Assessing the Risks to the Japanese Government Bond - IMF

WP/11/292

Assessing the Risks to the
Japanese Government Bond (JGB) Market

Waikei Raphael Lam and Kiichi Tokuoka

© 2011 International Monetary Fund WP/11/292

IMF Working Paper

Asia and Pacific Department

Assessing the Risks to the Japanese Government Bond (JGB) Market

Prepared by Waikei Raphael Lam and Kiichi Tokuoka

Authorized for distribution by Kenneth Kang

December 2011

This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily
represent those of the IMF or IMF policy. Working Papers describe research in progress by the
author(s) and are published to elicit comments and to further debate.

Abstract
Despite the rise in public debt, Japanese Government Bond (JGB) yields have remained
low and stable, supported by steady inflows from the household and corporate sectors,
high domestic ownership of JGBs, and safe-haven flows from heightened sovereign risks
in Europe. Over time, however, the market’s capacity to absorb new debt will likely
shrink as population ages and risk appetite recovers. In the short term, a decline in fund
supply from the corporate sector, where financial surpluses are abnormally high, and
spillovers from global financial distress could push up JGB yields. Fiscal reforms to
reduce public debt more quickly and lengthen the maturity of government bonds will help
limit these risks.

JEL Classification Numbers: E62

Keywords: Fiscal Sustainability, Sovereign Risk, Government yields, Financial Distress

Author’s E-Mail Address: [email protected]; [email protected]

2

Table of Contents
I. Introduction....................................................................................................................3 
II. Risks to the JGB Market from Shrinking Fund Supply, Global Spillovers, and

Market Volatility............................................................................................................5 
III. Risks from a Rise in JGB Yields .................................................................................16 
IV. Conclusions..................................................................................................................17 
References................................................................................................................................18 

Tables
1.  Impact of Loans and Deposits on Banks’ Holdings of Government Securities ............8 
2.  Correlation of Japanese Sovereign Yields ...................................................................12 
3  Factors Influencing Short-term JGB Yield Movements ..............................................14 
4.  Financial Indicators Influencing Sovereign Risk.........................................................15 

Figures
1.  Overview of the JGB Market.........................................................................................4 
2.  Japan: Financial Balance Sheets of the Non-financial Sectors......................................5 
3.  Fund Supply and Demand of Non-financial Sectors .....................................................7 
4.  Global Spillovers and Volatility of the JGB Market ...................................................13 

3

I. INTRODUCTION

Despite the rise in gross public debt to over 200 percent of GDP, yields on JGBs have
remained low and stable. After the earthquake in March 2011, despite expectations of
additional JGBs to finance reconstruction, 10-year JGB yields have remained around
1.0–1.2 percent. Auctions since the earthquake have been met with steady demand across all
key maturities, from banks which continue to purchase short-term securities to life insurers
looking to lengthen the duration of their bond portfolios to match their liabilities (Figure 1).

Nevertheless, over the medium term, the market’s capacity to absorb new debt is likely
to diminish as the population ages and risk appetite recovers. Japan’s large pool of
domestic savings, stable investor base, and high share of domestic ownership of JGBs have
helped maintain stability in the JGB market. But these favorable factors are likely to diminish
over time as population aging reduces household saving and risk appetite recovers. Without a
significant policy adjustment, the stock of gross public debt could exceed household financial
assets in around 10 years, at which point domestic financing may become more difficult.1

In the near term, the JGB market also faces domestic and external risks. Domestically,
the supply of funds for financing JGBs could decline as private spending to repair the
earthquake’s damage picks up. An increase in market volatility could also push banks to
shorten the maturity of their JGB holdings or reduce their JGB exposures to limit losses.
Given the high correlation between yields on JGBs and other sovereign debts, a sudden rise
in global risk premia could spillover and affect the JGB market. All these factors could
eventually contribute to a sustained rise in yields, worsen the public debt dynamics, and pose
a risk to financial stability.

To assess the risks to the JGB market, this paper examines:

 What are the key risks to stability in the JGB market? What are the possible channels
through which a domestic shock or global financial distress could affect the JGB market?

 What would be the implications of sustained high interest rates for public debt dynamics
and financial stability? What should be the policy priority to mitigate the risks to the JGB
market?

1 See Tokuoka (2010).

4

Figure 1. Overview of the JGB Market

Despite the rise in public debt, JGB yields have declined Since the earthquake in March 2011, 10-year JGB yields
and remained low. have stayed stable between 1.0-1.2 percent…

Japan: General Government Debt and JGB Yields 10-year Sovereign Yields

250 8 (In percent)
7
Gross debt (LS, in pct of GDP) 6 5 3
5 U.S. Germany U.K. Japan (RS) 2.5
200 Net debt (LS, in pct of GDP) 1/ 4 2
3 4.5 1.5
150 10-year JGB yields (RS, period 2 4 1
average, in pct) 1 0.5
0 3.5 0
100 2010 3

50 2.5
2
0
1.5
1990 1995 2000 2005 1

Source: IMF WEO database. 0.5
0
1/ Net debt refers to gross debt minus gross financial assets of the general Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11

government. Source: Bloomberg.

