Chapter

CHAPTER 1 Resolving the Crisis Legacy and Meeting New Challenges to Financial Stability

Author(s):
International Monetary Fund. Monetary and Capital Markets Department
Published Date:
May 2010
Share
  • ShareShare
Show Summary Details

A. How Has Global Financial Stability Changed?

The health of the global financial system has improved since the October 2009 Global Financial Stability Report (GFSR), as illustrated in our global financial stability map (Figure 1.1).1 However, risks remain elevated due to the still-fragile nature of the recovery and the ongoing repair of balance sheets. Concerns about sovereign risks could also undermine stability gains and take the credit crisis into a new phase, as nations begin to reach the limits of public sector support for the financial system and the real economy.

Figure 1.1.Global Financial Stability Map

Note: Closer to center signifies less risk, tighter monetary and financial conditions, or reduced risk appetite.

Macroeconomic risks have eased as the economic recovery takes hold, aided by policy stimulus, the turn in the inventory cycle, and improvements in investor confidence. The baseline forecast in the World Economic Outlook (WEO) for global growth in 2010 has been raised significantly since October, following a sharp rebound in production, trade, and a range of leading indicators. The recovery is expected to be multi-speed and fragile, with many advanced economies that are coping with structural challenges recovering more slowly than emerging markets. The improving growth outlook has reduced dangers of deflation, while inflation expectations remain contained as output gaps remain large in many advanced economies. In contrast, the need to address the consequences of the credit bubble has led to sharply higher sovereign risks amid a worsened trajectory of debt burdens (Figure 1.2).

Figure 1.2.Macroeconomic Risks in the Global Financial Stability Map

(Changes in notches since October 2009 GFSR)

Note: The indicators included in our assessment of macroeconomic risks (see Annex 1.1) are the IMF’s WEO growth projections, G-3 confidence indices, OECD leading indicators, and implied global trade growth (economic activity); mature and emerging market country breakeven inflation rates (inflation/deflation); and advanced country general government deficits and sovereign credit default swap spreads (sovereign credit).

With markets less willing or able to support leverage—be it on bank or government balance sheets—sovereign credit risk premiums have more recently widened across mature economies with fiscal vulnerabilities. Longer-run solvency concerns have, in some cases, telescoped into short-term strains in funding markets that can be transmitted to banking systems and across borders. The management of sovereign credit and financing risks therefore carries important consequences for financial stability in the period ahead (see Section B).

Quantitative- and credit-easing policies, extraordinary liquidity measures, and government-guaranteed funding programs have helped improve the functioning of short-term money markets and allowed a tentative recovery in some securitization markets. As a result, monetary and financial conditions have eased further, as market-based indicators of financial conditions largely reversed the sharp tightening seen earlier in the crisis. This has been accompanied by a decline in market and liquidity risks as asset prices have continued to recover across a range of asset classes (Figure 1.3).

Figure 1.3.The Crisis Remains in Some Markets as Others Return to Stability

Source: IMF staff estimates.

Note: The heat map measures both the level and one-month volatility of the spreads, prices, and total returns of each asset class relative to the average during 2003–06 (i.e., wider spreads, lower prices and total returns, and higher volatility). The deviation is expressed in terms of standard deviations. Dark green signifies a standard deviation under 1, light green signifies 1 to 4 standard deviations, yellow signifies 4 to 9 standard deviations, and magenta signifies greater than 9 standard deviations.

MBS = mortgage-backed security; RMBS = residential mortgage-backed security.

Supported by these more benign financial conditions, private sector credit risks have improved. Our estimates of global bank writedowns have declined to $2.3 trillion from $2.8 trillion in the October 2009 GFSR, reducing aggregate banking system capital needs. However, pockets of capital deficiency remain in segments of some countries’ banking systems, especially where exposures to commercial real estate are high. Banks face new challenges due to the slow progress in stabilizing their funding and the likelihood of more stringent future regulation, leading them to reassess business models as well as raise further capital and make their balance sheets less risky. Distress may resurface in banks that have remained dependent on central bank funding and government guarantees (see Section C).

The overall credit recovery will likely be slow, shallow, and uneven. The pace of tightening in bank lending standards has slowed, but credit supply is likely to remain constrained as banks continue to delever. Private credit demand is likely to rebound only weakly as households restore their balance sheets. Ballooning sovereign financing needs may bump up against limited lending capacity, potentially helping to push up interest rates (see Section D) and increasing funding pressures on banks. Policy measures to address supply constraints may therefore still be needed in some economies.

Emerging market risks have continued to ease. Capital is flowing to Asia (excluding Japan) and Latin America, attracted by strong growth prospects, appreciating currencies, and rising asset prices, and pushed by low interest rates in major advanced economies, as risk appetite continues to recover. Rapid improvements in emerging market assets have started to give rise to concerns that capital inflows could lead to inflationary pressure or asset price bubbles. So far there is only limited evidence of stretched valuations—with the exception of some local property markets. However, if current conditions of high external and domestic liquidity and rising credit growth persist, they are conducive to over-stretched valuations arising in the medium term (see Section E).

B. Could Sovereign Risks Extend the Global Credit Crisis?

The crisis has led to a deteriorating trajectory for debt burdens and sharply higher sovereign risks. With markets less willing to support leverage—be it on bank or sovereign balance sheets—and with liquidity being withdrawn as part of policy exits, new financial stability risks have surfaced. Initially, sovereign credit risk premiums increased substantially in the major economies most hit by the crisis. More recently, spreads have widened in some highly indebted economies with underlying vulnerabilities, as longer-run public solvency concerns have telescoped into strains in sovereign funding markets that could have cross-border spillovers. The subsequent transmission of sovereign risks to local banking systems and feedback through the real economy threatens to undermine global financial stability.

The crisis has increased sovereign risks and exposed underlying vulnerabilities. The higher budget deficits resulting from the crisis have pushed up sovereign indebtedness, while lower potential growth has worsened debt dynamics. For example, G-7 sovereign debt levels as a proportion of GDP are nearing 60-year highs (Figure 1.4). Higher debt levels have the potential for spillovers across financial systems, and to impact on financial stability. Some sovereigns have also been vulnerable to refinancing pressures that could telescope medium-term solvency concerns into short-term funding challenges (Figure 1.5).

Figure 1.4.Sovereign Debt to GDP in the G-7

(In percent)

Source: IMF, Fiscal Affairs Department database.

Note: Average using purchasing power parity GDP weights.

Figure 1.5.Sovereign Risks and Spillover Channels

Table 1.1 shows a range of vulnerability indicators for advanced economies that captures their current fiscal position, reliance on external funding, and banking system linkages to the government sector.2 It features not only economies that had credit booms and subsequent busts, but also those whose underlying vulnerabilities have come into greater focus, and which are perceived as having less flexibility—economically or politically—to address mounting debt burdens.3,4

Table 1.1.Sovereign Market and Vulnerability Indicators(Percent of GDP, unless otherwise indicated)
Sovereign CDS Spreads (bps)1,210-year Swap Spreads (bps)1,3Sovereign Credit Rating/Outlook1Fiscal and Debt FundamentalsExternal FundingBanking System Linkages
5- yearCDS curve slope(5-year minus 1-year)Change since 9/30/2009(Notches above speculative grade/outlook)4Rating actions(since 6/30/07)5General government structural deficit 6,7 FY2010 (p)Gross gen. govt. debt 6,8,9 FY2010 (p)Netgen. govt. debt 6,8,10 FY2010 (p)Gen. govt. securities < 1 year remaining maturity 11Gen. govt. debt held abroad12Current account balance6,13 2010 (p)Depository institutions’claims on gen. govt.14BIS reporting banks’consolidated claims on public sector15
(percent of 2009 GDP)(percent of depository institutions’ consolidated assets)
Australia3814−39239/Stablenone4.919.85.43.94.3−3.52.31.22.7
Austria582818310/Stablenone4.370.760.56.158.51.815.14.013.2
Belgium58332459/Stablenone4.3100.191.122.665.0−0.521.36.219.0
Canadan.a.n.a.−24-1410/Stablenone3.082.331.814.114.1−2.618.68.94.6
Czech Republic693463-585/Stable2 up/0 down3.737.6n.a.5.19.6−1.714.312.45.9
Denmark3422−16310/Stablenone1.751.23.14.417.93.18.21.76.2
Finland25198310/Stablenone1.949.9n.a.12.035.92.04.72.09.6
France502413810/Stablenone4.684.274.517.248.7−1.918.54.68.0
Germany3316−17610/Stablenone3.876.768.615.840.35.520.66.711.8
Greece427−2233812823/Neg0 up/6 down8.9124.1104.315.999.0−9.717.58.532.3
Iceland412−134n.a.n.a.0/Neg0 up/11 down4.8119.977.2n.a.n.a.5.4n.a.n.a.18.1
Ireland1552611908/Neg0 up/5 down7.978.847.83.347.20.45.80.69.0
Israel11260−505/Stable3 up/0 down-0.177.572.8n.a.14.53.94.77.11.1
Italy1252066137/Stablenone3.5118.6116.024.556.4−2.829.411.920.0
Japan6654−688/Negnone7.5227.3121.748.713.72.869.321.81.9
Korea823243−335/Stable1 up/0 down−1.433.3n.a.3.23.01.66.84.24.0
Netherlands34226310/Stablenone5.264.246.016.246.25.010.82.88.9
New Zealand46143429/Negnone2.031.33.44.912.9−4.65.62.85.9
Norway1913−68−2510/Stablenone7.353.6−153.612.127.516.8n.a.n.a.11.9
Portugal16032102657/Neg0 up/2 down7.185.981.613.060.2−9.010.23.223.0
Slovak Republic6041−67346/Stable2 up/0 down4.737.3n.a.3.512.6−1.819.321.75.9
Slovenia5337−65−318/Stablenone4.435.2n.a.n.a.19.6−1.511.07.36.2
Spain1303855239/Neg0 up/1 down7.366.957.512.426.9−5.320.66.37.2
Sweden3523−12710/Stablenone0.843.1−16.24.219.35.44.21.46.2
Switzerland4522−46210/Stablenone0.339.839.24.63.89.5n.a.n.a.5.0
United Kingdom7740174310/Negnone7.678.271.66.617.9−1.75.11.13.6
United States421621710/Stablenone9.292.666.217.924.7−3.38.25.52.7
Sources: Bank for International Settlements (BIS); Bloomberg, L.P.; IMF, International Financial Statistics, Monetary and Financial Statistics, and World Economic Outlook (WEO) databases; BIS-IMF-OECD-World Bank Joint External Debt Hub; and IMF staff estimates.Note: (p) = projected. CDS = credit default swap; bps = basis points.

The crisis has driven up market prices of sovereign risk.

The vulnerabilities outlined in Table 1.1 are being priced in to market assessments of sovereign risk. A cross-sectional regression over 24 countries indicates that higher current account deficits and greater required fiscal adjustment are correlated with higher sovereign credit default swap (CDS) spreads (Figure 1.6).5 In addition, BIS reporting banks’ consolidated cross-border claims on each country’s public sector as a proportion of GDP help to explain spreads, especially for those countries with wider spreads.6

Figure 1.6.Contributions to Five-Year Sovereign Credit Default Swap Spreads

(In basis points)

Sources: Bank for International Settlements (BIS); and IMF staff estimates. Note: Credit default swap spread (t-stats) =–2.35 (–1.89) current account balance + 4.45 (3.08) required fiscal adjustment + 4.14 (4.93) BIS bank claims. Adjusted R = 0.81, n = 24.

Sovereign risks have come to the fore in the euro zone.

The global financial crisis triggered several phases of unprecedented volatility in European government bond and swap markets (Figure 1.7).7 To chart the evolving nature of risk transmission among euro zone sovereigns, a model of swap spreads was estimated that takes account of joint probabilities of default, global risk aversion, and fiscal fundamentals (Box 1.1).

Figure 1.7.The Four Stages of the Crisis

(Ten-year sovereign swap spreads, in percent)

Source: Bloomberg L.P.

In the early stages of the crisis, the increase in global risk aversion benefited core sovereigns such as France and Germany, while spreads widened for sovereigns (Figure 1.7) perceived to be more risky. After Lehman’s collapse, the countries that weighed adversely on other sovereigns were those that had financial systems that were hit hard by the financial crisis (Austria, Ireland, and the Netherlands). As sovereigns stepped in with public balance sheets to support banks, there was a general narrowing of swap spreads as fears of systemic crisis subsided and global risk aversion fell. However, more recently, the source of spillovers has shifted to economies with weaker fiscal outlooks and financial strains, with these tensions most evident in Greece.

The recent turmoil in the euro zone also demonstrated how weak fiscal fundamentals coupled with underlying vulnerabilities can manifest themselves as short-term financing strains.

In the presence of outsized deficits and an unsustainable debt trajectory, heavy reliance on external demand for government obligations and large concentrated debt rollover requirements can shorten the timeline for addressing solvency challenges. Unlike local demand sources, nonresident buyers are naturally more attuned to sovereign risk and inclined to step back from further purchases in times of market stress. A debt profile with concentrated maturities also introduces “trigger dates” around which policymakers must navigate. These hurdles can constrain policy options and increase the likelihood of standoffs developing between the government and investors demanding higher risk premiums. Ultimately, an unresolved solvency crisis amid high near-term refinancing needs and political uncertainty could limit access to public debt capital markets.

Box 1.1.Explaining Swap Spreads and Measuring Risk Transmission among Euro Zone Sovereigns

What factors most affected swap spreads during the four phases of the crisis (see diagram) and how did sovereign risk transmission evolve during these phases? A model of swap spreads based on measures of sovereign risk, global risk aversion, and country-specific fiscal fundamentals was estimated to shed light on this question (see Annex 1.10 on the IMF GFSR website). The first figure summarizes the results of the model. It shows that during the initial phase of the crisis, the increase in global risk aversion helped lower swap spreads in core sovereigns as investors sought the relative safety of these bonds. However, as the crisis progressed, spreads widened in other sovereigns, driven by worsening fundamentals and spillovers. In recent months, spreads have continued to widen in those countries with the greatest fiscal pressures.

Sovereign risk transmission between two countries was derived from sovereign CDS spreads using the methodology developed by Segoviano (2006). Essentially, this measure represents the probability of distress in one sovereign given the distress in another. In order to determine whether the nature of risk transfer had changed, these joint probabilities of distress were averaged over each of the four phases of the crisis that are defined in the diagram.

During the systemic outbreak phase of the crisis (see first table), the main sources of risk transfer—shown by the sum of the percentage contributions in the last row—were Austria, Ireland, Italy, and the Netherlands. In other words, the euro zone members that faced the greatest concerns regarding their exposures to eastern Europe, domestic financial systems (e.g., Ireland), or general fiscal conditions (in the case of Italy) transmitted the most sovereign risk to other countries.

Contributions to Swap Spreads by Crisis Phase

(Average of changes in swap spreads in basis points)

Source: IMF staff estimates.

In contrast, during the latest sovereign risk phase (see second table), Greece, Portugal, and, to a lesser extent, Spain and Italy became the main contributors to inter-sovereign risk transfer, reflecting the shift in market concerns from financial sector vulnerabilities to fiscal vulnerabilities.

Contributions to Euro Area Distress Dependence, October 2008–March 2009(Percentage point contribution to total distress probability)
Contribution from:
GermanyFranceItalySpainNetherlandsBelgiumAustriaGreeceIrelandPortugalTotal
Contribution to:
Germany9.912.011.113.79.415.88.411.18.7100
France7.711.89.717.48.918.07.811.47.3100
Italy6.38.610.814.78.919.29.913.97.8100
Spain6.58.613.314.38.518.69.014.17.1100
Netherlands6.910.113.311.510.617.38.912.39.0100
Belgium6.18.111.39.214.819.09.414.57.5100
Austria5.77.914.112.611.410.611.814.411.5100
Greece5.37.012.810.511.09.518.416.19.3100
Ireland5.47.213.311.611.710.518.212.59.6100
Portugal5.87.611.69.012.88.421.09.813.8100
Total15.67.411.49.612.28.516.78.812.37.7100
Source: IMF staff estimates.
Contributions to Euro Area Distress Dependence, October 2009–February 2010(Percentage point contribution to total distress probability)
Contribution from:
GermanyFranceItalySpainNetherlandsBelgiumAustriaGreeceIrelandPortugalTotal
Contribution to:
Germany12.011.113.44.87.46.919.86.218.3100
France5.613.414.86.08.17.718.28.018.3100
Italy4.010.416.43.36.87.224.27.220.5100
Spain4.310.214.43.37.07.423.98.421.1100
Netherlands4.513.210.212.28.05.322.13.321.2100
Belgium4.310.310.912.94.67.622.68.118.8100
Austria3.78.710.812.53.07.026.56.021.8100
Greece4.17.514.215.74.27.810.515.720.3100
Ireland3.17.79.912.82.06.85.931.320.6100
Portugal4.28.513.715.74.67.410.023.612.3100
Total13.78.311.012.73.46.57.021.48.118.0
Source: IMF staff estimates.
Note: This box was prepared by Carlos Caceres, Vincenzo Guzzo, and Miguel Segoviano.

Financial channels can amplify sovereign risks.

Insufficient collateral requirements for sovereign counterparties in the over-the-counter (OTC) swap market can transmit emerging concerns about the credit risk of a sovereign to its counterparties. In contrast to most corporate clients, dealer banks often do not require highly rated sovereign entities to post collateral on swap arrangements.8 Dealers may attempt to create synthetic hedges for this counterparty risk by selling assets that are highly correlated with the sovereign’s credit profile, sometimes using short CDS (so-called “jump-to-default” hedging).

This hedging activity from uncollateralized swap agreements can put heavy pressure on the sovereign CDS market as well as other asset classes. For instance, heavy demand for jump-to-default hedges can quickly push up the price of short-dated CDS protection. With bond dealers also trying to offset some of the sovereign risk in their government bond inventory, many European sovereign CDS curves departed from their normal upward sloping configuration to significant flattening or outright inversion (Figure 1.8). Greece’s sovereign CDS curve inverted in mid-January as the funding crisis accelerated and jump-to-default hedging demand increased; Portugal’s CDS curve inverted two weeks later. These pressures can easily spill over into the domestic bond market and push yields higher.

Figure 1.8.Sovereign Credit Default Swap Curve Slopes

(Five-year credit minus one-year default swap spread, in basis points)

Source: Bloomberg L.P.

Yet sovereign CDS markets are still sufficiently shallow, especially in one-year tenors, that a large gross notional swap exposure may prompt a dealer to look to other, more liquid asset classes for a potential hedge for its exposure to sovereigns.9 Proxies such as corporate credit, equities, or even currencies are commonly used, putting pressure on other asset classes. If swap arrangements with sovereigns were adequately collateralized, there would be no need for such defensive hedges and there would be less potential for volatility to spread from swaps to other markets.10 However, steps to reduce transmission channels should avoid interfering with efficient market functioning and good risk management practices. Thus, recent proposals to ban “naked” CDS exposures could be counter-productive, as this presupposes that regulators can arrive at a working definition of legitimate and illegitimate uses of these products (see Section F) (Annex 1.2).

