## wpiea2025034-print-pdf

## Source details

**Canonical URL:** [wpiea2025034-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025034-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025034-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025034-print-pdf.pdf.json)

---

### Overview and motivation
- Euro area multinational banks hold large shares of banking sector assets in non-euro-area countries, e.g., nearly 80 percent in the Czech Republic and around 40 percent in Hungary (Figure 1).
- ECB’s Targeted Longer-Term Refinancing Operations (TLTRO) aimed to increase bank lending by offering access to central bank term funding at favorable conditions, conditional on banks’ lending to the real economy.
- Outstanding TLTRO volumes peaked at €2.2 trillion as of 2021 Q3 (see Appendix A).
- Sizable presence of euro-area owned subsidiaries creates a potential pathway for euro area monetary policy to generate spillover effects on banks and the wider domestic economy outside the euro area.

### Policy relevance and potential channels
- Spillovers from TLTRO could amplify or dampen domestic cyclical conditions in host countries, affecting the burden on domestic economic policies.
- Possible financial stability and structural effects in host-country financial systems from TLTRO spillovers:
  - Foreign subsidiaries accessing cheaper funding could undercut domestic banks, reducing domestic banks’ profitability and market share.
  - Pressure on domestic banks’ profitability could push them toward a riskier asset profile.
  - Stress among domestic banks could test weaker financial safety nets outside the euro area.
- Business cycle synchronization across European countries is variable; heterogeneity implies spillovers could be procyclical at certain times.

### Research question and empirical approach
- Objective: Investigate the impact of TLTRO on selected bank balance sheet indicators in non-euro-area countries, including funding costs, lending rates, profitability, and risk profiles.
- Two empirical approaches:
  - Difference-in-differences (DiD): foreign-owned subsidiaries of parent banks from the euro area = treatment group; domestic banks in host markets = control group.
  - OLS continuous-treatment approach: treating TLTRO funding as a continuous variable (TLTRO exposure).
- Key model specifications:
  - DiD: Yt,i = β0 + β1DT + β2Dafter + β3DT Dafter + ui,t, with β3 as the average treatment effect on the treated; Dafter = 1 for treatment period 2014–22.
  - Continuous-treatment: Yt,i = α0 + α1TLTROi,t + α2Xi,t + vi,t, with TLTROi,t measured either as total parent-country TLTRO take-up or asset-weighted parent TLTRO take-up; Xi,t includes host-country output gap, host-country short term interest rates, and parent-bank characteristics.
  - Destination heterogeneity: Yt,i = θ0 + θ1TLTROi,t + θ2Xi,t + θ3EM + θ4EM × TLTROi,t + ei,t (EM = dummy for emerging markets).
  - Source heterogeneity: Yt,i = η0 + η1TLTROi,t + η2Xi,t + η3Periphery + η4Periphery × TLTROi,t + εi,t (Periphery = dummy for periphery euro area parent countries).

### Data and sample
- Main dataset: Fitch Connect database of worldwide bank balance sheets, annual frequency; sample period 2006–2022.
- Final estimation dataset: unbalanced panel of 1167 banks located in 19 non-euro area European countries.
  - 1048 domestic banks; 119 subsidiaries of parent banks located in the euro area.
  - The 119 subsidiary banks have parent banks in 21 parent countries; five countries (Austria, Germany, France, Italy, and Netherlands) account for 73 subsidiaries.
- Macroeconomic controls: short-term interest rates and output gap from the IMF WEO database.
- Two measures of TLTRO exposure:
  - Total TLTRO take-up of the parent country at time t.
  - Total TLTRO take-up of the parent country weighted by the share of parent bank assets in total parent country assets (proxy for individual parent TLTRO exposure; assumes TLTRO take-up is proportional to parent bank size).
- Treatment period defined as 2014–22.

