## 1. Summary Statistics

## Source details

**Canonical URL:** [1. Summary Statistics](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12303.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12303.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12303.pdf.json)

---

### Regional context and motivation
- CESEE experienced large, synchronized credit and housing boom-bust cycles during the last decade; housing price inflation in Estonia, Latvia, and Lithuania was in the range of 120–160 percent between 2004:Q1 and 2007:Q1.
- Policymakers deployed a variety of macroprudential instruments (capital requirements, loan classification and provisioning rules, reserve or liquidity requirements, credit eligibility criteria) with differences in timing and intensity across countries.
- Some countries did not use MPPs to curb credit expansion; in some cases policies were tightened late, and in others policies were relaxed during the expansion (e.g., for harmonization upon joining the European Union).

### Data and methodological contributions
- Constructed a quarterly database of prudential measures grouped into 29 categories, covering sixteen CESEE countries from the late 1990’s or early-2000’s to end-2010.
- Devised scoring rules to quantify each measure’s intensity over time and across types; twenty-nine categories coded numerically with linear transformations or fixed increments (0.25, 0.5, 1) for tightenings/loosings.
- Focused empirically on housing price inflation as the outcome because:
  - housing booms are linked to financial instability in the literature;
  - domestic credit aggregates are often affected by foreign-exchange denomination or indexing, complicating cross-country quarterly real credit growth series;
  - housing price data drawbacks acknowledged (uneven quality, comparability issues, short series for some countries); foreign investor demand affected some markets.
- Quarterly housing price series compiled for 16 CESEE countries (Albania, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Serbia, Slovakia, Slovenia, Turkey, and Ukraine), generally beginning in the early 2000s; series deflated with national CPI and seasonally adjusted.
- MPP dataset sources: national central bank/supervisor documents, IMF Staff Reports, FSAP documents, cross-checked with country-level studies and book chapters.

### Modeling approach and econometric framework
- Modeled housing price dynamics using an error correction model with a long-run relationship between log real housing prices and log real GDP per capita; estimated long-run elasticity about one (model equivalent to housing price / income per capita).
- Panel error-correction specification for ∆(log real housing price) includes:
  - lagged changes in log real GDP per capita,
  - lagged domestic real interest rate,
  - lagged effective foreign currency real interest rate,
  - lagged changes in MPP variables,
  - change in working-age population,
  - country fixed effects and period dummies,
  - error-correction term from long-run relationship.
- Panel is unbalanced; sample ends in 2011:Q1. Boom defined as 2002:Q4–2008:Q3 (except Estonia and Latvia: boom ends 2007:Q3); bust from 2008:Q4–2011:Q1.
- Preliminary regressions include two lags of each policy variable and two lags of an aggregate index of remaining MPP changes; baseline regression includes core set of three MPPs separately and an aggregate for others.
- Estimation implemented with fixed effects (Stata xtpmg with dfe and cluster options).

### Macroprudential policy taxonomy (29 categories aggregated into 5 groups)
- Capital measures (12 types): includes mincap (minimum CAR), tgtmincap, cap, cgrcap, hhsc, fcsc, lsctot, and various risk-weight related measures.
- Provisioning measures: gp, dp, dpfc, dptot.
- Liquidity measures: rr, rrfc, rrbase, mrr (marginal reserve requirements on foreign funding), srr, cc, cgr, fclr, lr, and combined indices (mpprr, rrtot, mrrtot, cgrr).
- Loan eligibility requirements: ltv, ltvfc, dti, dtifc, and combined indices (ltv tot, dti tot, elig).
- Other quantitative restrictions: limits or bans on foreign-currency lending (otherfc).
- Tax and non-bank measures: tax, other, rest.

### Coding and construction of policy strength indices (selected operationalizations)
- mincap: score = quarterly change in minimum ratio (e.g., move from 8% to 10% yields score of 2).
- rwmol: operationalization = (rw mortgage loans - basel risk weights for mortgage loans) / 25.
- rwmolfc: operationalization = (rw mortgage loans in fc - rw for mortgage loans in lc) / 50.
- gp/dp/dpfc: index changes of 0.5 for tightening/loosening general or specific provisioning rules.
- rr: reserve requirement rate on lc deposits (percent); mpprr operationalized as (rr/10 + rrfc/10)/2 if rrfc>0 else rr/10.
- mrr: marginal reserve requirements on foreign funding operationalized as (mrr rate/10) / 2.
- cgr/cc: credit growth reserve and marginal reserve on credit growth operationalized as functions of thresholds and penalty rates (e.g., (10/annual threshold) * (penalty rate/100)).
- ltv: operationalization = (100 - maximum LTV) / 20 [default=100].
- dti: operationalization = (1 - (maximum DTI / 60)) * 4 [default=60].

