and 70 per cent of Canadian dollar/U.S. dollardaily settlement value, depending on the insti-tution. Banks not participating in CLS cite thelack of same-day settlement as significantly hin-dering the business case for their participation.Those banks participating in CLS share that con-cern, but feel that participation by Canadianbanks is important. All banks expressed a stronginterest in the possibility of multiple daily set-tlement sessions in CLS to accommodate settle-ment of FX trades for same-day value.
Regardless of its current inability to settle same-day trades, the use of CLS Bank by Canadianbanks continues to increase. CIBC began partic-ipating as a settlement member in September2006. Because CIBC is an important counter-party in the Canadian-dollar FX market, this isexpected to increase the total value settled throughCLS Bank by Canadian banks and by otherinternational users of CLS.15
Conclusion
Canada’s major banks are using a comprehen-sive framework to manage FXSR that focuses ongovernance, measurement, and control. Whilesome improvement has been observed since the1998 CPSS survey, there is still room for certainbanks to make further progress in managing thisrisk.
Gross non-PvP settlement continues to be theprimary source of exposure for Canadian banks;however, the proportion of their FX activity thatsettles through CLS Bank is increasing. Currently,four of the six major Canadian banks participatein this system. Greater use is hindered by theinability of CLS to settle same-day FX trades.
References
Aaron, M., J. Armstrong, and M. Zelmer. 2007.“An Overview of Risk Management atCanadian Banks.” Bank of Canada Finan-cial System Review (June): 39–47.
15. For a trade to be eligible for settlement through CLS,both counterparties must participate in the system.
77
Policy and Infrastructure Developments
Basel Committee on Banking Supervision(BCBS). 2000. Supervisory Guidance forManaging Settlement Risk in ForeignExchange Transactions. Basel: Bank forInternational Settlements (BIS).
Committee on Payment and Settlement Sys-tems (CPSS). 1996. Settlement Risk inForeign Exchange Transactions. Basel: Bankfor International Settlements (BIS).
———. 1998. Reducing Foreign Exchange Settle-ment Risk: A Progress Report. Basel: Bankfor International Settlements (BIS).
Goodlet, C. 2006. “Bank of Canada OversightActivities during 2005 under the PaymentClearing and Settlement Act.” Bank ofCanada Financial System Review (June):31–34.
———. 2007. “Bank of Canada Oversight Acti-vities during 2006 under the PaymentClearing and Settlement Act.” Bank ofCanada Financial System Review (June):33–37.
Miller, P. and C. A. Northcott. 2002. “CLS Bank:Managing Foreign Exchange SettlementRisk.” Bank of Canada Review (Autumn):13–25.
78
Research
Summaries
Financial System Review – December 2007
Introduction
ank of Canada staff undertakeresearch designed to improve overallknowledge and understanding of theCanadian and international financial
systems. This work is often pursued froma broad, system-wide perspective thatemphasizes linkages across the differentparts of the financial system (institutions,markets, and clearing and settlement systems),linkages between the Canadian financialsystem and the rest of the economy, andlinkages to the international environment,including the international financial sys-tem. This section summarizes some of theBank’s recent work.
Policies for providing liquidity vary from centralbank to central bank. But a generic feature of allthese policies is the restriction to a small numberof agents and a reliance on a market for liquidity.In the paper, The Provision of Central BankLiquidity under Asymmetric Information, JamesChapman and Antoine Martin consider a stylizedeconomy in which the central bank has less-pre-cise information about credit conditions than therest of the financial market to which it is providingliquidity. The authors find that the optimal policyin this model is for the central bank to restrict itsliquidity injections to the financial market to asmall subset of market participants and to makethese injections in a way that is sensitive to theunderlying market conditions. The authors thenbriefly describe the Bank of Canada’s policiesfor providing liquidity.
Collateral Portfolios and Adverse Dependenceby Alejandro García and Ramazan Gençaysummarizes the second of two papers that developa framework for calculating haircuts for assetsused as collateral. In their first paper (summarizedin the December 2006 FSR), García and Gençayproposed a framework for comparing different
B
methods of computing haircuts for individualassets. Particular attention was paid to selectinga method that would provide sufficient collateralin the case of low-probability events (large un-expected declines in asset prices) that might affectthe stability of the financial system while alsotaking into account the cost of pledging collateral.In the paper summarized here, they examine howhaircuts should be calculated in situations inwhich a variety of assets are pledged as collateral.This time, the focus is on the relationship amongthe prices of the different assets pledged as col-lateral and, in particular, how this relationshipcan change when markets are under stress. Thissituation is characterized as an event where thereis a change in the correlation among the returnsof the assets in the pool of collateral.In the article, Housing Market Cycles andDuration Dependence in the United Statesand Canada, Rose Cunningham and Ilan Koletexplore data on real house price cycles at theaggregate level and city level for the United Statesand Canada. Using data for 137 cities, the authorsexamine the duration, size, and correlations ofhousing market cycles in North America. Theyfind that North American housing cycles are long,averaging over five years of expansion and fouryears of contraction, and that there is a fairlyhigh degree of positive correlation in house pricecycles between U.S. and Canadian cities. Theauthors then estimate a model for expansionsand contractions in house prices that allowsthem to test for duration dependence. The resultssuggest that housing market expansions havepositive duration dependence—that is, the longerthat expansions in house prices continue, themore likely they are to end. At the same time,there seems to be no relationship between thelength of a contraction and the likelihood of itsending. Standard determinants of house prices(interest rates, income, and population growth)are included as controls.
81
Financial System Review – December 2007
The Provision of Central Bank Liquidityunder Asymmetric InformationJames T. E. Chapman and Antoine Martin*
entral banks provide liquidity in variouscontexts to promote the stability andefficient functioning of the financialsystem. While the exact institutional
aspects of liquidity provision vary among cen-tral banks, some basic features seem to be generic.First, the provision of central bank liquidity innormal periods is restricted to a small subset ofpossible agents who are encouraged to competefor liquidity with each other instead of automat-ically receiving liquidity from the central bank.Second, in extraordinary cases, the central bankhas the option of providing liquidity to a muchbroader range of agents, and this liquidity canbe provided independent of financial marketconditions.