### Key variables and descriptive statistics
- Main dependent variables: deposit rates, lending rates, profitability (operating ROA), and risk (bank z-score).
- Risk (z-score) defined as (ROA + equity/assets) divided by the standard deviation of the ROA.
- Pre-TLTRO (pre) means:
  - Deposit rate (%) — Domestic Banks: 3.70; Foreign subsidiaries: 4.08; p-value: .00
  - Loan rate — Domestic Banks: 7.79; Foreign subsidiaries: 8.58; p-value: .00
  - ROA (%) — Domestic Banks: 0.70; Foreign subsidiaries: 0.56; p-value: .20
  - Z-score — Domestic Banks: 3.35; Foreign subsidiaries: 3.04; p-value: .00
  - Total assets (USD 000s) — Domestic Banks: 5978; Foreign subsidiaries: 6073; p-value: .92
  - Gross loans (USD 000s) — Domestic Banks: 3392; Foreign subsidiaries: 3832; p-value: .37
  - Loan-to-asset ratio (%) — Domestic Banks: 65.01; Foreign subsidiaries: 62.42; p-value: .00
  - Liquid assets to total assets (%) — Domestic Banks: 19.82; Foreign subsidiaries: 20.71; p-value: .17
  - Deposits-to-liabilities ratio (%) — Domestic Banks: 76.41; Foreign subsidiaries: 64.13; p-value: .00
- After TLTRO (2014 onwards) means:
  - Deposit rate (%) — Domestic Banks: 1.92; Foreign subsidiaries: 1.73; p-value: .03
  - Loan rate — Domestic Banks: 6.09; Foreign subsidiaries: 6.24; p-value: .37
  - ROA (%) — Domestic Banks: 0.72; Foreign subsidiaries: 0.95; p-value: .01
  - Z-score — Domestic Banks: 3.32; Foreign subsidiaries: 2.97; p-value: .00
  - Total assets (USD 000s) — Domestic Banks: 7212; Foreign subsidiaries: 7317; p-value: .92
  - Gross loans (USD 000s) — Domestic Banks: 3909; Foreign subsidiaries: 4445; p-value: .31
  - Loan-to-asset ratio (%) — Domestic Banks: 60.25; Foreign subsidiaries: 57.77; p-value: .00
  - Liquid assets to total assets (%) — Domestic Banks: 22.13; Foreign subsidiaries: 20.02; p-value: .00
  - Deposits-to-liabilities ratio (%) — Domestic Banks: 84.72; Foreign subsidiaries: 76.89; p-value: .00
- Average changes (post vs pre):
  - Deposit rates declined by 235 basis points for foreign subsidiaries and by 178 basis points for domestic banks.
  - Lending rates declined by 234 basis points for foreign subsidiaries and by 170 basis points for domestic banks.
  - Profitability (ROA) increased by nearly 40 basis points for foreign subsidiaries and remained nearly unchanged for domestic banks.

### Main empirical findings — Difference-in-Differences
- Exposure to TLTROs through euro area parents of foreign subsidiaries is associated with:
  - Decreased deposit and lending rates for foreign subsidiaries relative to domestic banks.
  - Higher profitability among foreign subsidiaries relative to domestic banks.
  - No significant impact on risk.
- Magnitudes (DiD estimates, text summary and Table 2):
  - TLTRO exposure of foreign subsidiaries associated with a 25bps decrease in deposit rates.
  - Associated with a 54 bps decrease in lending rates.
  - Associated with a 0.36 percent increase in ROA.
  - All three results significant at the 5 percent confidence level.
- Selected regression coefficients (Table 2):
  - tltroi = 1 coefficients: Lending rates: -2.096*** (0.092); Deposit rates: -1.498*** (0.043); Profitability: 0.024 (0.051); Risk: -0.044** (0.019)
  - 1.tltroi#1.treated coefficients: Lending rates: -0.544** (0.256); Deposit rates: -0.246** (0.120); Profitability: 0.364** (0.142); Risk: -0.027 (0.053)
  - Observations: 9,939 (lending, deposit, profitability), 9,931 (risk)
  - R-squared: 0.062 (lending), 0.129 (deposit), 0.001 (profitability), 0.016 (risk)
  - Standard errors in parentheses; significance: *** p<0.01, ** p<0.05, * p<0.1