### Key empirical findings
- Short-run evidence indicates significant negative effects on housing price inflation from:
  - changes in the minimum capital adequacy ratio (minimum CAR),
  - changes in marginal reserve requirements on foreign funding,
  - changes in marginal reserve requirements linked to credit growth.
- No robust evidence that changes in standard average reserve requirements, provisioning rules, or eligibility criteria (loan-to-value ratio, debt-service-to-income ratio) had significant effects in this sample.
- Three policies had a significant impact when tightened; only changes to the minimum CAR had a significant impact when eased.
- The four policies analyzed had a strong impact during boom years; impacts during the bust were less robust.
- Endogeneity concerns and timing/anticipation of policy implementation may bias estimated policy effects downward and delay observable impacts beyond immediate quarters.

### Empirical results — selected coefficients and statistics (baseline regression)
- Long-run growth coefficient on GDP per capita: approximately 1.23.
- Error-correction coefficient: -0.06 (p-value 0.00).
- Autoregressive short-run terms:
  - ∆(log housing price index) t-1: -0.42*** (p-value 0.00) in baseline.
  - ∆(log housing price index) t-2: -0.17*** (p-value 0.00) in baseline.
- MPP coefficients (baseline column (1), p-values in parentheses):
  - ∆(minimum capital adequacy ratio) t-1: -4.32*** (0.00)
  - ∆(minimum capital adequacy ratio) t-2: -1.06 (0.12)
  - ∆(marginal reserve requirements on foreign funding) t-1: -1.62** (0.02)
  - ∆(marginal reserve requirements on foreign funding) t-2: -0.08 (0.84)
  - ∆(marginal reserve requirements on credit growth) t-1: -2.37*** (0.00)
  - ∆(marginal reserve requirements on credit growth) t-2: -2.37*** (0.00)
  - ∆(other policies) t-1: -0.23 (0.52); t-2: -0.31 (0.39)
- Model fit and sample:
  - R-sqr: 0.532; adj.R-sqr: 0.467
  - Number of observations: 541

### Economic magnitudes from dynamic multipliers and point estimates
- A one percentage point increase in the minimum capital requirement:
  - Maximum impact after four-to-five quarters: housing prices are 8.5 percent lower than they would have been without the policy change.
- Point estimate: 4.5 percent (standard errors large; precise quantification not possible within sample and framework).
- Countercyclical capital buffer note: under Basel III can reach up to 2.5 percentage points (corresponding to a score of 2.5 in the scoring system) and could potentially have a large impact on housing price dynamics in the short and medium term.
- Marginal reserve requirements on foreign borrowing: an increase by 20 percentage points lowers housing prices by only 3 percent at the peak.
- Marginal reserve requirements related to excessive credit growth: a one-point increase in the index (maximum observed in the sample is 1.18 points) lowers housing prices by 8 percentage points after one year.
- Dynamic multiplier timing: effects generally observed within the first year; some measures peak within the sample’s short- to medium-term horizon.

### Asymmetry and heterogeneity of effects
- Minimum CAR: effect stronger and more prolonged when regulation was loosened (examples: Latvia, Lithuania, Romania during the boom) than when tightened.
- Marginal reserve requirements (foreign funding and credit growth): tightening shows more robust effects; coefficients bigger and more statistically significant than in baseline.
- Tightening vs easing (selected coefficients, p-values in parentheses):
  - Policy tightening t-1: Minimum CAR -2.68*** (0.00); Foreign funding -3.49*** (0.00); Credit growth -2.18** (0.03)
  - Policy easing t-1: Minimum CAR -7.38** (0.01); Foreign funding -1.08 (0.19); Credit growth -2.48 (0.13)
- Boom period (selected t-1): Minimum CAR -4.52** (0.05); Foreign funding -3.89*** (0.00); Credit growth -2.72*** (0.00)
- Bust period (selected t-1 and t-2): mixed/sign-dependent coefficients (e.g., Minimum CAR t-1 -4.98*** (0.00); Credit growth t-2 -13.16*** (0.00)).

### Robustness, limitations, and interpretation caveats
- Endogeneity: policymakers may tighten MPPs in anticipation of housing price inflation, biasing estimated impacts downward.
- Timing and calibration: measures may be non-binding initially; effects may appear after several quarters or upon announcement rather than implementation.
- Measurement/coding judgment: numerical scoring involves judgment; interactions between different measures not modeled except through aggregate indices.
- Cross-border lending and non-bank channels sometimes circumvent domestic MPPs, complicating attribution when analyzing domestic credit volumes.
- Heterogeneity in housing price series (different property types, geographic coverage) addressed via country fixed effects.
- Reserve requirements may serve multiple objectives (e.g., sterilization of FX intervention) and have multidimensional features not fully captured.
- Cointegration tests: one of two Westerlund (2007) panel tests does not reject the null of no cointegration between housing prices and GDP per capita; excluding the error-correction term yields effects consistent with baseline.

