## Macroprudential Regulation and Sector-Specific Default Risk

## Source details

**Canonical URL:** [Macroprudential Regulation and Sector-Specific Default Risk](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022141-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022141-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022141-print-pdf.pdf.json)

---

### Introduction — purpose, scope, and main contributions
- Purpose and scope:
  - Examines transmission of macroprudential regulation across financial and non-financial sectors via impact on short- and long-term probabilities of default (PDs) at sector level.
  - Combines an ECB macroprudential policy database with sector-specific short- and long-term PDs for financial and non-financial sectors (basic materials, consumer (non-)cyclical, industrial, and technology) in European countries over the period 2008–2017.
  - Uses two empirical approaches: a dynamic panel analysis with an intensity-adjusted Macroprudential Policy Index (MPI) and a panel event-study focused on CRR/CRD IV (phased in June 2013) and BRRD (introduced May 2014).
- Main contributions:
  - Provides evidence that tighter macroprudential regulation benefits non-financial sectors on average, not only the financial sector.
  - Disaggregates effects by specific macroprudential reforms to identify heterogeneous channels and timings of transmission.

### Key data and index construction
- Sector PDs and sample:
  - Daily PDs with default horizons of 6 months and 5 years for six corporate sectors and 28 European countries; aggregated to monthly medians and averaged within each month; sample period 2008 to 2017.
  - CRI methodology: forward intensity model of Duan et al. (2012) with extensions; includes economy/sector-wide and firm-specific variables as inputs.
- MaPPED and MPI:
  - MaPPED covers macroprudential policies for 28 European countries between 1995 and 2017 across a closed list of 53 instruments in 11 categories.
  - Intensity-adjusted MPI weighting scheme:
    - Introduction or phasing out: weight = 1.00
    - Level changes: weight = 0.25
    - Scope/definition changes: weight = 0.10
    - Events maintaining level/scope: weight = 0.05
    - Multiply weights by +1 for tightening and −1 for loosening; ambiguous events excluded.
- Summary statistics for MPI (2008–2017):
  - Mean MPI non-EA countries: 18.13
  - Mean MPI EA countries: 15.49
  - Standard deviation MPI non-EA countries: 7.92
  - Standard deviation MPI EA countries: 5.98
  - Sample composition: 18 EA countries and 10 non-EA countries.

### Main empirical findings — dynamic panel (Section 3.2)
- Model features and estimation:
  - Dynamic panel regressions estimated separately by sector; dependent variable PDc,t(h) for h ∈ {6M; 5Y}; country fixed effects; 3 lags of PD; lagged cumulative MPI; macrofinancial controls; OLS with one-way country fixed effects; Driscoll Kraay standard errors; monthly data 2008–2017.
  - Long-run effect of a one unit change in MPI given by δ/(1 − ∑ρl 3 l=1).
- Financial sector:
  - 5-year horizon:
    - Short-run: a unit increase in MPI decreases the 5-year default probability by 0.425 bps.
    - Long-run: the decrease is 4.20 bps.
    - Robust across EA and non-EA subsamples.
  - 6-month horizon:
    - No statistical evidence of a significant impact of MPI.
- Non-financial sectors (5-year horizon):
  - Short-run MPI effects range from −0.24 bps (consumer cyclical) to −0.39 bps (industrial).
  - Long-run effects range from −1.79 bps to −4.7 bps.
  - These results mainly driven by EA countries in subsample analysis.
- Non-financial sectors (6-month horizon):
  - Statistically significant MPI effects for 6-month PDs with magnitudes commensurate with financial-sector estimates (especially in EA countries).
- Controls and other predictors:
  - First two PD lags significant in almost all cases; third lag necessary for financial and basic materials sectors.
  - Industrial production (IP) growth decreases PDs in financial and consumer cyclical sectors.
  - Yield curve slope: significant and negative.
  - Stock returns: negative relation with short-term PDs; positive relation with long-term PDs.
  - Inflation (Inf): positive and significant across sectors and horizons.
  - Inflation volatility (InfVol): associated with a significant decrease in default risk.
- Synthesis:
  - Strengthening macroprudential regulation reduces medium- and long-term default risk across financial and non-financial sectors, with transmission to non-financial sectors mainly via enhanced financial-sector resilience and credit-channel spillovers.

