## wpiea2025096-print-pdf ## Source details **Canonical URL:** [wpiea2025096-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025096-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025096-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025096-print-pdf.pdf.json) --- ### Baseline specification (Sections 3.1–3.5) - Regression framework (Equation (2)): - Log(Loans)_{l,i,t} = β1 MP_{l,t−1} + β2 MaPP_{l,t−1} + β3 MP_{l,t−1} × Nonbank_{l} + β4 MaPP_{l,t−1} × Nonbank_{l} + γ_{l} + μ_{i,t} + ε_{l,i,t} - Dependent variables: - Logarithm of the US dollar amount of new syndicated loans - Margin spread over LIBOR (in some specifications) - Key regressors and identification: - MP_{l,t−1} and MaPP_{l,t−1}: lagged MP and MaPP shocks at the lender-country level - Nonbank_{l}: dummy equal to one for nonbanks and zero for banks - Main coefficients of interest: β3 and β4 (differential lending from nonbanks relative to banks following contractionary MP and MaPP shocks) - Preferred controls: lender characteristics γ_{l} and firm×quarter fixed effects μ_{i,t} - Identification relies on a firm borrowing from at least one bank and one nonbank in a given quarter - Alternative demand controls: country×industry×quarter FE or ILST FE (industry-location-size-time) - Standard errors and clustering: - Cluster standard errors by firm - Results robust to alternative clustering combinations - Pricing specifications: - Years to maturity included when examining spread effects - Caveat: pricing data is less populated in Dealogic - Relationship lending (Equation (3)): - Log(Loans)_{l,i,t} = β1 MP_{l,t−1} + β2 MaPP_{l,t−1} + β3 MP_{l,t−1} × Nonbank_{l} + β4 MaPP_{l,t−1} × Nonbank_{l} + Relation_{l,i,t−1} × (β5 + β6 Nonbank_{l} + β7 MP_{l,t−1} + β8 MaPP_{l,t−1} + β9 MP_{l,t−1} × Nonbank_{l} + β10 MaPP_{l,t−1} × Nonbank_{l}) + γ_{l} + μ_{i,t} + ε_{l,i,t} - Relation_{l,i,t−1}: dummy = 1 if firm i borrowed from lender l in past five years; alternative: loan duration = log years since first loan - Timing: MP and MaPP shocks lagged by one quarter for comparability - Deal-level aggregation and bank-characteristic interactions (Equation (4)): - NBshr_{d,t} = β1 MP_{d,t−1} + β2 MaPP_{d,t−1} + Weak bank_{d,t−1} × (β3 + β4 MP_{d,t−1} + β5 MaPP_{d,t−1}) + α_{i} + ILST FE + ε_{d,t} - NBshr_{d,t}: share of loans from nonbanks in deal d at time t - Weak bank_{d,t−1}: dummy for syndicates with weaker banks (loan-weighted Tier 1 capital ratio in bottom quartile or NPLs top quartile); time-varying - MP_{d,t} and MaPP_{d,t}: deal-level shocks weighted by participating banks’ relative loan shares - Risk-augmented specification (Equation (5)): - NBshr_{d,t} = β1 MP_{d,t−1} + β2 MaPP_{d,t−1} + Risky_{i,t−1} × (β3 + β4 MP_{d,t−1} + β5 MaPP_{d,t−1}) + α_{i} + ILST FE + ε_{d,t} - Risky_{i,t−1}: dummy = 1 if firm i is a risky borrower at t−1 - Four proxies for risky borrowers: - (i) 5y avg. spread top quartile (within years-to-maturity buckets) - (ii) Leveraged loans: spread > 150 bps over Libor - (iii) High-PD: PD in top quartile of country-time distribution - (iv) Junk rating: S&P long-term issuer rating below BBB− - Firm-level real effects (Equation (6)): - Y_{i,t} = β1 MP_{c,t−1} + β2 MaPP_{c,t−1} + NBrelation_{i,t−1} × (β3 + β4 MP_{c,t−1} + β5 MaPP_{c,t−1}) + α_{i} + μ_{c,s,t} + ε_{i,t} - Y_{i,t}: Log total real debt; Net liquid assets (log); Log tangible investment; Log intangible investment; Log total employment; PD next 12 months - NBrelation_{i,t−1}: dummy = 1 if firm borrowed from nonbanks in previous five years - MP_{c,t−1} and MaPP_{c,t−1}: loan-weighted averages of lender-country shocks (weights = share of loans firm borrowed from each lender) - Controls: borrower FE α_{i}, country-industry-time FE μ_{c,s,t} ### Main results (Section 4.1) - Effects of monetary policy (MP) shocks: - Contractionary MP shocks lead to a 7.8 percent decline in global syndicated loans to nonfinancial firms (Column (1), Table 1). - With firm×time FE (Column (5), Table 1): a one-standard deviation MP tightening shock leads to a 2.3 percent decline in bank credit supply; nonbanks increase lending by 2.6 percent (nonbanks increase lending by 4.9 percent relative to banks). - Selected coefficients: - MP shock (Table 1, Col (1)): -0.078 ∗∗∗ (standard error 0.004). - MP shock×Nonbank (Table 1, Col (1)): 0.057 ∗∗∗ (standard error 0.004). - MP shock (Table 3, Col (4)): -0.021 ∗∗∗ (standard error 0.003). - MP shock×Nonbank (Table 3, Col (4)): 0.046 ∗∗∗ (standard error 0.003). - Pricing effects (Table 1, Col (6)): - MP shock increases spread by 0.189 ∗∗ (standard error 0.089). - MP shock×Nonbank (Col (6)): -0.473 ∗∗∗ (standard error 0.140) — nonbanks charge lower spreads relative to banks following MP tightening. - Interpretation: consistent with a credit lending channel of monetary policy; nonbanks partially offset bank lending contraction during MP tightenings. - Effects of macroprudential policy (MaPP) shocks: - One-standard deviation MaPP tightening shock: banks cut credit supply by 2.0 percent; nonbanks increase lending by 2.5 percent relative to banks (absolute nonbank response muted). - Selected coefficients: - MaPP shock (Table 2, Col (1)): -0.040 ∗∗∗ (standard error 0.003). - MaPP shock×Nonbank (Table 2, Col (1)): 0.036 ∗∗∗ (standard error 0.004). - MaPP shock (Table 3, Col (4)): -0.019 ∗∗∗ (standard error 0.002). - MaPP shock×Nonbank (Table 3, Col (4)): 0.016 ∗∗∗ (standard error 0.002). - Pricing effects (Table 2, Col (6)): - MaPP shock increases spread by 0.144 ∗∗ (standard error 0.065). - MaPP shock×Nonbank (Col (6)): -0.163 (standard error 0.127), less precise. - Interpretation: findings align with theory that tighter capital requirements reduce bank deposits and can lower nonbanks’ debt financing costs, prompting credit reallocation. - Combined MP and MaPP shocks: - Results robust when both shocks included (Table 3, Col (4) preferred): - Nonbanks expand lending by 4.6 percent relative to banks during MP shocks, compared to 1.6 percent during MaPP shocks. - In absolute terms, nonbank lending expands by 2.5 percent during MP shocks but is muted during MaPP shocks. - Selected coefficients (Table 3, Col (4)): - MP shock: -0.021 ∗∗∗ (standard error 0.003). - MaPP shock: -0.019 ∗∗∗ (standard error 0.002). - MP shock×Nonbank: 0.046 ∗∗∗ (standard error 0.003). - MaPP shock×Nonbank: 0.016 ∗∗∗ (standard error 0.002). - Spread (Table 3, Col (5)): - MP shock 0.166 ∗ (standard error 0.089); MaPP shock 0.124 ∗ (standard error 0.065) - MP shock×Nonbank -0.474 ∗∗∗ (standard error 0.157); MaPP shock×Nonbank -0.058 (standard error 0.140). - Robustness: - Positive/negative shock coefficients similar (Tables OA.2 and OA.3). - Robust to restricting to selected AEs (Table OA.4) and to controlling MP–MaPP endogeneity (pairwise correlation 0.0844; Column (7) Table OA.9). - Normalization via bank stock prices: - Method: regress log change of real bank stock prices on MP and MaPP shocks; normalize shocks so a normalized shock corresponds to a one percent decline in bank stock prices. - Normalized results (preferred specification): - MP shock calibrated to decrease bank stock prices by one percent → 5.5 percent relative increase in nonbank lending. - MaPP shock calibrated to decrease bank stock prices by one percent → 4.0 percent relative increase in nonbank lending. - Difference statistically significant at the five percent level (when normalized with bank stock prices). - Interaction and coordination between MP and MaPP: - Aggregate interaction (Table 4): MaPP and MP shocks complement each other in reducing credit supply when not distinguishing lender type; one-standard deviation increase in both shocks leads to additional 0.5 percent decline in lending to nonfinancial corporations. - By lender type (Table 4, Col (2)): - Tightening both policies mitigates credit contraction from banks by 0.3 percent (coefficient flips positive when isolating banks). - Expansion of nonbank credit is mitigated by 1.4 percent when both policies tighten (triple interaction). - Combined effect for nonbanks ~4.2 percent (reported as 4.2 = 4.6 + 1.0 - 1.4). - Tightening-loosening combinations (Table 4, Col (5) selected): - MP+ × MaPP+ on banks: 0.002 (standard error 0.003). - MP+ × MaPP− on banks: -0.006, -0.004, -0.005 (standard errors in table). - MP− × MaPP+ on banks: -0.014 ∗∗∗, -0.012 ∗∗∗, -0.007 ∗∗ (standard errors reported). - MP+ × MaPP+ × Nonbank: -0.033 ∗∗∗, -0.029 ∗∗∗, -0.023 ∗∗∗. - MP− × MaPP+ × Nonbank: 0.019 ∗∗, 0.020 ∗∗∗, 0.023 ∗∗∗. - Interpretation: divergence in policy direction typically leads to additional contraction in bank lending; coordinated tightening can deter nonbanks, while divergence can amplify nonbank expansion. - Key takeaways: 1. Differential increase in nonbank credit supply during contractionary MP shocks found in US and Denmark generalizes to a sample of 22 countries. 2. Nonbank lending expansion also occurs when MaPP tightens, indicating shocks that constrain banks shift credit to nonbanks. 