### Policy implications and takeaways
- Tightening minimum CAR, imposing marginal reserve requirements on foreign funding, and imposing marginal reserve requirements related to excessive credit growth appear effective in lowering housing price inflation in CESEE in the short run.
- Standard average reserve requirements, simple provisioning rules, and borrower-eligibility limits (LTV/DTI) show limited evidence of effectiveness in this sample and timeframe; sparse use and potential circumvention through cross-border credit may limit efficacy.
- Macroprudential measures appear most effective during boom periods; their impact during bust periods is less consistent.
- Policymakers should be mindful of endogeneity, calibration, timing, anticipation, and possible circumvention (e.g., cross-border substitution) when designing and evaluating MPPs.
- Suggested future research: explore alternative dependent variables (construction activity, real estate transaction volumes, credit growth once detailed currency-composition data available).

### Appendix — Selected descriptions of measures and scoring rules (from Appendix 2)
- CAPITAL (examples):
  - mincap: minimum required capital adequacy ratio (percent); series description: minimum capital adequacy ratio.
  - hhsc: maximum ratio of household loans to share capital; operationalization: index with change of 1 for implementing/abandoning, change of 0.5 for penalty/base changes.
- RISK WEIGHTS (examples):
  - rwmol: risk-weight on mortgage loans in lc (percent); operationalization = (rw mortgage loans - basel risk weights for mortgage loans) / 25.
  - rwmolfc: risk-weight on mortgage loans in fc; operationalization = (rw mortgage loans in fc - rw for mortgage loans in lc) / 50.
- PROVISIONING:
  - gp/dp/dpfc: indices with changes of 0.5 for tightening/loosening.
- LIQUIDITY (examples):
  - rr/rrfc: reserve requirement rates on lc/fc deposits (percent).
  - mrr: marginal reserve requirements on foreign funding (percent); operationalization = (mrr rate/10) / 2.
  - cgr/cc: credit growth reserve and marginal reserve on credit growth operationalized via (10/threshold) * (penalty rate/100) formulas and step functions.
- ELIGIBILITY:
  - ltv: Loan-to-value ceiling (percent); operationalization = (100 - maximum LTV) / 20 [default=100].
  - dti: Debt-service-to-income ceiling (percent); operationalization = (1 - (maximum DTI / 60)) * 4 [default=60].
- OTHER: tax and non-bank measures operationalized as indices changing by 0.5 for tightening/easing.

*Source: _wp12303 - 1. Summary Statistics*

### 1. Summary Statistics...................................................................................................

### 1. Summary Statistics

### Regional context and motivation
- The CESEE region experienced large and synchronized credit and housing boom-bust cycles during the last decade, making it a rich setting to study macroprudential policy (MPP) effectiveness.
- Example magnitude: housing price inflation in Estonia, Latvia, and Lithuania was in the range of 120–160 percent between 2004:Q1 and 2007:Q1.
- Policymakers in CESEE deployed a variety of instruments (capital requirements, loan classification and provisioning rules, reserve or liquidity requirements, credit eligibility criteria), with differences in timing and intensity across countries.
- Some countries did not use MPPs to curb credit expansion; in some cases policies were tightened late, and in others policies were relaxed during the expansion for reasons including harmonization upon joining the European Union.

### Data and methodological contributions
- Constructed a comprehensive quarterly database of all major prudential measures grouped into 29 categories, covering sixteen CESEE countries from the late 1990’s or early-2000’s to end-2010.
- Devised scoring rules to quantify each measure’s intensity over time and across types.
- Focused empirically on housing price inflation as the outcome variable because:
  - Housing booms are linked to financial instability in the literature.
  - Domestic credit aggregates in CESEE are frequently affected by foreign-exchange denomination or indexing, producing valuation effects that complicate cross-country quarterly real credit growth series; comprehensive currency-breakdown data are not widely available for about half the countries studied.
  - Acknowledged drawbacks of housing price data: uneven quality, cross-country comparability issues, and for some countries short time series; also acknowledged foreign investor demand affected housing prices in some markets.

### Modeling approach
- Modeled housing price dynamics using an error correction model with a long-run relationship between housing prices and output per capita.
- Estimated long-run elasticity is about one, making the model equivalent to one where the ratio of housing prices to income per capita (housing affordability) is a determinant.
- Short-run dynamics were estimated to detect the impact of MPP instrument changes on housing price inflation.

### Key empirical findings
- Short-run evidence indicates that the following instruments had an impact on housing price inflation:
  - Changes in the minimum capital adequacy ratio (CAR).
  - Changes in marginal reserve requirements on foreign funding.
  - Changes in marginal reserve requirements linked to credit growth.
- Quantitative magnitude: a change by one percentage point of the minimum CAR has on average a cumulative effect of 8.5 percent on housing prices after four quarters.