Sovereign crises can widen and cross borders as they spread to the banking system.

Due to the close linkages between the public sector and domestic banks, deteriorating sovereign credit risk can quickly spill over to the financial sector (Figure 1.9). On the asset side, an abrupt drop in sovereign debt prices generates losses for banks holding large portfolios of government bonds. On the liability side, bank wholesale funding costs generally rise in concert with sovereign spreads, reflecting the longstanding belief that domestic institutions cannot be less risky than the sovereign. In addition, the perceived value of government guarantees to the banking system will erode when the sovereign comes under stress, thus raising funding costs still higher. Multiple sovereign downgrades could precipitate increased haircuts on government securities or introduce collateral eligibility concerns for central bank or commercial repos.11

Figure 1.9.Sovereign Risk Spilling over to Local Financial Credit Default Swaps (CDS), October 2009 to February 2010

Sources: Bloomberg L.P.; and IMF staff estimates.

Financial sector linkages can transmit one country’s sovereign credit concerns to other economies. As higher domestic government borrowing in a country crowds out private lending, multinational banks may withdraw from cross-border banking activities. Likewise, other economies that are heavily reliant on international debt borrowing or on banks from countries under significant sovereign stress could be viewed as susceptible to financial sector instability. Figure 1.10 illustrates these linkages by showing how some countries in eastern Europe have proven more sensitive to changes in Western European sovereign credit risk.

Figure 1.10.Regional Spillovers from Western Europe to Emerging Market Sovereign Credit Default Swaps

Sources: Deutsche Bank; and IMF staff estimates.

Note: Sensitivities of sovereign credit default swaps (CDS) captured by regression betas estimated from daily spread changes between October 2009 and February 2010 in joint regression, using the Traxx Main Index and a reweighted SovX-Western Europe index that matches geographic profile of Traxx Main.

Thus, the skillful management of sovereign risks is essential for maintaining financial stability and preventing an unnecessary extension of the crisis.

C. The Banking System: Legacy Problems and New Challenges

The global banking system is coping with the legacy of the crisis and with the prospect of further challenges from the deleveraging process. Improving economic and financial market conditions have reduced expected writedowns and bank capital positions have improved substantially. But some segments of country banking systems remain poorly capitalized and face significant downside risks. Slow progress on stabilizing funding and addressing weak banks could complicate policy exits from extraordinary support measures, and the tail of weak institutions in some countries risks having “zombie banks” that will act as a dead weight on growth. Banks must reassess business models, raise further capital, shrink assets, and make their balance sheets less risky. Policymakers will need to ensure that this next stage of the deleveraging process unfolds smoothly and leads to a safe, competitive, and vital financial system.

Since the October 2009 GFSR, total estimated bank writedowns and loan provisions between 2007 and 2010 have fallen from $2.8 trillion to $2.3 trillion. Of this amount, around two-thirds ($1.5 trillion) had been realized by the end of 2009 (Table 1.2 and Figure 1.11). As explained in that previous GFSR, these estimates are subject to considerable uncertainty and considerable range of error.12 The sources of this uncertainty include the data limitations, measurement errors from consolidation, cross-country variations, changes in accounting standards, and uncertainty associated with our assumptions about exogenous variables. Differences between writedowns projected and realized reflect a number of factors, including the future path of delinquencies, differences in accounting conventions and reporting lags across regions, and the pace of loss recognition. In the current environment of near-zero interest rates, banks also face strong incentives to extend maturities and prevent delinquent loans from being reported as nonperforming.13

Table 1.2.Estimates of Global Bank Writedowns by Domicile, 2007–10
Estimated HoldingsEstimated Writedowns October 2009 GFSREstimated Writedowns April 2010 GFSRImplied Cumulative Loss Rate October 2009 GFSRImplied Cumulative Loss Rate April 2010 GFSRShare of Total Writedowns April 2010 GFSR
(billions of U.S. dollars)(billions of U.S. dollars)(billions of U.S. dollars)(percent)(percent)(percent)
U.S. Banks
Loans
Residential mortgage2,9812302047.76.823.0
Consumer1,11519518017.516.220.4
Commercial mortgage1,114100879.07.89.8
Corporate1,10472656.65.97.4
Foreign11,74557533.33.05.9
Total for loans8,0596545888.17.366.5
Securities
Residential mortgage1,49518916612.711.118.8
Consumer142000.00.00.0
Commercial mortgage196634832.024.55.4
Corporate1,11548174.31.51.9
Governments580000.00.00.0
Foreign197571667.36.77.4
Total for securities4,5023712968.26.633.5
Total for loans and securities12,5611,0258858.27.0100.0
U.K. Banks
Loans
Residential mortgage1,63647272.91.65.9
Consumer423666415.715.114.0
Commercial mortgage344394111.212.19.1
Corporate1,82883634.53.413.8
Foreign12,51426120310.48.144.6
Total for loans6,7444973987.45.987.5
Securities
Residential mortgage225271112.05.02.5
Consumer58427.42.80.4
Commercial mortgage5112823.515.01.7
Corporate2582579.52.71.5
Governments360000.00.00.0
Foreign167239295.84.46.4
Total for securities1,625107576.63.512.5
Total for loans and securities8,3696044557.25.4100.0
Euro Area Banks
Loans
Residential mortgage4,53047441.01.06.6
Consumer67527254.03.83.8
Commercial mortgage1,27240373.12.95.6
Corporate5,01885791.71.611.9
Foreign14,5002822566.35.738.4
Total for loans15,9944804423.02.866.4
Securities
Residential mortgage96613010413.510.815.7
Consumer271581.92.81.1
Commercial mortgage264624023.515.06.0
Corporate1,3162201.70.00.0
Governments2,146000.00.00.0
Foreign11,943113725.83.710.8
Total for securities6,9073332244.83.233.6
Total for loans and securities22,9018146653.62.9100.0
Other Mature Europe Banks2
Total for loans3,2411651345.14.186.0
Total for securities72936224.93.014.0
Total for loans and securities3,9702011565.13.9100.0
Asian Banks3
Total for loans6,15097841.61.473.5
Total for securities1,72869304.01.826.5
Total for loans and securities7,8791661152.11.5100.0
Total for all bank loans40,1891,8931,6474.74.172.4
Total for all bank securities15,4919166295.94.127.6
Total for loans and securities55,6802,8092,2765.04.1100.0
Sources: Bank for International Settlements (BIS); Bank of Japan; European Securitzation Forum; Keefe, Bruyette & Woods; U.K. Financial Services Authority; U.S. Federal Reserve; and IMF staff estimates. Note: Domicile of a bank refers to its reporting country on a consolidated basis, which includes branches and subsidiaries outside the reporting country. Bank holdings are as of the October 2009 GFSR. Mark-to-market declines in securities pricing are as of January 2010.

Figure 1.11.Realized and Expected Writedowns or Loss Provisions for Banks by Region

(In billions of U.S. dollars unless indicated)

Source: IMF staff estimates.

1 Includes Denmark, Iceland, Norway, Sweden, and Switzerland.

2 Includes Australia, Hong Kong SAR, Japan, New Zealand, and Singapore.

Expected writedowns from loans have declined with the improved economic outlook, but further deterioration lies ahead.

For U.S. banks, estimated loan writedowns and provisions for 2007–10 were revised down by $66 billion to $588 billion after growth turned positive and house prices stabilized in the second half of 2009 (Table 1.2). Nevertheless, serious mortgage delinquencies and foreclosures continue to rise, as unemployment persists at a high level and almost one-quarter of mortgage borrowers have negative housing equity. Loan charge-off rates are expected to peak between 2009 and 2011 depending on the asset class (Figure 1.12).

Figure 1.12.U.S. Bank Loan Charge-Off Rates

(In percent of total loans)

Sources: Federal Reserve; and IMF staff estimates.

For euro area banks, improvements in GDP growth and unemployment forecasts have brought down estimated total loan writedowns and provisions by $38 billion to $442 billion since the October 2009 GFSR. Total loan loss provisions are now expected to have peaked at 1 percent in 2009 and decline to 0.7 percent this year. Corporates in the euro area proved more resilient than expected as they adjusted their capital expansion/working capital requirements, and reduced labor costs through the use of flexible working arrangements. Larger corporates also issued record amounts of debt in capital markets.

For U.K. banks, estimated loan loss provisions have been revised down by $99 billion to $398 billion, reflecting improvements in expected losses on residential mortgages. The projected mortgage loss provision rate for the first half of 2009 (1.9 percent) is significantly below that projected in the October 2009 GFSR (2.7 percent). However, commercial real estate has deteriorated more rapidly than anticipated with peak-to-trough price declines of more than 40 percent now expected, notwithstanding some signs of a recent uptick in prices in some segments.14

Financial healing and market normalization have led to a substantial improvement in securities prices, further pushing down overall writedown estimates.

Estimated global securities writedowns in banks have dropped by $287 billion to $629 billion as a result of improvements in market pricing of liquidity and risk premia across the range of corporate, consumer, and real estate securities held by banks (Figure 1.13). The largest reduction in writedowns is in corporate securities, while improvements in real-estate-related securities were more uneven. For example, in the United States, prices of (private label) residential mortgage-backed securities (RMBS) remain under pressure. In Europe, top-rated U.K. RMBS prices recovered strongly in the latter half of 2009, but Spanish RMBS markets reflect the weak housing market.

Figure 1.13.Global Securities Prices

(Rebased, 2007:Q3 = 100)

Sources: Barclays Capital; European Securitization Forum; Markit; and IMF staff estimates.

Note: ABS = asset-backed security; CMBS = commercial mortgage-backed security.

In aggregate, bank capital positions have improved substantially …

Capital ratios of aggregate banking systems have improved substantially since the October 2009 GFSR (Table 1.3). Banks have continued to raise private capital, and in some cases a pick-up in earnings in 2009 has helped to bolster capital. Projected write-downs are mostly covered by earnings for the aggregate banking system.

Table 1.3.Aggregate Bank Writedowns and Capital(In billions of U.S. dollars, unless otherwise shown)
United States (ex-GSEs)Euro AreaUnited KingdomOther Mature Europe1
Total reported writedowns (to end-2009: Q4)268041535582
Total capital raised (to end-2009: Q4)32925622255
Tier 1/RWA capital ratios(at end-2009), in percent11.3 (+1.5)9.1 (+1.1)11.5 (+2.3)8.5 (+0.3)
Source: IMF staff estimates.Note: Capital-raising includes government injections net of repayments. Capital ratios reflect those repayments. Figures in parentheses reflect percentage point changes since end-2008. All figures are under local accounting conventions and regulatory regimes, making direct comparisons between countries/regions impossible. GSE = government-sponsored enterprise. Tier 1 = Tier 1 capital; RWA = risk-weighted assets.

…but some segments of country banking systems remain poorly capitalized and face significant downside risks.

The aggregate picture masks considerable differentiation within segments of banking systems, and there are still pockets where capital is strained; where risks of further asset deterioration are high; and/or which suffer from chronically weak profitability.

In the United States, real estate exposures still represent a significant downside risk. The regional banks with heavy exposure to real estate need to raise capital (Table 1.4).15 Some 12 institutions have commercial real estate (CRE) exposure in excess of four times tangible common equity.16 In addition, the mortgage government-sponsored enterprises (GSEs) already received $128 billion of capital from the Treasury as of end-2009 and analysts’ estimates of total capital likely to be needed stretch up to $300 billion, highlighting that in the United States a substantial proportion of mortgage credit risk and capital shortfall has been transferred to the government by placing the GSEs under conservatorship.17

Table 1.4.United States: Bank Writedowns and Capital(In billions of U.S. dollars, unless otherwise shown)
Four Largest Banks (by assets)Investment/ Processing BanksRegional BanksOther Banks1
Tier1/RWA at end-2009 (in percent)10.614.911.510.3
Expected writedowns (Q1:2010–Q4:2011)22814716
Gross drain on capital2 (Q1:2010–Q4:2011)50626
Tier 1 capital at end-2009514143120353
Source: IMF staff estimates.Note: RWA = risk-weighted assets.

Further pressure on real estate markets may lie ahead. The “shadow housing inventory” continues to rise as lenders retain ownership of foreclosed property and forbear on seriously delinquent borrowers (as shown by the rising gap between 90-day+ delinquencies and foreclosure starts in Figure 1.14). The ending of foreclosure moratoria, house purchase tax incentives, and the Federal Reserve’s agency MBS purchases could trigger another drop in housing prices.18 In addition, a mortgage principal modification program (or the passage of so-called “cramdown” legislation) would precipitate significant additional losses on both first- and second-lien loans, prompting further RMBS downgrades.19

Figure 1.14.U.S. Mortgage Market

(In percent of total mortgage loans, seasonally adjusted)

Source: Mortgage Bankers Association.

Concerns in real estate lending also present a challenge in some euro area economies. In Spain, the most vulnerable loans are to property developers, as nonperforming loans and repossessions of troubled real assets have increased sharply over the last two years. Problem assets comprised of nonperforming loans and repossessions are projected to rise further, although reserves and earnings provide substantial cushions against potential losses. Overall, our conclusion is that, in Spain, a small gross drain on capital is expected in both commercial and savings banks under the baseline, despite severe economic deterioration. Under our adverse scenario, the gross drain on capital could reach €5 billion and €17 billion at commercial and savings banks, respectively (see Table 1.5 and Annex 1.3). These estimates are subject to considerable uncertainty and are relatively small in relation to both overall banking system capital and, importantly, the funds set aside under the resolution and recapitalization program set up by the government under the Fund for the Orderly Restructuring of Banks (FROB) of €99 billion. So far, three restructuring plans have been approved under the FROB involving a total of eight savings banks. The existing FROB scheme is currently scheduled to expire by June 2010. It is therefore important that the comprehensive resolution and restructuring processes financed through the FROB be under way before that date.

Table 1.5.Spain: Bank Writedowns and Capital(In billions of euros, unless otherwise shown)
Commercial BanksSavings BanksCommercial BanksSavings Banks
Baseline scenarioAdverse-case scenario
Tier 1/RWA ratio at 2009:Q21 (in percent)8.99.08.99.0
Expected writedowns,2010–122132633
Net drain on capital, 2010–123−51−36−152
Gross drain on capital,2010–12416517
Tier 1 capital at 2009:Q2199789978
Source: IMF staff estimates.Note: RWA = risk-weighted assets; for details refer to Annex 1.3.

While the overall health of German banks has improved since the peak of the crisis, banks may still face substantial writedowns on both their loan books and securities holdings, and the pace of realization has been uneven across the different categories of banks. Among main banking categories, Landesbanken have the highest loan writedown rate.20 Commercial banks, Landesbanken, and other banks still hold relatively large amounts of structured products, which results in particularly high writedown rates on their overall securities holdings. Strong capital positions at end-2009 and advanced writedown realization by commercial banks ensure their adequate capitalization (Table 1.6 and Annex 1.4). In contrast, Landesbanken, other banks, and, to a lesser degree also savings banks, are yet to incur a substantial part of total estimated writedowns and are projected to have a net drain on capital. Raising additional capital could prove particularly difficult for the Landesbanken, many of which remain structurally unprofitable and thus vulnerable to further distress. The impending withdrawal of the government’s support measures could intensify these vulnerabilities, stressing the need for expedited consolidation and recapitalization in this sector.

Table 1.6.Germany: Bank Writedowns and Capital(In billions of U.S. dollars, unless otherwise shown)
Commercial BanksLandesbanken and Savings BanksOther Banks1
Tier 1/RWA ratio at end-20092 (in percent)11.07.98.3
Expected writedowns, 2010:Q1–2010:Q43−34721
of which, loans:19274
of which, securities−222016
Net drain on capital, 2010:Q1–2010:Q44−272214
Tier 1 capital at end-2009218415545
Source: IMF staff estimates.Note: Foreign exchange rate assumed at 1 euro =1.4 U.S. dollars; RWA = riskweighted assets; for details refer to Annex 1.4.

Central and eastern European banking systems should be able to absorb the near-term peak in nonperforming loans, but are very vulnerable to weaker economic growth.

All banking systems remain susceptible to downside economic scenarios and this is especially so in central and eastern Europe (CEE). Nonperforming loan (NPL) ratios appear likely to peak during 2010 in the region (see Box 1.2), and banks appear sufficiently capitalized to absorb the baseline increase. However, another acceleration in NPL formation, were a weaker economic scenario to unfold, would leave banks significantly weakened and ill-prepared to absorb losses. As experience from previous crises shows, NPL ratios typically remain elevated for several years after the onset of a crisis, and coverage ratios of loss provisions to NPLs have already fallen to an average of about 65 percent in the CEE region, from pre-crisis levels of about 90 percent.21

While banks are still coping with legacy problems, they now face significant challenges ahead, suggesting the deleveraging process is far from over.

Deleveraging has so far been driven mainly from the asset side as deteriorating assets have hit both earnings and capital. Going forward, however, it is likely to be influenced more by pressures on the funding or liability side of bank balance sheets, and as new regulatory rules act to reduce leverage and raise capital and liquidity buffers.

The new regulatory proposals—enhanced Basel II and proposed revisions to the capital adequacy frame-work—point in the direction in which banks must adjust. The proposals will greatly improve the quality of the capital base, strengthen its ability to absorb losses, and reduce reliance on hybrid forms of capital. The quantitative impact study that will help calibrate the new rules is ongoing and final rules are to be published before end-2010, with a view to implementation by 2012. The outcome seems likely to be significant pressure for increases in the quality of capital, a further de-risking of balance sheets, and reductions in leverage. Once known—and possibly earlier—markets will re-rate banks on their perceived ability to achieve the new standards. Prudent bank management should therefore continue to build buffers of high-quality capital now in anticipation of the more demanding standards.

Box 1.2.Nonperforming Loans in Central and Eastern Europe: Is This Time Different?

At what levels and when could nonperforming loan ratios be expected to peak in central and eastern Europe, based on experience from previous economic downturns?

Nonperforming loans (NPLs) have increased substantially in the central and eastern Europe (CEE) region since the onset of the global financial crisis.

This box presents a top-down framework for assessing the deterioration in bank asset quality and analyzing NPLs under different scenarios, based on historical experience in emerging markets.1

The estimation sample consists of annual data between 1994 and 2008 for Asian and Latin American economies, as well as South Africa and Turkey.2

The data reveal that emerging market NPL ratios tend to rise rapidly in a crisis, and remain more than twice as high as before the initial shock for more than four years (first figure). The technical details on the data and the estimations are given in Annex 1.6 on the IMF’s GFSR website.

