## wpiea2026023-source-pdf — Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5 ## Source details **Canonical URL:** [wpiea2026023-source-pdf — Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5](https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026023-source-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2026/english/wpiea2026023-source-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2026/english/wpiea2026023-source-pdf.pdf.json) --- ### Data: sample composition and lender/borrower coverage - Final sample spans 27 lender countries (21 AEs and 6 EMDEs) from 2005Q1 to 2023Q4. - Lenders and subsidiaries: - 963 unique banking groups holding 4,871 subsidiaries. - 75 percent of lender entities are bank subsidiaries and 25 percent are NBFIs. - Conditional on having at least one NBFI, each banking group has on average 12.4 bank subsidiaries and 6.2 NBFI subsidiaries. - Unconditional averages in the full sample: 3.8 bank subsidiaries and 1.3 NBFI subsidiaries. - 438 listed banking groups and 525 unlisted; 90 percent of lending is originated by listed banking groups. - Borrowers: - 52,380 unique nonfinancial firms across 150 borrower countries (38 AEs and 112 EMDEs). - Private NFCs account for 66 percent of total borrowing. - Expanded sample (Section 3.6): - 939 nonbanking parent companies to study independent NBFIs. - Independent NBFIs account for about two-thirds of all NBFI entities in the expanded sample. - Bank-affiliated NBFIs account for roughly 27 percent of syndicated loans, compared with 9 percent by other NBFIs. ### Syndicated-loan dataset, lender classification, and measurement - Primary source: Dealogic syndicated loan data covering nearly all primary-market loans issued worldwide by nonfinancial firms. - Main loan measure: US dollar amount of newly issued loans, deflated by each country’s CPI deflator. - Price metric used in some exercises: all-in drawn spread (interest rate margin plus fees over LIBOR). - Lender classification rules: - Banks: Dealogic SIC codes beginning with 60. - Nonbanks: SIC codes 61–67, excluding most codes starting with 65 and mortgage broker code 6162; allocated 6510, 6519, 6532 to nonbanks. - Text-based keyword methods flag banks for missing SIC codes (e.g., ‘bank,’ ‘banco,’ ‘banca,’ ‘banque’) and identify investment banks via ‘investment’. - International financial institutions and development banks excluded. - Imputation of missing participation: - Assign 50 percent of the loan to lead arrangers and distribute remainder equally among other participants. - Results robust to alternative imputations; Section 5 uses regression-based approach from Blickle et al. forthcoming. - Lender composition and loan characteristics: - NBFI sample mainly includes investment banks and broker-dealers (64 percent of bank-owned NBFI subsidiaries), plus investment funds, asset managers (including hedge funds, private equity, and other alternative investment vehicles), insurance companies, and pension funds. - NBFIs typically take larger tranche values, especially bank-owned, and charge higher spreads (Table A.5). - Median term lengths around five years across lender types. ### Balance-sheet, firm risk, and ownership data - Banking-group balance-sheet variables from S&P Compustat and Capital IQ: leverage ratios, log total assets, NPLs (percent of loans), ROA, Tier 1 capital ratios, 24-month PD from NUS-CRI. - 513 unique listed banks identified with Compustat balance-sheet information. - Firm-level risk proxies: - 24-month PD from NUS-CRI; leverage ratio (total debt/total assets); firm age. - Zombie dummy (ICR < 1, leverage > median in country-industry pair, negative real sales growth for at least two consecutive years). - ‘Vulnerable’ firms: top quintile debt and first quintile liquid assets. - Firms with loan-weighted average spread over past five years in top quartile; leveraged loans defined as spread >150 basis points over Libor. - Matched around 5.6k unique firms with Compustat data; roughly 19.3k firms with non-missing Dealogic spread data. - Novel time-varying bank ownership dataset (2000–2024): - Focus on top 250 bank and top 250 nonbank lender entities in Dealogic (top 250 NBFIs ~90 percent of NBFI lending; top 250 bank entities ~84 percent of bank lending). - Constructed from Dealogic ultimate parent field plus dated ownership-change information (FINRA BrokerCheck, SEC Forms 10‑K/8‑K/BD, Companies House, national registries, regulatory disclosures, annual reports, press releases, regulatory approvals, LEI data). - LLMs used to assist identification, standardization, and reconciliation of affiliate names. - Excludes subsidiaries that became government-owned (including U.S. Treasury TARP recipients in 2009). - Dataset limited to subsidiaries active in the syndicated loan market. ### Macroprudential (MaPP) shocks: construction and empirical properties - Primary MaPP source: iMaPP macroprudential database of Alam et al. (2025), recording changes in 17 MaPP instruments across 134 countries. - Analysis restricted to 27 countries and to measures directly constraining banks’ lending capacity, split into three categories: 1. Loan-supply measures: limits to credit growth, loan-loss provisioning requirements, loan restrictions, loan-to-deposit ratio caps, limits on FX lending. 2. Reserve requirements. 3. Prudential measures: stress testing, restrictions on profit distribution, structural measures (limits on exposures between financial institutions). - Coding: binary variables (1 tightening, −1 loosening, 0 no change). - Baseline MaPP shock extraction (purging endogenous responses): - Cumulate and aggregate binary indicators at parent-country level. - Regress MaPP index on country fixed effects and lagged macro and financial controls; residuals ε_{c,t} taken as exogenous MaPP shocks: - MaPP_{c,t} = β1 Macro_{c,t−1} + β2 Financial_{c,t−1} + α_{c} + ε_{c,t} - Lagged macro controls: real GDP growth, REER growth, year-on-year CPI inflation, five-year ahead real GDP forecast. - Financial controls: year-on-year change in real house prices, private credit-to-GDP growth, ten-year government bond yield, ten-year government bond yield gap (relative to equivalent U.S. bond yield), Chinn-Ito index, banks’ average Z-score. - Empirical properties: - Contractionary MaPP shock strengthens banks’ resilience but dampens real activity. - Following a one-standard-deviation MaPP tightening: bank Z-score improves; GDP growth, private credit, and bank stock prices decline. - High-frequency (HF) MaPP announcements compiled for six major economies used in Section 4 as a robustness test. - Summary statistics (Table A.6): - Baseline shocks — Obs.: 2,571; Mean: -0.02; S.D.: 0.96; P25: -0.76; P50: -0.03; P75: 0.67 - HF shocks — Obs.: 2,880; Mean: 0.02; S.D.: 0.97; P25: -0.47; P50: 0.27; P75: 0.76 ### Empirical model and identification - Unit of analysis: lender–borrower–quarter level; main sample restricted to banking groups to compare bank vs NBFI subsidiaries. - Primary regression: - Log(Loans)_{l,j,i,t} = β1 MaPP_{c,t−1} + β2 MaPP_{c,t−1} × NBFIsubs._{l,j} + β3 NBFIsubs._{l,j} + γ_{j} + μ_{i,t} + ε_{l,j,i,t} - Dependent variable: logarithm of US dollar amount of new syndicated loans by lender l (bank or NBFI subsidiary) in group j to firm i at time t. - MaPP_{c,t−1}: lagged MaPP shocks at banking group-country level. - NBFIsubs._{l,j}: dummy equal to one when lender is a NBFI subsidiary within a banking group. - γ_{j}: banking-group fixed effects; μ_{i,t}: firm × quarter fixed effects. - Interpretation: - β1: lending response of bank subsidiaries to a one-standard-deviation MaPP tightening. - β2: differential lending by NBFI subsidiaries relative to bank subsidiaries within same group, same borrower, same quarter. - Robustness/alternative controls: - ILST fixed effects (industry–location–size–time) to relax firm × quarter assumption. - Spread regressions aggregated at tranche–lender–borrower–quarter level use ILST and add years to maturity. - Standard errors clustered at firm level. - Extensions: Section 3.6 expands sample to include independent NBFIs (not owned by banking groups). ### Main empirical findings — intra-group reallocation (Section 3.2) - Bank lending in syndicated loan market falls by 0.022 (‑2.2 percent) after a one-standard deviation MaPP tightening (Table 1, column (1)). - Controlling for demand with firm×quarter fixed effects reduces decline to −0.004 (Table 1, column (2)). - Banking groups expand lending through their NBFI affiliates relative to bank subsidiaries: - MaPP shock × NBFI subs. = 0.020 (Table 1, column (3)). - In column (3): bank subsidiaries reduce lending by 0.010 (‑1.0 percent); NBFI affiliates expand lending by 0.020 (2.0 percent) relative to bank entities, and by 0.010 (1.0 percent) in absolute terms. - Back‑of‑the‑envelope offset using average NBFI lending share of 28.4 percent implies banking groups with NBFI affiliates offset more than half of contractionary MaPP effect: - Group-level reduction in syndicated lending: -0.43 percent (calculation provided in text). - Mitigation ratio: -0.57, or -57 percent (calculation provided in text). ### Heterogeneity, margins, pricing, and risk composition - Jurisdictional drivers: - Main effects driven by U.S. banking groups and to a lesser extent euro area groups. - NBFI types: - Investment