### Panel event-study results — CRR/CRD IV (Basel III transposition) and BRRD
- Methodology:
  - Standard panel event-study with two-way fixed effects; event windows of 5 quarters before and after intervention; tests for pre-event trends; lead exclusion restrictions (δ−1 = 0 and δ−2 = 0) applied; diagnostics report generally no statistical evidence of pre-event trends for main specifications.
- CRR/CRD IV (event centered June 2013; window March 2012–September 2014):
  - 6-month PD increases observed around the reform:
    - consumer non-cyclical: 1.68 bps increase
    - financial: 3.08 bps increase
    - industrial: 3.18 bps increase
    - These 6-month increases above median levels range from 34% (consumer non-cyclical) to 62% (financial).
  - 5-year PD increases observed around the reform:
    - consumer non-cyclical: 5.53 bps increase
    - financial: 8.26 bps increase
    - industrial: 11.05 bps increase
    - These 5-year increases above median levels range from 4% (basic material) to 7% (financial).
  - Dynamic post-event patterns:
    - Financial sector: initial spike followed by progressive decline; example decrease of 7.14 bps for the 5-year DP two quarters after the event; long-run impact (5+ quarters) yields a decrease in 6-month DP of 1.88 bps and a decrease in 5-year DP of 1 bps. The 1.88 bps decrease for the 6-month DP corresponds to a 38% decrease relative to the median level.
    - Non-financial sectors: sustained long-run increases in default risk after the reform; long-run 5-year increases range from 4.19 bps (industrial) to 12.05 bps (basic materials), corresponding to 2.5% to 7.6% above median levels. Consumer non-cyclical and technology sectors show long-run 6-month DP increases of 1.18 bps and 1.27 bps respectively (24% above medians).
  - Interpretation:
    - Banks may react to higher capital requirements by shrinking assets or RWA rather than raising capital, reducing credit supply and raising non-financial-sector default risks in the medium to long run.
  - Subsample robustness:
    - EA subsample confirms main conclusions; non-EA subsample shows pre-event trends for some sectors and less robust inference.
- BRRD (event centered December 2014; window September 2013–March 2016):
  - No statistical evidence of a pre-event trend.
  - Short-run impact muted; substantial long-run reductions in both 6-month and 5-year PDs across sectors.
  - Quantified reductions in 6-month PDs:
    - consumer cyclical: 2.09 bps
    - financial: 3.77 bps
    - industrial: 4.78 bps
    - basic materials: 9.62 bps (noted as substantial and possibly influenced by external factors)
  - Quantified reductions in 5-year PDs:
    - consumer non-cyclical: 9.30 bps
    - financial: 20.44 bps
    - basic materials: 46.5
  - Decreases relative to median PDs:
    - For 6-month PDs: from 47% (technology) to 76% (financial).
    - For 5-year PDs: from 6.7% (technology) to 17.3% (financial).
  - Interpretation and transmission:
    - BRRD’s credible bail-in framework increases market discipline, reduces banks’ risk-taking, improves banks’ funding conditions for BRRD-sensitive instruments, and enhances banks’ ability to supply credit — producing spillovers that lower default risk in non-financial sectors.
  - Comparison with CRR/CRD IV:
    - Higher capital requirements improve long-run financial-sector resilience but can raise non-financial-sector default risk via credit-supply contraction.
    - BRRD-style resolution reforms can reduce default risk across both financial and non-financial sectors by improving market discipline and credit supply.

### Policy implications and synthesis
- Aggregate conclusion:
  - Macroprudential regulation is on average beneficial to non-financial sectors through reduced default risk; economic- and statistical-significance comparable to effects in the financial sector.
- Instrument design matters:
  - Reforms strengthening resolution frameworks and market discipline (BRRD) reduce default risk across financial and non-financial sectors.
  - Reforms raising capital targets (CRR/CRD IV) can improve financial-sector resilience but may have unintended adverse effects on non-financial-sector default risk due to credit supply adjustments.
- Implementation considerations:
  - Monitor banks’ balance-sheet adjustments and credit supply responses when tightening capital requirements.
  - Consider complementary measures to mitigate short- to medium-term real-economy costs from higher capital requirements.
  - Account for subsample heterogeneity: many effects are stronger or primarily driven by euro area (EA) countries.

*Source: wpiea2022141-print-pdf — IMF Working Paper "Macroprudential Regulation and Sector-Specific Default Risk".*

### 1. I ntroduction .......................................................................................................