3. Divergence in MP and MaPP amplifies credit reallocation to nonbanks; coordinated tightening can mitigate but not necessarily eliminate nonbank expansion. - Policy implication: - Traditional MaPPs focused only on banks may not stop credit leakage to nonbanks during contractionary MP shocks. - Extending regulatory perimeter to nonbanks could help limit credit leakages, improve MP transmission, and bolster financial resilience (text cites Abbas et al. 2025). ### Relationship lending and nonbanks (Section 4.2) - Concepts: - Relationship lending accrues information over time and can smooth credit allocation during downturns. - Benefits appear largely confined to bank relationships, but nonbanks can also shield borrowers during MP shocks in some contexts. - Empirical strategy: - Relation (pre-relation dummy) = 1 if borrower borrowed from lender in past five years; alternative: Duration = log years since first loan - Estimate Equation (3) with interactions between shocks, Nonbank, and Relation/Duration - Standard errors clustered by firm - Main empirical findings (selected estimates and magnitudes): - Direct MP shock coefficients (various columns): -0.034, -0.037, -0.035, -0.031, -0.033, -0.035 (all ∗∗∗). SEs reported. - Direct MaPP shock coefficients (various): -0.023, -0.021, -0.022, -0.021, -0.020, -0.021 (all ∗∗∗). - Relation main effects: - Pre-relation Relation coefficients: 0.024, 0.062, 0.077 (all ∗∗∗). - Duration Relation coefficients: 0.020, 0.031, 0.031 (all ∗∗∗). - Interaction highlights: - MP shock × Nonbank: 0.037–0.044 (∗∗∗ across columns). - MaPP shock × Nonbank: 0.005–0.015 (several ∗∗∗ depending on column). - MP shock × Relation: 0.030–0.033 (columns (1)-(3), ∗∗∗) and 0.009–0.012 (columns (4)-(6), ∗∗∗). - MP shock × Nonbank × Relation: 0.045, 0.034, 0.032 (cols (1)-(3)) and 0.018, 0.016, 0.014 (cols (4)-(6)) — all ∗∗∗. - MaPP shock × Nonbank × Relation: mixed evidence (some ∗∗, some small or negative). - Economic illustrations: - Firms with prior bank relationship offset nearly 94 percent of decline in credit supply following MP tightening: 0.033/0.035. - Firms with prior bank relationship offset about one-fourth of decline during MaPP tightening: 0.006/0.022. - Sample size and fit: - Observations across columns: 749,629; 748,453; 739,866; 749,629; 748,453; 739,866. - R2: 0.727, 0.795, 0.877, 0.727, 0.795, 0.877. - Interpretation: - Relationship lending with banks largely smooths credit contractions from MP and MaPP tightenings; nearly full offset for MP tightening, partial for MaPP. - Relationship lending with nonbanks is important during MP shocks: triple interaction implies firms with existing nonbank relationships borrow 3.2–4.5 percent more compared to first-time nonbank borrowers when MP tightens. - Pricing: increased nonbank lending during shocks may come at higher spreads for borrowers with existing relationships, but during MP shocks borrowers from nonbanks still benefit from a decline in spreads relative to banks (see Table OA.7). ### Bank characteristics and credit migration to nonbanks (Section 4.3) - Main finding: - Nonbank credit increases relative to bank credit following contractionary MP and MaPP shocks; migration stronger in syndicates with weaker banks. - Evidence on nonbank share and weak banks (Table 6): - Nonbank share higher in syndicates with weaker banks. - Preferred specification (column (2), low bank capital): nonbank loan share rises by an additional 1.3 and 1.7 percentage points following a one-standard-deviation increase in MP and MaPP shocks, respectively, compared to syndicates with stronger banks. - Selected coefficients (columns 1–2 examples): - MP shock: 0.013 (0.006) col (1); 0.017 (0.007) col (2). - MaPP shock: 0.021 (0.005) col (1); 0.031 (0.005) col (2). - MP shock×Bank charact.: 0.011 (0.006) col (1); 0.013 (0.007) col (2). - MaPP shock×Bank charact.: 0.016 (0.004) col (1); 0.017 (0.005) col (2). - Observations: 47,615; 42,387; 47,615; 42,387. R2: 0.657–0.678. - Credit-risk migration to nonbanks (Table 7): - Nonbanks lend more to risky borrowers on average. - Contractionary MP and MaPP shocks do not cause nonbanks to disproportionally increase lending to risky borrowers within syndicates. - MaPP shock coefficients often positive and significant across risk proxies (examples): - MaPP shock: 0.053 (0.007) ***, 0.062 (0.007) ***, 0.055 (0.008) ***, 0.065 (0.008) *** (various columns). - Risky (unconditional): coefficients positive in several proxies (examples: 0.011 (0.006) *, 0.016 (0.006) ***, 0.061 (0.015) ***). - Observations range from 8,936 to 36,356; R2 range 0.635 to 0.817. - Real effects on firms (Table 8): - Firms with NB relation (borrowed from at least one nonbank in past five years) have higher debt, investment, employment, and higher PDs unconditionally. - After contractionary shocks, NB-relation firms show: - Higher capital expenditures and employment relative to firms without NB relations. - Higher intangible investment in response to MaPP shocks (statistically significant). - No evidence that NB relations reduce borrower riskiness after shocks; PDs appear to increase, especially during MaPP shocks. - Selected coefficients (standard errors in parentheses): - MP shock: 0.055 (0.045) Total Debt; 0.056 ** (0.028) Intangibles. - MaPP shock: -0.070 ** (0.032) Total Debt; -0.079 ** (0.035) Liquid assets; -0.079 *** (0.019) Capital exp.; -0.067 *** (0.021) Intangibles; -0.074 *** (0.018) Employment. - NB relation (unconditional): 0.166 *** (0.024) Total Debt; 0.105 *** (0.013) Capital exp.; 0.071 *** (0.015) Intangibles; 0.060 *** (0.010) Employment; 0.111 *** (0.026) PD. - MP shock×NB relation: 0.039 * (0.022) Capital exp.; 0.053 *** (0.019) Employment. - MaPP shock×NB relation: 0.033 ** (0.013) Capital exp.; 0.040 ** (0.018) Intangibles; 0.034 *** (0.010) Employment; 0.049 ** (0.022) PD. - Observations by dependent variable reported across columns; R2 range 0.611 to 0.982. - Bank lending to nonbanks and funding channel (Table 9): - Banks increase lending to nonbank borrowers relative to nonfinancial corporations after MaPP shocks: estimated differential 0.8–1.5 percent more loans to nonbanks following a one-standard-deviation MaPP shock, depending on demand controls. - No consistent evidence banks reallocate more to nonbanks following contractionary MP shocks. - Low-capitalized banks (LC bank) drive MaPP result: weaker banks increase lending to nonbanks during MaPP shocks. - Selected coefficients (standard errors): - MaPP shock: -0.020 *** (0.002) col (1); -0.018 *** (0.002) col (2); -0.015 *** (0.002) col (3); -0.014 *** (0.003) col (4). - MaPP shock×NB Borrower: 0.015 ** (0.007); 0.011 ** (0.006); 0.008 (0.005); -0.012 (0.011). - MaPP shock×NB Borrower×LC bank: 0.029 ** (0.013) col (4). - Observations: 547,476; 545,765; 533,115; 218,587. R2: 0.757–0.889. - Robustness and heterogeneity: - Results driven by nonbanks with unstable funding (investment banks, broker-dealers, finance companies, private equity, venture capital, hedge funds); pension funds and insurance companies classified as stable funding. - Robust across loan types (term loans vs credit lines), regression-based allocation of missing loan shares, extensive margin, exclusion of top three nonbanks, alternative MaPP constructions, lag structures, fixed effects, and clustering methods. - Splits: pre-GFC vs post-GFC, USD vs non-USD loans, cross-border vs domestic lending, AEs vs EMDEs — baseline results persist. - Mechanisms: - MP shocks: deposit outflows and widening policy–deposit rate spread constrain bank funding; nonbanks attract depositor funds and expand lending (deposits→nonbanks); some nonbanks increase borrowing from banks (credit line drawdowns). - MaPP shocks: tighter bank regulation constrains banks near regulatory limits, prompting reallocation to nonbanks; preferential regulatory treatment for exposures to nonbanks can incentivize bank shifts. - Financial stability implications and policy considerations: - Benefits: nonbanks cushion firms from credit contractions, supporting capex and employment. - Risks: nonbanks lend more to riskier borrowers, rely on unstable funding, are less regulated, often more procyclical, and lack central bank liquidity access. - Policy recommendations: - Close data gaps on nonbank sector. - Consider extending macroprudential regulations/perimeter to nonbanks (leverage limits, capital and liquidity requirements, stress-testing). - Monitor private credit markets (direct lending) where nonbanks have larger, less transparent roles. ### Appendix A: Data (selected) - List of lender countries (Table A.1) — borrower-country share (average loan share in estimation sample): - Austria AE 0.35% - Belgium AE 0.74% - Brazil EMDE 0.49% - Canada AE 6.70% - Cyprus AE 0.004% - Finland AE 0.12% - France AE 7.71% - Germany AE 6.98% - Greece AE 0.12% - India EMDE 1.54% - Ireland AE 0.44% - Italy AE 1.95% - Japan AE 11.13% - Lithuania AE 0.03% - Netherlands AE 3.71% - Norway AE 0.70% - Portugal AE 0.18% - Slovenia AE 0.01% - Spain AE 2.74% - Sweden AE 1.02% - United Kingdom AE 11.35% - United States AE 43.68% - Loan characteristics: nonbanks vs banks (Table A.2) — selected summary statistics: - Nonbanks: - Term length (years): Mean 5.13, STD 3.48, P25 3.00, P50 5.00, P75 6.00 - Tranche value ($ million): Mean 311.46, STD 524.41, P25 40.73, P50 121.84, P75 338.30 - Syndicate members: Mean 7.32, STD 6.74, P25 3.00, P50 5.00, P75 9.00 - Participation share: Mean 0.18, STD 0.19, P25 0.06, P50 0.10, P75 0.25 - All-in-pricing (BPs): Mean 266.86, STD 162.66, P25 150.00, P50 250.00, P75 350.00 - Banks: - Term length (years): Mean 5.00, STD 3.92, P25 3.00, P50 5.00, P75 6.00 - Tranche value ($ million): Mean 252.40, STD 472.00, P25 26.10, P50 86.30, P75 257.51 - Syndicate members: Mean 6.44, STD 6.19, P25 3.00, P50 5.00, P75 8.00 - Participation share: Mean 0.17, STD 0.19, P25 0.05, P50 0.10, P75 0.25 - All-in-pricing (BPs): Mean 208.32, STD 139.42, P25 102.50, P50 187.50, P75 275.00 - Figures on shocks and correlation: - Figure A.1: MP shocks median and IQR, 2000q1–2020q1. - Figure A.2: MaPP shocks median and IQR, 2000q1–2020q1. - Figure A.3: Correlation of MP and MaPP shocks — linear regression line; correlation 0.0844. ### Appendix B: Selected tables (high-level highlights) - Funding models of nonbanks (Table B.1): - MP shock: -0.021*** (0.003) to -0.021*** (0.002) across columns. - MaPP shock: -0.020*** (0.002) to -0.019*** (0.002). - MP shock × Stable Nonbank: -0.119*** (0.027) to -0.097*** (0.020). - MP shock × Unstable Nonbank: 0.060*** (0.004) to 0.049*** (0.003). - Observations: 740,206; 739,026; 730,335. R^2: 0.726; 0.794; 0.877. - Loan types and robustness (Table B.2): - Baseline MP shock: -0.021*** (0.002); MP shock × Nonbank: 0.046*** (0.003). - MaPP shock × Nonbank: 0.016*** (0.002). - Observations across columns vary (e.g., 739,866; 348,427; 522,995; 2,014,048). R^2 range up to 0.933 in ext. margin column. - Alternative MaPP constructions (Table B.3): - MP shock × Nonbank stable ≈ 0.046–0.047*** (0.003). - MaPP shock × Nonbank varies with construction: 0.012*** (0.002) to 0.026*** (0.003). - Split-sample robustness (Table B.4): - MP shock × Nonbank positive and significant across many splits (USD, Non-USD, Cross-border, Domestic, AE, EMDE). - Observations and R2 reported for each split (e.g., Observations 739,866; 276,509; 403,604; R2 0.877; 0.854; 0.893). - Online appendix tables (OA.2–OA.10) provide asymmetric effects, normalization with bank stock prices, alternative lag structures, clustering methods, and sample distribution across MP/MaPP combinations (Table OA.6 counts: MP+ × MaPP+ 211,198; MP+ × MaPP− 152,032; MP− × MaPP+ 199,881; MP− × MaPP− 176,755; Total 739,866). _Italic: Source: wpiea2025096-print-pdf (Working Paper No. WP/2025/096), Sections 3.1–3.5, 4.1–4.3, Appendices A–B._ ### 3.1 Baseline specification ### 3.1 Baseline specification ### Regression framework and variables - Main regression (Equation (2)): - Log(Loans)_{l,i,t} = β1 MP_{l,t−1} + β2 MaPP_{l,t−1} + β3 MP_{l,t−1} × Nonbank_{l} + β4 MaPP_{l,t−1} × Nonbank_{l} + γ_{l} + μ_{i,t} + ε_{l,i,t} - Dependent variables: - Logarithm of the US dollar amount of new syndicated loans - Margin spread over LIBOR (in some specifications) - Key regressors: - MP_{l,t−1} and MaPP_{l,t−1}: lagged MP and MaPP shocks at the lender-country level - Nonbank_{l}: dummy equal to one for nonbanks and zero for banks - Main coefficients of interest: β3 and β4 (differential lending from nonbanks relative to banks following contractionary MP and MaPP shocks) - Fixed effects and identification: - Preferred controls: lender characteristics γ_{l} and firm×quarter fixed effects μ_{i,t} (Khwaja and Mian (2008)) - Identification relies on a firm borrowing from at least one bank and one nonbank in a given quarter - Alternative controls for credit demand: country×industry×quarter fixed effects or industry-location-size-time (ILST) fixed effects (Degryse et al. 2019) - ILST fixed effects compare borrowers within the same two-digit industry, country, quarter, and similar size proxied by quartile bins of total borrowing volume within a country-year pair - Standard errors and clustering: - Cluster standard errors by firm - Results robust to alternative clustering combinations - Pricing specifications: - Years to maturity included as an additional control when examining spread effects - Caveat: pricing data is less populated in Dealogic ### 3.2 Relationship lending ### Augmented specification and interpretation - Augmented regression (Equation (3)): - Log(Loans)_{l,i,t} = β1 MP_{l,t−1} + β2 MaPP_{l,t−1} + β3 MP_{l,t−1} × Nonbank_{l} + β4 MaPP_{l,t−1} × Nonbank_{l} + Relation_{l,i,t−1} × (β5 + β6 Nonbank_{l} + β7 MP_{l,t−1} + β8 MaPP_{l,t−1} + β9 MP_{l,t−1} × Nonbank_{l} + β10 MaPP_{l,t−1} × Nonbank_{l}) + γ_{l} + μ_{i,t} + ε_{l,i,t} - Relation_{l,i,t−1}: - Dummy equal to one if firm i has borrowed from lender l in the past five years - Alternative measure: loan duration = logarithm of years since borrower first obtained a loan from a given lender (Banerjee et al. 2021) - Coefficient interpretations: - β7 and β8: differential effect of MP and MaPP shocks on lending by banks to borrowers with an established relationship vs other borrowers - β9 and β10: additional effect of shocks on loans to borrowers with a prior relationship with nonbanks - Total differential effect on lending to borrowers with a nonbank relationship relative to banks: - For MP shocks: β3 + β9 - For MaPP shocks: β4 + β10 - Timing/lags: - Both MP and MaPP shocks are lagged by one quarter for comparability - Justification: median difference between announcement and in-force date of macroprudential measures is one quarter for 28 EU countries (Budnik and Kleibl (2018)) - Sensitivity analysis with alternative lag structures yields very similar results ### 3.3 Bank characteristics and credit migration to nonbanks ### Deal-level aggregation and testing - Aggregated specification (Equation (4)): - NBshr_{d,t} = β1 MP_{d,t−1} + β2 MaPP_{d,t−1} + Weak bank_{d,t−1} × (β3 + β4 MP_{d,t−1} + β5 MaPP_{d,t−1}) + α_{i} + ILST FE + ε_{d,t} - Variables and construction: - NBshr_{d,t}: share of loans from nonbanks in each syndicated loan deal d at time t - Weak bank_{d,t−1}: dummy for syndicates with weaker banks - Proxied by loan-weighted Tier 1 capital ratio (or NPLs) across banks in a syndicated loan deal falling in the bottom (top) quartile of the loan deal-time distribution - Time-varying: banks can transition between states - MP_{d,t} and MaPP_{d,t}: deal-level MP and MaPP shocks constructed as country-specific shocks for each participating bank weighted by relative loan shares of each lender in the syndicate - Coefficient interpretations: - β4 and β5: marginal credit migration within a syndicate from weak banks to nonbanks relative to syndicates with strong banks in response to contractionary MP and MaPP shocks - Positive β4 or β5 implies tighter MP or MaPP prompts migration of credit from weaker banks to nonbanks within syndicates - Fixed effects: - Preferred specification controls for time-invariant borrower characteristics α_{i} and ILST fixed effects for time-varying credit demand - Firm×time fixed effects are not used at deal level because they assume a firm borrows from multiple syndicates within the same quarter ### 3.4 Borrowers’ credit risk migration to nonbanks ### Risky borrower interactions and proxies - Risk-augmented specification (Equation (5)): - NBshr_{d,t} = β1 MP_{d,t−1} + β2 MaPP_{d,t−1} + Risky_{i,t−1} × (β3 + β4 MP_{d,t−1} + β5 MaPP_{d,t−1}) + α_{i} + ILST FE + ε_{d,t} - Risky_{i,t−1}: - Dummy equal to one if firm i is a risky borrower at t−1 - Coefficients of interest: - β4 and β5: whether nonbanks increase lending disproportionately within syndicates that include risky borrowers following contractionary shocks - Four proxies for risky borrowers: - (i) Borrowers with a loan-weighted average spread over the past five years in the top quartile of the sample distribution in each quarter - Note: average spread compared across years-to-maturity buckets to account for loan duration effects - (ii) Leveraged loans: borrowers whose spread is over 150 bps over Libor - (iii) High-PD borrowers: PD in the top quartile of the country-time distribution - (iv) Borrowers rated ‘speculative’ by S&P: long-term issuer credit rating below BBB− - Data considerations: - PDs and credit ratings have a more restricted sample size relative to spread-based Dealogic measures - PDs and ratings are common indicators of financial constraints ### 3.5 Real effects at the firm-level ### Firm-level outcomes and specification - Firm-quarter regression (Equation (6)): - Y_{i,t} = β1 MP_{c,t−1} + β2 MaPP_{c,t−1} + NBrelation_{i,t−1} × (β3 + β4 MP_{c,t−1} + β5 MaPP_{c,t−1}) + α_{i} + μ_{c,s,t} + ε_{i,t} - Dependent variables Y_{i,t} (firm-specific balance sheet and outcome measures): - Logarithm of total real debt (short- and long-term debt) - Net liquid assets (logarithm of total liquid assets minus short-term liabilities) - Logarithm of tangible investment - Logarithm of intangible investment - Logarithm of total employment - PD over the next 12 months - NBrelation_{i,t−1}: - Dummy equal to one if a firm borrowed from nonbanks in the previous five years - Shock construction: - MP_{c,t−1} and MaPP_{c,t−1} computed as loan-weighted average of lender-country shocks, weights equal to share of loans a firm borrowed from each lender - Controls: - Borrower fixed effects α_{i} - Country-industry-time fixed effects μ_{c,s,t} to control for credit demand _Italic: Source: wpiea2025096-print-pdf, Section 3.1–3.5 (Baseline specification, Relationship lending, Bank characteristics and credit migration to nonbanks, Borrowers’ credit risk migration to nonbanks, Real effects at the firm-level)._ ### 4.1 Main results ### 4.1 Main results ### Effects of monetary policy (MP) shocks - Contractionary MP shocks lead to a 7.8 percent decline in global syndicated loans to nonfinancial firms (Column (1), Table 1). - Controlling tightly for credit demand (firm×time fixed effects, Column (5), Table 1): - A one-standard deviation MP tightening shock leads to a 2.3 percent decline in bank credit supply. - Nonbanks increase lending by 2.6 percent (nonbanks increase lending by 4.9 percent relative to banks). - MP shock coefficients and interactions (selected estimates from Table 1 and Table 3): - MP shock (Table 1, Col (1)): -0.078 ∗∗∗ (standard error 0.004). - MP shock×Nonbank (Table 1, Col (1)): 0.057 ∗∗∗ (standard error 0.004). - MP shock (Table 3, Col (4) preferred specification): -0.021 ∗∗∗ (standard error 0.003). - MP shock×Nonbank (Table 3, Col (4)): 0.046 ∗∗∗ (standard error 0.003). - Price effects (Table 1, Column (6)): - MP shock increases spread by 0.189 ∗∗ (standard error 0.089). - MP shock×Nonbank (Column (6)) shows -0.473 ∗∗∗ (standard error 0.140), indicating nonbanks charge lower spreads relative to banks following MP tightening. - Interpretation: - Results are consistent with a credit lending channel of monetary policy via tighter credit conditions. - Nonbanks partially offset the contraction in bank lending during MP tightenings, consistent with banks’ funding frictions and deposit outflows. ### Effects of macroprudential policy (MaPP) shocks - A one-standard deviation MaPP tightening shock leads banks to cut credit supply by 2.0 percent, while nonbanks increase lending by 2.5 percent relative to banks, though the absolute nonbank response is muted (Table 2 summary). - MaPP shock coefficients and interactions (selected estimates from Table 2 and Table 3): - MaPP shock (Table 2, Col (1)): -0.040 ∗∗∗ (standard error 0.003). - MaPP shock×Nonbank (Table 2, Col (1)): 0.036 ∗∗∗ (standard error 0.004). - MaPP shock (Table 3, Col (4) preferred specification): -0.019 ∗∗∗ (standard error 0.002). - MaPP shock×Nonbank (Table 3, Col (4)): 0.016 ∗∗∗ (standard error 0.002). - Price effects (Table 2, Column (6)): - MaPP shock increases spread by 0.144 ∗∗ (standard error 0.065). - MaPP shock×Nonbank (Column (6)) is -0.163 (standard error 0.127), estimated with less precision. - Interpretation: - Findings align with theory (e.g., tighter capital requirements reduce bank deposits, lowering nonbanks’ debt financing costs) and prior empirical work showing credit reallocation to nonbanks when banks’ lending capacity