## _wp13143

## Source details

**Canonical URL:** [_wp13143](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13143.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13143.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13143.pdf.json)

---

### Introduction and paper contributions
- Focus: link between short-term interest rates and bank risk taking (monetary policy’s “risk-taking” channel) using confidential loan-level data from the Federal Reserve’s Survey of Terms of Business Lending (STBL).
- Motivations: debate that low policy rates and unconventional monetary policies may have encouraged higher bank risk taking prior to the global financial crisis.
- Main contributions:
  - First to present evidence of an inverse relationship between interest rates and bank risk taking using disaggregated commercial bank loan data for the United States (use of detailed STBL data for this purpose).
  - Construct an ex-ante measure of bank risk taking using confidential loan-level data on the riskiness of loans of U.S. banks, focusing on banks’ risk assessment at loan origination rather than ex-post loan performance.
  - Link the strength of the relationship between policy rates and bank risk taking to the capital structure of the banking system.

### Theoretical model and testable implications
- Stylized model (based on Dell’Ariccia, Laeven, and Marquez (2013)):
  - Banks choose monitoring effort q; monitoring cost per dollar lent = (1/2)cq².
  - Financing: fraction k financed with owner’s money (capital), 1-k financed by deposits.
  - Deposits fully insured; deposit rate r_D = r^* (reference real risk-free rate).
  - Equity yield r_E = (r^* + ξ)/q with ξ ≥ 0.
  - Zero expected profits equilibrium determines lending rate r_L.
- Key model equations and expressions (as presented):
  - Π = q( r_L − r_D(1 − k) ) − r_E k − (1/2) c q² L(r_L).
  - Alternative: Π = ( q( r_L − r^*(1 − k) ) − ( r^* + ξ ) k − (1/2) c q² ) L(r_L).
  - First-order condition yields q̂ (paper notation, see equation (3)).
  - Lending rate from zero profits: r_L = r^*(1 − k) + ξ/(2 c k (r^* + ξ)). (paper’s equation (5))
- Channels through which r^* affects monitoring:
  - Risk-shifting channel: higher r^* raises deposit costs and can reduce monitoring incentives.
  - Pass-through channel: higher r^* raises lending rates r_L and can increase monitoring incentives.
- Capitalization implications:
  - Poorly capitalized banks (low k): risk-shifting effect stronger; changes in r^* reduce monitoring more.
  - Well-capitalized banks (high k): limited liability matters less; negative association between interest rates and risk taking is more pronounced.
- Testable implication stated:
  - Bank risk taking is negatively associated with interest rates if banks are not capital constrained; the negative relationship is less pronounced for poorly capitalized banks.

### Data, measurement, and descriptive statistics
- Data:
  - Source: confidential STBL loan-level data, quarterly, sample period 1997-2011.
  - STBL sample: stratified sample of about 400 banks; combined assets of responding banks in Q4 2011 represented about 60 percent of all assets of U.S. commercial banks.
  - STBL records loans made during the first full business week of the middle month in each quarter (February, May, August, November).
- Loan risk rating (STBL scale 1–5):
  - 1 = Minimal Risk, 2 = Low Risk, 3 = Moderate Risk, 4 = Acceptable Risk, 5 = Special Mention or Classified Asset.
- Complementary variables:
  - Bank balance-sheet: FFIEC 031 and 041 (Call Reports).
  - Market capitalization: end-of-quarter BHC market cap from CRSP for largest bank in BHC.
  - Regional controls: BEA personal income, BLS unemployment, OFHEO/FHFA housing prices; region CPI change from BLS.
  - Time-specific: three-month average target federal funds rate (nominal); FDIC fraction of U.S. bank failures.
- Summary statistics (Table 1; N = 994,287; Number of banks = 475):
  - Risk rating: Average 3.434, Standard deviation 0.850
  - Loan size (dollars): Average 512,481, Standard deviation 4,992,014
  - Dummy for loans secured by collateral: Average 0.809, Standard deviation 0.393
  - Loan maturity (years): Average 1.238, Standard deviation 1.768
  - Bank total assets (millions of dollars): Average 191,624, Standard deviation 279,194
  - Tier 1 capital ratio: Average 0.094, Standard deviation 0.027
  - Total capital ratio: Average 0.122, Standard deviation 0.023
  - Common stock to assets ratio: Average 0.003, Standard deviation 0.008
  - Equity to assets ratio: Average 0.101, Standard deviation 0.026
  - Target federal funds rate (%): Average 2.206, Standard deviation 1.853
  - State personal income (dollars): Average 312,407, Standard deviation 359,049
  - Change in region CPI (%): Average 0.596, Standard deviation 0.858
  - State unemployment rate (%): Average 5.830, Standard deviation 2.169
  - Change in state housing prices (%): Average 0.500, Standard deviation 2.198
- Descriptive relationship:
  - Aggregate weighted average loan risk rating and nominal federal funds rate show a strong negative relationship, statistically significant at the 1% level (quarterly data Q2 1997–Q4 2011). Detrending via linear time trend or HP filter yields similar results.

### Empirical methodology
- Regression framework (Section II.A):
  - Loan risk rating of loan k by bank i in quarter t regressed on:
    - federal funds rate at beginning of quarter t (r_t),
    - bank capital-asset ratio at beginning of quarter t (k_it),
    - loan-specific controls (X_kit: loan size, collateral indicator, loan maturity),
    - bank-specific controls (Z_it other than leverage),
    - regional time-varying controls (W_st),
    - bank fixed effects α_i and state fixed effects γ_s.
  - Standard errors clustered at the quarterly level.
- Coefficients of interest:
  - β on federal funds rate: expected negative.
  - Interaction γ on (capital × federal funds rate): expected negative (reductions in interest rates increase bank risk taking especially for banks with relatively high capital).
- Note: empirical analysis uses nominal federal funds rate; correlation between nominal and real effective federal funds rates over 1997-2011 = 0.9.

### Main empirical results
- Baseline OLS (loan-level), sample Q2 1997–Q4 2011; standard errors clustered by quarter.
- Core finding:
  - Significantly negative relationship between nominal federal funds rate and ex-ante bank risk taking (loan risk ratings).
- Representative coefficient magnitudes (Table 2; Table 3):
  - Target federal funds rate coefficients across specifications: -0.016*** [0.000], -0.016** [0.007], -0.021*** [0.006], -0.021*** [0.006], -0.031*** [0.009], -0.031*** [0.008].
  - Table 3 Panel A: Target federal funds rate = -0.031*** [0.008] across columns.
- Magnitude example:
  - Based on regression estimates in column 6 of Table 2: a one-standard deviation decrease in interest rates of 1.853 would suggest an increase in loan risk ratings of 0.057 (loan risk ratings SD = 0.85).
- Loan-characteristic effects (Table 3 Panel A):
  - Loan size (log): negative significant (examples -0.031*** [0.002], -0.024*** [0.002]).
  - Dummy for secured loans: positive significant (examples 0.241*** [0.008], 0.227*** [0.008]).
  - Loan maturity: negative significant (examples -0.022*** [0.002], -0.018*** [0.001]).
- Heterogeneity by loan subsamples (Table 3 Panel B):
  - Loan size above median: target federal funds rate -0.030*** [0.007] (N = 520,258).
  - Loan size below median: -0.034*** [0.010] (N = 474,029).
  - Loans secured by collateral: -0.027*** [0.009] (N = 804,163).
  - Loans not secured: -0.055*** [0.005] (N = 190,124).
  - Longer maturity: -0.032*** [0.008] (N = 510,972).
  - Shorter maturity: -0.028*** [0.009] (N = 483,315).