banks and broker-dealers primarily drive increase in lending; investment funds and asset managers do not increase syndicated-loan lending post-MaPP tightening. - Extensive margin: - NBFI subsidiaries more likely to form new lending relationships in response to MaPP tightening (Appendix Table B.2), though opposite found for U.K. and euro area nonbank subsidiaries in that table. - Pricing: - Bank-owned NBFI subsidiaries adjust facility-level spreads by an additional 1.5 basis points in response to a one‑standard deviation MaPP shock (Table B.3, column (3)); magnitude described as negligible and is additional to already higher average spreads charged by nonbanks. - Risk-taking and borrower composition: - Augmented specifications using proxies (PD, Spread 5y, Lev. loan, Lev. ratio, Age, Vuln, Zombie) show no statistical evidence that NBFI lending increases disproportionately to risky borrowers (Table 2). - In some specs, NBFIs may favor lending to less risky borrowers (lower spreads, not leveraged loans). - No evidence NBFI subsidiaries increase exposure to zombie firms following tighter regulation. - Overall: substitution from bank to nonbank subsidiaries appears to decrease, or at least maintain, overall portfolio risk for banking groups. ### Banking-group financial strength and substitution intensity - Bank strength proxies: PD, NPLs, Tier 1 capital ratio (T1R), Size (log total assets); weak groups defined by quartile thresholds. - Findings: - Weaker banking groups reduce lending through bank affiliates more and increase lending through affiliated NBFI subsidiaries by more (Table 3). - MaPP shock × NBFI subs. × Bank charact. coefficients: 0.025 (PD column), -0.013 (NPLs column, not significant), 0.025 (T1R column), 0.025 (Size column) — indicating stronger reallocation toward NBFIs for weaker/smaller/lower‑capitalized banks in several specifications. ### Between‑group versus within‑group substitution - Including individual lender fixed effects reveals inter‑group substitution: - Banking groups increase lending via NBFI affiliates relative to bank subsidiaries by ~0.9 percent (between‑group effect), roughly half the magnitude of the within‑group effect. ### Foreign subsidiaries and cross‑border lending (Section 3.4) - Foreign bank subsidiaries help cushion home-country MaPP tightening: - Foreign bank subsidiaries increase lending by 2.4 percent relative to domestic bank subsidiaries (1.1 percent in absolute terms). - NBFI location differences: - Domestic NBFIs expand lending by 2.1 percent relative to foreign counterparts. - Foreign NBFIs do not adjust lending; sum of coefficients yields statistically insignificant −0.4 percent. - Foreign NBFIs account for 2.2 percent of all banking group lending over sample and 0.9 percent in 2024. - Cross-border behavior: - Cross-border lending declines on average after a MaPP tightening in source country; foreign affiliates mitigate decline (Table 4). - In cross-border settings, foreign bank and foreign NBFI subsidiaries play larger offsetting roles than domestic NBFIs. - Domestic NBFIs used to sustain home-country lending; foreign bank and foreign NBFI subsidiaries used to support lending abroad. - Heterogeneity by market presence: - In core markets (top quartile of country loan share), foreign bank subsidiaries increase lending by an additional 5.5 percent; foreign NBFIs expand by 1.1 percent in core markets relative to other markets (Table 5). - In non-core markets, foreign subsidiaries cut lending relative to other markets. - Host-country MaPP shocks: - When MaPP tightens in borrower (host) country, banking-group credit supply falls on average; domestic and host NBFI subsidiaries expand lending relative to home bank subsidiaries (domestic NBFIs +2.1 percent; host NBFIs +1.6 percent). - Cross-border lending declines on average by 4 percent (Table 6, column 4). ### Bank‑owned NBFIs versus other NBFIs (Section 3.6) - Expanded sample composition: - 955 unique banking groups; 939 unique nonbanking groups. - Entities: 2,950 bank entities; 1,692 NBFIs not affiliated with banking groups; 824 bank-owned NBFI subsidiaries. - Sample shares: bank-affiliated NBFIs ~27 percent of all loans; other NBFIs ~9 percent. - Expanded model (individual lender fixed effects γ_l) interpretation: - β1: bank entities’ average response; β2: additional response for NBFIs not owned by banking groups; β3: additional response for bank-owned NBFI subsidiaries relative to other NBFIs. - Key empirical findings: - Other NBFIs reduce lending to NFCs by 0.7 percent relative to banks following a one-standard-deviation MaPP tightening (column 3, Table 7). - Bank-owned NBFI subsidiaries increase lending by 1.8 percent relative to other NBFIs (column 3, Table 7). - Selected coefficients (Table 7): - MaPP shock (col 1): -0.029 ∗∗∗ (standard error (0.003)). - MaPP shock×NBFI (col 3): -0.007 ∗∗ (standard error (0.003)). - MaPP shock×NBFI×BG subs. (col 3): 0.018 ∗∗∗ (standard error (0.003)). - U.S. subsample (col 4): MaPP shock -0.009 ∗∗∗ (standard error (0.003)); MaPP shock×NBFI -0.021 ∗∗∗ (standard error (0.007)); MaPP shock×NBFI×BG subs. 0.062 ∗∗∗ (standard error (0.007)). - R^2 values reported across columns: 0.303, 0.895, 0.895, 0.856, 0.847, 0.890; observations across columns include 782,594; 761,629; 761,629; 234,795; 32,599; 150,994. - Market-structure interaction (crowding-out) results (Table 8): - In countries where banking groups do not have a large presence, other NBFIs increase lending by 3.9 percent and 3.7 percent relative to bank and NBFI subsidiaries after tightening MaPP shocks. - Triple interaction MaPP_{c,t−1} × NBFI_{l,j} × Ctry shr_{w,t−1} shows other NBFIs reduce lending substantially relative to bank-owned NBFI and bank subsidiaries where banking groups have high lending share. - Selected coefficients (Table 8): MaPP shock×NBFI (All) 0.039 ∗∗∗ (s.e. 0.010); MaPP shock×NBFI×BG subs. (All) -0.037 ∗∗∗ (s.e. 0.010); MaPP shock×NBFI×Ctry shr (All) -0.051 ∗∗∗ (s.e. 0.010); MaPP shock×NBFI×BG subs.×Ctry shr (All) 0.060 ∗∗∗ (s.e. 0.011). - Mechanisms: - Bank-owned NBFIs likely benefit from internal capital markets, parental funding access, and implicit guarantees enabling expansion when parent banks face tighter MaPP. - Other NBFIs face competitive constraints where banking groups have large presence and may be crowded out. ### High‑frequency MaPP announcements and main robustness (Section 4.2) - Using HF MaPP surprises: - Parent banks curtail credit by 0.9 percent through bank affiliates (MaPP shock coefficient: -0.009 in Table 10, col 3). - NBFI subsidiaries expand lending by 1.5 percent relative to bank subsidiaries (MaPP shock×NBFI subs. 0.015 in Table 10, col 3). - Baseline (non-HF) vs HF estimates: - Baseline: MaPP shock -0.010 (Table 10, col 1); MaPP shock×NBFI subs. 0.020 (col 1). - HF-restricted sample: MaPP shock -0.016; MaPP shock×NBFI subs. 0.022 (Table 10, col 2). - Lending spreads: - NBFI subsidiaries increase spreads by ~2.9 basis points relative to bank subsidiaries (MaPP shock×NBFI subs. 2.878 in Table 10, col 8). - Robustness and MaPP subcomponents (Table 11): - Excluding reserve requirements (RR) and stress-testing (ST): MaPP shock×NBFI subs. remains positive; loan-supply measures have strong effects enabling substitution. - Capital-based measures: MaPP shock 0.004 (col 4) and 0.008 (HF col 5); MaPP shock×NBFI subs. -0.014 (col 4) and -0.019 (HF col 5), indicating capital-based, consolidated measures constrain scope for substitution toward NBFIs. - Additional robustness: - Results robust to alternative abnormal-return calculations, inclusion of country–banking group×time fixed effects, lagged bank-level controls, excluding large loans (USD 100 million) or very large U.S. bank holding company subsidiaries, alternative clustering, lag lengths, and post-origination participation adjustments (Blickle et al. forthcoming). ### Magnitudes and selected model fit statistics (Table 10 and Table 11 highlights) - Table 10 (log new syndicated loans): - Column (1): MaPP shock -0.010 ∗∗∗; MaPP shock×NBFI subs. 0.020 ∗∗∗; NBFI subs. 0.039 ∗∗∗; Observations 696,913; R2 0.8940. - Column (2): MaPP shock -0.016 ∗∗∗; MaPP shock×NBFI subs. 0.022 ∗∗∗; Observations 534,142; R2 0.8990. - Column (3): MaPP shock -0.009 ∗∗∗; MaPP shock×NBFI subs. 0.015 ∗∗∗; Observations 332,712; R2 0.9100. - Column (8) (spreads): MaPP shock -0.462 (s.e. 0.603); MaPP shock×NBFI subs. 2.878 ∗∗∗ (s.e. 0.614); Observations 273,263; R2 0.371. - Table 11 (alternative MaPP shocks): - Baseline (col 1): MaPP shock -0.010 ∗∗∗; MaPP shock×NBFI subs. 0.020 ∗∗∗; Observations 696,913; R2 0.8940. - Capital-based (col 4): MaPP shock 0.004 ∗∗∗; MaPP shock×NBFI subs. -0.014 ∗∗∗; Observations 696,913; R2 0.8940. - HF: capital-based (col 5): MaPP shock 0.008 ∗∗∗; MaPP shock×NBFI subs. -0.019 ∗∗∗; Observations 295,639; R2 0.926. ### Mechanisms, systemic implications, and interpretation - Mechanism: lender‑based MaPP announcements directly constrain parent banks’ lending to NFCs, creating incentives to shift credit intermediation toward less-regulated NBFI affiliates within banking groups. - Geographic heterogeneity: effects concentrated in the U.S. and, to a lesser extent, the euro area; UK and selected European lenders show no statistically significant responses. - Loan-supply vs capital-based measures: - Loan-supply