## _wp13167

## Source details

**Canonical URL:** [_wp13167](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13167.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13167.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13167.pdf.json)

---

### I. Motivation — core purpose and contribution
- Main contribution: provide a unified framework for assessing net benefits of macroprudential policy, with guidance on assessing policy leakages.
- Concept: adopt notion of long-run cost and benefit (from BCBS (2010a) with differences). Costs arise from an increase in the cost of intermediation and its effect on long-run output. Benefit derived from increased resilience—reduction in the probability of a crisis and output losses in the event of a crisis.
- Broader perspective: targets intermediate indicators tied to externalities identified in De Nicolo, Favara, and Ratnovski (2012) (DFR). The paper deals mostly with the first externality (strategic complementarities) and partly with the second (fire sales).
- Dataset and methods: uses IMF (2011c) survey and new dataset compiled by Lim et al (forthcoming); cross-country panel evidence and a U.S. time-series model with asymmetric macro-financial sensitivities; analysis of leakages and case studies.
- Objective: set up components of a core semi-structural model to provide real-time estimates of medium-to-long-term trend in real economic activity and interactions with intermediate macrofinancial indicators.

### II. Analytical recipe: the ingredients
- Framework components (preserve variables and symbols):
  - Integrates probability of crisis (p), loss given crisis (l and l*), interactions between financial and real activity (α), and effects of macroprudential policies on p and l.
  - Early warning indicators inform when to act; signals are imperfect and carry NSR (noise-to-signal ratios).
- Thresholds and definitions (exact formulations preserved):
  - GFSR threshold: "3-5 percentage point annual change in the credit-GDP ratio".
  - GFSR table note: threshold of changes in credit/GDP ratio of 3 percentage points.
  - Dell’Ariccia et al (2012) Boom definition:
    - Credit gap = percentage deviation of credit-to-GDP from a backward looking, rolling, cubic trend estimated over the period t-10 and t-1.
    - A credit boom is identified when: the deviation from trend is higher than 1.5 times its standard deviation and the annual growth of the credit-to-GDP exceeds 10 percent, or the annual growth of the credit-to-GDP exceeds 20 percent.
  - Empirical relationship between thresholds:
    - If a Boom is identified, then the GFSR threshold is breached for over 70 percent of the cases.
    - If a GFSR threshold is identified, the Boom threshold is breached in about 30 percent of the cases.
  - Raw snippet preserved: "Boom379/(379+146)  379/(379+921) No Boom= 0.72=0.299214415".
- Multi-model approach:
  - Combine models/indicators of risks within the financial sector, asset price models, crisis prediction models, and amplification channel analyses (SysMo project referenced).

### III. Conceptual framework: costs and benefits (formal expressions and insights)
- Policy mapping:
  - Policy lowers p → p* and l → l*; policy reduces intermediation and lowers the output forecast at t+2 by factor α (output becomes (1−α)·Yt+2).
- Expected output expressions (exact structures preserved in text interpretation):
  - Expected output without policy (expression (1)): expected output = (1−p)·Yt+2 + p·(1−l)·Yt+2.
  - Expected output with policy (expression (2)): expected output with policy = (1−p*)·(1−α)·Yt+2 + p*·(1−l*)·(1−α)·Yt+2.
  - Adoption rule: adopt policy only if (2) > (1); rearranged criterion referenced as expression (3.1).
- Key analytic insights:
  - Net benefits negligible when p is very low.
  - Net benefits extremely sensitive to α (sensitivity of forward-looking output measure to intermediate targets).
  - Net benefits rise with policy effectiveness (larger reductions in p and l).
  - Two unseen costs omitted from simple setup: imperfections of early warning indicators (NSR) and policy leakages reducing effectiveness.

### IV. Empirical building blocks and measurements (selected quantitative results)
- Trend and forecasting:
  - Uses Beveridge-Nelson (BN) trend for forward-looking trend estimates.
- Real-time sensitivities (U.S. bivariate model of GDP and credit-to-GDP gap):
  - A 1 percent increase in the credit-to-GDP gap improves GDP forecasts by about 0.2 percent on a 4-6 quarter-ahead horizon if there is no banking sector distress.
  - The same 1 percent increase in the credit-to-GDP gap reduces the GDP forecast by about 1 percent on the same horizon during large banking distress.
  - Reported estimate for α: α = 0.002 for every percentage point of credit growth (text: “α(=0.002 for every percentage point of credit growth)”).
  - Interpretation: reducing credit growth by 1 percentage point would dip the medium-term GDP forecast by about 0.2 percent if no crisis, but would improve the forecast by 1 percentage point if a crisis would otherwise occur.
- Probability of crisis (panel logit, 2-years ahead using credit growth and house price growth):
  - Threshold: credit-to-GDP growth of 3 percentage points used to detect rapid credit growth.
  - Heatmap examples:
    - 5 percentage point credit growth and 20 percent real house price growth → p = 14 percent.
    - 6 percentage point credit growth and 25 percent real house price growth → p = 19 percent.
- Output loss in event of banking crisis (cross-country evidence):
  - Cross-country data for 10 countries and 12 crisis events show GDP on average falls about 8 percent per year from the long-term forecast for five years from the beginning of the crisis.
  - Alternative measures show on average these are equivalent to 7 percent annual drop in actual output from some measure of the trend.
  - Cross-section model of 67 crisis cases: a 1 percentage point higher change in the credit-to-GDP ratio prior to the crisis is associated with a higher average yearly cost of a financial crisis of 0.6 percent.
- Illustrative baseline computation (text example):
  - Sample median credit growth 2 years before a crisis = 3 percentage points of GDP.
  - Average cost of crisis = 7 percent below potential for five years.
  - Example: country with 6 percentage point credit growth → expected crisis cost = 7 + 0.6*(6−3) = 8.8 percent below potential on average for five years.

### V. Policy response — instruments, transmission, and empirical effectiveness
- Instruments and primary channels:
  - Risk weights, Countercyclical Capital Buffer (CCB), Dynamic provisions: direct impact on economic capital.
  - Reserve requirements (RR): direct impact on funding structure.
  - Loan-to-Value (LTV) and Debt-to-Income (DTI): direct impact on assets (eligibility/flow).
- Loan pricing framework (Box 2, preserved equation and variable definitions):
  - r_L*(1-t) >= E*r_e + (D*r_d + C + A - O)*(1-t)
  - Definitions preserved for r_L, t, E, r_e, D, r_d, C, A, O.
  - Short-term assumption: assume macroprudential policies have a full impact on lending rates in the short term.
- Qualitative transmission highlights:
  - Risk weights/CCB: raise cost of assets; can raise lending rates; off-setting investor confidence effects possible.
  - RR: restricts funding available to back new loans; raises lending rates; applied to deposits, foreign liabilities, wholesale funds.
  - LTV/DTI: tighten eligibility flow; reduce mortgage credit growth and house price growth; reduce expected and unexpected losses, can lower lending rates for eligible borrowers.
  - Dynamic provisioning: may increase cost of capital through provisions booked; expected-loss accounting could change effects.
- Empirical evidence (survey and panel, 2000Q1–2011Q4):
  - LTV limits, reserve requirements and risk weights slow credit growth and house price growth; provisioning often not significant due to few users.
  - DTI limits slow loan/deposit growth and share of foreign liabilities—targets related to “fire-sale” externality.
  - RR lowers loan-deposit ratio growth; risk weights and LTV have larger effects on foreign asset-liability mismatches.
  - Prolonged impacts (historical averages):
    - Higher capital requirements: lowered credit growth by about 1 percentage point on impact and by 5 percentage points cumulatively in two years.
    - Lowered house price growth by 1 percentage point in the short-term and by 6 percentage points in two years.
    - RRs have prolonged impact on house price growth and loan-deposit growth.
  - Quality of buffers (Basel III) matters: higher quality loss-absorbing buffers are costlier and have larger effects on intermediate targets.

### VI. Quantitative example mapping tightening → intermediate targets → crisis parameters (Table 3 preserved numbers)
- Baseline assumptions used in example:
  - Baseline: Credit-to-GDP change = 5pp; Real house price growth = 20% => p = 0.14; l = 0.092. Average loss given crisis = 0.08.
- Two-year changes from tightening (preserved numbers):
  - Two-year change in credit growth (percentage points):
    - RR: -2.45
    - Capital Risk Weights: -5.04
    - LTV limits: -2.18
    - DTI limits: -2.63
  - Two-year change in house price growth (percentage points):
    - RR: -5.36
    - Capital Risk Weights: -5.79
    - LTV: -3.70
    - DTI: -1.98
  - p* (post-policy probability of crisis):
    - RR: 0.03
    - Capital Risk Weights: 0.03
    - LTV: 0.034
    - DTI: 0.032
  - Loss given crisis, l*:
    - RR: 0.065
    - Capital Risk Weights: 0.050
    - LTV: 0.067
    - DTI: 0.064
  - Cost on output forecast, α:
    - RR: 0.0049
    - Capital Risk Weights: 0.0101
    - LTV: 0.0044
    - DTI: 0.0053
  - Additional row preserved (unclear label): 0.0058, 0.0013, 0.0064, 0.0057
- Interpretation example for capital risk weights:
  - Tightening reduces probability of crisis from 14 percent to 3 percent and loss given crisis from 9.2 percent to 5 percent in illustrative example.
  - Note: For the United States, one percentage point lower credit growth reduces the output forecast by 0.2 percent (Annex 3).