### 1. Introduction

### Purpose and scope
- Examines transmission of macroprudential regulation across financial and non-financial sectors via impact on short- and long-term probabilities of default (PDs) at sector level.
- Combines an ECB macroprudential policy database with sector-specific short- and long-term PDs for financial and non-financial sectors (basic materials, consumer (non-)cyclical, industrial, and technology) in European countries over the period 2008–2017.
- Uses two empirical approaches:
  - A dynamic panel analysis with an intensity-adjusted macroprudential policy index (MPI).
  - A panel event-study focused on two European policy reforms: i) CRR/CRD IV (phased in June 2013) and ii) BRRD (introduced May 2014).

### Main contributions
- Provides evidence that tighter macroprudential regulation benefits non-financial sectors on average, not only the financial sector.
- Disaggregates effects by specific macroprudential reforms to identify heterogeneous channels and timings of transmission.

### Key empirical findings (MPI, dynamic panel)
- Financial sector:
  - A one unit increase in the MPI (a new activation of a macroprudential instrument) significantly reduces long-term default risk by 0.425 and 4.20 basis points (bps) at short and long horizons, respectively.
  - The long-horizon impact (4.20 bps) corresponds to a 3.5% reduction in long-term default risk compared to its median level.
- Non-financial sectors:
  - A one unit change in the MPI significantly decreases both short- and long-term default risks across non-financial sectors.
  - Long-term reductions at a long horizon range from 1.79 bps for the consumer cyclical sector to 4.7 bps for the industrial sector.
  - These long-term reductions correspond to a reduction in long-term default risk ranging from 1 to 3% relative to sector-specific median levels.
- Interpretation: Tighter macroprudential regulation yields economically and statistically meaningful reductions in sector-level default risk, comparable in magnitude between financial and non-financial sectors.

### Panel event-study results (CRR/CRD IV and BRRD)
- CRR/CRD IV (higher capital requirements; phased in June 2013):
  - Financial sector:
    - Initially increased both short- and long-term default risks.
    - Eventually led to a significant decrease in default probabilities at a horizon of 5 quarters after introduction and beyond.
    - Short-term default risk is 1.88 bps lower than before the reform — a 38% decrease compared to the median level.
  - Non-financial sectors:
    - Experienced continued increases in default risk, particularly in long-term PDs at horizon 5 quarters after reform:
      - Increases range from 4.19 bps for the industrial sector to 12.05 bps for the basic materials sector.
      - These correspond to 2.5 to 7.6% higher than corresponding median levels.
    - Marked long-run increases in short-term PDs for:
      - Consumer non-cyclical: 1.18 bps (24% increase above median).
      - Technology: 1.27 bps (24% increase above median).
  - Interpretation: Detrimental effects on non-financial sectors likely reflect banks reacting to higher capital requirements by shrinking assets (credit supply contraction) rather than raising capital, reducing credit access for non-financial borrowers.
- BRRD (strengthened resolution framework; May 2014):
  - Led to a substantial decrease in both short- and long-term PDs for financial and non-financial sectors at a horizon of 5 quarters after introduction and beyond.
  - Observed decreases:
    - Short-term PDs: from 2.09 to 4.78 bps.
    - Long-term PDs: from 9.3 to 20.44 bps.
  - Translations relative to median levels:
    - Short-term PD decreases of 47% (technology sector) to 76% (financial sector) below median levels.
    - Long-term PD decreases of 6.7 to 17.3% below median levels.
  - Interpretation: Transition toward a bail-in regime increases risk-monitoring incentives for shareholders and creditors, reduces banks’ risk-taking and improves banks’ ability to supply credit, thereby lowering default risk in non-financial sectors.

### Relation to literature and policy implications
- Extends literature on macroprudential regulation by showing beneficial effects on non-financial sector PDs as well as financial-sector resilience.
- Contributes to studies on higher capital requirements’ transitional costs and long-term benefits, documenting that reduced access to credit can have persistent adverse effects on non-financial sector default risk.
- Complements findings on bail-in regimes and resolution frameworks by documenting broader real-economy benefits of BRRD-style reforms.
- Highlights potential unintended consequences of macroprudential tightening (e.g., CRR/CRD IV) on credit supply and default risk in non-financial sectors, suggesting the importance of monitoring credit supply responses and considering complementary measures to mitigate short- to medium-term real-economy costs.

