Effects of Corporate and Government Bond Purchases on Credit Spreads and Their Transmission Mechanism: The Case of Japan Kenji Suganuma* and Yoichi Ueno** November 2017 * Deputy Director and Economist, Bank of Japan (E-mail: kenji.suganuma@boj.or.jp) ** Director and Senior Economist, Bank of Japan (E-mail: youichi.ueno@boj.or.jp) The views expressed in this paper are those of the authors and do not necessarily reflect the official views of the Bank of Japan 1
Questions How and to what extent do the BOJ s corporate and government bond purchases push down Japanese firms credit spreads? What are the determinants of credit spreads in Japan? 2
Contribution 1. The first estimation of GZ spread and excess bond premium in Japan 2. Finding of the effectiveness of the excess bond premium for forecasting real economic activities 3. The first empirical investigation of local and global supply effect of the bond purchases in corporate bond market 4. Quantitative evaluation of the BOJ s QEs on credit spreads 3
Empirical strategy 1. Extend Gilchrist and Zakrajšek s framework (AER, 2012) by incorporating factors related to monetary policy 2. Examine the relationship between the BOJ s corporate and government bond purchases and the factors related to monetary policy 4
GZ framework Credit spread = specific factors + common factor Specific factors: 1. Default probability of issuer firm 2. Liquidity of bond 3. Expected recovery rate given a default 5
Credit spread specific factors 1. Default probability Schaefer and Strebulaev (JFE, 2008): Merton (1974) s distance to default 2. Liquidity Longstaff et al. (JF, 2005), Mahanti et al. (JFE, 2008): maturity (-), issue size (+), age (-), credit rating (+) 3. Expected recovery rate Gilchrist and Zakrajšek (AER, 2012): Industry specific 6
Credit spread common factor Excess bond premium: Effective measure to forecast real economic activity Tightly link to financial intermediaries risk appetite Frequently used with GZ spread in recent literature 7
GZ spread in Japan: GZ t 383 issuer firms and 5,614 corporate bonds from 1997 to 2016 GZ t = 1 N t CS it [k] = 1 N t (y it [k] y it f [k]) i k i k CS it [k]: Credit spread of corporate bond k issued by firm i at time t y it [k]: Yield of the corporate bond at time t y it f [k]: Government bond yield with the same coupon rate, face value, and maturity as the corporate bond at time t N t : # of bond/firm observations at time t 8
Apr-97 Apr-98 Apr-99 Apr-00 Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Apr-06 Apr-07 Apr-08 Apr-09 Apr-10 Apr-11 Apr-12 Apr-13 Apr-14 Apr-15 Apr-16 GZ spread in Japan (cont.) 0.5 0.4 %P GZ spread (left) Conventional credit spread index (BBB-AA, right) %P 7 6 5 0.3 4 0.2 3 0.1 2 1 0.0 0 9
GZ framework: OLS estimation ln(cs it [j]) = α 0 + α 1 DD it + α 2,l ln(x ilt [j]) l EBP t + α 3,m m + ε it [j]. D rating,imt + α 4,n 1 N ε it[j]. t i j n D industry,in 10
Apr-97 Apr-98 Apr-99 Apr-00 Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Apr-06 Apr-07 Apr-08 Apr-09 Apr-10 Apr-11 Apr-12 Apr-13 Apr-14 Apr-15 Apr-16 Excess bond premium in Japan: OLS estimates Distance to default Log of age Log of issue size Log of maturity AAA or AA A -0.16 *** (0.001) 0.19 *** (0.001) -0.04 *** (0.003) 0.86 *** (0.002) -0.82 *** (0.006) -0.76 *** (0.005) BB or the lower 0.93 *** (0.017) Industry dummy yes # of observation 360770 Adjusted R square 0.46 a 1.5 1.0 0.5 0.0-0.5-1.0 Notes: Sample period: 1997:4-2016:10. Asymptotic standard errors are in parentheses. Asterisks indicate statistical significance at the 1% level. 11
Effectiveness of EBP for forecasting -0.7-0.6-0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 dynamic cross-correlation Output gap EBP GZ BBB-AA -4-3 -2-1 0 +1 +2 +3 +4 +5 +6 +7 +8 leading to credit indexes lagging to credit indexes -0.7-0.6-0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 0.3 dynamic cross-correlation Capital input gap EBP GZ BBB-AA -4-3 -2-1 0 +1 +2 +3 +4 +5 +6 +7 +8 leading to credit indexes lagging to credit indexes Labor input gap -0.7 dynamic cross-correlation -0.6-0.5-0.4-0.3-0.2-0.1 0.0 EBP GZ BBB-AA -4-3 -2-1 0 +1 +2 +3 +4 +5 +6 +7 +8 leading to credit indexes lagging to credit indexes 12