…amid steady demand as shown by recent auction results. Corporate and households sectors have been recording
large financial surpluses, which have been invested
mainly in JGBs through banks…

Japan: Auction Results for 10-year JGBs Japan: Financial Surpluses

6 (In percent of GDP)
Bid/offer ratio 6
After the
Bid/offer ratio (12-month moving earthquake 15.0

4 average) 10.0
Yields (%) 4

5.0

2 2 0.0

1998 2000 2002 2004 2006 2008 2010

-5.0

0 0 -10.0 Private corporate sector Household sector
Source: Ministry of Finance.
2007/1 General government net lending
2008/1
2009/1 -15.0
2010/1
2011/1
2011/4
2011/6
2011/8
2011/10
2011/12

Source: Bank of Japan Flow of Funds statistics

…and the market has been supported by stable domestic …with a very low reliance on foreign financing.
players…

Japan: Shares of JGB Holdings at end-2010 G5 Economies:
Share of Foreign Holdings of Government Bonds
(In percent) Others Central bank
9% 8% (In percent)

Overseas 70 70
60
5% 50
40
Households 60 30
20
5% 10
0
50

National 40
pension fund
Banks 30
10% 39%

20

Insurance and 10
pension
24% 0 UK France US Germany
Japan
Source: Bank of Japan Flow of Funds statistics
Source: Debt Management Report 2010 (Ministry of Finance).

5

II. RISKS TO THE JGB MARKET FROM SHRINKING FUND SUPPLY, GLOBAL SPILLOVERS,
AND MARKET VOLATILITY

Decline in Fund Supply

In Japan, large savings by the corporate and household sectors have provided a steady
supply of funds to the JGB market. At the macro level, lending and borrowing by the non-
financial sectors, which consist of the general government, the private corporate sector, the
household sector, and private non-profit institutions, are mostly intermediated by the
financial sector.2 At end-2010, financial assets held by the private corporate and household
sectors stood at ¥2,275 trillion (450 percent of GDP), exceeding liabilities by ¥840 trillion
(about 170 percent of GDP) (Figure 2).3 This large domestic surplus has contributed to
financing nearly 95 percent of the stock of JGBs domestically.

Figure 2. Japan: Financial Balance Sheets of the Non-financial Sectors1/

(End-2010, in trillion yen)

3000

54 Private non-profit institutions

2500 General 471 18
government

2000 Private 786 844 752 General
1500 corporate 299 government:
sector JGBs and FBs

General
government:
other liabilities

1000 Household Financial 1068 Private
sector sector corporate
500 1488 sector

0 361 Household
Asset side sector
Liability side

Source: Bank of Japan Flow of Funds statistics
1/ General government does not include Fiscal Investment and Loan Program (FILP).

Over the past decade, a gradual increase in deposits and a trend decline in corporate
loans have provided additional space for investments in JGBs. Since 2000, household
deposits have increased by ¥40 trillion or 8 percent of GDP (Figure 3), supported by
declining but still positive household saving rates. During the same period, the stock of
corporate loans has declined by about ¥100 trillion (20 percent of GDP). These two factors
have led to a decline in the loan-deposit ratio from 95 to 70 percent (bottom right chart of

2 Indeed, financial assets and liabilities are almost balanced in the financial sector (excluding the Bank of Japan)
with financial assets exceeding liabilities by only 3 percent of GDP at end-2010.

3 These numbers are calculated on an unconsolidated basis. For example, JGBs held directly by households are
not subtracted from the assets or liabilities.

6

Figure 3) and created significant space for financial institutions to increase their JGB
holdings. In addition, the Bank of Japan has also stepped up purchases of JGBs (currently at
¥21.6 trillion yen per year on a gross basis) since the beginning of the global financial crisis
and acquired government securities through the new asset purchase program, which has
contributed to stable yields.4

Over the medium term, however, the Japan: Household Financial Assets and Gross Public Debt 1/

market’s capacity to absorb new debt is (In trillions of yen)
likely to diminish as the population ages. 2,100
On a stock basis, the household sector has 1,900
1,700
1,500

been financing more than half of JGBs either 1,300
directly or indirectly through banks and other
financial intermediaries, but since the 1990s, 1,100 HH fin assets, assuming 3.5% HH saving rate
population aging has reduced the financial 900 Gross Public Debt including FILP Liabilities 2/
surpluses of the household sector.5 Going 700 Gross Public Debt excluding FILP Liabilities
forward, population aging is likely to reduce 500
household surpluses further, and without a 300 2010 2015 2020 2025
2005

Sources: IMF WEO database, BoJ, and authors' calculations.
1/ Gross debt of the general government including and excluding liabilities owed by the
Fiscal Investment and Loan Program (FILP).
2/ For 2011 and on, FILP liabilities are assumed to stay at the same level as in 2010.

significant policy adjustment, the stock of gross public debt could exceed household financial

assets (currently at 300 percent of GDP) in around 10 years, suggesting that Japan may need

turn to other sectors, including overseas, to finance its deficit.