*Source: IMF Working Paper — "Macroprudential Regulation and Sector-Specific Default Risk" (Introduction section).*

### 2. Data

### 2. Data

### 2.1. Sector-specific probabilities of default (PDs) — data sources and construction
- Source of PDs:
  - Credit Research Initiative (CRI) at the National University of Singapore provides daily updated PDs for 34,000 actively-listed firms in 128 economies.
  - CRI methodology: forward intensity model of Duan et al. (2012) with extensions from Duan and Fulop (2013); accounts for correlations of individual forward intensities through common risk factors and applies Nelson-Siegel functional restrictions across default horizons.
  - Economy/sector-wide variables used in the CRI model include: stock index return, short-term interest rate, and average distance-to-default of the financial and non-financial sector.
  - Firm-specific variables include: volatility-adjusted leverage, liquidity, profitability, relative size, market misvaluation, and idiosyncratic volatility.
- PD horizons and coverage used in this analysis:
  - Daily PDs with default horizons of 6 months and 5 years.
  - Six corporate sectors: basic materials, consumer cyclical, consumer non-cyclical, financial, industrial, and technology.
  - 28 European countries (the 27 countries for the European Union (EU) and the United Kingdom).
- Aggregation and sample:
  - For each country-sector pair, daily observations on the median of individual firms’ PDs are retrieved.
  - Aggregation to monthly frequency by taking the average over observations within a given month.
  - Sample period used in the empirical analysis: 2008 to 2017.
  - Rationale for sample window: balances maximizing country coverage for each sector and including the post Global Financial Crisis period; end-2017 conditioned by restrictions on macroprudential policy decision data availability (see Section 2.2).

### 2.1. Key empirical observations on sector PDs (from the sample 2008–2017)
- Long-term (5-year) default risk:
  - The financial sector’s 5-year default risk is materially lower than default risk in nonfinancial sectors.
  - The difference between the 5-year PDs of the consumer non-cyclical and financial sectors ranges from 15 to 30 bps, depending on the subsample considered and whether the mean or median is used.
- Short-term (6-month) default risk:
  - For the 6-month horizon, the financial sector’s default risk is comparable to the consumer non-cyclical and technology sectors.
  - Short-term mean default risk and variability (standard deviation) are higher in the subsample of euro area (EA) countries, consistent with the sample including the GFC and the euro area sovereign debt crisis which exerted pressure on the financial sector’s resilience.

### 2.2. The Macroprudential Policies Evaluation Database (MaPPED) — coverage and instrument classification
- MaPPED coverage:
  - Dataset on macroprudential policies in the banking sectors for 28 European countries between 1995 and 2017.
  - Provides lifecycle information for policy instruments: activation dates, changes in scope or level, and deactivation dates when applicable.
- Operational definition of a macroprudential instrument (Budnik and Kleibl (2018)) — an instrument satisfies at least one of:
  1. identified as macroprudential by the competent authority or in relevant legislation;
  2. implementation is motivated by macroprudential objectives (e.g., preserving financial stability);
  3. prudential tool is comparable in design or transmission channels to subsequently introduced macroprudential instruments and likely broadly impacts the banking sector.
- Instrument universe:
  - Closed list of 53 macroprudential policy instruments classified in 11 categories.
  - The 11 categories are: minimum capital requirements, capital buffers, risk weights, leverage ratios, loan-loss provisioning, lending standards restrictions, limits on credit growth and volumes, limits on large exposures and concentration, liquidity requirement and limits to currency mismatches, and other measures which contains mainly crisis-related measures and resolution tools.

### 2.2. Approaches to summarizing macroprudential stance and limitations
- Common approaches in the literature:
  - Activation/d eactivation dummy index: sums dummy variables indicating whether each instrument in a category is active at a given time.
  - Lifecycle-calibrated index with asymmetric treatment: accounts for recalibrations and changes in scope over an instrument’s life-cycle and assigns +/− signs to tightening/easing events to capture direction; can be aggregated into a Macroprudential Policy Index (MPI) by taking the difference between cumulative tightening and easing.
- Limitations of lifecycle-calibrated MPI:
  - Captures direction but provides limited information about intensity of variations.
  - Equal treatment of activations versus level/scope changes can produce large MPI movements even when underlying instruments change only marginally (e.g., frequent small recalibrations).

### 2.2. Intensity-adjusted Macroprudential Policy Index (MPI) — construction used in this analysis
- Weighting scheme for lifecycle events:
  - Introduction or phasing out of an instrument (first activation or deactivation): weight = 1.00.
  - Events affecting the level of the policy instrument: weight = 0.25.
  - Events affecting the definition/scope of the instrument: weight = 0.10.
  - Events resulting in maintaining the level and scope of a tool: weight = 0.05.
- Directional adjustment:
  - Multiply weights by +1 for policy tightening events and −1 for policy loosening events to reflect contribution to the overall macroprudential policy stance.
  - Events with ambiguous impact on the policy stance are not considered.