### Data limitations and caveats noted
- The database includes all major prudential policy measures that may affect the price or availability of credit; the authors do not claim that all measures were adopted for macroprudential reasons rather than microprudential ones.
- Cross-border lending and non-bank channels sometimes circumvented domestic MPPs, complicating attribution of effects when analyzing domestic credit volumes.
- Foreign investor demand and access to foreign financing influenced housing price dynamics in some markets.

*Source: _wp12303 - 1. Summary Statistics*

### 0.92 percent in our sample. We do not find robust evidence that changes in standard average

### _wp12303 - 0.92 percent in our sample. We do not find robust evidence that changes in standard average reserve requirements, provisioning rules, or eligibility criteria (loan-to-value ratio, debt-service-to-income ratio) had any significant effect.

### Main findings on macroprudential policy (MPP) effectiveness
- Changes in minimum capital adequacy ratio (minimum CAR), marginal reserve requirements related to foreign borrowing, and marginal reserve requirements related to credit growth had significant negative effects on housing price inflation in the short run (first two quarters) and in the baseline specification.
- Changes in standard average reserve requirements, provisioning rules, or eligibility criteria (loan-to-value ratio, debt-service-to-income ratio) do not show robust evidence of significant effects on housing price inflation in this sample.
- Three policies had a significant impact when tightened; only changes to the minimum CAR had a significant impact when eased.
- The four policies analyzed had a strong impact during boom years; the impact during the bust was less robust.
- Endogeneity concerns and timing/anticipation of policy implementation may bias estimated policy effects downward and delay observable impacts beyond the immediate quarters following implementation.

### Data sources and coverage
- Quarterly housing price series compiled for 16 CESEE countries (Albania, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Serbia, Slovakia, Slovenia, Turkey, and Ukraine), generally beginning in the early 2000s; series are deflated with national CPI and seasonally adjusted.
- Fundamental macroeconomic variables include real income per capita, domestic deposit rate (as domestic currency interest rate), effective foreign currency policy rate (Fed Funds for Russia, Turkey, Ukraine; ECB policy rate for others) adjusted by year-on-year appreciation of local currency.
- MPP dataset constructed quarterly for 16 CESEE countries from national central bank/supervisor documents, IMF Staff Reports, FSAP documents, and cross-checked with country-level studies and book chapters.

### Macroprudential policy taxonomy (29 categories aggregated into 5 groups)
- Capital measures: 12 types including minimum CAR, minimum target CAR, minimum CAR related to credit growth, definition of regulatory capital, maximum loans-to-households relative to capital, maximum foreign-currency loans to capital, and risk-weights on mortgage and corporate exposures.
- Provisioning measures: changes in general provisions and specific provisions (domestic and foreign currency).
- Liquidity measures: average reserve requirement ratios (domestic and foreign currency), reserve requirement base, marginal reserve requirements on foreign borrowing, special reserve requirements on securities, marginal reserve requirements related to credit growth, liquidity ratios, foreign currency liquidity requirements.
- Loan eligibility requirements: maximum loan-to-value (local and foreign currency) and maximum debt-service-to-income (local and foreign currency).
- Other quantitative restrictions: limits or bans on foreign-currency lending (stock or flow terms).

### Coding and construction of policy strength indices
- Twenty-nine categories coded numerically to capture relative changes over time and across measure types; for simple numeric regulations, a linear transformation of the regulatory number is used; for more complex rules, fixed increments of 0.25, 0.5, or 1 are used for tightenings/loosings.
- Examples:
  - Minimum CAR: score = quarterly change in minimum ratio (e.g., move from 8% to 10% yields score of 2).
  - Across-the-board risk-weight on domestic mortgages (Basel deviation): compute deviation from Basel I/II, divide by 25, take quarterly change (value of 2 when risk-weight 100 vs. 50 under Basel I).
  - Penalty on foreign-currency mortgages: difference between risk-weights on foreign vs. domestic divided by 50, quarterly change (score of 1 for a 50 percentage point penalty).
- Quarterly scores summed across categories to produce a change-in-strength series; cumulative sums used to represent cumulative policy stance changes.

### Econometric framework
- Panel error-correction model where ∆(log real housing price) is explained by:
  - lagged changes in log real GDP per capita,
  - lagged domestic real interest rate,
  - lagged effective foreign currency real interest rate,
  - lagged changes in MPP variables,
  - working-age population change,
  - country fixed effects and period dummies,
  - error-correction term from long-run relationship between log real housing prices and log real GDP per capita.
- Panel is unbalanced; sample ends in 2011:Q1. Boom defined as 2002:Q4–2008:Q3 (except Estonia and Latvia: boom ends 2007:Q3); bust from 2008:Q4–2011:Q1.
- Preliminary regressions include two lags of each policy variable and two lags of an aggregate index of remaining MPP changes; baseline regression includes a core set of three MPPs separately and an aggregate for others.
- Estimation implemented with fixed effects (Stata xtpmg with dfe and cluster options).