### VII. Leakages — channels, evidence, and policy responses (Annex 6)
- Leakage channels emphasized:
  - Crossborder arbitrage through direct crossborder lending.
  - Regulatory arbitrage via shadow banking (domestic nonbanks outside banking regulatory perimeter).
  - Domestic nonbank financial intermediation competing with regulated banks.
- Empirical crossborder leakage findings (preserved estimates):
  - A tightening of reserve requirements reduces credit growth by "0.4-0.5 of a percentage point" on impact but increases leakages through higher crossborder lending growth by "1.2 percentage points".
  - Over two years, credit growth decreases by "2.5 percentage points", and direct crossborder loan grows by "about 5 percentage points".
  - Provisioning policies attract direct crossborder flows but do not significantly impact credit aggregates.
- Evidence on foreign branches and nonbanks:
  - Example (U.K.): about one third of the reduction in loan supply after domestic capital tightening is offset by foreign branches in the U.K.
  - United States: nonbanks significant in consumer and mortgage markets; reliance on bank-only aggregates would have missed early signals—consumer credit-GDP ratio changes would have signaled by 2003.
- Policy implications to counter leakages (preserved case-based recommendations):
  - Financially open economies with foreign subsidiaries:
    - Capital tools and LTV could work well for "correlated risks"; RRs susceptible to crossborder leakages; DTI limits could be used alongside or instead of RRs for funding risks.
  - Systems with foreign bank branches:
    - LTV and DTI apply to activity and are useful; capital tools problematic for universal use because branches are home regulated; RRs could help if applicable to branch liabilities.
  - Systems with large nonbanks:
    - LTV and DTI tools coordinated across regulators recommended; capital tools and RRs difficult to impose.
- Standard prescriptions to plug evasions:
  - Regulate financial services, not sectors.
  - Strengthen macroprudential regulation of shadow banking.
  - Coordinate macroprudential tools across borders and apply reciprocity principles.
  - Monitor arbitrage opportunities regularly.
  - Address data gaps and expand credit aggregates to include all sources of credit.

### VIII. Measurement and modelling components (summary of core elements)
- Core Macrofinancial Model, α:
  - Country-specific time series model estimating sensitivities of output forecasts to intermediate targets (Annex 3, US example).
  - Main empirical result: 1 percent increase in credit-to-GDP ratio in normal times → GDP forecast improves by about 0.2 percent on 4–6 quarter horizon; in large distress → reduces GDP forecast by about 1 percent on same horizon.
- Probability of crisis, p:
  - Panel logit model 1970-2010 (Annex 5) relating credit growth and asset price growth to probability of a banking crisis.
- Loss given crisis, l and l*:
  - Cross-country model with 67 crisis cases (Annex 4) relating past systemic risk to output loss during a crisis.
- These measured components feed into evaluation of effects of macroprudential policies and net-benefit calculations per expressions in Section II.

### IX. Practical implementation notes, caveats, and takeaways
- Multi-step strategy reproduced (exact steps listed in source):
  - When to Act. Summarize the effectiveness of early warning indicators, and the role of credit aggregates in deciding when to act (Section II).
  - Concept. Set up a simple conceptual framework for understanding the net benefits of policy expressed in terms of parameters that can be estimated (Section II).
  - Policy Effectiveness. Evaluate the effectiveness of policy in lowering systemic risk indicators (credit growth, loan/deposit ratios, asset price growth, foreign liabilities growth) (Section III).
  - Benefits. Estimate the effect of change in systemic risk indicators on the probability of a crisis and depth of the output loss during a crisis from cross-country data (Section III).
  - Cost. Compute the cost of policy by estimating the effect of change in the intermediate target on the forecast for GDP with an empirical model with asymmetric macro-financial sensitivities (Section III).
  - Leakages. Discuss policy side-effects and leakages (Section IV).
  - Section V: conclude with takeaways and rules-of-thumb for policymakers.
- Key practical messages and numerical/qualitative conclusions:
  - Use two thresholds (IMF (2011b) and Dell’Ariccia et al (2012)) with policy action in between for judgment plus analytical support.
  - Policymakers need measures of intermediate targets inclusive of all sources of credit (banks and nonbanks).
  - Net benefits negligible when p is very low; p increases non-linearly when house prices and credit grow rapidly, especially when credit-to-GDP ratio grows by "3 percentage points of GDP or more".
  - For the United States, if policy measures are taken and there is no crisis, the output forecast drops by "0.2 percent for every 1 percentage point tightening of credit growth".
  - Most effective and prolonged instruments on credit growth and house price growth historically: reserve requirements, higher risk weights on capital, and LTV limits; provisioning policies have not shown significant effects where data exist.
  - Given evidence of leakages, tailor instrument choice to financial structure and coordinate across regulators and borders.

*Italic source: _wp13167 — Excerpted sections and annexes as provided (IMF working paper content).*

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

### _wp13167 - References .............................................................................................................

### I. MOTIVATION — core purpose and contribution
- Main contribution: provide a unified framework for assessing net benefits of macroprudential policy, with guidance on assessing policy leakages.
- Concept: adopt notion of long-run cost and benefit (from BCBS (2010a) with differences). Costs arise from an increase in the cost of intermediation and its effect on long-run output. Benefit derived from increased resilience—reduction in the probability of a crisis and output losses in the event of a crisis.
- Broader perspective: targets intermediate indicators tied to externalities identified in De Nicolo, Favara, and Ratnovski (2012) (DFR). The paper deals mostly with the first externality (strategic complementarities) and partly with the second (fire sales).
- Dataset and methods: uses IMF (2011c) survey and new dataset compiled by Lim et al (forthcoming); cross-country panel evidence and a U.S. time-series model with asymmetric macro-financial sensitivities; analysis of leakages and case studies.
- Objective: set up components of a core semi-structural model to provide real-time estimates of medium-to-long-term trend in real economic activity and interactions with intermediate macrofinancial indicators.

### Key context and framing points (preserve terminology and numeric precision)
- Macroprudential policies aim to limit system-wide (systemic) financial risk (IMF (2011a, 2013) viewpoint).
- Three externalities per DFR (2012): (1) strategic complementarities (excessive/correlated risk), (2) fire sales (decline in asset prices, balance-sheet deterioration), (3) contagion through networks. This paper focuses mostly on (1) and partly on (2).
- The analysis integrates: probability of crisis (p), loss given crisis (l and l*), interactions between financial and real activity (α), and effects of macroprudential policies on p and l.

### II. ANALYTICAL RECIPE: THE INGREDIENTS
#### A. Systemic Risk Indicators: When and Whether to Act
- Rationale: instruments target market failures/externalities; policies target intermediate indicators as manifestations of market failures (e.g., rapid credit growth as sign of strategic complementarities).
- Early warning indicators are imperfect:
  - Rapid credit growth and credit booms signal banking crises 2-3 years before the event (Laeven and Valencia, 2010; IMF, 2011b & 2012a; Lund-Jensen, 2012; Dell’Ariccia et al, 2012; BCBS, 2010c; Drehmann et al, 2012).
  - Out-of-sample analysis shows credit growth (and gap measures) produced very low (but increasing) probabilities of crisis before 2007 (IMF, 2011b).
- Policy trade-off: compare benefits of avoiding/reducing crises with current costs of tightening intermediation and the risk of over-regulation (false positives).

#### Thresholds and indicators (preserve figures and thresholds exactly)
- GFSR threshold: "3-5 percentage point annual change in the credit-GDP ratio" (text reference).
- Table note: GFSR refers to the threshold of changes in credit/GDP ratio of 3 percentage points.
- Dell’Ariccia et al (2012) Boom definition (exact formulation preserved):
  - Credit gap = percentage deviation of credit-to-GDP from a backward looking, rolling, cubic trend estimated over the period t-10 and t-1.
  - A credit boom is identified when: the deviation from trend is higher than 1.5 times its standard deviation and the annual growth of the credit-to-GDP exceeds 10 percent, or the annual growth of the credit-to-GDP exceeds 20 percent.
- Empirical relationship between thresholds:
  - If a Boom is identified, then the GFSR threshold is breached for over 70 percent of the cases.
  - If a GFSR threshold is identified, the Boom threshold is breached in about 30 percent of the cases.
  - Policy implication: express concern when GFSR threshold is breached and escalate action by the time the Boom threshold is breached to provide a meaningful range for policy implementation.
- Coverage of credit: policymakers should include credit from all lenders (not just commercial banks). Example: United States nonbanks significant in consumer and mortgage markets; commercial bank credit alone greatly underestimates crisis risks and signals—neither GFSR nor Boom crosses thresholds for the United States since 2000 (Annex 2). Changes in consumer credit-GDP ratio would have signaled by 2003.