### Continuous exposure (OLS/TWFE) results and intensity effects
- Deposit rates (Table 3):
  - $1 billion of parent-country TLTRO take-up associated with a reduction of 0.2 bps in deposit rates in the most stringent specification (Column (2) TWFE): TLTRO coefficient -0.002** (0.001).
  - OLS (Column (1)) TLTRO coefficient -0.005*** (0.001).
  - EM x TLTRO interaction negative and significant in some specifications: EM x TLTRO -0.006*** (0.002) in Column (3) and -0.011*** (0.001) in Column (7) for weighted measures.
  - Observations range: 8,840 to 8,942; R-squared up to 0.802 in TWFE specifications.
- Lending rates (Table 4):
  - $1 billion of parent-country TLTRO take-up associated with a reduction of 0.4 bps in lending rates in the most stringent specification (Column (2) TWFE): TLTRO coefficient -0.004*** (0.001).
  - OLS (Column (1)) TLTRO coefficient -0.007*** (0.001).
  - EM x TLTRO interaction negative and significant: EM x TLTRO -0.011*** (0.003) in Column (3).
  - Observations range: 8,840 to 8,942; R-squared up to 0.896 in TWFE specifications.
- Profitability (Table 5):
  - $1 billion of TLTRO exposure associated with higher ROAs by around 0.2 bps in some specifications: TWFE Column (2) TLTRO coefficient 0.002* (0.001).
  - OLS Column (1) TLTRO coefficient 0.003*** (0.001).
  - EM x TLTRO sometimes positive and significant: EM x TLTRO 0.002** (0.001) in weighted specifications.
  - Observations range: 8,840 to 8,942; R-squared around 0.460 in TWFE.
- Risk (Table 6):
  - No consistent significant effect of TLTRO on risk across specifications.
  - Some weighted specifications show negative coefficients but mixed significance; overall result: no significant impact on risk.
  - Observations range: 8,830 to 8,932; R-squared up to 0.606 in TWFE.

### Economic significance and robustness
- Larger total TLTRO take-up by euro area countries reached over $400 billion for Germany, France, and Italy at peak, and over $2 trillion in total at peak — implying economically significant spillovers when aggregated.
- Asset-weighted TLTRO exposure produces similar signs and larger magnitudes but can reduce statistical significance (TWFE coefficient can drop out of 5% significance), plausibly due to correlation between subsidiary fixed effects and parent characteristics used in weighting.
- For $1 billion of TLTRO take-up:
  - Deposit rates: reduction of 0.2 bps (TWFE most stringent).
  - Lending rates: reduction of 0.4 bps (TWFE most stringent).
  - Profitability: ROA increases about 0.2 bps in some specifications.
- Interpretation: TLTRO funds cheaper than deposits, enabling substitution of deposits with TLTRO funds and allowing lending rates to fall by more than deposit rates.

### Heterogeneity: destination and source effects
- Destination effects:
  - TLTRO exposure effects on lending and deposit rates stronger in emerging markets than in advanced economies (interaction terms significant), though differences muted for profitability and risk.
- Source effects:
  - Interaction for parent banks in euro area periphery countries not statistically significant for variables of interest; hypothesis of source effects in TLTRO spillovers is rejected.

### Conclusions and policy implications
- Empirical evidence using subsidiaries of euro-area banks operating outside the euro area indicates larger exposure to TLTROs is associated with:
  - Lower deposit rates.
  - Lower lending rates.
  - Higher profitability for foreign subsidiaries relative to domestic banks in host countries outside the euro area.
- Policy considerations:
  - Spillovers can loosen or tighten financial conditions in non-euro area countries depending on cyclical alignment; this affects calibration of domestic policies outside the eurozone.
  - Potential financial stability and structural effects: foreign-owned subsidiaries benefiting from TLTRO spillovers may become more competitive than domestic banks, potentially reshaping local financial landscapes; policymakers should monitor these effects.

