II. Japan and the Global Financial System: Spillovers and Systemic Linkages1
1. The global crisis has underscored the need to assess potentially systemic linkages across financial systems. The globalization and increased complexity of financial services has contributed to stronger linkages between international financial institutions. While these trends can contribute to economic growth by smoothing credit allocation and diversifying risk, they can also exacerbate the transmission of shocks. Indeed, the current crisis has demonstrated that systemic linkages can stem not just from solvency concerns, but also from liquidity squeezes and other stress events, and can quickly spillover across international financial systems.
2. This chapter uses two complementary approaches to assess risks from financial linkages between Japan’s financial system and the rest of the world. The analysis is focused on the banking system—which dominates financial intermediation in Japan—and on its linkages with other advanced economies, which account for the bulk of Japan’s overseas exposures. The first approach relies on network analysis, which tracks the spillover effects on banking systems of a credit event or liquidity squeeze through linkages in the interbank market. The second approach is based on the Co-Risk model, which uses market data to explore perceptions of direct and indirect linkages among financial institutions during stress events.
3. This work is part of an ongoing effort to further develop surveillance tools for cross-border analysis. The approaches used apply the work featured in Chapter 2 of the April 2009 GFSR. Such tools provide useful metrics for regulators to assess which systems and institutions could be a source of contagion, while highlighting potential spillover paths through the international financial system.
B. Cross-Border Financial Linkages in Japan: Recent Developments
4. Japan has played an active role in the integration of global financial markets. These connections are reflected in cross-border claims of Japanese banks, which have expanded significantly in recent years. While overseas exposures remain relatively low compared with U.S. and European banks, Japanese banks have been re-establishing international presence following the banking crisis of the late 1990s and early 2000s. Since mid-2002, Japanese banks’ cross-border claims have more than doubled from around $800 billion to nearly $2 trillion, with exposures to Western Europe accounting for more than 40 percent of the increase. Overall, advanced economies account for more than two-thirds of Japanese banks’ total overseas claims. By contrast, direct exposure to Eastern Europe—which has emerged as a potential flash point amid the current crisis—is very limited.
Consolidated Cross-Border Claims of Japanese Banks 1/
Source: BoJ, “Consolidated International Banking Statistics in Japan.”
1/ Consolidated cross-border claims in all currencies and local claims in non-local currencies of Japanese banks to each country and region. Asia/Pacific includes Hong Kong SAR and Singapore.
5. While also expanding in recent years, exposure to Asia remains relatively modest. Prior to the 1997–98 Asian crisis, Japanese exposures to Asia were similar to those of other regions. Subsequently, Japanese banks significantly cut back on their Asian operations in the wake of Japan’s banking crisis. Over the last six years, however, Japanese banks have also re-engaged with the rest of the region, with lending picking up sharply, in particular to Australia, China, India, New Zealand and Korea.2 Nevertheless, overseas claims of Japanese banks on the rest of Asia (including Hong Kong SAR and Singapore) are still only around one-third to one-quarter of those on Western Europe and the United States. As a result, risks of financial spillovers to and from the rest of the region appear low at present, and are likely to be much less significant than those associated with advanced economies outside Asia.
Japanese Banks’ External Credits to Asia 1/
Source: Bank of Japan.
1/ Figures in parenthesis are the shares in total claims at the end of 2008.
6. Cross-border bank exposures have fallen somewhat, over the last year reflecting factors such as unwinding yen carry trades and a retrenchment of Japanese banks’ lending abroad. Various proxies for overseas yen funding—including inter office accounts of foreign banks’ activities, foreign bank borrowing in Japanese money markets, foreign exchange margin trading and short-yen positions of non-commercial traders—suggest that carry trade positions have been unwinding as interest rate differentials have narrowed and exchange rate volatility has increased. Japanese banks’ overseas lending, which increased by nearly 20 percent (y/y) last June, has fallen significantly since the escalation of the global crisis, contracting by 3½ percent (y/y) in December.
Japan: Interoffice Accounts of Foreign Banks and Policy Rate Spreads
Sources: Bank of Japan and Haver Analytics.
1/ Simple average of US Fed Funds rate, ECB main refinancing operations: minimum bid rate, and RBA cash rate target minus BOJ call rate.