#### Multi-model approach and indicators
- Policymakers typically combine:
  - Models/indicators of risks building within the financial sector,
  - Asset price models to detect sectoral overheating,
  - Crisis prediction models to track probability of crisis,
  - Amplification channel analyses (e.g., house price increases affecting collateral constraints and lending).
- Reference to SysMo project (Blancher et al (2013) forthcoming): combine different models/methods.

- Table metadata (as provided):
  - Note: The table shows the number of occurances across 183 countries from 1966 to 2010 when thresholds were breached. GFSR refers to the threshold of changes in credit/GDP ratio of 3 percentage points. "Boom" refers to the threshold in Dell'Ariccia et al (2012). See Annex 2.
  - Raw items appearing in snippet: "Boom379/(379+146)  379/(379+921) No Boom= 0.72=0.299214415" (preserve exact textual fragments as presented).

#### B. Conceptual Framework for Assessing Net Benefits of Policy
- Ultimate objective: bolster long-term trend in real economic activity against permanent blows from financial crisis.
- Expected output path depends on probability of crisis (p) and loss given crisis (Y falling from Y* to Yc).
- Timeline and decision logic (in text):
  - At time t a signal indicates rising systemic risk and higher probability of crisis within next two years (example).
  - If no crisis, long-term output at t+2 is Y*; with probability p there will be a crisis and output drops to Yc.
  - Expected level of Yt+2 lies between Y* and Yc; higher p moves expectation closer to Yc.
- Components to estimate for net benefits (to be part of core model):
  - Probability of crisis, p (and p* with policy),
  - Loss given crisis, l (and l* with policy),
  - Policy effectiveness in reducing p and l,
  - Cost of policy via effect on intermediation and forecast GDP (including mis-identification / type II error).

### Multi-step strategy (exact steps listed)
- Steps as presented in the paper:
  - When to Act. Summarize the effectiveness of early warning indicators, and the role of credit aggregates in deciding when to act (Section II).
  - Concept. Set up a simple conceptual framework for understanding the net benefits of policy expressed in terms of parameters that can be estimated (Section II).
  - Policy Effectiveness. Evaluate the effectiveness of policy in lowering systemic risk indicators (credit growth, loan/deposit ratios, asset price growth, foreign liabilities growth) (Section III).
  - Benefits. Estimate the effect of change in systemic risk indicators on the probability of a crisis and depth of the output loss during a crisis from cross-country data (Section III), by extending the analysis in IMF (2011b) to a panel-logit model.
  - Cost. Compute the cost of policy by estimating the effect of change in the intermediate target (due to policies) on the forecast for GDP. This will be done with an empirical model with asymmetric macro-financial sensitivities based on whether the banking system is in a ‘normal’ or ‘distressed’ state (Section III).
  - Leakages. While leakages from macroprudential policies are not directly included in the core analytical model, some new findings on policy side-effects will be discussed (Section IV).
- Section V: conclude with takeaways and basic rules-of-thumb to provide practical guidance for policymakers.

### Figures, tables, boxes and annexes referenced (as presented)
- Tables listed include: "1. Thresholds for Credit Growth", "2. Comparison of Policy Tools—Expected Effects on Balance Sheets and Prices", "3. Effects of Policy Tightening: An Example".
- Figures listed include: "1. Structure of the Analytical Framework", "2. Policy Time Line", "3. Cost and Benefit in terms of output level—Schematic Representation", "4. United States: Time-varying Sensitivities of Real GDP forecast to Past Credit Growth", "5. Probability of Crisis and Early Warning Indicators—A Heattmap", "6. Crisis Cost", "7. Deriving a No-Policy Baseline: Some Measurements", "8. Baseline Expected Output", "9. Impact of Tightening of Macroprudential Policy Instruments on Intermediate Targets", "10. Summary of Measurements".
- Boxes listed include: Box 1 "Nonbank Lenders in the U.S.: Possibilities for Underestimating Systemic Risk", Box 2 "Effects of Macroprudential Policies on Lending Rates: What to Expect", Box 3 "Leakages through Direct Crossborder Lending: The Croatian Experience", Box 4 "Leakages through Foreign Branches: Empirical Evidence from the UK".
- Annexes listed include: 1. Other Studies on Costs and Benefits, 2. Signals Based on Two Different Thresholds for Bank Credit Aggregates, 3. An Output-Credit Forecasting Model, 4. Estimating the effect of intermediate targets on the cost of a financial crises, 5. Predicting the Probability of a Banking Crisis, 6. Effect of Macroprudential Policy Instruments on Intermediate Targets.

*Source: _wp13167 - References.*

### 14.      Macroprudential policy that affects an intermediate target, like credit growth,

### _wp13167 - 14.      Macroprudential policy that affects an intermediate target, like credit growth,

### Conceptual framework: costs and benefits
- Macroprudential policy is modeled as affecting:
  - the probability of crisis p → p* (policy lowers p to p*), and
  - the loss given crisis l → l* (policy lowers l to l*).
- Policy can also reduce intermediation and thus lower the output forecast at t+2 by a factor α (output becomes 2(1) t Yα in the text).
- Expected output without policy (expression (1) in the text) is:
  - 22 2 (1)(1)(1) (1) tt t pl Yp Y pl Y     
  - (interpreting: expected output = (1−p)·Yt+2 + p·(1−l)·Yt+2)
- Expected output with policy (expression (2) in the text) is:
  - *** 22 ** 2 (2)(1)(1)(1)(1) (1)(1) tt t plYpY p lY         
  - (interpreting: expected output with policy = (1−p*)·(1−α)·Yt+2 + p*·(1−l*)·(1−α)·Yt+2)
- Policy should be adopted only if expected output with policy exceeds expected output without policy, i.e., (2) > (1). The paper gives the criterion as expression (3) and a rearranged inequality (3.1):
  - (3.1) 11 (3.1)0 11 p lpl pl pl        
  - (Text interpretation: benefits (LHS first term) must outweigh costs (term involving α).)

### Key insights from the analytic expression
- Net benefits are negligible when the probability of crisis p is very low, everything else constant.
- Net benefits are extremely sensitive to α, the sensitivity of the forward-looking output measure to intermediate targets.
- Net benefits rise with policy effectiveness: policies that lower both probability p and depth l of crises (i.e., lower p* and l*) yield significantly higher net benefits.
- Two “unseen” costs not in the simple setup:
  - Imperfections of early warning indicators (noisiness or noise-to-signal ratios, NSR).
  - Policy leakages that reduce policy effectiveness (narrow the differences between p and p* and between l and l*).

### Empirical building blocks required
- To operationalize the framework policymakers need:
  a. An early warning system that alerts policymakers 2-3 years ahead of a crisis.
  b. Sensitivity of a forward-looking measure of output, 2t Y, to intermediate targets, α.
  c. An estimate of the probability of crisis, p, and the loss in level of output in the event of a crisis, l.
  d. The effectiveness of policy instruments on dampening growth in intermediate targets, and the effect of a reduction in intermediate targets on p* and l*.

### Measurements and empirical estimates (selected results from the paper)
- Trend and forecasting:
  - The paper uses the Beveridge-Nelson (BN) concept of the trend for forward-looking trend estimates.
- Real-time sensitivities (U.S. bivariate model of GDP and credit-to-GDP gap):
  - A 1 percent increase in the credit-to-GDP gap improves GDP forecasts by about 0.2 percent on a 4-6 quarter-ahead horizon if there is no banking sector distress.
  - The same 1 percent increase in the credit-to-GDP gap reduces the GDP forecast by about 1 percent on the same horizon during large banking distress.
  - The paper reports this as an estimate for α: α = 0.002 for every percentage point of credit growth (text: “α(=0.002 for every percentage point of credit growth)”).
  - Interpretation for policy: reducing credit growth by 1 percentage point would dip the medium-term GDP forecast by about 0.2 percent if no crisis, but would improve the forecast by 1 percentage point if a crisis would otherwise occur.
- Probability of crisis (panel logit model, 2-years ahead using credit growth and house price growth):
  - Threshold: credit-to-GDP growth of 3 percentage points is used to detect rapid credit growth.
  - Heatmap examples:
    - 5 percentage point credit growth and 20 percent real house price growth → p = 14 percent.
    - 6 percentage point credit growth and 25 percent real house price growth → p = 19 percent.
- Output loss in event of banking crisis (cross-country evidence):
  - Cross-country data for 10 countries and 12 crisis events show GDP on average falls about 8 percent per year from the long-term forecast for five years from the beginning of the crisis (Figure 6).
  - Alternative measures show on average these are equivalent to 7 percent annual drop in actual output from some measure of the trend.
  - Cross-section model of 67 crisis cases: a 1 percentage point higher change in the credit-to-GDP ratio prior to the crisis is associated with a higher average yearly cost of a financial crisis of 0.6 percent.
- Illustrative baseline computation (text example):
  - Sample median credit growth 2 years before a crisis = 3 percentage points of GDP.
  - Average cost of crisis = 7 percent below potential for five years.
  - Example: a country with 6 percentage point credit growth → expected crisis cost = 7 + 0.6*(6−3) = 8.8 percent below potential on average for five years (0.6 is the marginal effect of credit growth on crisis cost).