*Source: IMF Working Paper — Introduction and Appendices of "TLTRO Spillovers Outside the Euro Area" (content unit: wpiea2025034-print-pdf).*

### Introduction ...........................................................................................................

### Introduction

### Overview and motivation
- Euro area multinational banks hold large shares of banking sector assets in non-euro-area countries, e.g., nearly 80 percent in the Czech Republic and around 40 percent in Hungary (Figure 1).
- The sizable presence of euro-area owned subsidiaries creates a potential pathway for euro area monetary policy—especially policies with banks as the intended transmission channel—to generate spillover effects on banks and the wider domestic economy outside the euro area.
- ECB’s Targeted Longer-Term Refinancing Operations (TLTRO) aimed to increase bank lending by offering access to central bank term funding at favorable conditions, conditional on banks’ lending to the real economy.
- Outstanding TLTRO volumes peaked at €2.2 trillion as of 2021 Q3 (see Appendix A).

### Policy relevance and potential channels
- Spillovers from TLTRO could amplify or dampen domestic cyclical conditions in host countries, affecting the burden on domestic economic policies.
- Possible financial stability and structural effects in host-country financial systems from TLTRO spillovers include:
  - Foreign subsidiaries accessing cheaper funding could undercut domestic banks, reducing domestic banks’ profitability and market share.
  - Pressure on domestic banks’ profitability could push them toward a riskier asset profile.
  - Stress among domestic banks could test weaker financial safety nets outside the euro area.
- Business cycle synchronization across European countries is variable; heterogeneity implies spillovers could be procyclical at certain times (Arčabić, Panovska, and Tica (2024)).

### Research question and empirical approach
- Objective: Investigate the impact of TLTRO on selected bank balance sheet indicators in non-euro-area countries, including funding costs, lending rates, profitability, and risk profiles.
- Two empirical approaches:
  - Difference-in-differences (DiD): foreign-owned subsidiaries of parent banks from the euro area = treatment group; domestic banks in host markets = control group.
  - OLS continuous-treatment approach: treating TLTRO funding as a continuous variable (TLTRO exposure).

### Main findings (summary)
- Foreign-owned subsidiaries of euro-area banks operating in non-euro-area countries experienced:
  - Lower funding costs (deposit rates) relative to domestic banks.
  - Lower lending rates relative to domestic banks.
  - Statistically significant improvements in profitability.
  - No significant effects on risk were observed.

### Relation to literature
- Complements literature on ECB unconventional monetary policies (UMPs) and their spillovers outside the euro area; prior studies find APP and other UMP announcements affected equity prices, bond yields, and exchange rates with notable spillovers to non-euro-area countries.
- Extends TLTRO literature that focuses on domestic euro-area impacts (credit supply, lending rates, and margins) by examining cross-border spillovers to subsidiaries in non-euro-area host countries.
- Links to multinational banking literature on internal capital markets and international transmission of liquidity shocks; internal capital markets are a hypothesized pathway for TLTRO funding reaching foreign subsidiaries.

### Identification issues and scope
- Two primary expected effects from a parent-level funding shock:
  - Direct effect: cheaper internal funds from parent to subsidiary → reduction in foreign subsidiaries’ deposit rates.
  - Indirect (general equilibrium) effect: lower domestic deposit demand → possible decline in deposit costs for domestic banks.
- For lending rates, full pass-through from funding costs to lending rates is a baseline scenario under competition; actual pass-through depends on market imperfections and general equilibrium dynamics.
- Profitability effects depend on relative changes in deposit and lending rates; no prior on sign or magnitude.
- Heterogeneity arises if:
  - No internal capital market exists between parent and subsidiary.
  - Banks keep deposit rates high to retain deposits anticipating TLTRO expiry.
  - Market imperfections prevent deposit-rate changes translating into lending-rate changes.
- Scope limitations:
  - The paper aims to establish whether spillovers exist and their magnitude; it does not provide direct evidence of internal capital markets within banking groups or interpret findings as evidence of market imperfections.
  - If domestic banks are “contaminated” via indirect effects, DiD estimates would be biased toward zero (underestimated treatment effects); but such indirect effects would imply larger aggregate spillovers from a policy perspective.