This article summarizes Chapman and Martin(2007), in which we develop a stylized economicmodel that captures these features. In the model,the central bank has two instruments with whichto inject liquidity into a payments system: an in-strument whose use depends on prevailing mar-ket conditions (the market-sensitive instrument),and an instrument whose use does not dependon market conditions (the market-insensitiveinstrument). These two instruments have differenteffects on the behaviour of agents in the economy.
We find that when the central bank is modelledas having less-precise information than otheragents about what actions agents take to insurethemselves against credit risk, the optimal policyfor the central bank has the features noted above.
C
* Senior Economist at the Federal Reserve Bank ofNew York. The views expressed here are those ofthe author and do not necessarily reflect those ofthe Federal Reserve Bank of New York or the FederalReserve System.
The Model
The key features of the model borrow heavilyfrom the seminal work of Freeman (1996,1999). The model abstracts from many impor-tant features of real-world financial and pay-ments systems but contains the four criteria,stated by Zhou (2000), necessary to effectivelymodel a payments system: First, it captures theunderlying transactions that lead to a need forsome non-cash payments. Second, the debt in-struments used in trade for goods are differentfrom saving/investment debt. Third, there is apotential shortage of liquidity, for at least someagents, when payment debt is settled. Fourth,there exists credit risk that is generated endoge-nously by the choice of agents.
The model features two types of agents: debtorsand creditors, who interact with each other totrade money and short-term debt for goods andlater for money to settle the short-term debt.The debtors that trade for goods may default in-stead of settling their debt. To avoid this default,a creditor can pay the cost of monitoring thedebt and thus reduce the probability that thedebtor will default (credit risk).
The investment to reduce the probability of de-fault is observable only by other agents in theeconomy and is not observable by the centralbank. This assumption is consistent with tworeal-world characteristics: First, agents in the fi-nancial system can take actions to limit their ex-posure to credit risk. Second, since the centralbank is usually not an active participant in thefinancial system, its information about these ac-tions is less precise than that of other financialsystem agents. Thus, at the margin, participantsin the banking sector have better informationabout their counterparties than the central bank.
When these loans are settled, there is a coordi-nation problem in the timing of settlement.
83
Research Summaries
That is, there is a chance that the creditor who iswaiting for a debt to clear may have an unex-pected need for the funds before the debt is set-tled. When this happens, the creditor can borrowfunds (liquidity) from other creditors, usingtheir unsettled claims as collateral. From thepoint of view of the other creditors, the unset-tled debt may be unsettled either because of acoordination problem in the settlement of debtor because of default.
The market for liquidity
In the model, the interest rate at which creditorsare able to borrow is efficient in that it accuratelyreflects the credit risk inherent in the claims theyhold. The total supply of liquidity comes fromdebts that are already settled. But when the co-ordination problem is severe, the supply of fundsavailable is small relative to the demand forfunds, and there will be a liquidity shortage. Inthis case, the interest rate that equates the supplyof liquidity to the demand for liquidity will pri-marily reflect a liquidity premium and will notaccurately reflect credit risk. Previous work(Freeman 1996, 1999; Martin 2004) has shownthat such a liquidity shortage is suboptimal andrequires the central bank to intervene with atemporary injection of liquidity.
Central bank liquidity provision
If the model contained no credit risk, it wouldbe optimal for the central bank to intervene di-rectly to eliminate the liquidity shortage. Indeed,since the coordination problem in settling debtdoes not arise because of a choice made byagents, the central bank’s intervention wouldnot affect incentives to monitor. The problem isattributable to a missing market that would co-ordinate the settlement of funds at an exact timewithin a day. The central bank’s intervention, inthis case, can be viewed as an attempt to correctthe inefficiency arising from the missing market.
When there is credit risk and agents can take ontoo much of this risk, the optimal action for thecentral bank is not as straightforward. Agents inthe economy form rational expectations aboutthe effect that the central bank’s policy will have,and they will behave accordingly. If the centralbank’s policy on providing liquidity is too liberal,it will increase the credit risk in the financial sys-tem, since it will reduce the incentive for privateagents to monitor credit risk. This distortion of
84
incentives is caused by two factors: First, thecentral bank in the model will misprice liquiditybecause of its less-precise information. Second,if the central bank intervenes and provides li-quidity by extending uncollateralized loans, itwould distort the allocation of credit risk in thefinancial system by taking credit risk on its ownbooks at an incorrect price.
The market-sensitive instrument
If the central bank provides liquidity to all cred-itors in a way that is not conditional on anymarket variables, then creditors will have no in-centive to put any effort in avoiding credit risk,since the price that they are charged for liquidityfrom the central bank is not affected by theamount of monitoring they do. Since they gainno benefit from monitoring, agents will notmonitor their exposure to credit risk. And thecentral bank will again take on credit risk fromagents when it provides liquidity. It follows thatan optimal policy must be conditional on theunderlying market price for liquidity.
For liquidity provision to give the correct incen-tives to all creditors, liquidity must be providedto a subset of the creditors. This subset (i.e., cen-tral bank counterparties)1 has more-preciseinformation than the central bank about theamount of monitoring of credit risk. They usethis information when supplying liquidity tothe rest of the payments system, thereby chargingthe correct price. Agents in the economy whoare not central bank counterparties then knowthat the price they have to pay for liquidity willdepend on the amount of credit-risk monitoringthat they undertake. They will therefore choosethe amount of monitoring that equates the costof monitoring to the expected cost of obtainingliquidity.
The optimal policy should be set up to encouragecompetition between the central bank counter-parties. Without this competition, these coun-terparties would use their privileged position toearn economic rents. In addition, liquidity should
1. In the working paper, these are referred to as primarydealers. The term “central bank counterparty” is usedhere to avoid confusion, since the term “primarydealer” is used in Canada to denote distributors ofgovernment securities whose participation in primaryand secondary markets for Government of Canadabonds is above a certain threshold.
Financial System Review – December 2007
be provided to the central bank’s counterpartieson a collateralized basis, so that all credit riskresides with the agents in the economy and notwith the central bank.