is constrained. ### Combined MP and MaPP shocks (unified framework) - Results robust when both shocks are included (Table 3): - Nonbanks mitigate contraction in bank credit supply during contractionary shocks. - Nonbanks expand lending disproportionately more during MP shocks than MaPP shocks. - Preferred specification (Table 3, Col (4), firm×time FE): nonbanks expand lending by 4.6 percent relative to banks during MP shocks, compared to 1.6 percent during MaPP shocks. - In absolute terms, nonbank lending expands by 2.5 percent during MP shocks but is muted during MaPP shocks. - Selected coefficients from Table 3: - MP shock: -0.021 ∗∗∗ (standard error 0.003, Col (4)). - MaPP shock: -0.019 ∗∗∗ (standard error 0.002, Col (4)). - MP shock×Nonbank: 0.046 ∗∗∗ (standard error 0.003, Col (4)). - MaPP shock×Nonbank: 0.016 ∗∗∗ (standard error 0.002, Col (4)). - Spread (Table 3, Col (5)): MP shock 0.166 ∗ (standard error 0.089); MaPP shock 0.124 ∗ (standard error 0.065); MP shock×Nonbank -0.474 ∗∗∗ (standard error 0.157); MaPP shock×Nonbank -0.058 (standard error 0.140). - Robustness: - Coefficients for positive and negative MP and MaPP shocks are highly similar (Tables OA.2 and OA.3, Online Appendix). - Results robust to restricting to selected AEs (Table OA.4) and to controlling for endogeneity between MP and MaPP (pairwise correlation 0.0844; Column (7) of Table OA.9). ### Normalization via bank stock prices to compare shock magnitudes - Rationale: - One-standard deviation shocks are not directly comparable across MP and MaPP because economic meaning differs. - Bank stock prices capture interest rate changes, shifts in future profitability, risk-taking incentives, and regulatory constraints, providing a unified metric. - Method: - Regress log change of real bank stock prices on MP and MaPP shocks separately; normalize each shock by its coefficient so normalized shocks correspond to a one percent decline in stock prices. - Normalized results (preferred specification, Table (3) Col (4) rerun with normalized shocks; Table OA.5): - MP shock calibrated to decrease bank stock prices by one percent leads to a 5.5 percent relative increase in nonbank lending. - MaPP shock calibrated to decrease bank stock prices by one percent leads to a 4.0 percent relative increase in nonbank lending. - The difference between coefficients is statistically significant at the five percent level when normalized with bank stock prices. ### Interaction and coordination between MP and MaPP shocks - Aggregate interaction (Table 4): - When not distinguishing banks from nonbanks, MaPP and MP shocks complement each other in reducing credit supply (Table 4, Column (1)). - A one-standard deviation increase in both shocks leads to an additional decline in lending to nonfinancial corporations of 0.5 percent (economically small). - Effects by lender type (Table 4, Column (2) and subsequent rows): - Tightening of both policies mitigates the credit supply contraction from banks by 0.3 percent (coefficient flips positive when isolating banks). - The expansion of nonbank credit is mitigated by 1.4 percent when both policies tighten (triple interaction term). - Combined effect for nonbanks remains expansionary relative to banks: combined effect in the order of 4.2 percent (4.2 = 4.6 + 1.0 - 1.4 as reported in text). - Tightening-loosening combinations (selected coefficients from Table 4, tightest specification, Column (5)): - MP+ × MaPP+ (both tighten) on banks: 0.002 (standard error 0.003) — not precisely estimated in all columns. - MP+ × MaPP− on banks: -0.006, -0.004, -0.005 (standard errors in table). - MP− × MaPP+ on banks: -0.014 ∗∗∗, -0.012 ∗∗∗, -0.007 ∗∗ (standard errors reported). - Interpretation: lack of policy coordination (MP loosening and MaPP tightening, or vice-versa) typically leads to additional contraction in bank lending. - Nonbanks in policy-combination cases (selected rows, Table 4): - MP+ × MaPP+ × Nonbank: -0.033 ∗∗∗, -0.029 ∗∗∗, -0.023 ∗∗∗ (standard errors reported). - MP− × MaPP+ × Nonbank: 0.019 ∗∗, 0.020 ∗∗∗, 0.023 ∗∗∗ (standard errors reported). - Interpretation: coordinated tightening can deter nonbanks (possible market interpretation of stronger intent to rein in credit growth), while divergence in policy direction can amplify nonbank expansion relative to banks. - Magnitude and statistical significance are generally small for interaction terms relative to individual policy effects. ### Key takeaways and contribution - Three main contributions highlighted: 1. The differential increase in nonbank credit supply during contractionary MP shocks found in U.S. and Denmark holds across a large sample of 22 countries, suggesting this behavior is a feature of nonbanks’ business models. 2. Nonbank lending expansion also occurs when MaPP tightens, indicating that shocks directly affecting banks’ lending capacity shift credit to the nonbank sector. 3. Divergence in the direction of MP and MaPP amplifies credit reallocation to nonbanks; coordinated policy actions (both tighten) can mitigate nonbank expansion but may still leave nonbanks expanding relative to banks. - Policy implication noted: - Traditional MaPPs that focus only on banks may not be able to stem credit leakage to nonbanks during contractionary MP shocks. - Extending the regulatory perimeter to nonbanks could help limit credit leakages to nonbanks, improve transmission of monetary policy, and contribute to overall financial sector resilience (as suggested in the text, citing Abbas et al. 2025). *Source: 4.1 Main results (wpiea2025096-print-pdf).* ### 4.2 Relationship lending and nonbanks ### 4.2 Relationship lending and nonbanks ### Key concepts and motivation - Relationship lending: lenders collect substantial information on borrowers’ underlying credit risk over time, allowing adjustment of credit terms and better allocation of loans during bad times. - Benefits during downturns: higher credit supply and more favorable terms alleviate firms’ credit constraints, allowing nonfinancial firms to increase investment and employment (Bolton et al. 2016, Sette and Gobbi 2015, Beck et al. 2018, Banerjee et al. 2021). - Potential downside: weakly-capitalized banks may have an incentive to keep unproductive and unviable zombie firms alive when the cost of debt goes up to avoid recognition of losses (Albuquerque and Mao 2025). - Distinction between banks and nonbanks: benefits of relationship lending during downturns appear largely confined to bank relationships (Aldasoro et al. 2024), likely due to greater nonbank cyclicality and instability of nonbanks’ funding model (Irani et al. 2021, Fleckenstein et al. forthcoming). However, evidence from the US and Denmark suggests nonbanks can also shield borrowers from contractionary MP shocks (Elliott et al. 2022, 2024; Cucic and Gorea 2024). ### Empirical strategy for this section - Relationship measures: - Pre-relation: dummy = 1 if a borrower has borrowed from a given lender in the past five years. - Duration: logarithm of the number of years since the borrower got the first loan from a specific lender. - Estimation: Equation (3) with interactions between shocks (MP shock, MaPP shock), lender type (Nonbank), and relationship measures (Pre-relation, Duration). - Dependent variable: log of new syndicated loans. - Standard errors clustered by firm. ### Main empirical findings (table results and magnitudes) - Direct shock effects: - MP shock coefficients: -0.034, -0.037, -0.035, -0.031, -0.033, -0.035 (all marked ∗∗∗). Standard errors reported in table: (0.004)(0.003)(0.003)(0.004)(0.003)(0.003). - MaPP shock coefficients: -0.023, -0.021, -0.022, -0.021, -0.020, -0.021 (all marked ∗∗∗). Standard errors reported: (0.003)(0.002)(0.002)(0.003)(0.002)(0.002). - Relationship main effects: - Relation coefficients (columns (1)-(3) using pre-relation): 0.024, 0.062, 0.077 (all ∗∗∗). SEs: (0.004)(0.003)(0.003). - Relation coefficients (columns (4)-(6) using Duration): 0.020, 0.031, 0.031 (all ∗∗∗). SEs: (0.002)(0.001)(0.001). - Interaction effects: - Relation × Nonbank: -0.013 ∗∗ (0.005) in column (1); later columns show small, statistically insignificant values: -0.001, -0.002, -0.004, -0.000, -0.002 with SEs shown. - MP shock × Nonbank: 0.037, 0.044, 0.034, 0.035, 0.041, 0.032 (all ∗∗∗). SEs: (0.005)(0.004)(0.004)(0.005)(0.004)(0.004). - MaPP shock × Nonbank: 0.006, 0.013 ∗∗∗, 0.015 ∗∗∗, 0.005, 0.013 ∗∗∗, 0.014 ∗∗∗ with SEs: (0.004)(0.003)(0.003)(0.004)(0.003)(0.003). - MP shock × Relation: 0.030, 0.032, 0.033 (columns (1)-(3), ∗∗∗) and 0.009, 0.009, 0.012 (columns (4)-(6), ∗∗∗). SEs: (0.004)(0.004)(0.003) and (0.001)(0.001)(0.001). - MaPP shock × Relation: 0.006 ∗, 0.003, 0.006 ∗∗, 0.001, 0.000, 0.002 ∗ with SEs: (0.004)(0.003)(0.002)(0.001)(0.001)(0.001). - MP shock × Nonbank × Relation: 0.045, 0.034, 0.032 (columns (1)-(3)) and 0.018, 0.016, 0.014 (columns (4)-(6)) — all reported as ∗∗∗. SEs: (0.006)(0.005)(0.005)(0.002)(0.002)(0.002). - MaPP shock × Nonbank × Relation: 0.012 ∗∗, 0.003, -0.003, 0.004 ∗∗, 0.001, -0.001 with SEs: (0.006)(0.005)(0.004)(0.002)(0.002)(0.001). - Economic magnitudes illustrated in prose: - Firms with a prior lending relationship with banks offset nearly 94 percent of the decline in credit supply following MP tightening: 0.033/0.035. - Firms with a prior lending relationship with banks offset about one-fourth of the decline during MaPP tightening: 0.006/0.022. - Sample and fit statistics: - Observations: 749,629; 748,453; 739,866; 749,629; 748,453; 739,866 (columns (1)-(6)). - R2: 0.727, 0.795, 0.877, 0.727, 0.795, 0.877. - Fixed effects included (varies by column): - Firm FE, Lender FE, Country×Sector×Time FE, ILST FE, Firm×Time FE (specific columns indicated in table). ### Interpretation and substantive conclusions - Relationship lending with banks: - Allows nonfinancial firms to mitigate contraction in credit supply driven by both MP tightening and MaPP tightening. - The quantitative example shows nearly full offset for MP tightening (94 percent) and partial offset for MaPP tightening (about one-fourth). - Results using Duration (years of lending