### Methodology (models estimated)
- Difference-in-differences model:
  - Yt,i = β0 + β1DT + β2Dafter + β3DT Dafter + ui,t
  - DT = 1 for treated banks (foreign subsidiaries of euro-area parents), 0 otherwise.
  - Dafter = 1 for treatment period 2014–22, 0 otherwise.
  - β3 is the average treatment effect on the treated (effect of TLTRO exposure).
- Continuous-treatment (TWFE and OLS) specifications:
  - Yt,i = α0 + α1TLTROi,t + α2Xi,t + vi,t
    - TLTROi,t = exposure of bank i at time t to parent-country TLTRO (two measures described in Data).
    - Xi,t = control variables including host-country output gap, host-country short term interest rates, and parent-bank characteristics (asset size, deposits to liabilities, loans to assets).
    - Use combinations of country and parent-country fixed effects; one specification includes bank and time fixed effects (Two-Way Fixed Effects).
- Destination (emerging vs. advanced economies) effects:
  - Yt,i = θ0 + θ1TLTROi,t + θ2Xi,t + θ3EM + θ4EM × TLTROi,t + ei,t
  - EM = dummy for emerging markets; θ4 captures differential TLTRO effect in emerging markets.
- Source (core vs. periphery parent countries) effects:
  - Yt,i = η0 + η1TLTROi,t + η2Xi,t + η3Periphery + η4Periphery × TLTROi,t + εi,t
  - Periphery = dummy for periphery euro area parent countries; η4 captures differential TLTRO effect from periphery parents.

### Data
- Main dataset: Fitch Connect database of worldwide bank balance sheets, at annual frequency.
- Data preparation follows principles in Thibaut and Le Mathias (2015) for European bank coverage.
- Two measures of TLTRO exposure:
  - Total TLTRO take-up of the parent country at time t.
  - Total TLTRO take-up of the parent country weighted by the share of parent bank assets in total parent country assets (proxy for individual parent TLTRO exposure; assumes TLTRO take-up is proportional to parent bank size).
- Treatment period defined as 2014–22.

*Source: IMF Working Paper — Introduction section of "TLTRO Spillovers Outside the Euro Area" (contents as provided).*

### Appendix B). The final estimation dataset consists of an unbalanced panel of 1167 banks

### wpiea2025034-print-pdf - Appendix B). The final estimation dataset consists of an unbalanced panel of 1167 banks

### Dataset and sample scope
- Unbalanced panel of 1167 banks over 2006–2022 located in 19 non-euro area European countries: Albania, Belarus, Bosnia and Herzegovina, Bulgaria, Czech Republic, Denmark, Hungary, Moldova, Montenegro, North Macedonia, Norway, Poland, Romania, Russian Federation, Serbia, Sweden, Türkiye, Ukraine, and the United Kingdom.
- Of the 1167 banks:
  - 1048 banks are domestic banks in the non-euro area country.
  - 119 are subsidiaries of parent banks located in the euro area.
- The 119 subsidiary banks have parent banks in 21 parent countries; five countries (Austria, Germany, France, Italy, and Netherlands) account for 73 subsidiaries.
- Parent-bank and parent-country information supplemented manually where the original Fitch dataset lacked parent details.
- Country-level TLTRO uptake data from Eurostat used because bank-wise TLTRO uptake in the euro area is not publicly available.
- Macroeconomic controls included: short-term interest rates and the output gap from the IMF WEO database.

### Key variables and measures
- Main dependent variables: deposit rates, lending rates, profitability (operating ROA), and risk (bank z-score).
- Risk (z-score) defined as (ROA + equity/assets) divided by the standard deviation of the ROA.
- Deposit rates, lending rates, and operating ROA taken directly from the Fitch database.
- Macroeconomic controls: short-term interest rate (S/T rate) and output gap (IMF WEO).

