## sipea2023032

## Source details

**Canonical URL:** [sipea2023032](https://www.imf.org/-/media/files/publications/selected-issues-papers/2023/english/sipea2023032.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/selected-issues-papers/2023/english/sipea2023032.pdf.md)
- [Structured JSON version](/-/media/files/publications/selected-issues-papers/2023/english/sipea2023032.pdf.json)

---

### Introduction
- BoJ unconventional monetary policies (UMP) since 2013 include Yield Curve Control (YCC) policy that caps the 10-year rates at 0 percent within a narrow band.
- Headline inflation has been above the central bank’s target of 2 percent y/y since April 2022.
- BoJ widened the target band for 10-year JGB yields in December 2022 to address deteriorating long-term JGB market functioning.
- Objective: characterize the JGB yield curve (including slope) using a state-space model and estimate how changes in the slope affect bank profitability using fixed-effects panel estimation; examine lagged effects, heterogeneity across banks, and cross-border spillovers.

### Data and Characterization of the JGB Yield Curve
- Data:
  - Constant-maturity JGB yields with maturities from 3 months to 30 years.
  - Sample period for yields: January 2000 till October 2022.
  - Bank-level annual unconsolidated financial statements: 95 banks, sample period 2003-2021 (Fitch-Bankscope).
  - Macroeconomic variables from IMF World Economic Outlook; NIKKEI volatility from Thomson Reuters.
- State-space specification:
  - Nelson-Siegel representation with three latent factors: L_t (level), S_t (slope), C_t (curvature).
  - Parameter λ governs decay factor on factor loadings and is estimated jointly with latent factors.
  - Estimation via maximizing Gaussian likelihood with Kalman smoothing.
- Key characterization and model fit:
  - Level factor recovered from historically low levels in 2016 and increases further since end-2021.
  - Slope factor confirms sustained flattening over the past decade and recent steepening, mainly for maturities beyond 10 years given YCC.
  - Curvature has been negative over the sample and increased sharply with the introduction of YCC.
  - In-sample forecast performance (standard deviation of residuals):
    - 4-year yields: 0.7 basis points.
    - 10-year yields: 3 basis points.

### Empirical Results: How Slope Affects Bank Profitability
- Estimation framework:
  - Fixed-effects panel with dependent variables:
    - Net interest margin (relative to assets).
    - Net non-interest income (relative to gross revenues).
    - Provision expenses-to-total loans ratio.
    - Return on assets.
  - Key regressors include annual average slope of JGB yield curve (S_t), 3-month constant-maturity JGB yields (r_i,t), lagged bank controls ((log) total assets, equity-to-assets, liquid assets-to-total assets, non-performing loans-to-gross loans, securities-to-total assets, deposits-to-total liabilities), macro variables (real GDP growth, inflation), and NIKKEI volatility; bank fixed effects included.
- Main findings:
  - A steeper JGB yield curve is estimated to:
    - Improve net interest margins (NIMs).
    - Reduce net non-interest income (reflecting valuation losses on securities holdings and commission/fee income effects).
    - Increase provision expenses (reflecting higher credit risk).
  - Magnitude:
    - "A 100-basis-points (close to 1-standard-deviation) increase in the slope of the JGB yield curve is estimated to improve NIMs by 2 basis points (or 8 basis points in the long-run)."
  - Lag structure:
    - Positive impact on NIMs and overall profitability is stronger after a one-year lag; non-interest income losses and provision increases moderate after one year.
  - Heterogeneity and limitations:
    - Large heterogeneity across banks in exposures (maturity gaps), especially major vs regional banks; analysis assumes equal exposures due to data constraints.
    - Better identification requires granular maturity-gap data to assess differential effects across bank groups.
- Interaction with macro/market conditions:
  - The beneficial impact of a steeper yield curve on profitability is reduced if real economic growth is lower or stock market volatility is higher, mainly via:
    - Lowering banks’ non-interest income.
    - Increasing provision expenses.