prudential measures (targeting segments or products at subsidiary/local level) facilitate substitution toward NBFIs. - Capital-based consolidated measures reduce scope for substitution and may lead parents to curtail lending group-wide. - Potential consequences: - Banking groups can offset more than half of MaPP-induced group-level credit contraction by reallocating toward NBFI subsidiaries. - Increased bank–nonbank interconnectedness may weaken MaPP effectiveness and raise systemic risk, concentration risk, and rapid stress transmission potential from NBFIs to parent banks. - Bank support to nonbank affiliates during stress may produce inefficient capital allocation and higher funding costs. ### Policy discussion and recommendations - Regulatory implications: - Expanding the regulatory perimeter to better encompass nonbanks is recommended to improve understanding of credit risks and financial stability implications of NBFIs’ growing role in credit intermediation. - Several regulators (United Kingdom, Australia, Singapore) have started integrating system‑wide stress tests and scenario analyses to assess banks’ exposures to NBFIs. - International policy cooperation, regulatory harmonization, and enhanced information sharing are important to mitigate cross‑border loopholes and address data gaps. - Research implications: - Analysis centers on syndicated loan market; open question whether MaPP tightening contributed to expansion of private credit in opaque, lightly regulated markets. ### Appendix A: selected data tables and figures (highlights) - Table A.1 and Table A.2: selected bank and nonbank subsidiary parent changes over time (examples include Credit Suisse → UBS Group AG → 2023; Silicon Valley Bank → First Citizens BancShares Inc → 2023). - Table A.3 (bank‑owned NBFIs counts and shares): - Investment banks, broker-dealers: 784 — 64.2 - Business credit institutions: 26 — 2.1 - Insurance and pension funds: 26 — 2.1 - Investment funds & asset managers: 152 — 12.4 - Other NBFI subsidiaries: 234 — 19.2 - Total: 1,222 — 100 - Table A.4 (lender-country average loan shares): - USA — 35.13; GBR — 10.29; JPN — 12.16; CAN — 8.40; FRA — 6.43; DEU — 5.65; others listed. - Table A.5 (loan-level characteristics by lender type, selected entries): - Banks: Term Length mean 5.19; Tranche Value mean 241.84 (US million); All-in-Pricing mean 232.68 (bps). - Bank-owned NBFI subs.: Term Length mean 4.89; Tranche Value mean 351.44 (US million); All-in-Pricing mean 249.63 (bps). - Other NBFIs: Term Length mean 6.17; Tranche Value mean 270.79 (US million); All-in-Pricing mean 386.80 (bps). - Figure A.3 and Table A.6 summarize MaPP shock distributions and HF vs baseline shock statistics. *Source: wpiea2026023-source-pdf — Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5* ### Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5 ### wpiea2026023-source-pdf - Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5 ### Data: overview - Final sample spans 27 lender countries (21 AEs and 6 EMDEs) from 2005Q1 to 2023Q4. - Covers 963 unique banking groups, which together hold 4,871 subsidiaries. - Among these lender entities, 75 percent are bank subsidiaries and 25 percent are NBFIs. - Conditional on having at least one NBFI, each banking group has on average 12.4 bank subsidiaries and 6.2 NBFI subsidiaries. - Unconditional averages in the full sample: 3.8 bank subsidiaries and 1.3 NBFI subsidiaries. - In total, 438 are listed banking groups, while 525 are unlisted. - The bulk of the lending (90 percent) is originated by listed banking groups. - Borrower coverage: 52,380 unique nonfinancial firms across 150 borrower countries (38 AEs and 112 EMDEs). - Private NFCs account for 66 percent of total borrowing. - Expanded sample (Section 3.6) includes 939 nonbanking parent companies to study independent NBFIs. - In the expanded sample, independent NBFIs account for about two-thirds of all NBFI entities. - Bank-affiliated NBFIs account for roughly 27 percent of syndicated loans, compared with 9 percent by other NBFIs. ### Syndicated loans dataset and construction - Primary source: Dealogic syndicated loan data covering nearly all primary-market loans issued worldwide by nonfinancial firms. - Main variable of interest: US dollar amount of newly issued loans, deflated by each country’s CPI deflator. - Price metric used in some exercises: all-in drawn spread (interest rate margin plus fees over LIBOR). - Lender classification: - Banks: Dealogic SIC codes beginning with 60 (depository institutions). - Nonbanks: SIC codes 61–67, excluding most codes starting with 65 (real estate firms) and certain mortgage brokers (6162). - Allocated codes starting in 65 to nonbanks: 6510, 6519, and 6532. - For lenders without SIC codes, text-based keyword methods flag banks (e.g., ‘bank,’ ‘banco,’ ‘banca,’ ‘banque’) and identify investment banks through ‘investment’. - International financial institutions and development banks are excluded. - Imputation of missing loan participation amounts: - Assign 50 percent of the loan to lead arrangers and distribute the remainder equally among other participants. - Results are robust to alternative imputation techniques; Section 5 uses a regression-based approach from Blickle et al. forthcoming. - Lender composition and behavior: - Sample captures 963 unique banking groups and 4,871 subsidiaries. - NBFI sample mainly includes investment banks and broker-dealers (64 percent of the sample of NBFI subsidiaries), and also investment funds, asset managers (including hedge funds, private equity, and other alternative investment vehicles), insurance companies, and pension funds. - NBFIs typically take larger tranche values, especially those bank-owned, and charge higher spreads (Table A.5 in Appendix A). - Other NBFIs (not bank-owned) charge considerably higher spreads, potentially signaling greater risk appetite. - Median term lengths are around five years across lender types. ### Balance sheet data (banking groups and firms) - Sources: S&P Compustat and Capital IQ; matched using Capital IQ identifiers in Dealogic and the Capital IQ–Compustat link. - Banking group variables retrieved: - Leverage ratios (debt to equity or debt to assets), size (log total assets), nonperforming loans as percent of total loans (NPLs), return on assets (ROA), Tier 1 capital ratios. - Probability of default (PD) over the next 24 months from NUS-CRI (modified Merton distance-to-default). - Identified 513 unique listed banks with Compustat balance sheet information. - Nonfinancial firm variables and risk proxies: - PD over the next 24 months from NUS-CRI. - Leverage ratio: total debt (short- and long-term) as a share of total assets. - Firm age. - Zombie dummy from Albuquerque and Iyer (2024): ICR < 1, leverage > median in country-industry pair, real sales growth negative, all for at least two consecutive years. - ‘Vulnerable’ firms: top quintile debt and first quintile liquid assets (Albuquerque 2024). - Firms with loan-weighted average spread over past five years in the top quartile of the sample distribution in each quarter. - Leveraged loans: borrowers whose spread is over 150 basis points over Libor. - Matching coverage: - Matched around 5.6k unique firms with Compustat data. - Roughly 19.3k firms with non-missing Dealogic data on spreads. ### Bank ownership structure dataset - Novel time-varying dataset linking bank and nonbank subsidiaries from 2000 to 2024. - Focus: top 250 bank and top 250 nonbank lender entities in Dealogic. - Top 250 NBFIs account for around 90 percent of total lending by NBFIs in the full sample. - Top 250 bank entities account for roughly 84 percent of total lending by banks. - Construction steps and sources: - Start with the ultimate parent field from Dealogic (static snapshot of >50% control). - Compile dated ownership-change information from FINRA’s BrokerCheck, SEC filings (Forms 10-K, 8-K, and BD), Companies House (U.K.), national business registries, regulatory disclosures, annual reports, press releases, regulatory acquisition approvals, and global Legal Entity Identifier relationship data. - Cross-validate ownership transitions and distinguish mergers, internal reorganizations, partial divestitures, and genuine changes in ultimate control. - Leverage LLMs to assist in identification, standardization, and classification of affiliates and to reconcile naming inconsistencies and detect missing links. - Capabilities and limitations: - Reconstruct a continuous ownership timeline for each lender and record ownership changes (mergers, takeovers, restructurings, rebrandings). - Example events recorded: BB&T and SunTrust Banks combination into Truist Financial Corporation and Truist Bank in 2019; Bank of Scotland into Lloyds Banking Group in 2009; Fortis Bank into BNP Paribas Fortis in 2008; Credit Suisse First Boston ownership until June 12, 2023 and absorption into UBS Group following government-facilitated rescue. - Exclude subsidiaries that became government-owned (including those receiving U.S. Treasury TARP capital injections in 2009). - Dataset limited to subsidiaries active in the syndicated loan market and may be incomplete beyond that scope. - Appendices: Appendix A contains full lists of ownership changes (Table A.1 for bank subsidiaries, Table A.2 for nonbank subsidiaries) and the time-varying banking group ownership dataset is made available. ### Macroprudential (MaPP) shocks: construction and properties - Primary MaPP source: iMaPP macroprudential database of Alam et al. (2025), recording changes in 17 MaPP instruments across 134 countries. - Analysis restricts to 27 countries and measures directly constraining banks’ lending capacity, split into three categories: 1. Measures focusing on loan supply: limits to credit growth, loan-loss provisioning requirements, loan restrictions, loan-to-deposit ratio caps, and limits on FX lending. 