relationship) show similar patterns with smaller coefficients. - Findings are consistent with Hachem (2011) theory and Berger et al. (2024) empirical results that relationship lending smooths the credit channel from contractionary MP shocks. - Relationship lending with nonbanks: - Relationship lending is even more important for firms that borrow from nonbanks during MP shocks. - Triple interaction MP shock × Nonbank × Pre-relation ranges between 0.032 and 0.045 in columns (1)-(3), implying firms with an existing lending relationship with nonbanks borrow 3.2-4.5 percent more compared to first-time nonbank borrowers when MP tightens. - This suggests nonbanks play an important role in allocating funding to borrowers when interest rates increase, particularly to firms with established lending relationships. - Evidence on additional protection offered by nonbanks relative to banks during MaPP shocks is more mixed and depends on controls for time-varying firm characteristics. - Pricing and margins: - Increased lending from nonbanks when MP or MaPP tighten may come at a greater price for borrowers with an existing lending relationship, relative to those without one (reference to Table OA.7 in the Online Appendix). - However, during MP shocks, borrowers from nonbanks still benefit from a decline in spreads compared to those borrowing from banks, indicating intensive margin effects where nonbanks attract new borrowers by offering preferential credit terms. ### Open questions and contribution - Gap in literature: authors are not aware of prior research focusing on the role of relationship lending (with banks or nonbanks) during contractionary MaPP or regulatory shocks; this analysis aims to fill that gap. - External validity question: whether findings from the US and Denmark on nonbank protection during MP shocks generalize to the authors’ sample of 22 country lenders is an open question addressed in the analysis. *Source: wpiea2025096-print-pdf - 4.2 Relationship lending and nonbanks — https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025096-print-pdf.pdf* ### 4.3 Bank characteristics and credit migration to nonbanks ### 4.3 Bank characteristics and credit migration to nonbanks ### Main finding - Nonbank credit consistently increases relative to bank credit following contractionary MP and MaPP shocks: nonbanks fill a funding gap created by the reduction in bank credit. - Credit migration from banks to nonbanks is stronger in syndicates that include weaker banks. ### Bank characteristics and nonbank share (evidence from Table 6) - The nonbank share is typically higher in syndicates with weaker banks (third row of Table 6). - The share of nonbank loans increases in response to contractionary policy shocks (first two rows of Table 6). - The increase in nonbank participation after MP and MaPP shocks is more pronounced in loan deals involving weaker banks. - In the preferred specification (column (2), low bank capital), the nonbank loan share rises by an additional 1.3 and 1.7 percentage points following a one-standard-deviation increase in MP and MaPP shocks, respectively, compared to syndicates with stronger banks. - These effects are economically large given the mean difference in the nonbank share between syndicates with strong and weak banks is around one percentage point. - Selected coefficients from Table 6 (coefficients and clustered standard errors in parentheses): - MP shock: 0.013 (0.006) in column (1); 0.017 (0.007) in column (2). - MaPP shock: 0.021 (0.005) in column (1); 0.031 (0.005) in column (2). - Bank charact.: 0.005 (0.004) in column (1); 0.011 (0.004) in column (2). - MP shock×Bank charact.: 0.011 (0.006) in column (1); 0.013 (0.007) in column (2). - MaPP shock×Bank charact.: 0.016 (0.004) in column (1); 0.017 (0.005) in column (2). - Observations and fit: - Observations: 47,615 (column 1), 42,387 (column 2), 47,615 (column 3), 42,387 (column 4). - R2: 0.657, 0.678, 0.658, 0.678 (columns 1–4). ### Credit-risk migration to nonbanks (evidence from Table 7) - Nonbanks provide more lending to risky borrowers on average. - Contractionary MP and MaPP shocks do not lead nonbanks to disproportionately increase lending to risky borrowers relative to non-risky borrowers. - Risk proxies used (four alternatives): - 5y avg. spread (top quartile). - Leveraged loans (spread over 150 bps over Libor). - High PD (top quartile of country-time PD distribution). - Junk rating (S&P long-term issuer rating below BBB−). - Selected coefficients from Table 7 (columns correspond to alternative risk proxies; standard errors in parentheses): - MP shock: 0.007 (0.010), 0.009 (0.011), 0.011 (0.011), 0.010 (0.012), -0.002 (0.018), 0.003 (0.021), -0.024 (0.025), -0.013 (0.026). - MaPP shock: 0.053 (0.007) ***, 0.062 (0.007) ***, 0.055 (0.008) ***, 0.065 (0.008) ***, 0.031 (0.012) ***, 0.043 (0.014) ***, 0.074 (0.016) ***, 0.084 (0.017) ***. - Risky (unconditional higher nonbank share for risky borrowers): 0.011 (0.006) *, 0.007 (0.007), 0.016 (0.006) ***, 0.015 (0.006) **, 0.013 (0.007) **, 0.021 (0.008) **, 0.061 (0.015) ***, 0.050 (0.018) ***. - MP shock×Risky and MaPP shock×Risky coefficients show no systematic disproportionate increase to risky borrowers (see table for column-specific estimates). - Observations and fit (selected): Observations range from 8,936 to 36,356 across specifications; R2 ranges from 0.635 to 0.817. ### Real effects on firms (evidence from Table 8) - Firms that borrow from nonbanks (NB relation = 1 if borrowed from at least one nonbank in past five years) tend to have higher debt, investment, employment, and higher PDs unconditionally. - After contractionary MP and MaPP shocks, firms with prior nonbank relationships show: - Higher capital expenditures and employment relative to firms without nonbank relationships. - Higher intangible investment in response to MaPP shocks (statistically significant). - No evidence that nonbank relationships reduce borrower riskiness after shocks; PDs appear to increase, especially during MaPP shocks. - Selected coefficients from Table 8 (standard errors in parentheses): - MP shock: 0.055 (0.045) for Total Debt; -0.002 (0.050) Liquid assets; 0.021 (0.027) Capital exp.; 0.056 ** (0.028) Intangibles; -0.010 (0.026) Employment; 0.014 (0.061) PD. - MaPP shock: -0.070 ** (0.032) Total Debt; -0.079 ** (0.035) Liquid assets; -0.079 *** (0.019) Capital exp.; -0.067 *** (0.021) Intangibles; -0.074 *** (0.018) Employment; -0.029 (0.035) PD. - NB relation (unconditional): 0.166 *** (0.024) Total Debt; 0.004 (0.027) Liquid assets; 0.105 *** (0.013) Capital exp.; 0.071 *** (0.015) Intangibles; 0.060 *** (0.010) Employment; 0.111 *** (0.026) PD. - MP shock×NB relation: 0.022 (0.037) Total Debt; 0.024 (0.042) Liquid assets; 0.039 * (0.022) Capital exp.; -0.004 (0.023) Intangibles; 0.053 *** (0.019) Employment; 0.035 (0.041) PD. - MaPP shock×NB relation: 0.015 (0.026) Total Debt; 0.033 (0.030) Liquid assets; 0.033 ** (0.013) Capital exp.; 0.040 ** (0.018) Intangibles; 0.034 *** (0.010) Employment; 0.049 ** (0.022) PD. - Observations by dependent variable: 239,871; 705,724; 538,144; 55,230; 39,400 (note: table lists Observations 239871705724538144552303919400 across columns); R2 range: 0.611 to 0.982 (see table for column-specific values). ### Bank lending to nonbanks and potential funding channel (evidence from Table 9) - Hypothesis: banks may reallocate lending toward nonbanks when constrained by MaPP shocks (regulatory tightening). - Results: - Banks increase lending to nonbank borrowers relative to nonfinancial corporations after MaPP shocks. - Estimated differential lending: 0.8–1.5 percent more loans to nonbanks following a one-standard-deviation MaPP shock, depending on demand controls. - Column (3) coefficient marginally insignificant at p = 11 percent for one specification. - No consistent evidence that banks reallocate more lending to nonbanks following contractionary MP shocks — banks tend to cut credit broadly during MP shocks. - Low-capitalized banks are primary drivers of the MaPP result: weaker banks increase lending to nonbanks relative to other borrowers during MaPP shocks. - Selected coefficients from Table 9 (standard errors in parentheses): - MP shock: 0.002 (0.004) column (1); 0.001 (0.003) column (2); 0.001 (0.003) column (3); 0.016 *** (0.005) column (4). - MaPP shock: -0.020 *** (0.002) column (1); -0.018 *** (0.002) column (2); -0.015 *** (0.002) column (3); -0.014 *** (0.003) column (4). - MP shock×NB Borrower: -0.013 (0.009); -0.011 (0.008); -0.010 (0.006); -0.006 (0.016). - MaPP shock×NB Borrower: 0.015 ** (0.007); 0.011 ** (0.006); 0.008 (0.005); -0.012 (0.011). - MP shock×LC bank: -0.015 *** (0.005) (column 4). - MaPP shock×LC bank: -0.007 * (0.004) (column 4). - MP shock×NB Borrower×LC bank: 0.012 (0.014) (column 4). - MaPP shock×NB Borrower×LC bank: 0.029 ** (0.013) (column 4). - Observations and fit: - Observations: 547,476; 545,765; 533,115; 218,587 (columns 1–4). - R2: 0.757, 0.815, 0.889, 0.877. ### Robustness checks and heterogeneity - Nonbanks with unstable funding drive the main results during contractionary MP and MaPP shocks (pension funds and insurance companies classified as stable funding; investment banks, broker-dealers, finance companies, private equity, venture capital firms, hedge funds classified as unstable funding). - Results robust across: - Loan types: term loans versus credit lines. - Regression-based allocation of missing loan shares. - Extensive margin of lending. - Excluding top three nonbanks (JP Morgan Securities LLC, BofA Securities, Citigroup Global Markets Inc. together account for ~10 percent of syndicated lending by nonbanks over 2000-19). - Alternative macroprudential measures: - Baseline MaPP focused on loan-targeted measures and stress testing. - Adding reserve requirements, capital requirements, conservation buffers, countercyclical capital buffers yields broadly robust results; adding capital-based requirements produces smaller short-run effects on credit growth. - Additional sensitivity checks: - Lags of MaPP and MP shocks; inclusion of additional lags and interactions; different fixed-effect specifications; inclusion of MP shocks directly in MaPP estimation — main results remain robust. - U.S.