7. However, the strains in Japanese financial markets during the current global crisis highlight the importance of cross-border spillovers, and suggest the need to closely monitor cross-border linkages and vulnerabilities. While direct bank-to-bank linkages remain relatively low, Japan is closely integrated with global financial markets, as highlighted by the significant strains induced by the collapse of Lehman Brothers last Fall. These events highlight the need for a closer monitoring of Japanese cross-country exposures. In the same vein, regulators overseas should monitor risks that their financial systems may face from other economies, including Japan. Motivated by this need and the dominance of Western Europe and the United States in Japanese banks’ overseas exposure, the rest of this chapter examines some key questions that are likely to be of interest to policymakers and regulators both in and outside Japan, namely:
Assessment of spillover risks and systemic linkages. How vulnerable is the Japanese financial system to distress in overseas financial institutions, and vice versa?
Tracking impacts across economies. What are the potential contagion paths, including indirect effects?
Dynamics. How have these systemic linkages changed since the onset of the crisis, particularly as yen carry trades have been unwinding?
C. Network Analysis: Assessing Systemic Linkages using Banking Exposures3
8. Methodologies to assess potentially systemic institutions include those that rely primarily on institutional data, such as network analysis, and those that rely primarily on market data, such as the Co-Risk model which is briefly discussed in the next section. Network analysis relies primarily on institutional data. The basis for this analysis is the construction of a matrix of gross institutional exposures (domestic and cross-country). Researchers then track the network spillovers resulting from hypothetical credit events to specific institutions. Recent important extensions include the analysis of liquidity (or funding) events as in Chan-Lau et al. (2009 a, b). Most of the network literature has focused on the interbank credit market because interbank loans represent a large fraction of banks’ balance sheets in many countries (Upper, 2007).
9. Network analysis is a potentially powerful surveillance tool. Information extracted from the analysis of spillover effects can be used to produce and track several vulnerability indicators such as a hazard rate (or the number of times an institution would fail in the face of alternative stress events), the path and rounds of contagion resulting from alternative “trigger” institutions and the capital impairment (or the percentage of the pre-event capital an institution would lose once all the contagion rounds are accounted for). In this section, we perform some comparative network spillover analysis for Japan.
10. We use network analysis to track hypothetical network spillovers for the Japanese banking system arising from severe credit and liquidity events. The exercise relies on bilateral consolidated cross-border bank exposures for a sub-sample of sixteen BIS reporting countries for three dates: December 2007, March 2008 and December 2008.
11. Specifically, two scenarios are considered—a hypothetical severe credit shock and a combination of this credit plus a funding shock.
Severe credit event. The first scenario tracks the domino effects triggered by the individual default of each country’s cross-border interbank claims. It is assumed that a country’s banking sector losses are fully absorbed by its capital, and a country’s banking sector is said to fail when its aggregate capital is not sufficient to fully cover the losses induced by the default on its cross-border interbank claims.
Severe credit plus liquidity event. The second scenario consists of analyzing the case where the failure of a banking system also induces a liquidity squeeze, forcing a reduction in other banking sectors, balance sheets and, consequently, a fire sale of assets.
Although these hypothetical analyses may be extreme and highly unlikely, they help to illustrate systemic interconnections.
12. The Japanese banking system is one of the most resilient in the sample.
We find that:
As for most countries in the sample, the combination of a credit and liquidity events raises the vulnerability of the Japanese banking system.
However, Japan is much less exposed than other countries. For example, in March 2008, only a credit event triggered by the U.S., would have fully depleted the capital of the Japanese banking sector (Table II. 1). If this credit event induces a generalized liquidity squeeze, a default in both the U.S. and the U.K.’s cross-country interbank exposures would have completely depleted Japanese bank capital (Table II.2).
The magnitude of the impact of these shocks on the Japanese banking system is also relatively low compared to the other economies in our sample. In addition to the recent decrease in their gross exposures, this perhaps reflects an increase in diversification of Japanese banks’ activities globally (Table II.3).
Interestingly, when measured by capital impairment, the vulnerability of the Japanese banking sector to both shocks has generally decreased from its March 2008 peak.
