Journal Issue
Share
Article

2011 Spillover Report—Selected Issues

Author(s):
International Monetary Fund
Published Date:
July 2011
Share
  • ShareShare
Show Summary Details

I. Euro Area Output Spillovers1

This note analyzes output spillovers from the Euro Area (EA) using a global vector autoregression (GVAR) and a macroeconometric model of the G-20. These two approaches are used to assess the impact of shocks in the EA on other systemically important economies.

In the first instance, a GVAR model is used to shed light on spillover effects across countries. The approach uses a dynamic multi-country framework for the analysis of the international transmission of shocks and is based on the GVAR toolbox, launched in December 2010, and sponsored by the ECB.2 It comprises 26 economies, with the EA as one of the economies covered. The model is constructed by combining separate models for each of the 26 economies linking core variables within each economy with corresponding trade-weighted foreign variables. EA variables are GDP-weighted aggregates of eight countries (Austria, Belgium, Finland, France, Germany, Italy, Netherlands and Spain). The model has both real and financial variables: real GDP, inflation, the real equity price, the real exchange rate, short and long-term interest rates, and the oil price. All the data are observed at the quarterly frequency, from 1979Q2 to 2009Q4.

GVAR estimates show that output spillovers from the EA to other S-4 economies are meaningful (Figure 1). Output spillovers are measured as ratios of the peak impulse responses of output to the peak impulse response of output in the EA. The spillover coefficients can be thought of as elasticity measures. The EA has the strongest spillovers to the neighboring UK, and the least to the US. These spillovers to the rest of the S-4 range from about 10 percent for spillovers to the US to about a quarter for spillovers to UK. In turn, output spillovers from the US to the EA are stronger than spillovers from the EA to the US. The US impact is most pronounced on the UK.3

Figure 1.GVAR Results: Peak Impulse Responses of Output

(Relative to each S-4 economy)

Source: Staff calculations.

To gauge the size of output spillovers from another angle, simulations are also undertaken using a macroeconometric model.4 Estimates are derived from a macroeconometric model which features extensive linkages between the real and financial sectors within and across the G-20 economies. The EA is represented in the model by France, Germany, Italy and Spain. The variables under consideration are the GDP deflator, a consumption price index, real GDP, real domestic demand, a short term nominal interest rate, a long term nominal interest rate, an equity price index, the nominal bilateral exchange rate, and the prices of energy and non-energy commodities.

The macroeconometric model simulations are largely consistent with the GVAR results of positive, but modest, spillovers, with spillover strength depending on proximity. Five scenarios are generated with supply, demand, monetary policy, term premium, and equity risk premium shocks in the EA. These shocks are calibrated to raise output in the EA by one percentage point in all five scenarios. A negative equity risk premium shock, for example, would increase wealth and demand in the EA, resulting in stronger demand for exports from other regions via the trade channel. Through financial linkages, it would also raise output elsewhere with the size depending on the dependence on the financial sector and the strength of linkages with EA financial markets. The peak impulse responses from these five scenarios are then averaged.

The results are similar to those obtained from the GVAR model (Figure 2). Spillovers from the EA are strongest to the UK, while the US and the UK have the strongest impact on the EA.

Figure 2.G-20 Model Results: Average Peak Impulse Responses of Output

(Relative to each S-5 economy)

Source: Staff calculations.

Note: Depicts the average peak impulse responses of output to supply, demand, monetary policy, term premium, and equity risk premium shocks in China , the Euro Area , Japan , the United Kingdom , and the United States which raise output there by one percent.

The peak impulse responses of output to shocks in the EA are increasing with geographical proximity to the EA, reflecting the strength of trade and financial linkages. The highest spillovers are to Russia and the UK, but the magnitude is moderate with average peak impulse response ratios of about a quarter (Figure 3). In the case of Russia, this results reflects its strong trade and commodity price linkages with the EA, while for the UK it reflects its strong trade and financial linkages. The trade channel is also important for Turkey and Saudi Arabia, with the latter having significant exposure via the commodity price channel.

Figure 3.Average Peak Impulse Responses of Output to Shocks in the EA

(Relative to the EA)

Source: Staff calculations.

Note: Depicts the average peak response of output in each S-5 economy to shocks in each of the other S-4 economics which raise output there by one percent.

II. Contribution of the Euro Area to Common Risk in Global Financial Markets5

This note analyzes the relative contributions of different regions and assets to common risk across global financial markets. In doing so, it makes use of a simple principal component analysis on the risk premiums of a variety of assets.

A simple principal component analysis is used to estimate the extent to which unobservable shifts in common risk factors contribute to observed changes in asset-specific expected returns. As international investors react to shocks by rebalancing their portfolios in asset markets that would otherwise be unrelated, any change in the willingness of global investors to bear risk—or any common shock—is deemed to raise the co-movement across asset returns. By assuming that risk premiums embedded in selected asset yields differentials are determined jointly in the market and influenced by both asset-specific factors and a common factor, the latter component can be identified and—thereby—stripped out. In other words, if there is an increase in the (risk-neutral) probability of default for all asset considered—which most likely happened during the global financial crisis—this would likely be picked up in the principal component, along with shifts in investors’ attitude toward risk.

The methodology is used to assess the contribution of Euro Area (EA) asset markets to the estimated common risk component by adding up the contribution to the common risk component of all asset markets in the region. In addition, the analysis allows to gauge the extent of volatility spillovers from individual EA asset markets across borders and across markets, once we abstract from “risk commonalities”.

The analysis relies on the set of risk premiums embedded in the following yield differentials:

  • U.S. asset-backed commercial paper (versus the 3-month U.S. Treasury bond yield);
  • Three-month U.S. dollar, euro, sterling, and yen London interbank offered rates (versus their corresponding overnight index swap rates);
  • U.S., euro-area, UK, and Japanese high-yield financial and industrial corporate bonds (versus their respective benchmark 10-year government bond yields);
  • U.S., euro-area, UK, and Japanese equities (whose implied risk is computed as the earning price ratio versus their respective benchmark 10-year government bond yields);
  • 10-year sovereign bonds (over Bunds) for peripheral euro-area countries (including Greece, Ireland, Spain, and Portugal);
  • Asia, Europe, and Latam emerging markets bonds (whose implied risk is given by their global EMBI+ spread versus the 10-year U.S. Treasury bond yield).

As shown in the chart below, the estimated unobserved factor indicates that the common risk component increased sharply during Spring 2010 as sovereign pressures in peripheral European countries intensified, but the rise in risk commonalities was not as severe as it was during the time of the Lehman bankruptcy.6

Measuring risk commonalities: alternative proxies

(basis points, LHS; index, RHS)

Sources: Bloomberg and staff calculations.

The analysis suggests that European financial markets play an important role in transmitting financial shocks to the rest of the world. In particular, the EA is estimated to contribute to over one-fifth of the changes in risk commonalities—a contribution which is second only to that of the US.

Contribution of individual regions to risk commonalities

(basis points)

Sources: Bloomberg and staff calculations.

Contribution of individual regions to risk commonalities

(percentage, average over August 2007 - February 2011)

Source: Staff calculations

Volatility spillovers from the EA periphery sovereign bond market to the rest of the world appear to be limited. Specifically, while looking at “raw” cross-market correlations (blue bars in the left chart below) indicates that “observed” volatility co-movements across assets and across borders tend to be widespread, examining cross-correlations of spreads where the common risk component has been stripped out (red bars in the left chart below) reveals that significant volatility spillovers from the EA sovereign bond market are likely to be felt only in the bond market of EA financial corporations, signaling a surge in EA banks’ perceived riskiness. To a lesser extent, increases in EA sovereign risk are also expected to adversely affect the perceived riskiness of EA non-financial corporations and emerging European countries.

Conditional correlations vs peripheral EA sovereign bond spreads

Source: Staff calculations.

Conditional correlations vs EA financial corporate bond spreads

Source: Staff calculations.