-specific robustness using Drechsel and Miura (2025) high-frequency bank regulation shocks (normalized to induce a one percent decline in bank stock prices) yields consistent expansion in nonbank lending relative to bank lending. - Additional splits: - Pre-GFC (until 2007Q4) versus post-GFC (starting 2010Q1): baseline results persist. - USD-denominated vs non-USD loans: increases in nonbank credit following shocks are greater in USD lending. - Cross-border vs domestic lending: baseline results hold. - Borrowers in AEs vs EMDEs: baseline results hold. - Alternative clustering approaches for standard errors (firm and country-time; lender and country-time; firm, lender, and time) do not alter main results. ### Interpretation and mechanisms - For MP shocks: - Deposit outflows and a widening spread between policy rate and deposit rate exacerbate funding frictions for banks, constraining lending and facilitating a shift of funding toward nonbanks (deposits→nonbanks). - Nonbanks (e.g., money market funds) may pass through higher interest rates and attract depositor funds, increasing their funding and enabling expansion of lending when banks contract. - Evidence from Denmark: nonbanks increase long-term funding during contractionary MP shocks while banks experience long-term funding outflows. - Some nonbanks (e.g., BDCs) increase borrowing from banks via credit line drawdowns and renegotiations, allowing expanded private lending despite tighter conditions. - For MaPP shocks: - Tighter bank regulation post-GFC (e.g., Basel III, EBA exercises) may have constrained banks—especially those near regulatory limits—prompting a reallocation of credit to nonbanks and thereby expanding nonbank lending to corporates. - Preferential regulatory treatment (lower risk weights under IRB, lower PD/LGD assignments) for exposures to nonbanks can incentivize banks to shift lending toward nonbank borrowers. - Low-capitalized banks increase lending to nonbanks during MaPP shocks, consistent with incentives to protect capital buffers. ### Financial stability implications and policy considerations - Benefits of nonbanks: - Nonbanks help cushion firms from credit contractions during MP and MaPP tightening, supporting capital expenditures and employment. - Risks of nonbanks: - Nonbanks unconditionally lend more to riskier borrowers and rely more on unstable funding sources. - Nonbanks are less regulated, typically have higher leverage, and often lack access to central bank liquidity facilities. - Nonbank lending is more procyclical; firms borrowing more from nonbanks suffer steeper declines in credit supply during banking crises. - Increased bank-to-nonbank lending during MaPP shocks may amplify bank-nonbank interconnectedness and systemic vulnerabilities. - Policy recommendations and implications highlighted in the text: - Close data gaps on the nonbank sector and extend macroprudential regulations to encompass nonbanks. - Consider expanding the regulatory perimeter: leverage limits, capital and liquidity requirements, stress-testing for nonbanks. - Recognize that macroprudential measures targeting banks may inadvertently accelerate nonbank expansion and credit leakage to less-regulated sectors. - Monitor private credit markets (direct lending) where nonbanks play a larger, less transparent role; this is an important avenue for future research. *Italic source attribution: Content from "4.3 Bank characteristics and credit migration to nonbanks" (wpiea2025096-print-pdf).* ### Appendix A: Data ### Appendix A: Data ### List of lender countries (Table A.1) - Austria AE 0.35% - Belgium AE 0.74% - Brazil EMDE 0.49% - Canada AE 6.70% - Cyprus AE 0.004% - Finland AE 0.12% - France AE 7.71% - Germany AE 6.98% - Greece AE 0.12% - India EMDE 1.54% - Ireland AE 0.44% - Italy AE 1.95% - Japan AE 11.13% - Lithuania AE 0.03% - Netherlands AE 3.71% - Norway AE 0.70% - Portugal AE 0.18% - Slovenia AE 0.01% - Spain AE 2.74% - Sweden AE 1.02% - United Kingdom AE 11.35% - United States AE 43.68% - Notes: AE refers to Advanced Economies, and EMDE refers to Emerging Market and Developing Economies. Loan share is the average loan share for each lender country in our estimation sample. ### Loan characteristics: nonbanks versus banks (Table A.2) - Sample restricted to nonfinancial borrowers based on SIC code classification. - Nonbanks (summary statistics) - Term length (years): Mean 5.13, STD 3.48, P25 3.00, P50 5.00, P75 6.00 - Tranche value ($ million): Mean 311.46, STD 524.41, P25 40.73, P50 121.84, P75 338.30 - Syndicate members: Mean 7.32, STD 6.74, P25 3.00, P50 5.00, P75 9.00 - Participation share: Mean 0.18, STD 0.19, P25 0.06, P50 0.10, P75 0.25 - All-in-pricing (BPs): Mean 266.86, STD 162.66, P25 150.00, P50 250.00, P75 350.00 - Margin pricing (BPs): Mean 266.57, STD 161.38, P25 150.00, P50 250.00, P75 350.00 - Banks (summary statistics) - Term length (years): Mean 5.00, STD 3.92, P25 3.00, P50 5.00, P75 6.00 - Tranche value ($ million): Mean 252.40, STD 472.00, P25 26.10, P50 86.30, P75 257.51 - Syndicate members: Mean 6.44, STD 6.19, P25 3.00, P50 5.00, P75 8.00 - Participation share: Mean 0.17, STD 0.19, P25 0.05, P50 0.10, P75 0.25 - All-in-pricing (BPs): Mean 208.32, STD 139.42, P25 102.50, P50 187.50, P75 275.00 - Margin pricing (BPs): Mean 204.62, STD 135.67, P25 100.00, P50 175.00, P75 275.00 ### Monetary policy (MP) and Macroprudential policy (MaPP) shocks (Figures A.1–A.3) - Figure A.1: Monetary policy shocks over time - Displays median (red line) and interquartile range (blue area) of MP shocks from 2000q1 to 2020q1. - Figure A.2: MaPP shocks over time - Displays median (red line) and interquartile range (blue area) of MaPP shocks from 2000q1 to 2020q1. - Figure A.3: Correlation of MP and MaPP shocks - Notes: Red line is the linear regression line between the two series (correlation of 0.0844). *Source: Appendix A: Data (Tables A.1–A.2 and Figures A.1–A.3).* ### Appendix B: Tables (Selected regression results) ### Funding models of nonbanks (Table B.1) - Dependent variable: log of new syndicated loans. - Key coefficients (standard errors in parentheses; asterisks denote significance: *, **, *** at 10%, 5%, 1%): - MP shock: -0.021*** (0.003) in column (1); -0.023*** (0.003) in column (2); -0.021*** (0.002) in column (3) - MaPP shock: -0.020*** (0.002); -0.020*** (0.002); -0.019*** (0.002) - MP shock × Stable Nonbank: -0.119*** (0.027); -0.109*** (0.025); -0.097*** (0.020) - MP shock × Unstable Nonbank: 0.060*** (0.004); 0.061*** (0.004); 0.049*** (0.003) - MaPP shock × Stable Nonbank: -0.047*** (0.018); -0.030* (0.015); -0.017 (0.013) - MaPP shock × Unstable Nonbank: 0.017*** (0.003); 0.019*** (0.003); 0.018*** (0.002) - Fixed effects and fit: - Firm FE (columns vary), Lender FE ✓✓✓, Country×Sector×Time FE ✓ (where indicated), ILST FE ✓ (where indicated), Firm×Time FE ✓ (where indicated) - Observations: 740,206; 739,026; 730,335 - R^2: 0.726; 0.794; 0.877 - Notes: Stable nonbanks refer to nonbanks with stable funding, namely pension funds and insurance companies. Unstable nonbanks are all the other nonbanks. Standard errors clustered by firm. ### Loan types, regression-based approach, and extensive margin (Table B.2) - Dependent variable: log of new syndicated loans. - Columns: (1) Baseline; (2) Term loans; (3) Credit lines; (4) Reg. approach; (5) Ext. margin; (6) Exc. top 3 nonbanks. - Key coefficients: - MP shock: -0.021*** (0.002) baseline; -0.019*** (0.003) term loans; -0.022*** (0.003) credit lines; -0.022*** (0.003) reg. approach; -0.007*** (0.001) ext. margin; -0.016*** (0.002) excluding top 3 nonbanks - MaPP shock: -0.019*** (0.002); -0.019*** (0.002); -0.020*** (0.002); -0.016*** (0.002); -0.007*** (0.001); -0.018*** (0.002) - MP shock × Nonbank: 0.046*** (0.003); 0.033*** (0.004); 0.056*** (0.003); 0.067*** (0.004); 0.016*** (0.002); 0.029*** (0.003) - MaPP shock × Nonbank: 0.016*** (0.002); 0.012*** (0.003); 0.021*** (0.002); 0.030*** (0.003); 0.007*** (0.001); 0.0005 (0.002) - Fixed effects and fit: - Lender FE ✓ for all columns; Firm×Time FE ✓ for all columns - Observations: 739,866; 348,427; 522,995; 739,866; 2,014,048; 689,635 - R^2: 0.877; 0.899; 0.896; 0.816; 0.933; 0.879 ### Alternative MaPP shocks (Table B.3) - Dependent variable: log of new syndicated loans. Columns compare MaPP constructions: - Column (1) baseline MaPP shocks; (2) measures targeting loan supply; (3) adds reserve requirements to baseline; (4) adds capital requirements, conservation buffers, and countercyclical capital buffers to baseline. - Key coefficients: - MP shock: -0.021*** (0.002); -0.023*** (0.002); -0.019*** (0.002); -0.022*** (0.002) - MaPP shock: -0.019*** (0.002); -0.027*** (0.002); -0.020*** (0.002); -0.015*** (0.002) - MP shock × Nonbank: 0.046*** (0.003); 0.047*** (0.003); 0.046*** (0.003); 0.047*** (0.003) - MaPP shock × Nonbank: 0.016*** (0.002); 0.026*** (0.003); 0.016*** (0.002); 0.012*** (0.002) - Fixed effects and fit: - Lender FE ✓✓✓✓; Firm×Time FE ✓✓✓✓ - Observations: 739,866 for all columns - R^2: 0.877 for all columns ### Robustness checks: split-sample specifications (Table B.4) - Dependent variable: log of new syndicated loans. - Columns: (1) Base; (2) Pre-GFC; (3) Post-GFC; (4) USD; (5) Non-USD; (6) Cross-border; (7) Domestic; (8) AE; (9) EMDE. - Key coefficients (selected): - MP shock (column values): -0.021*** (0.002) base; -0.004 (0.005) pre-GFC; -0.006* (0.003) post-GFC; -0.024*** (0.003) USD; -0.014*** (0.004) Non-USD; -0.017*** (0.003) Cross-border; --0.022*** (0.002) Domestic; -0.015** (0.007) AE - MaPP shock: -0.019*** (0.002) base; 0.002 (0.005) pre-GFC; -0.009*** (0.002) post-GFC; -0.018*** (0.002) USD; -0.015*** (0.003) Non-USD; -0.012*** (0.002) Cross-border; --0.020*** (0.002) Domestic; -0.007 (0.006) AE - MP shock × Nonbank: 0.046*** (0.003) base; 0.015* (0.009) pre-GFC; 0.007* (0.004) post-GFC; 0.058*** (0.004) USD; 0.030*** (0.005) Non-USD; 0.028*** (0.004) Cross-border; 0.073*** (0.006) Domestic; 0.048*** (0.003) AE; 0.032*** (0.010) EMDE - MaPP shock × Nonbank: 0.016*** (0.002) base; 0.016* (0.010) pre-GFC; 0.008*** (0.003) post-GFC; 0.023*** (0.003) USD; -0.002 (0.004) Non-USD; 0.013*** (0.003) Cross-border; 0.019*** (0.004) Domestic; 0.016*** (0.002) AE; 0.027*** (0.009) EMDE - Fixed effects and fit: - Lender FE ✓ for all columns; Firm FE×Time ✓ for all columns - Observations: 739,866; 276,509; 403,604; 448,201; 305,467; 236,391; 475,751; 696,949; 42,306 - R^2: 0.877; 0.854; 0.893; 0.831; 0.909; 0.837; 0.892; 0.880; 0.820 - Notes: Column (1) baseline specification; column (2) pre-GFC sample (up to 2007Q4); column (3) post-GFC (after 2010Q1); column (4) USD loans only; column (5) non-USD loans only; column (6) cross-border lending; column (7) domestic lending; column (8) borrowers in advanced economies (AEs); column (9) borrowers in emerging markets and developing economies (EMDEs). Standard errors clustered by firm. Asterisks, *, **, and ***, denote statistical significance at the 10%, 5%, and 1% levels. *Source: Appendix B: Tables (Tables B.1–B.4).