A role for the market-insensitiveinstrument
To be effective, the optimal policy described bythe model (a market-sensitive policy) has tworequirements. First, it needs a well-functioningmarket for liquidity. Second, it requires that thecentral bank know exactly how much liquidityto supply to its counterparties. The lack of eitherof these requirements implies a role for a market-insensitive policy to supplement the market-sensitive policy.
In certain situations, however, the market for li-quidity may be disrupted. In these cases, thefirst requirement is missing. When this happens,liquidity must be provided using a market-insensitive policy, since the market among thecentral bank’s counterparties is not functioningproperly.
If the second requirement is missing, then thecentral bank in the model does not know theamount of liquidity demanded by the market,and it must forecast the amount of liquidity toinject. Large errors in the central bank’s forecastwill cause distortions in the pricing of creditrisk. A market-insensitive policy that is set sothat it is inactive in normal market conditionswill help limit such distortions; it will providean upper bound on the effect that errors in theforecast of liquidity can have.
Liquidity Provision by theBank of Canada
In general, the provision of liquidity by theBank of Canada to the financial system is cen-tred on its monetary policy framework.2
Liquidity provision by the Bank shares some ofthe key features implied by the model, althoughit is significantly more complex. First, in normalcircumstances, the model suggests that the
2. Details of the Bank of Canada’s framework for imple-menting monetary policy may be found in Bank ofCanada (2007). For a description of how the Bank ofCanada has recently used some of these facilities, seeBox 3 on p. 12.
central bank should use a market-sensitive policy,which is intended for a small subset of all marketparticipants. In the case of the Bank of Canada,open market buyback operations (special pur-chase and resale agreements and sale and repur-chase agreements) and the Large Value TransferSystem (LVTS) cash setting are essentially market-sensitive policies. The use of open market buy-back operations is based on market conditions(including importantly, observed rates in theovernight market); they are transacted with onlya subset of the market; and they are carried outin such a way that virtually no credit risk is as-sumed by the Bank of Canada. The Bank canadjust the targeted level of settlement balancesdepending on actual and expected conditions inthe overnight market (Arjani and McVanel 2006).Access to these settlement balances is restrictedto direct participants in the LVTS.
Second, when it is difficult to accurately forecastthe level of liquidity needed, the model suggeststhat the central bank should provide liquiditythrough a market-insensitive policy. This policyshould be designed in such a way that it encour-ages participants to transact with each other fortheir liquidity needs and use the market-insen-sitive instrument only for unexpected shortfalls.In the case of the Bank of Canada, the StandingLiquidity Facility (SLF) is available to LVTS di-rect participants experiencing temporary unex-pected shortfalls in their end-of-day settlementbalances. The rate paid on loans from the SLFencourages direct participants in the LVTS toseek liquidity from each other rather than fromthe SLF.3
Finally, the model suggests that in extraordinarycircumstances the central bank should provideliquidity to a larger set of participants through amarket-insensitive policy. In cases of extraordi-nary stress, the Bank provides Emergency LendingAssistance (ELA) to member institutions in theCanadian Payments Association, not only to thedirect participants in the LVTS, under the restric-tions set out in its policy.4
3. The rate paid to use the SLF is 25 basis points abovethe target overnight rate, while the rate that the Bankof Canada pays on balances left with it overnight is25 basis points less than the target overnight rate.
4. For a fuller description, see Daniel, Engert, andMaclean (2004–05).
85
Research Summaries
Conclusion
Our model suggests that central bank liquidityis best provided through a tiered structure: Thecentral bank provides liquidity to a subset of themarket that, in turn, provides liquidity to others.This is fundamentally because the provision ofliquidity by the central bank can distort the priceof credit risk in the market to which the liquidityis provided. The model implies that a centralbank that has relatively less information thanmarket participants should effectively delegatethe monitoring of credit risk to a subset of themarket.
The Bank of Canada’s policy for liquidity provi-sion shares many of the policy features that areoptimal in this model. In particular, it has theaspects of limited access and market sensitivityin normal circumstances and wider access inextraordinary circumstances.
References
Arjani, N. and D. McVanel. 2006. “A Primer onCanada’s Large Value Transfer System.”Available at <http://www.bankofcanada.ca/en/financial/lvts_neville.pdf>.
Bank of Canada. 2007. “A Primer on the Imple-mentation of Monetary Policy in the LVTSEnvironment.” Available at <http://www.bankofcanada.ca/en/lvts/lvts_primer_2007.pdf>.
Chapman, J. T. E. and A. Martin. 2007. “Redis-counting under Aggregate Risk with MoralHazard.” Bank of Canada Working PaperNo. 2007-51 and Federal Reserve Bank ofNew York Staff Report No. 296.
Daniel, F., W. Engert, and D. Maclean. 2004–05.“The Bank of Canada as Lender of LastResort.” Bank of Canada Review (Winter):3–16.
Freeman S. 1996. “The Payment System,Liquidity, and Rediscounting.” AmericanEconomic Review 86 (5): 1126–38.
———. 1999. “Rediscounting under AggregateRisk.” Journal of Monetary Economics 43(1): 197–216.
86
Martin, A. 2004. “Optimal Pricing of IntradayLiquidity.” Journal of Monetary Economics51 (2): 401–24.
Zhou, R. 2000. “Understanding Intraday Creditin Large-Value Payment Systems.” FederalReserve Bank of Chicago Economic Perspec-tives 24 (3): 29–44.
Financial System Review – December 2007
Collateral Portfolios and AdverseDependenceAlejandro García and Ramazan Gençay
s financial markets and their supportinginfrastructure continue to develop overtime, banking professionals and regu-lators are taking steps to make them
safer. Many of those steps involve the use of col-lateral to manage financial risks.1 But collateralitself may consist of risky assets whose value canchange over time. Consequently, a pledge ofcollateral must be large enough to adequatelycover all losses in the event of a counterparty de-fault. Thus, the initial value of collateral is dis-counted. In other words, the amount of collateralpledged must be larger than the amount owing.This discount, often referred to as the “haircut,”lowers the risk associated with a transaction.But because collateral is costly to pledge, theframework established for setting haircuts mustrecognize the inherent trade-off between costsand risks associated with collateral. This frame-work could also provide useful information todetermine the desirable allocation of the portfolioof collateral.