However, if stress spreads from the periphery to core EA financial institutions, the potential for spillovers is much larger. In particular, once the common risk component is stripped out, specific volatility spillovers from the EA financial sector have the potential to increase significantly the perceived riskiness of EA non-financial corporations, Japanese and U.S. banks (red bars in the right chart above), even though “raw” conditional correlations between spreads of EA and Japanese financial institutions are seen to remain below the 0.2 threshold. Incidentally, the analysis also seems to suggest that shocks to EA financial corporate bond spreads tend to be negatively correlated with those to EMBI+ bond spreads, once global risk commonalities have been set aside.7

III. Market Assessment of Spillovers from the Euro Area Based on Conditional Distress Probabilities8

This note analyzes Euro Area (EA) sovereign and financial distress spillovers using market information, including credit default swaps, equity prices and sovereign yields. The results are presented in heat-maps showing the sensitivity of non-EA countries to developments in the EA as well as the importance of groups of EA countries in generating spillovers to the rest of the world. The sample consists of the following countries: Austria, Belgium, France, Germany, Greece, Ireland, Netherlands, and Portugal (EA), and Brazil, China, Hungary, India, Indonesia, Japan, Korea, Russia, Turkey, United Kingdom and United States (non-EA). When studying banks, one major bank from each country is used.9

Spillovers are measured by averages of estimated Conditional Probabilities of Distress (CoPoD) in non-EA sovereigns and banks given distress in EA sovereigns and banks. Distress is defined as a (hypothetical) credit event that triggers CDS contracts.10 For example, if the CoPoD in Bank A given distress in Sovereign B is 0.5, CDS market implied probability suggests that there is a 50-percent probability that a (hypothetical) credit event in Sovereign B would be followed by a CDS event in Bank A. CoPoDs represent the market’s assessment of potential spillovers through a variety of channels such as direct exposure to governments and banks, deleveraging and market confidence.

The CoPoDs are estimated using linear and non-linear dependence between individual probabilities of distress.11 Probabilities of distress (PoDs) are first derived from market quotes of five-year sovereign and bank CDS spreads in U.S. dollar.12 Pairwise CoPoDs are calculated for a group of sovereigns and banks. A matrix of distress dependence between sovereigns/banks can be derived. Although conditional probabilities do not imply causation, the set of pairwise conditional probabilities can provide important insights into interlinkages and the likelihood of contagion between the sovereigns/banks in the system. The results shown below are taken from averages of CoPoDs from January 2010 to April 2011, covering a period of market turmoil for European sovereigns and banks.

The analysis shows that EA spillovers can be substantial. The impact is largest in neighboring Europe and smallest in Asia. Sovereign distress in the EA program countries could have knock-on effects on banks, including in the core EA, which would likely in turn have global systemic effects.

Sovereign-to-Sovereign Spillovers

Staff analysis suggests that sovereign credit events in peripheral EA program countries have larger spillovers to neighboring sovereigns than to elsewhere. The estimated spillovers from peripheral EA program country sovereigns tend to be largest in Eastern Europe, reflecting close links, through market confidence, but meaningful spillovers also occur to some other advanced and emerging markets such as Brazil (chart below).

Conditional Probability of Distress of each Non-Euro Area sovereign given Euro Area Program Country Sovereigns Fall in Distress

Source: Staff calculations.

Sovereign-to-Banks Spillovers

Sovereign credit events in the EA program countries would likely mainly affect banks in those countries. There is a larger impact on countries with close links to the EA such as Hungary and Turkey, while China and Japan are affected less, reflecting their lower exposure.

Conditional Probability of Distress of Global Banks Given Euro Area Program Country Sovereigns Fall in Distress

Source: Staff calculations.

Bank-to-Bank Spillovers

Estimated spillovers from EA program country banks are large for other banks in the region but smaller for banks elsewhere. The strongest distress spillovers are within Europe itself, including in Hungary, Turkey, Russia and the UK. Brazil is also indirectly affected, reflecting the large presence of Spanish banks. As expected, spillovers to the Asian and US banking systems are relatively less significant, in part due to their lower direct exposures to the program country banks but also because their earnings and profitability are supported by strong local economic conditions (Asia) or driven by investment banking returns which remained robust over 2010 helped by increased asset prices and renewed risk appetite (US). Conditional probability of default for the U.S. hovers near the low end of the 0.2-0.4 range, while Brazil is at the upper end.

Conditional Probability of Distress of Global Banks Given Euro Area Program Country Banks Fall in Distress

Source: Staff calculations.

Core EA bank distress would, however, be a systemic event impacting banks globally. Spillovers from core EA banks are very large for Hungary, Turkey, the UK, Brazil and Russia. Distress in EA core banks would represent a systemic event that could impact banking systems far beyond the European region, including the US and Asia (The conditional probability of default for the U.S. increases to the upper end of the 0.2-0.4 range under such conditions).

Conditional Probability of Distress of Global Banks Given Core Euro Area Banks Fall in Distress

Source: Staff calculations.

Japanese and Chinese banks remain the least affected by spillovers from the EA. The relatively weaker impact on Japanese and Chinese banks is a result of their lower direct exposure and limited holdings of bonds issued by European sovereigns and banks. Moreover, they tend to have strong liquidity positions based on a substantial local deposit base, and are therefore relatively insulated from funding withdrawal pressures due to systemic concerns. Perhaps the most significant impact of EA banks distress on Asian banks stems from their impact on global risk aversion, trade and growth prospects.

Summing up, the global impact of distress in program country sovereigns and banks is potentially sizable, but much smaller than the estimated global spillovers that would arise if stress were to also spread to core EA banks. This critical result can also be seen by mapping unconditional distress probabilities across global banks. While implied unconditional PoDs (derived from current market CDS in April 2011), are lower than the implied conditional probability in the event of distress in the program countries (light blue bars in text chart below), the implied probability in the event core EA banks are affected is much higher for all countries (dark blue bars in text chart), and would generally be highest in countries that are geographically close to the EA and whose banking systems are most closely linked to that of the EA. For most non-EA banks in the sample, the estimated conditional probability of distress given distress in a major core EA bank is as high or higher than the peak implied unconditional probability of distress in the period after the Lehman bankruptcy.

Current implied probability of distress compared to expected probability of distress in non-EA banks…

… given distress a) in GIP sovereign, b) in core EA bank

Source: Staff calculations.

IV. Cross-border Deleveraging Spillovers of the EA Sovereign Debt Crisis13

This note presents simulations of cross-border spillovers of the Euro Area (EA) sovereign debt crisis. The simulations are performed based on BIS foreign claims data and the model of bank deleveraging developed in Tressel (2010).14 The behavioral assumption is that banks maintain a target minimum capital-to-asset ratio by contracting their balance sheet when experiencing sudden losses.

Two scenarios are considered (Table 1). The events triggering these scenarios are assumed to take place over a short time span during which banks cannot raise equity as they correspond to episodes of market stress. In these scenarios, claims held in the trading books of international banks bear losses of 30 percent as a result of a sudden and sharp increase in bond yields, with 20 percent of the claims assumed to be in the trading books of banks. The increase in yields affects claims on governments and banks, with the EA program countries (Greece, Ireland, and Portugal; GIP) affected in the first scenario and other EA banks also affected in the second.

Table 1.Scenario assumptions
Domestic
ScenarioShockImpact onholdings ofDeleveraging
CountriesSectorLossbalance sheetsovereign debtby subsidiaries
sovereignTrading books
1GIP& banks30 percent(20 percent)Same loss20 percent
GIP sovereign
and banks +sovereignTrading books
2other EA banks& banks30 percent(20 percent)Same loss20 percent
Source: Staff calculations.
Source: Staff calculations.

Exposures to EA program countries suggest that the spillovers will be mainly channeled by German and French banks. Because of their size and high leverage, French and German banks are the main sources of spillovers within the EA and between the EA and other regions. According to the most recently available sectoral bilateral exposures of the BIS consolidated banking statistics, German and French banks have large exposures in percent of their equity to the sovereigns and banks of Greece, Ireland, and Portugal (Figure 1).15

Figure 1.Exposures of international banks to GIP sovereign and banks

(In percent of bank equity)

Sources: Bank of England; BIS Consolidated Banking Statistics; Bankscope; IMF, International Financial Statistics; and IMF staff calculations.