In the near term, fund supply to the JGB market from the corporate sector could also

decline. Corporate financial surpluses, which amounted to 7 percent of GDP in 2010, are an

important source of JGB funds through the banks. After the global financial crisis, these

surpluses rose sharply as corporates Japan: Net Increase in Holdings of JGBs and FBs during
postponed investment and capped wages. 2009-10 (In percent of 2010 GDP)

Looking ahead, these surpluses could decline 14.0 14.0
as corporates undertake investment for 12.0 12.0
reconstruction or expansion overseas. 10.0 10.0
8.0
8.0

At the same time, demand for JGBs from 6.0 6.0
pension funds could also weaken if pension
payouts accelerate. Amid population aging, 4.0 4.0
one of the largest institutional investors, the
National Pension Fund, has already begun 2.0 2.0
reducing assets to make payouts to retirees.
0.0 Banks Private Private National Overseas 0.0
Bank of insurance pension pension -2.0
-4.0
-2.0 Japan fund

-4.0

Source: Bank of Japan Flow of Funds Statistics

4 See Bank of Japan (2011), Berkmen (2011), Lam (2011), and Ueda (2011).
5 See Tokuoka (2010).

7

Figure 3. Fund Supply and Demand of Non-financial Sectors

Japan: Breakdown of Deposits (in trillion yen) Japan: Outstanding Loans and Banks' Government

Security Holdings (in trillion yen)

1400 1400 1400 Household loans
1200 1200
Household deposits Corporate deposits Other deposits

1200 Private corporate loans
Banks' government security (JGBs and FBs) holdings

1000 1000 1000

800 800 800

600 600 600

400 400 400

200 200 200

00 0
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Source: Bank of Japan Flow of Funds statistics Source: Bank of Japan Flow of Funds statistics

Japan: Private Corporate Financial Surpluses Japan: Loan-Deposit Ratio of Domestic Banks 1/

(In percent of GDP) (In percent)
100
14.0 Increase in currency and deposits 14.0 100
12.0 Decline in loans 12.0 95 95
10.0 Others 10.0 90 90
8.0 85 85
8.0 Financial surpluses 6.0 80 80
4.0 75 75
6.0 2.0 70 70
0.0 65 65
4.0 2010
-2.0 Source: Haver Analytics.
2.0 -4.0 1/ Does not include Japan Post Bank.

0.0 2002 2004 2006 2008 2000
2000 2001
2002
-2.0 2003
2004
2005
2006
2007
2008
2009
2010
2011

-4.0
Source: Bank of Japan Flow of Funds statistics

Estimating a basic demand function for government securities can help assess the
impact of a decline in corporate and household financial surpluses on banks’ JGB
holdings. Here we estimate the following equation:

govtsec = β1 loans + β2 deposits + β3 control variables,

where govtsec is banks’ holdings of central government securities (JGBs and FBs), 6 loans is
the stock of bank loans, and deposits is the sum of corporate and household sector deposits
(all in percent of GDP). Control variables include real GDP growth, spreads between long-
term prime lending rates and 10-year JGB yields, and CPI inflation.7 This equation estimates
how much govtsec would increase when loans decline and deposits rise. While there is an
endogeneity issue between govtsec and loans,8 Granger causality tests suggest that a decline
in loans leads to an increase in govtsec, and not in the opposite direction.9 Financial surpluses
of the corporate and household sectors channeled through the banking sector are observed
when loans decrease or deposits increase, or both. Thus, if financial surpluses of these

6 Excluding Japan Post Bank due to data constraints.

7 These variables are included to control for business cycles and risk appetite. Including other variables (e.g.,
equity returns) to control for risk appetite does not change the results much.

8 For example, banks may reduce loans to purchase government bonds.

9 Using first differences, the hypothesis that loans do not Granger-cause govtsec is rejected at the 1 percent level,
while the hypothesis of no Granger-causality in the reverse direction is not rejected at the 10 percent level.

8

sectors have a positive impact on banks’ holdings of central government securities, the
coefficients in the regressions should read as β1 < 0 and β2 > 0. We run regressions in level
form assuming cointegration (where estimates are robust to endogeneity).10

The results suggest that a decline in financial surpluses of the corporate and household
sectors could significantly reduce banks’ purchases of central government securities.
The estimates indicate that a 1 percent of GDP increase in loans would reduce banks’
holdings of central government securities by 0.3–0.6 percent of GDP, while a similar decline
in deposits would cut banks’ holdings of these securities by 0.5–0.9 percent of GDP
(Table 1). The last column of the table shows the results with the main independent variables
interacted with the post-Lehman dummy. These additional terms are statistically insignificant,
suggesting the effects of loans and deposits have not changed substantially before and after

Table 1. Impact of Loans and Deposits on Banks’ Holdings
of Government Securities 1/, 2/

Sample period: Q4 1997-
(quarterly data)

Dependent variable: banks’ (1) (2) (3) (4)

holdings of central government -0.42
(0.026)
securities
0.51
loans -0.56 - -0.35 (0.040)
(0.067) - (0.031) -0.12
(0.124)
deposits - 0.86 0.67
- (0.065) (0.044) 0.12
(0.086)
loans * post Lehman dummy - - -
- - -

deposits * post Lehman dummy - - -
- - -

Num of obs 53 53 53 53
R-squared 0.75 0.86 0.96 0.97

Source: Bank of Japan Flow of Funds Statistics; Haver; CEIC.

1/ Cointegration is assumed. Other regressors include a lag of quarterly growth (SA), spreads between long-
term prime lending rates and 10-year JGB yields, quarterly CPI inflation, and quarter dummies.

2/ Figures in parentheses indicate (robust) standard errors. Numbers in bold indicate a 5 percent level of
significance.