### Practical implementation notes and limitations
- The paper cautions that the simple static framework must be extended for practical use:
  - Interactions between output, risk and policy are dynamic and policy criteria may be time-varying (e.g., discounted sum of output over a medium-to-long run horizon).
  - A binary crisis/no-crisis indicator at a fixed time is a simplification; a realistic model should use time-varying multivariate probability distributions.
- NSR (noise-to-signal ratios) and policy leakages should be incorporated where possible:
  - NSR quantifies false signals or missed crises; assigning NSRs is difficult due to wide ranges of thresholds and model/judgment heterogeneity.
  - Policy leakages reduce effectiveness and therefore reduce the differences between p and p* and between l and l*.
- Model choices:
  - The bivariate model combines local linear projection method and smooth-transition technique to capture how real-financial transmission changes between normal and distress states.
  - The systemic financial stress index used as a state proxy is defined as the fraction of big banks with negative equity returns below the 5th percentile of the joint distribution, plus cumulative negative excess returns over the following two weeks to qualify as a stress episode.

### Steps to derive a no-policy baseline (summary)
- Estimate a real-time forward-looking forecast for output at t+2 (the “trend”).
- Estimate the probability of crisis p using early warning indicators (e.g., panel logit with credit growth and house price growth).
- Estimate loss given crisis l from cross-country and country-specific evidence.
- Combine p and l in expression (1) to compute the baseline expected output without policy, noting that baseline expected output is conditional on credit growth and shifts down when credit/GDP growth is more rapid.

*Italic source: Excerpt from _wp13167 - 14. Macroprudential policy that affects an intermediate target, like credit growth (PDF chapter/section).*

### 30.      The next section looks at the effectiveness of macroprudential policies. Expected

### _wp13167 - 30.      The next section looks at the effectiveness of macroprudential policies. Expected

### B. Policy Response — overview of instruments and transmission

- Different policy instruments affect intermediate targets and build resilience in different ways.
- Five typically-used tools and their primary balance-sheet/price channels:
  - Risk weights, Countercyclical Capital Buffer (CCB), Dynamic provisions: direct impact on economic capital.
  - Reserve requirements (RR): direct impact on funding structure.
  - Loan-to-Value (LTV) and Debt-to-Income (DTI): direct impact on assets (eligibility/flow).
- Pricing effects analyzed using the loan-pricing equation (Elliott and Santos (2012), Box 2): a unit of loan must provide return sufficient to cover cost of equity, debt and deposits backing the loan adjusted for expected losses and incidental costs.
- Intended policy assessment takes into account:
  - (a) policy effectiveness on intermediate targets, and relation of intermediate targets to the GDP forecast; and
  - (b) extent to which lowering the intermediate target lowers the probability and the cost of the crisis, that is p* and l*.

### Policy to intermediate targets: What to expect (qualitative transmission)

- Capital-based tools (risk weights, CCB):
  - Risk weights increased on mortgage loans, overheating sectors, foreign-currency loans to unhedged borrowers.
  - Can be applied to stock or flow of loans.
  - Increase in RWA raises the cost of assets (raising equity is costly) if capital ratios cannot decline.
  - Negative effect on credit growth can operate through higher lending rates as banks recover cost of raising equity.
  - CCB could have similar effects but may be less targeted than RWA.
  - Offsetting effect: investors may accept lower return on capital due to perception of higher resilience from larger buffers.
- Reserve requirements (RR):
  - Applied to deposits, foreign liabilities, wholesale funds; on stock or flows.
  - Restricts funding available to back new loans and requires setting aside assets as reserves -> less to lend; direct effect on credit growth and loan/deposit ratio.
  - Increase can lead to higher lending rates as cost of equity is higher than cost of debt or deposits.
- Loan-to-Value (LTV) and Debt-to-Income (DTI):
  - Tighten eligibility and constrain flow of new loans; reduce mortgage credit growth and house price growth.
  - Lower LTV/DTI excludes some new borrowers and reduces borrowing amounts for eligible borrowers (larger down-payments).
  - More equity at stake discourages defaults -> lowers probability of default (PD) and affects loss distribution under Basel II advanced approach (expected and unexpected losses fall).
  - Lower expected losses imply lower credit spreads and lower lending rates (loan pricing equation, Box 2).
  - Reduction in unexpected losses increases risk-based capital ratios; if banks reduce capital to keep ratios constant, cost of capital funding a loan falls, further lowering lending rates.
  - Net: quantitative impact on credit growth and house prices usually not offset by lowering of lending rates on eligible borrowers.
- Dynamic provisioning:
  - Sets a general provision buffer against fluctuations in specific provisions (incurred-loss approach causes cyclical provision patterns).
  - General provisions booked on liability side (reduce net income) or in shareholder equity (reduce distributed dividends) -> could increase cost of capital and lending rates.
  - Expected-loss accounting changes (forthcoming) will recognize upfront expected losses, potentially helpful macroprudentially.
- Table 2 (schematic): maps instruments to balance-sheet items, stock/flow application, credit-growth, house-price growth, loan/deposit, foreign liabilities/foreign assets, eligibility, and lending rates (qualitative marks preserved in source).

### Policy to intermediate targets: Empirical evidence (2000Q1–2011Q4, Survey-based)

- Data and approach:
  - Uses IMF Survey (IMF, 2011c) and extension by Lim et al (forthcoming); instruments: LTV limits, DTI limits, reserve requirements, provisioning, altering risk-weights on regulatory capital.
  - Dynamic panel regressions of intermediate target on its own lag, controls and policy variable over 2000Q1-2011Q4 for countries using the measure.
  - Policy effects estimated over and above policy interest rate or lending rate and GDP growth.
- Main empirical findings:
  - During application of LTV limits, reserve requirements and risk weights, credit growth and house price growth—intermediate targets related to the “correlated-risk-taking” externality—slow down; provisioning policy is not always significant (few users).
  - DTI limits are powerful in slowing loan/deposit growth and the share of foreign liabilities in foreign asset growth—intermediate targets related to the “fire-sale” externality.
  - RR also lowers loan-deposit ratio growth; risk weights and LTV have on average larger effects on foreign asset-liability mismatches.
- Prolonged impacts (short-term and medium-term via dynamic structure):
  - Higher capital requirements, on average, historically:
    - Lowered credit growth by about 1 percentage point on impact and by 5 percentage points cumulatively in two years.
    - Lowered house price growth by 1 percentage point in the short-term and by 6 percentage points in two years.
  - Tighter DTIs lower asset-liability funding mismatches.
  - Capital requirements effective in three of four intermediate targets featured.
  - RRs have prolonged impact on house price growth and loan-deposit growth.
- Quality of buffers matters:
  - During upswing, capital ratios buoyant due to profitability, but Basel III requires higher quality loss-absorbing buffers; ensuring higher quality with quantity is costlier and would have larger effects on intermediate targets, creating greater resilience.

### Quantitative example and mapping to crisis probability and loss (Table 3 example)

- Baseline assumptions:
  - Baseline: Credit-to-GDP change = 5pp; Real house price growth = 20% => p = 0.14; l = 0.092. (See Figure 5 and Annex 5 for estimates of p and p*, given credit growth and house price growth. See Annex 4 and Figure 8 for l.)
  - Average loss given crisis is 0.08. With slowing credit growth, loss is lowered.
- Table 3: Average effects of tightening (two-year changes and implied crisis parameters) — preserved numbers from source table:
  - Two-year change in credit growth (in percentage points):
    - Reserve Requirements (RR): -2.45
    - Capital Risk Weights: -5.04
    - Loan-to-Value (LTV) limits: -2.18
    - Debt-to-Income (DTI) limits: -2.63
  - Two-year change in house price growth (in percentage points):
    - RR: -5.36
    - Capital Risk Weights: -5.79
    - LTV: -3.70
    - DTI: -1.98
  - p* (probability of crisis after policy tightening):
    - RR: 0.03
    - Capital Risk Weights: 0.03
    - LTV: 0.034
    - DTI: 0.032
  - Loss given crisis, l*:
    - RR: 0.065
    - Capital Risk Weights: 0.050
    - LTV: 0.067
    - DTI: 0.064
  - Cost on output forecast, α:
    - RR: 0.0049
    - Capital Risk Weights: 0.0101
    - LTV: 0.0044
    - DTI: 0.0053
  - Additional row (unclear label in table but preserved numbers): 0.0058, 0.0013, 0.0064, 0.0057
  - Note: For the United States, one percentage point lower credit growth reduces the output forecast by 0.2 percent. (See Annex 3.)
  - Expression for net benefits referenced: (1 - p*l*)/(1-pl)-(1/1-α) ≧ 0? (See expression 3.1 in text for expression on net benefits.)
- Interpretation of the example (capital risk weights):
  - Tightening capital requirements by raising risk weights, on average, reduces credit-to-GDP ratio growth by a little more than 1 percentage point on impact and 5 percentage points in two years.
  - Real house price growth reduced by 5.8 percent over two years.
  - These reductions together reduce probability of crisis from 14 percent to 3 percent and loss given crisis from 9.2 percent to 5 percent (in the illustrative example).