### Capital interactions and heterogeneity (Table 4, Panels A–C; Table 5)
- Interaction results (Table 4 Panel A):
  - Tier 1 capital ratio × target federal funds rate: -0.317*** [0.082]
  - Total capital ratio × target federal funds rate: -0.239*** [0.077]
  - Common stock-assets ratio × target federal funds rate: -0.541*** [0.127]
  - Stock market capitalization-assets ratio × target federal funds rate: -0.077** [0.034]
- Subsample by Tier 1 above vs below median (Panel B):
  - Tier 1 above median: target federal funds rate -0.048*** [0.016]; Observations 490,064; Number of banks 455; R2 0.227
  - Tier 1 below median: target federal funds rate -0.019*** [0.004]; Observations 504,223; Number of banks 184; R2 0.155
- Quarter-year fixed effects (Panel C):
  - Tier 1 capital ratio: 1.109*** [0.372]
  - Tier 1 capital ratio × target federal funds rate: -0.389*** [0.081]
  - Total capital ratio: 1.670*** [0.357]
  - Total capital ratio × target federal funds rate: -0.279*** [0.085]
  - Common stock-assets ratio × target federal funds rate: -0.313* [0.157]
- Economic magnitudes (examples reported):
  - Column 2 of Table 4: one standard deviation decrease in interest rates evaluated at one standard deviation below sample mean of Tier 1 capital ratio → increase in loan risk ratings of 0.06.
  - Same evaluated at one standard deviation above sample mean of Tier 1 capital ratio → increase of 0.08 (one-tenth of loan risk ratings SD).
  - Inflection points (Panel C column 1):
    - Tier 1 capital ratio: increases in Tier 1 translate into a decrease in bank risk taking when target federal funds rate exceeds 2.85 percent.
    - Total capital ratio: decreases in risk taking when target federal funds rate exceeds 5.99 percent.
- Loans under prior commitment (~25% of observations; Table 5):
  - Excluding these loans generally increases absolute magnitude of interaction coefficients; complete sample retained for subsequent analysis.

### Robustness and subsample analyses
- Small banks (assets below top quintile; Table 6; Observations 217,513; Number of banks 414; R2 0.305):
  - Tier 1 capital ratio × target federal funds rate: -0.396*** [0.119]
  - Total capital ratio × target federal funds rate: -0.260** [0.120]
  - Common stock-assets ratio × target federal funds rate: -0.824** [0.310]
- States without large banks (Table 7; Observations 598,017; Number of banks 443; R2 0.214):
  - Target federal funds rate: -0.061*** [0.014]
  - Tier 1 capital ratio × target federal funds rate: -0.209* [0.110]
- State cyclicality (Table 8):
  - Target federal funds rate for states with high correlation with US GDP: -0.037** [0.014]
  - Target federal funds rate for low-correlation states: -0.021*** [0.004]
  - Tier 1 capital ratio × target federal funds rate significant in several subsamples (examples: -0.349*** [0.095]; -0.436*** [0.151])
- Regional monetary policy measures (Table 10):
  - Regional measure: target federal funds rate minus change in region CPI: -0.031*** [0.008]
  - Deviation from regional Taylor rule: -0.031*** [0.008]
  - Tier 1 capital ratio × (target federal funds rate − change in region CPI): -0.274*** [0.083]
  - Tier 1 capital ratio × (deviation from regional Taylor rule): -0.217*** [0.080]
- Measurement error and economic-state controls (Table 11):
  - Interactions between federal funds rate and state of the economy (real U.S. GDP growth; NBER recession dummy) leave capital × rate interactions roughly unchanged.
  - Target federal funds rate × NBER recession dummy example: -0.028** [0.013]
- Crisis vs non-crisis periods (Table 9):
  - Crisis window defined as 2008–2010.
  - Interaction between capital ratios and interest rates more pronounced during non-crisis periods; can turn mildly positive during crisis periods.
  - Example: non-crisis (column 2): at one-standard deviation above sample mean of Tier 1 capital ratio, one-standard deviation decrease in interest rates → increase in loan risk ratings of 0.10 (loan risk ratings SD = 0.85).
  - Inclusion of FDIC fraction of bank failures: reductions in interest rates have disproportionately positive effect on bank risk taking during periods with relatively few bank failures.
- Long low-rate spells (Table 12):
  - Measure: number of consecutive quarters without an increase in the target federal funds rate.
  - Findings:
    - Longer spells without increases associated with higher risk ratings of new loans (specification (1): 0.005* [0.002]).
    - Effect more pronounced for well-capitalized banks (periods without increase × Tier 1: 0.058* [0.029]; 0.080** [0.033] in other specs).
    - Robust to controlling for level of federal funds rate.
    - Periods without increase × target federal funds rate: -0.004** [0.002] (specification 3).
  - Consistent with prior literature: holding rates low for prolonged periods softens lending standards.

### Key empirical patterns and magnitudes
- Repeated empirical regularities:
  - Higher target federal funds rates often associated with lower loan risk ratings (examples: -0.031*** [0.008], -0.016*** [0.004], -0.021*** [0.006]).
  - Interaction terms show higher bank capital amplifies the negative association between federal funds rate and loan risk ratings (examples: Tier 1 × rate -0.317*** [0.082]; Total capital × rate -0.239*** [0.077]; Common stock-assets × rate -0.541*** [0.127]).
  - Heterogeneity across loan types: unsecured loans and small loans show larger negative coefficients (e.g., unsecured: -0.055*** [0.005]).
  - Subsample differences: effects stronger in non-crisis years and during prolonged low-rate spells; measurement-error concerns would bias coefficients toward zero, so detected effects are not driven by simple optimism bias in loan officer reporting.