10 Unit root is not rejected for govtsec, deposits, or loans, but is rejected for the residual in the estimated
equation. This suggests that the three variables (govtsec, deposits, and loans) are cointegrated and that the
results are not subject to endogeneity bias.

9

the peak of the global financial crisis. The Japan: Household Financial Surpluses

estimates in the table in turn imply that if (net) (In percent of GDP) 14.0
repayment of loans and accumulation of deposits 14.0 12.0
12.0 Increase in currency and deposits 10.0
Securities (excluding government securities) and shares 8.0
of the corporate sector cease—as happened at the 10.0 Outward investment in securities 6.0
4.0
peak of the previous business cycle in 2007— 8.0 Other 2.0
Financial surpluses 0.0
2010
and corporate financial surpluses decline by 6.0 -2.0
-4.0
4 percent of GDP, banks’ net government 4.0

security purchases could fall by 1–3 percent of 2.0

GDP. This would be a sizeable reduction, 0.0 2000 2002 2004 2006 2008
compared to the annual net government debt 1998
issuances in recent years (10 percent of GDP).
-2.0

-4.0
Source: Bank of Japan Flow of Funds statistics

A decline in corporate financial surpluses does not necessarily lead to higher JGB yields.
This is because a decline in corporate financial surpluses is typically accompanied by a
recovery in domestic demand and higher tax revenue, which would reduce the need for debt
financing. This also occurred during Japan’s previous expansion between 2003 and 2007
(middle right chart of Figure 1), when corporate financial surpluses fell from nearly
10 percent of GDP in 2003 to zero in 2007 as business investment boomed. However, overall
fiscal deficits also declined by 5½ percent of GDP thanks to a cyclical tax recovery and
spending cuts, while the household sector maintained its financial surpluses. Consequently,
the JGB market experienced little funding pressure, with 10-year JGB yields staying below
2 percent even at the peak of the recovery.

There are three risks to such a “good” scenario in the current business cycle.

 The corporate sector might accelerate its Japan: Private Corporate Outward Financial Investment 4.0
overseas expansion. In recent years, 3.5
corporate outward direct investment (In percent of GDP) 3.0
(equity acquisition) has remained around 4.0 2.5
1 percent of GDP, but corporates are 2.0
increasingly looking overseas given the 3.5 outward security investment 1.5
strong yen and shrinking domestic 3.0

2.5
2.0 outward direct investment

1.5

market.11 A further shift in investment 1.0 2002 2004 2006 2008 1.0
overseas would reduce corporate 0.5 0.5
surpluses held in bank deposits to finance 0.0 0.0
JGBs.12 -0.5 2000 2010 -0.5
-1.0 -1.0

Source: Bank of Japan Flow of Funds statistics

 Projected declines in fiscal deficits might
not be enough to offset the impact of lower corporate financial surpluses. In particular,

even under the authorities’ plan, overall fiscal deficits would narrow only by 3 percent of

11 According to a recent survey by Teikoku Data Bank (July 2011), 25 percent of manufacturing companies
believe that oversea investment will accelerate. Overseas mergers and acquisitions activities by Japanese
corporations have already increased significantly to a record of 3 trillion yen (0.6 percent of GDP) in the first
six months of 2011.

12 Over time, oversea investments will contribute to corporate surpluses through repatriation of profits, but in
the short term, an increase in oversea investments is likely to result in net cash outflows.

10

GDP during the next 5–6 years,13 compared to 5½ percent of GDP during the previous
expansion period of 2003–07. This reflects more limited room for expenditure cuts than
in the past.

 A shift in households’ asset portfolio could also reduce demand for JGBs. For example,
the estimates in Table 1 imply that if households’ net purchases of securities (excluding
government securities) and shares bounce back to 2 percent of GDP (2007 level) as risk
appetite recovers, that could reduce banks’ purchases of government securities by
1½–2 percent of GDP through slower accumulation of deposits.

Based on historical trends, a decline in corporate and household financial surpluses
would likely have a modest impact on yields, but a more substantial response cannot be
ruled out. Japan’s historical data suggest that the immediate impact on yields from a decline
in corporate financial surpluses even to zero would be at most about 10 basis points.14
However, the response of yields to a funding shock could be nonlinear and more significant
once public debt exceeds a certain threshold.15

Market Volatility

Banks’ large and increasing holdings of Average Annual Return 4
JGBs are a key source of vulnerability. 2
Since mid-2008, banks have increased JGB (In percent) 0
-2
4 -4
2 -6
-8
holdings by ¥40 trillion or 8 percent of GDP 0 2005-10 2000-10 -10
amid a flight to safety and increasing private -2 2008-10 -12
-4 10-year JGB -14
-16
sector surpluses as discussed above. During -6 Avg stock return in Japan 1/ -18
this period, they have earned higher returns -8
-10 10-year US Treasury (in yen

from JGBs than from alternative investments -12 terms)
(for example, nominal returns from U.S. -14
-16

Treasuries in yen terms have been negative -18
due to the yen’s appreciation and narrowing
Sources: Japan Securities Research Institute; IMF WEO and IFS databases.
1/ Consisting of capital gains and dividend payouts.

interest differential). However, with banks’

(excluding Japan Post Bank) outstanding JGB holdings rising to ¥150 trillion (more than

15 percent of their total assets), they now face higher interest rate risk.