Overall, network analysis suggests that the Japanese banking system is highly resilient and that risks to external stability arising from cross-border banking spillovers to Japan have fallen over the last year.
|Impairment in percent of initial capita|
|United States||Fully impaired||-84.3|
|Impairment in percent of initial capita|
|United Kingdom||Fully impaired||-49.5|
|United States||Fully impaired||-97.5|
|(Impairment in percent of initial capital)|
13. The current crisis has also highlighted the need to analyze the full implications of stress events, beyond the “point of impact” and including indirect effects. For example, a combined credit and liquidity event triggered by a hypothetical default on German banks’ cross-country interbank liabilities, would end up inducing a 33 percent capital loss for the Japanese banking sector after taking into account the full set of domino effects (Table II.4). This is significantly higher than the initial 9.2 percent loss induced “on impact.” Similarly, the trigger country may not necessarily inflict the highest degree of capital losses to Japan. In this example, France would have induced a higher degree of capital losses (9.7 percent of the original) for the Japanese banking sector.
|Trigger Country||Affected Countries||Contribution to Final Capital Impairment (in percent of initial capital)||Japan’s Final Capital Impairment (in percent of initial capital)|
|United Kingdom||United Kingdom||-12.0|
|United States||United States||-48.0|
14. Potential spillovers from Japan to the rest of the world also appear relatively limited at present. The failure of the Japanese banking system would not trigger the failure of any other banking system, but would lead to significant capital losses in Switzerland, France, Netherlands, Ireland, and Australia (Table II.5). By contrast, spillovers from Japan to the German, U.K. and U.S. banking systems are notably smaller than their impacts on the Japanese banking system, e.g., while the failure of the Japanese banking system would lead to a 16.5 percent decline in the capital of the German banking system, the impact of a hypothetical failure of the German banking system would impose a higher burden on Japan (depleting banking system capital by as much as 33 percent, as shown in Table II.2). Notably, while the spillover impact on the United States is seemingly modest and nowhere close to full impairment, the Japanese banking system has the fourth largest impact on the U.S. banking system, after the United Kingdom, France and Germany (Table II.3).
|impairment in percent of initial capital|
D. The Co-Risk Model: Market Perceptions of Spillovers21
15. Notwithstanding the usefulness of network analysis to assess potentially systemic financial linkages, it is important to supplement it with alternative methodologies. In addition to network interconnections through interbank exposures, there are likely to be other linkages between financial institutions arising from exposure to common risks factors, such as the adoption of similar business models (e.g., similar risk management systems or portfolio holdings), common accounting practices, market perceptions of financial institutions’ coincidence of fortunes, and so on. One method of extracting this information focuses on tracking market perceptions of how the credit risk of one institution affects other institutions’ credit risk (in particular, see Brunnermeier et al. (2009)).
16. A major appeal of these type of methodologies is their reliance on high frequency, publicly available data. The data includes institutions’ CDS spreads, Moody’s KMV EDFs, corporate bond spreads, distance-to-default measures or the VaR of their trading portfolio. Under efficient markets, co-movement of these variables should convey information about both direct and indirect linkages across financial institutions.22
17. Conditional Co-Risk estimates suggest that the largest Japanese financial institutions are more likely to be impacted by global financial institutions than the other way around. Drawing from CDS time series data on global financial institutions, bilateral conditional Co-Risk measures are estimated to assess which institutions were perceived by the market to be more connected to each other. Figure II.1 presents a graphical representation of some of the conditional co-risk estimates. The direction of the arrows represent the percentage increase in the CDS spread that stress in a ‘source’ institution would have induced on another institutions’ CDS spread.23 For example, as of September 2008, had Mitsubishi been under stress, it would have induced a 15 percent increase in Citigroup’s CDS spread. On the other hand, had Citigroup been under stress, Mitsubishi’s CDS spread would have increased by 43 percent. This pattern is broadly similar to that observed for the other two Japanese global financial institutions in our estimation. These results are consistent with those of the network analysis which suggest that spillovers arising from Japan onto major advanced economies—including Germany, the United States and the United Kingdom—are likely to be smaller than those affecting Japan.
Figure II.1.Japan: A Diagrammatic Depiction of Co-Risk Feedbacks between U.S., European, and Japanese Banks
Sources: Bloomberg, L.P.; Primark Datastream; and IMF staff estimates.
Note: This diagram presents the conditional Co-Risk estimates between pairs of selected financial institutions.
E. Conclusion and Future Research
18. The analysis suggests that risks from cross-border banking spillovers—both to and from Japan—are limited overall and have generally fallen over the past year. Network analysis suggests that, among advanced economies, the Japanese banking system is one of the most resilient to spillovers. Furthermore, the Co-Risk model suggests that Japanese banks are more likely to be impacted by market perceptions of changes in other institutions’ credit risk profile, than vice versa.