### Empirical results — preliminary regressions (selected outcomes)
- Among capital measures:
  - Minimum CAR: significant effect in the first quarter following change (expected negative direction).
  - Maximum household loans/capital: significant effect in expected direction for first lag.
  - Risk-weight changes on loans to households: do not show consistent effects.
- Provisioning measures: little evidence of impact on housing price inflation in short run.
- Liquidity measures:
  - Average reserve requirements: no significant effect on housing price inflation.
  - Marginal reserve requirements on foreign funding: significant and associated with housing price declines.
  - Marginal reserve requirements on credit growth: significant and associated with housing price declines.
- Eligibility measures (LTV/DTI): generally insignificant in CESEE sample; across-the-board LTV and DTI have signs consistent with expectations but limited statistical support.
- Selection of core set for baseline regression: changes in minimum CAR, marginal reserve requirements on foreign funding, and marginal reserve requirements on credit growth (each changed between seven and nine times in sample).

### Baseline regression (selected coefficients and statistics)
- Long-run growth coefficient on GDP per capita approximately 1.23 (reported without p-value in table).
- Error-correction coefficient: -0.06 (p-value 0.00) in baseline.
- Autoregressive short-run terms:
  - ∆(log housing price index) t-1 coefficient: -0.42*** (p-value 0.00) in baseline; 0.42*** (p-value 0.00) in regression without EC term.
  - ∆(log housing price index) t-2 coefficient: -0.17*** (p-value 0.00) in baseline; 0.13** (p-value 0.05) without EC term.
- MPP coefficients (baseline column (1), p-values in parentheses):
  - ∆(minimum capital adequacy ratio) t-1: -4.32*** (0.00)
  - ∆(minimum capital adequacy ratio) t-2: -1.06 (0.12)
  - ∆(marginal reserve requirements on foreign funding) t-1: -1.62** (0.02)
  - ∆(marginal reserve requirements on foreign funding) t-2: -0.08 (0.84)
  - ∆(marginal reserve requirements on credit growth) t-1: -2.37*** (0.00)
  - ∆(marginal reserve requirements on credit growth) t-2: -2.37*** (0.00)
  - ∆(other policies) t-1: -0.23 (0.52); t-2: -0.31 (0.39)
- Model fit and sample:
  - R-sqr: 0.532; adj.R-sqr: 0.467
  - Number of observations: 541

### Economic magnitudes (from dynamic multipliers)
- A one percentage point increase in the minimum capital requirement:
  - Maximum impact after four-to-five quarters: housing prices are 8.5 percent lower than they would have been without the policy change.
  - The effect decays thereafter; cumulative decline after ten quarters reported but truncated in available excerpt.

### Robustness, limitations, and interpretation caveats
- Endogeneity: policymakers may tighten MPPs in anticipation of housing price inflation, which can bias estimated impacts downward.
- Timing and calibration: some measures may be non-binding initially and effects may appear after several quarters; anticipation of measures may generate effects before official implementation dates.
- Measurement/coding judgment: numerical scoring of policy strength involves judgment; interactions between different measures are not modeled except through aggregate indices.
- Heterogeneity in housing price series across countries (different types of residential real estate, geographic coverage) addressed by country fixed effects.
- Reserve requirements may be used for sterilization of FX intervention or have multidimensional features not fully captured (eligibility of assets, averaging rules), which can complicate inference on their effectiveness.

### Policy implications and takeaways
- Tightening minimum CAR, imposing marginal reserve requirements on foreign funding, and imposing marginal reserve requirements related to excessive credit growth appear to be effective tools for lowering housing price inflation in CESEE in the short run.
- Standard average reserve requirements, simple provisioning rules, and borrower-eligibility limits (LTV/DTI) show limited evidence of effectiveness in this sample and timeframe; sparse use and potential circumvention through cross-border credit may limit efficacy.
- Macroprudential measures appear most effective when applied during boom periods; their impact during bust periods is less consistent.
- Policymakers should be mindful of endogeneity, calibration, timing, anticipation, and possible circumvention (e.g., cross-border substitution) when designing and evaluating MPPs.

*Source: Authors’ calculations and analysis as presented in the supplied excerpt of the IMF working paper.*

### 4.5 percent. Standard errors around this point estimates, however, are quite large, indicating

### _wp12303 - 4.5 percent. Standard errors around this point estimates, however, are quite large, indicating

### Key empirical findings on MPP effectiveness
- Point estimate: 4.5 percent. Standard errors around this point estimates, however, are quite large, indicating that a precise quantification of the magnitude of the effect is not possible within our sample and empirical framework.
- Countercyclical capital buffer implication: The countercyclical capital buffer under Basel III, which can reach up to 2.5 percentage points (corresponding to a score of 2.5 in our scoring system), could potentially have a large impact on housing price dynamics in the short and medium term.
- Marginal reserve requirements on foreign borrowing: An increase by 20 percentage points lowers housing prices by only 3 percent at the peak.
- Marginal reserve requirements related to excessive credit growth: A one-point increase in the index (maximum observed in the sample is 1.18 points) lowers housing prices by 8 percentage points after one year.