Note: The exposures were adjusted using data from the Bank of Ireland to account for the fact that a significant portion of the claims are claims on foreign banks domiciliated in Ireland.

1 EA3: Greece, Ireland, and Portugal.

2 Other EA includes Austria, Belgium, Ireland, Portugal, and the Netherlands.

3 The exposures are calculated in percent of the equity of banks that have foreign exposures. Banks that do not have exposures to Greece, Ireland, and Po are not included in the computation.

Losses of major international banks become large when market stress spreads to more countries. Losses of banks appear manageable as long as the crisis is contained to Greece, Ireland and Portugal. However, losses of trading books become large and exceed 10 percent of equity of French and German banks under scenario 2, when market sentiment about the core EA also turns negative.

As a result, deleveraging in absolute terms would be the largest within the EA itself (Table 2). The deleveraging would negatively affect intra-EA financial integration (Figure 2). In absolute terms, the US and the UK would be the most affected after the EA, followed by Central and Eastern European countries. Most of deleveraging would be caused by EA banks.

Table 2.Reduction in Liabilities to Foreign Banks(In bil. US$)
scenario:12
Advanced countries120324
of which:
Euro area 1/54154
Japan613
United Kingdom1445
United States2871
Emerging Markets1336
of which:
Central and Eastern Europe619
Asia37
Latin America25
Other emerging markets24

Excluding Luxembourg Source: Staff calculations

Excluding Luxembourg Source: Staff calculations

Figure 2.Reduction in foreign liabilities to foreign banks (in bil. US$)

(Scenario 2)

Source: Staff calculations.

Advanced economies and emerging markets in Europe are particularly vulnerable to deleveraging. In scenario 2, countries that are the most susceptible to experience capital outflows by international banks outside of the EA are mostly in Europe: Poland, Czech Republic, Hungary, Nordic and Baltic countries. Other emerging markets are resilient to such shock. Figures 3 and 4 illustrate the estimated deleveraging under scenarios 1 and 2, respectively.

Figure 3.Estimated Deleveraging by International Banks, Scenario 1

(In percent of GDP)

Source: Staff calculations.

Figure 4.Estimated Deleveraging by International Banks, Scenario 2

(In percent of GDP)

Source: Staff calculations.

Taking account of indirect exposures to the EA program countries could significantly affect the mapping of potential spillovers. Recent data from the BIS provides some information on total direct and indirect gross exposure, including through guarantees via CDS contracts (text table). According to these data, the total exposure U.S. banks, for example, may be several times as large as just the direct exposure, while the indirect exposures of U.K. banks may be as large as their direct exposures. In contrast, indirect exposures of French and German banks, while significant, are generally several times smaller than their direct exposures.

Debt issued by:
GreeceIrelandPortugal
France
Direct gross exposures 1/17.112.114.3
Indirect gross exposures8.3126.45.24
Germany
Direct gross exposures 1/24.931.723.5
Indirect gross exposures5.940.413.8
US
Direct gross exposures 1/313.83.6
Indirect gross exposures34.15441.2
UK
Direct gross exposures 1/6.022.86.8
Indirect gross exposures5.059.24.7
Japan
Direct gross exposures 1/0.93.11.4
Indirect gross exposures0.11.30.6
Source: Consolidated Banking Statistics, Table 9E, Q4 of 2010 (as reported in the BIS Quarterly Review of June 2011)

Exposures to banks and sovereign only

Source: Consolidated Banking Statistics, Table 9E, Q4 of 2010 (as reported in the BIS Quarterly Review of June 2011)

Exposures to banks and sovereign only

V. Euro Area Spillovers: Global Projection Model Analysis

This note analyzes spillovers from the Euro Area using the Fund’s Global Projection Model (GPM) 16. The GPM model is a six-region non-linear rational expectations model comprising the US, the Euro Area (EA), Japan, Emerging Asia, Latin America and a group of remaining countries. It features two types of demand shocks: an idiosyncratic demand shock that originates in a particular block and propagates to other regions via the output gap over time, and instantaneous global shocks—e.g., the sovereign market distress in Europe—that are applied instantaneously to all blocks in the model. Additionally, GPM incorporates real-financial linkages—for example, bank lending tightness variables—and spillover channels such as demand shocks, exchange rates, inflation and interest rate in its estimates.

In the EA spillover analysis, GPM constructs two scenarios:

  • A baseline or “tremor” scenario, analyzes spillovers to the rest of the world from the ongoing sovereign debt problems in EA program countries. Under this scenario, shocks to EA financial conditions (i.e., bank lending conditions) and domestic demand are equal to about half of those experienced in the Lehman crisis.
  • A second, downside, “earthquake” scenario measures possible spillovers should sovereign risk premiums, growth declines and contagion be even larger. Under this scenario, the magnitude of the shocks are doubled from the tremor scenario. In other words, bank lending is affected in a manner similar to Lehman episode and domestic demand also doubled.

GPM model results from the “tremor” scenario indicate that a milder shock arising from the ongoing sovereign debt crisis in EA program countries would elicit a GDP reduction in the region, with only modest spillover elsewhere. The projected downturn in growth stems primarily from program country governments’ fiscal responses, which would be expected to play a significant role in restraining near-term growth in those countries. However, the impact on the rest of the EA and elsewhere would be limited as the real economy of the core EA would be little affected by financial headwinds that buffet the program countries.

In the “earthquake” scenario, GPM measures the tail downside spillover risk should a worst case scenario occur, in which strains spread to the core EA. In this scenario, insufficiently rapid and strong policy action would lead to significant financial market losses, resulting in a substantial decline in capital ratios in all EA countries. Under these conditions, the overall impact on world growth would be substantially higher. Compared to the pre-crisis baseline, EA and U.S. growth in 2011 would be lower by 1.4 and 0.7 percentage points, respectively. Spillover to other regions would also be higher compared to the tremor scenario though the magnitude of impact would be smaller relative to the EA and the US (see table below17).

Effect of Euro Area Turbulence on GDP Growth(Deviation from pre-crisis baseline, in percentage points)
“Tremor” Scenario“Earthquake” Scenario
Difference from Pre-crisis BaselineDifference from Pre-crisis Baseline
2011201220112012
Country/RegionAnnualAnnualAnnualAnnual
U.S.-0.2-0.2-0.7-0.7
Euro Area-0.4-0.2-1.4-1.3
Japan0.0-0.1-0.1-0.4
Emerging Asia-0.1-0.1-0.3-0.4
Latin America-0.1-0.1-0.2-0.3
Remaining GPM countries-0.2-0.2-0.6-0.8
World 1/-0.2-0.1-0.4-0.5
Source: WEO Update (January 2011).

GPM world represents 87.5 percent of world GDP by PPP (2007-2010 average).

Source: WEO Update (January 2011).

GPM world represents 87.5 percent of world GDP by PPP (2007-2010 average).

VI. Spillovers from the Euro Area’s Monetary Policy and Liquidity Operations18

This note analyzes the global implications of the ECB’s policies. It considers spillovers from the Euro Area (EA) under two scenarios: (i) faster-than-anticipated monetary policy tightening; and (ii) a withdrawal of the ECB’s exceptional liquidity provision to banks. Staff analyses show that the first scenario would generate modest spillovers. By contrast, the repercussions from an early withdrawal of exceptional liquidity could be significant, especially for the UK and some CEE countries.

Monetary policy tightening

Under the scenario below, it is assumed that the short-term nominal interest rate returns to its neutral level six quarters sooner than anticipated by markets. The baseline scenario assumes that monetary tightening in the EA proceeds at the pace expected by the euribor futures market. Under the alternative scenario, the ECB tightens monetary policy faster than under the baseline to control inflation. Both scenarios abstract from monetary policy stabilization in the rest of the world. These scenarios are simulated with a refined version of the structural macroeconometric model of the G-20 documented in Vitek (2010).19

Accelerated monetary tightening is estimated to generate a modest output loss in the EA. Taking the difference between the alternative and baseline scenarios, a cumulative output loss of 0.9 percent is estimated over the period 2011-2016. This output loss is generated by the interest rate and exchange rate channels of monetary policy, amplified by an international financial accelerator mechanism.