### Cross-Border Spillovers
- Japan’s net foreign asset position: "3.2 trillion US dollars as of 2022Q3" (components: "750 billion USD as net portfolio assets, 1.5 trillion USD as net FDI, and the rest including mainly reserve assets").
- Japan is the largest foreign holder of US Treasuries, accounting for "about 15 percent of all holdings outside the US" as of end-September 2022 (down from 17 percent a year ago).
- Foreign holdings of JGBs: "7.1 percent of outstanding JGBs (14.1 percent including Treasury discount bills) as of end-September 2022."
- VAR evidence on yield-spillovers:
  - "A 100-basis-points increase in the JGB yield curve slope is associated with an increase in the US Treasury yield curve slope by close to 10 basis points on average."
  - The impact appears short-lived, dissipating in five to six months.
  - Estimated spillovers from US Treasury to JGB slope are more persistent and significantly higher beyond short term.
  - One-year cumulative response of US Treasury slope to a shock to JGB slope is "about one-third of the spillover from the US to Japan."
- Additional channels: interbank exposures, foreign exchange, output; potential amplification if accompanied by elevated fiscal concerns or sharp domestic equity declines.

### Policy Implications and Recommendations
- Overall assessment:
  - A steeper JGB yield curve would help improve banks’ overall profitability, but the strength of the effect hinges on macroeconomic and financial market responses.
- Policy and supervisory recommendations:
  - Central bank should provide clear and careful communication on normalization strategy to avoid excessive market volatility.
  - Policymakers and supervisors should intensify monitoring of potential vulnerabilities from higher domestic interest rates, including:
    - Duration risk of financial institutions.
    - Strength of borrower cash flows, especially for those hit more strongly by the pandemic and those with variable rate loans.
  - Financial sector policies to mitigate short-term vulnerabilities if the JGB yield curve steepens could include:
    - Strengthening engagement with financial institutions with relatively high exposure to interest rate movements.
    - Encouraging reduction of duration risk where appropriate; some large banks have started reducing duration risk.
    - Supervisory efforts to reduce potential short-term adverse impacts to better harness medium-term profitability benefits.
- Caveats:
  - Banks generally have higher JGB holdings post-pandemic, suggesting larger yen interest rate risk on banking books, particularly for smaller banks.
  - Results are conditional on available data and linear approximations; non-linearities and broader global reactions could alter effects.

### Table Estimates and Interpretive Implications (regression details and key coefficients)
- Slope of the JGB Yield Curve (contemporaneous):
  - Column (1): 0.045* (0.025)
  - Column (2): 0.022*** (0.007)
  - Column (3): -1.776** (0.717)
  - Column (4): 0.105*** (0.019)
- Slope of the JGB Yield Curve (lagged):
  - Column (1): 0.135*** (0.022)
  - Column (2): 0.017*** (0.006)
  - Column (3): 1.905*** (0.493)
  - Column (4): 0.053*** (0.017)
- Short-term Interest Rate:
  - Column (1): -0.158** (0.071)
  - Column (2): -0.340*** (0.077)
  - Column (3): 0.025 (0.020)
  - Column (4): 0.0045 (0.018)
  - Column (5): 5.286** (2.286)
  - Column (6): 0.986 (2.321)
  - Column (7): 0.069 (0.055)
  - Column (8): 0.026 (0.060)
- Real GDP Growth:
  - Column (1): 0.010* (0.006)
  - Column (2): 0.004 (0.006)
  - Column (3): 0.003*** (0.001)
  - Column (4): 0.002 (0.001)
  - Column (5): 0.229 (0.140)
  - Column (6): 0.156 (0.134)
  - Column (7): -0.012*** (0.003)
  - Column (8): -0.016*** (0.003)
- Inflation:
  - Column (1): -0.018 (0.019)
  - Column (2): -0.006 (0.016)
  - Column (3): 0.007** (0.003)
  - Column (4): 0.005* (0.003)
  - Column (5): -1.796*** (0.446)
  - Column (6): -0.924** (0.382)
  - Column (7): 0.023** (0.009)
  - Column (8): 0.006 (0.008)
- NIKKEI Volatility Index (log):
  - Column (1): -0.315*** (0.049)
  - Column (2): -0.232*** (0.049)
  - Column (3): -0.005 (0.009)
  - Column (4): 0.013 (0.009)
  - Column (5): -5.527*** (1.832)
  - Column (6): -4.372** (1.906)
  - Column (7): 0.128*** (0.041)
  - Column (8): 0.191*** (0.045)
- Model controls and fit:
  - Bank Controls: Yes (all columns).
  - Bank F.E.: Included (all columns).
  - Observations:
    - Columns (1) and (2): 1,296
    - Columns (3) and (4): 1,565
    - Columns (5), (6), (7): 1,565
    - Column (8): 1,544
  - R-squared:
    - Column (1): 0.233
    - Column (2): 0.258
    - Column (3): 0.967
    - Column (4): 0.966
    - Column (5): 0.649
    - Column (6): 0.650
    - Column (7): 0.357
    - Column (8): 0.329
  - Number of banks reported: 95 (as printed).
  - Notes: All columns include lagged dependent variable. Standard errors are clustered at the bank level. *** p<0.01, ** p<0.05, * p<0.1.
- Interpretive implications:
  - Lagged slope estimates are often larger and more statistically significant than contemporaneous estimates.
  - Short-term interest rate coefficients vary by specification; some large point estimates have sizable standard errors.
  - NIKKEI Volatility Index (log) has consistent and often significant negative associations in several specifications.