### Dynamic multipliers and timing
- Figure description: Each shock represents an increase by one unit in the intensity of the policy variable. The cumulative change in house prices is shown on the vertical axis (in percent). Time (in periods) is on the horizontal axis.
- Effects reported for three core MPPs over 10 periods: Minimum capital adequacy ratio; Marginal reserve requirements on foreign funding; Marginal reserve requirements related to credit growth.
- Timing: Effects are generally observed within the first year, with some measures peaking within the sample’s short- to medium-term horizon.

### Asymmetry of MPP effects
- Minimum CAR:
  - Effect on housing prices stronger and more prolonged when regulation was loosened (examples: Latvia, Lithuania, Romania during the boom) than when it was tightened.
- Marginal reserve requirements (foreign funding and credit growth):
  - Only tightening seems to have had a robust effect; coefficients are bigger and more statistically significant than in the baseline.
- Endogeneity caveat:
  - Asymmetries may reflect endogeneity biases. Example: capital requirements tightened in response to expected acceleration in housing prices, while loosening may have occurred for exogenous reasons (e.g., harmonization with EU minima following EU entry). This would bias tightening coefficients differently from loosening coefficients.

### Regression robustness and alternative specifications
- Long-run relationship: The estimated long-run relationship between housing prices and GDP per capita is not statistically significant in the baseline regression.
- Cointegration test: One of the two Westerlund (2007) panel tests does not reject the null of no cointegration between housing prices and GDP per capita.
- Alternative regression: Excluding the error-correction term (i.e., same variables as baseline except for the error-correction term) yields significance and order of magnitude of effects for the four MPPs in the core set consistent with baseline results (see Column (2) of Table 3).

### Tabulated coefficients (selected results from Table 4)
- Note: Dependent variable is the log difference of the real housing price index. Regressions include time and country fixed effects. P-values in parentheses. *, **, and *** denote statistical significance at the 10 percent, 5 percent and 1 percent confidence levels respectively.
- Baseline / Policy change coefficients:
  - Minimum capital adequacy ratio
    - Policy change t-1: -4.32*** (0.00)
    - Policy change t-2: -1.06 (0.12)
  - Marginal reserve requirements on foreign funding
    - Policy change t-1: -1.62** (0.02)
    - Policy change t-2: -0.08 (0.84)
  - Marginal reserve requirements related to credit growth
    - Policy change t-1: -2.37*** (0.00)
    - Policy change t-2: -2.37*** (0.00)
- Tightening vs easing (selected):
  - Policy tightening t-1: Minimum CAR -2.68*** (0.00); Foreign funding -3.49*** (0.00); Credit growth -2.18** (0.03)
  - Policy easing t-1: Minimum CAR -7.38** (0.01); Foreign funding -1.08 (0.19); Credit growth -2.48 (0.13)
- Boom period (selected):
  - Policy change t-1: Minimum CAR -4.52** (0.05); Foreign funding -3.89*** (0.00); Credit growth -2.72*** (0.00)
  - Policy change t-2: Minimum CAR -1.35** (0.02); Foreign funding -2.39** (0.02); Credit growth -1.98*** (0.00)
- Bust period (selected):
  - Policy change t-1: Minimum CAR -4.98*** (0.00); Foreign funding -1.03 (0.21); Credit growth 1.65 (0.22)
  - Policy change t-2: Minimum CAR 1.72** (0.04); Foreign funding 0.50 (0.24); Credit growth -13.16*** (0.00)

### Policy implications and caveats
- Policy implications:
  - Raising the minimum CAR was followed by a significant deceleration in housing prices—supportive evidence for use of the countercyclical capital buffer.
  - Allowing banks to hold less capital (typically following EU entry) was followed by a sizeable acceleration in housing prices—relevant for debates on maximum harmonization under the EU’s new Capital Requirements Directive and banking union.
  - Standard reserve requirements showed no impact on housing price inflation, but marginal reserve requirements targeted at specific excesses (credit growth, foreign funding) had measurable effects.
- Caveats and limitations:
  - Endogeneity may bias coefficients downwards; some tighter policies appearing ineffective may reflect policy timing (policies adopted when accelerating housing prices were anticipated).
  - Effects may materialize outside the two-quarter horizon used in the empirical framework, or they may occur upon announcement rather than implementation.
  - Coding intensity of policy measures relies on subjective judgment and may be imperfect.
- Suggested future research:
  - Explore alternative dependent variables to capture the boom, such as construction activity, real estate transaction volumes, or credit growth once detailed data on currency composition of loans becomes available.