2. Reserve requirements. 3. Prudential measures focusing on stress testing, restrictions on profit distribution, and structural measures (e.g., limits on exposures between financial institutions). - Coding: All MaPP measures coded as binary variables: one for tightening, minus one for loosening, zero for no change. - Baseline MaPP shock extraction (purging endogenous response): - Cumulate and aggregate binary indicators at the country level of the parent company. - Regress MaPP index on country fixed effects and lagged macroeconomic and financial controls, take residuals as exogenous MaPP shocks: - MaPP_{c,t} = β1 Macro_{c,t−1} + β2 Financial_{c,t−1} + α_{c} + ε_{c,t} - Lagged macroeconomic controls: real GDP growth, real effective exchange rate (REER) growth, year-on-year CPI inflation, five-year ahead real GDP forecast. - Financial controls: year-on-year change in real house prices, private credit-to-GDP growth, ten-year government bond yield, ten-year government bond yield gap (relative to equivalent U.S. bond yield), Chinn-Ito index of financial openness, banks’ average Z-score. - Resulting shock series balanced between tightening and loosening episodes (Figure A.3 in Appendix A). - Empirical properties: - A contractionary MaPP shock strengthens banks’ resilience but dampens real activity. - Following a one-standard-deviation MaPP tightening: - Bank balance sheet health improves (higher Z-score). - GDP growth, private credit, and bank stock prices decline (Figure B.1 in Appendix B). - High-frequency MaPP announcements compiled for six major economies are used in Section 4 as a robustness test. ### Main results: empirical model and identification - Unit of analysis: lender–borrower–quarter level, sample restricted to banking groups to test differential lending by NBFI subsidiaries relative to bank subsidiaries within the same banking group. - Primary regression specification: - Log(Loans)_{l,j,i,t} = β1 MaPP_{c,t−1} + β2 MaPP_{c,t−1} × NBFIsubs._{l,j} + β3 NBFIsubs._{l,j} + γ_{j} + μ_{i,t} + ε_{l,j,i,t} - Dependent variable: logarithm of US dollar amount of new syndicated loans of lender l belonging to banking group j granted to nonfinancial firm i at time t. - MaPP_{c,t−1}: lagged MaPP shocks at banking group-country level. - NBFIsubs._{l,j}: dummy equal to one when loans are intermediated by a NBFI subsidiary within a banking group. - γ_{j}: banking-group fixed effects. - μ_{i,t}: firm × quarter fixed effects (controls for time-varying borrower characteristics). - Interpretation: - β1 indicates the lending response of bank subsidiaries following a one-standard-deviation MaPP tightening shock. - β2 measures the differential lending provided by NBFI subsidiaries relative to bank subsidiaries within the same banking group after the MaPP shock. - Identification of β2 relies on variation in lending across subsidiary types (bank versus nonbank) that belong to the same parent, lend to the same borrower, and in the same quarter. - Robustness and alternative controls: - Robustness checks relax the firm × quarter assumption by controlling for credit demand using industry–location–size–time (ILST) fixed effects (Degryse et al. 2019). - ILST fixed effects compare borrowers within the same two-digit industry, country, and quarter, and account for firm size by grouping firms into quartile bins of total borrowing volume within each country-year pair. - In spread (price) regressions: - Data aggregated at the tranche–lender–borrower–quarter level. - Use ILST fixed effects rather than firm × quarter fixed effects to preserve cross-lender variation. - Add years to maturity as an additional control variable. - Standard errors are clustered at the firm level to address within-firm correlation and potential dependence across multiple loans to the same borrower. - Extensions: - Section 3.6 expands the sample to include NBFIs not owned by banking groups to contrast behavior of independent NBFIs with bank-affiliated NBFIs. *Source: wpiea2026023-source-pdf - Section 4 discusses the high-frequency MaPP identification as a robustness test. Section 5* ### 3.2 Banks’ regulatory-induced reallocation? Lending through NBFIs ### 3.2 Banks’ regulatory-induced reallocation? Lending through NBFIs ### Main empirical finding: intra-group reallocation of credit - Bank lending in the syndicated loan market falls by 0.022 (‑2.2 percent) after a one-standard deviation MaPP tightening shock (Table 1, column (1)). - Controlling for demand with firm×quarter fixed effects reduces the magnitude of the bank lending decline (Table 1, column (2): MaPP shock = -0.004). - Banking groups expand lending through their NBFI affiliates relative to their bank subsidiaries: MaPP shock × NBFI subs. = 0.020 (Table 1, column (3)). - Bank subsidiaries reduce lending by 0.010 (‑1.0 percent) (column (3)). - NBFI affiliates expand lending by 0.020 (2.0 percent) relative to bank entities, and by 0.010 (1.0 percent) in absolute terms (column (3)). - Back‑of‑the‑envelope offset calculation using the average NBFI lending share of 28.4 percent implies banking groups with NBFI affiliates were able to offset more than half of the contractionary effect of MaPP tightening. - Group-level reduction in syndicated lending: -0.43 percent (calculation: 1.0×0.284 - 1.0×0.716 = -0.43 percent). - Mitigation ratio: -0.57, or -57 percent (calculation: (-0.43 + 1.0)/-1.0 = -0.57). ### Heterogeneity by jurisdiction and nonbank type - Main effects driven by U.S. banking groups (Table 1, column (4)) and to a lesser extent by euro area banking groups (column (6)). - Among NBFIs, investment banks and broker-dealers primarily drive the increase in lending (Appendix Table B.4). - Investment funds and asset managers (including hedge funds, private equity, and other alternative investment vehicles) do not increase syndicated‑loan lending following tighter MaPP shocks; they may be more active in private or bilateral credit markets. ### Extensive margin and pricing responses - Reallocation also materializes along the extensive margin: NBFI subsidiaries are more likely to form new lending relationships in response to MaPP tightening (Appendix Table B.2), though the opposite is found for U.K. and euro area nonbank subsidiaries in that table. - Bank‑owned NBFI subsidiaries pass on higher regulatory costs to borrowers, but the economic magnitude is negligible: - Relative to bank subsidiaries, NBFI subsidiaries adjust facility‑level spreads by an additional 1.5 basis points in response to a one‑standard deviation MaPP shock (Table B.3, column (3)). - This negligible increase is in addition to already higher average spreads charged by nonbanks. ### Risk‑taking and borrower risk composition - Augmented specification (Equation (3)) tests whether NBFI lending increases disproportionately to risky borrowers using time‑varying proxies for Risky borrowers (PD, Spread 5y, Lev. loan, Lev. ratio, Age, Vuln, Zombie). - Across alternative risk indicators, no statistical evidence that the increase in lending by NBFI subsidiaries during MaPP shocks is disproportionately channeled to risky borrowers (Table 2). - In some specifications (Table 2, columns (2) and (3)), NBFIs may favor lending to less risky borrowers—firms with lower spreads and not leveraged loans. - Although NBFIs tend to be more exposed to risky borrowers on average (Table 2 shows higher lending to high‑PD firms, highly leveraged firms, and firms with limited liquid assets), their exposure does not increase in response to MaPP tightening shocks. - No evidence that NBFI subsidiaries increase exposure to zombie firms due to tighter regulation. - Overall: substituting lending from bank to nonbank subsidiaries appears to decrease, or at least maintain, overall portfolio risk for banking groups. ### Banking group financial strength and differential responses - We examine heterogeneity by parent banking‑group strength using four proxies: PD, NPLs, Tier 1 capital ratio (T1R), and Size (log total assets). - Weak groups defined as PD or NPLs in top quartile, or T1R or Size in bottom quartile. - Weaker banking groups reduce lending through bank affiliates more following MaPP tightening (Table 3, third row). - Weaker banking groups increase lending through affiliated NBFI subsidiaries by more (Table 3, fourth row). - MaPP shock × NBFI subs. × Bank charact. coefficients: 0.025 (PD column), -0.013 (NPLs column, not significant), 0.025 (T1R column), 0.025 (Size column) — indicating stronger reallocation toward NBFIs for weaker/ smaller/ lower‑capitalized banks in several specifications. - Interpretation: financially weaker banking groups—higher distress risk, lower capitalization, or smaller size—appear to shift lending more toward NBFI affiliates that are less directly