*sipea2023032 — JGB Yield Curve and Macro-Financial Stability: How Would a Steeper JGB Yield Curve Affect Bank Profitability? (SIP/2023/032).*

### Section 1

### JGB Yield Curve and Macro-Financial Stability: How Would a Steeper JGB Yield Curve Affect Bank Profitability?

### A. Introduction
- BoJ unconventional monetary policies (UMP) since 2013 include Yield Curve Control (YCC) policy that caps the 10-year rates at 0 percent within a narrow band.
- Headline inflation has been above the central bank’s target of 2 percent y/y since April 2022, with considerable uncertainty around the inflation outlook.
- BoJ widened the target band for 10-year JGB yields in December 2022 to address deteriorating long-term JGB market functioning.
- Objective of the note: characterize the JGB yield curve (including slope) using a state-space model and estimate how changes in the slope affect bank profitability using fixed-effects panel estimation; examine lagged effects, heterogeneity across banks, and cross-border spillovers.

### B. Data and Characterization of JGB Yield Curve
- Data:
  - Constant-maturity JGB yields with maturities from 3 months to 30 years.
  - Sample period for yields: January 2000 till October 2022.
  - Bank-level annual unconsolidated financial statements: 95 banks, sample period 2003-2021 (Fitch-Bankscope).
  - Macroeconomic variables from IMF World Economic Outlook; NIKKEI volatility from Thomson Reuters.
- State-space specification:
  - Nelson-Siegel representation with three latent factors: L_t (level), S_t (slope), C_t (curvature).
  - Parameter λ governs decay factor on factor loadings and is estimated jointly with latent factors.
  - Estimation via maximizing Gaussian likelihood with Kalman smoothing.
- Key characterization and model fit:
  - Level factor recovered from historically low levels in 2016 and increases further since end-2021.
  - Slope factor confirms sustained flattening over the past decade and recent steepening, mainly for maturities beyond 10 years given YCC.
  - Curvature has been negative over the sample and increased sharply with the introduction of YCC.
  - In-sample forecast performance (standard deviation of residuals):
    - 4-year yields: 0.7 basis points.
    - 10-year yields: 3 basis points.