Effectiveness of EBP for forecasting (cont.) -0.5-0.4-0.3-0.2-0.1 0.0 CI dynamic cross-correlation EBP 0.1 0.2 GZ BBB-AA 0.3-12 -9-6 -3 0 +3 +6 +9 +12 +15 +18 +21 +24 leading to credit indexes lagging to credit indexes -0.5-0.4-0.3-0.2-0.1 IIP dyanamic cross-correlation 0.0 0.1 0.2 EBP GZ BBB-AA 0.3-12 -9-6 -3 0 +3 +6 +9 +12 +15 +18 +21 +24 leading to credit indexes lagging to credit indexes Unemployment rate 0.5 dyanamic cross-correlation 0.4 0.3 0.2 0.1 0.0-0.1-0.2 EBP GZ BBB-AA -0.3-12 -9-6 -3 0 +3 +6 +9 +12 +15 +18 +21 +24 leading to credit indexes lagging to credit indexes 13
Effectiveness of EBP for forecasting (cont.) p h Y t+h = α + β i Y t i + γ 1 TS t + γ 2 RPR t + γ 3 Z t + ε t+h i=1 h Y t+h ln(y t+h ) ln(y t 1 ) Y t ln(y t ) ln(y t 1 ) Y t : Real economic variable TS t : Term spread RPR t : Real interest rate Z t : Excess bond premium, GZ spread, or conventional credit spread index 14
Effectiveness of EBP for forecasting (cont.): OLS estimates Dependent variabl: Composite index (coincident) in indexes of business conditions Three month ahead forecast Term spread -0.008-0.004-0.007-0.002 0.018 * 0.000 (0.011) (0.010) (0.011) (0.010) (0.009) (0.008) Real interest rate 0.005 0.004 0.006 0.005 0.000 0.007 (0.004) (0.004) (0.004) (0.004) (0.003) (0.004) BBB-AA spread - 0.006-0.007 - - (0.005) (0.005) GZ spread - - -0.024-0.058 - - (0.036) (0.044) Predicted GZ spread - - - - - 0.365 (0.260) Excess bond premium - - - - -0.031 *** -0.022 ** (0.007) (0.009) Adjusted R square 0.203 0.213 0.201 0.216 0.332 0.230 Notes: Sample period: 1997:8 2016:6. Each specification includes a constant and p lags of Y t i (not reported), where p is determined by the AIC. HAC consistent standard errors are in parentheses. Asterisks *, **, and *** respectively indicate statistical significance at the 10%, 5%, and 1% level. 15
Effectiveness of EBP for forecasting (cont.): OLS estimates Dependent variabl: Composite index (coincident) in indexes of business conditions Twelve month ahead forecast Term spread 0.039 0.034 0.032 0.035 0.104 *** 0.062 (0.048) (0.049) (0.045) (0.048) (0.039) (0.047) Real interest rate -0.009-0.004-0.007-0.003-0.027 ** -0.009 (0.015) (0.013) (0.015) (0.014) (0.013) (0.015) BBB-AA spread - 0.035 ** - 0.036 * - - (0.018) (0.018) GZ spread - - 0.046-0.031 - - (0.127) (0.125) Predicted GZ spread - - - - - 1.491 * (0.824) Excess bond premium - - - - -0.073 *** -0.056 ** (0.022) (0.026) Adjusted R square 0.054 0.103 0.046 0.099 0.493 0.107 Notes: Sample period: 1997:8 2016:6. Each specification includes a constant and p lags of Y t i (not reported), where p is determined by the AIC. HAC consistent standard errors are in parentheses. Asterisks *, **, and *** respectively indicate statistical significance at the 10%, 5%, and 1% level. 16
Channels from QEs to credit spreads 1. Default risk channel Krishnamurthy and Vissing-Jorgensen (BPEA, 2011) 2. Local and global supply channels D Amico and King (JFE, 2013), Greenwood and Vayanos (RFS, 2014), Greenwood, Hanson, and Liao (2017) 3. Risk taking channel Adrian and Shin (HME, 2011), He and Krishnamurthy (AER, 2013) 17
BOJ s corporate bond outright purchase program Amount to be purchased (maximum) Criteria Credit Rating Maturity 19-Feb-09 one trillion yen A or higher within one year 31-Dec-09 28-Oct-10 0.5 trillion yen abolishment 14-Mar-11 2 trillion yen one to two year 4-Aug-11 BBB or higher 2.9 trillion yen 27-Apr-12 30-Oct-12 3.2 trillion yen one to three year 18
Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 BOJ s stock of corporate and government bonds 3.5 3.0 2.5 2.0 1.5 1.0 trillion yen Corporate bond (left) Government bond (right) trillion yen 400 350 300 250 200 150 100 0.5 50 0.0 0 19