*Source: Authors' calculations.*

### Appendix 2: Description of Macroprudential Policy Measures and the Scoring Rules to

### Appendix 2: Description of Macroprudential Policy Measures and the Scoring Rules to Measure Their Intensity

### CAPITAL MEASURES (EXCEPT RISK_WEIGHTS)
- mincap — Minimum required capital adequacy ratio
  - measure description: minimum required capital adequacy ratio (in percent)
  - series description: minimum capital adequacy ratio
- tgtmincap — (Target) capital adequacy ratio below which restrictions are imposed
  - measure description: capital adequacy ratio below which restrictions are imposed (in percent)
  - note: [in the database but not in the sample]
- cap — Capital eligibility
  - measure description / series description: index with a change of 1 for tightening/easing the capital base calculation
- cgrcap — Minimum required capital adequacy ratio as a function of credit growth (if above a threshold)
  - measure description: minimum required capital adequacy ratio (in percent)
  - series description: annual threshold (in percent); penalty rate (in percent)
  - operationalization: (10/annual threshold) * [minimum required capital-mincap+((penalty rate-100)/100)]
- hhsc — Maximum ratio of household loans to share capital
  - series description / operationalization: index with a change of 1 for implementing/abandoning the measure and increase/decrease of the maximum ratio, change of 0.5 for a change in penalties and for changes in the base
- fcsc — Maximum ratio of fc loans to own funds
  - measure description: maximum ratio of fc loans to own funds
  - operationalization: 3/ratio
- lsctot — Maximum ratio of a targeted type of loans to capital
  - operationalization: hhsc + hhscfc

### RISK WEIGHTS MEASURES
- rwmol — Risk weights / mortgage loans
  - measure description: risk-weight on mortgage loans in lc (in percent)
  - operationalization: (rw mortgage loans - basel risk weights for mortgage loans) / 25
  - special case [for Bulgaria only]: risk-weight on mortgage loans in lc above LTV threshold (in percent) and LTV threshold
    - operationalization: (100 - threshold ltv)/50 * (rw mortgage loans - basel risk weight for mortgage loans)/25
- rwmolfc — Risk weights surcharge/ FC mortgage loans
  - measure description: risk-weight on mortgage loans in fc (in percent)
  - operationalization: (rw mortgage loans in fc - rw for mortgage loans in lc) / 50
  - special case [for Bulgaria only]: risk-weight on mortgage loans in fc above LTV threshold (in percent) and LTV threshold
    - operationalization: (100 - threshold ltv)/50 * (rw mortgage loans - basel risk weight for mortgage loans)/25
- rwmoltot — Risk weights / mortgage loans (combined)
  - measure description: risk-weight on mortgage loans (in percent)
  - operationalization: rwmol + rwmolfc
- rwcons — Risk weights / consumer loans
  - measure description: risk-weight on consumer loans in lc (in percent)
  - operationalization: (rw consumer loans - basel risk weights for consumer loans) / 25
- rwconsfc — Risk weights surcharge/ fc consumer loans
  - measure description: risk-weight on consumer loans in fc (in percent)
  - operationalization: (rw consumer loans in fc - rw for consumer loans in lc) / 50
- rwconstot — Risk weights / consumer loans (combined)
  - operationalization: rwcons + rwconsfc
- rw — Risk weights / mortgage and consumer loans in lc (combined)
  - operationalization: rwmol + rwcons
- rwfc — Risk weights surcharge / mortgage and consumer loans in fc (combined)
  - operationalization: rwmolfc + fwconsfc
- rwtot — Risk weights / mortgage and consumer loans (combined)
  - operationalization: rw + rwfc
- rwcorpfc — Risk weights on fc corporate loans
  - measure description: Risk weight on fc corporate loans (in percent)
  - note: [Not included in regression]
- rwcc — Risk weights on loans above a threshold related to credit growth
  - measure description: risk-weights on loans above threshold (in percent) and annual threshold (in percent)
  - operationalization: (10/annual threshold) * (risk weight-100) / 50
- rwtot2 — Risk weights (total except corporate)
  - operationalization: rw + rwfc + rwcc

### PROVISIONING MEASURES
- gp — Rules for general provisions
  - operationalization: index with a change of 0.5 for tightening/loosening general provisioning rules (index=0 if measure is dropped)
- dp — Rules for specific provisions
  - operationalization: index with a change of 0.5 for tightening/loosening loan loss provisioning or loan classification
- dpfc — FC-loans rules for specific provisions
  - operationalization: index with a change of 0.5 for tightening/loosening fc-loan loss provisioning or fc-loan classification above that for lc-loan classification/provisioning
- dptot — Provisioning rules (combined)
  - operationalization: gp + dp + dpfc