19. However vulnerabilities could increase, particularly as the crisis abates. Looking ahead, if risk appetite were to recover and yen carry trades resume, cross-border exposures could rise swiftly. Similarly, spillover risks could increase in the future as Japanese banks resume the process of re-establishing their international links, which had been progressing at a brisk pace in the years before the current crisis. Indeed, Japanese banks could be in a strong position to take advantage of the global recovery by investing in financial institutions and economies abroad, as suggested by the rise in their overseas lending prior to the Lehman collapse. Hence, prudent surveillance calls for continued careful assessment of potential vulnerabilities and spillover paths, including indirect effects.
20. Looking ahead, methods such as those discussed in the chapter could be tailored to help policymakers in Japan and other parts of the world to strengthen their multilateral surveillance efforts. The chapter makes the case for the development and the tracking of metrics to assess direct and indirect systemic linkages. It illustrates how the analysis could aid in identifying those institutions more likely to be systemic or vulnerable under alternative credit and liquidity events. The Bank of Japan conducts comprehensive and rigorous stress tests of the banking sector on a regular basis, and is actively considering incorporating analysis of potentially systemic linkages.24 To better track the systemic implications of stress events, more detailed information will be needed by global regulators, including off-balance sheet information and data on non-bank financial institution exposures. In addition, enhanced monitoring of systemic linkages in a globalized world is likely to require information-sharing agreements between countries on cross-market and cross-border linkages.
Prepared by Marco Espinosa-Vega, Juan Solé, and Murtaza Syed.
See Chapter IV of the 2006 Selected Issues paper for more details.
The results presented in this section draw on work by Jorge Chan-Lau in Chapter 2 of the April 2009 GFSR.
Although recent events and the relatively low liquidity of Japan’s CDS markets make the results more uncertain, it is important to keep in mind that the analysis was based on long time-series data covering the period before the severe financial dislocations induced by the Lehman collapse (July 1, 2003 to September 12, 2008).
An institution is said to be under stress if its CDS spread is above the 95th percentile of its distribution.
The results of these tests are featured biannually in the BoJ’s Financial System Report, and focus on: credit risk, for which the bank relies on a credit migration model using the internal data-sets of obligor-rating transition matrices of Japanese banks; market risk for which the BoJ simulates both expected valuation changes and the unexpected losses associated with equity holdings conditioned on the stressed macro-scenarios; and interest rate risk, for which the BoJ estimates the impact of yield curve shifts and twists on banks’ portfolios including bond, loan and deposit portfolios.
BossMichaelHelmutElsingerMartinSummer and StefanThurner2005 “Network Topology of the Interbank Market” Quantitative Finance Volume 4 pp. 677–684 (Austria: Department of Finance-Complex Systems Research Group).
BrunnermeierMarkusAndrewCrockettCharlesGoodhartAvinash D.Persaud and HyunShin2009. The Fundamental Principles of Financial Regulation. Geneva Reports on the World Economy 11. International Center for Monetary and Banking Studies.
Chan-LauJorgeEspinosa-VegaMarcoGieseckeKay and SoléJuan2009a “Assessing Cross-Border Financial Linkages: A Network Analysis” IMF working paper forthcoming.
Chan-LauJorgeEspinosa-VegaMarcoGieseckeKay and SoléJuan2009b “Co-Risk Measures to Assess Financial Systemic Linkages” IMF working paper forthcoming.
ElsingerHelmutAlfredLehar and MartinSummer2006 “Risk Assessment for Banking Systems” Management Science Vol. 52 no. 9 pp. 1301–14.
FurfineCraig H.2003 “Interbank Exposures: Quantifying the Risk of Contagion” Journal of Money Credit and Banking Vol. 35 no. 1.
International Monetary Fund2009Global Financial Stability Report Chapter 2April (Washington).
Márquez-Diez-CanedoJavier and SerafinMartinez-Jaramillo2007 “Systemic Risk: Stress Testing the Banking System” The International Conference on Computing in Economics and Finance.
MemmelChristoph andIngridStein2008 “Contagion in the German Interbank Market” (Frankfurt: Deutsche Bundesbank).
MüllerJeanette2006 “Interbank Credit Lines as a Channel of Contagion” Journal of Financial Services Research 29:1 37–60 2006 (Swiss National Bank).
UpperChristian2007 “Using Counterfactual Simulations to Assess the Danger of Contagion in Interbank Markets” BIS Working Paper No. 234 (Basel: Bank for International Settlements).
UpperChristian and AndreasWorms2004 “Estimating Bilateral Exposures in the German Interbank Market: Is There a Danger of Contagion?)” European Economic Review 48 (2004) pp. 827-849.