### From intermediate targets to lowering probability and depth of crisis (p* and l*)

- Tightening policies affect intermediate targets that in turn lower the probability and depth of output loss in case of crisis.
- Empirical caution:
  - Policy instrument coded as categorical variable (+1 tightening, -1 loosening) — interpretation requires care.
  - Risk weights applied to whole stock of credit are costly for banks and more likely to be transferred to customers via higher credit spreads or other tightening (including DTI).
  - DTIs impact credit growth only if binding on flow of new customers.
- Real-world considerations:
  - Responses likely larger in practice because capital requirements and provisions build buffers that increase resilience, affect foreign liabilities growth, and have multiplicative effects on p*.
  - Confidence effects on the banking sector could reduce crisis risk but are difficult to capture in this framework.
- Policy costs:
  - Depend on sensitivity of output forecast to intermediate targets had there been no crisis.
  - If crisis-risks were zero and policy tightened, the GDP forecast level would be reduced modestly for the United States compared to no-crisis, no-policy level. Example: in Table 3, if a crisis did not occur, the GDP forecast would be reduced by (α) 1 percent.
- Net benefits:
  - Components can be combined to evaluate net benefits using the framework in Section II.
  - For the Table 3 risk-weights example, net benefit of policy remains positive when combining ingredients, but estimates require substantial judgment.
  - Policy leakages and financial-structure specifics are important considerations (analyzed in next section of source).

### Measurement and modelling components summarized (Figure 10)

- Core Macrofinancial Model, α:
  - Country-specific time series model estimating sensitivities of output forecasts to intermediate targets (Annex 3, US example).
  - Model shows: 1 percent increase in credit-to-GDP gap increases level of output-forecast modestly if no banking system distress, but decreases output-forecast one-for-one if there is a crisis.
- Probability of crisis, p:
  - Panel logit model 1970-2010 (Annex 5) relating credit growth and asset price growth to probability of a banking crisis.
- Loss given crisis, l and l*:
  - Cross-country model with 67 crisis cases (Annex 4) relating past systemic risk to output loss during a crisis.
- These measured components feed into evaluation of effects of macroprudential policies.

*Italic line: Source: IMF working paper section as provided in the content unit.*

### Annex 6

### Annex 6

### Leakages: overview
- Macroprudential policies can produce unintended side-effects through leakages when market players other than regulated institutions exploit incentives for risk-taking.
- Leakages occur when supervision and regulation do not match actual activities and forms of intermediation. When policies clamp down on banking activities, other institutions or crossborder channels can pick up the slack.
- Three core leakage channels emphasized:
  - Crossborder arbitrage through direct crossborder lending.
  - Regulatory arbitrage through parts of the domestic financial sector outside the banking regulatory perimeter (shadow banking).
  - Domestic nonbank financial intermediation that competes with regulated banks.

### Key empirical findings on crossborder leakages
- Re-estimating the panel data model on policy instruments with direct crossborder loans to the private sector as the dependent variable yields specific effects for certain instruments (Annex 6).
- Reserve requirements and provisioning policies tend to incentivize foreign parents to lend directly to the private sector, bypassing subsidiaries.
- Exact reported empirical effects:
  - A tightening of reserve requirements reduces credit growth by "0.4-0.5 of a percentage point" on impact but increases leakages through higher crossborder lending growth by "1.2 percentage points".
  - Over two years, credit growth decreases by "2.5 percentage points", and direct crossborder loan grows by "about 5 percentage points".
- Local bank provisioning policies did not have significant impacts on credit aggregates but do attract direct crossborder flows.
- Reasons why RRs increase leakages:
  - RRs are imposed on banks’ liabilities and can be circumvented if liabilities are foreign (from banks’ foreign parents) by providing direct crossborder loans instead of channeling funds through domestic branches or subsidiaries.
  - Capital requirements (higher risk weights) typically require additional capital injections by parents or lower profit distributions and are less substitutable by direct crossborder lending.
- LTV and DTI limits primarily affect loan demand and are applied mostly to mortgage loans to households; substitution to foreign loans is harder for households, reducing crossborder leakage for these instruments.

### Foreign branches and nonbank financial institutions: evidence and implications
- Foreign bank branches and nonbank intermediaries can offset the intended effects of macroprudential tightening on banks.
- Empirical example: In the U.K., when capital requirements are tightened for U.K. banks, loan supply diminishes, but about a third of the effect is offset by foreign branches in the U.K.
  - Policy implication: tightenings should be coordinated with home supervisors so foreign branches face similar constraints on U.K. exposures.
- United States case:
  - A significant share of consumer and mortgage intermediation occurs in nonbanks (shadow banks).
  - Measures of excesses based on bank credit alone failed to signal a crisis; using consumer credit growth from all sources breaches at least one threshold early in the 2000s and would have signaled a crisis.
  - Nonbanks can circumvent macroprudential policies aimed only at banks.
- Policy tools that target activities rather than institutions can reduce leakages:
  - LTV and DTI limits, applied to products across institutions, can control the flow of risky credit even if the stock of high-LTV loans remains.
  - To address the stock problem, higher risk weights on existing high-LTV loans could be applied, but effectiveness requires foreign parents to follow host rules.
  - Close coordination with other regulators is needed to address nonbank activity.

### Policy implications to counter leakages (three financial-structure cases)
- Financially open economies with foreign subsidiaries:
  - Capital tools and LTV could work well for containing the “correlated risks” externality.
  - Reserve requirements (RRs) are subject to leakages via substitution to direct crossborder lending for domestic lending.
  - Since DTIs and RRs are usually effective at containing funding risks (loan/deposit growth) linked to the “fire sales” externality, DTI limits could be used instead or alongside RRs.
- Financial systems with foreign bank branches:
  - Combination of LTV and DTI could work well as they apply to activity not institution.
  - Capital tools could be problematic for universal use as branches are regulated by home supervisors.
  - RRs could help if they can be applied to branches (for example broad money liabilities of branches).
- Financial systems with a large share of nonbanks:
  - Combinations of LTV and DTI tools could work, coordinated closely with other regulators.
  - For consumer credit, coordination between the macroprudential supervisor and the consumer protection agency is recommended if one exists.
  - Capital tools and RRs would be difficult to impose in such systems.

### Standard policy prescriptions to plug evasions and circumventions
- Regulate financial services, not sectors.
- Strengthen macroprudential regulation of the (shadow) banking sector.
- Coordinate macroprudential tools and ensure consistent application across borders, including through reciprocity principles.
- Monitor arbitrage opportunities on a regular basis.
- Address data gaps and expand credit aggregates to include all sources of credit for crisis signaling and early warning. 

### Summary: effectiveness, costs, and design considerations
- The paper provides a framework and new evidence on the effectiveness and unintended consequences of macroprudential policy instruments.
- Important numerical and qualitative conclusions:
  - The quality of early warning models matters most for net benefits; mis-use can be costly.
  - Use two thresholds (IMF (2011b) and Dell’Ariccia et al (2012)) with policy action in between to combine judgment with analytical results.
  - Policymakers need measures of the intermediate target inclusive of all sources of credit (banks and nonbanks).
  - Net benefits are negligible when the probability of crisis is very low; probability of crisis increases non-linearly when house prices and credit grow rapidly, especially when credit-to-GDP ratio is growing by "3 percentage points of GDP or more".
  - Policy mistakes (false positives) can be costly when macro-financial linkages are strong; for the United States, if policy measures are taken and there is no crisis, the output forecast drops by "0.2 percent for every 1 percentage point tightening of credit growth".
  - Net benefits rise with policy effectiveness provided costs are contained; benefits depend on large differences between pre- and post-policy probabilities of crisis (p and p*) and pre- and post-policy loss given crisis (l and l*).
  - Most effective and prolonged instruments on credit growth and house price growth are reserve requirements, higher risk weights on capital, and LTV limits; provisioning policies have not shown significant effects where data exist.
  - Historical use: LTVs have tended to have less bite than capital requirements and RRs (Section III, Annex 6).
  - Given evidence of leakages, policymakers should tailor instrument choice to the financial structure of the economy (examples summarized above).

*Annex 6, IMF staff*

### Box 2. Effects of Macroprudential Policies on Lending Rates: What to Expect

### Box 2. Effects of Macroprudential Policies on Lending Rates: What to Expect

### Loan pricing framework
- Lending decisions can be reduced to a pricing equation that determines whether a loan provides sufficient return:
  - r_L*(1-t) >= E*r_e + (D*r_d + C + A - O)*(1-t)
- Definitions of variables in the equation:
  - r_L = effective interest rate on the loan, including the annualized effect of fees;
  - t = marginal tax rate for the bank;
  - E = proportion of equity backing the loan;
  - r_e = required rate of return on the marginal equity;
  - D = proportion of debt and deposits funding the loan, assumed to be the amount of the loan minus E;
  - r_d = effective marginal interest rate on D, including indirect costs of raising funds, such as from running a branch network;
  - C = the credit spread, equal to the probability-weighted expected loss;
  - A = administrative and other expenses related to the loan;
  - O = other income and expense items related to the loan.