Historical Dynamics of Emerging Market Nonperforming Loan Ratios around Large Increases in Year t

Source: IMF staff estimates.

Note: Average of indices for Argentina, Chile, Colombia, Dominican Republic, Indonesia, Malaysia, Philippines, Turkey, and Uruguay.

Nonperforming loans in the CEE region have developed largely in line with patterns observed in previous emerging market downturns.

Simulations for the CEE region starting in 2008 indicate that bank asset quality has developed largely as would be expected based on historical experience in emerging markets, considering the size of the GDP shocks that hit the CEE region.3 The model-based projections fairly accurately predict the increase in NPL ratios across subregions in the CEE region during 2009, with the largest increase predicted in the Baltic countries and the smallest in the CE-3 countries (second figure).4 However, the model simulations envisage sharp currency depreciations in response to the large negative GDP shocks that have hit most countries in the CEE region. This explains why the model overpredicts the increase in NPL ratios, especially in the Baltic countries, as CEE exchange rates have successfully been stabilized on the back of international policy coordination and financial backstops.5

Simulated Average Nonperforming Loan Ratios

(In percent)

Source: IMF staff estimates.

Note: CE-3 = Czech Republic, Hungary, and Poland; CIS = Russia and Ukraine; SEE = Bulgaria, Croatia, and Romania.

Simulations suggest that NPL ratios will peak during 2010 in most CEE countries under the WEO baseline scenario for GDP growth.

The simulations indicate that most of the increase in NPL ratios occurred during 2009, but suggest that bank asset quality will improve only gradually in 2011 for most countries, even if GDP growth recovers during 2010 as projected in the World Economic Outlook (WEO). In the Commonwealth of Independent States (CIS), the simulations suggest a decline in the NPL ratio by the end of 2010 on the back of a more vigorous projected economic recovery. However, loans that have been restructured may turn up in the official NPL statistics with a delay, when interest rates are normalized and rolling over of NPLs becomes more costly in terms of interest revenue forgone, which could mean that reported asset quality in the CIS may also continue to deteriorate in 2010.

In a weaker growth scenario, NPL ratios would continue to increase substantially in 2010.

In an adverse scenario where GDP is 4 percentage points lower than the WEO baseline in 2010 and 2 percentage points lower in 2011, the simulations indicate that NPL ratios would increase by around one-third during 2010 in all subregions except the CIS, and would remain elevated in 2011.

Note: This box was prepared by Kristian Hartelius.1 The approach taken is to estimate coefficients for the relationship between GDP growth, exchange rate movements, and the ratio of NPLs to total loans for economies outside the CEE region, and then project NPL ratios for the CEE region based on those coefficients. The approach has the advantage of overcoming data limitations in NPL time series for the CEE region, which are often too short to capture full credit cycles. The approach cannot be expected to deliver very precise country-level forecasts, but can serve as a useful complement to country-specific, bottom-up stress tests.2 The economies included in the estimation sample are Argentina, Chile, Colombia, the Dominican Republic, Indonesia, Malaysia, Mexico, Peru, the Philippines, South Africa, Taiwan Province of China, Thailand, Turkey, Uruguay, and Venezuela.3 Although foreign bank ownership and foreign currency lending reached extreme levels in the CEE region in the runup to the current crisis, they were also important elements in many emerging market crises in the past two decades, which enables the model to explain the European data relatively well.4 The group labeled Baltics comprises Estonia, Latvia, and Lithuania. The group labeled CE-3 comprises the Czech Republic, Hungary, and Poland. The group labeled SEE comprises Bulgaria, Croatia, and Romania, and the group labeled CIS comprises Russia and Ukraine. There is considerable variation in NPL ratios within these groupings, as detailed in Table 24 of the Statistical Appendix.5 As noted in Annex 1.6, the model predictions fit the Baltic data better, when controlling for actual exchange rate developments.

Few banks can expect retained earnings alone to lift them to the new capital standards …

Some banks are confident that they will be able to raise prices to maintain their recent high returns on equity, but history suggests they may struggle to do so. To assess this, U.S. bank lending rates were regressed on a number of macroeconomic and structural variables.22 The results suggest that the wide margins and pricing power banks have enjoyed in recent quarters is likely to dissipate as the yield curve flattens (Figure 1.15).

Figure 1.15.Banks’ Pricing Power—Actual and Forecast

(In percent)

Sources: Federal Reserve; Federal Deposit Insurance Corporation; and IMF staff estimates.

For the few banks that have significant capital markets operations, investment banking revenues are unlikely to provide the bonanza they did in 2009, as interest rates and exceptional liquidity conditions normalize and competition returns. Some corporate issuance in 2009 was precautionary to take advantage of low historical rates, and is unlikely to be repeated. The decline is unlikely to be fully offset by a rise in mergers and acquisition activity. At the same time, the move to central counterparty clearing of many contracts that were previously traded over the counter (at relatively wide spreads) could put downward pressure on one important revenue stream for the larger banks.

…and funding pressures are set to mount, pushing up costs.

The April 2009 GFSR cautioned that large banks generally needed to extend the maturity of their debt. However, they have seemingly been deterred by the historically high spreads at which they would issue, and the availability of ample, cheap central bank funding. The wall of refunding needs is now bearing down on banks even more than before, with nearly $5 trillion in bank debt due to mature in the coming 36 months (Figure 1.16). This will coincide with heavy government issuance and follow the removal of central bank emergency measures. In addition, banks will have to refinance securities they structured and pledged as collateral at various central bank liquidity facilities that are ending.

Figure 1.16.Bank Debt Rollover by Maturity Date

(In billions of U.S. dollars)

Source: Moody’s.

Banks must move further to reduce their reliance on wholesale markets, particularly short-term funding, as part of the deleveraging process. The investor base for bank funding instruments has been permanently impaired as structured investment vehicles (SIVs) and conduits have collapsed, and banks are significantly less willing to fund one another unsecured. Central banks have provided a substitute with their liquidity facilities, but extraordinary support is set to be scaled back over time. This could put pressure on spreads, and particularly in those markets where the large retained securities portion of bank assets highlights the continuing disruption of mortgage securitization markets (Figure 1.17). However, a significant portion of these securities are being funded through the Bank of England and European Central Bank facilities. In contrast, the U.S. Federal Reserve has purchased securities outright—largely through the quantitative-easing program—and has thus assisted banks through a more durable asset transfer process (see Annex 1.8 on the IMF’s GFSR website).

Figure 1.17.Government-Guaranteed Bank Debt and Retained Securitization

(As percent of national banking system deposits)

Sources: Autonomous Research; European Central Bank; and IMF staff estimates.

If banks fail to shrink their assets to reduce their need for funding or do not issue sufficient longer-term wholesale funding, they will inevitably be competing for the limited supply of deposit funding (Autonomous Research, 2009).

Indeed, there are already signs that deposit funding is becoming more expensive. The funding spread—the difference between the LIBOR market and what banks pay for deposits—is already heavily negative in the United States and United Kingdom. Even in the euro area, where the funding spread has typically been a positive 175 basis points in normal times, it has now turned negative (Figure 1.18). As a result, even though spreads on assets have widened further in recent months, bank top-line profitability is under pressure in all these regions.23

Figure 1.18.Euro Area Banking Profitability

(In basis points, on volume-weighted new business, excluding overdrafts)

Sources: Autonomous Research; and IMF staff estimates.

Notes: Funding spread = three-month Euribor less volume-weighted average of rates paid on new deposits to households and corporates. Asset spread = interest income on volume-weighted average of rates paid on new lending to households and corporates, less three-month Euribor.

Slow progress on stabilizing funding and addressing weak banks could complicate policy exits from extraordinary support measures.

The planned exit from extraordinary liquidity measures may be complicated by the need for banks generally to extend the maturity of their liabilities and by the presence of a tail of weak banks in the system. Although LIBOR-overnight index swap (OIS) spreads have narrowed, there are ample other signs that money markets have yet to return to normal functioning. The contributions of LIBOR and EURIBOR panel banks to their respective benchmarks remain more dispersed than before the crisis; credit lines for medium-sized banks, and banks that required substantial public support, have generally not yet been reinstated; and turnover in the repo market for any collateral other than higher-rated sovereign paper remains low.

Although substantially improved, there are lingering signs that some institutions remain dependent on central bank liquidity facilities. National central bank data (Figure 1.19) indicate that a number of euro area banks have increased their reliance on European Central Bank (ECB) funding over recent quarters, suggesting their demand is to meet genuine funding needs rather than simply to finance attractive carry trades. Some widening of both financial and sovereign CDS spreads is likely as the withdrawal of extraordinary ECB measures draws nearer. In the United States, borrowing at the Federal Reserve’s discount window has fallen steadily but remains well above pre-crisis levels.24

Figure 1.19.Net European Central Bank (ECB) Liquidity Provision and Credit Default Swap (CDS) Spreads

(Changes December 31, 2006–October 31, 2009)

Sources: Bloomberg L.P.; and euro area national central banks.

Note: Changes in net liquidity provisions are expressed as a percent of bank total assets, while the squares reflect the change in sovereign credit default swap (CDS) spreads between December 1, 2006, and October 31, 2009.

What does this mean for financial policies?

The consequence of these deleveraging forces will be to highlight the extent of overcapacity in the financial system as costs rise, push up competition for stable funding sources, and intensify pressure on weak business models (Figure 1.20). Thus, policy will need to ensure that this next stage of the deleveraging process unfolds smoothly and ends in a safe, vital, and more competitive financial system. This will include addressing too-important-to-fail institutions in order to ensure fair pricing power throughout the financial system and to guard against rising concentration as the size of financial systems shrinks (see Annex 1.5).

Figure 1.20.Bank Credit to the Private Sector

(In percent of nominal GDP)

Sources: Haver Analytics; and IMF staff estimates.

Note: Dotted lines are estimates. Year of credit peak in parentheses.

The viability of weaker segments of banking systems is likely to come into question given new regulations, deleveraging forces, and the withdrawal of extraordinary central bank support facilities. In a number of countries, a significant part of the banking system lacks a viable business model, or suffers from chronic unprofitability. In the case of the European Union, the need for rationalization of the sector can be seen in the striking variability of banking returns (Figure 1.21). The German system, for example, suffers from weak overall profitability, and a large tail of unprofitable banks—primarily the nation’s Landesbanken. Moreover, care will be needed to ensure that too-important-to-fail institutions in all jurisdictions do not use the funding advantages their systemic importance gives them to consolidate their positions even further.

Figure 1.21.Bank Return on Equity and Percentage of Unprofitable Banks, 2008

(In percent)

Sources: Bankscope; EU Banking Supervision; Federal Deposit Insurance Corporation; and IMF staff estimates.

Note: Size of circle corresponds to relative size of bank loan stock at end-2008. Return on equity is as defined by the Banking Supervision Committee (BSC) of the European System of Central Banks in each of its reports. Some countries were reporting under national accounting standards in the earlier BSC reports. For the United States and Japan, return on equity is net income divided by total equity according to Federal Deposit Insurance Corporation and Bankscope data, respectively.

If excess banking capacity is maintained, the costs are felt across the whole economy and are not just limited to support costs faced by taxpayers. Weak banks normally compete aggressively for deposits (on the back of risk-insensitive and underpriced deposit insurance), wholesale funding, and scarce lending opportunities, so squeezing margins for the whole system. Unless tightly constrained, institutions that are either government-owned, or have explicit or implicit government backing, have also demonstrated in many cases a tendency to invest in risky assets of which they have little experience—some of the German Landesbanken being only the latest examples—so adding to systemic risks and the likelihood of future bailouts.

Japan presents a telling example of the challenges banks face in a crowded sector amid low growth and muted or negative inflation. The exceedingly low nominal rates leave banks increasingly pressed to maintain profitability. Over the past 20 years, the average return on bank assets has been negative, partly owing to the disposal of nonperforming loans after the bubble burst. Low returns on assets make it hard for banks to rely on loan revenues to absorb credit losses, and volatility in the values of equity holdings leads to large fluctuations in bank profits (Figure 1.22). Tangible equity at the largest banks is low, and is likely to be put under further pressure by the latest Basel proposals. Options for improving profitability—taking greater market risks, offshore expansion, higher lending margins, or balance sheet shrinkage—all have their difficulties, both economically and politically. Thus, improving profitability is a critical challenge for Japanese banks.

Figure 1.22.Banking System Profitability Indicators

(In percent, average over 2001–08)

Sources: Bankscope; and IMF staff estimates.

Note: Different industry structures and accounting conventions make comparison across countries/regions difficult.

D. Risks to the Recovery in Credit

The credit recovery will be slow, shallow, and uneven. Credit supply remains constrained as banks continue to repair balance sheets. Notwithstanding the weak recovery in private credit demand, ballooning sovereign needs may bump up against supply. Policy measures to address capacity constraints, along with the management of fiscal risks, should help to relieve pressures on the supply and demand for credit.

Credit availability is likely to remain limited …

Two years ago, the GFSR described the possibility that credit growth might drop to near zero in the major economic areas affected by the crisis, as has now happened. For example, in the United States, real credit growth has fallen sharply when compared with past recessions (Figure 1.23).25

Figure 1.23.Real Nonfinancial Private Sector Credit Growth in the United States

(In percent, year-on-year)

Sources: Haver Analytics; National Bureau of Economic Research; and IMF staff estimates.

Note: This figure compares recent real nonfinancial private sector credit growth to that in past recessions, from 1970 to 2001. Past recession dates are from the National Bureau of Economic Research. For this figure, the end of the recent recession is assumed to be 2009:Q3, the first quarter of positive growth.

The last few rounds of bank lending surveys, however, have indicated that lending conditions are tightening at a slower pace, and in some sectors have already begun to register an outright easing. Figure 1.24 indicates that credit growth has lagged lending conditions by around four quarters, suggesting that the worst of the credit contraction may be over. Nevertheless, as discussed in Section C, it is likely that bank credit will continue to be weak as balance sheets remain under strain and funding pressures increase. Banks’ reluctance to lend is evident in still-elevated borrowing costs and strict lending terms (for example, stringent covenants and short maturities) in some sectors.

Figure 1.24.Average Lending Conditions and Growth in the Euro Area, United Kingdom, and United States

Sources: Haver Analytics; central bank lending surveys; and IMF staff estimates.

Companies have increasingly drawn on nonbank sources of credit in recent quarters as banks have tightened credit supply (Figure 1.25).26 However, nonbank credit has only provided a partial substitute for bank lending and total credit growth has fallen. In general, in addition to households, small and mediumsized enterprises (SMEs) tend to be largely reliant on bank lending and so still face credit constraints. Furthermore, the supply of credit that has been available from central banks during the crisis is set to wane this year.27 Central bank commitments imply under $400 billion of securities purchases in the euro area, United Kingdom, and United States, in total, compared with around $1.9 trillion in 2009. So even though we expect nonbank capacity to increase over the next two years, as economies start to recover, total credit supply, including bank lending, is set to recover slowly (Figure 1.26).

Figure 1.25.Contributions to Growth in Credit to the Nonfinancial Private Sector

(In percent, year-on-year)

Sources: Haver Analytics; and IMF staff estimates.

Figure 1.26.Nonfinancial Private Sector Credit Growth

(In percent, year-on-year)

Sources: Haver Analytics; and IMF staff estimates.

Note: The dotted lines show projected credit growth. If credit demand is estimated to exceed capacity, after meeting sovereign borrowing needs, then credit is assumed to be constrained by available capacity, including the impact of government and central bank policies.

… and sovereign needs are set to dominate credit demand …

Sovereign issuance surged in 2009 to record levels in all three regions as crisis-related interventions and fiscal stimulus packages led to an unprecedented increase in government borrowing requirements (Figure 1.27). Government borrowing will remain elevated over the next two years, with projected financing needs for both the euro area and the United Kingdom well above previous expectations in the October 2009 GFSR. Burgeoning public sector demand risks crowding out private sector credit if funds are diverted to public sector securities. In addition, as discussed in Section B, a rise in sovereign risk premia could raise private sector borrowing costs.

Figure 1.27.Total Net Borrowing Needs of the Sovereign Sector

(In percent of GDP)

Sources: National authorities; and IMF staff estimates.

Notwithstanding these risks, private sector demand growth is likely to remain subdued as households and corporates restore balance sheets. The need for private sector deleveraging varies across region and sector (Figure 1.28). For instance, in the United States, households are at the beginning of the deleveraging process, while nonfinancial companies have less of a need to reduce leverage. By contrast, in the euro area and the United Kingdom, nonfinancial corporate debt as a share of GDP is much higher, having experienced a rapid run-up during the pre-crisis period. This, together with the increase in household leverage, means that the United Kingdom’s nonfinancial private sector debt, at over 200 percent of GDP, is one of the highest among mature economies.28

Figure 1.28.Credit to GDP

(In percent)

Source: IMF staff estimates.

Note: Dashed lines represent forecasts.

… which is likely to result in financing gaps.

Updating the analysis of credit demand and capacity in the October 2009 GFSR suggests that ex ante financing gaps will remain in place for all three regions in 2010 (Table 1.7).29 There is some uncertainty around our estimates for both credit demand and capacity, so the size of the financing gap, which is the difference between these two estimates, is approximate. Nevertheless, the work is useful in highlighting the relative size of the ex ante financing gaps. As in the October 2009 GFSR, the analysis suggests that the United Kingdom could have the largest gap (around 9 percent of GDP over 2010–11) as weak bank capacity struggles to keep up with surging sovereign issuance. We expect smaller financing gaps in the euro area in 2010 (around 2 percent of GDP), and a similar gap in the United States in 2010, which is closed by remaining central bank commitments to purchase securities.30

Table 1.7.Projections of Credit Capacity for and Demand from the Nonfinancial Sector
20102011
AmountGrowthAmountGrowth
Euro Area
Total credit capacity available for the nonfinancial sector5402.89004.6
Total credit demand from the nonfinancial sector6903.51,0405.1
Credit surplus (+)/shortfall (–) to the nonfinancial sector−150−140
Memo: Central bank and government committed purchases130
Credit surplus (+)/shortfall (–) in percentage of GDP−2−1
United Kingdom
Total credit capacity available for the nonfinancial sector501.31804.7
Total credit demand from the nonfinancial sector2005.13007.4
Credit surplus (+)/shortfall (–) to the nonfinancial sector−150−120
Memo: Central bank and government committed purchases110−120
Credit surplus (+)/shortfall (–) in percentage of GDP−10−8
United States
Total credit capacity available for the nonfinancial sector1,7205.22,4507.1
Total credit demand from the nonfinancial sector2,0005.82,5006.8
Credit surplus (+)/shortfall (–) to the nonfinancial sector−280−50
Memo: Central bank and government committed purchases1360
Credit surplus (+)/shortfall (–) in percentage of GDP−20
Source: IMF staff estimates.Note: Amount is in billions of local currency units rounded to the nearest ten. Growth is in percent.