### Descriptive statistics (pre- and post-TLTRO means, Table 1)
- Before TLTRO (pre):
  - Deposit rate (%) — Domestic Banks: 3.70; Foreign subsidiaries: 4.08; p-value: .00
  - Loan rate — Domestic Banks: 7.79; Foreign subsidiaries: 8.58; p-value: .00
  - ROA (%) — Domestic Banks: 0.70; Foreign subsidiaries: 0.56; p-value: .20
  - Z-score — Domestic Banks: 3.35; Foreign subsidiaries: 3.04; p-value: .00
  - Total assets (USD 000s) — Domestic Banks: 5978; Foreign subsidiaries: 6073; p-value: .92
  - Gross loans (USD 000s) — Domestic Banks: 3392; Foreign subsidiaries: 3832; p-value: .37
  - Loan-to-asset ratio (%) — Domestic Banks: 65.01; Foreign subsidiaries: 62.42; p-value: .00
  - Liquid assets to total assets (%) — Domestic Banks: 19.82; Foreign subsidiaries: 20.71; p-value: .17
  - Deposits-to-liabilities ratio (%) — Domestic Banks: 76.41; Foreign subsidiaries: 64.13; p-value: .00
- After TLTRO (2014 onwards):
  - Deposit rate (%) — Domestic Banks: 1.92; Foreign subsidiaries: 1.73; p-value: .03
  - Loan rate — Domestic Banks: 6.09; Foreign subsidiaries: 6.24; p-value: .37
  - ROA (%) — Domestic Banks: 0.72; Foreign subsidiaries: 0.95; p-value: .01
  - Z-score — Domestic Banks: 3.32; Foreign subsidiaries: 2.97; p-value: .00
  - Total assets (USD 000s) — Domestic Banks: 7212; Foreign subsidiaries: 7317; p-value: .92
  - Gross loans (USD 000s) — Domestic Banks: 3909; Foreign subsidiaries: 4445; p-value: .31
  - Loan-to-asset ratio (%) — Domestic Banks: 60.25; Foreign subsidiaries: 57.77; p-value: .00
  - Liquid assets to total assets (%) — Domestic Banks: 22.13; Foreign subsidiaries: 20.02; p-value: .00
  - Deposits-to-liabilities ratio (%) — Domestic Banks: 84.72; Foreign subsidiaries: 76.89; p-value: .00
- Average changes (post vs pre, text summary):
  - Deposit rates declined by 235 basis points for foreign subsidiaries and by 178 basis points for domestic banks.
  - Lending rates declined by 234 basis points for foreign subsidiaries and by 170 basis points for domestic banks.
  - Profitability (ROA) increased by nearly 40 basis points for foreign subsidiaries and remained nearly unchanged for domestic banks.

### Main empirical findings — Difference-in-Differences (Table 2)
- Exposure to TLTROs through euro area parents of foreign subsidiaries is associated with:
  - Decreased deposit and lending rates for foreign subsidiaries relative to domestic banks.
  - Higher profitability among foreign subsidiaries relative to domestic banks.
  - No significant impact on risk.
- Magnitudes (text summary and Table 2 coefficients):
  - TLTRO exposure of foreign subsidiaries associated with a 25bps decrease in deposit rates.
  - Associated with a 54 bps decrease in lending rates.
  - Associated with a 0.36 percent increase in ROA.
  - All three results significant at the 5 percent confidence level.
- Table 2 regression highlights (columns correspond to Lending rates, Deposit rates, Profitability, Risk):
  - tltroi = 1 coefficients: Lending rates: -2.096*** (0.092); Deposit rates: -1.498*** (0.043); Profitability: 0.024 (0.051); Risk: -0.044** (0.019)
  - 1.tltroi#1.treated coefficients: Lending rates: -0.544** (0.256); Deposit rates: -0.246** (0.120); Profitability: 0.364** (0.142); Risk: -0.027 (0.053)
  - Observations: 9,939 (lending, deposit, profitability), 9,931 (risk)
  - R-squared: 0.062 (lending), 0.129 (deposit), 0.001 (profitability), 0.016 (risk)
  - Standard errors in parentheses; significance: *** p<0.01, ** p<0.05, * p<0.1