This article summarizes the second of two pa-pers that explore a framework for calculatinghaircuts for different assets. The first, Garcíaand Gençay (2006), proposed a framework forcomparing different methods for computinghaircuts for individual assets. Particular atten-tion was paid to selecting a method that wouldaccomplish two goals. First, it would providesufficient collateral in the case of low-probabilityevents (large unexpected declines in asset prices)that might affect the stability of the financialsystem. Second, it would take into account thecost of pledging collateral. The second paper,García and Gençay (2007), examines how hair-cuts should be calculated in situations where avariety of assets are pledged as collateral. Here,
1. According to Khan (2007), the use of collateralto mitigate counterparty credit risk has increasedsubstantially.
A
the focus is on the relationship among the pricesof the different assets pledged as collateral and,in particular, how this relationship can changewhen markets are under stress. We refer to thischange as a change in the dependence structure,2which is caused by an event that changes the re-lationship between the returns on the assets inthe pool of collateral. For example, during normalmarket conditions, a given pool of collateralmay exhibit diversification benefits. However,during extreme market conditions, few, if any,such benefits may be evident for the same pool.
Financial Risks duringExtreme Events
When collateral consists of a variety of assets,note should be taken of two effects generally as-sociated with extreme events. The first is associ-ated with the individual security, and the secondwith the portfolio as a whole. The former is re-ferred to as the individual effect, the latter as theportfolio effect. The individual effect occurs whenthere is a negative return on an asset pledged ascollateral, but the dependence structure of theportfolio does not change significantly. Theportfolio effect occurs when there is a change inthe relationship among the various assets pledgedas collateral; that is, the dependence structurebetween the assets changes and exhibits smallerdiversification benefits than observed histori-cally.3 To illustrate the portfolio effect, considertwo hypothetical securities, x and y, that arepledged as collateral; and two states of the world:
2. This is usually referred to as a change in correlation,but this is not always correct, since there can be achange in the dependence structure without a changein the correlation.
3. Chan et al. (2005) refer to this as a “phase-locking”effect. The authors offer an explanation for theseeffects from a financial-engineering perspective.
87
Research Summaries
Chart 1 Scatter Plot of Losses during Normal Periods
Per cent
Security y
-20 -10 0 10 20-20
-10
0
10
20
-20
-10
0
10
20
Security x
o
o
o
o oo
o
o
oo
oo
o
oo
o
o
o
o
o
o oo
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
oo o
o
oo
o
oo o o
oo
o
oo o o
o
o
oo o
oo
oo
o
o
o
oo
o
o
o
o
o
o
o
ooo o
o
o
oo
o oo
o
o
oo
o
oo
o
o
o
o
o
oo
o
o
o
o
oo
oo
ooo
o
o
o
o
o
o
o
oo
oo
o
o
o
ooo
oo
o
o
oo
o
o
o oo
oo
o
o oo
o
ooo
oo o
o
o
oo
o
o
o
o
oo oo
oo
o o
oo
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
oo
o
o
o
o
o
o
o o
oo
o
o
o o
o
oo
o
o
o
oo
oo
oo
o
oo
o
oo
o
o
o
o
o
oo o
oo
o
o
o
oo
o
o
o
o
o
o
o
o
o oo
o
o
o
o
ooo
o
o
oo
o
oo
o
o
oo
o
o
o
o
o
ooo o
o
o
oo
o
o
o
o
o
o
o
o
o
oo
oo
oo
o
o
o
o
o
o
o
o
o
o
o
o
ooo
o oo
o
o
o
o
oo
o
o
o
o
o
o
o oo
oo
oo
o
oo
o
o
oo o
o
o
oo
o
o
o
o
o
o
o
o
o
oo
o
o
oo
oo
oo
o
o
o
o
oo
o
o
oo
oo
o
oo
o
o
oo
o
o oo
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o o
oooo
o
oo
o
o
o
oo oo
oo
oo
o
o
o oo
o o
o
o
o
oo
o
oo
o
o
oo
o
oo o
o
oo
o
o
o
o oo
oo o
o
o
o
o
o
oo ooo
oo
o
o
ooo
o
o
oo
oo
o
o
o
o
o
o
o
o
o
o
o
o
oo
oo
o o
o
oo
o
o
o
o
o
o o
o
o
o
oo
oo
o
o
oo
o
o
o
ooo o
oo
oo
ooo
o
o
o
o
o
o
o
o
oo
o
o
o
o oo o
o
o
o
o
o oo
o
o
o
oo
o
oo
oo
oo
o
o
o
ooo
o
o
oo
o
o
o o
o
o
o
oo o oo
oo
o
o
o
o
ooo
ooo
o
o
o
o
oo
o
oo
oo
o
o o
o
o
o oo
o
o
o
o
o
oo
o
o
oo
o
o
o
oo
o o
o oo
oo o
o
o
o
o
o
o
o
o o o
o
o
oo o
o o
oo
o
o o
ooo
oo
oo
o
o
o
o
o
oo
o
o
o
o
o
o
o
oo
o
o
o
oo
oo
o
o o
o
oo
o oo
o
o o
o
o
o
oo
o
o
oo
oo
o
oo
o
oo o
o
o
o
o
oo
o
oo
oo