A rise in market volatility that prompts banks to unwind their JGB holdings could be
triggered in several ways.

 Shortening of maturities. Major banks have been shortening the maturities of their JGB
holdings to an average of about 2 years in FY2010 (from 3.2 years in 2002–03) in

13 For example, the Cabinet Office projected in August 2011 that assuming an increase in the consumption tax
rate to 10 percent by FY2015, the general government overall fiscal deficit (excluding the social security fund)
would narrow by only 3 percent of GDP between FY2010-15.
14 Estimated using regression results in Tokuoka (2010), which report that a decline in corporate or household
financial net worth of 1 percent of GDP would raise 10-year JGB yields by 1–2 basis points.

15 There is some empirical evidence consistent with the view that the impact of a rise in debt on yields is
nonlinear and becomes significant once the debt exceeds a certain threshold (e.g., Faini, 2006; Ardagna, Caselli,
and Lane, 2004).

11

response to higher interest rate risk. Higher interest rate volatility could further push
banks—particularly the regional banks, which hold longer maturity JGBs (about
3½ years)—to further shorten the maturity.

 Banks’ risk management practice. Higher interest rate volatility could induce a JGB sell-
off if banks’ risk exposures exceed the calculated thresholds of their risk management
model. A notable example was the so-called ‘VaR shock’ in June 2003, when 10-year
JGB yields more than tripled over three months, surging from a historically low of
0.5 percent to 1.6 percent (Figure 4).16 Although banks have now strengthened risk
management practices by including qualitative assessment in addition to the quantitative
risk measures in VaR models, banks’ JGB holdings are significantly larger, compared to
10 years ago.

 Rating downgrade. Although recent sovereign downgrades have had limited impact on
JGB yields, a further rating downgrade or a series of weak JGB auctions that push up
yields volatility could induce banks to reduce the duration of JGB holdings or to sell
JGBs to limit losses. This could in turn lead foreign investors to unwind their positions in
the futures and swaps markets.17

 Rollover risk. The rollover risks of JGBs Advanced Economies: Average Maturity of Public Debt 14
have risen along with the government’s 12
annual financing requirement, which now (In years) 10
amounts to about 55 percent of GDP 14 8
(including financing bills)—the highest 12 6
among advanced economies. The large 10 4
financing needs reflect not only the high 2
debt stock but also their relatively short 8 0
average maturity, which is still around 6
5½-6 years (including financing bills) 4
despite the recent lengthening.18 Given the 2
0

Source: IMF Fiscal Monitor (May 2010).

16 This episode was termed the “VaR shock” because the rise in volatility increased risk measures in banks’
internal value-at-risk (VaR) models and led to one-sided selling by banks as they attempted to shed risk (Bank
of Japan, 2010).

17 A large portion of JGB holdings are held in banks’ balance sheet as ‘available-for-sale’ or ‘held-to-maturity’
accounts outside the trading book. Banks would need to provide impairments for the valuation losses depending
on the magnitude of the losses (in practice if market value falls below 70 percent of the book value). In case
banks apply internal risk-based method and use zero risk weight on their JGB holdings, market investors could
price in the perceived increase in risk on bank valuation. A sustained rise in sovereign yields could pose interest
rate risks to banks’ balance sheets. Unexpected illiquidity in the JGB market and the uncertain prospect of fiscal
consolidation could also dampen their prices.

18 According to the authors’ estimates, the average maturity rose from 4.7 years at end-FY2005 (March 2006) to
5.8 years at end-FY2010.

12

large amount of bonds that need to be rolled over, any increase in uncertainty over the
supply and demand of JGBs could disrupt the smooth absorption of new issuances and
push up JGB yields.19

Global Spillovers

Global financial distress could have Correlation of JGB yields with Other Advanced Countries
negative spillover effects on the JGB
market through the banking system. (6-month rolling correlation on daily 10-year bond yields)
Japan’s sovereign yields are sensitive to 1.0 1.0

0.8 0.8

0.5 0.5

global risks, such as the investors’ risk 0.3 0.3

appetite. For example, the correlation between 0.0 France 0.0
10-year yields (returns) on JGBs and U.S. -0.3 Germany -0.3
Treasuries ranges from 0.37 to 0.58 (Table 2). -0.5 US -0.5
-0.8 Simple average correlation 1/ -0.8

In response to capital losses on their foreign -1.0 -1.0

bond portfolios, Japanese banks could reduce Jan-07
Jul-07
Jan-08
Jul-08
Jan-09
Jul-09
Jan-10
Jul-10
Jan-11
Jul-11

the maturity to minimize losses. For example, Source: Bloomberg.
in late 2010, the sudden rise in JGB yields 1/ Average pairwise correleations: Canada, France, Italy, Germany, UK, US.

mirrored those in U.S. Treasuries, as Japanese banks sold off some of their long-term JGB

holdings and shortened maturities in response to losses on their U.S. Treasuries. So far the

European turmoil has had limited impact on the JGB market. JGB yields—along with the U.S.

and Germany sovereigns—declined during the recent European sovereign distress due to ‘safe-

haven’ flows. However, if sovereign distress spreads more globally, that could also raise the

risk premium on JGBs.