### Channels: how macroprudential policies affect lending rates
- Macroprudential policies change the proportion of debt and equity funding a loan, thereby affecting:
  - cost of capital (E and r_e),
  - other funding costs (D and r_d),
  - and ultimately lending rates (r_L).
- The loan rate must cover:
  - the cost of capital and other funding sources,
  - expected credit losses (C),
  - administrative expenses (A),
  - and be adjusted for other income/expense items (O) and taxes (t).

### Offsetting adjustments and short-term assumption
- Macroprudential policies may trigger offsetting adjustments to one or more variables in the loan pricing equation that reduce the impact of policies on lending rates.
- In the case of a full offset—as predicted by the Modigliani-Miller proposition—countercyclical capital would have no effect on lending rates.
- Given the empirical evidence in the next section that macroprudential policies have an effect on reducing systemic risk in the short run and that offsetting effects in the loan pricing equation may be of a long-term nature, this paper assumes that macroprudential policies have a full impact on lending rates in the short term.

### Uses in literature and modeling
- The simple loan pricing equation is an important building block in many studies because it translates regulatory costs into loan rate increases and GDP changes.
- Studies and institutions using variations of this loan pricing (accounting-based) approach include:
  - Elliott (2009 and 2010a), BCBS (2010a, 2010b), Slovik and Cournede (2011), and IIF (2011).
- Individual country studies building on the pricing equation include:
  - Koop et al. (2010), Schanz et al. (2011), de-Ramon et al. (2012).
- The pricing equation is especially useful in economic and econometric models that do not feature explicit bank capital or liquidity behavior, and for analyzing interactions between capital and liquidity standards and macroprudential policies.

*Prepared by Andre O. Santos.*

### Annex 3. An Output-Credit Forecasting Model

### Annex 3. An Output-Credit Forecasting Model

### Model and stress indicator
- The systemic financial stress (SFS) index is the fraction of banks that have negative equity returns (vis-à-vis market returns) below the 5th percentile of returns given by the joint distribution of all banks, with cumulatively negative returns for the following two weeks (see Arsov et al, 2013).
- Forecast horizons considered: h = 1, ..., H, with H = 6 quarters.
- Endogenous vector: the log of the level of real GDP, and the gap in the log of the credit-to-GDP ratio.

### Methodology
- The approach combines:
  - Local linear projection method (Jorda, 2005), also known as direct forecasts, to improve robustness against misspecification, especially in episodes of large financial distress.
  - Smooth-transition technique (Weise, 1999) to allow transmission between real economic activity (real GDP) and macro-financial developments (credit-to-GDP) to change when the economy switches from normal times to distress.
- Regression coefficients are allowed to change with an observed state variable (the index of bank distress) via a monotonically increasing sigmoid curve that maps the stress indicator into [0, 1].
  - The sigmoid function is relatively flat for low values of the stress indicator (observed most of the time) to smooth out irregular fluctuations.
  - After a threshold (determined by parameters in the sigmoid), the function rises non-linearly so transmission characteristics can change abruptly if supported by the data.
- The model is non-recursive: a separate equation is estimated for each forecast horizon; forecasts are not iterated forward mechanically (unlike VAR models).

### Forecast experiment design
- Sample period used for the experiment: 2002:1 through 2012:2.
- For each horizon h, forecasts yhat_h are computed using actually observed RHS data.
- An artificial scenario is constructed by increasing the level of credit-to-GDP ratio on the RHS by 1 percent in each period. Denote the new RHS vector by x̃ and the recalculated forecasts by yhat_h_tilde.
- The reported effect is the difference Δy_h = yhat_h_tilde − yhat_h.
- The impact of the credit-to-GDP increase on forecasts varies with the position of the economy in the financial cycle through the sigmoid function and the estimated polynomial coefficient matrix.

### Main empirical result
- The role of credit for the GDP forecast depends on the state of the banking sector:
  - A 1 percent increase in the credit-to-GDP ratio in normal times amounts to improvements in the GDP forecasts by about 0.2 percent on a 4-6 quarter-ahead horizon.
  - The same 1 percent increase reduces the GDP forecast by about 1 percent (on the same 4-6 quarter-ahead horizon) in times of large distress.

### Coefficient estimates and visualization
- Estimates of regression coefficients are depicted in Figure A3.1:
  - For each forecast horizon (t+1, t+2, t+3, t+4, t+5, t+6) regression coefficients are plotted as a function of the distress index for values between 0 percent and 25 percent (approximately the observed range).
  - Figure layout:
    - First row: GDP equation.
    - Second row: credit-to-GDP gap equation.
    - First column: sum of coefficients on current and lagged GDP terms.
    - Second column: sum of coefficients on current and lagged credit-to-GDP terms.

*Prepared by Jaromír Beneš. — Annex 3 (content unit)._wp13167*

### Annex 6. Effect of Macroprudential Policy Instruments on Intermediate Targets

### Annex 6. Effect of Macroprudential Policy Instruments on Intermediate Targets

### Cross-country and country-evidence summary
- Recent studies indicate macroprudential policy can mitigate systemic risk, especially in credit and real estate booms.
- Findings from the literature cited:
  - LTVs and DTIs
    - Almeida, Campello, and Liu (2005): LTV limits affect the financial accelerator; housing prices more sensitive to income shocks with higher maximum LTV ratios.
    - Wong et al (2011): LTV policy reduced household leverage and systemic risk in Hong Kong and other countries.
    - Ahuja and Nabar (2011): LTV limits slow property price growth; both LTVs and DTIs can slow mortgage credit growth (49 countries).
    - IMF (2011d): High LTV ratio strengthens the effect of real GDP growth on house price growth; government participation tends to exacerbate house price swings.
    - Kuttner and Shim (2012): Changes in maximum LTV and/or DTI ratios have strong effects on house prices and housing credit (57 economies).
  - Other instruments
    - Lim et al (2011): Caps on LTV, DTI, ceilings on credit, reserve requirements, countercyclical capital, and time-varying provisioning can reduce credit growth procyclicality.
    - Dell’Ariccia et al (2012): Composite macroprudential measures reduce incidence of credit booms and lower probability that booms end badly.
    - Vandenbussche, Vogel, and Detragiache (2012): Capital requirement and liquidity measure changes impacted housing price inflation in Central, Eastern, and Southeastern Europe.
    - Tovar et al (2012): Average reserve requirement and a composite of other macroprudential instruments had moderate and transitory effects on credit growth in five Latin American countries.
  - Country studies (examples)
    - Ahuja and Nabar (2011), Craig and Hua (2011), Igan and Kang (2011): LTVs, DTIs, and stamp duties in Hong Kong associated with lower transaction volumes and slower price growth.
    - Krznar and Medas (2012): Tightening LTVs associated with lower mortgage credit and house price growth in Canada.
    - Galac (2010): Credit growth ceilings in Croatia reduced domestic private but not total private sector credit (substitution to foreign debt).
    - Jimenez et al (2012): Dynamic provisioning in Spain (2000) mitigated credit supply cycles and improved aggregate and firm-level credit availability.
    - Wang and Sun (2013): Reserve requirement changes curbed credit growth and house price growth in China.

### Our approach (methodology and data)
- Objective: Estimate quantitative impact of five most frequently used macroprudential instruments on systemic-risk-related intermediate targets: credit, credit/GDP, house prices, liquidity, leverage, capital flows, and leakages.
- Instruments examined: LTVs, DTIs, risk weights (capital requirements), reserve requirements, provisioning requirements.
- Sample: 38 countries, 2000:1–2011:4; quarterly, seasonally adjusted data.
- Estimation: Dynamic panel regressions using one-step GMM Arellano-Bond estimator with instrumental variables (lags) for policy instruments to address selection bias and endogeneity. A step function variable is used for each MaPP instrument (takes +1 at the time the instrument is tightened and does not change until the subsequent change).
- Controls included: GDP growth (y-o-y growth rate of real GDP) and lending rate (average interest rate on short-and medium-term financing needs of the private sector). Regressions include individual (country) effects; time effects excluded due to high correlation with macroprudential policy variable.

### Key empirical findings (panel GMM estimates, 2000–2011)
- General patterns
  - Lagged dependent variables are highly persistent across specifications (statistically significant coefficients around 0.67–0.90 depending on dependent variable and specification).
  - Control variables (GDP growth and lending rates) generally have expected signs across specifications.

- Effects on Credit/GDP growth (Table A6.1)
  - Credit/GDP growth (t-1): coefficients 0.83***, 0.89***, 0.88***, 0.90***, 0.71*** (standard deviations 0.02, 0.01, 0.02, 0.01, 0.02 respectively).
  - GDP Growth (t): 0.33*** (0.04) in I; other specifications report smaller coefficients with respective stddevs.
  - Lending rates (t): mixed signs; specification II shows -0.14*** (0.03).
  - Reserve requirement: -0.54** (0.20) in specification I.
  - Risk weights: -0.89*** (0.25) in specification II.
  - Provisioning: -0.38 (0.31) in specification III (not statistically significant).
  - LTV: -0.39** (0.16) in specification IV.
  - DTI: -0.82*** (0.26) in specification V.
  - Number of observations by spec: 638, 631, 542, 705, 374. Number of countries by spec: 15, 15, 13, 17, 9.