### C. Empirical Results: How Slope Affects Bank Profitability
- Estimation framework:
  - Fixed-effects panel model with dependent variables:
    - Net interest margin (relative to assets);
    - Net non-interest income (relative to gross revenues);
    - Provision expenses-to-total loans ratio;
    - Return on assets.
  - Key regressors include annual average slope of JGB yield curve (S_t), 3-month constant-maturity JGB yields (r_i,t), lagged bank controls ((log) total assets, equity-to-assets, liquid assets-to-total assets, non-performing loans-to-gross loans, securities-to-total assets, deposits-to-total liabilities), macro variables (real GDP growth, inflation), and NIKKEI volatility; bank fixed effects included.
- Main findings:
  - A steeper JGB yield curve is estimated to:
    - Improve net interest margins (NIMs).
    - Reduce net non-interest income (reflecting valuation losses on securities holdings and commission/fee income effects).
    - Increase provision expenses (reflecting higher credit risk).
  - Magnitude:
    - "A 100-basis-points (close to 1-standard-deviation) increase in the slope of the JGB yield curve is estimated to improve NIMs by 2 basis points (or 8 basis points in the long-run)."
  - Lag structure:
    - Positive impact on NIMs and overall profitability is stronger after a one-year lag; non-interest income losses and provision increases moderate after one year.
  - Heterogeneity and limitations:
    - Large heterogeneity across banks in exposures (maturity gaps), especially major vs regional banks; analysis assumes equal exposures due to data constraints.
    - Better identification requires granular maturity-gap data to assess differential effects across bank groups.
- Interaction with macro/market conditions:
  - The beneficial impact of a steeper yield curve on profitability is reduced if real economic growth is lower or stock market volatility is higher, mainly via:
    - Lowering banks’ non-interest income (e.g., smaller unrealized/realized gains on securities).
    - Increasing provision expenses (higher credit risk).

### D. Cross-Border Spillovers
- Japan’s net foreign asset position: "3.2 trillion US dollars as of 2022Q3" (components: "750 billion USD as net portfolio assets, 1.5 trillion USD as net FDI, and the rest including mainly reserve assets").
- Japan is the largest foreign holder of US Treasuries, accounting for "about 15 percent of all holdings outside the US" as of end-September 2022 (down from 17 percent a year ago).
- Foreign holdings of JGBs: "7.1 percent of outstanding JGBs (14.1 percent including Treasury discount bills) as of end-September 2022."
- VAR evidence on yield-spillovers:
  - A "100-basis-points increase in the JGB yield curve slope is associated with an increase in the US Treasury yield curve slope by close to 10 basis points on average."
  - The impact appears short-lived, dissipating in five to six months.
  - Estimated spillovers from US Treasury to JGB slope are more persistent and significantly higher beyond short term.
  - One-year cumulative response of US Treasury slope to a shock to JGB slope is "about one-third of the spillover from the US to Japan."
- Additional channels for spillovers noted: interbank exposures, foreign exchange, output; potential amplification if accompanied by elevated fiscal concerns or sharp domestic equity declines.

### E. Policy Implications and Recommendations
- Overall assessment:
  - A steeper JGB yield curve would help improve banks’ overall profitability, but the strength of the effect hinges on macroeconomic and financial market responses.
- Policy and supervisory recommendations:
  - Central bank should provide clear and careful communication on normalization strategy to avoid excessive market volatility.
  - Policymakers and supervisors should intensify monitoring of potential vulnerabilities from higher domestic interest rates, including:
    - Duration risk of financial institutions.
    - Strength of borrower cash flows, especially for those hit more strongly by the pandemic and those with variable rate loans.
  - Financial sector policies to mitigate short-term vulnerabilities if the JGB yield curve steepens could include:
    - Strengthening engagement with financial institutions with relatively high exposure to interest rate movements.
    - Encouraging reduction of duration risk where appropriate; some large banks have started reducing duration risk.
    - Supervisory efforts to reduce potential short-term adverse impacts to better harness medium-term profitability benefits.
- Caveats:
  - Banks generally have higher JGB holdings post-pandemic (amid fiscal stimulus), suggesting larger yen interest rate risk on banking books, particularly for smaller banks.
  - Results are conditional on available data and linear approximations; non-linearities and broader global reactions could alter effects.

*JGB Yield Curve and Macro-Financial Stability: How Would a Steeper JGB Yield Curve Affect Bank Profitability? (SIP/2023/032).*

### Section 2

### sipea2023032 - Section 2

### Key analytical points on global spillovers and bank balance sheets
- Depending on their risk-taking capacity, foreign investors could also be a source of global spillovers arising from changes in JGB yields.
- The overall impact would depend on additional factors such as maturity ladder of bank balance sheets and the degree at which banks adjust deposit rates in response to changes in market rates.