### Interpretation, caveats, and policy implications
- Interpretation caveats:
  - Theoretical model and empirical results do not by themselves determine whether additional risk taking under lower rates is “excessive” (Stein, 2012 discussed).
  - Short-run vs medium-run: holding rates low briefly may improve portfolio quality; prolonged low rates could raise default risk over medium term (consistent with Jimenez et al., 2008; Maddaloni and Peydro, 2011).
- Policy implications articulated:
  - Low short-term interest rate environment increases bank risk taking as measured at origination; the effect depends on bank capitalization.
  - A one-standard deviation decrease in interest rates evaluated at one standard deviation above sample mean of Tier 1 capital ratio → increase in loan risk ratings of about one-tenth its standard deviation.
  - Whether monetary policy should internalize financial stability concerns depends on shock type:
    - Near cycle peak: tightening may reduce leverage and risk taking while containing price pressures — no trade-off.
    - Low inflation with excessive risk taking (e.g., run-up to recent crisis): tightening reduces risk taking but may cause undesired contraction or deflation — trade-off exists; other macroprudential tools may be needed.
  - Results support arguments for integrating macro-prudential regulation into monetary policy frameworks or centralizing macro-prudential responsibilities within monetary authority, though trade-offs with credibility, accountability, and political interference remain open.

*Source: _wp13143 - Section II and References*

### References .............................................................................................................

### _wp13143 - References .............................................................................................................

### Introduction and paper contributions
- Focus: link between short-term interest rates and bank risk taking (monetary policy’s “risk-taking” channel), using confidential loan-level data from the Federal Reserve’s Survey of Terms of Business Lending (STBL).
- Motivation: debate that low policy rates and unconventional monetary policies may have encouraged higher bank risk taking prior to the global financial crisis.
- Three main contributions:
  - First to present evidence of an inverse relationship between interest rates and bank risk taking using disaggregated commercial bank loan data for the United States (To the best of our knowledge, we are the first to use detailed STBL data for the purpose of studying the link between interest rates and bank risk taking.2).
  - Construct an ex-ante measure of bank risk taking using confidential loan-level data on the riskiness of loans of U.S. banks, focusing on banks’ risk assessment at loan origination rather than ex-post loan performance.
  - Link the strength of the relationship between policy rates and bank risk taking to the capital structure of the banking system.

### Key empirical finding (as stated)
- Bank risk taking—as measured by the risk ratings of the bank’s loan portfolio—is negatively associated with short-term interest rates—as proxied by the federal funds rate4—and this negative relationship is more pronounced for highly capitalized banks.

### Relation to existing literature (selected points)
- Theoretical literature predicts banks may engage in riskier activities when monetary policy is expansive and agency problems are severe (examples cited: Adrian and Shin, 2010; Freixas, Martin, and Skie, 2011; Diamond and Rajan, 2012; Dell’Ariccia, Laeven, and Marquez, 2013; Acharya and Naqvi, 2012).
- Empirical studies vary by measure and geography:
  - Lown and Morgan (2006): U.S. lending standards tend to tighten after monetary contractions, but results not statistically significant.
  - Maddaloni and Peydro (2011): euro area lending survey data show credit standards loosen when overnight rates are lowered; holding rates low for prolonged periods softens lending standards further.
  - Altunbas, Gambacorta, and Marquez-Ibañez (2010): increases in interest rates and negative Taylor rule residuals are positively associated with default risk measures.
  - Paligorova and Santos (2012): syndicated loan pricing shows riskier borrower pricing is more favorable during loose monetary policy; effect stronger for banks with greater risk appetite.
  - Jimenez et al. (2008), Ioannidou, Ongena, and Peydro (2009): credit registry evidence for Spain and Bolivia; low interest rates at origination associated with greater lending to risky borrowers and with increased leverage; also find low rates decrease the riskiness of banks’ overall loan portfolios in the short run but may increase default risk over the medium term.
- Distinct advantages of the STBL + Call Report U.S. data used in this paper:
  - STBL captures absolute level of risk on new business loans and includes information on marginal pricing of loans.
  - Larger and more heterogeneous sample of banks with long time series relative to many other country datasets.

### The simple model: setup and intuition (stylized version from Dell’Ariccia, Laeven, and Marquez (2013))
- Environment:
  - Banks exert monitoring effort q to increase loan repayment probability; monitoring cost per dollar lent equals (1/2)cq².
  - Bank financing: fraction k financed with owner’s money (capital/equity) and 1-k financed by deposits.
  - Deposits are fully insured; deposit rate r_D = r^*, the reference (real risk-free) rate.
  - Equity yield r_E = (r^* + ξ)/q, with ξ ≥ 0 (equity premium).
  - Banks operate under limited liability and compete so that lending rate r_L is set with zero expected profits in equilibrium (free-entry competitive equilibrium).
- Bank expected profit (per the paper’s notation)
  - Π = q( r_L − r_D(1 − k) ) − r_E k − (1/2) c q² L(r_L).   (equation (1) in paper)
  - Alternative expression in paper: Π = ( q( r_L − r^*(1 − k) ) − ( r^* + ξ ) k − (1/2) c q² ) L(r_L).   (equation (2) in paper)
- First-order condition for monitoring q yields (paper expression):
  - q̂ = ( r_L − r^*(1 − k) )(1/c)^(1/3) ? (equation (3) in paper) — (paper presents q̂ in its own notation; see equation (3)).
- Lending rate consistent with zero profits (paper expression):
  - r_L = r^*(1 − k) + ξ/(2 c k (r^* + ξ))? (presented as equation (5) in the paper; full expression in paper).
  - (Paper’s equation (5) explicitly: r_L = r^*(1 − k) + ξ/(2 c k (r^* + ξ)).)
- From the model:
  - Changes in the reference rate r^* affect monitoring through two channels:
    - Risk-shifting channel: higher r^* raises deposit costs, reducing bank profits in the success state and lowering incentives to monitor.
    - Pass-through channel: higher r^* increases lending rates r_L, raising bank rewards in the success state and improving incentives to monitor.
  - The net effect depends on bank capitalization k:
    - For poorly capitalized banks (low k), the classical risk-shifting effect is stronger and reduces the net effect of a change in r^*.
    - For well-capitalized banks (high k), limited liability matters less, and the negative association between interest rates and risk taking is more pronounced.
- Testable implication stated:
  - Bank risk taking is negatively associated with interest rates if banks are not capital constrained; the negative relationship is less pronounced for poorly capitalized banks.

### Empirical methodology and data (overview as described)
- Approach: panel regression analysis to test model predictions using confidential STBL loan-level risk ratings combined with bank balance sheet data (Call Reports).
- Focus: ex-ante loan risk at origination, federal funds rate as proxy for short-term policy rate, heterogeneity by bank capitalization and other bank characteristics.
- Empirical tests include robustness checks and extensions (details and numerical results are in later sections of the paper).

### Interpretation and caveats (authors’ points)
- Theoretical model and empirical results do not by themselves determine whether additional risk taking under lower rates is “excessive” (see Stein, 2012 for discussion on excessiveness).
- Holding rates low for short periods may improve loan portfolio quality; holding rates low for prolonged periods could increase loan default risk over the medium term (evidence and interpretation drawn from cited literature such as Jimenez et al., 2008).