### 2.2. Summary statistics for the intensity-adjusted MPI (period 2008–2017)
- MPI prevalence and variability by subsample:
  - Mean MPI in non-euro area (non-EA) countries: 18.13.
  - Mean MPI in euro area (EA) countries: 15.49.
  - Standard deviation of MPI in non-EA countries: 7.92.
  - Standard deviation of MPI in EA countries: 5.98.
- Sample composition:
  - The sample includes 18 EA countries and 10 non-EA countries; results for the full panel may overweight EA countries due to their larger share.
- Empirical strategy implication:
  - Analysis will be carried out on the full sample and separately on the EA and non-EA subsamples to identify potential differences in impact and transmission of macroprudential regulation.

*Source: IMF Working Paper — 2. Data (section content from wpiea2022141-print-pdf).*

### Section 3.2 sum m arizes the results.

### Section 3.2 — Empirical results (Macroprudential Regulation and Sector-Specific Default Risk)

### Model and estimation details
- Dynamic panel regression estimated separately for each sector; dependent variable is sector-specific default probability 푃푃푃푃
푐푐,푡푡
(ℎ) at horizons h ∈ {6M; 5Y} with country-specific fixed effect η
푐푐
.
- Specification features:
  - Lags of the dependent variable (3 lags used).
  - Lagged macroprudential policy index (MPI) as cumulative measure of macroprudential policy stance.
  - Macrofinancial and financial controls: year-on-year change in industrial production excluding construction (IP), year-on-year change in HICP (Inf), trailing one-year inflation volatility (InfVol), one-year trailing stock market return (StockRet), stock market volatility (StockVol), Δ ShortRate, slope of government bond yield curve.
- Estimation approach and sample:
  - Estimation carried out by OLS with one-way country fixed effect dummies.
  - Estimation period is 2008-2017 on monthly data (10 years of monthly observations).
  - All regressors are considered as lagged values.
  - Standard errors: Driscoll Kraay standard errors (robust to general forms of cross-sectional and temporal dependence; clustered at the country level).
- Rationale for lagged MPI and cumulative measure:
  - Captures announcement impact and progressive transmission due to implementation lags and transitional regimes.
  - Long-run effect of a one unit change in MPI given by 훿훿/(1− ∑ 휌휌푙푙
3
푙푙=1
).
- Endogeneity and identification considerations:
  - Monthly frequency and lagging of MPI and controls mitigate contemporaneous reverse causality concerns.
  - Inclusion of macrofinancial controls further addresses policy reaction to evolving risks.
  - Nickell (1981) bias considered not material given 10-year monthly sample.

### Main empirical findings: impact of macroprudential policy (MPI)
- Financial sector (5-year horizon):
  - A unit increase in the MPI decreases the 5-year default probability by 0.425 bps in the short-run.
  - The long-run decrease in the 5-year default probability is 4.20 bps.
  - Result is robust across EA and Non-EA subsamples.
- Financial sector (6-month horizon):
  - No statistical evidence of a significant impact of MPI on the 6-month default probability.
  - Interpretation: macroprudential policies tend to address longer-term aspects of default risk.
- Non-financial sectors (5-year horizon):
  - A unit change in MPI short-run effects range from −0.24 bps (consumer cyclical) to −0.39 bps (industrial).
  - Corresponding long-run effects range from −1.79 bps to −4.7 bps.
  - Subsample analysis indicates these results are mainly driven by euro area (EA) countries.
  - Basic materials sector shows an even stronger 5-year impact in the full sample, but that estimate is driven by non-EA countries and is not statistically significant in the subsample.
- Non-financial sectors (6-month horizon):
  - Statistically significant MPI effects for 6-month DPs with magnitudes commensurate with financial sector estimates (especially driven by EA countries).
- Overall interpretation:
  - Tighter macroprudential regulation reduces medium- and long-term default risk in financial and non-financial sectors.
  - Transmission to non-financial sectors occurs mainly indirectly via the credit capacity and resilience of the financial sector.