Table 2. Correlation of Japanese Sovereign Yields

Correlation with 10-year JGB yields 1/ 10-year US 10-year German Average yields of
Treasury yields
bond yields adv. countries 2/

Entire sample: (Jan 2000 - May 2011) 0.58 0.37 0.49

Before Jan 2008 0.61 0.19 0.44

After Lehman crisis 0.62 0.87 0.75

Source: Bloomberg.
1/ Correlation coefficients refer to the correlation of 10-year JGB yields in levels and they are
all statistically significant at the 5 percent level.
2/ Average yields refer to the average of 10-year yields on U.S. Treasury, German sovereign bonds,
and U.K. Treasury bonds.

Another channel for global spillover could be through the derivatives markets where
foreign participation is high. Despite low foreign ownership of JGBs (5 percent of the total
outstanding), foreign investors are active in the JGB futures market, holding about one-third of
outstanding contracts.20 Compared to domestic players, foreign investors also appear to be
more sensitive to Japanese sovereign risk, as indicated by the rise in spreads on JGB CDS

19 For example, so called “FILP shock” took place in 1998 when yields spiked due to confusion over the
purchases of JGBs by the Fiscal Investment and Loan Program (FILP) Special Account.
20 Statistical analysis, however, does not point to a particular direction of causality.

13

contracts—traded mostly among foreign investors.21 Any distress sell-offs in the futures market
could affect the JGB cash market given the close arbitrage links. Overseas financial distress
could lead to a rise in global yields, which in turn could amplify pressures on JGB yields
through the derivatives markets (Figure 4).

Figure 4. Global Spillovers and Volatility of the JGB Market

Share of Foreign investors in the JGB Secondary Market 1/ 5-year Sovereign CDS Spreads

(in percent of total transactions ) in trillion yen (Basis points)
180
55 5000 180
Total turnover in JGB market (RHS) 4500
4000 160 Japan United States 160
50 Total turnover in JGB futures market (RHS) 3500
45 Share of foreign investors in JGB market (LHS) 3000 140 Germany United Kingdom 140
2500
Share of foreign investors in JGB futures market (LHS) 2000 120 120
40 1500
35 1000 100 100
30
25 80 80
20
15 60 60
10
40 40

5 500 20 20

00

2005 2006 2007 2008 2009 2010 0 0
Jun-08 Dec-08 Jun-09 Dec-11
Sources: Japan Securities Dealers Assocation and MoF. Source: Bloomberg. Dec-09 Jun-10 Dec-10 Jun-11

1/ Excluding repo transactions and Treasury bills. Numbers refer to annual turnover that includes both purchases

and sales of JGBs.

Japan: Sovereign Ratings by S&P Ten-Year Japanese Government Bond Yields and
(Local Currency; LT-Debt)
Historical Volatility
AA+
(2000M2) 2.5 Yield (percent) 60-day volatility (RHS) 0.3
0.25
2 0.2
0.15
AA AA 1.5 0.1
(2001M11) (2007M4) 0.05
1
AA- AA-
(2002M4) (2011M1) 0.5

0 0
Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan-
2000M1 2001M7 2003M1 2004M7 2006M1 2007M7 2009M1 2010M7 00 01 02 03 04 05 06 07 08 09 10 11
Source: Bloomberg.
Source: Bloomberg.

Estimating the sensitivity of the JGB yields to global risk factors can help assess the
potential impact from global spillover. We estimate the global spillover channel by using a
time-series regression and taking into account global factors and investors’ risk appetite.22
Granger-causality tests show that movements in global yields generally precede those of
JGBs, while the reverse causality from JGB yields to global yields does not appear to be
statistically significant.23 The results indicate that U.S. Treasury and German sovereign yields

21 Japan’s CDS market is not very liquid and consists mainly of foreign hedge funds. Foreign investors looking to
short JGBs typically acquire short positions on JGB futures, or buy out-of–the-money put options on interest rate
swaps.
22 The analysis uses daily data from 2005 with an ARIMA specification that accounts for the auto-regressive and
heteroscedastic features of short-term yield movement. Lagged variables are used as explanatory variables. An
ARIMA model applied as a statistical test on sovereign yields suggests that the time series are non-stationary. The
regression includes U.S. Treasury and German Bund yields, and the implied volatility of JGB yields as a proxy for
investor’s risk appetite. Other risk factors include exchange rate volatility and term premia. These risk factors in
essence capture both domestic and external risks.

23 The hypothesis that 10-year U.S. Treasury yields and 10-year German sovereign yields do not Granger-cause
10-year JGB yields are rejected with F-statistics equal to 51.7 and 35.4 (both p-values close to zero), indicating the
statistical significance at the 5 percent level. However, the reverse causality from 10-year JGB yields to 10-year

(continued…)

14

are significant at the 5 percent level (Table 3). These estimates imply, for example, that a one
percentage point increase in U.S. Treasury yields (or a change in global risk factors that raise
U.S. Treasury yields by one percentage point) could increase JGB yields by nearly 15 basis
points. Other things being held constant, JGB yields were more closely driven by U.S.
Treasury yields after the global financial crisis. In addition, uncertainty in the financial
markets, such as measured by the implied volatility of JGBs, also have a strong impact on
JGB yields. This would imply that a rise in global uncertainty, which is reflected in a higher
volatility of JGBs, could raise the risk premium in JGBs. Based on the estimates, a rise of
implied volatility, similar to what took place after the Lehman crisis, could push up JGB
yields by more than 40 basis points, holding other variables constant. 24