*Source: _wp13143 - References .............................................................................................................*

### Section II.

### Section II.

### A. Regression Model of Bank Risk Taking
- Model specification (equation (1)): loan risk rating ߪ௞௜௧ of loan k by bank i in quarter t is regressed on:
  - federal funds rate at the beginning of quarter t (ݎ௧),
  - capital-asset ratio of bank i at the beginning of quarter t (ܭ௜௧),
  - loan-specific controls ܺ௞௜௧ (loan size, indicator for collateral, loan maturity),
  - bank-specific controls ܻ௜௧ (other than leverage),
  - regional time-varying controls ܼ௝௧,
  - bank fixed effects ߙ௜ and state fixed effects ߣ௝.
- Standard errors clustered at the quarterly level.
- Coefficient of interest: ߚ on the federal funds rate; expected sign: negative.
- Interaction specification (equation (2)): includes interaction ܭݎ௧ between bank capital and interest rates. Expected coefficient on interaction ߜ: negative — reductions in interest rates increase bank risk taking especially for banks with relatively high capital (low leverage).
- Alternative specification (equation (2a)): replace interest rate with quarter-specific fixed effects ࢚࣎.
- Note on real vs nominal rates:
  - Model cast in real rates; empirical approach uses nominal federal funds rate.
  - Over the sample period (1997-2011), correlation between nominal and real effective federal funds rates is 0.9.

### B. Financial Stability Considerations and Monetary Policy: Evidence from FOMC Minutes
- Identification assumption: interest rate changes induced by monetary policy are exogenous to bank risk taking (monetary policy does not respond to financial stability).
- FOMC minutes keyword analysis:
  - 14 keywords related to financial stability were counted (total appearances and number of reports; frequency computed as appearances divided by number of quarters).
  - Findings:
    - Prior to 2007, most financial-stability-related keywords (including “financial stability,” “bank risk,” and “systemic risk”) did not appear even once.
    - Since 2007, keywords related to financial stability appeared more frequently, though increases were small for most keywords.
  - Interpretation: until recently, financial stability considerations played a limited direct role in setting U.S. monetary policy; protecting financial stability was primarily assigned to supervisory and regulatory policy while interest rate policy focused on price stability and output stabilization.

### IV. DATA AND DESCRIPTIVE STATISTICS
- Data source: confidential loan-level data from the Federal Reserve’s Survey of Terms of Business Lending (STBL), quarterly, sample period 1997-2011.
  - STBL sample: stratified sample of about 400 banks.
  - Combined assets of responding banks in Q4 2011 represented about 60 percent of all assets of U.S. commercial banks.
  - STBL records loans made during the first full business week of the middle month in each quarter (February, May, August, November).
- Loan risk rating:
  - STBL scale 1 to 5: 1=Minimal Risk, 2=Low Risk, 3=Moderate Risk, 4=Acceptable Risk, 5=Special Mention or Classified Asset.
  - Loan-level fields include risk rating, bank identifier, size (dollars), maturity (years), secured indicator, rate and pricing details.
- Prime rate supplement: collected from all STBL respondents; prime rate defined as “the administered rate used for pricing business and other credit.”
- STBL revisions and sample details:
  - Risk-rating item added in 1997; respondent panel expanded in 1997 to include U.S. branches and agencies of foreign banks.
  - Minimum loan size increased from $1,000 to $3,000 in 2006.
  - STBL micro-level data are confidential to the Federal Reserve System.
- Complementary datasets and variables:
  - Bank balance-sheet: Consolidated Reports of Condition and Income (FFIEC 031 and 041) — bank total assets, regulatory capital ratios, common stock.
  - Market capitalization: end-of-quarter BHC market cap from CRSP for largest bank in BHC (used in some specifications).
  - Bank location: headquarters from National Information Center (NIC).
  - Regional controls (state level where available): personal income (BEA), unemployment rate (BLS), change in housing prices (OFHEO/FHFA). Region-level control: quarter-over-quarter change in CPI (BLS).
  - Time-specific variables: three-month average target federal funds rate (nominal); fraction of U.S. bank failures from FDIC (relative to number of insured banks).
- Descriptive statistics (Table 1):
  - Average loan risk rating: 3.43.
  - Standard deviation of loan risk rating: 0.85.
  - Aggregate data show a strong negative relationship between weighted average loan risk rating and the nominal federal funds rate, statistically significant at the 1% level (quarterly data Q2 1997–Q4 2011). Detrending used linear time trend or Hodrick-Prescott filter yields similar results.

### V. EMPIRICAL RESULTS
A. Main Results
- Estimation: OLS regressions at the loan level, standard errors clustered by quarter, sample Q2 1997 to Q4 2011.
- Baseline result: significantly negative relationship between short-term interest rates (nominal federal funds rate) and ex-ante bank risk taking (loan risk ratings).
- Magnitude examples:
  - Based on regression estimates in column 6 of Table 2: a one-standard deviation decrease in interest rates of 1.853 would suggest an increase in loan risk ratings of 0.057 (loan risk ratings SD = 0.85).
- Loan-characteristic controls (Table 3): controlling for loan size (logs), secured dummy, and loan maturity; sample splits by loan size, collateral, and maturity continue to show significantly negative relationship between interest rates and loan risk ratings.
- Capital interaction results (Table 4):
  - Interaction between federal funds rate and capital measures (Tier 1 capital ratio; Total capital ratio; Common stock-assets ratio; Stock market capitalization-assets ratio) yields a statistically significant negative coefficient on the interaction across measures.
  - Example magnitudes:
    - Column 2 of Table 4: a one standard deviation decrease in interest rates evaluated at one standard deviation below the sample mean of the Tier 1 capital ratio translates into an increase in loan risk ratings of 0.06.
    - Same estimate evaluated at one standard deviation above the sample mean of the Tier 1 capital ratio: increase of 0.08 (one-tenth of the loan risk ratings SD).
  - Market-capitalization measure produces more muted economic effect than book capital measures.
  - Panel C (time-fixed effects) produces similar or somewhat larger economic magnitudes than models using the federal funds rate level.
  - Inflection points from column 1 of Panel C:
    - Tier 1 capital ratio: increases in Tier 1 translate into a decrease in bank risk taking when target federal funds rate exceeds 2.85 percent.
    - Total capital ratio: decreases in risk taking when target federal funds rate exceeds 5.99 percent.
- Loans made under prior commitment (~25% of observations; Table 5):
  - Excluding these loans generally increases the absolute magnitude of coefficients on interactions between capital ratios and federal funds rates.
  - Complete sample retained in subsequent analysis.