- Effects on Real House Price growth (Table A6.2)
  - Real house price (t-1): 0.86***, 0.84***, 0.84***, 0.81***, 0.77*** (stddevs 0.02, 0.01, 0.02, 0.01, 0.02).
  - GDP Growth (t): positive and significant in all specs (e.g., 0.36*** (0.06) in I; 0.16*** (0.07) in V).
  - Lending rates (t): negative and often significant (e.g., -0.13*** (0.05) in II; -0.67*** (0.10) in IV).
  - Reserve requirement: -1.07** (0.26) in specification I.
  - Risk weights: -1.24*** (0.25) in specification II.
  - Provisioning: -0.16 (0.35) in specification III (not significant).
  - LTV: -0.86** (0.23) in specification IV.
  - DTI: -0.52** (0.24) in specification V.
  - Number of observations by spec: 433, 431, 428, 593, 307. Number of countries by spec: 11, 12, 11, 15, 8.

- Effects on Liquidity growth (non-core funding: bank credit to deposits) (Table A6.3)
  - Liquidity (t-1): 0.8***, 0.89***, 0.88***, 0.86***, 0.67*** (stddevs 0.02, 0.02, 0.02, 0.02, 0.02).
  - GDP Growth (t): positive and significant across specs (e.g., 0.27*** (0.05) in I).
  - Lending rates (t): negative and significant in some specs (e.g., -0.12*** (0.03) in I; -0.13*** (0.03) in III).
  - Reserve requirement: -0.5** (0.25) in specification I.
  - Risk weights: -0.03 (0.20) in specification II (not significant).
  - Provisioning: -0.2 (0.24) in specification III (not significant).
  - LTV: 0.14 (0.19) in specification IV (not significant).
  - DTI: -0.38* (0.23) in specification V.
  - Number of observations by spec: 560, 550, 493, 635, 317. Number of countries by spec: 15, 15, 13, 17, 9.

- Effects on Capital Flows (ratio of foreign liabilities to foreign assets for bank institutions) (Table A6.4)
  - Capital flows (t-1): 0.76***, 0.86***, 0.75***, 0.77***, 0.74*** (stddevs 0.03, 0.02, 0.03, 0.02, 0.03).
  - GDP Growth (t): coefficients and significance vary; e.g., 0.79* (0.45) in I; 1.04*** (0.15) in III; 1.07** (0.47) in V.
  - Lending rates (t): mixed signs and large standard deviations.
  - Reserve requirement: -2.91 (2.05) in specification I (not significant).
  - Risk weights: -2.15* (1.24) in specification II.
  - Provisioning: -3.62 (2.89) in specification III (not significant).
  - LTV: -3.26*** (1.28) in specification IV.
  - DTI: -3.63* (2.01) in specification V.
  - Number of observations by spec: 536, 520, 462, 572, 280. Number of countries by spec: 14, 14, 12, 16, 9.

- Effects on Leakages (direct cross-border credit to private sector proxied by IMF BOP items) (Table A6.5)
  - Leakages (t-1): 0.81***, 0.75***, 0.81***, 0.79***, 0.83*** (stddevs 0.03, 0.04, 0.03, 0.03, 0.04).
  - GDP Growth (t): positive and significant across specs (e.g., 0.84*** (0.14) in I).
  - Interest rates (t): negative and sometimes significant (e.g., -0.17** (0.09) in I; -0.36** (0.15) in II).
  - Reserve requirement: 1.21** (0.65) in specification I — evidence of positive association with leakages.
  - Risk weights: -2.13 (1.61) in specification II (not significant).
  - Provisioning: 1.84** (1.01) in specification III — positive association with leakages.
  - LTV: -1.85 (1.26) in specification IV (not significant).
  - DTI: -0.81 (0.78) in specification V (not significant).
  - Number of observations by spec: 270, 232, 331, 348, 197. Number of countries by spec: 10, 8, 10, 10, 6.

### Interpretation and policy-relevant implications (based on results in this Annex)
- Effectiveness
  - Tightening of LTVs, DTIs, reserve requirements, and risk weights is associated with reductions in credit/GDP growth and house price growth in the panel regressions.
  - Provisioning does not show a significant impact on credit/GDP or house price growth in the presented specifications.
  - Some macroprudential measures also affect liquidity and capital flows; notable significant negative associations include LTV and capital flows, and risk weights and capital flows in some specifications.
- Leakages and cross-border substitution
  - Reserve requirement tightening and higher provisioning are associated with increased leakages (positive coefficients: reserve requirement 1.21** (0.65); provisioning 1.84** (1.01)), suggesting potential cross-border substitution or circumvention when those instruments are tightened.
  - Other instruments show either negative or insignificant associations with leakages in the available specifications.
- Robustness notes
  - Results are based on a step function coding of instrument changes and account for endogeneity using GMM with lagged instruments.
  - The regression sample composition (number of observations and countries) varies across specifications and dependent variables.

*Prepared by Ivo Krznar; Annex 6. Effect of Macroprudential Policy Instruments on Intermediate Targets*

### References

### _wp13167 - References

### Macroprudential policy, instruments, and frameworks
- Ahuja, A. and M. Nabar, 2011: “Safeguarding Banks and Containing Property Booms: Cross-Country Evidence on Macroprudential Policies and Lesson from Hong Kong SAR”, IMF Working Paper, WP/11/284.
- Aiyar, S., Calomiris, C. W., and T. Wieladek, 2012, “Does macro-pru leak? Evidence from a UK policy experiment”, NBER Working Paper 17822.
- Basel Committee of Banking Supervision (BCBS), 2010c “Guidance for National Authorities Operating the Countercyclical Capital Buffer,” December (Bank for International Settlements).
- Committee on the Global Financial System (CGFS), 2012, “Operationalising the selection and application of macroprudential instruments”, CGFS Papers No 48.
- Crowe, C., G. Dell’Ariccia, D. Igan, and P. Rabanal, 2011, “Policies for Macrofinancial Stability: Options to Deal with Real Estate Booms,” IMF Staff Discussion Note, SDN/11/.
- De Nicolò, Gianni, Giovanni Favara, and Lev Ratnovski, “Externalities and Macroprudential Policy”, IMF Staff Discussion Note 12/05.
- Demirgüç-Kunt, A., E. Detragiache, and P. Gupta, 2006, “Inside the crisis: An empirical analysis of banking systems in distress,” Journal of International Money and Finance, Elsevier, vol. 25(5), pages 702–18, August.
- Lim, C., F. Columba, A. Costa, P. Kongsamut, A. Otani, M. Saiyid, T. Wezel, and X. Wu, 2011: “Macroprudential Policy: What Instruments and How to Use Them? Lessons from Country Experiences”, IMF Working Paper No. 11/238 (Washington: International Monetary Fund).
- Lim, C., I. Krznar, F. Lipinsky, A. Otani, and X.Wu, 2013: “The Macroprudential Framework: Policy Responsiveness and Institutional Arrangements, IMF Working Paper.
- Osinski, Jacek, Katherine Seal, and Lex Hoogdin, 2013, “Macroprudential and Microprudential Policies: Toward Cohabitation,” IMF Staff Discussion Note 13/05, (Washington: International Monetary Fund).
- Vandenbussche, J., U. Vogel, and E. Detragiache, 2012, “Macroprudential Policies and Housing Prices—A New Database and Empirical Evidence for Central, Eastern, and Southeastern Europe,” IMF Working Paper No 12/303.
- Wang, Bin, and Tao Sun, 2013, “How Effective are Macroprudential Policies in China,” IMF Working Paper 13/75. (Washington: International Monetary Fund)

### Capital, liquidity, and regulatory impact studies
- Bank of England, 2010, “The long-term economic impact of higher capital levels” Financial Stability Report.
- Basel Committee of Banking Supervision (BCBS), 2010a, “An assessment of the long-term economic impact of stronger capital and liquidity requirements.”
- Basel Committee of Banking Supervision (BCBS), 2010b, “Assessing the macroeconomic impact of the transition to stronger capital and liquidity requirements” (Interim and Final reports).
- _______, 2011, Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems (Basel: Bank for International Settlements; June 1). www.bis.org/publ/bcbs189.htm.
- European Parliament, 2011, “Assessment of the cumulative impact of various regulatory initiatives on the European Banking sector,” PE 464.439.
- Federal Reserve Bank of New York, 2011, “Basel III: Long-Term Impact on Economic Performance and Fluctuations,” Staff Report No. 485.
- Financial Services Authority, 2009, “Optimal regulation of bank capital and liquidity: how to calibrate new international standards,” Occasional Paper Series No. 38.
- ICFR 2011 “What is the price of Basel III? An analysis of the impact (and impact studies) of Basel III on the real economy.”
- Roger, Scott and Jan Vlcek, 2011, Macroeconomic Costs of Higher Bank Capital and Liquidity Requirements, IMF WP/11/103.
- Santos, André Oliveira and Douglas Elliott, 2012, “Estimating the Costs of Financial Regulation,” IMF Staff Discussion Note 12/11, September (Washington: International Monetary Fund).
- Slovik, Patrick, and Boris Cournède, 2011, Macroeconomic Impact of Basel III. OECD Economics Department Working Papers. (OECD Publishing).
- Institute of International Finance, 2010, “The net cumulative economic impact of banking sector regulation: some new perspectives.”