Table 3. Factors Influencing Short-term JGB Yield Movements 1/ 2/

Sample period: (1) (2) (3)
Jan 2006 - May 2011
Dependent variable: 10-year JGB
yields

10-year U.S. Treasury yields 0.16 0.13 0.12
(0.01) (0.01) (0.01)
10-year German sovereign bond
yields - 0.10 0.10
- (0.02) (0.02)
Implied volatility of JGBs 3/
0.05 0.05 0.05
Equity returns (Nikkei) (0.00) (0.01) (0.01)
-0.07 -0.09 -0.09
Term premium 3/ (0.03) (0.03) (0.03)

Dummy*U.S Treasury yields 4/ - 0.14 0.14
- (0.05) (0.05)
-
- - 0.01
- (0.00)

Source: Bloomberg
1/ All variables included in the regression refer to the first lags.
2/ Figures in parentheses indicate the standard errors. Numbers in bold indicate
a 5 percent level of significance.
3/ Implied volatility refers to 30-day implied volatility of 10-year JGBs as calculated
based on underlying options. Term premium refers to the slope between 2-year
and 5-year JGBs.
4/ The dummy variable spans from September 2008 to April 2009 to include
the peak of the global financial crisis.

U.S. Treasury yields or 10-year German yields is not statistically significant, with p-values close to 0.3 and 0.2,
respectively.
24 Alper and Forni (2011) suggest a notable spillover of government bond yields from advanced countries by as
much as 30 basis points on average across the advanced and emerging economies, after controlling for domestic
and global fundamentals.

15

Further regressions also suggest that market Relations between CDS vs Public Debt
risk to the JGB market is subject to global
factors. Sovereign risks as measured by the (CDS average over Jan-Oct 2011; public debt as of 2010)

600 Greece (165, 1332)
CDS (basis points)

CDS spreads are in general positively 500 Ireland
Portugal

correlated with the fiscal positions. Higher 400 y = 2.6065x - 10.182
public debt ratio as a percent of GDP is usually 300 R² = 0.1035
associated with higher sovereign CDS spreads. 200
In the case of Japan, the relations between the Spain

Italy

CDS spreads and fiscal variables are less clear 100 United Kingdom France Japan

than other advanced countries. Nevertheless, its Germany United States

0

sovereign CDS spreads are highly correlated to 0 50 100 150 200 250
Source: Bloomberg.
Public debt/GDP (%)

developments in global financial markets,

particularly in the United States and Europe. A rise in CDS spreads in the United States and

Europe, and lower global equity returns are found to be correlated with an increase in CDS

spreads in Japan at the 5 percent significance level. Specifically, a one-percentage rise in the

composite CDS spreads in advanced countries could raise Japan’s CDS spreads by 30 basis

points.

Table 4. Financial Indicators Influencing Sovereign Risk1/ 2/

Sample Period:

May 2009 - June 2011 (1) (2) (3)

Dependent variable: 5-year

Sovereign CDS for Japan

5-year sovereign CDS spreads 0.49 0.30 0.30
(advanced countries) 3/ (0.07) (0.09) (0.08)

Sovereign yields differential -5.39 -3.47 -3.57
between U.S. and Japan 3/ (1.39) (1.61) (1.63)

Global equity returns - -0.29 -0.28
Implied volatility of Nikkei 3/
LIBOR-OIS spreads - (0.08) (0.08)

- 0.14 0.13

- (0.03) (0.03)

- - 0.24

- - (0.12)

Source: Bloomberg
1/ All variables included in the regression refer to the first lags.

2/ Figures in parentheses indicate the standard errors. Numbers in bold indicate a 5
percent level of significance.

3/ Sovereign spreads for advanced countries refer to the composite sovereign CDS
spreads including the U.S., and the 10 largest European countries. The sovereign yields
differential is the 5-year U.S. Treasury net of 5-year JGB yields (presumably, this captures
domestic factors that may drive risk premium). Implied volatility is calculated on
underlying option prices.

16

III. RISKS FROM A RISE IN JGB YIELDS

A significant rise in yields would leave the Japan: Net Public Debt 1/
fiscal position extremely vulnerable. If
sovereign yields rise by 100 basis points over (In percent of GDP)
the next 5 years, the net debt-to-GDP ratio
would remain at high levels over the long 250 250
term, even after a 10 percentage points of 225
GDP adjustment in the structural fiscal 225 no adjustment + yield hike by 100 200
balance to anchor sustainability.25 The high basis points 2/
debt levels would leave the fiscal position
vulnerable to interest rate or funding shocks 200
and risk undermining public confidence.
175 175

150 150
10 pct points adjustment in the structural
125 125
balance 3/ + yield hike by 100 basis points 2/

100 100

75 75

50 50

2005 2010 2015 2020 2025 2030

Sources: Cabinet Office; IMF WEO database; and authors' estimations.
1/ Net debt of the general government including the social security fund.
2/ Yields hike by 100 basis points gradually over 5 years between 2011-16. As a result, nominal interest rate
growth differential is assumed to converge to 225 basis points over the long term (125 basis points (pre-
global financial crisis average since 2000) + 100 basis points).
3/ Assumes a 10 percent of GDP improvement of the structural primary balance between 2010-20.