Short Term Nominal Interest Rate

Source: Bloomberg, Staff calculations.

Cumulative Output Losses: 2011 – 2016

Source: Staff calculations.

Accelerated monetary tightening in the EA is estimated to generate moderate spillovers to the rest of the world. Taking the difference between the alternative and baseline scenarios, estimated cumulative output losses for other advanced economies range from 0.0 to 0.3 percent over the period 2011-2016, while for emerging economies they range from 0.0 to 0.2 percent. These spillovers primarily reflect reduced export demand from the EA, mitigated by real effective currency depreciations in the rest of the world.

ECB’s liquidity provision

Simulations are first undertaken to estimate cross-border spillovers of the ECB full allotment liquidity provision to the periphery EA’s banks. The simulations provide a counterfactual of the positive spillover effect of full allotment liquidity provision at a fixed rate. The analysis is performed based on the model of bank deleveraging based on the behavioral assumption that banks maintain a target minimum leverage ratio by contracting their balance sheet (they “deleverage”) or recapitalizing when experiencing losses or an increase in the required minimum capital to asset ratio. 20

The ECB started providing unlimited liquidity at a fixed rate in October 2008. The ECB Governing Council decided to change the procedure of the weekly main refinancing operations to a tender procedure with full allotment at a fixed rate to remedy the malfunctioning of the money market.21 This helped lower the refinancing cost of the marginal borrower with high refinancing needs. Given renewed market tensions, in December 2010, the ECB decided to prolong the 3-month Long Term Refinancing Operation (LTRO) as fixed rate full allotment tenders at least until July 2011. As a result, the ECB’s exposures to the periphery EA’s banks are significant.22 Other non-periphery EA banks also remain significantly exposed to the banks of the periphery.

ECB liquidityForeign Interbank
fundingLiabilities
(Q4 of 2010)(Q3 of 2010)
Greece130.710.1
Ireland124.8193.2
Portugal54.049.2
Spain 1/72.7248.9
Note: in billions of US$

ECB financing as of end January 2011

Sources: BIS, ECB, DLX, and Central Bank of Ireland
Note: in billions of US$

ECB financing as of end January 2011

Sources: BIS, ECB, DLX, and Central Bank of Ireland

The scenario assumes that the full allotment liquidity provision by the ECB has prevented excessive movements in the funding costs of the banks of the periphery. The liquidity provision at full allotment is assumed to have lowered funding costs of the banks of the periphery and of subsidiaries by 500 basis points, and overall funding needs are approximated by the sum of (i) the liquidity provision by the ECB to these banks; and (ii) total interbank liabilities.23 In other words, the simulation illustrates the counterfactuals, i.e. periphery banks’ funding costs, both from ECB refinancing facility and from the interbank market, would increase by 500 bps in the absence of the ECB liquidity provision. The scenario assumes that subsidiaries of banks of the periphery deleverage in the same proportion as their parent in response to a withdrawal of the ECB exceptional liquidity provision. To assess robustness, we also consider an alternative scenario in which the funding shocks differs across countries. Funding costs are assumed to increase by 700 bps in Greece and Ireland, 400 bps in Portugal, and 250 bps in Spain. Finally, to gauge how the simulated deleveraging is affected by the size of the funding shock, we consider two additional scenarios, in which funding costs rise by respectively 250 bps and 750 bps.

The funding shock is derived from the increase in credit risk premia during recent market stress since 2008. The assumed 500 bps is within the range of the decrease in euribor in late 2008 at the time the exceptional liquidity provision was initiated and with the range of the increase in long-term rates in 2010, both of which reflect increased credit risks (Figure 1). The first chart shows the decline in euribor as a result of policy reactions at the onset of the financial crisis in 2008. The second chart shows movements in government bond yields of the periphery. The main scenario thus assumes that funding costs of ECB refinancing or the interbank market would increase by this amount should the ECB return to competitive auctions for its refinancing facilities. Given uncertainties in quantifying a counterfactual, we considered three additional scenarios to ascertain robustness of the main conclusions.

Figure 1.Movements in short-term and long-term financing costs

Source: Datastream.

Estimated losses in absence of exceptional liquidity provision

(In percent of bank equity)

Source: Staff calculations

The simulation shows that exceptional liquidity provision has helped contain funding costs significantly. Should exceptional liquidity provision be ended, the funding shock would be substantial for Greek, Portuguese and Irish banks, resulting in losses exceeding 20 percent of bank capital. The shock would be more manageable for Spanish banks.

Positive spillovers from ECB refinancing operations occur mainly through the operations of Greek, Portuguese and Irish banks. In the main scenario, a sudden rise in funding costs would result in a large deleveraging by Irish banks (36.5 percent of foreign claims), Portuguese banks (about 15 percent of foreign claims) and Greek banks (about 5 percent of their claims). In the alternative scenario, the impact would be broadly similar. However, if the sudden increase in funding costs reaches 750 bps, the deleveraging would be much larger, as Irish, Portuguese, Greek and Spanish banks would reduce their foreign assets by respectively 68 percent, 28 percent, 18 percent and 1 percent.

Reduction in Foreign Claims(Bil US$)
Reduction in ForeignAlternativeScenarioScenario
Claims (Bil US$)Main scenarioscenario(250 bps)(750 bps)
Advanced countries21635136416
of which:
Euro area 1/7912113151
Japan69110
United Kingdom8314014160
United States3456665
EA program countries1012220
Emerging Markets917129
of which:
Central and Eastern Europe715021
Asia0000
Latin America2107
Other emerging markets0101

Excluding Luxembourg

Excluding Luxembourg

Most spillovers would occur within the EA and between the EA and other advanced economies (Figure 2). Spillovers within the EA would account for a large share of the contraction in foreign activities of international banks, in particular in the second scenario in which ECB liquidity provisioning plays a broader stabilizing role in the interbank market. Because of its role as a financial center and interconnectedness with other European countries, the UK would be significantly affected by the withdrawal of ECB exceptional liquidity provision. Spillovers to emerging markets would be small on average, except for a small number of CEE countries. While these spillovers are large, it should be recognized that ongoing exceptional liquidity provision can also have drawbacks, notably delayed restructuring of weak banks.

Figure 2.Estimated deleveraging in the main scenario

(In percent of GDP)24

Source: Staff calculations.

VII. Euro Area Fiscal Consolidation: Model Simulation Results25

This note analyzes spillovers from planned fiscal consolidation in the Euro Area. The analysis is based on scenarios simulated with the latest 6-region version (including U.S., Euro Area, Japan, China, Emerging Asia, and Rest of the World) of the IMF’s dynamic general equilibrium model—Global Integrated Monetary and Fiscal Model (GIMF)—and a refined version of the structural macroeconometric model of the Group of Twenty(G-20 model).26 These alternative models provide complementary analyses of spillovers from fiscal consolidation in the Euro Area. The GIMF model features a more detailed fiscal transmission mechanism, while the G-20 model features a higher level of disaggregation across economies.

The first scenario reflects the overall Euro Area fiscal adjustment projected by the April 2011 World Economic Outlook. Staff projects the consolidation to be quite strong in 2011, on the order of 1.7 percent of GDP (of which 0.9 percent is the cyclically-adjusted component), followed by more moderate adjustments averaging 0.5 percent of GDP in subsequent years until 2016. In structural terms, the implied cumulative fiscal consolidation would amount to 2.4 percent of GDP by end-2016 (see text chart). Under the baseline scenario, the change in the structural balance is assumed to be permanent in the year it is implemented, but future changes in the structural balance are not anticipated and do not affect behavior until they actually occur. Further, the consolidation is prevented from having an impact on interest rates over the first five years in all regions.

Euro Area - Projected Fiscal Consolidation

(in percent of GDP)

Source: IMF Wodd Economic Outlook, April 2011.