### Findings on control variables and other predictors
- Persistence in default probabilities:
  - First two PD lags are significant in almost all cases; a third lag is necessary for financial and basic materials sectors.
- Industrial production (IP):
  - Increased economic activity (IP) significantly decreases short- and long-term default risk in the financial and consumer cyclical sectors.
  - For the industrial sector, yearly growth in IP is not significant despite negative sign on coefficient (information may be embedded in lagged PD dynamics).
- Yield curve slope:
  - Coefficient estimates are significant and negative, consistent with a steepening yield curve predicting stronger future real economic activity and lower default risk.
- Stock market effects:
  - Short-term default risk: negative relationship between stock market performance and default risk.
  - Long-term default risk: positive relationship between stock returns and default risk (consistent with Duffie et al. (2007), Chava and Purnanandam (2010), Friewald et al. (2014)).
- Short-term interest rates (Δ ShortRate):
  - In the non-EA subsample for consumer non-cyclical, industrial, and technology sectors, changes in short-term risk-free interest rates show a negative relationship with default risk (consistent with literature on monetary loosening and bank risk-taking).
- Inflation (Inf) and inflation volatility (InfVol):
  - Inflation coefficient estimates are positive and significant across sectors and horizons — authors conjecture Inf proxies for trend component given its downward trend in the sample period.
  - Higher inflation volatility (InfVol) is associated with a significant decrease in default risk.
- Subsample heterogeneity:
  - Many of the MPI and control variable effects are stronger or primarily driven by euro area (EA) countries in subsample analysis.

### Synthesis and interpretation
- The evidence indicates that strengthening macroprudential regulation:
  - Decreases 5-year default probabilities in the financial sector substantially (short-run: 0.425 bps; long-run: 4.20 bps).
  - Reduces both short- and long-term default risks in non-financial sectors, with short-run MPI effects between −0.24 bps and −0.39 bps and long-run effects between −1.79 bps and −4.7 bps.
- Transmission mechanism:
  - Primary channel appears to be via enhanced resilience of the financial sector, which lowers credit-related vulnerabilities in the real economy and yields cross-sectoral spillovers.
- Contribution to literature:
  - Confirms and complements Meuleman and Vander Vennet (2020) by demonstrating aggregate-level reductions in bank systemic/default risk, and extends the analysis by documenting beneficial spillovers to non-financial sectors.
  - Consistent with literature linking financial sector systemic risk buildup to downside risks for the real economy and with evidence on cross-sectoral contagion in default clustering.

*Source: wpiea2022141-print-pdf - Section 3.2 summarizes the results.*

### 4. Panel event-study on the transmission of

### 4. Panel event-study on the transmission of macroprudential policy reforms

### 4.1. The methodology
- Empirical design: standard panel event-study regression with two-way fixed effects (TWFE) using sector-specific probability of default at horizons of 6 months and 5 years, 푃퐷퐷푐,푡(ℎ), as the outcome variable.
- Treatment indicator: 푑푐,푡푗 for an event at country-specific time êc located within an event window indexed by j ∈ [푗,푗], with j < 0 before the event and j > 0 after the event.
- Controls and fixed effects:
  - 푋𝐶,푡−1: vector of lagged macro-financial controls (introduced in Section 3).
  - 휇𝑐: country fixed effect.
  - 휃𝑡: year fixed effect.
- Identification and restrictions:
  - Follow Freyaldenhoven et al. (2019) and define 푑푐,푡푗 accordingly.
  - To avoid under-identification and perfect multicollinearity issues, exclude the first lead of treatment indicators (푑𝑐,𝑡−1), equivalent to imposing 훿−1 = 0; coefficients 훿𝑗 for j ≥ 0 interpreted relative to one period prior to the intervention.
  - Because all countries are treated simultaneously in this application, additionally exclude the second lead of treatment indicators (푑𝑐,𝑡−2), equivalent to imposing 훿−2 = 0.
  - Dynamic effects are implicitly restricted to be homogeneous across countries; 훿𝑗 estimates are average pre/post effects across countries.
- Endogeneity and pre-event trends:
  - Diagnostic approach: test hypothesis of no pre-event trend as 훿𝑗 = 훿𝑗+1 = . . . = 훿−3 = 0; complement with post-event trend test 훿0 = 훿1 = . . . = 훿𝑗 = 0.
  - Freyaldenhoven et al. (2019) 2SLS correction is not used because “no statistical evidence supporting the presence of pre-event trends in the sector-specific probabilities of default is found.”19
- Event window choice for policy analyses: 5 quarters before and after the chosen reference date (total 10 quarters) to capture pre-event trends and progressive transmission of announced changes.