constrained by bank regulation. ### Between‑group versus within‑group substitution - Including individual lender fixed effects (Appendix Table B.5) reveals inter‑group credit supply substitution: - Banking groups increase lending via their NBFI affiliates relative to bank subsidiaries by approximately 0.9 percent (between‑group effect). - This between‑group effect is roughly half the magnitude of the within‑group effect documented in Table 1. ### Regulatory context and mechanisms - Regulatory perimeter differences facilitate reallocation: - In the United States, commercial banks are regulated by the Federal Reserve, OCC, and FDIC and face stringent capital, liquidity, activity, and safety‑and‑soundness requirements. - Most NBFI affiliates (broker‑dealers, insurers, asset managers) are not subject to bank‑like capital or liquidity regulation and fall under functional regulators (e.g., SEC, FINRA) whose mandates emphasize investor protection and market integrity rather than systemic risk. - Regulation Y subjects U.S. bank holding companies to consolidated supervision by the Federal Reserve, but nonbank subsidiaries are not directly subject to bank‑level capital, liquidity, or other macroprudential requirements. - Resulting regulatory architecture: macroprudential tightening primarily binds insured depository institutions, while nonbank affiliates are affected only indirectly through consolidated constraints—providing scope to reallocate syndicated‑loan exposures toward NBFI arms. *Source: 3.2 Banks’ regulatory-induced reallocation? Lending through NBFIs — wpiea2026023-source-pdf* ### 3.4 Foreign subsidiaries and cross-border lending ### 3.4 Foreign subsidiaries and cross-border lending ### Core empirical findings on subsidiaries and NBFIs - Foreign bank subsidiaries help parent banks cushion the impact of tighter home-country macroprudential policies: - Foreign bank subsidiaries increase lending by 2.4 percent relative to domestic bank subsidiaries (1.1 percent in absolute terms). - NBFI response differs by location: - Domestic NBFIs expand lending by 2.1 percent relative to their foreign counterparts. - Foreign NBFIs do not adjust lending in response to MaPP shocks (sum of all coefficients yields a statistically insignificant point estimate of –0.4 percent). - Foreign NBFIs account for a small and declining share of banking group lending in the syndicated loan market: - 2.2 percent of all banking group lending over the estimation sample. - 0.9 percent in 2024. - Interpretation: the overall mitigation effect of NBFIs following home-country MaPP tightening is driven primarily by domestic NBFIs rather than by foreign NBFIs. ### Cross-border lending and the role of foreign affiliates - Cross-border lending behavior after MaPP tightening in the source country: - Cross-border lending declines on average after a MaPP tightening shock in the source country (negative MaPP shock effect reported in Table 4). - Foreign affiliates mitigate the fall in cross-border lending (Table 4, column 5). - In cross-border settings, both foreign bank and foreign NBFI subsidiaries play a larger role than domestic NBFIs in offsetting the impact of tighter regulation (Table 4, column 6). - Distinct uses of entities by parent banks: - Domestic NBFIs are primarily used to sustain lending to home-country borrowers. - Foreign bank and foreign NBFI subsidiaries are used primarily to support lending abroad. - Importance of internal capital markets: increased lending activity by domestic NBFIs and foreign affiliates highlights the role of banking groups’ internal capital markets in reallocating funds across borders. ### Heterogeneity by lenders’ country presence (core vs non-core markets) - Definition: for each banking group, the average share of loans to a given country relative to total cross-border lending; “high country loan share” = top quartile, “low country loan share” = bottom quartile. - Evidence from Table 5: - In response to tighter MaPP, banking groups use foreign affiliates to mitigate contractions in markets where they have a large presence (column 1). - The increased lending response in core markets is mainly driven by foreign bank subsidiaries and, to a lesser extent, by foreign NBFI subsidiaries (column 2). - Foreign bank subsidiaries increase lending by an additional 5.5 percent in core markets relative to other markets (sum of third and fourth columns). - Foreign NBFIs expand lending by 1.1 percent in core markets relative to other markets (0.4 + 5.1 - 3.7 - 0.7). - In non-core markets, foreign subsidiaries cut lending relative to other markets (columns 3-4). - Overall interpretation: banking groups rely on a dual strategy: - Foreign bank and NBFI subsidiaries mitigate the effect of regulation on cross-border lending—especially in core markets. - Domestic NBFIs cushion the effect on lending at home. ### MaPP shocks in the host country (subsidiary-country tightening) - Setup: re-estimate with MaPP shocks matched at borrower-country (host-country) level; distinguish ‘host’ subsidiaries (in the country where MaPP tightens), ‘home’ subsidiaries (in the parent’s home/source country), and ‘foreign’ subsidiaries (in a third country). - Main findings (Table 6): - Banking groups’ credit supply falls following MaPP tightening shocks in the host country (negative average effect). - Relative to home bank subsidiaries, both domestic and host NBFI subsidiaries expand lending: - Domestic NBFI subsidiaries expand lending by 2.1 percent relative to home bank subsidiaries. - Host NBFI subsidiaries expand lending by 1.6 percent relative to home bank subsidiaries. - Host bank subsidiaries contract lending in a manner similar to domestic bank subsidiaries (the coefficient of 0.001 in column 2 is not statistically significant, but the average effect is negative). - Foreign bank affiliates located in a third country increase lending differentially by 1.6 percent relative to home bank subsidiaries. - Foreign NBFIs curtail lending somewhat (and account for a very small share of total lending). - Cross-border restriction results (Table 6, columns 4–6): - Cross-border lending declines on average by 4 percent (column 4). - The decline is broad-based across different subsidiaries, suggesting regulatory shocks abroad lead parent banks to privilege domestic lending. - Robustness: adding country–lender MaPP shocks does not materially change these results (column 3 of Table 6). ### Overall interpretation and contribution - Novel intra-group adjustment channel documented: - Banking groups rely not only on foreign bank affiliates (as in prior literature) but also on domestic NBFIs to buffer the impact of macroprudential tightening on syndicated loan supply. - Location-specific strategies: - When regulation tightens at home: parent banks reallocate cross-border lending toward foreign affiliates and use domestic NBFIs to sustain lending at home. - When regulation tightens in host countries: parent banks partially offset reductions by channeling lending through domestic and host NBFI subsidiaries, and through bank subsidiaries located in third countries. - Implication: entities with looser regulation, or outside the regulatory perimeter, play an important role in maintaining syndicated lending when regulation tightens. *Source: 3.4 Foreign subsidiaries and cross-border lending (wpiea2026023-source-pdf)* ### 3.6 Bank-owned NBFIs vs other NBFIs ### 3.6 Bank-owned NBFIs vs other NBFIs ### Overview - Extends the analysis to include NBFIs not owned by banking groups to test whether the observed post-MaPP increase in lending by NBFIs is specific to banking-group affiliation or applies more broadly to the NBFI business model. - Focus on differential lending responses of: bank entities, NBFIs not affiliated with banking groups, and bank-owned NBFI subsidiaries. ### Data and sample composition - 955 unique banking groups. - 939 unique nonbanking groups. - Entities in sample: - 2,950 bank entities. - 1,692 NBFIs not affiliated with banking groups. - 824 bank-owned NBFI subsidiaries. - Share of syndicated loans (sample averages): - Bank-affiliated NBFIs: around 27 percent of all loans. - Other NBFIs: 9 percent of all loans. ### Econometric specification (expanded model) - Expanded model (Equation (4)) estimated with individual lender fixed effects γ_l: - Log(Loans)_{l,j,i,t} = β1 MaPP_{c,t−1} + NBFI_{l,j} × (β2 MaPP_{c,t−1} + β3 MaPP_{c,t−1} × BG subs_j + β4 BG subs_j) + γ_l + μ_{i,t} + ε_{l,j,i,t}. - Interpretation of coefficients: - β1: average lending response of bank entities to a MaPP shock. - β2: additional lending response for NBFIs not owned by banking groups. - β3: additional lending response of bank-owned NBFI subsidiaries relative to other NBFIs. ### Key empirical findings - Aggregate estimated effects (text summary and Table 7 highlights): - Other NBFIs (not affiliated with banking groups) reduce lending to NFCs by 0.7 percent relative to banks following a one-standard-deviation MaPP tightening shock (column 3 of Table 7). - Bank-owned NBFI subsidiaries increase lending by 1.8 percent relative to other NBFIs. - Selected coefficients reported in Table 7: - MaPP shock (column 1): -0.029 ∗∗∗ (standard error (0.003)). - MaPP shock×NBFI (column 3): -0.007 ∗∗ (standard error (0.003)). - MaPP shock×NBFI×BG subs. (column 3): 0.018 ∗∗∗ (standard error (0.003)). - MaPP shock (column 4, U.S.): -0.009 ∗∗∗ (standard error (0.003)). - MaPP shock×NBFI (column 4, U.S.): -0.021 ∗∗∗ (standard error (0.007)). - MaPP shock×NBFI×BG subs. (column 4, U.S.): 0.062 ∗∗∗ (standard error (0.007)). - Results are driven by lenders from the United States (column 4 of Table 7), consistent with earlier findings. - R^2 values reported in Table 7: 0.303, 0.895, 0.895, 0.856, 0.847, 0.890 across columns reported; observations range (e.g., 782,594; 761,629; 761,629; 234,795; 32,599; 150,994). ### Market-structure interaction and crowding-out analysis - Hypothesis: other NBFIs face tighter competitive constraints where banking groups have a large presence; bank-owned NBFIs may crowd out other NBFIs in these markets. - Extended specification (Equation (5)) introduces Ctry shr_{w,t−1}, a dummy for countries where the share of syndicated loans originated by banking groups in country w is in the top quartile. - Key results (Table 8 highlights): - In countries where banking groups do not have a large presence, after a tightening MaPP shock: - Other NBFIs increase lending by 3.9 percent and 3.7 percent respectively relative to bank and NBFI subsidiaries (column (1) textual summary). - The triple interaction MaPP_{c,t−1} × NBFI_{l,j} × Ctry shr_{w,t−1} indicates other NBFIs reduce lending substantially relative to bank-owned NBFI and bank subsidiaries in countries where banking groups have a high share of lending. - Effects driven entirely by domestic lending (column (2) of Table 8). - Selected coefficients from Table 8: - MaPP shock×NBFI (All): 0.039 ∗∗∗ (standard error (0.010)). - MaPP shock×NBFI×BG subs. (All): -0.037 ∗∗∗ (standard error (0.010)). - MaPP shock×NBFI×Ctry shr (All): -0.051 ∗∗∗ (standard error (0.010)). - MaPP shock×NBFI×BG subs.×Ctry shr (All): 0.060 ∗∗∗ (standard error (0.011)). - R^2 in Table 8: 0.895, 0.911, 0.858 across columns (observations: 761,626; 446,693; 281,083). ### Mechanisms and interpretation - Competitive advantage of bank-owned NBFIs likely driven by access to internal capital markets and parent resources: - Parent banks can reallocate funds across affiliates when their capital position is affected, supporting expansion of NBFI subsidiaries. - Even if NBFI subsidiaries cannot collect deposits, affiliation provides indirect access to the parent’s deposit base or wholesale lines and potentially implicit guarantees. - Literature support: - Banking-group affiliation documented to provide capital support, faster recovery, and lower failure probability for affiliates (Ashcraft 2008 and cited literature). - Bai et al. (2025) find nonbank affiliates receive more favorable credit conditions from parent banks than bank affiliates. ### Implications - The increase in lending by nonbanks after MaPP shocks is driven exclusively by NBFIs belonging to banking groups, not by other NBFIs—implying that lighter regulation alone does not explain broad-based NBFI expansion after MaPP tightening. - Where banking groups hold large domestic market shares, bank-owned NBFIs appear to protect or expand market share at the expense of other NBFIs, consistent with a crowding-out/competition effect. - Internal capital market advantages and parent support are plausible channels for bank-owned NBFIs’ differential behavior. *Source: 3.6 Bank-owned NBFIs vs other NBFIs, wpiea2026023-source-pdf* ### 4.2 Main results ### 4.2 Main results ### Empirical findings on lending reallocation - Using HF MaPP surprises, parent banks curtail credit by 0.9 percent through their bank affiliates (MaPP shock coefficient: -0.009 in column (3) of Table 10). - NBFI subsidiaries expand lending by 1.5 percent relative to bank subsidiaries in response to MaPP announcements (MaPP shock×NBFI subs. coefficient: 0.015 in column (3) of Table 10). - Baseline estimates using non-HF MaPP shocks: MaPP shock coefficient -0.010 (column (1) of Table 10) and MaPP shock×NBFI subs. 0.020 (column (1) of Table 10). - Estimates restricted to HF sample produce qualitatively similar results (column (2) of Table 10: MaPP shock -0.016; MaPP shock×NBFI subs. 0.022). - The reallocation effect is driven primarily by U.S. banking groups; no statistically significant responses are found for UK and the selected European lenders (Table 10, columns (4)–(7) and accompanying text). ### Lending spreads and pricing - The increase in lending by NBFI subsidiaries is accompanied by an increase in spreads charged to NFCs of around 2.9 basis points relative to bank subsidiaries (column (8) of Table 10: MaPP shock×NBFI subs. 2.878). - Interpretation: banking groups attempt to pass on higher regulatory costs via higher loan prices through NBFI affiliates, though magnitudes are described as negligible. ### Robustness checks and alternative MaPP specifications - Results are robust to using two-day changes in excess returns of bank stock prices and to using estimated daily abnormal returns instead of excess returns (reference to Table B.6; qualitative result preserved). - Excluding reserve requirements (RR) and stress-testing (ST) from the baseline MaPP specification: - Column (2) of Table 11 (Base excl. RR): MaPP shock -0.008; MaPP shock×NBFI subs. 0.018. - Column (3) of Table 11 (Base excl. RR and ST): MaPP shock -0.017; MaPP shock×NBFI subs. 0.040. - Interpretation: loan-supply measures appear to have particularly strong effects on curtailing bank credit, creating opportunities for NBFI subsidiaries. - Capital-based measures (leverage limits, countercyclical buffers, conservation buffers, capital requirements): - Column (4) of Table 11 (Capital-based): MaPP shock 0.004; MaPP shock×NBFI subs. -0.014 (NBFI subsidiaries reduce lending by 1.4 percent relative to bank subsidiaries). - Column (5) of Table 11 (HF: capital-based): MaPP shock 0.008; MaPP shock×NBFI subs. -0.019. - Interpretation: capital-based measures at the consolidated level constrain banks’ scope to shift lending toward NBFI affiliates; these measures may induce parent banks to curtail lending across the group. - Additional robustness checks (summarized): - Inclusion of country–banking group×time fixed effects yields highly similar estimates. - Adding lagged bank-level controls (log total assets, ROA, NPLs, Tier 1 capital ratio, leverage ratio, two-year PD) leaves estimates virtually unchanged (sample size declines). - Banking group×time fixed effects do not overturn main findings. - Excluding loans above USD 100 million and excluding subsidiaries of U.S. bank holding companies with total consolidated assets above USD 100 billion (FR Y-14Q reporting concerns) confirm main findings despite smaller samples (Table B.8). - Adjusting for post-origination participation shares following Blickle et al. (forthcoming) does not eliminate the documented regulatory-induced lending reallocation (Table B.9), though imputation caveats are noted. - Adding time fixed effects, using MaPP shocks lagged two, three, and four quarters, and alternative clustering approaches (firm and time; banking group; firm, banking group and time; country-banking group×time) all leave baseline results qualitatively unchanged (Tables B.10–B.12). - Notes on abnormal returns computation: abnormal returns computed as residuals from regressions of bank stock prices on a constant and the aggregate stock market index of each country (relaxing baseline beta=1 assumption). ### Magnitudes, sample and model fit (selected statistics from tables) - Table 10 (dependent variable: log of new syndicated loans): - Column (1): MaPP shock -0.010 ∗∗∗; MaPP shock×NBFI subs. 0.020 ∗∗∗; NBFI subs. 0.039 ∗∗∗; Observations 696,913; R2 0.8940. - Column (2): MaPP shock -0.016 ∗∗∗; MaPP shock×NBFI subs. 0.022 ∗∗∗; Observations 534,142; R2 0.8990. - Column (3): MaPP shock -0.009 ∗∗∗; MaPP shock×NBFI subs. 0.015 ∗∗∗; Observations 332,712; R2 0.9100. - Column (8) (spreads, aggregated at tranche-lender-borrower-quarter level): MaPP shock -0.462 (standard error 0.603); MaPP shock×NBFI subs. 2.878 ∗∗∗ (standard error 0.614); Observations 273,263; R2 0.371. - Table 11 (alternative MaPP shocks, dependent variable: log of new syndicated loans): - Column (1) Baseline: MaPP shock -0.010 ∗∗∗; MaPP shock×NBFI subs. 0.020 ∗∗∗; Observations 696,913; R2 0.8940. - Column (4) Capital-based: MaPP shock 0.004 ∗∗∗; MaPP shock×NBFI subs. -0.014 ∗∗∗; Observations 696,913; R2 0.8940. - Column (5) HF: capital-based: MaPP shock 0.008 ∗∗∗; MaPP shock×NBFI subs. -0.019 ∗∗∗; Observations 295,639; R2 0.926. ### Mechanisms, heterogeneity, and interpretation - Mechanism: lender-based MaPP announcement measures directly constrain parent banks’ lending to nonfinancial corporates, strengthening incentives for banking groups to shift credit intermediation toward NBFI affiliates. - Geographic heterogeneity: effects are concentrated in the U.S. and, to a lesser extent, the euro area; UK and selected European lenders show no statistically significant responses. - Capital-based vs loan-supply measures: - Loan-supply prudential measures often target specific market segments or products at the subsidiary/local level, facilitating substitution toward NBFIs. - Capital-based measures largely apply at the consolidated group level, prompting parent banks to curtail lending across all subsidiaries and limiting scope for substitution. - Potential consequences of