Under the alternative scenario, the impact of the same consolidation plan is allowed to flow through to interest rates as easing concerns about debt sustainability in the Euro Area lower sovereign risk premia. Under this scenario, the planned fiscal adjustment is seen as sufficiently strong to bring about a gradual and permanent reduction in sovereign risk premia, as financial conditions and market confidence improve. Specifically, sovereign risk premia in the Euro Area are assumed to fall gradually by a cumulative 50 basis points by 2016.

The decline of investors’ concerns regarding sovereign risk issues in the Euro Area is assumed to spread to the rest of the world, mirrored by a cumulative reduction of other regions’ sovereign risk premia on the order of 20 basis point by 2016—roughly one third of what is assumed for the Euro Area. Both scenarios abstract from monetary policy accommodation in the Euro Area and in the rest of the world over the period 2011-16, in order to highlight the impact of fiscal policy and its spillovers. Of course, this assumption overstates the “true” contractionary effect on output that would actually take place, since monetary policy stabilization would moderate the impact.

According to model simulations, planned fiscal consolidation in the Euro Area will generate meaningful output losses in this region. In particular, under the baseline scenario where the consolidation does not bring down sovereign spreads, cumulative output losses over 2011-16 are found to range from 2.2 to 2.4 percent (see text chart below). The scenario that incorporates lower sovereign risk premia—both in the Euro Area and elsewhere—results in smaller cumulative output losses for the Euro Area over the same period, ranging from 0.9 to 0.8 percent. In other words, according to our model simulations, the generalized fall in risk premia partly offsets the Euro Area output losses arising from its planned fiscal adjustment (at least until end-2015).

Cumulative Effect of Euro Area Fiscal Consolidation on Output: 2011-2016

GIMF and G20 Models

Source: Staff calculations.

Model simulations envisage moderate spillovers to the rest of the world, which are negative in the absence of risk premia reductions and largely positive in the event that risk premia do fall. Assuming no risk premia reductions, model simulations reckon a cumulative output drop over 2011-16 in the rest of the world on the order of 0.5 percent—with regional output losses ranging from 0.1 to 1.1 percent. By contrast, growth spillovers from Euro Area fiscal consolidation plans would generally turn positive if the fiscal adjustment is successful in lowering global risk aversion and interest rates respond (e.g., bringing down risk premia worldwide). Under this scenario, estimated cumulative output gains are on the order of 0.2 percent, with region-specific output changes ranging from -0.3 to 0.5 percent. Differences in the magnitude of spillover effects across countries and country groups reflect trade linkages—those with greater trade openness and a higher share of trade with the Euro Area would tend to be affected more. The somewhat larger spillover effects under the GIMF model simulations (relative to those simulated using the G-20 model) are largely due to the strength of trade links, which are magnified by distinguishing between trade in final versus intermediate goods.

VIII. Spillovers from Higher Bank Capital Requirements in the Euro Area27

This note analyzes spillovers from regulatory increases in capital adequacy requirements in the Euro Area (EA). The novelty of this analysis is that it integrates and focuses on cross-border spillovers, and therefore serves as a complement to the MAG studies.28 The first section analyzes spillovers within the framework of a global model of bank balance sheets. The second section analyzes macroeconomic spillovers with the framework of a global macroeconometric model. In both sections, the focus is on direct potential spillovers only, abstracting from the potentially large positive impact of new higher capital requirements on underlying bank stability.

Bank Deleveraging Spillovers

This section presents simulations of bank deleveraging resulting from a tightening of capital adequacy regulations that lead to a large reduction in bank leverage. The simulations are performed in the model of bank deleveraging based on the simple behavioral assumption that banks maintain a target minimum leverage ratio by contracting their balance sheet (they “deleverage”), or by raising equity.29 The calibration relies on BIS consolidated banking statistics for the exposures of international banks to various countries.30 This reduction in leverage is assumed to stem from higher bank capital requirements and market pressures to reduce risk taking and leverage. To keep the exercise tractable, the risk composition of bank portfolios is assumed to remain constant, hence deleveraging is proportionally distributed across all asset classes. It is also assumed, for the purposes of the model, that the reduction in leverage does not lower funding costs.

We consider three scenarios under various amplification mechanisms and transmission to local affiliates. The main assumption, common to all scenarios, is that international banks experience a uniform 2 percentage point increase in their minimum capital to asset ratios (to reach a capital to asset ratio of 6 percent), resulting in a symmetric shock to all asset classes. Another common assumption is that banks are able to realize 80 percent of the required increase in the capital (given total assets) by increasing equity, and 20 percent by deleveraging. The baseline scenario assumes that there is full pass-through of the deleveraging to local affiliates of international banks. The second scenario adds moderate liquidity funding shocks to the baseline scenario under the assumption of 20 percent haircuts associated with fire sales. The third scenario modifies the baseline scenario by assuming only a 30 percent pass-through of the deleveraging to local affiliates, as opposed to 100 percent under the baseline.31

In the baseline scenario, the deleveraging is broad-based, and affects mostly large European financial institutions. The reduction in total assets is significant for French, German, Swiss and Swedish banks, where reductions equivalent to 6-8 percent of assets take place. This reflects the very high initial leverage of these banks. By contrast, Austrian, Canadian, U.S., Italian and Japanese banks do not deleverage significantly following the increase in required capital.

From a host country perspective, spillover effects are large mostly in advanced economies. The deleveraging exceeds 5 percent of GDP for a group of advanced economies (Belgium, Cyprus, Iceland, Ireland, and New Zealand), a few emerging markets (the three Baltic countries, Hong Kong SAR), and several small countries (see Table 1).