### 4.2. Impact of higher capital requirements in the banking system (CRR/CRD IV; Basel III transposition)
- Policy event and event window:
  - Basel III standards implemented in European law in June 2013 via CRR/CRD IV (Capital Requirements Regulation 575/2013 and Directive 2013/36/EU).
  - Event window chosen: March 2012 to September 2014 (10 quarters centered on June 2013).
  - Implementation timing: 11 countries announced that Basel III minimum capital requirements would be effective from January 2014; remaining countries announced effective date January 2015. Twelve countries announced a fully phased-in CCoB (2.5% of RWA) during the event window; 5 of them also activated at least one other capital buffer.22
- Summary diagnostics:
  - Generally no statistical evidence supporting pre-event trends (mitigates endogeneity concerns).23
  - Reject null hypothesis of no post-event trend on the 6-month DPs in all cases (statistically significant short-term impact).
  - For 5-year DPs, reject null of no post-event trend only for consumer cyclical and technology sectors.
- Estimated impacts (around introduction of CRR/CRD IV):
  - 6-month DPs: increases observed
    - consumer non-cyclical: 1.68 bps increase
    - financial: 3.08 bps increase
    - industrial: 3.18 bps increase
  - 5-year DPs: increases observed
    - consumer non-cyclical: 5.53 bps increase
    - financial: 8.26 bps increase
    - industrial: 11.05 bps increase
- Economic significance relative to median PD levels (comparison with Table 1):
  - For 6-month horizon: increases above median levels range from 34% (consumer non-cyclical) to 62% (financial).
  - For 5-year horizon: increases above median levels range from 4% (basic material) to 7% (financial).
- Dynamic post-event patterns:
  - Financial sector:
    - Short-term spike in default risk followed by progressive decline in post-event period.
    - Example: decrease of 7.14 bps for the 5-year DP two quarters after the event.
    - Long-run impact (estimate of 5+ in equation (2)): decrease in default risk of 1.88 bps for the 6-month DP and 1 bps for the 5-year DP.
    - The 1.88 bps decrease for the 6-month DP corresponds to a 38% decrease relative to the median level.
  - Non-financial sectors:
    - Sustained increase in default risk following the introduction of the policy package.
    - Long-run impact on 5-year DPs ranges from 4.19 bps (industrial sector) to 12.05 bps (basic material sector).
    - These long-run 5-year increases correspond to increases above median levels ranging from 2.5% to 7.6%.
    - Consumer non-cyclical and technology sectors also show long-run 6-month DP increases of 1.18 bps and 1.27 bps respectively, corresponding to a 24% increase compared to their median levels.
  - Consumer cyclical sector: noted exception—not affected by sustained non-financial trend.24
- Interpretation and mechanisms:
  - Financial sector:
    - Initial rise in default risk consistent with short-term concerns about banks’ ability to meet higher capital requirements, particularly for less well-capitalized or less profitable banks.
    - Over time, CRR/CRD IV strengthens resilience, reducing default risk in the long run—consistent with Jordà et al. (2020) on better-capitalized banking systems better withstanding crises and maintaining credit supply.
  - Non-financial sectors:
    - Banks may shrink RWA or reallocate credit to meet higher CET1-to-RWA requirements (Gropp et al. (2018); Juelsrud and Wold (2020)).
    - Reduced credit supply can lead to lower investment, employment, asset growth, and sales for firms (Gropp et al. (2018); Fraisse et al. (2020)).
    - The observed pattern—mainly long-term increases in default risk—likely reflects structural changes in access to credit after phasing-in of the new regime.
    - Short-term increases in consumer non-cyclical and technology sectors suggest vulnerability to reduced bank credit supply.
    - Findings support literature that short-term transitional costs can partially offset long-term benefits of higher capital requirements (Mendicino et al. (2020)).
- Robustness by subsample (EA vs non-EA):
  - EA (euro area) subsample: confirm lack of statistical evidence of pre-event trend for most cases; main conclusions on statistically and economically significant impact confirmed in EA sub-sample.25
  - non-EA subsample: find statistical evidence of pre-event trends in short- and long-term default risks for basic material, consumer cyclical, and financial sectors.
    - Pre-event trend in financial sector for non-EA countries may reflect anticipatory adjustments over the long implementation cycle of Basel III (2010 to 2013–2015) rather than confounding factors.
  - Main conclusions hold in EA sample but not in non-EA sample.