At face value, ex ante financing gaps imply that ex post either borrowing needs to be scaled back to equalize the lower supply, or that market interest rates will need to rise. Any increases in interest rates, however, are unlikely to be uniform, and certain sectors, such as SMEs and less creditworthy borrowers, may face higher borrowing costs. In particular, given the surge in public sector borrowing and expected deleveraging by the banking sector, upward pressure on interest rates is likely to result.

Policy action could help to relieve these pressures. For example, the authorities should carefully assess the implications of their policy actions and exit strategies, as well as their timing, on the quantity of credit available to support the economic recovery. The implementation of measures to manage fiscal risks and limit rises in public sector credit demand, along with policies to address weaknesses in the banking system—such as strengthening securitization markets, as discussed in the October 2009 GFSR—should also be considered. There is the possibility that central bank support measures, including purchases of securities, may still be needed in some cases to offset the retrenchment in credit capacity.

E. Assessing Capital Flows and Bubble Risks in the Post-Crisis Environment31

Prospects for strong growth, appreciating currencies, and rising asset prices are pulling capital flows into Asia-Pacific (excluding Japan) and Latin American countries, while push factors—particularly low interest rates in major advanced economies—are also key. Against this backdrop, this section assesses the drivers of recent portfolio capital flows, and both the near- and medium-term prospects of systemic asset price bubbles forming. It finds no evidence of systematic bubbles in advanced and emerging market economies and across asset classes in the near term. However, if the current environment of low interest rates, abundant liquidity, and capital flows persists, history suggests that bubbles could form in the medium term. Moreover, vigilance is warranted given that it is notoriously difficult to identify such financial imbalances ex ante.32

Last year saw a welcome recovery in portfolio capital flows toward emerging markets and other advanced economies. “Pull factors” such as relative growth differentials, appreciating currencies, and rising asset prices are driving the resurgence. The flows have been targeted to countries perceived by investors to have better cyclical and structural growth prospects, like Brazil, China, India, and Indonesia, as well as their trading and financial partners, including commodity exporters.

However, “push factors,” such as low interest rates in major advanced economies and much-improved funding market conditions, are also key drivers of capital flows.33 Low policy rates have encouraged investors to shift their precautionary cash holdings into riskier assets. For example, U.S. money market mutual fund assets have fallen by over half a trillion dollars since March 2009, as central bank policy and operations helped to put downward pressure on broader money market interest rates and risk premiums (Figure 1.29).

Figure 1.29.Low Short-Term Interest Rates Are Driving Investors Out of Cash

Sources: Bloomberg L.P.; and Investment Company Institute. Note: OIS = overnight indexed swap.

When taken together, these push and pull factors may create a conducive environment for future asset price appreciation, and this, in turn, has heightened concerns about asset price bubbles forming. The surge in portfolio inflows also raises concerns about vulnerabilities to sudden stops, once global monetary and liquidity conditions are tightened or if risk appetite were to diminish.

Although portfolio flows were strong in 2009, other capital flows, which include cross-border bank lending, and direct investments have not recovered to the same extent. This reflects the persistent deleveraging by mature market banks and the still-added tepid desire by firms for cross-border mergers and acquisitions and green field development. For example, the nonportfolio, non-FDI (foreign direct investment) category of the capital accounts of Brazil, Korea, and Russia remained negative in the data available for 2009, and FDI remains subdued in Korea and Russia.34

Further flows could emerge as the crisis has led investors to reconsider the balance of risk and return in emerging and other advanced economies.

The crisis has altered perceptions about risk and return in mature relative to emerging markets. Perceptions of sovereign credit risks have moved in favor of emerging markets and some other advanced economies, primarily due to unfavorable debt dynamics in the major advanced economies and southern Europe (see Section B). In contrast, the average credit rating of issuers in JPMorgan’s Emerging Market Bond Index improved to the lowest investment grade rating during the crisis, reflecting upgrades to some emerging market sovereigns, notably Brazil. Additionally, emerging market equities continued to register higher volatility-adjusted returns than developed markets during and after the fall of 2008 (Figure 1.30).

Figure 1.30.Emerging Market Returns Better on a Volatility-Adjusted Basis

(In percent)

Sources: Bloomberg L.P.; MSCI Barra; and IMF staff estimates.

Note: Volatility-adjusted returns = three-year rolling log returns/three-year historical standard deviation of returns.

The favorable performance of emerging market assets relative to mature market assets has prompted growing interest by global investors in raising their asset allocations to emerging markets and other advanced economies. For example, retail investors and hedge funds are adding to their emerging market portfolios in the near term, facilitated by the increasing development of exchange-traded funds (ETFs) targeting emerging markets broadly and countries like Brazil and China.35 In debt markets, the outstanding stock of emerging market debt has grown to over $7 trillion, compared to under $2 trillion in the mid- to late 1990s, and benchmark bond indices are garnering greater acceptance by institutional investors.36

However, recent surveys indicate that institutional investors’ home bias has only changed in a gradual fashion over the years.37 Some estimate that emerging market equities account for just 5 to 9 percent of global equity exposures, far lower than their share of global market capitalization of 12 percent, and the 27 percent share implied by a GDP-weighted global equity index.38 Nevertheless, even small shifts in portfolio allocations could translate into significant capital inflows to emerging markets and other advanced economies. They also could add to market volatility and test an individual market’s capacity to absorb inflows, especially if flows are concentrated in particular asset classes or in a short period of time.

Portfolio flows have rebounded strongly …

Strong portfolio equity flows into emerging markets and other advanced economies in 2009 primarily reflect a recovery trade from the deep retrenchment in 2008 as shown by the green bars in Figure 1.31. However, Latin America was the only region where 2009 inflows exceeded 2008 outflows by a wide margin as shown by the higher ratio of net flows. In general, regions viewed as having lower growth prospects and structural challenges are receiving smaller inflows. For example, equity funds with exposure to Europe, the Middle East, and Africa recovered less than one-half of the outflows in 2008, and funds continued to flow out of major advanced economy equity funds. Within these broad regions, however, some countries have experienced a rapid surge in portfolio inflows; for example, Brazil was responsible for a large portion of flows to Latin America.

Figure 1.31.Cumulative Retail Net Flows to Equity and Debt Funds

(In percent of initial assets under management)

Sources: Emerging Portfolio Fund Research, Inc.; Investment Company Institute; and IMF staff estimates.

Note: Numbers underneath bars represent ratio of net flows in 2009 to those in 2008. “+” (”–”) when net inflows turned positive (negative) in 2009 from negative (positive) in 2008. EMEA represents Europe, Middle East, and Africa. Core mature markets include Japan, United States, and western Europe. U.S. corporate represents inflows into U.S. mutual funds invested primarily in corporate debt.

Investor flows into global corporate and emerging market external bonds and notes have also been strong in 2009, reflecting the reopening of global credit markets and an expected compression in credit spreads after extreme default scenarios were priced in at the height of the crisis.39 Inflows into U.S. investment-grade and high-yield funds in 2009 were multiples above their 2008 outflows, but those to emerging market debt funds had not yet fully recovered. Even though emerging market external debt issuance reached a record of over $200 billion, part of this issuance was required to meet the large refinancing needs that were highlighted in the October 2009 GFSR. Indeed, emerging market corporates and banks still face refinancing needs of about $450 billion for foreign-currency-denominated debt over the next two years, with a concentration of maturities this year (Figure 1.32).

Figure 1.32.Refinancing Needs for Emerging Market and Other Advanced Economies Remain Significant

(In billions of U.S. dollars)

Sources: Bloomberg L.P.; and IMF staff estimates.

Notes: Repayment of principal and coupon on bonds and principal only on foreign currency loans. Asia = China, India, Indonesia, Malaysia, Korea; Latin America = Argentina, Brazil, Chile, Mexico; Europe and CIS = Hungary, Kazakhstan, Poland, Russia, Turkey, Ukraine; Middle East and Africa = South Africa, United Arab Emirates. CIS = Commonwealth of Independent States.

… but have portfolio flows caused asset prices to reach excessive valuations?

Compared with prior crisis episodes, asset prices have moved along a broadly similar recovery path (Figure 1.33). For example, the price of emerging market equities in real terms has recovered to the median level of historical correction episodes. Also, the depth of the trough and the pace of recovery during the Asian crisis were similar to those during the current crisis.

Figure 1.33.Emerging Market Real Equity Prices: Historical Corrections

(Pre-correction peak = 100)

Sources: Bloomberg L.P.; World Economic Outlook database; and IMF staff estimates.

Note: Median, maximum, and minimum based on five correction episodes following peaks in July 1990, August 1994, July 1997(Asian crisis), February 2000, and March 2002, but excluding this crisis that started in October 2007, and are based on the MSCI Emerging Markets Index.

A few asset classes have attracted particular attention—equity and property prices, local sovereign yield, and external sovereign credit spreads—but we find little evidence that bubbles have formed in these segments in the near term (Table 1.8).40 The table is not meant to be a definitive predictor of a bubble in an individual market or across markets, but rather to be a useful tool to compare valuations across time and economies in order to make a preliminary identification of potential hot spots that bear deeper investigation.41 For advanced economies, equity valuations are within historical norms.42 Forward-looking valuations are generally below the peaks prior to the collapse of Lehman Brothers as well as the bursting of the U.S. tech bubble in 2000. There are also few signs of overvaluation in local sovereign debt markets (with the exception of Japan), including in mature economies, where official bond purchase programs have been pursued after controlling for monetary and financial conditions.43

Table 1.8.Asset Class Valuations(Z -score)
EquityResidential Real EstateLocal Sovereign YieldLocal Corporate CreditExternal Sovereign Credit
Backward-lookingForward-looking
Shorter horizonLonger horizonPrice to rentPrice to income
Asia-Pacific
Australia−0.30.0−2.11.91.5−0.1
China0.6−0.11.9−1.4
Hong Kong SAR0.30.62.12.0
India0.80.70.20.4−1.0
Indonesia1.10.2−1.3−1.3−0.6−0.5
Japan−1.8−1.1−2.6−1.9−2.01.6
Korea0.6−0.60.6−0.8−0.6
Malaysia0.0−0.4−1.8−0.90.50.2
Philippines−0.20.0−0.9−1.30.80.2
Taiwan Province of China−0.2−0.80.3−1.0
Thailand−0.1−2.7−2.3−0.5
Europe, Middle East and Africa
Austria−1.0−0.7−0.1−1.2−0.30.4
Belgium0.40.3−0.31.01.40.4
Czech Republic−0.4−0.80.61.6−0.2
Denmark0.40.21.51.0
France−1.8−0.7−1.12.21.70.00.4
Germany−0.7−1.0−1.3−1.7−1.60.10.4
Greece−0.4−1.4−1.9−0.70.90.4
Hungary−0.20.0−1.10.6−1.3
Ireland−0.92.10.91.10.8−0.70.4
Israel0.0−0.6−0.61.0
Italy−1.0−1.0−0.61.00.6−0.70.4
Netherlands0.0−0.4−1.01.51.40.4
Norway−0.4−0.51.91.3
Poland−0.80.1−0.4−1.0−0.7−0.2
Portugal−1.3−0.4−0.50.4
Russia−0.2−0.4−1.1−0.3−2.90.5
South Africa0.10.2−0.10.2−1.10.7
Spain−0.9−0.90.21.51.40.70.4
Sweden−0.10.00.22.60.8
Switzerland−0.8−0.60.9
Turkey−0.10.31.40.3
United Kingdom−0.4−0.8−0.91.11.4−0.2
Americas
Argentina0.1−1.5−0.4−0.3
Brazil0.81.80.10.1
Canada−0.5−0.20.41.91.3−0.2
Chile1.30.7−1.70.4
Colombia1.21.9−2.01.5−0.70.0
Mexico0.41.20.3
Peru0.70.2−2.40.7
United States−0.6−0.6−0.11.3−0.40.51.8
Sources: Bloomberg L.P.; IBES; OECD; and IMF staff estimates.Note: A z-score represents the deviation of latest observation from either the period average or model value expressed in the number of standard deviations. Green signifi es less than 1.5 standard deviations above, pink 1.5−2 standard deviations above, and magenta greater than 2 standard deviations above. Backward-looking equity valuation is calculated as the unweighted average of z-scores of dividend-yield and price-to-book ratios. Forward-looking equity valuation represents z-score of 12-month forward price-to-earnings ratios (shorter horizon) and z-score of dividend discount model estimates (longer horizon). Valuation of local sovereign yields, local corporate spreads, and external sovereign spreads are based on z-score of the deviation from econometric model value. For methodologies see Annex 1.9 on the s GFSR website.

In credit markets, the narrowing of spreads appears to be consistent with macroeconomic fundamentals and reduced risk aversion in Europe, though the extent of credit spread compression is somewhat greater than model predictions in the United States. Emerging market sovereign external credit spreads appear broadly consistent with fundamentals. In the foreign exchange markets, the recent pick-up in cross-border financial flows to emerging economies has not led to substantial changes in real effective exchange rates, as economies have generally preferred to build up reserves in response to inflows.44

There are some valuation hot spots in a few economies that have attracted significant portfolio investment. For example, in two Latin American economies, 12-month forward price-to-earnings ratios exceed historical averages by 1.5 standard deviations or more. There are also signs that property prices may be stretched in some Asia-Pacific economies with price-to-rent and/or price-to-income ratios 1.5 or more standard deviations beyond historical averages.45Box 1.3 takes a closer look at the Asia-Pacific real estate markets, where housing prices and transaction volumes have surged to very high levels. However, these are primarily occurring in the high-end market.

Rising asset prices and portfolio flows have coincided with some pick-up in leverage.

The financial flows in 2009, especially to emerging markets and other advanced economies, have primarily been attributed to portfolio reallocation by unlevered institutional and retail investors. Leveraged investors, such as hedge funds, remain smaller and less leveraged than before the financial crisis, but they have recouped a significant amount of their crisis-related losses in 2009. With $2.1 trillion under management at the end of 2009, the hedge fund universe has returned to three-quarters of its pre-crisis peak.

Additionally, the available evidence suggests that the incentives for “carry trade” have increased steadily over the past year, but they are yet to reach the high levels of 2006 and 2008. For Australia, carry trade indicators have not changed significantly since late 2008 (Figure 1.34).46 Furthermore, mature market banks’ willingness to lend is only gradually improving, and the growth of domestic bank credit in most emerging market and other advanced economies is only beginning to turn around. The exception is in China, where credit growth soared through mid-2009 and remains at a fast pace, although decelerating (Figure 1.35).

Figure 1.34.Incentives for Foreign Currency Carry Trades Are Recovering

Sources: Bloomberg L.P.; and IMF staff estimates.

Figure 1.35.Real Domestic Credit Growth and Equity Valuation

(Z-score)

Sources: IMF, International Financial Statistics database; and IMF staff estimates.

Note: Credit growth is 12-month moving average of year-on-year growth of real credit to the private sector (or other sectors). Deflated by consumer price index. Historical data go back to the early 1980s. Equity valuation is 12-month price/earnings ratios in February. Historical data go back to the late 1980s.

What could put asset prices on a bubble trajectory?

Although there is only limited evidence of stretched valuations across countries in the near term, current conditions could give rise to potential for bubbles to form in the medium term. Typically, for bubbles to have a systemic impact requires substantial overvaluation in several risk assets for a protracted period that is supported by excessive leverage, often in the form of concentrated bank lending (see Box 1.4). Indeed, the abundant liquidity that remains within advanced country banking systems, if unlocked, has the potential to boost the prices of risk assets, unless carefully monitored and controlled.

Expansionary financial conditions could fuel asset price inflation, potentially setting off an upward cycle of asset prices and credit through a financial accelerator mechanism.47 The challenge of managing the consequences of capital flows is particularly acute for countries with limited exchange rate flexibility. Such regimes may exacerbate the impact of capital flows on local liquidity conditions, while attracting inflows on expectations of future currency appreciation.48

Policymakers have responded to the rising capital flows, but continued vigilance is needed as current conditions remain supportive of further inflows. Governments have started to lean against increasing asset price pressures by beginning to remove some of the support to the financial system with the aim of reining in high credit growth. Thus, close monitoring and a variety of macroprudential actions are warranted to help ensure that leverage and concentration do not reach excessive levels. Chapter 4 discusses the policy options and previous experience in addressing capital inflows. It notes that there have been varying degrees of success with different types of measures and controls to mitigate their impact on asset prices and inflation.

F. Policy Implications

The health of the global financial system has improved, and the world has avoided a full-blown depression. However, risks remain elevated due to the still-fragile nature of the recovery and the ongoing repair of balance sheets. Attention has shifted toward sovereign risks that could undermine stability gains and take the credit crisis into a new phase, as we begin to reach the limits of public sector support for the financial system and the real economy. Bank funding pressures are emerging as the key risk from the ongoing deleveraging process, and may replace capital as the dominant constraint to the normalization of credit. To maintain the momentum in the reduction of systemic risks, and to prepare for exits from extraordinary policy support, further action is required of policymakers in several key areas.

Box 1.3.Asian Residential Real Estate Markets: Bubble Trouble?

Asian real estate markets rebounded quickly in the second half of 2009 from their 2008 downturn, distinguishing this region from the other parts of the world (first figure). While much of the world continued to grapple with the housing bust, housing prices and transaction volumes recovered in certain eastern Asian economies (notably China, Hong Kong SAR, Korea, and Singapore) and closely linked advanced economies (Australia and New Zealand).1 In particular, prices for high-end properties in major metropolitan areas exceeded their 2008 peaks, gradually spilling over to the broader market. This development echoes the rally in other risky assets such as regional equities and bonds.

The rebound has been mainly driven by unprecedented policy measures to mitigate the impact of the global financial crisis and the ensuing return of risk appetite. First, mortgage rates are at historical lows as central banks around the globe have cut policy rates. Second, reviving real estate loan growth helped pull the markets out of the trough (second figure), especially in China. Third, governments in China and Korea introduced housing-related tax initiatives in late 2008 to revive domestic real estate markets. Finally, capital inflows have played an important role. In Singapore, foreigners and companies accounted for 12.5 percent of the third-quarter home purchases in 2009, rising from 8 percent in the previous quarter. In Hong Kong SAR, an influx of buyers from mainland China pushed prices up, especially for luxury apartments.

Metrics of affordability are mixed, but on balance suggest that valuations risk becoming stretched (third and fourth figures). Although the average price-to-income index for the east Asian economies has risen only modestly, the price-to-rent index is elevated. As typically happens in housing bubbles, many purchasers may have been buying in the expectation of price appreciation, rather than simply for dwelling purposes.