o
o
o
oo
oo o
o
o
o
oo
oo
o
o
o
o
oo
o
o
ooo o
o
o
o
ooo
o
o
o
o
o
o
o
oo
o
o
oo
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
oo
oo
o
o
o
o
o
o
o oo
o
o
o
o
o
ooo
o
o
o
o o
o
o
oo
o
oo
o
o
o
oo
o
o
o
oo
oo
o
o
o
oo
oo
o
o
o
o
oo
o
oo
o
o
o
o
o
oo o
o
o
o
o o
oo
oo
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o oo
oo
o
o
o
oo
o
ooo o
ooo
o
o
oo
oo
o
o
o
o
o o
oo
o o
o o
o
o
o
oo
o
o
o
o
o
o o
o
o
o
oo
o o
o
oo o
o
oo
o
o
oo
o
oo
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o oo
o
oo
oo
o
o
o
o
oooo
o
oo
o
o
oo
o
o o
o
o
o
oo
oo
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o oo
o
o
o
oo
o
o
o o
o
o
oo
oo
o
o
oo o o
oo
o
o
o
oo
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o o
oo
o
o
o
oo
o
o
ooo
o
o
oo
oo
oo
o
o
o
o
oo
o
o
o o
o
o
o
oo
o
o
o
o
o
o
o
o
oo
o
o
oo
o
o
o
o
o
oo
o
oo
oo o
oooo
o
o
o
o
o
o
o
oo
oo
o
o
o
o
o
o
o
o
oo
o
o o
o
oo
oo
o
o
o
o
o
oo
oo
o
o
oo
o
o
oo oo
o
o
o
o
o
o
o
o
oo
o
oo
o
o
oo
ooo
oo
o
oo
o
o
o
ooo
o
o
o
o
o
ooo
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
oo
o
o
o
o
o
o
o
oo
o
o
o
o
o
oo
o
oo
oo
oo
o
o
o
oo
o
oo
o
o
ooo o o
oo
o
oooo
oooo
o
o
o
o
o
oo
o
o
oo o
oo
o
o
o
o
oo
oo
oo
o o
oo
o
o
o
o
ooo
o
o o
o
o
o
o
o
o
oo
o
o ooo
oo
o
o
o
o o
oo
o
o
o
o
o
o
o
o
o
o
o oo o
ooo o
o
o
o
o
o
o
o
oo
o
oo
o
oo
o
oo
o
o
o
o
o
o o
o
o
oo
o
o
oo
o
oo
oo
o
o
o
oo
o
o
o
o
o
o
o
o
o
oo
o
oo
o
o
o
o
oo
o
o
o
o
o
o
o
o
oo
ooo
oo
o
oo
o
o oo
o
o
o
o
o
o
oo
o
oo
o
o
o
o
o
o
o
o o
o
o
o
o
oo
oo
o
o
oo
oo o
o
oo oo
o
oo
oo o
o
o
oo
o
ooo
o
o
o
o
o
o
oo
o oo
oo
oo
o
o
oo
o o
o
o o
o
oo
o
o
o
oo
o
ooo
oo o
o
o oo o
o
o
o
o
o
o
o
o
oo
oo
o
o
o
o
o
oo
oo
o
oo
o oo
o
o
o
ooo
o
o
o
o
oo
o
o
o
oo
o
oo
o
oo o
o
o
oo
o
o
o
oo
o
o
o
oo
oo
o
o
o
o
oo
o
ooo
o
o
o
o
o
o
o
o
o
oo
o oo
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
oo
o
o
o
o
o
o o
o
o
o
o oo
oo
o
o
o
o
oo
o
o
ooo
o
o
oo
o
o
o
o
o
o
oo
o
oo
oo
ooo
oo
o
o o
o
o
oo o
o o
o
o
o
o
oo
o
o
oo
o
o
o
oo
o
o
o
o
o
oo
ooo
oo
o oo oo
oo
o
o
o
o
o
oo
o
o
o oo
o
o
o
o
o
o
oo o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o oo
o
o
o
o
oo
o o
o
o
o
oo
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
oo
o
oo
o
oo o
o
o
o
o
oo
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o o
o
oo
oo
o
o
oo o
o
o
o
o
o
o
o
o
oo
o
o
o o
oo
ooo
o
o
o
oo o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o
oo
o
o
o
o
oo
o
oo
o
o
oo
o
o
ooo
o
o
o
oo
oo
oo
o
oo o
o
ooo
o
o
o
oo
o
ooo
o
Scatter plot of 2,000 points representing losses from two hypotheticalsecurities with normal marginals N(3,4) and the dependence structureof a Normal copula with correlation 0.1.
Chart 2 Scatter Plot of Losses during Extreme Events
Per cent
Security y
-20 -10 0 10 20-20
-10
0
10
20
-20
-10
0
10
20
Security x
o
o o
o
o
oo
o
oo
o
oo
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
ooo
o
oo
o
o
oo
o
o
o
o
oo o
o
o
o
o
oo
o
o
o o
oo
o
o
o
o
o
o
o
ooo
oo
o
o
o
o
o
ooo
o
o
o
o
o
o
o
o
oo
o
o
oo
o
o
ooo
o o
o
oo o
o
oo o
oo
o
o
o
o
oo
o
o
ooo
o
o
ooo
o
o
ooo o
o
o
o
o
o
o
o o
o o
oo
o
o
o
o
oo
o
oo
o
o
o
o
o
o
o
oo
o
o
o o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o oo
o o
oo
o
o
oo
o
o
o
oo
o
oo
oo
o
oo o
o
oo
oo
oo
ooo
o
o
oo
oo o
o
o
o
oo
o
o
oooo oo
o
o
ooo
oo
o
o
o
o
o
o
oo
o
o
o
oo
o
oo
oo
oo
o
oo
oo
o
oo
o
oo
o
o
o
o
o o
o
o o
o
o
o
o
o
o
o
o
o
o
ooo
o
o
oo
oo
oo
oo
oo
o
o
ooo
oo
oo
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
oo
ooo o
oo
o
o oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o o
o
o
ooo
o
o
ooo
o
o
ooo
oo
oo
oo
o
oo
oo
o
o
o
o
oo
o
o
o
o
o
oooo
o
o
o
oo
o
o
o oo
oo
o
o
o
o oo o
ooo
o
oo o
oo
ooo
o
o
o
o
o
o
ooo oo
o
o
oo
o
o
oo
o
oo
o
o o
o
o
o
o
o
o
oo
o
ooo
o
o
o
o o
oo
o
o
o
o
o
o
o
o o
o
o
o
o
o
o
o ooo
o
oo
o
o
oo
o
o
o
oooo
o
o
o
oo
o
o
o
o
o
ooo
oooo
o o o
o
o
o
o
oo
o
o
o
o
o
o
o
o
oo oo
o
o
o
o
oo o
o
o
o
o
o
o
o
oo
o
o
o
oo
oo
oo
oo
oo
oo
o
oo
o
oo
o
o
o
ooo
o
o
ooo
o
oo
o
o
o
o
ooo
oo
oo
o
o
oo
o
o
o
o
o
o
oo
oo
o
o
o
o
oooo
o
o
o
o
o
oo
o
o
o
o
oo
o
o
o
oo
oo
o
o
oo
o o
o
o
o
oo
oo
o
oo
oo
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
oo
o
o
o
o
oooo
o
o
o
oo
o
o o oo
o
o
o
o
o
ooo
o
o
o
oo
oo
o
o
o
o
oo
oo
o
o
oo
oo
o
oo
oo
o
o
o
o ooo
o oo
ooo
o
o
oo
o oo
o
o
o
o
o
ooo
o
oo
oo
oo
oo
o
o
o
o
oo