Default risk channel QEs Stock price = Distance to default Credit spread Extraction of a common factor in firms distances to default DD it = α + δ m m D rating,imt + θ n n D industry,in + d t D time,t + ε it t 20
Default risk channel (cont.): OLS estimates Dependent variable: Distance to default 2.69 *** AAA or AA dummy (0.01) 1.03 *** A dummy (0.01) BB or the lower -2.17 *** dummy (0.03) Industry dummy yes Time dummy yes # of observation 361180 Adjusted R square 0.56 Notes: Sample period: 1997:4-2016:10. Asymptotic standard errors are in parentheses. Asterisks indicate statistical significance at the 1% level. 21
Apr-97 Apr-98 Apr-99 Apr-00 Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Apr-06 Apr-07 Apr-08 Apr-09 Apr-10 Apr-11 Apr-12 Apr-13 Apr-14 Apr-15 Apr-16 Default risk channel (cont.) d t 4 3 2 1 0-1 -2-3 -4 22
Default risk channel (cont.): OLS estimates h d t+h = α + β i d t i + Term spread Real interest rate Log of the BOJ's stock of JGBs and CBs ( change over the last three months) p i=1 γ 1 TS t + γ 2 RPR t + γ 3 Z t + ε t+h Three month ahead Six month ahead 0.63 ** 1.50 *** (0.31) (0.49) -0.03-0.07 (0.10) (0.15) 5.80 *** 12.88 *** (2.09) (3.01) Adjusted R square 0.18 0.31 Notes: Sample period: 1997:8 2016:6. Each specification includes a constant and p lags of Y t i (not reported), where p is determined by the AIC. HAC consistent standard errors are in parentheses. Asterisks **, and *** respectively indicate statistical significance at the 5% and 1% level. 23
Local and global supply channel: Within CB market Decline of CB supply Reduction of credit spread Effective range of bond purchase depends on arbitragers or financial intermediaries risk appetite. No QE effect No QE effect Local supply effect Global supply effect eligible maturity maturity 24
Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Local and global supply channel: Across JGB and CB market Relative supply of CB to JGB Credit spread 35 % 30 25 20 15 10 5 25
Local supply channel: Clue from traded bond data AA or higher credit rating corporate bond yields 0.8 % QQE QQE with NIRP QQE with YCC (Nov-15 to Jan-16) (Feb- to Oct-16) (Nov-16 to Jan-17) 0.8 % 0.8 % 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0-0.2-0.2-0.2-0.4 0 1 2 3 4 5 maturity -0.4 0 1 2 3 4 5 maturity -0.4 0 1 2 3 4 5 maturity Note: Traded bond data is available only since Nov-15. 26
Local supply channel: Clue from our estimation results Distribution of idiosyncratic excess bond premium ε it [j] EBP t Before the BOJ's program After the BOJ's program 0.70 0.70 0.60 Within 0.60 Within 0.50 0.40 Outside 0.50 0.40 Outside 0.30 0.30 0.20 0.20 0.10 0.10 0.00-3.5-3.0-2.5-2.0-1.5-1.0-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.00-3.5-3.0-2.5-2.0-1.5-1.0-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 27
FIs risk appetite Two proxy variables for financial intermediaries risk appetite 1. Lending attitude of financial institutions in Tankan Survey to firms about the level of FIs risk appetite Used in estimation after controlling fund demand side condition D.I. such as business condition and financial position D.I. 2. D.I. for credit standards in Senior Loan Officer Opinion Survey on Bank Lending Practices at Large Japanese Banks ("Loan Survey") Survey to banks about the change of their risk appetite Used in estimation after accumulating and adjusting the sample mean 28
FIs risk appetite and EBP Tight link between FIs risk appetite and EBP 2.0-30 0.4-5 Excess bond premium (level, left) 1.5 Tankan (regression residual, right) 1.0 Tight 0.5 0.0-0.5-1.0 1997 99 01 03 05 07 09 11 13 15-25 -20-15 -10-5 0 5 10 0.3 Tightened 0.2 0.1 0.0-0.1-0.2-0.3 Excess bond premium (dif., left) -0.4 Loan survey (D.I. for credit standards, right) -0.5 1997 99 01 03 05 07 09 11 13 15 0 5 10 15 20 25 Correlation:-0.29 Correlation:-0.24 29