The booming Asian real estate markets may pose risks to financial stability as banks are increasingly vulnerable to a price correction (fifth figure).2 In addition, because the majority of mortgage loans in Asian economies carry floating rates, the widely anticipated rate hikes in the region will increase the burden on household balance sheets.3 Moreover, as many municipal budgets in China tend to rely heavily on revenue from land sales, a real estate market downturn may put their fiscal situation into question.4

In light of these potential risks, authorities in the region have taken measures to cool real estate markets, including tighter requirements on mortgage lending, increasing land supply, and re-imposition of higher transaction taxes. The average loan-to-value ratio of new mortgage loans in Hong Kong SAR has dropped significantly from its peak in June, and banks in mainland China have started to tighten their mortgage criteria. Furthermore, growth rates of transaction values in these booming markets all slowed down sharply in December (sixth figure). However, the declines may have been contaminated by seasonality close to the year-end, and transactions had accelerated earlier as buyers rushed to take advantage of the stimulus measures before their expiration. Therefore, the full-fledged effects of the cooling measures are still to be seen in the coming quarters. The authorities may also need to fine-tune their policies in response to new market developments to maintain a delicate balance between leaning against housing bubbles and ensuring a solid economic recovery.

Real House Prices

Sources: OECD; Global Property Guide; and national authorities.

Note: The indices started in June 2002.

Real Estate Loan Growth

(In percent, year-on-year)

Sources: CEIC; national authorities; and IMF staff estimates.

Price-to-Income Ratio Indices

Sources: OECD; and national authorities.

Note: The indices started in 2001.

Price-to-Rent Ratio Indices

Sources: OECD; and national authorities.

Note: The indices started in September 2007.

Real Estate Loans as a Portion of Net New Bank Lending

(In percent)

Sources: CEIC; national authorities; and IMF staff estimates.

Note: Real estate loans include construction loans and mortgages. Trough was in 2008 for Korea and Singapore and in 2009Q1 for China and Hong Kong SAR. Latest was in 2009Q3 for Korea, and 2009Q4 for other economies.

1 Net new bank lending in Hong Kong SAR was negative in 2009, real estate loans in the year are presented as a share of quarterly average of 2008 net new bank lending.

2 Net new bank lending in Singapore was negative in the first quarter of 2009; data for this quarter represent real estate loans as a share of quarterly average of 2009 net new bank lending.

Transaction Value Growth in Response to Measures to Cool the Real Estate Market

(In percent, year-on-year)

Sources: CEIC; national authorities; and IMF staff estimates.

Note: Korean data represent units of transactions. Measures shown in the figure are the first major cooling policies announced by these Asian economies.

Note: This box was prepared by Deniz Igan and Hui Jin. Heejin Kim provided data support.1 India does not appear to exhibit the same dynamics; housing market conditions remain soft in most regions.2 It should be noted that these economies are only modestly levered with an average 45 percent mortgage-to-GDP ratio, compared to the 77 percent average of the advanced economies in the first figure. In addition, bank exposures to the property sector generally remain within regulatory limits. However, the increasing exposure to real estate is a worrisome trend.3 This applies more to China and Korea given the heterogeneity of monetary policy mandates in different Asian economies.4 Revenue from land sales in 2009 was estimated to be about one-third of total revenue in major cities in China.

Careful management of sovereign risks is essential for financial stability in the period ahead.

Sovereign risks have been transformed in a number of important ways. As the public sector stepped in to support financial institutions, distinctions between sovereign and private liabilities have been blurred and public exposure to private risks has increased. Channels of transmission among weaker mature sovereign credits have been revealed. Regional and global financial stability could be threatened if sovereign shocks are transmitted to banking systems and across borders. Thus, deteriorating fiscal fundamentals need to be credibly addressed.

In most cases, the success of ambitious fiscal adjustment that is required to reduce government debt to sustainable levels will depend on securing broad political support. Plans for medium-term fiscal consolidation should be developed and made public, including contingency measures if the deterioration in public finances is greater than predicted. Where necessary, these should be combined with a strengthening of fiscal institutions and improvement in public debt management frameworks. Other structural reforms to improve external competitiveness and growth prospects may also be necessary. Major economies, in particular, should be vigilant in maintaining medium-term fiscal discipline to avoid the risks of ratings downgrades and higher interest rates, which could spill over to other countries as well as increase funding costs for domestic banks and corporates.

Even as these reforms are implemented, risks will remain high in the short term and countries will remain susceptible to macroeconomic shocks and shifts in market sentiment. Immediate steps should therefore be taken to reduce the potential for the telescoping of longer-term sovereign credit risks into short-term financing concerns. This can be avoided through improved debt management practices, such as lengthening the maturity of public debt, to reduce near-term pressures. This will provide additional time for medium-term structural reforms to take effect.

In addition, authorities should endeavor to mitigate the transmission of sovereign risk through financial markets, for example by reducing the distortions from ratings triggers in statutory guidelines, and by strengthening collateral policies for OTC derivative exposures. However, steps to reduce transmission channels should avoid interfering with efficient market functioning and good risk management practices. Thus, recent proposals to ban “naked” CDS exposures could be counterproductive, as this presupposes that regulators can arrive at a working definition of legitimate and illegitimate uses of these products. A general definition of “naked shorts” remains elusive for both market participants and regulators, reflecting the wide spectrum of activity that can constitute naked positions, ranging from hedging activity to outright speculation. Even though sovereign CDS may at times influence underlying bond markets, particularly during periods of distress, banning “naked shorts” would be ineffective and difficult to enforce. A prohibition against the use of certain derivatives may simply transfer selling pressure to related cash market instruments, such as government bonds, equities, or foreign exchange, and make hedging of exposures more costly and complex.

The focus of policymakers should be on improving already-existing CDS data sources to monitor markets, and on continuing to strengthen the market’s operational infrastructure. Policymakers should push to move bilateral OTC derivative contracts on to central counterparties (CCPs), and to advocate more consistent and uniform collateral practices on bilateral contracts. This would reduce the need to use sovereign CDSs as synthetic hedges against private sector counterparty risk, and possibly reduce volatility in the sovereign CDS market. These reforms would also promote global financial stability, while allowing market mechanisms to determine the ultimate usage of sovereign CDS. Chapter 3 discusses the role that CCPs can play in making OTC markets safer.

Policymakers need to ensure that this next stage of the deleveraging process unfolds smoothly and results in a safer, competitive, and vital financial system.

Bank deleveraging has been driven mainly from the asset side thus far, as mounting losses have prompted banks to reduce exposures to riskier assets. Going forward, however, the deleveraging process will be dominated by pressures on the funding or liability side of bank balance sheets. New regulatory rules will act to reduce leverage and raise capital and liquidity buffers. While the key banking systems most affected by the crisis likely now have sufficient capital, in aggregate, to meet expected future losses, there is significant variation across individual institutions within these systems. Some have a weak tail of thinly capitalized institutions that are highly dependent on cheap central bank funding. These impaired institutions compete for funding with more profitable and better-capitalized institutions, thereby squeezing margins and limiting the ability of healthier banks to finance their loan portfolios. If left unaddressed, this could ultimately act as a brake on the recovery of credit.

Box 1.4.Could Conditions in Emerging Markets Be Building a Bubble?

There is a growing body of literature that suggests banking crises often result from the build-up of financial imbalances.1 These imbalances develop over a number of years through a simultaneous boom in asset prices and credit. Rapid credit growth alone or the development of an asset price bubble by itself may not create vulnerabilities. It is the coexistence of credit and asset price booms that increases the likelihood of future financial stress. This is because at some point, if the boom turns to bust, the economy will be left saddled with large debts backed by assets with falling value. As the recent crisis has shown, a vicious circle of falling asset prices and reductions in leverage can form, potentially leading to widespread instability in the financial system. Such a financial crisis is likely to be associated with a deep and protracted slowdown in economic activity, particularly if there is distress in the banking sector.2

One common way of assessing the development of imbalances is to create a set of indicators that measure the deviation of key variables from their trend. This method is used to capture the cumulative process whereby imbalances build up steadily over time. The first figure shows that in the years before past episodes of financial stress, a strong increase in credit relative to its trend was associated with a rise in asset prices and growth in portfolio capital inflows. Interestingly, credit appears to stay at a high level even after asset prices have started to fall sharply. This may be because only a small proportion of loans will mature or default at any point in time, so the level of credit will decline relatively slowly. It could also reflect companies drawing down previously agreed precautionary credit lines, as happened during the 2007–09 global financial crisis.

More recently, there is some evidence to suggest that asset price pressures may be building in some emerging markets. The second figure shows the deviation in trend for credit, portfolio capital inflows, and asset prices in Brazil, Russia, India, and China. This shows that, following the latest boom and bust where all three series rose and fell sharply, there has been a resumption of a build-up in capital flows, particularly in China and India. In addition, credit did not fall back as sharply as the other two indicators in 2008 and remains high relative to trend, albeit lower than the peak in 2008. If credit remains at this level and if portfolio flows continue to build, this could create conditions in which asset prices could boom and, over time, potentially lead to the development of financial system vulnerabilities.

Asset Prices, Credit, and Capital Flows in Past Crises

(In standard deviations from mean)

Sources: Haver Analytics; national authorities; and IMF staff estimates.

Note: This figure shows the average deviation in real asset prices, real private credit, and real cumulative capital flows for developed countries in the early 1990s and around the 2007–09 global credit crisis, as well as for southeast Asian countries (excluding Japan) around the 1997 crisis. The data shown are a four-quarter moving average of a z-score calculated over the period 1991–2005.

Asset Prices, Capital Flows, and Credit in Brazil, Russia, India, and China

(In standard deviations from mean)

Sources: Haver Analytics; national authorities; and IMF staff estimates.

Note: This figure shows the deviation in the series from their estimated trend. The series are the average for Brazil, Russia, India, and China, where figures are available. All data are in real terms. The figures are shown as a z-score calculated over the period 1991–2005.

Note: This box was prepared by William Kerry.1 See Borio and Lowe (2002); Borio and Drehmann (2009); Alessi and Detken (2009); and Gerdesmeier, Reimers, and Roffia (2009).2Chapter 4 of the October 2008 WEO discusses this in more detail.

Going forward, funding pressures are likely to intensify for banks, as the wall of shorter-duration debt issued during the crisis matures, as banks compete with sovereigns to issue longer-dated debt, as central banks reduce their extensive liquidity support—thereby returning lower-quality collateral to banks—and as banks compete more aggressively for deposits to meet new liquidity requirements. Swift resolution of nonviable institutions and restructuring of those with a commercial future is thus a vital component of the deleveraging process. This will help to ensure that once public support measures are removed, a healthy core of viable financial institutions remains, able to withstand normal competitive forces and resume lending. Measures to restructure and resolve weak institutions also facilitate the withdrawal of extraordinary support measures and the normalization of central bank liquidity facilities. The sooner weakened institutions recognize losses and are either resolved, restructured, or recapitalized by existing or new investors, the sooner the financial system can return to health.49 Continuing to strengthen the capital base will also help prepare the financial system for timely implementation of the more stringent requirements of the new enhanced Basel II regime and other changes to the capital adequacy framework. At the same time, greater clarity is needed in defining the new financial system framework, including financial sector taxation, to give banks more certainty over their future business models. These measures will need to be taken in conjunction with addressing the issue of “too-important-to-fail” institutions, to solve moral hazard problems, and to restore healthy and fair competition.

Policies may still be needed to ensure adequate flows of credit to the private sector.

Credit availability is likely to remain limited as banks continue to reduce leverage. Notwithstanding the weak recovery in private credit demand as households restore balance sheets, ballooning sovereign financing needs may bump up against supply constraints and exacerbate funding pressures, further constraining credit supply. Accordingly, measures to strengthen the recovery of safer securitization markets may be necessary (see the October 2009 GFSR). Furthermore, targeted support to ensure adequate lending to the SME sector may be warranted in some economies. There is the possibility that central bank support measures, including purchases of securities, may still be needed to offset the retrenchment in credit capacity by the bank and nonbank sectors in selected cases.

The necessity of further deleveraging in a number of countries can make the task of exiting from extraordinary support and liquidity measures a delicate one. In general, policymakers should seek to implement coherent and credible exit strategies once normalcy has returned to financial markets. Unnecessary delay risks private sector institutions becoming dependent on offcial support, distortions in market prices, and an undermining of central bank credibility regarding price stability. However, premature withdrawal risks jeopardizing economic recovery by exacerbating the deleveraging process. Policymakers need to formulate exit strategies suitable to their economic circumstances—coordinated where necessary across fiscal, monetary, and regulatory authorities—and credibly communicate them to market participants. The withdrawal of financial sector support can be facilitated by using built-in market incentives (e.g., a rising premium charged for guarantees) and the judicious use of termination dates.

Emerging market policymakers will need to deploy a wide range of policy tools to address the challenges arising from capital inflows.

The strong rebound in emerging market portfolio inflows, while welcome, is leading to concerns over inflationary pressures or asset price bubbles in receiving countries. Although there is only limited evidence at this time of stretched valuations across countries—with the exception of some local property markets—current conditions of high external and domestic liquidity and rising credit growth have the potential to stoke inflation and give rise to bubbles over a multi-year horizon. In addition to macro-policy adjustment (including measures supporting exchange rate appreciation), possible policy tools include liquidity management operations to mop up domestic liquidity; prudential tools to restrict banks’ ability to fuel a credit boom and restrict a build-up of excessive leverage; and measures to target specific asset prices and markets. Chapter 4 discusses the use of capital controls as part of the macroprudential policy mix.

Addressing too-important-to-fail banks is critical for restoring market discipline and insulating sovereign balance sheets.

Excess capacity in the financial system and significant concentration of power in “too-important-to-fail” institutions remain to be addressed as the financial system undergoes further deleveraging. Market discipline and fair competition will be supported by addressing the significant advantages in funding markets enjoyed by too-important-to-fail institutions.50 This is critical to avoid even greater concentration as the financial system shrinks.51 Importantly, to protect sovereign balance sheets and to reduce the risks of recurrence, such institutions must have adequate capital and liquidity buffers plus robust risk management systems and capacities. Policymakers must also reduce the potential and actual moral hazard associated with too-important-to-fail institutions.

Box 1.5.Proposals to Address the Problem of Too-Important-to-Fail Financial Institutions

“Too-important-to-fail” (TITF) firms are those believed to be so large, interconnected, or critical to the workings of the wider financial system or economy that their disorderly failure would impose significant costs on third parties. This status engenders expectations that, if failure were to loom, the authorities would be forced to prevent the collapse of these institutions, thereby shielding creditors from loss, reducing borrowing costs, and encouraging additional leveraged risk-taking by TITF firms. The policy response to the financial crisis—entailing selective bailouts favoring TITF firms and assisted mergers—has exacerbated this already-serious moral hazard problem in the United States and Europe. Proposals made by the Basel Committee on increased capital for market risk and liquidity requirements and improvements to clearing infrastructure (see Chapter 3) would reduce systemic risk across the financial sector. In addition, a range of policy responses has been suggested to address the specific issue of TITF institutions and is under consideration by the Financial Stability Board:

  • Tougher supervisory standards for TITF firms. An element in the U.S. administration’s proposal for systemic firms is for regulators to require tougher minimum capital, liquidity, and risk management requirements, effectively under Pillar 2 of the Basel framework. This has the advantage of flexibility but relies on regulators identifying sources of systemic risk accurately while maintaining robust independence from TITF firms.

  • Resolution mechanisms (TITF insolvency regimes; “living wills”). The crisis highlighted the absence of legal powers in many jurisdictions to intervene in, or wind up, troubled TITF institutions in an orderly way outside standard bankruptcy procedures. Such mechanisms are vital to give credibility to the threat of failure. Requiring the preparation of “living wills” by TITF firms would force their boards to understand the complexities of their legal structures while providing some assistance to regulators in insolvency. Unless a robust cross-border resolution regime for TITF firms can be implemented, jurisdictions may seek the safer option of resolving subsidiaries they host rather than allow cross-border branching of TITF entities.

  • Additional capital requirements linked to systemic risks. In addition to the higher levels of better quality capital for internationally active banks proposed by the Basel Committee, additional requirements could be calibrated to penalize firms’ attributes that make them TITF and thus internalize the costs these institutions impose on the system. Chapter 2 illustrates how systemic-risk-based capital surcharges can be made operational. Such requirements should be set to motivate TITF firms to divest activities and shrink assets to raise their return on equity, while favoring new entrants and greater competition.

  • Taxes or levies to pay for costs of resolving TITF entities. While initially intended to “claw back” the costs of crisis bailout, such taxes could be used to encourage TITF firms to reduce systemic risks. To fully address the problem, such taxes or levies would need to be calibrated to exceed the cost of capital benefit that TITF firms derive from their status. Policymakers should ensure that in the event of a failing TITF firm, there is appropriate burden-sharing so that shareholders lose their investment, unsecured creditors incur losses through haircuts, and management is replaced.

  • Limits on market share or asset size. To confine TITF firms to a manageable size for crisis management and competition purposes, additional capital requirements and leverage ratios could be combined with caps on relative market share (as with the United States’ 10 percent limit on insured deposits), balance sheet size, or counter-party exposures. Such basic rules of thumb prevent TITF firms arbitraging risk-based measures and recognize the need to cap sovereign risk posed by the failure of any one firm.

  • Restrictions on activities. Some recent proposals have included the exclusion of own-account proprietary trading from all institutions with access to deposit insurance and lender-of-last-resort facilities (to address existing conflicts of interest, moral hazard, and skewed competition—the “Volcker rule”). To avoid unintended consequences, “proprietary trading” would need to be carefully defined to exclude market-making, hedging, and client-driven trading activities.

Note: This box was prepared by Paul Mills.

There have been a number of policy instruments proposed to address the problem (see Box 1.5) but little consensus on which are most advantageous. Available options range from higher capital requirements linked to systemic importance, to imposing limits on the size and scope of institutions, with regulatory authorities tailoring their approach to reflect specific country circumstances. Whatever option is chosen, the simple metric of effectiveness will be whether too-important-to-fail institutions reduce their contribution to systemic risk and do so in a matter that is internationally consistent. The window of opportunity for real reform of too-important-to-fail institutions is rapidly closing, so policymakers should take bold steps to ensure this topic stays on the reform agenda, and meaningful progress is made.

Annex 1.1. Global Financial Stability Map: Construction and Methodology52

The further improvements in global financial stability and underlying conditions are illustrated in our global financial stability map (Figure 1.1). The changes in indicators are highlighted in Figure 1.36 and the specific indicators used are noted in Table 1.9. The rest of this annex outlines key features of the global financial stability map (GFSM) and reviews its experience through the crisis.