B. Robustness Checks
- Small banks (assets below top quintile; Table 6): negative relationship and negative interaction between capital and rates persist; magnitude similar or larger than full sample.
- States with small banking systems (exclude states with banks in top 1% of asset distribution; Table 7): negative interaction remains significant.
- States with business cycles less correlated with national cycle (Table 8):
  - Split by correlation of state income growth with U.S. GDP growth; results roughly stable and significant across subsamples.
  - Coefficient on federal funds rate smaller and interacted term larger in low-correlation subsample; differences not statistically significant.
  - Similar findings when splitting by state income volatility.
- Crisis vs non-crisis periods (Table 9):
  - Crisis period defined as years 2008-2010.
  - Negative effect of interaction between capital ratios and interest rates is more pronounced during non-crisis periods.
  - During crisis periods, the coefficient on the interaction can turn mildly positive.
  - Example magnitude (column 2 of Table 9): at one-standard deviation above sample mean of Tier 1 capital ratio, a one-standard deviation decrease in interest rates translates into an increase in loan risk ratings of 0.10 (loan risk ratings SD = 0.85 for non-crisis sample).
  - Inclusion of FDIC fraction of bank failures (column 5): reductions in interest rates have a disproportionately positive effect on bank risk taking during periods with relatively few bank failures.
- Regional monetary policy measures (Table 10):
  - Two alternatives tested:
    1. nominal target federal funds rate minus change in regional CPI;
    2. deviation between nominal target federal funds rate and nominal rate implied by simple regional Taylor rule: (regional inflation + 0.5×(regional inflation – 2%) + 0.5×(log(state income) – trend log(state income))).
  - Results: negative relationship between interest rates and bank risk taking persists; link less pronounced for poorly capitalized banks.
- Measurement error concerns:
  - If loan officers are more optimistic in expansions, reported risk may be underestimated during expansions; because federal funds rate co-moves with cycle (higher in expansions), such measurement error would bias coefficient on federal funds rate toward zero — working against finding an effect.
  - Additional regressions (Table 11) control for interactions between federal funds rate and state of the economy (real U.S. GDP growth; NBER recession dummy). Coefficients on interactions between federal funds rate and banks’ capital ratios remain roughly unchanged; level effect of federal funds rate becomes more pronounced during recessions.
- Overall: robustness checks mitigate endogeneity concerns and support bank leverage as key factor driving the risk taking channel of monetary policy.

C. Long Periods with Low Interest Rates
- Measure of duration: number of consecutive quarters without an increase in the target federal funds rate by the FOMC.
- Findings (Table 12):
  - Longer periods without increases in interest rates are associated with higher risk ratings of new loans (column 1).
  - Effect is more pronounced for well-capitalized banks (column 2); robust to controlling for level of federal funds rate (column 3).
  - Negative effect of the interaction between capital ratios and interest rates on bank risk taking is more pronounced when interest rates are not raised for prolonged periods (column 4).
  - Consistent with Maddaloni and Peydro (2011): holding rates low for prolonged periods softens lending standards.
- Conclusion: both level and duration of low interest rates matter for the link between bank leverage and risk taking.

### VI. CONCLUSIONS
- Main conclusions:
  - A low short-term interest rate environment increases bank risk taking (as measured by loan risk ratings).
  - The effect depends on bank capitalization: the effect of interest rates on bank risk taking is less pronounced for poorly capitalized banks.
  - Economic magnitude: a one-standard deviation decrease in interest rates evaluated at one standard deviation above the sample mean of the Tier 1 capital ratio translates into an increase in loan risk ratings of about one-tenth its standard deviation.
  - Effect is more pronounced in non-crisis times and during prolonged periods without rate increases.
- Policy implications and interpretation:
  - Findings relate to debate on integrating macro-prudential regulation into monetary policy frameworks to pursue price and financial stability jointly.
  - Whether a trade-off between price and financial stability arises depends on shock type:
    - Near cycle peak: tightening may reduce leverage and risk taking while containing price pressures — no trade-off.
    - Low inflation with excessive risk taking (e.g., run-up to recent crisis): tightening reduces risk taking but may cause undesired contraction or deflation — trade-off exists; other macroprudential tools may be needed.
  - Results can be viewed as an argument for centralization of macro-prudential responsibilities within the monetary authority, given the interaction complexity; trade-offs with credibility, accountability, and political interference remain open questions.

*Source: _wp13143 - Section II.*

### References

### _wp13143 - References

### References on bank liquidity, risk-taking, and monetary policy
- Acharya, Viral, and Hassan Naqvi, 2012, “The Seeds of a Crisis: A Theory of Bank Liquidity and Risk-Taking over the Business Cycle,” Journal of Financial Economics, Vol. 106, No. 2, pp.  349-66.
- Adrian, Tobias, and Hyun Song Shin, 2009, “Money, Liquidity and Monetary Policy,” American Economic Review, Papers and Proceedings, Vol. 99, pp. 600-05.
- Adrian, Tobias, and Hyun Song Shin, 2010, “Financial Intermediaries and Monetary Economics,” in: Benjamin M. Friedman and Michael Woodford (eds.), Handbook of Monetary Economics, New York, N.Y.: Elsevier.
- Borio, Claudio, and Haibin Zhu, 2008, “Capital Regulation, Risk-Taking and Monetary Policy: A Missing Link in the Transmission Mechanism?” BIS Working paper No. 268.
- Freixas, Xavier, Antoine Martin, and David R. Skeie, 2011, “Bank Liquidity, Interbank Markets, and Monetary Policy,” Review of Financial Studies, Vol. 24, No. 8, pp. 2656-692.
- Dell’Ariccia, Giovanni, Luc Laeven, and Robert Marquez, 2013, “Monetary Policy, Leverage, and Bank Risk-Taking,” Journal of Economic Theory, forthcoming.
- Paligorova, Teodora, and Joao A.C. Santos, 2012, “Monetary Policy and Bank Risk-Taking: Evidence from the Corporate Loan Market,” working paper, Federal Reserve Bank of New York.
- Jimenez, Gabriel, Steven Ongena, Jose Luis Peydro, and Jesus Saurina, 2008, “Hazardous Times for Monetary Policy: What Do Twenty-Three Million Bank Loans Say About the Effects of Monetary Policy on Credit Risk-Taking?” Banco de España documento de Trabajo 0833.
- Ioannidou, Vasso P., Steven Ongena, and Jose Luis Peydro, 2009, “Monetary Policy, Risk-Taking, and Pricing: Evidence from a Quasi-Natural Experiment,” CentER - Tilburg University, mimeo.
- Stein, Jeremy, 2012, “Monetary Policy as Financial-Stability Regulation,” Quarterly Journal of Economics, Vol. 127, pp. 57-95.
- Mishkin, Frederic, 2010, “Monetary Policy Strategy: Lessons from the Crisis,” Prepared for the ECB Central Banking Conference, “Monetary Policy Revisited: Lessons from the Crisis,” Frankfurt, November 18-19.
- Rajan, Raghuram, 2010, “Why We Should Exit Ultra-Low Rates: A Guest Post,” The New York Times: Freakonomics, August 25.