### Continuous exposure (OLS) results and intensity effects (Tables 3–6)
- Deposit rates (Table 3):
  - $1 billion of parent-country TLTRO take-up associated with a reduction of 0.2 bps in deposit rates in the most stringent specification (Column (2) TWFE): TLTRO coefficient -0.002** (0.001).
  - In OLS (Column (1)), TLTRO coefficient -0.005*** (0.001).
  - EM x TLTRO interaction negative and significant in some specifications: EM x TLTRO -0.006*** (0.002) in Column (3) and -0.011*** (0.001) in Column (7) for weighted measures.
  - Output gap, S/T rate, asset size, deposits to liabilities, loans to assets coefficients reported across columns with exact values in tables.
  - Observations range: 8,840 to 8,942; R-squared up to 0.802 in TWFE specifications.
- Lending rates (Table 4):
  - $1 billion of parent-country TLTRO take-up associated with a reduction of 0.4 bps in lending rates in the most stringent specification (Column (2) TWFE): TLTRO coefficient -0.004*** (0.001).
  - OLS (Column (1)) TLTRO coefficient -0.007*** (0.001).
  - EM x TLTRO interaction negative and significant: EM x TLTRO -0.011*** (0.003) in Column (3).
  - Observations range: 8,840 to 8,942; R-squared up to 0.896 in TWFE specifications.
- Profitability (Table 5):
  - $1 billion of TLTRO exposure associated with higher ROAs by around 0.2 bps in some specifications: TWFE Column (2) TLTRO coefficient 0.002* (0.001).
  - OLS Column (1) TLTRO coefficient 0.003*** (0.001).
  - EM x TLTRO sometimes positive and significant: EM x TLTRO 0.002** (0.001) in weighted specifications.
  - Observations range: 8,840 to 8,942; R-squared around 0.460 in TWFE.
- Risk (Table 6):
  - No consistent significant effect of TLTRO on risk across specifications.
  - Some weighted specifications show negative coefficients but mixed significance; overall text reports no significant impact on risk.
  - Observations range: 8,830 to 8,932; R-squared up to 0.606 in TWFE.

### Economic significance and additional robustness notes
- Larger total TLTRO take-up by euro area countries reached over $400 billion for Germany, France, and Italy at peak, and over $2 trillion in total at peak — implying economically significant spillovers when aggregated.
- Asset-weighted TLTRO exposure produces similar signs and larger magnitudes but can reduce statistical significance (TWFE coefficient can drop out of 5% significance), plausibly due to correlation between subsidiary fixed effects and parent characteristics used in weighting.
- For $1 billion of TLTRO take-up:
  - Deposit rates: reduction of 0.2 bps (TWFE most stringent).
  - Lending rates: reduction of 0.4 bps (TWFE most stringent).
  - Profitability: ROA increases about 0.2 bps in some specifications.
- Interpretation: TLTRO funds cheaper than deposits, enabling substitution of deposits with TLTRO funds and allowing lending rates to fall by more than deposit rates.

### Heterogeneity: destination and source effects
- Destination effects:
  - TLTRO exposure effects on lending and deposit rates stronger in emerging markets than in advanced economies (interaction terms significant), though differences muted for profitability and risk.
- Source effects:
  - Interaction for parent banks in euro area periphery countries not statistically significant for variables of interest; hypothesis of source effects in TLTRO spillovers is rejected.

### Conclusions and policy implications
- TLTROs are ECB term lending operations to increase lending within the euro area by offering favorable borrowing conditions.
- Using cross-sectional variation in exposure through subsidiaries of euro area banks operating outside the euro area, larger exposure to TLTROs is associated with:
  - Lower deposit rates.
  - Lower lending rates.
  - Higher profitability for foreign subsidiaries relative to domestic banks in host countries outside the euro area.
- Policy considerations:
  - Spillovers can loosen or tighten financial conditions in non-euro area countries depending on cyclical alignment; this affects calibration of domestic policies outside the eurozone.
  - Potential financial stability and structural effects: foreign-owned subsidiaries benefiting from TLTRO spillovers may become more competitive than domestic banks, potentially reshaping local financial landscapes; policymakers should monitor these effects.