Figure 1.36.All Risks to Global Financial Stability and Its Underlying Conditions Have Improved

(In notch changes since the October 2009 GFSR)

Note: Overall notch changes are the simple average of notch changes in individual indicators. The number next to each legend indicates the number of individual indicators within each subcategory of risks and conditions. For lending standards, a positive value represents a slower pace of tightening or faster pace or easing.

Table 1.9.Global Financial Stability Map Indicators
Monetary and Financial Conditions
Monetary conditionsG-7 real short rates
G-3 excess liquidity
Growth in official reserves
Financial conditionsFinancial conditions index
Lending conditionsG-3 lending conditions
Risk Appetite
Investor surveyMerill Lynch investor risk appetite survey
Institutional allocationsState Street investor confidence index
Emerging market assetsEmerging market fund flows
Relative asset returnsGlobal risk appetite index1
Macroeconomic Risks
Economic activityWorld Economic Outlook global growth risks
G-3 confidence indices
OECD leading indicators
Implied global trade growth
Inflation/deflationGlobal breakeven inflation rates
Sovereign creditMature market sovereign CDS spreads
Advanced country general government balance2
Emerging Market Risks
SovereignsFundamental EMBIG spread
Sovereign credit quality
InflationMedian inflation volatility
Corporate sectorCorporate spreads
Credit Risks
Corporate sectorGlobal corporate bond index spread
Credit quality composition of corporate bond index
Speculative-grade corporate default rate forecast
Banking sectorBanking stability index
Household sectorConsumer and mortgage loan delinquencies
Household balance sheet stress
Market and Liquidity Risks
Market positioningHedge fund estimated leverage
Net noncommercial positions in futures markets
Common component of asset returns
Equity valuationsWorld implied equity risk premia
VolatilitiesComposite volatility measure
Funding and liquidityFunding and market liquidity index
Source: IMF staff estimates. For a detailed description of each indicator, see Annex 1.1 of the October 2009 GFSR.

The GFSM was designed to assess the risks and conditions that impact financial stability.53 The GFSM is intended to provide a summary, graphical representation of the IMF’s assessment of financial stability, capturing a diverse range of potential sources of instability, contagion among different segments of financial markets, and nonlinearities in the underlying factors. The philosophy underpinning the GFSM is that financial stability cannot be distilled into a single indicator, and is better understood by separating the underlying risks and conditions that could give rise to a systemic threat. The aim is to extract diagnostically useful information from economic and financial metrics, supplemented by judgment based on market intelligence and the IMF’s assessment of risks.

The GFSM tracks four broad risks and two underlying conditions considered relevant for financial stability and the IMF’s remit in supporting financial stability.

Macroeconomic risks affect financial stability through various channels—three elements are captured here. The global growth outlook underpins income—the borrower’s ability to pay and overall market perceptions of credit risk. Inflation/deflation risk can destabilize fixed-income markets and impact real debt burdens and is thus a source of financial stability risk. Sovereign risk results from unsustainable fiscal paths, and rising debt burdens can be a significant source of financial instability, potentially culminating in a sovereign default.

Emerging market risks capture underlying fundamentals in emerging markets—and are therefore closely related to macroeconomic risks described above, but conceptually separate as they focus only on emerging markets—and vulnerabilities to external shocks. Indicators include models that translate economic, financial, and political variables into a sovereign external credit risk spread. Underlying indicators of credit and inflation performance capture risks related to financial policies and are leading indicators of future vulnerabilities. Market perceptions of corporate credit risks are also included.

Credit risks measure credit stress in household and corporate balance sheets. Indicators attempt to capture risks in both banking and nonbanking systems. Risks in core financial institutions and contagion are assessed using models based on credit derivatives. Pressures in corporate debt markets are captured using delinquency rates and expected defaults. Market risks assess the potential for heightened pricing risks that could result in spillovers and/or mark-to-market losses, while liquidity risks measure stress in funding markets as well as liquidity conditions in secondary markets. These indicators highlight potential for vulnerabilities that arise from excessive leverage—risks that markets might correct abruptly and risks that a liquidity or funding crisis could spill over and impact markets more broadly, including credit risks.

Monetary and financial conditions gauge the stance of monetary policy and the cost and availability of funding. Measures include short-term real interest rates, as well as estimates of excess liquidity. The willingness and capacity of banks to lend is a key input as is the market-based indicator of financial conditions.

Risk appetite gauges the willingness of investors to increase (or shed) risk. Such “animal spirits” can greatly influence spread developments as well as market and liquidity risks. Gauges of risk appetite include survey- and market-based measures of risk appetite, as well as normalized flows into emerging markets.

The choice of specific indicators to assess these risks and conditions is guided by their relevance and various practical considerations. The indicators within each ray of the GFSM should be sufficient to capture potential sources of risk, but limited in number to avoid overlaps and canceling out of pertinent indicators. The indicators should be sufficiently forward-looking to have predictive powers for a 6–24 month window. A balance of economic, market-based, and survey-based indicators, as well price and quantity measures is sought to achieve these aims (Table 1.9). The indicators should be of relatively high frequency and have sufficient history to provide enough information through (in)stability cycles. The reliability of the indicators is periodically assessed and adjustments are made so that the GFSM adequately captures underlying risks and conditions at any given time.

Current conditions and risks are summarized in a scale of 0 to 10, with higher values signifying higher risks and easier conditions relative to their respective historical norms. Assessments of the contemporaneous values of the indicators are made relative to their own history in terms of percentile rankings.54 To construct the GFSM, we first determine the percentile rank of the current level of each subindicator relative to its history.55 The individual indicator rankings are aggregated into each of the six rays of the GFSM using equal weights. Judgment and technical adjustment are often used to attach greater importance to a particular set of indicators based on risks considered to be most relevant at a given time. In particular, technical adjustment is used when events that surpass historical experience raise (lower) some associated risk or condition indicators to the highest (lowest) level. The final choice of positioning on the GFSM represents the best judgment of IMF staff.

The GFSM tracked broad developments well during the global financial crisis that culminated in 2009 (Figure 1.37).56

Figure 1.37.Evolution of the Global Financial Stability Map, 2007–09

Note: Away from the center signifies higher risks, easier monetary and financial conditions, and higher risk appetite.

Monetary and financial conditions: The GFSM signaled very easy conditions from 2003 to 2006, suggesting the potential for a build-up of large imbalances ahead of the crisis. The pairing of relatively easy monetary and financial conditions and high levels of risk appetite reinforced this signal.

Risk appetite: This set of indicators captured the rise in levels of risk appetite in the run-up to the crisis, as well as the sharp contraction in risk appetite from very high levels ahead of the crisis.

Macroeconomic risks: Indicators signaled exceedingly low perceptions of risks at the onset of the crisis, and captured deteriorating conditions throughout the crisis as well.

Emerging market risks: These indicators suggested very low perceptions of risks in 2005–07, and a realization of risks only in late 2008 following the collapse of Lehman Brothers. This reflected the fact that the crisis originated in mature markets and the relatively resilient position of emerging markets was only threatened once the financial crisis spread to cross-border funding channels and the real economy.

Credit risks: Perception of risks increased from very low levels prior to the global financial crisis, signaling rising risks of a credit bubble and strains at the core of the financial system.

Market and liquidity risks: This set of indicators tracked the rise in risks to financial stability throughout the crisis period, reaching its highest level after the collapse of Lehman Brothers. Some of the subindicators on market positioning also pointed to increased high risk-taking ahead of the crisis in mid-2007.

Annex 1.2. Assessing Proposals to Ban “Naked Shorts” in Sovereign Credit Default Swaps57

Strains in Greek government bond markets have been partly blamed on speculative positioning through buying sovereign CDS protection. This has highlighted the need for further investigation and led to a discussion of the merits of a ban on “naked shorts.” Even though sovereign CDS may at times influence underlying bond markets, particularly during periods of distress, banning “naked shorts” would be ineffective and difficult to enforce. In addition, “naked shorts” may be hard to define and such bans may hamper legitimate financial activity. Instead, transparency and collateral practices in CDS markets could be substantially improved to reduce risks.

After a decade of static market share relative to the broader CDS market, sovereign CDS underwent a rapid expansion in 2009 and into 2010. Gross sovereign CDS notional leapt 31 percent (versus a 4 percent increase in total CDS gross outstanding) (Table 1.10). The more relevant sovereign net notional exposure increased 23 percent compared with a 10 percent contraction in total net notional positions.58 The number of sovereign CDS contracts also grew more than twice as fast as the entire market.

Table 1.10.Ten Largest Sovereign Credit Default Swap (CDS) Referenced Countries(In billions of U.S. dollars, as of February 5, 2010)
Gross NotionalNet Notional
Outstanding (billions of dollars)Year-on-year growth (percent)Outstanding (billions of dollars)Year-on-year growth (percent)
ltaly223.83524.840
spain102.04614.523
Germarry61.54712.927
Brazil141.52811.616
portugal60.11059.472
Austria41.5809.487
France44.8768.645
Mexico104.0446.437
lreland34.2776.036
Total sovereign2,174.331196.123
Total CDS15,026.741,128.4-10
Sources: Depository Trust & Clearing Corporation; and IMF staff estimates.

Sovereign CDS has unlikely exerted a significant influence on government bond markets, for Greece or other sovereigns.

The size of the sovereign CDS market and amount of net protection sold are negligible compared to government debt outstanding. For the market as a whole, gross sovereign default protection is $2 trillion in notional value, just 6 percent of the $36 trillion global government bond market. By contrast, corporate CDS are roughly equivalent in size to the global corporate bond market.

Net exposure represents only 0.5 percent of government debt, at $196 billion notional amount. Among the 20 largest sovereign CDS markets, the share of net notional CDS outstanding to government debt averages 2 percent and does not exceed 7 percent in any country (Figure 1.38).

Figure 1.38.Net Notional Credit Default Swaps Outstanding as a Share of Total Government Debt

(In percent)

Sources: Bank for International Settlements; Depository Trust & Clearing Corporation; and IMF staff estimates.

Could the tail (CDS spreads) wag the dog (bond yield spreads)?

In normal market conditions, CDS tend to move in tandem with bond yield spreads, as arbitrage conditions link the bond and derivatives markets.59 But in periods of funding stress and poor bond liquidity, CDS can decouple from bond yield spreads and might even lead the bond market. A simple test is to ask whether changes in sovereign CDS today influence—i.e., are correlated positively with—bond yield spreads tomorrow (Figure 1.39). In the case of Greece, the correlation of both instruments with changes one or more days ahead was generally nil or slightly negative, except during the peak points of the crisis as bond market liquidity evaporated.60,61

Figure 1.39.Correlation of Daily Changes in Five-Year Greek Credit Default Swap (CDS) and Bond Yield Spreads

(In rolling 60-day periods)

Sources: Bloomberg L.P.; and IMF staff estimates.

Sovereign CDS markets can be prone to distortions because of relatively shallow liquidity. For instance, banks often attempt to create synthetic hedges for counterparty risk to sovereigns due to low (or nonexistent) collateral requirements. When looking for assets that are highly correlated with the sovereign’s credit profile, banks resort to short-term CDS (so-called “jump-to-default” hedging). This hedging activity from uncollateralized swap agreements can distort the sovereign CDS market as well as other asset classes. For instance, heavy demand for jump-to-default hedges can quickly push up the price of short-dated CDS protection and cause sovereign CDS curves to invert, as happened in Greece and Portugal. These pressures can easily spill over into the domestic bond market and contribute to higher bond yields, especially for new debt issues.

The influence of sovereign CDS on government bond markets, minor in normal conditions and possibly greater under periods of stress, cannot be separated from the inefficacy of an outright ban on “naked shorts.” As discussed later in the policy section, more productive reforms would be using already-existing CDS data sources to monitor markets and continuing to improve the market’s operational infrastructure.

“Naked shorts” in sovereign CDS should not be banned.

Some argue that the very premise of CDS as a financial insurance product is inherently flawed and should be more tightly regulated. Buyers of CDS protection do not need an “insurable interest” to acquire protection (promoting adverse incentives) and nonbank sellers are not regulated or required to hold loss reserves (false sense of protection). In other words, CDS is an insurance-like product without insurance-like supervision.

This debate fails to consider an asset in the broader portfolio context and the nature of economic exposure. The correlation of risk factors defines economic exposure, not just ownership of a specific asset. As such, a portfolio manager may have an “insurable interest” in shorting an asset because of the portfolio’s risk exposures, even if that asset is not included in the portfolio. Sovereign CDS is not only “credit insurance,” but another tradable instrument in the risk management tool kit.

Speculation or hedging?

Recent activity in CDS relates more to concerns about counterparty or broad portfolio hedging than to sovereign default credit protection for holders of the underlying government bonds.

Counterparty hedging: As mentioned above, large banks generally do not require highly rated sovereign entities to post collateral for swap arrangements, introducing a significant unhedged counterparty exposure.62

Hedging country corporate exposure: Bank risk managers often aggregate individual corporate credit risks into acceptable country exposures that necessitate mitigation if breached. Sovereign CDS can offset those exposures by providing country-level risk diversification.

Proxy hedging: Investors also use sovereign CDS as a hedge against existing equity or corporate bond positions. This proxy hedge introduces basis risk (the sovereign’s profile could improve as the corporate’s worsens), but may be preferable due to greater liquidity or cheaper cost. Market sources cited such proxy hedgers as significant buyers of Greek sovereign CDS because individual Greek bank CDS were less liquid.

Hedging portfolio liquidity and market risk: A risk manager may desire to reduce daily portfolio value-at-risk (VaR) by looking for an uncorrelated macro hedge to the underlying debt or equity positions. Buying short-dated sovereign CDS protection could accomplish that objective much in the same way as a long gold position reflects a safe-haven bet.

Macro hedging and speculation: Macro funds are reportedly turning to sovereign CDS to express directional views on economic fundamentals and offset overall portfolio risk, especially via the new sovereign CDS indices. Yet since the launch of the Traxx SovX last year, the overall index has traded between 2–8 bps tighter than the intrinsic spread of the 15 underlying sovereigns CDS. This negative basis points to demand for individual-name CDS remaining stronger than demand for tradable sovereign CDS indices, suggesting that macro hedging is not a major mover of sovereign CDS markets.

Dealers represent about 90 percent of the sovereign CDS market and are net sellers of credit protection, according to the Depository Trust & Clearing Corporation (DTCC): By implication, this means that investors (real money and hedge funds) are net buyers of protection. Trading motivations cannot be entirely discerned from the DTCC classifications, but most dealer flows likely relate to hedging as part of market making activities. From a risk management perspective and business rationale, dealers are less inclined to take large directional bets in CDS. Nondealers generated just 15 percent of January’s trading in sovereign CDS and even less in November-December (Figure 1.40).

Figure 1.40.Sovereign Credit Default Swap Volumes

(In billions of U.S. dollars)

Source: Depository Trust & Clearing Corporation.

A “naked shorts” ban would not work.

The current discussion of a ban for “naked shorts” in sovereign CDS presupposes that regulators can arrive at a working definition of legitimate and illegitimate uses of these products. A general definition of “naked shorts” remains quite elusive for both market participants and regulators, reflecting the wide spectrum of activity that can constitute covered versus naked positions.

An outright ban on “naked shorts” in sovereign CDS would also be ineffective and inconsistent with wider ramifications for financial markets.

Not effective: Given that most sovereign CDS flows likely reflect hedging activity, an outright ban would merely prompt substitution to another asset correlated with sovereign risk. The most direct method would be to short the underlying bond, simply transferring more pressure to the cash market. Alternatively, to the extent that proxies are available (such as local equities, corporate CDS, or currency), pressure is transmitted to related markets, such as Greek bank equities or CDS. The short-selling bans on bank equities seemed to provide little relief to bank share prices.

Easily circumvented: “Creative” financial engineering could replicate default protection in another form. Alternatively, CDS business can be rerouted offshore or to dealers in another regulatory jurisdiction.

Inconsistent regulatory practice: Treating sovereign CDS differently than corporate CDS or any defensive derivative strategy introduces regulatory inconsistencies. After all, why consider sovereign CDS differently than corporate CDS or shorting bonds overall?

Section F explores appropriate measures for greater sovereign CDS transparency and mechanisms to reduce banks’ reliance on them for hedging purposes.

Annex 1.3. Assessment of the Spanish Banking System63

This annex attempts to estimate the impact of the financial crisis on the Spanish banking sector, looking separately at commercial banks and savings banks (cajas). We find that the overall Spanish banking system under our baseline case is likely to withstand consequences of the crisis, despite severe economic deterioration. Under our adverse-case scenario, three years of earnings are projected to cover future losses for the commercial banking sector, leaving the capital base intact, but the savings banking sector is projected to have a net drain on capital. Furthermore, the country’s banking system is highly differentiated in terms of holdings of bad loans and distressed real assets. After accounting for this cross-bank differentiation, small gross drain on capital is expected in both commercial and savings banks under the baseline. Under our adverse-case scenario, gross drain on capital is estimated at €5 billion for commercial banks and €17 billion for savings banks. These estimates compare against Tier 1 capital of €99 billion and €78 billion for commercial and savings banks, respectively.

The pace of house price deterioration and the extent of broad economic downturn in Spain have been more severe than in the euro area, on average. These developments have led many commentators to question whether the Spanish banking sector’s provisions are sufficient to withstand potential losses.

The analysis is divided in two parts: in the first part we estimate the net impact of current and expected losses of Spanish commercial and savings banks on their earnings stream over the 2010–12 period under our baseline and adverse-case scenarios; in the second part, we examine cross-bank differentiation in terms of real asset repossessions and assess what share of the system may need additional capital.64

The first part of the analysis benefited from collaboration with the Bank of Spain. Spain has pioneered the use of dynamic provisions since 2000 to mitigate credit procyclicality. This helped Spanish credit institutions to accumulate a significant buffer of loan loss provisions by the beginning of the crisis (IMF, 2008c, Box 1). Box 1.6 explains how losses from nonperforming loans (NPLs) are forecasted.

NPLs at commercial and savings banks are projected to peak at 6.3 percent and 6 percent, respectively, in 2010:Q3, and then come down to 5.1 percent and 5 percent, respectively, by the end of 2011 (Figure 1.41). The outcomes of forecasts using equations (1) and (2) are dependent on lag specifications. For example, for commercial banks, the selection of different lags resulted in the peak values between 5.5 percent and 7.4 percent, and the presented specification roughly corresponds to our median forecast. The forecast peaks in NPLs in 2010 are lower than those in the previous crisis episode in 1993–94, because of much lower interest rates during this crisis (5.3 percent in 2008 vs. 14.3 percent in 1992) and lower unemployment rates (18.8 percent in 2009 vs. 24.6 percent in 1994). The econometric approach does not capture an additional risk factor related to private leverage, which has dramatically increased over the 10 years of credit boom. Another weakness of the econometric approach comes from the use of historical data, which predicts a higher peak for NPLs at commercial banks, based on the historical experience and slowing NPLs at savings banks in 2009.65

Figure 1.41.Spain: Nonperforming Loans

(In percent of total loans)

Source: IMF staff estimates.