### Credit, housing, and real estate dynamics
- Almeida, H., M. Campello and C. Liu, "The Financial Accelerator: Evidence from International Housing Markets." Review of Finance, 10, pp. 1–32, 2006.
- Crowe, C., G. Dell’Ariccia, D. Igan, and P. Rabanal, 2011, “Policies for Macrofinancial Stability: Options to Deal with Real Estate Booms,” IMF Staff Discussion Note, SDN/11/.
- Dell'Ariccia, Giovanni, Igan, Deniz, Laeven, Luc, and Tong, Hui, " Policies for Macrofinancial Stability: How to Deal with Credit Booms," IMF Staff Discussion Note 12/06.
- Enoch, C. and I. Ötker-Robe, 2007, “Rapid Credit Growth in Central and Eastern Europe: Endless Boom or Early Warning?”, Palgrave Macmillan, New York, and International Monetary Fund.
- Igan, D. and H. Kang, 2011: “Do Loan-to-Value and Debt-to-Income Limits Work? Evidence from Korea”, IMF Working Paper No. 11/297.
- Jiménez, G., S. Ongena, J-O. Peydro, and J. Saurina, 2012, “Macroprudential Policy, Countercyclical Bank Capital Buffers and Credit Supply: Evidence from the Spanish Dynamic Provisioning Experiments,” European Banking Center Discussion Paper No. 2012-011.
- Kuttner, Kenneth and Ilhyock Shim, 2012. “Taming the Real Estate Beast: The Effects of Monetary and Macroprudential Policies on Housing Prices and Credit,” RBA Annual Conference Volume.
- Wong, E., T. Fong, K. Li, and H. Choi, 2011: “Loan-to-Value Ratio as a Macroprudential Tool: Hong Kong’s Experience and Cross-Country Evidence”, Hong Kong Monetary Authority Working Paper No. 01/2011.
- Vandenbussche, J., U. Vogel, and E. Detragiache, 2012, “Macroprudential Policies and Housing Prices—A New Database and Empirical Evidence for Central, Eastern, and Southeastern Europe,” IMF Working Paper No 12/303.

### Financial crises, banking distress, and systemic risk measurement
- Aziz, Jahangir, Francesco Caramezza, and Ranil Salgado, 2000, “Currency Crises: In Search of Common Elements,” IMF Working Paper 00/67 (Washington: International Monetary Fund).
- Bordo, M., B. Eichengreen, D. Klingebiel, M. S. Martinez-Peria and A. Rose, 2001, “Is the Crisis Problem Growing More Severe?,” Economic Policy Vol. 16, No. 32 (April, 2001).
- Borio, C., and M. Drehmann, 2009, "Assessing the risk of banking crises˗˗revisited," BIS Quarterly Review, Bank for International Settlements, March.
- Caprio, G., and D. Klingebiel, 1996, “Bank Insolvencies: Cross-Country Experience,” World Bank Policy Research Working Paper No. 1620.
- Caprio, G., and D. Klingebiel, 1999, “Episodes of Systemic and Borderline Financial Crises,” World Bank Policy Research Working Paper No. 1620.
- Laeven, Luc, and Fabian Valencia, 2008, "Systemic Banking Crises: A New Database," IMF Working Paper 08/224, (Washington: International Monetary Fund).
- Laeven, Luc, and Fabian Valencia, 2010, “Resolution of Banking Crises: The Good, the Bad, and the Ugly,” IMF Working Paper 10/146 (Washington: International Monetary Fund).
- Reinhart, C., and K. Rogoff, 2009. "The Aftermath of Financial Crises," American Economic Review, American Economic Association, vol. 99(2), pages 466–72, May.
- Hutchison, M., and I. Noy, 2005. "How Bad Are Twins? Output Costs of Currency and Banking Crises," Journal of Money, Credit and Banking, Blackwell Publishing, vol. 37(4), pages 725–52, August.
- Arsov, Ivailo, Elie Canetti, Laura Kodres and Srobona Mitra, 2013, “Near-Coincident Indicators of Systemic Financial Stress,” IMF Working Paper 13/115.
- Drehmann, Mathias, Claudio Borio & Kostas Tsatsaronis, 2011. “Anchoring Countercyclical Capital Buffers: The role of Credit Aggregates,” International Journal of Central Banking, vol. 7(4), pages 189–240, December.

### Empirical methods, time series, and macro-financial interaction
- Beveridge, Stephen and Charles R. Nelson, 1981, “A New Approach to the Decomposition of Economic Time Series into Permanent and Transitory Components with Particular Attention to Measurement of the Business Cycle,” Journal of Monetary Economics 7, pp. 151–74.
- Nelson, Charles R., 2008, “The Beveridge–Nelson Decomposition in Retrospect and Prospect,” Journal of Econometrics, Vol. 146 (2), pages 202–06.
- Jorda, O, 2005 “Estimation and Inference of Impulse Responses by Local Projections,” American Economic Review, 2005, 95(1), 161–82.
- Schularick, M., and A. M. Taylor, 2012, “Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and Financial Crises, 1870–2008,” American Economic Review, 102 (April 2012): pp 1029–61.
- Claessens, Stijn, M. Ayhan Kose, and Marco E. Terrones, 2011a, “How Do Business and Financial Cycles Interact?” IMF Working Paper 11/88 (Washington: International Monetary Fund).
- Claessens, Stijn,  Zoltan Pozsar, Lev Ratnovski, and Manmohan Singh, 2012, “Shadow Banking: Economics and Policy,” IMF Staff Discussion Note 12/12.

### Country and central bank studies, crisis management, and special topics
- Aiyar, S., 2011, “How did the crisis in international funding markets affect bank lending? Balance sheet evidence from the UK”, Bank of England Working Paper, No 424.
- Galac, T., 2010, “The Central Bank as Crisis Manager in Croatia—A Counterfactual Analysis”, Croatian National Bank Working Paper W -27, December 2010.
- Krznar, I. and P. Medas, 2012: “Recent experience with macroprudential tools in Canada: effectiveness and options moving forward”, Canada: Selected Issues, Country report No. 13/41.
- Lund-Jensen, Kasper, 2012, “Monitoring Systemic Risk Using Dynamic Thresholds,” IMF Working Paper 12/159, (Washington: International Monetary Fund). http://www.imf.org/external/pubs/ft/wp/2012/wp12159.pdf
- Galac, T., 2010, “The Central Bank as Crisis Manager in Croatia—A Counterfactual Analysis”, Croatian National Bank Working Paper W -27, December 2010.

### IMF policy papers and Global Financial Stability Report chapters
- International Monetary Fund (IMF), 1998, “Financial Crises—Characteristics and Vulnerabilities,” Chapter IV in the World Economic Outlook, www.imf.org/external/pubs/ft/weo/weo0598/.../0598ch4.pdf.
- _____, 2009, “From Recession to Recovery: How Soon and How Strong?” Chapter 3 in the World Economic Outlook, http://www.imf.org/external/pubs/ft/weo/2009/01/.
- _____, 2011a, “Macroprudential Policy: An Organizing Framework,” IMF Policy Paper (Washington). www.imf.org/external/np/pp/eng/2011/031411.pdf
- _____, 2011b, “Towards Operationalizing Macroprudential Policy—When to Act?” Chapter 3 in the Global Financial Stability Report, September (Washington: International Monetary Fund), http://www.imf.org/External/Pubs/FT/GFSR/2011/02/index.htm.
- _____, 2011c, “Macroprudential Policy: An Organizing Framework Background Paper,” http://www.imf.org/external/np/pp/eng/2011/031411a.pdf
- _____, 2011d, “Housing Finance and Financial Stability—Back to Basics??” Chapter 3 in Global Financial Stability Report, September (Washington: International Monetary Fund), http://www.imf.org/external/pubs/ft/gfsr/2011/01/index.htm.
- _____, 2012a, “Changing Global Financial Structures: Can They Improve Economic Outcomes?” Chapter 4 in the Global Financial Stability Report, October (Washington: International Monetary Fund), http://www.imf.org/External/Pubs/FT/GFSR/2012/02/index.htm.
- _____, IMF, 2012b, “The Reform Agenda: An Interim Report on Progress Toward a Safer Financial System,” Chapter 3 in the Global Financial Stability Report, October (Washington: International Monetary Fund), http://www.imf.org/External/Pubs/FT/GFSR/2012/02/index.htm.
- _____, IMF, 2013, “The Interaction Between Monetary and Macroprudential Policies,” IMF Policy Paper, January, http://www.imf.org/external/np/pp/eng/2013/012913.pdf .

*Compiled from the References section of _wp13167*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13167.pdf_