### References on capital regulation, supervision, and moral hazard
- Allen, Franklin, Elena Carletti, and Robert Marquez, 2011, “Credit Market Competition and Capital Regulation,” Review of Financial Studies, Vol. 24, No. 4, pp. 983-1018.
- Altunbas, Yener, Leonardo Gambacorta, and David Marquez-Ibanez, 2010, “Does Monetary Policy Affect Bank Risk-Taking?” BIS Working Paper No. 298.
- Bank of England, 2013, “The Financial Policy Committee’s Powers to Supplement Capital Requirements,” A Draft Policy Statement, January 2013.
- Hellmann, Thomas, Kevin Murdock, and Joseph Stiglitz, 2000, “Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough?” American Economic Review, Vol. 90, No. 1, pp. 147-65.
- Cordella, Tito, and Eduardo Levy-Yeyati, 2003, “Bank Bailouts: Moral Hazard vs. Value Effect,” Journal of Financial Intermediation, Vol. 12, No. 4, pp. 300-30.
- Repullo, Rafael, 2004, “Capital Requirements, Market Power, and Risk-Taking in Banking,” Journal of Financial Intermediation, Vol. 13, pp. 156-82.
- Borio, Claudio, and Haibin Zhu, 2008, “Capital Regulation, Risk-Taking and Monetary Policy: A Missing Link in the Transmission Mechanism?” BIS Working paper No. 268. (also listed under liquidity and monetary policy)

### References on bank behavior, competition, and lending channels
- Berger, Allen N., Anil K. Kashyap, and Joseph M. Scalise, 1995, “The Transformation of the U.S. Banking Industry: What a Long, Strange Trip It’s Been,” Brookings Papers on Economic Activity, No. 2, pp. 55-201.
- Black, Lamont K., and Lieu N. Hazelwood, 2013, “The Effect of TARP on Bank Risk-Taking,” Journal of Financial Stability, forthcoming.
- Black, Lamont K., and Richard J. Rosen, 2007, “How the Credit Channel Works: Differentiating the Bank Lending Channel and the Balance Sheet Channel,” Federal Reserve Bank of Chicago, Working Paper Series: WP-07-13.
- Dell’Ariccia, Giovanni, and Robert Marquez, 2006, “Competition among Regulators and Credit Market Integration,” Journal of Financial Economics, Vol. 79, pp. 401-30.
- Maddaloni, Angela, and Jose Luis Peydro, 2011, “Bank Risk-Taking, Securitization, Supervision and Low Interest Rates: Evidence from the Euro Area and U.S. Lending Standards,” Review of Financial Studies, Vol. 24, No. 6, pp. 2121-165.
- Morgan, Donald P., and Adam B. Ashcraft, 2003, “Using Loan Rates to Measure and Regulate Bank Risk: Findings and an Immodest Proposal,” Journal of Financial Services Research, Vol. 24, 2-3, pp. 181-200.
- Lown, Cara, and Donald P. Morgan, 2006, “The Credit Cycle and the Business Cycle: New Findings Using the Loan Officer Opinion Survey,” Journal of Money, Credit and Banking, Vol. 38, 1575-597.
- Buchanan, Claudia M., Sandra Eickmeier, and Esteban Prieto, 2011, “In Search for Yield? New Survey-Based Evidence on Bank Risk Taking,” CESIfo Working Paper No. 3375, March 2011. (listed as Buch in source)

### Theoretical contributions and systemic risk
- Diamond, Douglas, and Raghuram Rajan, 2012, “Illiquid Banks, Financial Stability, and Interest Rate Policy,” Journal of Political Economy, Vol. 120, pp. 552-91.
- Farhi, Emmanuel, and Jean Tirole, 2012, “Collective Moral Hazard, Maturity Mismatch and Systemic Bailouts,” American Economic Review, Vol. 102, pp. 60-93.
- Freixas, Xavier, Antoine Martin, and David R. Skeie, 2011, “Bank Liquidity, Interbank Markets, and Monetary Policy,” Review of Financial Studies, Vol. 24, No. 8, pp. 2656-692.
- Stein, Jeremy, 2012, “Monetary Policy as Financial-Stability Regulation,” Quarterly Journal of Economics, Vol. 127, pp. 57-95.

### Empirical analyses, working papers, and surveys
- Carpenter, Seth B., William Whitesell, and Egon Zakrajšek, 2001, “Capital Requirements, Business Loans, and Business Cycles: An Empirical Analysis of the Standardized Approach in the New Basel Capital Accord,” Board of Governors of the Federal Reserve System, Finance and Economics Discussion Series: 2001-48.
- Buch, Claudia M., Sandra Eickmeier, and Esteban Prieto, 2011, “In Search for Yield? New Survey-Based Evidence on Bank Risk Taking,” CESIfo Working Paper No. 3375, March 2011.
- Paligorova, Teodora, and Joao A.C. Santos, 2012, “Monetary Policy and Bank Risk-Taking: Evidence from the Corporate Loan Market,” working paper, Federal Reserve Bank of New York.
- Black, Lamont K., and Lieu N. Hazelwood, 2013, “The Effect of TARP on Bank Risk-Taking,” Journal of Financial Stability, forthcoming.
- Jimenez, Gabriel, Steven Ongena, Jose Luis Peydro, and Jesus Saurina, 2008, “Hazardous Times for Monetary Policy: What Do Twenty-Three Million Bank Loans Say About the Effects of Monetary Policy on Credit Risk-Taking?” Banco de España documento de Trabajo 0833.

### Policy speeches, commentaries, and books
- Bernanke, Ben, 2002, “Asset-Price ‘Bubbles’ and Monetary Policy,” speech delivered at the New York Chapter of the National Association for Business Economics, New York, N.Y., October 15.
- Bernanke, Ben, 2011, “The Effects of the Great Recession on Central Bank Doctrine and Practice,” Keynote address at the Federal Reserve Bank of Boston 56th Economic Conference ‘Long-Term Effects of the Great Recession,’ Boston, October, pp. 18–19.
- Taylor, John, 2009, Getting Off Track: How Government Actions and Interventions Caused, Prolonged, and Worsened the Financial Crisis, Hoover Press.
- Mishkin, Frederic, 2010, “Monetary Policy Strategy: Lessons from the Crisis,” Prepared for the ECB Central Banking Conference, “Monetary Policy Revisited: Lessons from the Crisis,” Frankfurt, November 18-19.

### Figure caption excerpt
- Figure 1. Interest Rates and Bank Risk Taking
  - This figure plots the weighted average loan risk rating from the U.S. Survey of Terms of Business Lending against the nominal federal funds rate based on quarterly data from the second quarter of 1997 until the fourth quarter of

*Source: _wp13143 - References*

### 2011.  Both variables are detrended by subtracting their linear time trend. The blue line indicates a regression line

### _wp13143 - 2011.  Both variables are detrended by subtracting their linear time trend. The blue line indicates a regression line