### LIQUIDITY MEASURES
- rr — Reserve requirements rate on lc deposits
  - measure description: minimum reserve requirement ratio on lc demand deposits (in percent)
- rrfc — Reserve requirements rate on fc deposits
  - measure description: minimum reserve requirement ratio on fc demand deposits (in percent)
- mpprr — mpprr
  - operationalization: Reserve requirement rate if rrfc>0: (rr/10 + rrfc/10)/2; if rrfc=0: rr/10
- rrbase — Reserve requirements base
  - operationalization: index with a change of 0.5 for change in reserve base, a change of 0.25 for a change in ratio on any other than demand deposits in domestic and foreign currency (if different from change in ratio for demand deposits)
- rrtot — Reserve requirements (combined)
  - operationalization: rr + rrbase
- lr — Liquidity regulation
  - operationalization: Index with a change of 0.5 for a tightening/easing of the regulation
- fclr — Foreign currency liquidity requirement
  - measure description: ratio of liquid fc assets to fc liabilities (in percent)
  - operationalization: (fclr rate/10) / 4 +/- 0.5 for change in the base
- mrr — Marginal reserve requirements
  - measure description: marginal reserve requirements on foreign funding (in percent)
  - operationalization: (mrr rate/10) / 2
- srr — Special reserve requirements
  - measure description: special reserve requirements on funds raised by domestic bond issuance to nonresidents (in percent)
  - operationalization: (srr rate/10) / 8
- mrrtot — Marginal and special reserve requirements
  - operationalization: mrr + srr
- cgr — Credit growth reserve (banks need to hold low-yield CB bills if their credit growth is above a threshold)
  - series description: annual threshold (in percent); Penalty rate (in percent)
  - operationalization: (10/annual threshold) * (penalty rate/100)
- cc — Marginal reserve requirements on credit growth above a threshold
  - series description: penalty rate is a step function with up to three thresholds.
  - operationalization: [(10/lowest threshold) * (lowest penalty rate/100)] + [(10/second threshold) * (second penalty rate/100)]+ [(10/third threshold) * (third penalty rate/100)]
- cgrr — Credit-growth-related reserve requirements
  - operationalization: cgr + cc

### ELIGIBILITY MEASURES
- ltv — Loan-to-value ceiling
  - measure description: Loan-to-value ceiling (in percent)
  - operationalization: (100 - maximum LTV) / 20 [default=100]
- ltvfc — FC loan-to-value ceiling
  - operationalization: index with a change of 0.5 for tightening/easing the ltv ratio for loans in fc relative to the ltv-ratio for loans in lc
- ltvtot — Loan-to-value ceiling (combined)
  - operationalization: ltv + ltvfc
- dti — Debt-service-to-income ceiling
  - measure description: Debt-service-to-income ceiling (in percent)
  - operationalization: (1 - (maximum DTI / 60)) * 4 [default=60]
- dtifc — FC debt-service-to-income ceiling
  - operationalization: index with a change of 0.5 for tightening/easing the dti ratio for loans in fc relative to the dti-ratio for loans in lc
- dtitot — Debt-service-to-income ceiling (combined)
  - operationalization: dti + dtifc
- elig — All eligibility measures combined
  - operationalization: ltv + ltvfc + dti + dtifc

### OTHER BANK REGULATORY MEASURES
- otherfc — Other quantitative limits on fc-lending as a share of total lending
  - operationalization: index with a change of 0.5 for measures taken to impede fc-lending and a change of 3 - ltvfc - dtifc - dpfc - rwfc - hhscfc for prohibiting fc-lending or fc-mortgage lending

### TAX POLICY AND NON-BANK REGULATORY POLICY MEASURES
- tax — Tax measures regarding real estate / mortgages
  - operationalization: index that changes by 0.5 if taxation of real estate/mortgages (e.g. interest rate subsidies, property tax) changes
- other — Regulatory measures on non-banks
  - operationalization: index that changes by 0.5 for other tightening/easing measures on regulation of non-bank credit institutions
- rest — Tax and non-bank measures combined
  - operationalization: tax + other

### Appendix 3: Macroprudential Policy Measures and Countries Where They Were Used (selection)
- The appendix provides a country-variable matrix indicating use (filled squares) of measures across countries. Example entries:
  - CAPITAL MEASURES: mincap used in multiple countries; tgtmincap used in one country; cap used in multiple countries; cgrcap used in one country; hhsc, fcsc used in selected countries.
  - RISK-WEIGHTS: rwmol, rwmolfc, rwcons, rwconsfc, rwcorpfc, rwcc used in various countries.
  - PROVISIONING: gp, dp, dpfc used in multiple countries.
  - LIQUIDITY: rr, rrfc, rrbase used extensively across countries; lr, fclr, mrr, srr, cgr, cc used in selected countries.
  - ELIGIBILITY: ltv, ltvfc, dti, dtifc used in selected countries.
  - OTHER BANK REGULATORY MEASURES: otherfc used in several countries.

*Appendix 2 and Appendix 3 extracted from the supplied PDF content.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12303.pdf_