Table 1.Reduction in Foreign Liabilities to GDP
Reduction in Foreign Liabilities to GDPReduction in Foreign Liabilities to GDP
by euro area banksby non euro area banks
Countryscenario 1scenario 2scenario 3Countryscenario 1scenario 2scenario 3
Algeria0.31%0.31%0.17%Algeria0.00%0.00%0.00%
Argentina0.29%0.29%0.17%Argentina0.06%0.06%0.04%
Australia1.09%1.11%0.65%Australia0.55%0.57%0.35%
Austria2.57%2.62%1.70%Austria0.29%0.30%0.29%
Belgium6.36%6.44%4.54%Belgium0.46%0.47%0.43%
Bosnia and Herzegovina0.21%0.22%0.10%Bosnia and Herzegovina0.00%0.00%0.00%
Brazil0.57%0.58%0.29%Brazil0.21%0.22%0.14%
Bulgaria1.71%1.73%0.86%Bulgaria0.05%0.05%0.04%
Canada0.56%0.57%0.39%Canada0.28%0.29%0.20%
Chile1.78%1.81%1.03%Chile0.14%0.15%0.11%
China0.08%0.08%0.06%China0.07%0.07%0.04%
Colombia0.23%0.23%0.10%Colombia0.03%0.03%0.02%
Costa Rica0.16%0.16%0.16%Costa Rica0.04%0.04%0.03%
Croatia1.42%1.45%0.79%Croatia0.05%0.05%0.05%
Cyprus5.55%5.63%4.81%Cyprus1.90%1.94%1.49%
Czech Republic2.61%2.68%1.31%Czech Republic0.05%0.06%0.03%
Denmark1.71%1.73%1.50%Denmark3.17%3.21%1.35%
Dominican Republic0.22%0.23%0.21%Dominican Republic0.04%0.04%0.04%
Ecuador0.04%0.04%0.04%Ecuador0.04%0.04%0.03%
Egypt0.77%0.78%0.47%Egypt0.16%0.17%0.08%
El Salvador0.04%0.04%0.04%El Salvador0.02%0.02%0.02%
Estonia0.37%0.37%0.30%Estonia7.65%7.75%4.92%
Finland0.82%0.83%0.76%Finland3.53%3.60%1.43%
France0.83%0.84%0.74%France0.46%0.48%0.37%
Germany1.03%1.05%0.75%Germany0.52%0.53%0.39%
Greece2.33%2.37%1.74%Greece0.14%0.15%0.09%
Guatemala0.03%0.03%0.03%Guatemala0.07%0.07%0.07%
Hong Kong SAR1.75%1.77%0.99%Hong Kong SAR3.43%3.61%1.38%
Hungary2.97%3.04%1.71%Hungary0.05%0.05%0.05%
Iceland5.58%5.68%5.09%Iceland0.43%0.44%0.41%
India0.27%0.28%0.22%India0.19%0.20%0.12%
Indonesia0.18%0.18%0.15%Indonesia0.15%0.16%0.11%
Ireland8.30%8.48%7.46%Ireland2.67%2.78%1.79%
Israel0.27%0.27%0.20%Israel0.11%0.12%0.10%
Italy2.41%2.44%1.56%Italy0.13%0.14%0.11%
Jamaica0.12%0.12%0.12%Jamaica0.12%0.13%0.12%
Japan0.33%0.33%0.18%Japan0.17%0.18%0.11%
Jordan0.15%0.16%0.15%Jordan0.06%0.06%0.06%
Kazakhstan0.26%0.26%0.25%Kazakhstan0.14%0.14%0.11%
Korea0.55%0.56%0.41%Korea0.44%0.46%0.24%
Latvia0.85%0.87%0.51%Latvia4.26%4.32%1.84%
Lebanon0.24%0.25%0.24%Lebanon0.24%0.24%0.24%
Lithuania0.57%0.58%0.34%Lithuania3.17%3.21%2.36%
Luxembourg47.41%48.19%37.37%Luxembourg14.77%14.98%12.96%
Malaysia0.35%0.36%0.27%Malaysia0.48%0.50%0.23%
Malta4.03%4.10%3.84%Malta0.49%0.49%0.43%
Mexico0.69%0.70%0.35%Mexico0.14%0.15%0.08%
Morocco1.89%1.92%1.24%Morocco0.02%0.02%0.02%
Netherlands2.98%3.03%2.82%Netherlands0.88%0.91%0.81%
New Zealand0.86%0.87%0.45%New Zealand4.15%4.21%3.65%
Norway1.03%1.05%0.91%Norway2.67%2.72%1.18%
Pakistan0.04%0.04%0.04%Pakistan0.24%0.25%0.11%
Panama4.39%4.46%3.78%Panama3.53%3.55%3.48%
Peru0.44%0.45%0.23%Peru0.05%0.05%0.04%
Philippines0.31%0.32%0.29%Philippines0.17%0.18%0.13%
Poland2.18%2.22%1.05%Poland0.20%0.20%0.10%
Portugal4.16%4.23%3.23%Portugal0.34%0.36%0.25%
Romania1.39%1.40%0.64%Romania0.04%0.05%0.04%
Russia0.42%0.43%0.31%Russia0.11%0.11%0.08%
Serbia0.86%0.87%0.57%Serbia0.02%0.02%0.01%
Singapore2.56%2.60%1.57%Singapore1.94%2.01%0.92%
Slovakia1.53%1.57%0.73%Slovakia0.05%0.05%0.03%
Slovenia1.47%1.50%0.94%Slovenia0.04%0.04%0.04%
South Africa0.31%0.31%0.26%South Africa0.79%0.84%0.31%
Spain2.26%2.30%1.78%Spain0.26%0.28%0.19%
Sri Lanka0.11%0.11%0.11%Sri Lanka0.05%0.05%0.03%
Sweden1.09%1.11%1.03%Sweden0.92%0.96%0.61%
Switzerland2.01%2.05%1.81%Switzerland0.29%0.31%0.26%
Thailand0.26%0.26%0.21%Thailand0.25%0.26%0.12%
Tunisia0.90%0.91%0.56%Tunisia0.03%0.03%0.03%
Turkey0.78%0.79%0.56%Turkey0.13%0.13%0.09%
Ukraine0.93%0.95%0.55%Ukraine0.32%0.32%0.20%
United Kingdom3.91%3.98%2.51%United Kingdom0.96%0.97%0.54%
United States0.76%0.77%0.47%United States0.62%0.64%0.33%
Uruguay0.45%0.46%0.25%Uruguay0.09%0.09%0.08%
Vietnam0.35%0.36%0.28%Vietnam0.07%0.07%0.05%
Source: BIS, staff calculations.
Source: BIS, staff calculations.

Spillovers under this scenario take place mostly within the EA. Most of the deleveraging would be driven by EA banks and would, therefore, affect other EA countries (Figure 1). In absolute terms, the US and the UK would experience the second largest deleveraging, but the spillover would remain small in percent of their GDP. While small in absolute terms, the spillovers would be the largest for Central and Eastern European countries in percent of their GDP (Table). Spillovers to the rest of the world would remain small.

Figure 1.Deleveraging by Region and Home Country of Banks

Source: Staff calculations

Liquidity funding shocks (scenario 2) only marginally increase spillovers to other countries. With 20 percent haircuts on fire sales of assets, the amplification of deleveraging caused by the reduction in interbank funding remains small, but higher haircuts would have much larger effects, as the amplification effect is highly non-linear (see Table).

Partial transmission to local affiliates of international banks significantly reduces cross-country spillovers (scenario 3). If the shock is only partially transmitted to local affiliates, the reduction in foreign claims of international banks exceeds or is about 5 percent of GDP in a few countries.

Macroeconomic Spillovers

This section analyzes spillovers from a regulatory increase in capital adequacy requirements in the EA. The focus is on the transitional costs of higher capital requirements, as opposed to the permanent net benefits accruing from less frequent and severe banking crises. This analysis is based on scenarios simulated with an extended and refined version of the macroeconometric model of the world economy, disaggregated into its fifteen largest national economies, documented in Vitek (2009).32 These scenarios abstract from monetary policy stabilization and assume that the macroeconomic effects of this regulatory measure are transmitted exclusively via a permanent increase in the spread between commercial bank lending and deposit rates.

We estimate that a one percentage point increase in capital adequacy requirements in the EA would generate modest spillovers. A capital requirement shock is analogous to a permanent monetary policy shock, and is transmitted in the model via the interest and exchange rate channels of monetary policy. In the EA, we estimate a peak output loss of 0.30 to 0.32 percent, depending on the speed of implementation (Figure 2). These estimates are based on a 0.12 percent increase in the interest rate spread following MAG (2010), and are approximately linearly increasing in the capital adequacy requirement increase.33 In other advanced economies, estimated peak output losses range from 0.01 to 0.09 percent, while in emerging economies they range from 0.02 to 0.14 percent, reflecting their greater trade openness and less flexible exchange rate regimes. These output losses primarily reflect reduced export demand from the EA, mitigated by real effective currency depreciations in the rest of the world.

Figure 2.Peak Output Losses

Source: Staff calculations

IX. Spillovers from Structural Reforms34

This note analyzes spillovers from structural reforms in the Euro Area (EA). It is based on the exercise undertaken for the G-20 mutual assessment process. Simulations are performed using the IMF’s dynamic general equilibrium model, the Global Integrated Monetary and Fiscal Model (GIMF).35 Structural reforms covered comprise labor and product market reforms. The combined reforms are estimated to raise the steady state level of EA real GDP by almost 6 percent. Spillovers to the rest of the world depend heavily on the strength of trading relationships. Those to the United States and Japan, although nontrivial, are modest. Spillovers to emerging Asia and the “rest of world” block, however, are more substantial.36

Labor Market Reforms

The labor market reform scenarios are based on recent work by the Organization for Economic Co-operation and Development (OECD) 37. Four components of the OECD labor market reforms were included: an increase in active labor market policies (ALMP), a reduction of the average replacement rate (ARR), an increase in the standard retirement age by two years, and a move to actuarial neutrality for worker ages 60-65. In particular:

  • The ALMP reform increases ALMP spending per unemployed worker relative to GDP per capita to the average level prevailing in a group of high ALMP spending OECD countries (Denmark, Austria, Netherlands, Norway, Sweden, and Switzerland).
  • The ARR reform implies a cut in average replacement rates to the average prevailing in a group of low ARR spending OECD countries (Australia, Canada, Japan, New-Zealand, United Kingdom, and United States).
  • The move to actuarial neutrality implies that the implicit tax rate on continued employment is reduced to zero. The pension system is said to be “actuarially neutral” if the cost in terms of foregone pensions and contributions paid is exactly offset by an increase in future pension benefits. If this cost is not offset, there is an implicit tax on continued work.
  • The ALMP labor market reforms are implemented in 2011; whereas, the other reforms are implemented gradually starting in 2012. All reforms are perceived to be fully credible by 2012.