### Summary statistics (Table 1)
- Observations: 994,287
- Number of banks: 475
- Reported averages and standard deviations:
  - Risk rating: Average 3.434, Standard deviation 0.850
  - Loan size (dollars): Average 512,481, Standard deviation 4,992,014
  - Dummy for loans secured by collateral: Average 0.809, Standard deviation 0.393
  - Loan maturity (years): Average 1.238, Standard deviation 1.768
  - Bank total assets (millions of dollars): Average 191,624, Standard deviation 279,194
  - Tier 1 capital ratio: Average 0.094, Standard deviation 0.027
  - Total capital ratio: Average 0.122, Standard deviation 0.023
  - Common stock to assets ratio: Average 0.003, Standard deviation 0.008
  - Equity to assets ratio: Average 0.101, Standard deviation 0.026
  - Target federal funds rate (%): Average 2.206, Standard deviation 1.853
  - State personal income (dollars): Average 312,407, Standard deviation 359,049
  - Change in region CPI (%): Average 0.596, Standard deviation 0.858
  - State unemployment rate (%): Average 5.830, Standard deviation 2.169
  - Change in state housing prices (%): Average 0.500, Standard deviation 2.198

### Main regression results: Federal funds rate and loan risk (Table 2; Tables 3 Panel A & B)
- Core finding across specifications:
  - Target federal funds rate coefficient frequently negative and statistically significant:
    - Table 2 (specifications (1)–(6)): coefficients reported as -0.016***, -0.016**, -0.021***, -0.021***, -0.031***, -0.031*** with standard errors in brackets [0.000], [0.007], [0.006], [0.006], [0.009], [0.008]
  - Table 3 Panel A (columns (1)–(5)): Target federal funds rate = -0.031*** (standard error [0.008]) across columns
- Loan-level controls (Table 3 Panel A):
  - Loan size (log): negative significant effect where reported; e.g., Loan size coefficient -0.031*** with [0.002] in one specification and -0.024*** with [0.002] in another
  - Dummy for secured loans: positive and significant; e.g., 0.241*** [0.008] and 0.227*** [0.008]
  - Loan maturity: negative and significant; e.g., -0.022*** [0.002], -0.018*** [0.001]
- Heterogeneity by loan characteristics (Table 3 Panel B):
  - Target federal funds rate coefficients by subsample:
    - Loan size above median: -0.030*** [0.007]
    - Loan size below median: -0.034*** [0.010]
    - Loans secured by collateral: -0.027*** [0.009]
    - Loans not secured by collateral: -0.055*** [0.005]
    - Loans with maturity longer than median: -0.032*** [0.008]
    - Loans with maturity shorter than median: -0.028*** [0.009]
  - Observations by subsample: 520,258; 474,029; 804,163; 190,124; 510,972; 483,315
  - R-squared values reported (Panel A examples): 0.170 to 0.183 across columns

### Bank capital interactions (Table 4; Panels A–C; Table 5)
- Interaction effects between capital ratios and federal funds rate (Table 4 Panel A):
  - Tier 1 capital ratio × target federal funds rate: -0.317*** [0.082]
  - Total capital ratio × target federal funds rate: -0.239*** [0.077]
  - Common stock-assets ratio × target federal funds rate: -0.541*** [0.127]
  - Stock market capitalization-assets ratio × target federal funds rate: -0.077** [0.034]
- Panel B (split by Tier 1 capital above vs below median):
  - Banks with Tier 1 above median: Target federal funds rate -0.048*** [0.016]; Observations 490,064; Number of banks 455; R2 0.227
  - Banks with Tier 1 below median: Target federal funds rate -0.019*** [0.004]; Observations 504,223; Number of banks 184; R2 0.155
- Panel C (quarter-year fixed effects):
  - Tier 1 capital ratio: 1.109*** [0.372]
  - Tier 1 capital ratio × target federal funds rate: -0.389*** [0.081]
  - Total capital ratio: 1.670*** [0.357]
  - Total capital ratio × target federal funds rate: -0.279*** [0.085]
  - Common stock-assets ratio × target federal funds rate: -0.313* [0.157]
- Loans not under commitment (Table 5):
  - Sample size: Observations 743,665 (various specifications)
  - Target federal funds rate coefficients vary: -0.016*** [0.004], 0.005 [0.008], 0.023** [0.009], -0.013*** [0.004], 0.057*** [0.015]
  - Tier 1 capital ratio × target federal funds rate: -0.236*** [0.079] (specification 2)
  - Total capital ratio × target federal funds rate: -0.327*** [0.067] (specification 3)
  - Common stock-assets ratio × target federal funds rate: -1.194*** [0.300] (specification 4)
  - Market capitalization ratio × target federal funds rate: -0.162*** [0.043] (specification 5)

### Results for small banks and states without large banks (Tables 6–7)
- Small banks (Table 6; assets below top quintile; Observations 217,513; Number of banks 414; R2 0.305):
  - Tier 1 capital ratio × target federal funds rate: -0.396*** [0.119]
  - Total capital ratio × target federal funds rate: -0.260** [0.120]
  - Common stock-assets ratio × target federal funds rate: -0.824** [0.310]
  - Target federal funds rate coefficient examples: -0.020* [0.010], 0.022 [0.014], 0.01 [0.016], -0.016 [0.010]
- States without large banks (Table 7; Observations 598,017; Number of banks 443; R2 0.214):
  - Target federal funds rate: -0.061*** [0.014] (specification 1)
  - Tier 1 capital ratio × target federal funds rate: -0.209* [0.110]
  - Total capital ratio × target federal funds rate: -0.219* [0.113]
  - Common stock-assets ratio × target federal funds rate: -0.387* [0.221]

### State cyclicality, regional monetary policy, and macro conditions (Tables 8, 10–11)
- State cyclicality (Table 8):
  - Target federal funds rate coefficients differ by state correlation with U.S. GDP and volatility:
    - Example: States with high correlation with US GDP (col 1): -0.037** [0.014]; low correlation (col 2): -0.021*** [0.004]
  - Tier 1 capital ratio × target federal funds rate significant in several subsamples:
    - E.g., column with high income growth volatility: -0.349*** [0.095]; column with low income growth volatility for small banks: -0.436*** [0.151]
- Regional monetary policy conditions (Table 10):
  - Regional measure: Target federal funds rate minus change in region CPI: coefficient -0.031*** [0.008] (col 1)
  - Deviation from regional Taylor rule: -0.031*** [0.008] (col 3)
  - Interactions:
    - Tier 1 capital ratio × (Target federal funds rate minus change in region CPI): -0.274*** [0.083]
    - Tier 1 capital ratio × Deviation from regional Taylor rule: -0.217*** [0.080]
- State of the economy (Table 11):
  - Target federal funds rate × NBER recession dummy: negative and significant in multiple specifications; example -0.028** [0.013]
  - Target federal funds rate × GDP growth coefficients reported (not significant): -0.003 [0.002] in multiple columns
  - Tier 1 capital ratio × target federal funds rate: -0.329*** [0.084] (specification including interactions)