Local and global supply channel: OLS estimates eligible dummy eligible dummy FIs' risk appetite eligible dummy log of amount of the BOJ's stock of CBs eligible dummy log of amount of the BOJ's stock of CBs FIs' Ratio of CBs supply to JGBs supply Ratio of CBs supply to JGBs a b c d e - -0.56 *** - - - (0.005) - - - -0.046 *** -0.164 *** (0.003) (0.006) - - - -0.019 *** -0.027 *** (0.001) (0.000) - - - 0.003 *** 0.010 *** (0.000) (0.000) - - 5.611 *** 3.148 *** 4.123 *** (0.023) (0.029) (0.024) - - - -0.011 *** -0.194 *** (0.000) (0.002) supply FIs' risk appetite Credit rating dummy yes yes yes yes yes Industry dummy yes yes yes yes yes # of observation 360770 360770 357024 328770 357024 Adjusted R square 0.46 0.48 0.55 0.53 0.57 Notes: Sample period for a and b is 1997:4-2016:10, that for c and e is 1998:1-2016:10, and that for d is 2000:2-2016:10. Asymptotic standard errors are in parentheses. Asterisks *** indicate statistical significance at the 1% level. 30
Risk taking channel: Identification BOJ s JGB and CB purchases mitigating FIs capital restriction increase of FIs risk appetite reduction of credit spreads in particular of lower-grade corporate bonds 31
Risk taking channel (cont.): OLS estimates eligible dummy FIs' risk appetite eligible dummy log of amount of the BOJ's stock of CBs eligible dummy log of amount of BOJ's stock of CBs FIs' risk appetite FIs' risk appetite Ratio of CBs supply to JGBs supply Ratio of CBs supply to JGBs supply FIs' risk appetite AAA or AA dummy FIs' risk appetite A dummy FIs' risk appetite BB or the lower dummy FIs' risk appetite f g - -0.049 *** -0.158 *** (0.003) (0.006) - -0.021 *** -0.022 *** (0.001) (0.000) - 0.003 *** 0.011 *** (0.000) (0.000) - -0.005 *** -0.024 *** (0.000) (0.001) - 2.847 *** 4.599 *** (0.034) (0.029) - -0.001-0.073 *** (0.001) (0.004) - 0.005 *** 0.005 *** (0.000) (0.001) - 0.002 *** -0.020 *** (0.000) (0.001) - -0.003 *** -0.027 *** (0.000) (0.004) Credit rating dummy yes yes yes Industry dummy yes yes yes # of observation 360770 328770 357024 Adjusted R square 0.46 0.54 0.58 a 32
Risk taking channel (cont.): OLS estimates p h RA t+h = α + β i RA t i + γ 1 TS t + γ 2 RPR t + γ 3 SP t + γ 4 Z t + ε t+h i=1 Loan Survey Tankan 1Q ahead 2Q ahead 1Q ahead 2Q ahead Term spread Real interest rate 13.99 *** -1.28 21.50 *** -2.61 * 3.45 *** 0.06 3.92 0.23 (4.37) (0.77) (5.85) (1.40) (3.43) (0.83) (2.89) (0.84) TOPIX 8.38 17.38 6.43 11.88 ( Quarterly log Diff. ) (7.09) (13.19) (6.45) (7.11) BOJ holdings of JGB and CB 78.71 *** 112.27 *** 23.37 *** 29.23 ** ( Quarterly log Diff. ) (20.71) (35.66) (12.76) (11.54) Adjusted R square 0.66 0.65 0.07 0.19 Notes: Sample period: 2001:1 2016:4. Each specification includes a constant and p lags of Y t i (not reported), where p is determined by the AIC. HAC consistent standard errors are in parentheses. Asterisks *, **, and *** respectively indicate statistical significance at the 10%, 5% and 1% level. 33
Risk taking channel (cont.): Quantitative evaluation An increase of ten percent in the BOJ s stock of JGBs and CBs Loan Survey Tankan 1Q ahead 2Q ahead 1Q ahead 2Q ahead AAA or AA 0.3% 0.5% -4.3% -5.3% A -1.8% -2.6% -10.2% -12.7% BBB -3.7% -5.3% -5.5% -6.9% BB or the lower -5.9% -8.4% -12.0% -15.0% 34
Comprehensive assessment An increase of ten percent in the BOJ s stock of JGBs and CBs Loan Survey Tankan 1Q ahead 2Q ahead 1Q ahead 2Q ahead Default risk channel -7.0% -15.6% -7.0% -15.4% Risk taking channel -1.5% -2.1% -8.1% -10.2% Local and global supply channel Within the BOJ's criteria -0.6% -0.5% 1.5% 1.1% Outside the BOJ's criteria 2.1% 3.4% 1.9% 1.5% Total Within the BOJ's criteria -9.1% -18.1% -13.6% -24.5% Outside the BOJ's criteria -6.4% -14.3% -13.1% -24.1% 35
Contribution of this paper 1. The first estimation of GZ spread and excess bond premium in Japan 2. Finding of the effectiveness of the excess bond premium for forecasting real economic activities 3. The first empirical investigation of local and global supply effect of the bond purchases in corporate bond market 4. Quantitative evaluation of the BOJ s QEs on credit spreads 36