The assumptions about the loss given default ratio (LGD) are derived from previous studies and analyst estimates. The baseline scenario is based on 25 percent LGDs for both commercial banks and savings banks, which correspond to internal estimates of downturn LGDs according to the Bank of Spain’s assessment and are in line with other euro area average LGDs.66 Losses on securities’ holdings are estimated at €4 billion for commercial banks and €1 billion for savings banks.67

We also consider the effect of repossessed real assets (Figure 1.42).68 Over the last two years, given the ailing state of the real estate and the construction sectors, Spanish banks have increased the use of debt-for-property swaps to manage their credit portfolios efficiently, trying to maximize asset value recovery. This practice helps banks in managing of their credit risk portfolios and minimizes losses, provided that property prices stabilize in the medium term and banks can sell those assets at their book value. However, if house price deterioration continues, banks under pressure may need to sell properties within a short period of time, resulting in substantial losses.

Figure 1.42.Spain: Real Asset Repossessions

(In billions of euros)

Source: IMF staff estimates.

Box 1.6.Estimating Potential Losses from Nonperforming Loans for Spain

In this exercise, we assume that potential losses are equal to flows of provisions in 2010–12, which are computed as flows of provisions in 2010 plus expected losses after 2010. In turn, expected losses after 2010 are estimated as additional provisions after 2010 that are necessary to cover expected losses in excess of accumulated loan loss reserves as of 2010:

Expected Losses after 2010 = NPL in 2010 x LGDStock of Provisions in 2010,

where NPL is the stock of nonperforming loans, and LGD is the cumulative loss given default ratio over the next two years. Drain on capital is calculated as potential losses minus future earnings in 2010–12.

We forecast nonperforming loans based on business cycle variables, loan costs, and house prices. GDP and the unemployment rate are used as business cycle indicators, the 12-month euro LIBOR is used for loan costs because it is a common benchmark for mortgages and other loans, and house prices are an indicator for the mortgage and the construction sectors. The dependent variable is obtained using the logit transformation: npl ≡ LN(NPL/(1 – NPL)).

Since the dependent variable has a unit root, the regression is estimated in first differences.1 Real GDP growth is ultimately removed from the regression, because of its collinearity with the unemployment rate and house prices. As a result, the following specifications (1) and (2) are obtained for commercial banks and savings banks, respectively.

where D. is the first difference operator, L. is the lag operator, npl_c and npl_s are NPLs for commercial and savings banks using the logit transformation above, U is the unemployment rate, I is the LIBOR rate, H is yearly changes in house prices. The constants are suppressed due to their insignificance. The regressions are estimated over 1987:Q4–2009:Q4. Forecasts for 2010 are produced using WEO data for the unemployment rate, while the LIBOR and house prices work with lags based on historical values.

Note: This box was prepared by Sergei Antoshin.1 The difference form also implies inertia of NPLs in levels.

Estimates of banks’ acquired or repossessed real estate assets vary significantly. Our own estimates are €22 billion and €37 billion for commercial banks and savings banks, respectively, in 2009Q3.69 Our time series on repossessions are augmented by the Bank of Spain’s estimates of €23 billion and €36 billion for 2009Q4. Repossessions surged over the last two years, adding €11 billion of troubled real assets in 2009 to the balance sheet of commercial banks and €21 billion for savings banks. We project that the pace of increases in repossessions will slow in 2010 to €10 billion for commercial banks and to €20 billion for savings banks. LGDs for repossessed assets are subject to a high degree of uncertainty because the distribution of repossessed assets by type is unknown for the overall system and because it is not likely that banks will recognize losses by selling these assets within the next three years. Since repossessed assets include land and unfinished construction with very high expected loss rates, we assume LGDs of 40 percent and 45 percent, which correspond to lower bounds of market estimates. Spanish banks are required to set aside provisions for repossessed assets, to account for the possible loss in value of that asset depending on the number of years that is maintained on the balance sheet before it is finally realized. We use the Bank of Spain’s estimates for stock of provisions for repossessions: €6 billion for commercial banks and €7 billion for savings banks.

Based on the forecasted NPLs and repossessions, and the assumed LGDs, expected losses in excess of end-2009 stock of loan loss provisions are computed in Table 1.11. Under the baseline scenario, stock of provisions at commercial and savings banks exceed expected losses by €10 billion and €12, respectively (line (6) in Table 1.11). Repossessions add €7 billion and €15 billion in expected losses after accounting for provisions for commercial and savings banks, respectively (line (12) in Table 1.11).

Table 1.11.Spain: Baseline and Adverse-Case Scenarios(In billions of euros, unless otherwise shown)
Commercial BanksSavings BanksCommercial BanksSavings Banks
Baseline scenarioAdverse-case scenario
(1)Total loans798882798882
(2)Stock of NPL in 2010/2011150536262
(3)Loan loss reserves23262123
(4)LGD for NPLs (percent)25254545
(5)Expected losses from NPL(2)*(4)−13−13−28−28
(6)Loan loss reserves–loan losses(3)+(5)1012−7−5
(7)Losses from securities−4−1−4−1
Adding repossessions
(8)Repossessions in 2010/2011131483656
(9)Reserves for repossessions6767
(10)LGD for repossessions (percent)40455560
(11)Expected losses from repossessions(8)*(10)−13−22−20−34
(12)Repossession reserves–losses(9)+(11)−7−15−14−27
(13)Total reserves–total losses(6)+(7)+(12)−1−3−26−33
(14)Pre-provision earnings in 2010–1252394131
(15)Net drain on capital2−(13)−(14)−51−36−152
(16)Memo: Tier 1 capital (Q2 2009)99789978
Source: IMF staff estimates.Note: NPL = nonperforming loan; LGD = loss given default.

Pre-provision net earnings are expected to decline 10 percent each year during 2010–12, due to a sharp fall in interest income, funding pressures in the medium term, and slowing deposit growth. Despite these declines, banks’ earnings stream over the next 3 years will be sufficient to cover those expected losses. In sum, under our baseline scenario, loan loss reserves and earnings are sufficient to fully absorb expected losses for the overall commercial banking and the savings banking sectors.

Our adverse-case scenario corresponds to a double-dip case, with the unemployment rate climbing to 24.5 percent in 2011 (as during the last crisis period in 1994) and house prices falling a further 15 percent year-on-year in 2010. (The impact of the LIBOR will take effect only in 2012 due to the lag structure of the estimated forecasting equation.) Under these circumstances, NPLs are forecasted to peak in 2011 at 7.8 percent and 7.1 percent for commercial and savings banks, respectively. LGDs for nonperforming loans are assumed at 45 percent for both commercial and savings banks, respectively, and LGDs for repossessed properties are at 55 percent and 60 percent, respectively. The assumed LGDs correspond to upper bounds of analysts’ estimates under downturn scenarios. Pre-provision net earnings are expected to drop 25 percent in 2010, 15 percent in 2011, and 15 percent in 2012. We also assume that banks will set aside 10 percent of the current stock of provisions. Under these assumptions, the remaining stock of provisions and earnings at commercial banks are still sufficient to cover future losses. However, the savings banking sector is projected to have net drain on capital of €2 billion (line (15) in Table 1.11).

The results from the first part of the analysis correspond to the overall banking sectors and ignore a high level of differentiation in terms of real asset repossessions and NPLs across banks. In the second part of the analysis, we attempt to estimate what portion of the system may need capital under the baseline and the adverse-case scenarios. We base our analysis on differentiation in repossessions across banks and extend the same level of differentiation on banks’ NPLs which are often unavailable on an individual bank basis, especially for savings banks. NPLs for individual banks are expected to grow twice as slowly as repossessions for the overall system, using the same level of differentiation as in 2009:Q3.70 Individual banks’ earnings are assumed to grow at the same rate as the system under the baseline. Table 1.12 shows that the cutoff rates for repossessions in 2010 in percent of customer loans for banks that are projected to have drain on capital are 8.5 percent for commercial banks and 8.4 percent for savings banks in 2010 under the baseline (line (8) in Table 1.12). Gross drain on capital is estimated at €1 billion and €6 billion for commercial and savings banks, respectively, under the baseline (line (16) in Table 1.12). The larger drain on capital for savings banks compared to commercial banks can be explained by weaker earnings of savings banks and a greater proportion of savings banks with very large amounts of repossessions.

Table 1.12.Spain: Calculations of Cutoff Rates for Banks with Drain on Capital(In percent of total loans, unless otherwise shown)
Commercial BanksSavings BanksCommercial BanksSavings Banks
Baseline scenarioAdverse-case scenario
(1)Total loans100100100100
(2)Stock of NPL in 2010/201119.37.79.27.4
(3)Loan loss reserves2.92.92.62.6
(4)LGD for NPLs (percent)50455045
(5)Expected losses from NPL(2)*(4)−4.6−3.5−4.6−3.4
(6)Reserves–losses(3)+(5)−1.8−0.6−2.0−0.7
(7)Losses from securities−0.5−0.1−0.5−0.1
Adding repossessions
(8)Repossessions in 2010/201118.58.45.76.4
(9)Reserves for repossessions0.80.80.80.8
(10)LGD for repossessions (percent)60556055
(11)Expected losses from repossessions(8)*(10)−5.1−4.6−3.4−3.5
(12)Repossession reserves–losses(9)+(11)−4.3−3.9−2.6−2.7
(13)Total reserves–total losses(6)+(7)+(12)−6.6−4.5−5.2−3.6
(14)Pre-provision earnings in 2010–126.54.45.13.5
(15)Drain on capital2−(13)–(14)0.20.10.10.1
(16)Gross drain on capital (billions of euros)316517
(17)Memo: Tier 1 capital (end-2009, billions of euros)99789978
Source: IMF staff estimates.Note: NPL = nonperforming loan; LGD = loss given default.

Under the adverse case scenario, the cutoff rates for repossessions for banks with drain on capital are lower, so larger portions of the sectors are expected to come under pressure. Gross drain on capital is estimated at €5 billion and €17 billion for commercial and savings banks, respectively (line (16) in Table 1.12). These capital drain amounts—€5 billion for commercial banks and €17 billion for savings banks—can be interpreted as capital required to bring the respective Tier 1 capital ratios back to the levels at end-2009, assuming that risk-weighted assets remain constant in 2010.

Main Implications

Our conclusion is that a small gross drain on capital is expected in both commercial and savings banks under the baseline, despite severe economic deterioration. Under our adverse scenario, gross drain on capital could reach €5 billion and €17 billion at commercial and savings banks, respectively. These estimates are subject to considerable uncertainty and are relatively small in relation to both overall banking system capital, and importantly, the funds set aside under the resolution and recapitalization program set up by the government under the FROB of €99 billion. So far, three restructuring plans have been approved under the FROB involving a total of eight savings banks. The existing FROB scheme is currently scheduled to expire by June 2010. It is therefore important that the comprehensive resolution and restructuring processes financed through the FROB be under way before that date.

Annex 1.4. Assessment of the German Banking System71

This annex provides an assessment of potential writedowns on loans and securities, and estimates drains on capital for three major categories of German banks. The results of the exercise show that commercial banks have recognized most of the estimated total writedowns and appear to be adequately capitalized. In contrast, Landesbanken and savings banks, and other banks are yet to record a substantial part of total estimated writedowns, and are expected to have a net drain on capital.

Our estimation of potential losses and the impact on capital benefited from collaboration with the Bundesbank. The analysis focuses on the three main banking sectors: commercial banks, Landesbanken72 and savings banks, and other banks. The exercise consists of three parts: econometric forecasting of loan losses, sample-based estimation of securities’ write-downs, and the calculation of the impact on capital.

The estimates of losses on loans and securities for the three banking sectors are summarized in Table 1.13. Two sets of assumptions pertaining to the uncertainty in prices of collateralized debt obligation (CDO) securities are presented.73 Our loss estimates for the baseline case show that total bank writedowns for 2007–10 may reach a combined $314 billion. Under the adverse case assumptions, the writedowns are estimated at $338 billion for the overall banking system (Table 1.13).

Table 1.13.Estimates of German Bank Writedowns by Sector, 2007–10(In billions of U.S. dollars, unless otherwise shown)
EstimatedEstimated WritedownsEstimated WritedownsImplied Cumulative Loss RateImplied Cumulative Loss Rate
Holdings(Baseline)(Adverse Case)(Baseline, in percent)(Adverse, in percent)
Commercial Banks
Total for loans1,76566663.73.7
Total for securities1346667791.222.3
Total for Loans and Securities2,1111321436.26.8
Landesbanken and Savings Banks
Total for loans1,8061021025.75.7
Total for securities66341496.17.3
Total for Loans and Securities2,4701431515.86.1
Other Banks
Total for loans55717173.13.1
Total for securities2148222714.918.1
Total for Loans and Securities70539445.66.3
All Banks
Total for loans4,1281851854.54.5
Total for securities1,15712915211.213.2
Total for Loans and Securities5,2863143385.96.4
Note: Totals may not exactly match sum due to rounding.

Among the three banking categories, the Landesbanken and savings banks group has the highest loan loss rate, owing largely to the large losses that occurred at the Landesbanken. Landesbanken hold 50 percent of the second sector’s total loans and are characterized by relatively higher loan loss rates. Securities losses are driven by significant holdings of RMBS and CDO securities, which comprise between 50–70 percent of all structured products held by the three categories. Within the Landesbanken and savings banks group, securities losses are mostly attributed to Landesbanken, which hold over 90 percent of structured products and represent 60 percent of total securities holdings in the sector. As further analysis shows, it is the variability in the pace of recognition of these losses that results in different outcomes for the adequacy of capitalization.

Loan Loss Estimation

The methodology for loan loss estimation using dynamic panels for the three groups of banks is described in detail in Box 1.7. The forecasts are obtained assuming that bank-specific variables are constant and using WEO projections for GDP growth and the market-based forward yield curve slope (Figure 1.43). The overall loan loss rate is estimated to have peaked in 2009 at 2 percent and is projected to decline to 1.3 percent in 2010.74 The 2009 peaks of loan loss rates for commercial and savings banks have exceeded the previous peaks in 2002–03, due to their high sensitivity to GDP growth. Figure 1.44 shows how these provision rates translate into euro losses.

Figure 1.43.Germany: Loan Loss Rates

(in percent of total loans; population-based)

Source: IMF staff estimates.

Figure 1.44.Germany: Loan Losses

(In billions of euros)

Source: IMF staff estimates.

Securities Writedowns

The estimation methodology for securities losses in Germany is similar to that for the euro area described in the previous GFSRs.75 The data on holdings of securitized assets was obtained from the central bank’s quarterly survey of 18 major banks, and accounted for over 90 percent of all such holdings by German credit institutions. The survey data was broken down into the following asset categories:76 RMBS, CMBS, Consumer ABS, CDOs, and other securitized products. In order to determine securities’ loss rates, we used the CMBS and RMBS price indices from the European Securitisation forum and the euro area Aggregate Corporate benchmark index for corporate securities.

Expected Writedowns, Earnings, and Capital Requirements

Based on supervisory annual reports and our estimates for loans losses for 2009, banks will report $261 billion in writedowns by end-2009 (Table 1.14). Commercial banks had a Tier 1 capital ratio of 11 percent, the highest among the sectors. The pace of loss recognition has varied considerably across the three categories. While commercial banks have recognized all combined losses on loans and securities, Landesbanken and savings banks are likely to face an additional $47 billion in losses in 2010, and the other banking category is expected to record a further $21 billion loss.77

Table 1.14.Germany: Bank Capital, Earnings, and Writedowns(In billions of U.S. dollars, unless otherwise shown)
Commercial BanksLandesbanken and Savings BanksOther Banks1Total
Estimated Capital Positions at and 2000
Total reported and estimated writedowns at end-2009214010021261
Tier1/RWA at end 2009, in percent11.07.98.38.6
Scenario brinig Forward Expected Earnings and writedowns(Q1:Q4 2010)
Expected writedowns(2010:Q1:04)3(1)-34721
of which, loans:19274
of which, secuinties-222016
Expected net retained earning through 2010(2)24256
Net drain on capital4(3)=(1)-(2)27221436
Tier 1 capital end-2009518415545200
Source: IMF staff estimates.Note: Foreign exchange rate assumed: 1 euro =1.4 U.S. dollars.

Box 1.7.Loan Loss Estimation for Germany

The data used for loan loss estimation are from supervisory annual reports. The approach to estimation was broadly similar to the one described in the 2009 Bundesbank’s Financial Stability Review with modifications to the estimation equation and separate procedures for three banking sectors: commercial banks, Landesbanken and savings banks, and other banks.

The sample used for estimation consists of 117 commercial banks (in 2008) representing 83 percent of total assets in the data set, 440 Landesbanken and savings banks (99.6 percent of total assets), and 1,060 other banks (97 percent of total assets), with the sample of annual observations for 1993–2008.

In order to capture bank-level differentiation in cross-section and time variations, we regress the loan loss rates on its lags, banks’ total assets (size effect), the nonperforming loan ratio (a proxy for credit risk), the lending ratio (total loans to total assets), real GDP growth and its lags, the unemployment rate and its lags, and the slope of the yield curve. The final representations are presented below.

For commercial banks:

For Landesbanken and savings banks:

For other banks:

where LN(LLRATEit) is the log of the loan loss rate for bank i at time t, L. is the lag operator, LN(SIZEit) is the log of total assets, LN(NPLit) is the log of NPLs in percent of total loans, LN(LRit)is the log of the total-loans-to-total-assets ratio, DGDPt is GDP growth, and DIFF_YIELDt is the slope of the yield curve (10-year minus 1-year).

The unemployment rate was insignificant when included with GDP, and was removed from the final specifications.

Note: This box was prepared by Sergei Antoshin.

Banks’ earnings recovered in 2009, supported by the steep yield curve, reviving credit markets, and extensive government support measures. Going forward, interest income is expected to reverse these gains in 2010, due to shrinking lending margins. We assume that net interest income will decline 10 percent in 2010, given a significant flattening of the yield curve. Non-interest income and expenditures are expected to remain relatively stable, in line with the long-term trend.

For commercial banks, strong capital positions at end-2009 and faster loss recognition are expected to have a positive effect on capital levels and ratios in 2010. In contrast, Landesbanken and savings banks are projected to have sizable losses in 2010, leaving them with a net drain on capital of $22 billion. A larger portion of the drain resides in Landesbanken. Other banks are expected to have a net drain of $14 billion. These capital drain amounts—$22 billion for Landesbanken and savings banks and $14 billion for other banks—can be interpreted as capital required to bring the respective Tier 1 capital ratios back to the levels at the end of 2009, assuming that risk-weighted assets remain constant in 2010.