reallocation: - Banking groups can offset more than half of the contractionary impact of MaPP on overall group-level credit supply by reallocating lending toward NBFI subsidiaries (aggregate effect summarized in Conclusion). - Increased bank–nonbank interconnectedness may weaken macroprudential policy effectiveness and raise systemic risk, concentration risk within groups, and potential for rapid stress transmission from NBFIs to parent banks. - NBFI distress can lead parent banks to reallocate capital to support nonbanks, potentially producing inefficient capital allocation and higher funding costs. ### Policy discussion and implications - Regulatory implications: - Expanding the regulatory perimeter to better encompass nonbanks is recommended to improve understanding of underlying credit risks and the financial stability implications of NBFIs’ expanding role in credit intermediation. - Several regulators (United Kingdom, Australia, Singapore) have begun integrating system-wide stress tests and scenario analyses to assess banks’ exposures to NBFIs. - International policy cooperation, regulatory harmonization, and enhanced information sharing are important to mitigate cross-border regulatory loopholes and address data gaps. - Research implications: - The paper’s analysis centers on the syndicated loan market; an open question remains whether tighter macroprudential policies have contributed to the expansion of private credit, which operates in relatively opaque and lightly regulated markets and raises additional financial stability concerns. *Source: wpiea2026023-source-pdf — section 4.2 Main results* ### Appendix A: Data ### Appendix A: Data ### A.1 Bank subsidiaries’ parent changes over time (Table A.1) - This table reports changes in the ultimate parent bank entities over time for selected banks in the Dealogic syndicated loan data sample. - Selected entries (Bank Name → Parent Name → Year): - ABN AMRO Bank NV → ABN AMRO Holding NV → 2000 - ABN AMRO Bank NV → RFS Holdings BV → 2007 - ABN AMRO Bank NV → State of the Netherlands → 2008 - ABN AMRO Bank NV → ABN AMRO Group NV → 2010 - ABN AMRO Bank NV → ABN AMRO Bank NV → 2019 - BBVA Compass → BBVA Compass → 2000 - BBVA Compass → Banco Bilbao Vizcaya Argentaria SA → 2007 - BBVA Compass → PNC Financial Services Group Inc → 2021 - Bank of Scotland → HBOS plc → 2001 - Bank of Scotland → Lloyds Banking Group plc → 2009 - Bank of the West → BNP Paribas SA → 2000 - Bank of the West → Bank of Montreal → 2023 - Bankia SA → Banco Financiero y de Ahorros SA - BFA → 2010 - Bankia SA → Bankia SA → 2012 - Bankia SA → CaixaBank SA → 2021 - Credit Suisse → Credit Suisse International → 2000 - Credit Suisse → UBS Group AG → 2023 - Silicon Valley Bank → Silicon Valley Bank → 2000 - Silicon Valley Bank → First Citizens BancShares Inc → 2023 - MUFG Union Bank NA → Mitsubishi UFJ Financial Group Inc → 2000 - MUFG Union Bank NA → US Bancorp → 2022 - Union Bank NA → Mitsubishi UFJ Financial Group Inc → 2005 - Union Bank NA → US Bancorp → 2022 - Notes: The table includes many additional bank subsidiary parent changes beyond the selected entries above. ### A.2 Nonbank subsidiaries’ parent changes over time (Table A.2) - This table reports changes in the ultimate nonbank parent entities over time for selected nonbank subsidiaries. - Selected entries (Nonbank Name → Parent Name → Year): - ABN AMRO Capital USA LLC → ABN AMRO Holding NV → 2000 - ABN AMRO Capital USA LLC → RFS Holdings BV → 2007 - ABN AMRO Capital USA LLC → ABN AMRO Group NV → 2010 - ABN AMRO Capital USA LLC → ABN AMRO Bank NV → 2019 - Abbey National Treasury Services plc → Abbey National plc → 2000 - Abbey National Treasury Services plc → Banco Santander SA → 2004 - Alcentra Ltd → Bank of New York Mellon Corp → 2000 - Alcentra Ltd → Franklin Resources Inc → 2022 - Bear Stearns & Co Inc → Bear Stearns & Co Inc → 2000 - Bear Stearns & Co Inc → JPMorgan Chase & Co → 2008 - BlueMountain Capital Management LLC → BlueMountain Capital Management LLC → 2003 - BlueMountain Capital Management LLC → Assured Guaranty Ltd → 2019 - BlueMountain Capital Management LLC → Sound Point Capital Management LP → 2023 - Credit Suisse (Singapore) Ltd → Credit Suisse International → 2000 - Credit Suisse (Singapore) Ltd → UBS Group AG → 2023 - Lehman Brothers International (Europe) → Lehman Brothers Holdings Inc → 2000 - Lehman Brothers International (Europe) → Nomura Holdings Inc → 2008 - Lehman Brothers North America → Lehman Brothers Holdings Inc → 2000 - Lehman Brothers North America → Barclays plc → 2008 - Wachovia Capital Markets LLC → Wachovia Corporation → 2000 - Wachovia Capital Markets LLC → Wells Fargo & Co → 2008 - Western Asset Management Co → Legg Mason Wood Walker Capital Markets → 2000 - Western Asset Management Co → Franklin Resources Inc → 2020 - Notes: The table includes many additional nonbank subsidiary parent changes beyond the selected entries above. ### A.3 Bank-owned NBFIs: counts and shares (Table A.3) - Number and share of unique bank-owned NBFIs: - Investment banks, broker-dealers: 784 — 64.2 - Business credit institutions: 26 — 2.1 - Insurance and pension funds: 26 — 2.1 - Investment funds & asset managers: 152 — 12.4 - Other NBFI subsidiaries: 234 — 19.2 - Total: 1,222 — 100 - Notes: Number and share of unique bank-owned NBFIs. ### A.4 Macroprudential measures and MaPP shocks (Figures A.1, A.3; Table A.6) - Figure A.1: Net cumulative sum of macroprudential measures (index) plotted 1999–2023 for: - World, AEs, EMs. - Notes: Alam et al. (2025) iMaPP database and authors’ calculations. The red vertical line marks the start of the Covid-19 pandemic in 2020Q1. - Figure A.3: MaPP shocks over time (median and IQR) plotted 2005q1–2025q1. - Notes: Blue solid line is the median MaPP shock sample values; grey area is the interquartile range. - Table A.6: Summary statistics for baseline and high-frequency (HF) shocks: - Baseline shocks — Obs.: 2,571; Mean: -0.02; S.D.: 0.96; P25: -0.76; P50: -0.03; P75: 0.67 - HF shocks — Obs.: 2,880; Mean: 0.02; S.D.: 0.97; P25: -0.47; P50: 0.27; P75: 0.76 - Notes: Summary statistics of baseline vs HF shocks. ### A.5 Lending to NFCs in the syndicated loan market: U.S. lenders (Figure A.2) - Left panel: Outstanding stock of loans (USD (billion)) plotted 2005q1–2025q1 broken down by: - Domestic Bank Entities (blue area) - Foreign Bank Entities (red area) - Domestic NBFI Subs. (green area) - Foreign NBFI Subs. (yellow area) - Right panel: NBFI subsidiaries’ share in banking group lending (Share (%)) plotted 2005q1–2025q1. - Notes: Left panel shows volume of outstanding syndicated loans from banking groups to NFCs by lender type. Right panel shows share of syndicated loans originated by NBFI subsidiaries in total banking group lending. ### A.6 Lender countries and loan shares (Table A.4) - Loan share is the average loan share for each lender country in the estimation sample. - Country — Share: - AUS — 1.90 - AUT — 0.40 - BEL — 0.30 - BRA — 0.19 - CAN — 8.40 - CHE — 2.52 - CHN — 3.09 - DEU — 5.65 - DNK — 0.31 - ESP — 3.16 - FIN — 0.63 - FRA — 6.43 - GBR — 10.29 - IND — 0.76 - IRL — 0.30 - ITA — 2.37 - JPN — 12.16 - KOR — 0.18 - MYS — 0.22 - NLD — 3.03 - NOR — 0.51 - PRT — 0.12 - SGP — 0.86 - SWE — 0.71 - THA — 0.12 - USA — 35.13 - ZAF — 0.27 ### A.7 Loan-level characteristics by lender type (Table A.5) - Banks (Obs; Mean; STD; P25; P50; P75): - Term Length — 220,526; Mean: 5.19; STD: 4.15; P25: 3.00; P50: 5.00; P75: 6.00 - Tranche Value (US million) — 228,666; Mean: 241.84; STD: 461.27; P25: 23.32; P50: 78.87; P75: 242.40 - # Lender — 231,400; Mean: 5.86; STD: 5.33; P25: 2.00; P50: 4.00; P75: 7.00 - Lender Share — 916,147; Mean: 0.18; STD: 0.19; P25: 0.06; P50: 0.12; P75: 0.25 - All-in-Pricing (bps) — 404,973; Mean: 232.68; STD: 147.81; P25: 125.00; P50: 200.00; P75: 300.00 - Margin-Pricing (bps) — 285,307; Mean: 225.72; STD: 142.56; P25: 125.00; P50: 200.00; P75: 300.00 - Bank-owned NBFI subsidiaries: - Term Length — 122,253; Mean: 4.89; STD: 2.83; P25: 3.00; P50: 5.00; P75: 5.50 - Tranche Value (US million) — 126,270; Mean: 351.44; STD: 558.34; P25: 48.22; P50: 143.75; P75: 393.77 - # Lender — 127,083; Mean: 7.30; STD: 6.10; P25: 3.00; P50: 6.00; P75: 9.00 - Lender Share — 295,478; Mean: 0.18; STD: 0.17; P25: 0.07; P50: 0.12; P75: 0.25 - All-in-Pricing (bps) — 200,189; Mean: 249.63; STD: 152.96; P25: 140.00; P50: 225.00; P75: 325.00 - Margin-Pricing (bps) — 168,558; Mean: 244.12; STD: 150.84; P25: 137.50; P50: 200.00; P75: 325.00 - Other NBFIs: - Term Length — 57,174; Mean: 6.17; STD: 4.27; P25: 4.00; P50: 5.00; P75: 7.00 - Tranche Value (US million) — 61,044; Mean: 270.79; STD: 475.07; P25: 31.98; P50: 96.48; P75: 286.72 - # Lender — 62,914; Mean: 7.28; STD: 6.79; P25: 3.00; P50: 5.00; P75: 9.00 - Lender Share — 107,160; Mean: 0.23; STD: 0.23; P25: 0.07; P50: 0.12; P75: 0.25 - All-in-Pricing (bps) — 70,058; Mean: 386.80; STD: 183.72; P25: 250.00; P50: 375.00; P75: 500.00 - Margin-Pricing (bps) — 57,185; Mean: 391.58; STD: 175.53; P25: 275.00; P50: 375.00; P75: 500.00 - Notes: Summary statistics of loan-level characteristics by lender type, restricted to loans to nonfinancial borrowers. *Source: Appendix A: Data, wpiea2026023-source-pdf* ### References ### References ### Bank–nonbank linkages and systemic risk - BCBS (2025), Banks’ interconnections with non-bank