*IMF WORKING PAPERS — TLTRO Spillovers Outside the Euro Area (content unit: wpiea2025034-print-pdf - Appendix B).*

### Appendix A.TLTRO Operations

### Appendix A.TLTRO Operations

### Operational summary
- Cross-border group formation was possible under the rules including if banks were in a parent-subsidiary relationship.
- The target area for the operations was loans within the euro area.
- The stock of eligible loans against which financing could be availed ranged from 7 percent to 30 percent as of certain cut-off dates.
- The terms of the financing varied between TLTROs.

### Contextual notes
- The Table below (in source) summarizes the main operational details of the TLTRO I, II, and III operations.
- Targeting and eligibility were specific to loans within the euro area and to stocks of eligible loans as of specified cut-off dates.

### Appendix B. Fitch Data

### Data preparation steps (deduplication and selection rules)
- 1. Confined the market sectors of firms in Fitch Connect to banks while excluding other institutions, e.g., central banks. Included market sectors: ’Banks,’ ’Credit Union,’ ’Other Banks,’ ’Retail & Consumer Banks,’ ’Universal Commercial Banks.’ ’Trading & Investment Banks,’ ’Wholesale Commercial Banks,’ ‘Development Banks.’
- 2. Excluded inflation-adjusted balance sheets.
- 3. Included only annual balance sheets and income statement. For most banks, the fiscal year ends in December. The year of each balance sheet was identified as follows: if the fiscal year ends in the second half of the calendar year, the year of the balance sheet in question is the same as the fiscal year; conversely, if the fiscal year ends in the first half of the calendar year, the year of a balance sheet is the fiscal year minus one.
- 4. When multiple balance sheets exist for one year, kept consolidated balance sheets when available; otherwise used unconsolidated balance sheets.
- 5. Dropped balance sheets with statement types of ‘Estimated,’ ‘Forecast,’ ‘Preliminary,’ ‘Partial’ or ‘Pro-forma’ when the balance sheets are annual.
- 6. When multiple balance sheets each year followed different accounting practices, selected IFRS, if available, or Local GAAP otherwise. If multiple balance sheets were neither IFRS nor Local GAAP, kept the balance sheet with the accounting system most frequently used by the same bank.
- 7. When multiple balance sheets had different fiscal year-end dates within a year (with year defined per step 3), kept the balance sheets which have the latest fiscal year end-dates within the year.
- 8. Some banks release multiple balance sheets with different ‘Audited/Qualified.’
- 9. ‘Flag’ within a year. Kept the balance sheets with the highest level.
- 10. ‘Audited/Qualified Flag’ within the year (the highest level corresponds to ‘Unqualified opinion’).
- 11. For some bank-year observations, multiple balance sheets had different ‘Fitch Nickname,’ which uniquely identify a bank’s statements with the same type of disclosure, including accounting system, consolidation, inflation adjusted currency, consolidation level etc. Picked the ‘Fitch Nickname’ that is most frequently used among the ‘Fitch Nicknames’ available in the latest observation of a bank. In years with multiple balance sheets for a bank, only kept the reserved ‘Fitch Nickname’ if available.
- 12. For some bank-year observations, when there were multiple balance sheets but no balance sheets with the reserved ‘Fitch Nickname,’ kept the ‘Fitch Nicknames’ that are most frequently used among all ‘Fitch Nicknames’ available in that year.

### Appendix C. Charts

### Figures referenced
- Figure 1. Euro Area-owned Subsidiaries’ Share in Banks’ Assets Outside Euro Area
- Figure 2. Outstanding TLTRO Volumes

*Source: TLTRO Spillovers Outside the Euro Area, Working Paper No. WP/2025/034*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025034-print-pdf.pdf_