Annex 1.5. United States: How Different Are “Too-Important-To-Fail” U.S. Bank Holding Companies (BHCs)?

Sources: SNL Financial; and IMF staff estimates. This annex was prepared by Andrea Maechler.

Notes: SCAP =Supervisory Capital Assessment Program excluding GMAC and Metlife. TARP = Troubled Asset Relief Program. TLGP = Temporary Liquidity Guarantee Program.

References

Note: This chapter was written by a team led by Peter Dattels and comprised of Sergei Antoshin, Alberto Buffadi Perrero, Phil de Imus, Joseph Di Censo, Alexandre Chailloux, Martin Edmonds, Simon Gray, Ivan Guerra, Vincenzo Guzzo, Kristian Hartelius, Geoffrey Heenan, Silvia Iorgova, Hui Jin, Matthew Jones, Geoffrey Keim, William Kerry, Vanessa Le Lesle, Andrea Maechler, Rebecca McCaughrin, Paul Mills, Ken Miyajima, Christopher Morris, Jaume Puig, Narayan Suryakumar, and Morgane de Tollenaere.

Annex 1.1 details how indicators that compose the rays of the map in Figure 1.1 are measured and interpreted. The map provides a schematic presentation that incorporates a degree of judgment, serving as a starting point for further analysis.

Reliance on foreign bank financing is measured by the consolidated claims on an immediate borrower basis of Bank for International Settlements (BIS) reporting banks on the public sector as a proportion of GDP.

It should be noted that near-term risks associated with Japan’s elevated public debt are low due to a number of Japan-specific features, including high domestic savings, low foreign participation in the public debt market, strong home bias, and stable institutional investors (Tokuoka, 2010).

For a more in-depth review of fiscal vulnerabilities, see IMF (2010b).

Estimates of required fiscal adjustment are drawn from IMF (2010c). These estimates are based on illustrative scenarios, in which the structural primary balance is assumed to improve gradually from 2011 until 2020; thereafter, it is maintained constant until 2030. Specifically, the estimated adjustment provides the primary balance path needed to stabilize debt at the end-2012 level if the respective debt-to-GDP ratio is less than 60 percent; or to bring the debt-to-GDP ratio to 60 percent in 2030. The scenarios for Japan are based on its net debt, and assume a target of 80 percent of GDP. For Norway, maintenance of primary surpluses at their projected 2012 level is assumed. The analysis is illustrative and makes some simplifying assumptions: in particular, beyond 2011, an interest rate–growth rate differential of 1 percent is assumed, regardless of country-specific circumstances.

As of early March, the regression significantly under-predicted Greek spreads, which arguably reflected heightened liquidity concerns and policy uncertainty not captured in the model.

Swaps are used as a numeraire to compare sovereign credit risk across multiple countries. Swap spreads refer to the yield differential between a specific maturity government bond and the fixed rate on an interest-rate swap with an equivalent tenor.

Collateral requirements represent the most commonly used mechanism for mitigating credit risk associated with swap arrangements by offsetting the transaction’s mark-to-market exposure with pledged assets.

Gross sovereign default protection is $2 trillion in notional value, just 6 percent of the $36 trillion global government bond market. The more relevant net exposure (true economic transfer in case of default) represents only 0.5 percent of government debt, at $196 billion notional amount.

There is also potential for stricter collateral requirements among dealers, and between dealers and monoline insurers, and highly rated corporates and banks.

Bank earnings also potentially suffer from heightened sovereign credit risk. Sovereign ratings downgrades can increase banks’ risk-weighting for government debt holdings; fiscal and monetary tightening can lead to asset quality deterioration; and higher taxes can directly reduce bank profitability.

See Box 1.1. of the October 2009 GFSR.

Differences in the speed of realization of writedowns or loss provisions between the euro area and the United States may reflect a lag in the credit cycle in the euro area; the higher proportion of securities on U.S. banks’ balance sheets; accounting differences between International Financial Reporting Standards (IFRS) and U.S. Generally Accepted Accounting Principles (U.S. GAAP); time lags between data collection and publication by national supervisors; and differences in the frequency of reporting.

New loans became more leveraged in the run-up to the crisis (often nonamortizing) and, as leases terminate in the next few years, many owners are unlikely to find new tenants.

Foreign institutions operating in the United States are generally lightly capitalized and reliant on capital support from foreign parents. A move toward requiring more localized capital holdings by foreign operations from regulators would entail substantial capital injections from their parents (principally European banks).

$1.4 trillion of CRE loans are due to roll over in 2010–14, almost half of which are now in negative equity (Azarchs and Mattson, 2010; Congressional Oversight Panel, 2010).

This does not include the likely recapitalization of the Federal Housing Administration (FHA), whose reserves are well below the 2 percent level mandated by Congress. While it has tightened some lending standards for low-quality borrowers and raised insurance fees, the FHA is caught between the objectives of propping up the housing market and rebuilding its reserves.

The backlog of 5 million foreclosures (and short-sales) now represents one year’s total sales. The U.S. Treasury Home Affordable Modification Program (HAMP) is rapidly qualifying mortgage borrowers for trial payment modifications, but these are proving slow to convert into permanent modifications, and the program shows little sign of fundamentally changing housing market dynamics.

Monoline insurers that have guaranteed RMBS may be forced into bankruptcy if losses continue to mount. Counterparties with unhedged, unwritten-off positions to those monolines, or those unable to replace hedges, would face additional market losses.

Landesbanken are regionally oriented. Their ownership is generally divided between the respective regional savings banks associations, on the one hand, and the respective state governments and related entities, on the other. The relative proportions of ownership vary from institution to institution.

The NBER Debt Enforcement Database (Djankov and others, 2008), based on an international survey of bankruptcy attorneys, indicates that the average recovery rate on corporate NPLs in the CEE region should be around 35 percent, with significantly lower recovery rates for some countries. Market estimates of recovery rates on mortgages in the region range between 40 and 80 percent, depending on the extent to which real estate prices have declined and how well the debt collection process functions.

Using quarterly Federal Reserve and Federal Deposit Insurance Corporation (FDIC) data covering the period from 1992–2009, an equation of the form:

S = 1.2 + 0.096 (0.000) steepness + 2.36 (0.000) conc–0.048 (0.001) credgrowth explained 79 percent of the movement, where S is the spread over the Fed Funds rate; steepness is the steepness of the U.S. Treasury yield curve between three months and 10 years; conc is an index of U.S. banking system concentration constructed from FDIC data, credgrowth is the growth of credit to the private sector as shown in Figure 1.26, and the figures in parentheses after each coefficient indicate significance after applying Newey-West autocorrelation correction.

In the euro area, the total spread on new business is at roughly half its level of a year ago.

In February, the Federal Open Market Committee decided to increase the rate charged to banks borrowing at the discount window by 25 basis points to 0.75 percent.

In Japan, total bank credit growth did not increase to the same extent as in the United States and Europe during the pre-crisis period, and, by the same token, has not experienced as significant a credit withdrawal. For this reason Japan is not included in our credit projections.

The nonbank sector—primarily insurance companies, pension funds, mutual funds, and foreign central bank reserve managers—plays an important role in supplying credit to the economy, for example through purchases of corporate and government debt securities. There are two main channels through which this can occur. First, a portion of households’ and companies’ savings can provide credit, either directly through investments in debt securities or indirectly through investments made on their behalf by asset managers. The second channel occurs through foreign investment in debt issued in the economy.

Annex 1.8 on the IMF’s GFSR website discusses the impact of large-scale asset purchase programs on the cost of credit.

McKinsey Global Institute (2010) estimates. Only Spain’s nonfinancial private sector leverage ratio is higher, at 221 percent of GDP, which compares with 193 percent in Switzerland, 174 percent in the United States, 163 percent in Japan, 154 percent in France, 138 percent in Canada, 128 percent in Germany, and 121 percent in Italy.

The ex ante financing gap is the excess of projected financing needs of the public and private nonfinancial sectors relative to the estimated credit capacity of the banks and the nonbank financial sector. There can only be an ex ante gap, as ex post, a rise in interest rates and/or credit rationing will bring credit demand and supply into balance.

Annex 1.7 on the IMF’s GFSR website explains the methodology used to estimate the financing gap and compares the latest projections for 2010 with those in the October 2009 GFSR.

Chapter 4 provides an overview of the global liquidity expansion, its effects on receiving countries, and options available to policymakers in response to surges in capital inflows. The chapter also discusses the effectiveness of different types of capital controls.

Borio and Lowe (2002) discuss these challenges, and offer a preliminary empirical investigation of the factors that can increase the vulnerability of the financial system, using a small set of useful indicators of asset prices, credit, and investment.

This reflects the extraordinarily low monetary policy rates of the G-4 central banks (Bank of England, Bank of Japan, ECB, and Federal Reserve) and their generous liquidity providing operations, which has led to low interest rates and money market risk premiums, as well as high excess liquidity. Chapter 4 finds strong links between global liquidity expansion and asset prices in capital flow recipient countries.

Bank lending is recovering more slowly than portfolio flows. There was a 24 percent decline in the gross issuance of emerging markets’ and other advanced economies’ syndicated loans in 2009, and a still-negative net change in combined exposures of BIS reporting banks to countries in Europe, the Middle East, and Africa. In contrast, BIS exposures to Latin America and Asia increased in the third quarter of 2009 (the latest available data), after falling sharply during the height of the crisis.

In 2009, global ETF assets with dedicated exposure to emerging market equities increased 130 percent, compared to 24 and 52 percent, respectively, for North American and European equities, according to Blackrock, one of the leading provider of ETFs.

See Peiris (2010) and CGFS (2007). Also, JPMorgan estimates that total assets under management benchmarked to its family of emerging market debt indices increased 19 percent in 2009 to about $280 billion.

Studies by MSCI Barra indicate that home bias has only gradually been reduced over the last decade. Most institutional investors tend to partition domestic from international equity allocations, with few using a more global approach to asset allocation.

According to MSCI’s all-country world investable and GDP-weighted indices.

At the height of the crisis, for example, investment-grade corporate bonds were trading at credit spreads that only previously had been priced into high-yield bonds, and overall credit spreads were affected by the stress in market functioning, which elevated trading liquidity risk premiums.

We assess equity valuations based on forward- and backward-looking price multiples as well as a dividend discount model, which relies on longer-term expectations of earnings and real yields. Several valuation ratios were used to assess property price valuation, while different econometric approaches were employed to gauge valuation of fixed-income assets. Mature market valuations are also assessed, as emerging market assets often trade in close relation.

We acknowledge that historical and cross-economy comparisons may ineffectively capture the current state of a particular market given structural changes in markets over time and differences in market structures between economies. Moreover, Table 1.8 does not include all the factors that may contribute to the formation of financial imbalances, such as measure of credit, financial system liquidity, or investment.

Forward-looking price-to-earnings ratios of Ireland appear elevated due largely to sharp downward revisions in earnings projections.

To assess the value of local sovereign debt in selected mature and emerging economies, local government yields have been modeled using a set of standard domestic factors representing monetary policy stance, fiscal conditions, and economic activity, as well as external factors. It does not use domestic savings or the microstructure of specific bond markets as explanatory variables, which may be particularly relevant for some economies like Japan. See Tokuoka (2010).

See the April 2010 WEO for a more detailed discussion of exchange rates.

A cautionary note, these real estate ratios can also be driven by larger relative movements in the denominator not just the numerator, and high ratios may also still reflect the high valuation built up between 2003 and 2007 that is still in the process of correction. So, it is key to analyze real estate markets at a economy-specific level. In the context of Table 1.8, the indicators allow us to make comparisons across economies and guide us to where further analysis may be required.

The carry trade indicator used is the difference between one-year swap rates between the investment and funding currencies, divided by the one-year volatility implied in exchange rate options. This attempts to capture both expectations of short-term rates in a forward horizon and changes in pricing of risk and risk appetite in the currency market.

Higher global liquidity tends to boost equity inflows to emerging markets and domestic asset valuation, particularly when the receiving country’s exchange rate regime is not flexible. See Chapter 4.

N’Diaye (2009) examines the impact of U.S. monetary policy and operation on Hong Kong SAR.

Too little competition can be as damaging as too much: a balance needs to be struck in which competition is sufficient to deliver innovative and competitive financial services that support growth, but is not so intense that it depresses returns for the entire financial sector. In general, “zombie banks”—those that have lost their commercial raison d’être, but are kept in existence for political reasons or by regulatory forbearance—engage in little innovation that is supportive of growth, but depress profits for the sector, and ultimately threaten financial stability.

U.S. data highlight that the largest banks generally entered the crisis will the lowest capital ratios while enjoying a lower cost of funding, suffered the greatest losses, and enjoyed the most government support and subsidy. Crisis mergers have meant that the top four banks have sharply increased their asset size relative to GDP and other bank assets (see Annex 1.5). Through the higher credit ratings arising from perceived government support, the five largest U.K. banks are calculated to have benefited by a total of £55 billion per year during 2007–09 just from preferential wholesale funding rates (Haldane, 2010).

In the European Union, the Commission’s Competition Directorate is requiring banks as a condition of significant state aid to cancel or defer coupons on preferred shares and hybrid instruments and dispose of banking units and subsidiaries to reduce concentration and encourage entry into banking markets. While not fully addressing the too-important-to-fail problem, this process goes some way toward redressing the moral hazard consequent upon crisis assistance. The absence of a similar process in the United States, Japan, and Switzerland leaves such sovereigns more exposed to contingent liabilities from more concentrated banking systems than otherwise.

This annex was prepared by Peter Dattels, Ken Miyajima, Rebecca McCaughrin, and Jaume Puig (see Dattels and others, forthcoming).

The GFSM was first introduced in the April 2007 GFSR.

The GFSM raises early warning signals when risks are excessively low and conditions loose, gauged against historical norms. During crises, the GFSM generally captures the worsening of risks and conditions contemporaneously (Dattels and others, forthcoming).

Moving averages are often used for higher frequency data to extract the trend and identify inflection points.

The description of the GFSM’s results before its introduction in the April 2007 GFSR is based on a reconstruction of the model’s results with past observations for the indicators used in the October 2009 GFSR (see also Dattels and others, forthcoming).

This annex was prepared by Joe Di Censo and Manmohan Singh.

Gross notional is the sum of CDS contracts bought. The aggregate net notional exposures shown herein reflect the net amount of protection bought for all net purchasers of CDS. This net exposure represents the maximum economic transfer in the event of default.

In this discussion, the bond yield spread refers to the yield differential between Greek government debt and equivalent maturity German bunds.

In contrast, contemporaneous changes in Greek CDS and cash spreads were positively correlated (0.27).

The difficulty of shorting bonds in order to sell CDS protection and arbitrage the bond-derivative basis suggests that CDS may actually “pull” bond yield spreads tighter, rather than “push” them wider. Assuming risk neutrality, any CDS premium should equal the cash credit spread of a par fixed-coupon bond of the same maturity. If the CDS spread exceeded the credit yield spread, an investor could sell CDS in the derivatives market and synthetically replicate that position by shorting a par fixed-coupon bond (on the same reference entity with the same maturity as the swap’s tenor) and invest the proceeds in a like-maturity risk-free security. In reality, shorting bonds is difficult. So CDS moving the cash market wider is less likely than the reverse scenario of bond yield spreads “pulling” CDS tighter.

Collateral requirements represent the most commonly used mechanism for mitigating credit risk associated with swap arrangements by offsetting the transaction’s mark-to-market exposure with pledged assets. Yet most sovereigns and foreign provinces/municipalities do not post collateral. This practice is due primarily to the lack of legal clarity surrounding enforcement of collateral rights against sovereigns.

This annex was prepared by Sergei Antoshin and Narayan Suryakumar and draws extensively upon Giustiniani (2009) and subsequent works.

The three-year horizon corresponds to the period over which most loans are completely written off under the Spanish accounting rules. Mortgages are written off over six years, which leaves the possibility of using earnings after 2012 to absorb losses.

As the analysis below shows, we view real asset repossessions as an additional risk factor affecting future losses. When NPLs and repossessions are combined, the share of problem assets in percent of total loans is higher for savings banks.

The above assumptions often correspond to lower bounds of market estimates.

The methodology for estimating securities’ losses is consistent with the approach to the euro area outlined in the previous GFSRs and is based on securities’ holdings provided by the Bank of Spain. All of the estimated losses are expected to originate from holdings of foreign securities. The relatively small loss figure can be attributed to the strong improvement in corporate securities prices over the past year and the marginal exposure of both the commercial and savings banks to toxic assets. Banks’ holdings of retained asset-backed securities are treated as loans because Spanish banks have retained nearly all the asset-backed and mortgage-backed securities they have originated over the past two years, in order to use them as collateral in tapping ECB facilities.

This part of the analysis benefited from the use of data on Spanish banks from Analistas Financieros Internationales. All estimates are those of the authors.

Repossessions of real assets are calculated as the sum of item 9 “Activos no corrientes en venta,” item 13.2 “Inversiones inmobiliarias,” and item 16.1 “Existencias” from the Consolidated Balance Sheets for commercial banks and savings banks, obtained from the banking associations. For the analysis, we use flows between 2007:Q2 and 2009:Q4.

The assumption is based on repossessions being viewed as the overall risk factor, which can also be extended to some degree (in our case, 50 percent) to NPLs. In other words, banks use both repossessions and NPLs to manage credit risks. However, a counterargument can be made that banks that bring real assets onto balance sheets effectively reduce their NPLs. The results of the exercise are likely to change under the inverse relationship assumption, generating a lower estimate for the impact on capital.

This annex was prepared by Sergei Antoshin and Narayan Suryakumar.

Landesbanken are regionally oriented. Their ownership is generally divided between the respective regional savings banks associations on the one hand and the respective state governments and related entities on the other. The relative proportions of ownership vary from institution to institution.

CDO prices are characterized by the highest loss rates across security classes and have a significant impact on the overall estimates of losses on securities. In our baseline case, we assume that loss rates for CDOs are 50 percent, while in the adverse case, they are assumed at 70 percent.

The ratio of the overall loss rate in 2009 to the overall loss rate in 2008 is 3.3, which is similar to the respective ratio for our sample of German listed banks whose 2009 loan loss provisions are already publicly available.

The aggregated balance sheet data, including the composition of the securities holdings, the profit and loss accounts, and capital bases for the different banking categories, were obtained from the Bundesbank.

The proportion of structured products to total securities holdings is roughly 60 percent for commercial banks, 65 percent for other banks, and 18 percent for Landesbanken and savings banks.

The remaining securities losses for savings and other banks are assumed to be recognized through the profit and loss account in 2010. Given that banks need not mark to market their entire securities portfolio, our assumption on the impact on earnings and capital is a conservative one.

    Other Resources Citing This Publication