* ### References ### wpiea2025096-print-pdf - References ### Bibliographic references (selected entries) - Abbas, N., Albuquerque, B., Garrido, J. M., Gautam, D., Mosk, B., Piontek, T., Rosha, A., Tressel, T. and Yokoyama, A. (2025), Corporate Sector Vulnerabilities and High Levels of Interest Rates, Departmental Paper DP/2025/001, International Monetary Fund. - Acharya, V. V., Cetorelli, N. and Tuckman, B. (2024), Where Do Banks End and NBFIs Begin?, NBER Working Papers 32316, National Bureau of Economic Research, Inc. - Agarwal, I., Hu, M., Roman, R. and Zheng, K. (2023), Lending by Servicing: Monetary Policy Transmission Through Shadow Banks, Working Papers 23-14, Federal Reserve Bank of Philadelphia. - Ahnert, T., Forbes, K., Friedrich, C. and Reinhardt, D. (2021), ‘Macroprudential FX regulations: Shifting the snowbanks of FX vulnerability?’, Journal of Financial Economics 140(1), 145–174. - Alam, Z., Alter, A., Eiseman, J., Gelos, G., Kang, H., Narita, M., Nier, E. and Wang, N. (2025), ‘Digging Deeper–Evidence on the Effects of Macroprudential Policies from a New Database’, Journal of Money, Credit and Money. - Albuquerque, B. (2024), ‘Corporate debt booms, financial constraints and the investment nexus’, Journal of Applied Econometrics 39(5), 766–789. - Albuquerque, B., Becker, J. and Firat, M. (2025), Bonding through crisis: the role of nonbanks. mimeo. - Albuquerque, B. and Iyer, R. (2024), ‘The Rise of the Walking Dead: Zombie Firms Around the World’, Journal of International Economics 152, 104019. - Albuquerque, B. and Mao, C. (2025), The Zombie Lending Channel of Monetary Policy, Available at https://ssrn.com/abstract=4572534, SSRN. - Aldasoro, I., Doerr, S. and Zhou, H. (2024), Non-bank lending during crises, CEPR Discussion Papers 18989, C.E.P.R. Discussion Papers. - Altavilla, C., Laeven, L. and Peydró, J.-L. (2020), Monetary and macroprudential policy complementarities: evidence from European credit registers, Working Paper Series 2504, European Central Bank. - Anguren, R., Jiménez, G. and Peydró, J.-L. (2024), ‘Bank capital requirements and risk-taking: Evidence from Basel III’, Journal of Financial Stability 74, 101292. - Araujo, J., Patnam, M., Popescu, A., Valencia, F. and Yao, W. (2024), ‘Effects of macroprudential policy: Evidence from over 6000 estimates’, Journal of Banking & Finance 169, 107273. - Banerjee, R. N., Gambacorta, L. and Sette, E. (2021), ‘The real effects of relationship lending’, Journal of Financial Intermediation 48, 100923. - Beck, T., Degryse, H., De Haas, R. and van Horen, N. (2018), ‘When arm’s length is too far: Relationship banking over the credit cycle’, Journal of Financial Economics 127(1), 174–196. - Bednarek, P., Briukhova, O., Ongena, S. and von Westernhagen, N. (2023), Effects of bank capital requirements on lending by banks and non-bank financial institutions, Discussion Papers 26/2023, Deutsche Bundesbank. - Begenau, J. and Landvoigt, T. (2022), ‘Financial Regulation in a Quantitative Model of the Modern Banking System’, The Review of Economic Studies 89(4), 1748–1784. - Berger, A. N., Bouwman, C. H., Norden, L., Roman, R. A., Udell, G. F. and Wang, T. (2024), Relationship Lending and Monetary Policy Shocks: Evidence from the U.S., Available at https://ssrn.com/abstract=4945534, SSRN. - Bluwstein, K. and Patozi, A. (2024), The effects of macroprudential policy announcements on systemic risk, Bank of England working papers 1080, Bank of England. - Bolton, P., Freixas, X., Gambacorta, L. and Mistrulli, P. E. (2016), ‘Relationship and Transaction Lending in a Crisis’, The Review of Financial Studies 29(10), 2643–2676. - Bruno, V. and Shin, H. S. (2015), ‘Capital flows and the risk-taking channel of monetary policy’, Journal of Monetary Economics 71, 119–132. - Buchak, G., Matvos, G., Piskorski, T. and Seru, A. (2018), ‘Fintech, regulatory arbitrage, and the rise of shadow banks’, Journal of Financial Economics 130(3), 453–483. - Budnik, K. and Kleibl, J. (2018), Macroprudential regulation in the European Union in 1995-2014: introducing a new data set on policy actions of a macroprudential nature, Working Paper Series 2123, European Central Bank. - Cerutti, E., Claessens, S. and Laeven, L. (2017), ‘The use and effectiveness of macroprudential policies: New evidence’, Journal of Financial Stability 28, 203–224. - Cerutti, E., Hale, G. and Minoiu, C. (2015), ‘Financial crises and the composition of cross-border lending’, Journal of International Money and Finance 52, 60–81. - Cerutti, E. and Zhou, H. (2018), Cross-border Banking and the Circumvention of Macroprudential and Capital Control Measures, IMF Working Papers 18/217, International Monetary Fund. - Chari, A., Dilts-Stedman, K. and Forbes, K. (2022), ‘Spillovers at the extremes: The macroprudential stance and vulnerability to the global financial cycle’, Journal of International Economics 136, 103582. - Chernenko, S., Erel, I. and Prilmeier, R. (2022), ‘Why Do Firms Borrow Directly from Non-banks?’, The Review of Financial Studies 35(11), 4902–4947. - Chernenko, S., Ialenti, R. and Scharfstein, D. (2025), Bank Capital and the Growth of Private Credit, Available at https://ssrn.com/abstract=5097437, SSRN. - Choi, S., Willems, T. and Yoo, S. Y. (2024), ‘Revisiting the monetary transmission mechanism through an industry-level differential approach’, Journal of Monetary Economics 145, 103556. - Cizel, J., Frost, J., Houben, A. and Wierts, P. (2019), ‘Effective Macroprudential Policy: Cross-Sector Substitution from Price and Quantity Measures’, Journal of Money, Credit and Banking 51(5), 1209–1235. - Claessens, S. (2015), ‘An overview of macroprudential policy tools’, Annual Review of Financial Economics 7, 397–422. - Claessens, S., Cornelli, G., Gambacorta, L., Manaresi, F. and Shiinad, Y. (2023), ‘Do Macroprudential Policies Affect Non-bank Financial Intermediation?’, International Journal of Central Banking 19(5), 185–236. - Correa, R., Paligorova, T., Sapriza, H. and Zlate, A. (2022), ‘Cross-border bank flows and monetary policy’, The Review of Financial Studies 35(1), 438–481. - Cucic, D. and Gorea, D. (2024), Nonbank Lending and the Transmission of Monetary Policy, Available at https://ssrn.com/abstract=3974863, SSRN. - De Haas, R. and Van Horen, N. (2013), ‘Running for the Exit? International Bank Lending During a Financial Crisis’, The Review of Financial Studies 26(1), 244–285. - Degryse, H., De Jonghe, O., Jakovljević, S., Mulier, K. and Schepens, G. (2019), ‘Identifying Credit Supply Shocks with Bank-Firm data: Methods and Applications’, Journal of Financial Intermediation 40(C). - Den Haan, W. J. and Sterk, V. (2011), ‘The Myth of Financial Innovation and the Great Moderation’, The Economic Journal 121(553), 707–739. - Doerr, S. and Schaz, P. (2021), ‘Geographic diversification and bank lending during crises’, Journal of Financial Economics 140(3), 768–788. - Drechsel, T. and Miura, K. (2025), The macroeconomic effects of bank regulation: New evidence from a high-frequency approach, Available at https://econweb.umd.edu/~drechsel/papers/bank_regulation_shocks.pdf, Working Paper. - Drechsler, I., Savov, A. and Schnabl, P. (2017), ‘The Deposits Channel of Monetary Policy’, The Quarterly Journal of Economics 132(4), 1819–1876. - Drechsler, I., Savov, A. and Schnabl, P. (2022), ‘How monetary policy shaped the housing boom’, Journal of Financial Economics 144(3), 992–1021. - Duchin, R. and Sosyura, D. (2014), ‘Safer ratios, riskier portfolios: Banks’ response to government aid’, Journal of Financial Economics 113(1), 1–28. - Duprey, T. and Tuzcuoglu, K. (2025), ‘High-frequency effects of macroprudential policy announcements’, Economics Letters 250, 112264. - Elliott, D., Meisenzahl, R., Peydró, J.-L. and Turner, B. (2022), Nonbanks, Banks, and Monetary Policy: U.S. Loan-Level Evidence since the 1990s, Working Paper 2022-27, Federal Reserve Bank of Chicago. - Elliott, D., Meisenzahl, R. R. and Peydró, J.