The fiscal impact of the labor reforms are based upon IMF staff estimates. The fiscal savings in pension outlays and ARR reforms constitute a reduction in government transfers. The ALMP reforms require an increase in government consumption expenditure to implement the programs. The increase in labor tax revenues that results from the higher employment induced by the reforms occurs endogenously in the GIMF simulation analysis. The overall impact on the fiscal balance of the ARR, ALMP, and pension reform are roughly neutral in the first few years. However, fiscal gains are realized gradually from a reduction in transfer outlays and higher labor income tax revenue.

The ALMP reform has the largest direct impact on real GDP in the first few years due to the increase in government consumption expenditure and the relatively quick response of household labor supply. The gains from the pension reforms are realized over several years as the effects on labor supply are more gradual. The implementation of the ARR reforms was delayed to mitigate the negative income effect in the first few years from the reduction in transfer outlays and the slow reaction of labor supply to this reform measure.

Product Market Reforms

Product market reforms result in multifactor productivity growth gains in both the tradable and nontradable sectors that are calibrated to match OECD estimates38. The reforms include countries moving to the best regulatory practices in upstream sectors observed in the average of the three most competitive countries in the OECD. The reforms are implemented in 2012, and the productivity growth gains are realized over a 10-year period. The OECD reforms are scaled depending on how much of the reforms have already been already implemented.

Spillovers of the Reforms onto the Rest of the World

The implementation of the labor and product market reforms, detailed above, increase productivity and employment permanently in the EA. This results in higher average household incomes and an increase in the demand for goods. Some of the higher demand spills over into imports and increases exports in all other regions. In the medium term, the spillovers are small as the rise in EA and world interest rates offset the boost from the trade channel. In the long run, however, the spillovers become more significant as the supply response in the EA kicks in and crowding-out effects fade. The effects, while modest in the US and Japan, are more substantial in emerging Asia and quite significant in the “rest of the world” block. This is likely due to the fact that neighboring countries (e.g. non-EA EU countries, Turkey and Russia) which have strong trade ties to the EA account for about 40 percent of the rest of the world block (in 2007 GDP in US dollar).

Real GDP impact from Euro Area structural reforms,

(percent deviation from steady state baseline)

Source: Staff calculations

The estimated spillover coefficients to the U.S. and Japan are broadly in line with earlier work assessing the impact of increasing euro area product and labor market competition to U.S. levels.39 That analysis concluded that such increase would raise long-run output in the EA by about 12 percent, while the impact on the rest of the world, calibrated using features of the U.S. economy, would amount to just under 1 percent. International spillovers arise in the model from the appreciation of the rest of the world’s terms of trade resulting from the EA’s increased output and exports, as well as higher EA propensity to import due to a shift in spending from consumption toward investment (which has a higher foreign component). Reduced goods markups—reflecting improved product market competition—account for around two thirds of the gains. The spillover coefficient, which may be measured as the estimated gain in the rest of the world relative to that in the EA, was found to be larger for cuts in goods markups than for wage markups because of the former’s relatively stronger output—and hence terms of trade—impact.

X. Spillovers from a Reduction in Tariff Protection in the Euro Area40

This note assesses the impact of tariff reductions in the Euro Area on trading partners’ exports. The tariff cuts considered for this exercise are based on current offers in the ongoing Doha round, but other countries’ tariffs are held fixed for this analysis in order to focus on the spillover effects of EA tariff policy. The results indicate that a 50 percent reduction in EA tariffs would raise global exports to the EA by more than 1 percent.

A partial-equilibrium model of international trade is used to assess the likely impact of a reduction in tariff protection in Euro Area (EA) countries on trading partners. The model consists of a large number of exporting countries. Each exporter decides how much to sell to the rest of the world based on the net-of-tariff price received by exporting to a particular importing region. In each importing region, the demand for imports from a given exporting country depends on the price of imports (inclusive of the tariff) from a particular exporter, as well as the tariff-inclusive price of imports from other, competing exporters. In equilibrium, the quantity of exports from a given country must equal the quantity of imports demanded by an importing region.

The model is used to simulate the impact of tariff reductions being considered under the Doha Round. Specifically, the model assesses the impact of a 50 percent reduction in trade-weighted average tariff rates applied by EA countries on exports and real GDP in 155 exporting countries.41 Tariff rates applied by importers in the rest of the world are assumed to remain unchanged.42

Figure 1.Vulnerability of Partner Exports to the Euro Area

(Exports to the Euro Area as Share of Total Exports)

Source: Staff calculations.

Preliminary estimates suggest that a 50 percent reduction in average tariff rates by EA countries would result in an increase in aggregate exports to the area by more than 1 percent, (Figure 2). Overall, export volume to the EA would increase the most for countries that currently face the highest tariff rates in the EA, such as China, Pakistan, Brazil, Vietnam, New Zealand, and Australia. The majority of countries would experience an increase in export volume to the EA. For these countries, export volume declines to the rest of the world, as tariff rates there remain unchanged, but aggregate export volume increases. Countries with relatively large export shares to the EA would experience the largest growth increase.

Figure 2.Impact of a Reduction in Protection in the European Union

(Percent change in export volume to the Euro Area)

Source: Staff calculations.

However, some countries would end up exporting less to the EA following such a unilateral tariff reduction because of preference erosion. The latter is due to the “Everything But Arms” (EBA) initiative of the European Union (EU), which allows exports from 49 developing economies to enter the EU duty free and quota free.43 As a consequence, if the EA countries were to reduce their most-favored nation (MFN) tariff rates, the margin of preference enjoyed by the EBA countries in the EU market would become smaller or “eroded.” In addition, export volume from 16 countries that are not eligible for preferences under the EBA would decline, as these countries have a preferential trade agreement with EU countries or face tariff rates that are below MFN rates.44 These negative impacts should not, however, be exaggerated as in practice nearly all EU trading partners get some sort of preference in the EU market, reducing the “effective” preference margins for many LICs. Also, due to the bureaucratic hurdles of complying with EU rules of origin, exporters are discouraged from claiming preferences on a substantial proportion of products. Finally, many LICs will benefit from the sharper cuts under Doha for items such as agricultural products and textiles.