### Empirical results: The Impact of a Steeper JGB Yield Curve on Bank Profitability and Underlying Drivers
- Table headline: The Impact of a Steeper JGB Yield Curve on Bank Profitability and Underlying Drivers.
- Dependent Variable columns labeled (1) (2) (3) (4) (5) (6) (7) (8).
- Key regressors and estimated coefficients (standard errors in parentheses). Significance indicated as in source: *** p<0.01, ** p<0.05, * p<0.1.

- Slope of the JGB Yield Curve:
  - Column (1): 0.045* (0.025)
  - Column (2): 0.022*** (0.007)
  - Column (3): -1.776** (0.717)
  - Column (4): 0.105*** (0.019)

- Slope of the JGB Yield Curve (lagged):
  - Column (1): 0.135*** (0.022)
  - Column (2): 0.017*** (0.006)
  - Column (3): 1.905*** (0.493)
  - Column (4): 0.053*** (0.017)

- Short-term Interest Rate:
  - Column (1): -0.158** (0.071)
  - Column (2): -0.340*** (0.077)
  - Column (3): 0.025 (0.020)
  - Column (4): 0.0045 (0.018)
  - Column (5): 5.286** (2.286)
  - Column (6): 0.986 (2.321)
  - Column (7): 0.069 (0.055)
  - Column (8): 0.026 (0.060)

- Real GDP Growth:
  - Column (1): 0.010* (0.006)
  - Column (2): 0.004 (0.006)
  - Column (3): 0.003*** (0.001)
  - Column (4): 0.002 (0.001)
  - Column (5): 0.229 (0.140)
  - Column (6): 0.156 (0.134)
  - Column (7): -0.012*** (0.003)
  - Column (8): -0.016*** (0.003)

- Inflation:
  - Column (1): -0.018 (0.019)
  - Column (2): -0.006 (0.016)
  - Column (3): 0.007** (0.003)
  - Column (4): 0.005* (0.003)
  - Column (5): -1.796*** (0.446)
  - Column (6): -0.924** (0.382)
  - Column (7): 0.023** (0.009)
  - Column (8): 0.006 (0.008)

- NIKKEI Volatility Index (log):
  - Column (1): -0.315*** (0.049)
  - Column (2): -0.232*** (0.049)
  - Column (3): -0.005 (0.009)
  - Column (4): 0.013 (0.009)
  - Column (5): -5.527*** (1.832)
  - Column (6): -4.372** (1.906)
  - Column (7): 0.128*** (0.041)
  - Column (8): 0.191*** (0.045)

- Model controls and fit:
  - Bank Controls included: Yes (all columns)
  - Bank F.E. Included: Included (all columns)
  - Observations:
    - Columns (1) and (2): 1,296
    - Columns (3) and (4): 1,565
    - Columns (5), (6), (7): 1,565
    - Column (8): 1,544
  - R-squared:
    - Column (1): 0.233
    - Column (2): 0.258
    - Column (3): 0.967
    - Column (4): 0.966
    - Column (5): 0.649
    - Column (6): 0.650
    - Column (7): 0.357
    - Column (8): 0.329
  - Number of banks reported: 95 95 95 95 95 95 (as printed)
  - Underlying Drivers listed in the table: Profitability; Net Interest Margin; Net Non-Interest Income-to-Gross revenues; Provisions-to-gross loans.
  - Notes: All columns include lagged dependent variable. Standard errors are clustered at the bank level, and provided in parantheses. *** p<0.01, ** p<0.05, * p<0.1.

### Interpretive implications (as implied by results)
- A steeper JGB yield curve is associated with statistically significant effects on several bank outcome measures, with direction and magnitude varying by specification.
- Lagged slope of the JGB yield curve often shows larger and highly significant associations relative to contemporaneous slope estimates.
- Short-term interest rate effects vary across specifications, including negative and positive coefficients, and large point estimates in columns (5) and (6) with sizable standard errors.
- Macroeconomic controls such as Real GDP Growth and Inflation show small but sometimes significant associations with bank outcomes; NIKKEI Volatility Index (log) has consistent and often significant negative associations in several specifications.

*Source: sipea2023032 - Section 2*

---


_Source: https://www.imf.org/-/media/files/publications/selected-issues-papers/2023/english/sipea2023032.pdf_