Yield increases could also pose a risk to Interest rate risk in the Banking System
banks. With banks holding a large amount of
(100 basis point value; in percent of tier-1 capital)

JGBs (more than 15 percent of total assets, 40 Major banks Regional banks
excluding Japan Post Bank), a rise in yields 35
would generate capital losses. For example, 30
according to the BOJ, a 100 basis point 25
20

increase in interest rates across all maturities 15

rose the value of interest rate risk (including 10
from loans) by around ¥500 billion at the 5
major banks and about ¥400 billion at the 0

regional banks in FY2010. The total value of

interest rate risk (including from loans) Source: Bank of Japan.

corresponds to 10 percent of major banks’ tier 1 capital and more than 30 percent of regional

banks’ tier 1 capital, respectively.26

Potential spillovers from a sovereign stress in Japan would hit a large segment of the
Japanese financial sector. Using a probability-based distress model, we estimate the
potential impact arising from a hypothetical sovereign distress event in Japan on Japanese
financial institutions in general (not just banks).27 Compared to periods before the global
financial crisis, the impact of a sovereign distress on individual financial institutions has
significantly risen, partly driven by higher JGB holdings in the financial system and the
higher sensitivity of market investors to the linkages between sovereign risks and distress
among financial institutions.

25 See IMF (2011) for a list of possible measures to achieve a 10 percentage points adjustment.
26 The IMF’s Global Financial Stability Report (October 2010) and the Bank of Japan’s Financial System
Report (September 2010).

27 The probability-based distress model was developed by Segoviano (2006) and Segoviano and Goodhart
(2009) to analyze the interconnectedness and the common dependence on specific shocks. The analysis uses
daily data of the equity prices and CDS spreads of large Japanese financial institutions from November 2006 to
June 2011. Distress refers to the case when the CDS spreads of sovereign or individual financial institutions
exceed the tail 5 percent VaR threshold that is implied by the data.

17

IV. CONCLUSIONS

Stabilizing and reducing public debt is critical to maintaining confidence in the JGB
market as the factors holding down JGB yields could diminish over time. A decline in
fund supply, particularly from the corporate sector, higher market volatility, and spillovers
from global financial distress could put upward pressure on JGB yields. To limit these risks,
fiscal policy should aim to reduce public debt quickly. Given the limited scope for
expenditure cuts, fiscal adjustment should take a balanced approach that involves both
raising revenue and curbing expenditure growth. Lengthening maturities of JGBs would also
lock in low interest rates, while reducing roll-over risks.

18

REFERENCES

Alper, Emre, and Lorenzo Forni, 2011, “Public Debt in Advanced Economies and its
Spillover Effects on Long-term Yields,” IMF Working Paper No. 11/210 (Washington:
International Monetary Fund).

Ardagna, Silvia, Francesco Caselli, and Timothy Lane, 2004, “Fiscal Discipline and the Cost
of Public Debt Service: Some Estimates from OECD Countries,” NBER Working
Paper Series, No. 10788 (Cambridge, Massachusetts, National Bureau of Economic
Research).

Bank of Japan, 2010, Financial System Report, September.

Berkmen, S. Pelin, 2011, “Bank of Japan’s Quantitative and Comprehensive Easing: Are
They Now More Effective?,” forthcoming IMF Working Paper.

Bolton, Patrick and Olivier Jeanne, 2011, “Sovereign Default Risk and Bank Fragility in
Financially Integrated Economies,” NBER Working Paper No. 16899, March 2011,
(Cambridge, Massachusetts, National Bureau of Economic Research).

Faini, Riccardo, 2006, “Fiscal Policy and Interest Rates in Europe,” Economic Policy,
Vol. 21, No. 47, pp. 443–489.

International Monetary Fund, 2011, Japan: Staff Report for the 2011 Article IV Consultation,
IMF Country Report No.11/181 (Washington: International Monetary Fund).

International Monetary Fund, 2010, Global Financial Stability Report, October.

Kallestrup, Rene, David Lando, and Agatha Mugoci, 2011, “Financial Sector Linkages and
the Dynamics of Bank and Sovereign Credit Spreads,” mimeo, July.
http://forskerskolen.rente.nhh.no/portals/0/Files/lando.pdf

Lam, Raphael W., 2011, “Bank of Japan’s Monetary Easing Measures: Are they Powerful
and Comprehensive?,” IMF Working Paper 11/264 (Washington: International
Monetary Fund).

Segoviano, Miguel A., and Charles Goodhart, 2009, “Banking Stability Measures,” IMF
Working Paper No. 09/4 (Washington: International Monetary Fund).

Segoviano, Miguel A., 2006, “Portfolio Credit Risk and Macroeconomic Shocks:
Applications to Stress Testing Under Data-Restricted Environments,” IMF Working
Paper No. 06/283 (Washington: International Monetary Fund).

Tokuoka, Kiichi, 2010, “The Outlook for Financing Japan’s Public Debt,” IMF Working
Paper 10/19 (Washington: International Monetary Fund).

Ueda, Kauzo, 2011, “Japan's Deflation and the Bank of Japan's Experience with Non-
traditional Monetary Policy,” CARF F-Series CARF-F-235, Center for Advanced
Research in Finance, The University of Tokyo.