1Prepared by Sergejs Saksanov and Francis Vitek.
2Smith, L.V. and A. Galesi (2010), GVAR Toolbox 1.0, http://www-cfap.ibs.cam.ac.uk/research/gvartoolbox/index.html.
3Japan’s spillovers to other countries might appear stronger than expected. This is likely due to the fact that the estimates are based on long time series, which cover periods in the 1980s and 1990s when Japan’s role was larger.
4Based on Vitek (2010).
5Prepared by Silvia Sgherri.
6The importance of risk commonalities varies over time, being greater at time of generalized stress in financial markets. On average, over the sample August 2007-February 2011, risk commonalities are found to explain about one-third of the total volatility for the portfolio considered.
7This could either be due to investors’ hedging strategies in the asset markets considered or to (opposite) fluctuations in the respective benchmark 10-year government bond yields.
8Prepared by Sally Chen, Mali Chivakul, Siret Dinc, Ola Melander, Mohamed Norat, and Malika Pant.
9While the note investigates spillovers from the EA as a whole, the effects of developments in the core and periphery EA are studied separately. Core EA, for the purposes of this analysis, comprises of Austria, Belgium, France, Germany, and the Netherlands, and EA program countries are Greece, Ireland, and Portugal. Both sovereigns and banks are studied for all G-20 countries except India, Indonesia, Mexico and South Africa (due to insufficient data for the sovereign for India and banks for others), and for Hungary (taken because of the availability of bank CDS data as representative of new EU member states in terms of their links—and hence as potential spillover recipients—with the Euro Area). Only one major bank from each country is used due to technical limits of the model. The EA banks are as follows: Austria - Erste, Belgium - Dexia, France - BNP Paribas, Germany - Deutsche, Greece - Alpha Bank, Ireland - Allied Irish, Netherlands - ING, Portugal - BCP, Other countries’ banks are as follows: Brazil - Itau Unibanco, China - Bank of China, Hungary - OTP Bank, India - State Bank of India, Japan - Nomura, Korea - Shinhan, Russia - Sberbank, Turkey - Akbank, UK -Barclays, US - Citigroup.
10A credit event could be a failure to pay on schedule, default, or, more broadly, a restructuring where bondholders are forced to bear losses.
11The CoPoDs are estimated as in Segoviano (2006), “Consistent Information Multivariate Density Optimizing Methodology,” Financial Markets Group, London School of Economics, Discussion Paper 557; Segoviano (2006), “The Conditional Probability of Default Methodology,” Financial Markets Group, London School of Economics, Discussion Paper 557; and Segoviano and Goodhart (2009), “Banking Stability Measures,” IMF Working Paper, WP/09/04.
12Transformation of CDS spreads to PoDs is done through a Matlab program which assumes a constant recovery rate of 40 percent. The transformation function is based on work on credit default swap such as O’Kane, D. and S. Turnbull. “Valuation of Credit Default Swaps.”Lehman Brothers, Fixed Income Quantitative Credit Research. April, 2003.
13Prepared by Thierry Tressel.
14Tressel, T, 2010, “Financial Contagion through Bank Deleveraging: Stylized Facts and Simulations Applied to the Financial Crisis”, IMF Working Paper 10/236.
15The simulations are based on Q3 of 2010 data (ultimate risk basis), including for sectoral bilateral exposures. However, when the sectoral data were not available (for example for French banks), the simulations were based on Q2 of 2010 data disclosed in the December 2010 BIS Quarterly Review. For German banks, Q3 of 2010 data on immediate risk basis were used; this method may overestimate German exposure to Ireland.
16Prepared by the GPM team, Research Department. The two scenarios and their results were published in WEO Update January 2011. http://www.imf.org/external/pubs/ft/weo/2011/update/01/index.htm
17In GPM:
  • Emerging Asia includes China, Hong Kong, India, Indonesia, South Korea, Malaysia, Philippines, Singapore, Taiwan and Thailand;
  • Latin America includes Brazil, Chile, Mexico, Columbia, Peru;
  • The remaining GPM countries include Argentina, Australia, Bulgaria, Canada, Denmark, Estonia, Israel, New Zealand, Norway, Russia, South Africa, Sweden, Switzerland, Turkey, UK and Venezuela.
18Prepared by Francis Vitek and Thierry Tressel.
19Vitek, F. (2010), Monetary policy analysis and forecasting in the Group of Twenty: A panel unobserved components approach, IMF Working Paper, 152.
20Tressel, T, 2010, “Financial Contagion through Bank Deleveraging: Stylized Facts and Simulations Applied to the Financial Crisis”, IMF Working Paper 10/236.
21Measures adopted by the ECB included fixed rate tenders with full allotment in all liquidity-providing operations; additional refinancing operations with one-month and three-month maturities, as well as the provision of funding at longer maturities of six months and one year, and a broadening of the collateral program.
22Refinancing by the ECB is backed by the provision of collateral, including sovereign debt securities and bank bonds. The ECB is also directly exposed to sovereigns of the periphery through the Securities Market Programme.
23Hence, the analysis differentiate banking systems according to their dependence on ECB refinancing and interbank financing. Interbank liabilities of the banking systems of the four countries are approximated by the total consolidated claims on domestic banks by banks of other nationalities (source: BIS Consolidated Banking Statistics, Q3 of 2010, and BIS Quarterly Review Dec 2010).
24More negative numbers mean larger deleveraging.
25Prepared by Silvia Sgherri, Francis Vitek, Derek Anderson, and Stephen Snudden.
26A comprehensive overview of the theoretical structure of GIMF is provided in Michael Kumhof, Douglas Laxton, Dirk Muir, and Susanna Mursula (2010), “The Global Integrated Monetary and Fiscal Model (GIMF) -Theoretical Structure,” International Monetary Fund Working Paper, 10/34. The structural macroeconometric model of the Group of Twenty is documented in Vitek, F. (2010), “Monetary policy analysis and forecasting in the Group of Twenty: A panel unobserved components approach,” International Monetary Fund Working Paper, 10/152.
27Prepared by Thierry Tressel and Francis Vitek.
28Basel Committee on Banking Supervision, “Results of the Comprehensive Quantitative Impact Study”, December 2010.
29See Tressel, T. (2010), Financial contagion through bank deleveraging: Stylized facts and simulations applied to the financial crisis, IMF Working Paper, WP/10/236.
30The simulations are based on 2010Q2 data.
31The pass-through to subsidiaries may be partial because they are significantly financed by local deposits.
32Vitek, F. (2009), Monetary policy analysis and forecasting in the world economy: A panel unobserved components approach, International Monetary Fund Working Paper, 09/238.
33Macroeconomic Assessment Group (2010), Assessing the macroeconomic impact of the transition to stronger capital and liquidity requirements, Financial Stability Board and Basel Committee on Banking Supervision Final Report.
34Prepared by Mali Chivakul and Stephen Snudden.
35See Kumhof and others (2010), “The Global Integrated Monetary and Fiscal Model (GIMF)-Theoretical Structure,” IMF Working Paper, 10/34.
36The model contains 5 blocks: the US, the EA, Japan, emerging Asia and the rest of the world. Emerging Asia block comprises China, Hong Kong S.A.R., India, Indonesia, the Republic of Korea, Malaysia, Philippines, Singapore, and Thailand. The rest of the world block comprises the remaining 167 countries in the world.
37The reform scenarios and country-by-country parameters are taken from Bouis and Duval (2011), “Raising Potential Growth After the Crisis: A Quantitative Assessment of the Potential Gains from Various Structural Reforms in the OECD Area and Beyond,” OECD Economics Department, Working Papers No. 835. The impact analyses, however, are simulated through GIMF.
38Bourlès and others (2010), “The Impact on Growth of Easing Regulations in Upstream Sectors”, CESifo Dice Report, Journal of International Comparisons, Vol. 8, No. 3.
39See Bayoumi and others (2004), “Benefits and Spillovers of Greater Competition in Europe: A Macroeconomic Assessment,” NBER Working Paper 10416.
40Prepared by Stephen Tokarick.
41This is the average reduction in EU tariffs used by Laborde, D., W. Martin, and D. van der Mensbrugghe, (2010), “Implications of the 2008 Doha Draft Agricultural and Non-agricultural Market Access Modalities for Developing Countries,” Washington: World Bank, to assess the likely impact of the December 2008 draft modalities being considered in the Doha Round.
42The 11 EU member countries that do not use the euro were excluded from this analysis due to likely conflicting effects on exports and GDP. On the negative side, these 11 EU countries would suffer from relative preference erosion under the proposed EU-wide tariff reductions. However they could benefit indirectly from their own reduced tariff rates.
43Countries eligible under the EBA are: Afghanistan, Angola, Bangladesh, Benin, Bhutan, Burkina Faso, Burundi, Cambodia, Cape Verde, Central African Republic, Chad, Comoros, Democratic Republic of the Congo, Djibouti, East Timor, Equatorial Guinea, Eritrea, Ethiopia, Gambia, Guinea, Guinea-Bissau, Haiti, Kiribati, Laos, Lesotho, Liberia, Madagascar, Malawi, Maldives, Mali, Mauritania, Mozambique, Nepal, Niger, Rwanda, Samoa, São Tomé and Principe, Senegal, Sierra Leone, Solomon Islands, Somalia, Sudan, Tanzania, Tuvalu, Togo, Uganda, Vanuatu, Yemen, and Zambia. Only 43 of these countries are included in the model, due to data limitations.
44Even though exports from a given country to the EA might decline, real GDP in that country could rise, depending on whether exports to other markets rise and the relative importance of exports to different destinations, as can be seen by comparing Figures 2 and 3.

Other Resources Citing This Publication