financial intermediaries, Basel Committee on Banking Supervision, Bank for International Settlements, Basel. Published 10 July 2025. - Bochmann, P., Dieckelmann, D., Grodzicki, M., Horan, A., Larkou, C. and Lenoci, F. (2025), Systemic risks in linkages between banks and the non-bank financial sector, Financial Stability Review – Special Feature (November), European Central Bank. - Cerutti, E., Claessens, S. and Laeven, L. (2017), ‘The use and effectiveness of macroprudential policies: New evidence’, Journal of Financial Stability28, 203–224. - Cetorelli, N., Landoni, M. and Lu, L. (2023), Non-Bank Financial Institutions and Banks’ Fire-Sale Vulnerabilities, Staff Report 1057, Federal Reserve Bank of New York. - Cetorelli, N. and Prazad, S. (2025), The nonbank footprint of banks, Staff Reports 1118, Federal Reserve Bank of New York. - Franceschi, E., Grodzicki, M., Kagerer, B., Kaufmann, C., Lenoci, F., Mingarelli, L., Pancaro, C. and Senner, R. (2023), Key linkages between banks and the non-bank financial sector, Financial Stability Review (May), European Central Bank. - Financial Stability Board (2025), Global Monitoring Report on Nonbank Financial Intermediation 2025, Technical report, Financial Stability Board. - 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 Economics149, 103908. - Buchak, G., Matvos, G., Piskorski, T. and Seru, A. (2018), ‘Fintech, regulatory arbitrage, and the rise of shadow banks’, Journal of Financial Economics130(3), 453–483. - 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 Studies34(5), 2181–2235. - Lee, H., Lee, S. and Paluszynski, R. (2024), ‘Capital Regulation and Shadow Finance: A Quantitative Analysis’, The Review of Economic Studies91(5), 3047–3084. ### Macroprudential policy, announcements, and spillovers - 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 Money57(5), 1135–1166. - Bluwstein, K. and Patozi, A. (2024), The effects of macroprudential policy announcements on systemic risk, Bank of England working papers 1080, Bank of England. - Claessens, S., Cornelli, G., Gambacorta, L., Manaresi, F. and Shiinad, Y. (2023), ‘Do Macroprudential Policies Affect Non-bank Financial Intermediation?’, International Journal of Central Banking19(5), 185–236. - Duprey, T. and Tuzcuoglu, K. (2025), ‘High-frequency effects of macroprudential policy announcements’, Economics Letters250, 112264. - 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 Economics136, 103582. - 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. - Fabiani, J. and Neanidis, K. C. (2025), ‘The bank-lending channel of macroprudential policy: Evidence from cross-border bank flows’, Journal of International Economics157, 104135. - BCBS (2025), Banks’ interconnections with non-bank financial intermediaries, Basel Committee on Banking Supervision, Bank for International Settlements, Basel. Published 10 July 2025. ### Bank capital, regulation, and lending channels - 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. - Aiyar, S., Calomiris, C. W., Hooley, J., Korniyenko, Y. and Wieladek, T. (2014), ‘The international transmission of bank capital requirements: Evidence from the UK’, Journal of Financial Economics113(3), 368–382. - Bednarek, P., Briukhova, O., Ongena, S. and von Westernhagen, N. (2025), ‘Effects of bank capital requirements on lending by banks and non-bank financial institutions’, Journal of Financial Intermediation63, 101167. - Begenau, J. and Landvoigt, T. (2022), ‘Financial Regulation in a Quantitative Model of the Modern Banking System’, The Review of Economic Studies89(4), 1748–1784. - Chernenko, S., Ialenti, R. and Scharfstein, D. (2025), Bank Capital and the Growth of Private Credit, Available at https://ssrn.com/abstract=5097437, SSRN. - Duchin, R. and Sosyura, D. (2014), ‘Safer ratios, riskier portfolios: Banks’ response to government aid’, Journal of Financial Economics113(1), 1–28. - Schwert, M. (2018), ‘Bank capital and lending relationships’, The Journal of Finance73(2), 787–830. - Silva, A. F. (2019), ‘Strategic liquidity mismatch and financial sector stability’, The Review of Financial Studies32(12), 4696–4733. - 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. - Bai, J., Campello, M. and Muthukrishnan, P. (2025), Loan-funded Loans: Asset-like Liabilities inside Bank Holding Companies, Available at https://ssrn.com/abstract=5260219, SSRN. - Ashcraft, A. B. (2008), ‘Are Bank Holding Companies a Source of Strength to Their Banking Subsidiaries?’, Journal of Money, Credit and Banking40(2-3), 273–294. - Avraham, D., Selvaggi, P. and Vickery, J. I. (2012), ‘A structural view of U.S. bank holding companies’, Federal Reserve Bank of New York Economic Policy Review18(2), 65–81. - Houston, J., James, C. and Marcus, D. (1997), ‘Capital market frictions and the role of internal capital markets in banking’, Journal of Financial Economics46(2), 135–164. ### Cross-border banking, regulatory arbitrage, and international spillovers - Claessens, S. and Van Horen, N. (2015), ‘The Impact of the Global Financial Crisis on Banking Globalization’, IMF Economic Review63(4), 868–918. - Cetorelli, N. and Goldberg, L. S. (2011), ‘Global Banks and International Shock Transmission: Evidence from the Crisis’, IMF Economic Review59(1), 41–76. - Cetorelli, N. and Goldberg, L. S. (2012), ‘Liquidity management of U.S. global banks: Internal capital markets in the great recession’, Journal of International Economics88(2), 299–311. - Demirgüç-Kunt, A., Horváth, B. L. and Huizinga, H. (2023), ‘Regulatory arbitrage and loan location decisions by multinational banks’, Journal of International Economics145, 103840. - Franch, F., Nocciola, L. and Zochowski, D. (2021), ‘Cross-border effects of prudential regulation: Evidence from the euro area’, Journal of Financial Stability53, 100820. - De Haas, R. and Van Horen, N. (2013), ‘Running for the Exit? International Bank Lending During a Financial Crisis’, The Review of Financial Studies26(1), 244–285. - De Haas, R. and van Lelyveld, I. (2010), ‘Internal capital markets and lending by multinational bank subsidiaries’, Journal of Financial Intermediation19(1), 1–25. - Houston, J. F., Lin, C. and Ma, Y. (2012), ‘Regulatory Arbitrage and International Bank Flows’, Journal of Finance67(5), 1845–1895. - Karolyi, G. A. and Taboada, A. G. (2015), ‘Regulatory arbitrage and cross-border bank acquisitions’, The Journal of Finance70(6), 2395–2450. - Ongena, S. R. G., Popov, A. A. and Udell, G. F. (2013), ‘‘When the Cat’s Away the Mice Will Play’: Does Regulation at Home Affect Bank Risk-Taking Abroad?’, Journal of Financial Economics108(3), 727–750. - Frame, W. S., Mihov, A. and Sanz, L. (2020), ‘Foreign investment, regulatory arbitrage, and the risk of us banking organizations’, Journal of Financial and Quantitative Analysis55(3), 955–988. - Demirgüç-Kunt, A., Horváth, B. L. and Huizinga, H. (2023), ‘Regulatory arbitrage and loan location decisions by multinational banks’, Journal of International Economics145, 103840. ### Corporate sector, private credit, and firm-level effects - Albuquerque, B. (2024), ‘Corporate debt booms, financial constraints and the investment nexus’, Journal of Applied Econometrics39(5), 766–789. - Albuquerque, B. and Iyer, R. (2024), ‘The Rise of the Walking Dead: Zombie Firms Around the World’, Journal of International Economics152, 104019. - Albuquerque, B. and Mao, C. (forthcoming), ‘The Zombie Lending Channel of Monetary Policy’, Management Science. - Albuquerque, B., Cerutti, E., Chen, N. and Firat, M. (2025), From banks to nonbanks: macroprudential and monetary policy effects on corporate lending, IMF Working Papers 25/96, International Monetary Fund. - 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 Intermediation34, 17–31. - Sufi, A. (2007), ‘Information Asymmetry and Financing Arrangements: Evidence from Syndicated Loans’, The Journal of Finance62(2), 629–668. - Ivashina, V. (2009), ‘Asymmetric information effects on loan spreads’, Journal of Financial Economics92(2), 300–319. - Gao, (references to corporate-sector-related work are included across multiple entries above). ### Methodology, data, and working papers - 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. and Zhou, H. (2018), Cross-border Banking and the Circumvention of Macroprudential and Capital Control Measures, IMF Working Papers 18/217, International Monetary Fund. - 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 Intermediation40(C). - Duprey, T. and Tuzcuoglu, K. (2025), ‘High-frequency effects of macroprudential policy announcements’, Economics Letters250, 112264. - Bai, J., Campello, M. and Muthukrishnan, P. (2025), Loan-funded Loans: Asset-like Liabilities inside Bank Holding Companies, Available at https://ssrn.com/abstract=5260219, SSRN. - Chernenko, S., Ialenti, R. and Scharfstein, D. (2025), Bank Capital and the Growth of Private Credit, Available at https://ssrn.com/abstract=5097437, SSRN. - 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. - Bhardwaj, A. and Javadekar, A. (2025), How Does Bank Lending To Non-Banks Affect Credit Allocation And Systemic Risk, Available at https://ssrn.com/abstract=4946216, SSRN. - 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. *Banking on Nonbanks — Working Paper No. WP/2026/023* --- _Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026023-source-pdf.pdf_