o
oo
o
o oo
o
o
ooo
oo
o
oo
o
oo
oo
oo
ooo
o
o
oo
oo oo
o
o
o
oo
o
o
o
o
o
o oo
o
o
o
o
o
o
o
o
o
oo
o
o
o
oo
o
oo
o
oo o
o
o
oo
oo
o
o
ooo
o
o
oo
o
oo
o
o
o
o
o
oo
oo
o
oo
o
o
oo
o oo
oooo
o
o
o
o
o
o
o
o
o
o
o
oo
o
oo
o
o
o
o
o oo
oo
o
o
o
o
o o
oo
o
ooo
o
oo
o
o
o
o
o
o
o
o
o o
oo o
o
o
o o
o
o
o
o
o
o
o
ooo
o o
o
o
o
o
oo
o
o
oo
o
o
o
o
o
oo
oo
oo
oo
o
ooo
o
o
o
o
o
o
o
oo o
o o
o
o
o
ooo
oo
o
o
oo
o
oooo
o
o
oo
o
o
oo
o o
oo
o
oo
o
o
o
o
o
o
oo
o
o
oo
o
o
o
o
o
o
o
o
o
o o
o o
o
o
oo
o
o
o
oo
o
o
oo
o
oo
o
o
o
oo
o
o
oo
oo
oo o
oo
o
o
oo
o
oo
o
o
o
oo
oo
oo
o
o
oo
o
o
o
o
o
o
o o
o
o
ooo
o
oo
o
o
o
o
o
o
ooo
o
o
o
o
o
o
o
o
oo
o
o
o
o
oo
ooo
o
oo
o
o
o o
oo
o
o
o
o
o
o
o
o
ooo
o
o
o
o
o
oo
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
oo
o
o
oo
o
o
oo
o
o o
o
o
o
o
oo
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
ooo
oo o
oo
o
oo
o
o
o
o
oooo
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
oo
o
o
oo
oo
o
o
oo
o
o
o
o
o
oo
ooo
o
o
o
o
o
o
o
o o
o
o oo
oo o
oo
o
o
oo
ooo
o
o
o
o
o
o
o
oo
ooo
o
o
ooo
oo
o
o
oo
o
o
oo
o
o
oo
o
o
o
oo
o
o
oo
oo
o
o
o
o
o
o
ooo
o
o
oo
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
oo
o
o
oo o
o
oo
o
oo
oo
o
o
o
oo
ooo
oo
o
o
o o
o
o o
oo
o
o
oo
o o
o
o
o
o
o
oo
oo
o
o
o
o
oo
o
o
o
oo
o
o
o
o
o
o
o
o
oo
oo
o
o
o
o
o
o
oo
o
oo o
o
oooo
oo
o
o oo
o
o
o
o
oo
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
o
o o
oo
oo
o
oo oo
o
o
oo
o o
o
oo
oo
o
oo o
ooo
oo
o
o
ooo
o oo
o
o
o
o
o
ooo
o
o
oo
o
ooo
oo
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
oo oo
oo
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
ooo
o
o o
o
oo
o
o
o
o
o
o
o
o
o
oo
oo
o o
o
o
oo
oo
o
oo
ooo oo
o
o
o
o
o o
o
o
o
o
o
o
o
o
o
o
o
oo
o o
o
o
o
o
o
ooo
o
o
oo
o
o
o
o
oooooo
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
ooo
o
o
oo
oo
o
oo
oo
o
o
o
o
oo
o
oo
o
o
o
o
o
o o
o
o
o
o
o
o
o
o
o
ooo
oo
o
oo
o
oo
oo o
o
o
oo
ooo
oo
o
o
o
o
o
o
o
o
o
ooo
o oo
o
o
o o
o
oo
o
o
o
o
oo
o
o
o
oo
ooo
o
o
o
o
o
o
oo
ooo
oo
o
o
o
o
oo
o
o
oo
o o
o
oo o
o
oo
o
o
o o
oo
o o
o
o
o
o
oo
o
oo
o
o
oo
oo
o
o
oo
o
o
o
o
o
o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
ooo o
o
o o
oo
o o
o
o
oo
ooo
o
o
oo
o
o
o
o
o
o
o
o
ooo
o
o
o
o
o o
o
o
o
o
oo
o
o
o
o
o
o
o
o
o
o
o
oo
o o
o
o
o
o
o
o
o
o
o o
o
Scatter plot of 2,000 points representing losses from two hypotheticalsecurities with normal marginals N(3,4) and the dependence structureof a Gumbel copula G(4).
a normal state and an extreme state. During thenormal state, the scatter plot of per cent lossesfor x and y may be represented by Chart 1;during the extreme state, it can be representedby Chart 2.4 In this example, we assume that thedistribution of returns for each asset was thesame in both states, but that the dependencestructure between the assets changes.
Chart 1 illustrates that, in a normal state, thereare many instances in which a large loss for oneasset does not coincide with a large loss for theother. In contrast, as Chart 2 shows, in an ex-treme event, the diversification benefits are sig-nificantly reduced compared with those observedin normal periods. Chart 2 shows a greater de-gree of positive dependence, that is, large lossesin one asset coincide with large losses for theother asset.5
Managing Portfolio Effects
To manage the financial risks associated withthe portfolio effect, the dependence among as-sets must be modelled in a way that reflects whatcould happen if there were an extreme event.We accomplish this using copulas—multivariatedistributions that are very useful in financial-engineering problems involving modelling twoor more random variables. Because copulas al-low the multivariate distribution of returns forthe portfolio to have a wide range of marginals(i.e., the distribution of returns for each asset)and dependence structures, they allow us to sep-arate the behaviour of the dependence structurefrom the behaviour of individual asset prices.This separation is not possible with traditionalrepresentations of multivariate distributionfunctions and may lead to a misspecification ofthe multivariate distribution. The use of copulasthus facilitates the aggregation of risk acrosssecurities that may have different return distri-butions.