### 4.3. Impact of a strengthening in the resolution framework for failing banks (BRRD)
- Policy event and event window:
  - Bank Recovery and Resolution Directive (BRRD, Directive 2014/59/EU) introduced in May 2014 to establish a European framework for resolution of failing banks and to shift from bailout to bail-in regimes.
  - Key BRRD elements: recovery and resolution planning; resolution authorities granted legal powers to intervene and use resolution tools; bail-in tool places losses first on shareholders and creditors before public restructuring.
  - Event center point chosen: December 2014 (latest date for transposition into national laws).
  - Event window: September 2013 to March 2016 (10 quarters centered on December 2014).
- Rationale for event window:
  - Window captures adoption into European law, the period around January 2015 when BRRD provisions not related to bail-in came into force, and January 2016 when Directive became fully binding for regulation of EU banks’ bail-ins.
  - Evidence (Cutura (2021)) that investors perceived enforcement of bail-in powers as credible: unsecured corporate bonds maturing after January 2016 (subject to bail-in) began trading with a positive spread over control bonds maturing before 2016.
- (Analysis of BRRD impacts on sector-specific default risks begins in the following pages beyond the supplied excerpt.)

*Source: IMF Working Paper — "Macroprudential Regulation and Sector-Specific Default Risk" (excerpt: Section 4).*

### introduction of the BRRD.

### introduction of the BRRD

### Event-study findings and timing
- No statistical evidence of a pre-event trend in default risk prior to the introduction of the BRRD.
- Short-run impact of the BRRD is muted, followed by a substantial long-run reduction in both short- and long-term default risks across all sectors.
- The event window captures a period of 10 quarters around the event; 68% confidence intervals are constructed based on standard errors clustered at the country level.

### Quantified impacts on probabilities of default (PDs)
- Reductions in 6-month PDs:
  - 2.09 bps for the consumer cyclical sector
  - 3.77 bps for the financial sector
  - 4.78 bps for the industrial sector
  - Basic materials sector: 9.62 bps (noted as substantial and possibly influenced by external factors)
- Reductions in 5-year PDs:
  - 9.30 bps for the consumer non-cyclical sector
  - 20.44 bps for the financial sector
  - Basic materials sector: 46.5 (reported alongside 9.62 bps for the 6-month and 5-year PDs)
- Decreases expressed relative to median PD levels:
  - For 6-month PDs: from 47% (technology sector) to 76% (financial sector)
  - For 5-year PDs: from 6.7% (technology sector) to 17.3% (financial sector)

### Transmission channels and interpretation
- Strengthened resolution framework (BRRD) credibly increased investors’ expectations about the likelihood of bail-ins, leading to increased market discipline and reduced banks’ risk taking.
- Improved balance-sheet quality in banks translates into reduced funding costs on BRRD-sensitive instruments and lower sector-level default risk.
- Spillovers to non-financial sectors:
  - Healthier bank balance sheets improve access to credit for non-financial firms, contributing to reduced default risk in those sectors.
  - This mechanism aligns with evidence that less risky banks can supply more credit due to better performance prospects and easier access to uninsured funding.

### Comparison with capital-requirement reforms (CRR/CRD IV)
- Higher capital requirements:
  - Improve the resilience of the financial sector in the long run.
  - Tend to raise long-term default risk in non-financial sectors, as banks react by reducing RWA rather than increasing capital, leading to reduced credit supply.
- Strengthened resolution framework (BRRD) vs. higher capital targets:
  - BRRD-style reforms that incentivize shareholders and creditors to monitor banks’ risk-taking may be more effective in reducing default risk in the financial sector once impacts on non-financial sectors are considered.
  - The direction of the transmission to non-financial sector default risk depends on how the policy affects banks’ credit supply: increased market discipline can lower banks’ risk and increase credit supply, while higher capital requirements can reduce credit supply.

### Robustness and subsample analysis
- Results are confirmed in both EA and non-EA subsamples (see Figures 4 and 6 of Appendix A).
- Estimates for the non-EA subsample exhibit lower precision, likely due to a smaller cross-section of countries and higher heterogeneity.

### Policy implications
- Macroprudential regulation is on average beneficial to non-financial sectors through reduced default risk; economic- and statistical-significance is comparable to effects in the financial sector.
- Choice of instrument matters:
  - Reforms strengthening resolution frameworks and market discipline can reduce default risk across financial and non-financial sectors.
  - Reforms that raise capital targets may improve financial-sector resilience but can have unintended adverse effects on default risk in non-financial sectors via reduced credit supply.
- Effective policy design should consider cross-sectoral transmission channels, particularly banks’ balance-sheet adjustments and credit supply responses.

*IMF Working Paper — Macroprudential Regulation and Sector-Specific Default Risk (introduction of the BRRD).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022141-print-pdf.pdf_