-L. (2024), ‘Nonbank lenders as global shock absorbers: Evidence from us monetary policy spillovers’, Journal of International Economics 149, 103908. - Erel, I. and Inozemtsev, E. (2025), ‘Evolution of Debt Financing Toward Less-Regulated Financial Intermediaries in the United States’, Journal of Financial and Quantitative Analysis 60(3). - Financial Stability Board (2024), Global Monitoring Report on Non-Bank Financial Intermediation, Technical report, Financial Stability Board. - Fleckenstein, Q., Gopal, M., Gallardo, G. G. and Hillenbrand, S. (forthcoming), ‘Nonbank Lending and Credit Cyclicality’, The Review of Financial Studies. - Gambacorta, L. (2005), ‘Inside the bank lending channel’, European Economic Review 49(7), 1737–1759. - Gambacorta, L. and Murcia, A. (2020), ‘The impact of macroprudential policies in Latin America: An empirical analysis using credit registry data’, Journal of Financial Intermediation 42, 100828. - Gambacorta, L. and Shin, H. S. (2018), ‘Why bank capital matters for monetary policy’, Journal of Financial Intermediation 35, 17–29. - Gebauer, S. and Mazelis, F. (2023), ‘Macroprudential regulation and leakage to the shadow banking sector’, European Economic Review 154, 104404. - Giannetti, M. and Laeven, L. (2012), ‘Flight home, flight abroad, and international credit cycles’, American Economic Review 102(3), 219–224. - Gómez, E., Murcia, A., Lizarazo, A. and Mendoza, J. C. (2020), ‘Evaluating the impact of macroprudential policies on credit growth in Colombia’, Journal of Financial Intermediation 42, 100843. - Hachem, K. (2011), ‘Relationship lending and the transmission of monetary policy’, Journal of Monetary Economics 58(6), 590–600. - Haque, S., Jang, Y. S. and Wang, J. J. (2025), Indirect Credit Supply: How Bank Lending to Private Credit Shapes Monetary Policy Transmission, Available at https://ssrn.com/abstract=5125733, SSRN. - Imbierowicz, B., Löffler, A. and Vogel, U. (2021), ‘The transmission of bank capital requirements and monetary policy to bank lending in Germany’, Review of International Economics 29(1), 144–164. - IMF (2024), The Rise and Risks of Private Credit, in ‘Global Financial Stability Report Spring 2024’, Chapter 2, International Monetary Fund, pp. 53–76. - Irani, R. M., Iyer, R., Meisenzahl, R. R. and Peydró, J.-L. (2021), ‘The rise of shadow banking: Evidence from capital regulation’, The Review of Financial Studies 34(5), 2181–2235. - Ivashina, V. (2009), ‘Asymmetric information effects on loan spreads’, Journal of Financial Economics 92(2), 300–319. - Ivashina, V., Scharfstein, D. S. and Stein, J. C. (2015), ‘Dollar funding and the lending behavior of global banks’, The Quarterly Journal of Economics 130(3), 1241–1281. - Jiménez, G., Ongena, S., Peydró, J.-L. and Saurina, J. (2012), ‘Credit Supply and Monetary Policy: Identifying the Bank Balance-Sheet Channel with Loan Applications’, American Economic Review 102(5), 2301–26. - Jiménez, G., Ongena, S., Peydró, J.-L. and Saurina, J. (2017), ‘Macroprudential Policy, Countercyclical Bank Capital Buffers, and Credit Supply: Evidence from the Spanish Dynamic Provisioning Experiments’, Journal of Political Economy 125(6), 2126–2177. - Khwaja, A. I. and Mian, A. (2008), ‘Tracing the Impact of Bank Liquidity Shocks: Evidence from an Emerging Market’, American Economic Review 98(4), 1413–1442. - Kim, S., Plosser, M. C. and Santos, J. A. (2018), ‘Macroprudential policy and the revolving door of risk: Lessons from leveraged lending guidance’, Journal of Financial Intermediation 34, 17–31. - Krainer, J., Vaghef, F. and Wang, T. (2024), Bank Lending to Nonbanks: A Robust Channel Fueled by Constrained Capital?, Available at https://www.fdic.gov/system/files/2024-09/vaghefi-paper-090624.pdf, Working Paper. - Nelson, B., Pinter, G. and Theodoridis, K. (2018), ‘Do Contractionary Monetary Policy Shocks Expand Shadow Banking?’, Journal of Applied Econometrics 33(2), 198–211. - Peters, R. H. and Taylor, L. A. (2017), ‘Intangible capital and the investment-q relation’, Journal of Financial Economics 123(2), 251–272. - Rojas, D., Vegh, C. and Vuletin, G. (2022), ‘The macroeconomic effects of macroprudential policy: Evidence from a narrative approach’, Journal of International Economics 139, 103644. - Romer, C. D. and Romer, D. H. (2004), ‘A New Measure of Monetary Shocks: Derivation and Implications’, American Economic Review 94(4), 1055–1084. - Sette, E. and Gobbi, G. (2015), ‘Relationship lending during a financial crisis’, Journal of the European Economic Association 13(3), 453–481. - Sufi, A. (2007), ‘Information Asymmetry and Financing Arrangements: Evidence from Syndicated Loans’, The Journal of Finance 62(2), 629–668. - Xiao, K. (2020), ‘Monetary Transmission through Shadow Banks’, The Review of Financial Studies 33(6), 2379–2420. ### Online appendix: figures, tables, and key numeric values - Figure OA.1: Loan amount outstanding: raw data versus sample - Vertical axis labelled Loan Outstanding ($ Billion). - Time span: 1990q1 to 2025q1. - Note: "The blue line is the loan amount outstanding in the full Dealogic dataset, and the red line is the loan amount outstanding in our estimation sample." - Table OA.1: Sources of monetary policy shocks - Columns: Country | Identification | Source | Start Date | End Date - United States: High-Frequency | Jarocinski and Karadi (2020) | 1990Q1–2024Q1 - Euro Area (14 countries): High-Frequency | Jarocinski and Karadi (2020) | 1999Q1–2023Q4 - United Kingdom: High-Frequency | Cesa-Bianchi et al. (2020) | 1997Q1–2015Q4 - Sweden: High-Frequency | Amberg et al. (2022) | 1999Q1–2018Q4 - Japan: High-Frequency | Kubota and Shintani (2022) | 1992Q1–2020Q4 - India: High-Frequency | Lakdawala and Sengupta (2021) | 2003Q1–2020Q4 - Canada: CBFD (a la R&R 2004) | Champagne and Sekkel (2018) | 1974Q1–2015Q4 - Brazil: CBFD (a la R&R 2004) | Alberola et al. (2021) | 1974Q1–2015Q4 - Norway: CBFD (a la R&R 2004) | Holm et al. (2021) | 1990Q1–2018Q4 - Note: "CBFD refers to Central Bank Forecasts Deviations." - Figure OA.2: Net cumulative sum of macroprudential actions - Index plotted 1990–2020. - Series shown: World, AEs, EMs. - Note: "Alam et al. (2025) iMaPP database and authors’ calculations. The red vertical line marks the start of the Covid-19 pandemic in 2020Q1." - Figure OA.3: Country-level responses to MaPP shocks - Panels for Real GDP growth and Real private credit growth. - "Cumulative impulse responses over 20 quarters following a one-standard deviation increase in MaPP shocks. Blue line is the point estimate, and dark (light) grey areas refer to the associated 68% (90%) confidence bands. Standard errors clustered by country." - Table OA.2: Asymmetric effects of monetary policy shocks (selected coefficients and counts) - Dependent variable: log of new syndicated loans (columns 1-5) and spread in bps (column 6). - MP+ coefficient examples: -0.089*** (SE 0.005), -0.110*** (SE 0.006). - MP− Shock examples: 0.068*** (SE 0.005), 0.081*** (SE 0.005). - MP+ Shock×Nonbank: 0.068*** (SE 0.005). - MP− Shock×Nonbank: -0.047*** (SE 0.005). - Observations: 756,531; 751,549; 749,629; 748,453; 739,866; 386,185 (varies by column). - R2 values: 0.689, 0.687, 0.727, 0.795, 0.877, 0.988. - Note: "Standard errors clustered by firm. Asterisks, *, **, and ***, denote statistical significance at the 10%, 5%, and 1% levels." - Table OA.3: Asymmetric effects of macroprudential policy shocks (selected coefficients and counts) - MaPP+ coefficient examples: -0.047*** (SE 0.006), -0.069*** (SE 0.006). - MaPP− Shock examples: 0.034*** (SE 0.005), 0.039*** (SE 0.005). - MaPP+ Shock×Nonbank: 0.066*** (SE 0.007). - MaPP− Shock×Nonbank: -0.010; -0.037*** (SE 0.006). - Observations: same pattern as Table OA.2. - R2 values: 0.688, 0.687, 0.726, 0.795, 0.877, 0.988. - Table OA.4: Robustness checks: major country lenders (selected entries) - Baseline MP shock: -0.021*** (SE 0.002). - MP shock×Nonbank: 0.046*** (SE 0.003). - MaPP shock×Nonbank: 0.016*** (SE 0.002). - Observations across columns: 739,866; 287,195; 37,690; 36,275; 159,575; 668,573. - R2 values: 0.877; 0.842; 0.821; 0.871; 0.868; 0.881. - Table OA.5: Rescaling coefficients with bank stock prices, and using Drechsel and Miura (2025) bank regulation shocks - Baseline MP shock: -0.021*** (SE 0.002). - Base: normalized MP shock: -0.025*** (SE 0.003). - MP shock×Nonbank examples: 0.046*** (SE 0.003), 0.055*** (SE 0.004), 0.128*** (SE 0.009). - MaPP shock×Nonbank examples: 0.016*** (SE 0.002), 0.040*** (SE 0.006). - Observations: 739,866; 739,818; 287,195; 287,195; 202,117. - R2 values: 0.877; 0.877; 0.842; 0.842; 0.852. - Table OA.6: Distribution of sample across shocks combination - MP+ × MaPP+: 211,198 - MP+ × MaPP−: 152,032 - MP− × MaPP+: 199,881 - MP− × MaPP−: 176,755 - Total: 739,866 - Note: "Total number of observations for each monetary policy-macroprudential policy shocks combination." - Table OA.7: Relationship lending and margin spread (selected coefficients and counts) - Relation coefficients: Relation -1.209** (SE 0.526); Relation×Nonbank -1.900*** (SE 0.534). - MP shock×Nonbank: -3.175*** (SE 0.612); -2.824*** (SE 0.553). - MaPP shock×Nonbank: 2.274*** (SE 0.506); 2.040*** (SE 0.468). - Observations: 388,592; 388,439; 386,185. - R2 values: 0.840; 0.855; 0.988. - Table OA.8: Real effects (selected coefficients and counts) - Dependent variables: (1) Total Debt, (2) Liquid assets, (3) Capital exp., (4) Intangibles, (5) Employment, (6) PD. - MP shock examples: 0.077* (SE 0.040) for Total Debt; 0.056** (SE 0.024) for Intangibles. - MaPP shock examples: -0.074** (SE 0.030) for Total Debt; -0.082*** (SE 0.031) for Liquid assets. - NB relation: 0.141*** (SE 0.019) for Total Debt. - Observations by column: 239,871; 70,572; 45,381; 44,552; 303,919; 400. - R2 values: 0.916; 0.898; 0.974; 0.963; 0.982; 0.610. - Note: "NB relation dummy equals one if the firm borrowed from a nonbank in the syndicate loan market in the past two years, and zero otherwise." - Table OA.9: Alternative lag structures and shock specification (selected coefficients) - MP shock coefficients across lags: -0.021*** (Lag 1), -0.020*** (Lag 2), -0.021*** (Lag 3), -0.022*** (Lag 4). - MaPP shock coefficients across lags: -0.019*** (Lag 1) through -0.019*** (Lag 4). - MP shock×Nonbank: 0.046*** (Lag 1) to 0.054*** (Lag 3). - Observations vary from 716,867 to 739,866 depending on specification. - R2 values: 0.877 to 0.880 across columns. - Table OA.10: Alternative clustering methods (selected coefficients) - MP shock: -0.021*** (column 1 SE 0.002) to -0.021** (column 4 SE 0.008). - MaPP shock: -0.019*** (SE 0.002) consistent across columns. - MP shock×Nonbank: 0.046*** (column 1 SE 0.003) to 0.046** (column 4 SE 0.019). - MaPP shock×Nonbank: 0.016*** (column 1 SE 0.002) to 0.016* (column 4 SE 0.010). - Observations: 739,866 for all columns. - R2: 0.877 for all columns. - Note: "Standard errors clustered by firm (column 1), firm and country-time (column 2), lender and country-time (column 3), and by firm, lender and time (column 4)." *From banks to nonbanks: macroprudential and monetary policy effects on corporate lending Working Paper No. WP/2025/096* --- _Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025096-print-pdf.pdf_