### Bank distress, interest-rate spell length, and episodic results (Tables 9, 12)
- Periods of bank distress (Table 9):
  - Split sample crisis years vs non-crisis years (2008–2010 as crisis window):
    - Crisis years: Target federal funds rate examples -0.031* [0.016], -0.042* [0.022]
    - Tier 1 capital ratio × target federal funds rate positive in some crisis specifications (e.g., 0.263* [0.133]) and negative and significant in non-crisis specifications (e.g., -0.549*** [0.109])
  - Years with many bank failures vs few:
    - Tier 1 capital ratio × target federal funds rate reported as 0.063 [0.187] and -0.394*** [0.096] across columns (7)–(8)
- Long periods of low interest rates (Table 12):
  - Periods without increase in rates (number of consecutive quarters without increase) coefficients:
    - Specification (1): 0.005* [0.002]
    - Specification (2): -0.001 [0.003]
    - Specification (3): -0.007* [0.003]
    - Specification (4): 0.006* [0.003]
  - Periods without increase in rates × tier 1 capital ratio:
    - Reported as 0.058* [0.029] (specification 2) and 0.080** [0.033] (specification 3)
  - Periods without increase in rates × target federal funds rate: -0.004** [0.002] (specification 3)
  - Target federal funds rate main effect reported as -0.031*** [0.008] (specification 1) and -0.024*** [0.008] (specification 2)

### Key empirical patterns (synthesis of reported coefficients and subsample results)
- In many specifications higher target federal funds rates are associated with lower reported loan risk ratings (negative coefficients such as -0.031*** [0.008], -0.016*** [0.004], etc.).
- Interaction terms typically show that higher bank capital (Tier 1 and total capital ratios, common stock-assets ratio, market capitalization-assets ratio) amplify the negative association between the federal funds rate and loan risk ratings (examples: Tier 1 × rate -0.317*** [0.082]; Total capital × rate -0.239*** [0.077]; Common stock-assets × rate -0.541*** [0.127]).
- Heterogeneity: the negative association between target federal funds rate and loan risk ratings is present across loan sizes, collateralization status, maturities, but magnitudes differ (e.g., not secured loans: -0.055*** [0.005]).
- Subsamples indicate varying sensitivity during crisis years, in states with different cyclical properties, among small banks, and depending on length of low-rate spells.

*Italicized source: _wp13143 - 2011.  Both variables are detrended by subtracting their linear time trend. The blue line indicates a regression line*

### Appendix Table 1. Frequency of Keyw

### Appendix Table 1. Frequency of Keywords Appearing in FOMC Minutes

### Overview
- Table reports keyword usage in FOMC meetings across three time spans:
  - 1997Q2—2011Q4 (total counts)
  - 1997Q2—2006Q4
  - 2007Q1—2011Q4
- For each span the table provides:
  - Conservative = the number of reports the word appears in (if a word appears several times in a report, that's not counted).
  - Liberal = the total number of times the word appears in the reports.
- Frequency = number of times a word has been used within a time period divided by the number of quarters in that time period.

### Keyword frequencies and counts
- Bank risk
  - # of times used 1997Q2—2011Q4: Conservative 0, Liberal 0
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 0, Liberal 0
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.000, Liberal 0.000

- Banking risk
  - # of times used 1997Q2—2011Q4: Conservative 0, Liberal 0
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 0, Liberal 0
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.000, Liberal 0.000

- Banking sector
  - # of times used 1997Q2—2011Q4: Conservative 10, Liberal 14
  - # of times used 1997Q2—2006Q4: Conservative 1, Liberal 1
  - # of times used 2007Q1—2011Q4: Conservative 9, Liberal 13
  - Frequency 1997Q2—2006Q4: Conservative 0.026, Liberal 0.026
  - Frequency 2007Q1—2011Q4: Conservative 0.450, Liberal 0.650

- Banking system
  - # of times used 1997Q2—2011Q4: Conservative 15, Liberal 19
  - # of times used 1997Q2—2006Q4: Conservative 3, Liberal 3
  - # of times used 2007Q1—2011Q4: Conservative 12, Liberal 16
  - Frequency 1997Q2—2006Q4: Conservative 0.077, Liberal 0.077
  - Frequency 2007Q1—2011Q4: Conservative 0.600, Liberal 0.800

- Condition of the banking system
  - # of times used 1997Q2—2011Q4: Conservative 2, Liberal 2
  - # of times used 1997Q2—2006Q4: Conservative 2, Liberal 2
  - # of times used 2007Q1—2011Q4: Conservative 0, Liberal 0
  - Frequency 1997Q2—2006Q4: Conservative 0.051, Liberal 0.051
  - Frequency 2007Q1—2011Q4: Conservative 0.000, Liberal 0.000

- Financial conditions
  - # of times used 1997Q2—2011Q4: Conservative 112, Liberal 351
  - # of times used 1997Q2—2006Q4: Conservative 74, Liberal 187
  - # of times used 2007Q1—2011Q4: Conservative 39, Liberal 167
  - Frequency 1997Q2—2006Q4: Conservative 1.897, Liberal 4.795
  - Frequency 2007Q1—2011Q4: Conservative 1.950, Liberal 8.350

- Financial stability
  - # of times used 1997Q2—2011Q4: Conservative 14, Liberal 17
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 14, Liberal 17
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.700, Liberal 0.850

- Financial system
  - # of times used 1997Q2—2011Q4: Conservative 11, Liberal 19
  - # of times used 1997Q2—2006Q4: Conservative 1, Liberal 2
  - # of times used 2007Q1—2011Q4: Conservative 10, Liberal 17
  - Frequency 1997Q2—2006Q4: Conservative 0.026, Liberal 0.051
  - Frequency 2007Q1—2011Q4: Conservative 0.500, Liberal 0.850

- Health of the banking system
  - # of times used 1997Q2—2011Q4: Conservative 0, Liberal 0
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 0, Liberal 0
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.000, Liberal 0.000

- Risks to the financial system
  - # of times used 1997Q2—2011Q4: Conservative 1, Liberal 1
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 1, Liberal 1
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.050, Liberal 0.050

- Stability of the financial system
  - # of times used 1997Q2—2011Q4: Conservative 2, Liberal 3
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 2, Liberal 3
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.100, Liberal 0.150

- Systemic
  - # of times used 1997Q2—2011Q4: Conservative 2, Liberal 4
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 2, Liberal 4
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.100, Liberal 0.200

- Systemic risk
  - # of times used 1997Q2—2011Q4: Conservative 0, Liberal 0
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 0, Liberal 0
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.000, Liberal 0.000

- Troubles of the banking system
  - # of times used 1997Q2—2011Q4: Conservative 1, Liberal 1
  - # of times used 1997Q2—2006Q4: Conservative 0, Liberal 0
  - # of times used 2007Q1—2011Q4: Conservative 1, Liberal 1
  - Frequency 1997Q2—2006Q4: Conservative 0.000, Liberal 0.000
  - Frequency 2007Q1—2011Q4: Conservative 0.050, Liberal 0.050

### Notes (as reported in the table)
- Frequency is determined as the number of times a word has been used within a time period divided by the number of quarters in that time period.
- Conservative = the number of reports the word appears in (if a word appears several times in a report, that's not counted).
- Liberal = the total number of times the word appears in the reports.
- Source: FOMC Minutes

*Source: FOMC Minutes*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13143.pdf_