We use the copula-based method to determinewhether a collateral pool contains assets that
4. Charts 1 and 2 represent losses as positive values andprofits as negative values. This is a standard conven-tion in statistics, since actuarial risk theory is a theoryof positive random variables.
5. Note that other outcomes are possible during extremeevents. For example, if the portfolio is composed of arisk-free asset and a risky asset, the result could be amore negative dependence.
88
Financial System Review – December 2007
have a low probability of joint losses.6 This isdone to assess whether, during extreme events,the returns of the assets in the portfolio are likelyto continue to be as diversified as they werehistorically.
Note that while a receiver of collateral wouldprefer to have a collateral portfolio with largediversification benefits, the collateral pledgorwould normally decide which assets will bepledged, subject to the rules of the collateral-pledging agreement. A copula-based methodol-ogy could be an input in determining maximumlimits for classes of assets that can be pledged ascollateral (i.e., sector limits), creating incentivesfor those pledging collateral to supply a diversi-fied pool. In the event of a counterparty default,having a diversified pool of collateral could re-duce the costs associated with selling (liquidating)the collateral portfolio, because it may be easierto find counterparties to take those assets thatstill exhibit diversification benefits. In contrast,a portfolio with lower diversification benefitsmay require a significant discount in order tosell the assets in time to cover the losses.
Stress Testing PortfolioDependence
García and Gençay (2007) also present a meth-odology for examining the performance of theportfolio in the face of an event that negativelyaffects the dependence structure. The collateralpool in question is subjected to stress tests inwhich the dependence is changed by (i) using acomprehensive set of copula families that repre-sent different dependence structures and (ii) in-creasing the degree of positive dependence foreach copula. When conducting stress tests, weassume that the characteristics of the individualassets in the portfolio do not change, only theirdependence on each other. We estimate the dis-tributions for each asset based on historical da-ta, and, for the dependence structure, we startwith a scenario based on historical observationsof the dependence. This approach provides arange of the possible adverse dependencies (andtheir associated losses) that may occur duringextreme events. For example, using various copula
6. Based on Carmona (2004) and Zivot and Wang(2006), our copula-based method uses a semi-parametric approach to model the marginals and acopula to model the dependence.
models to capture the dependence between theprice changes of two Canadian investment-gradeassets, we observe that the portfolio losses (neg-ative returns) can vary by as much as half of apercentage point. This result, coupled with thesubstantial size of collateral portfolios, maytranslate into a large discrepancy between thedifferent models in dollar terms.7
Conclusion
This work, together with García and Gençay(2006), proposes: (i) a framework for calculat-ing haircuts for individual assets, (ii) a methodfor monitoring changes in the dependence struc-ture of assets, and (iii) a method for stress testingand measuring the possible effects of adversedependence structures on portfolio value.
This research has two policy implications. First,there is a need for caution when considering theextent to which a haircut should be reduced totake account of diversification benefits in a col-lateral pool. While those benefits may be evi-dent in normal situations, they may declinesignificantly during extreme events. This couldlead to uncollateralized exposures, or even losses,if collateral has to be liquidated in a period whenmarkets are under stress. Second, when the num-ber of assets accepted as collateral increases, it isimportant to consider not only the individualcharacteristics of the asset in question, but alsoits effect on the overall dependence structure ofthe portfolio of collateral.
References
Carmona, R. 2004. Statistical Analysis of Finan-cial Data in S-PLUS. New York: Springer.
Chan, N. T., M. Getmansky, S. Haas, andA. W. Lo. 2005. “Systemic Risk and HedgeFunds.” MIT Sloan Research PaperNo. 4535-05. School of ManagementWorking Paper.
García, A. and R. Gençay. 2006. “Risk-CostFrontier and Collateral Valuation in Secu-rities Settlement Systems for Extreme Mar-ket Events.” Bank of Canada WorkingPaper No. 2006-17.
7. Note that this result is specific to the portfolioexamined.
89
Research Summaries
García, A. and R. Gençay. 2007. “ManagingAdverse Dependence for Portfolios of Col-lateral in Financial Infrastructures.” Bankof Canada Working Paper No. 2007-25.
Khan, N. 2007. “Infrastructure Developmentsfor Over-the-Counter Derivatives.” Bankof Canada Financial System Review (thisissue).
Zivot, E. and J. Wang. 2006. Modeling FinancialTime Series with S-PLUS, 2nd ed. NewYork: Springer.
90
Financial System Review – December 2007
Housing Market Cycles and DurationDependence in the United States andCanadaRose Cunningham and Ilan Kolet
ttention has recently been focused onthe deterioration in the housing andcredit markets in the United Statesand the effects that this may have on
financial institutions and, more broadly, on fi-nancial stability. After an extended upswing, therate of increase in U.S. house prices has slowedsharply, and prices have even declined in someareas. House price cycles play an important rolein the consumption and savings decisions ofconsumers. Developments in housing marketscan have a significant impact on performance inthe banking sector and, thus, on the financialsystem, since residential mortgages account fora large share of the loan portfolios of Canadianbanks. Therefore, it is important for policy-makersto have a good understanding of the cycles inhouse prices.
Canada and, until recently, the United Stateshave been experiencing the longest period ofrising house prices on record. A natural ques-tion is whether such a long expansion is moreor less likely to end than a shorter one; that is,whether cycles in the housing market exhibitduration dependence. If there is duration de-pendence in housing cycles, then turning pointscan, to some extent, be predicted by the lengthof the phase. Thus, duration could prove to be auseful indicator for policy-makers. This is par-ticularly interesting in the current context, giventhat the surge in house prices in the United Statesseems to be over, while Canadian house pricescontinue to increase.
The aim of this study is to examine house priceexpansions and contractions in the UnitedStates and Canada using a panel data sampleof 137 cities, spanning a period of at least20 years.1 The goal is to improve our
1. A panel data sample creates more variability by com-bining variation across micro units with variationover time. With this more informative data, moreefficient estimation is possible.
A
understanding of housing market cycles inNorth America. First, we compare housingmarket cycles in the United States and Canadawith respect to duration, size, and correlation.We then estimate a model2 to test for durationdependence during periods of expansion andcontraction in house prices.This study builds on the housing cycle literatureand, to the best of our knowledge, is the onlyone that tests specifically for duration depen-dence in housing market cycles. One reason forthe lack of such an approach may be the lack oflong-term time series for aggregate house prices,which makes time-series econometric estimatesunreliable. To address this problem, we use apanel data estimation technique.
Descriptive Analysisof House Price Cycles
Housing market cycles in Canada and the UnitedStates exhibit a number of differences. First,there have been fewer national housing cyclesin the United States than in Canada (two U.S.contractions since 1975 compared with four inCanada since 1980). Second, aggregate realhouse prices have been considerably less vola-tile in the United States than in Canada, with astandard deviation in growth of 3.5 per cent,compared with 6.5 per cent in Canada. Finally,
2. We construct indexes of U.S. real house prices usingthe city- and national-level nominal house priceindexes from the Office of Federal Housing Enter-prise Oversight deflated by the CPI data published bythe Bureau of Labor Statistics. Canadian house pricescan be measured using data on the average sellingprice compiled from the multiple listing service(MLS). The MLS series are then deflated using theCanadian national consumer price index from Statis-tics Canada. Both of these data sources have limita-tions that are described in detail in Cunningham andKolet (2007).
91
Research Summaries
housing market cycles in Canada have tended tobe shorter and sharper than those in the UnitedStates, particularly during periods of decliningprices. The longest nationwide contraction inCanada occurred in the early 1980s and lastedthree years, and the accompanying price de-clines were generally as large or larger than inthe United States.3
An analysis of aggregate housing market cyclesis not sufficient, however, because housing isinherently local. The analysis of city-level datashows that local housing cycles in the two coun-tries have been quite similar overall: the averageexpansion lasts 5.8 years, with an average in-crease in real prices of about 32 per cent in bothcountries, and during a typical contraction, realprices decline by about 10 per cent in both coun-tries. Contractions are shorter in Canadian cities,however, lasting an average of 3.5 years com-pared with 4.4 years in U.S. cities.
It is important to note that unconditional corre-lations between housing cycles in Canadaand the United States do not imply causalitybetween the two countries.
A Model of DurationDependence
To more formally examine the full course ofhousing cycles, we estimate a duration model.Also known as survival analysis, this technique,commonly used in microeconomics, has alsobeen used in several studies of economic cycles,most notably stock market and business cycles.It is particularly relevant for our work becausea natural question regarding house prices is,“Given the recent increase in home prices, whatis the probability of the expansion ending?”
We estimate separate discrete-time survivalmodels for expansions and contractions inhousing cycles:
The dependent variable is a binary variable, ,which represents the phase that city is in attime . In the model for expansions, , if
3. Commodity price shocks have had larger effects onthe Canadian economy than on the U.S. economyand may explain some of the differences in housingand business cycles in the two countries.
Pr yit 1=( ) Φ(DUR3...DUR10UP GINC GPOP DRM).,,,=
yiti
t yit 1=
92
city is in an expansion phase, and , ifit exits the expansion phase in period . Thisdependent-variable phase is estimated using astandard probit model in which the right-handside contains a variable that measures the dura-tion in the current phase ( ),along with other variables that control for fun-damental factors affecting the duration of hous-ing cycles (income, ; population, ;and mortgage rates, ).
A non-zero coefficient on the duration variableindicates duration dependence. More specifically,a statistically significant positive coefficient im-plies that the longer the current phase has lasted,the more likely it is to continue. Conversely, asignificant negative coefficient on the durationvariable implies that the longer the current phasehas gone on, the more likely it is to end. A sta-tistically insignificant coefficient means that thephase is duration independent.
Results and Implications
We find that the longer a housing expansionlasts, the more likely it is to move into a con-traction phase.4 In contrast, contractions seemto have no duration dependence, but the resultsare sensitive to the particular specification. Thecontrol variables (i.e., the fundamental factors—income, population, and interest rates) explainmost of the transition dynamics of contractions,but there is a role for duration to help us predictexpansions. The asymmetric nature of our find-ings on duration dependence may be due to thefact that duration acts as a proxy for other vari-ables that could explain the transition out ofhousing market expansions. One potential inter-pretation is that the duration dependence in ex-pansion cycles may be a proxy for speculativeactivity. Speculation may only appear in expan-sion phases because, unlike other asset markets,short selling of houses is not possible.
These results are interesting for policy-makersfor several reasons. First, the findings and esti-mation results suggest that fundamental factors,notably interest rates, have a significant impact
4. The duration dependence results for expansions areremarkably robust, but the contraction results aremore sensitive. Furthermore, we find the magnitudeof the duration dependence in expansions to beeconomically significant.
i yit 0=t
DUR3...DUR10UP
GINC GPOPDRM
Financial System Review – December 2007
on the transition out of both contractions andexpansions.5 Second, the fact that duration issignificant for expansion phases could prove tobe a useful indicator in predicting the length ofhousing market expansions. Since financial in-stitutions in Canada are exposed to the housingmarket through their residential mortgage loanportfolio, the ability to predict the length ofhousing market expansions could be useful inassessing the expected impact of housing mar-ket developments on these financial institutionsand on the financial system as a whole.
References
Cunningham, R. and I. Kolet. 2007. “HousingMarket Cycles and Duration Dependencein the United States and Canada.” Bank ofCanada Working Paper No. 2007-2.
5. Rising interest rates tend to decrease the survivalprobability of expansions. We find that the real pol-icy rate variable is robust to all our specificationchanges, and its effect on survival probabilities is sta-tistically and economically significant. In particular,for contractions, the change in the real policy rate hasa large effect of roughly the same magnitude asgrowth in income per capita.
93
