## _wp13106

## Source details

**Canonical URL:** [_wp13106](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13106.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13106.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13106.pdf.json)

---

### Introduction and scope
- Real credit in Latin America grew by an average of 9 percent from primo 2004 to ultimo 2011.
- Primo 2004 to 2007: average growth in real banking credit to the private sector was 12.4 percent; during 2008-09 it receded to 2.7 percent; in 2010-11 average real credit growth almost reached a near pre-crisis level of 10.2 percent.
- In 2011, real credit growth in Latin America was faster than both Emerging Europe and Emerging Asia.
- Countries included in the analysis: Argentina, Brazil, Bolivia, Chile, Colombia, Costa Rica, the Dominican Republic, El Salvador, Ecuador, Guatemala, Honduras, Nicaragua, Mexico, Panama, Paraguay, Peru, Uruguay and Venezuela.

### Channels, benefits, and risks of financial deepening
- Potential benefits:
  - Deeper credit markets can bolster capital accumulation and total factor productivity (references in source: Acemoglu and Zillibotti, 1997; Schumpeter, 1912; Levine, 1997; King and Levine, 1993).
- Risks of rapid credit growth:
  - Macroeconomic overheating: domestic demand above potential, inflationary pressure, asset price inflation, eroded competitiveness, widening external imbalances.
  - Asset-side vulnerabilities: overestimation of borrower creditworthiness, stretched loan processing, poor loan portfolio diversification.
  - Liability-side vulnerabilities: excessive reliance on unstable funding to sustain rapid credit growth.

### Data construction and analytic objectives
- Dataset disaggregates credit by market segments, sectors, currencies, maturities, and bank origin.
- Two complementary methods to assess “excessive” credit:
  - Gap methodology (Borio and Lowe, 2002; Borio and Drehmann, 2009): deviations of credit-to-GDP ratios and equity prices from estimated long-run trends to capture asset-side imbalances.
  - Panel cointegration between credit and deposits (inspired by Hahm et al., 2011): long-run relationship between credit and deposits; large deviations may indicate liability-side imbalances (reliance on non-deposit funding).
- A short-run error-correction model is used to assess drivers of deviations from long-run equilibrium.

### Stylized facts on credit expansion (2004–11)
- Sub-periods:
  - Primo 2004–2007: average real credit growth 12.4 percent per year; median 13.7 percent per year.
  - 2008–09: average real credit growth 2.7 percent per year; Brazil showed no slow-down; Paraguay and Peru accelerated.
  - 2010–11: average real credit growth 10.2 percent per year.
- Credit-to-GDP change:
  - Regional credit-to-GDP rose from 31.0 percent of GDP in primo 2004 to 38.5 percent of GDP in ultimo 2011.
  - Paraguay and Brazil: credit-to-GDP increased by 25 and 31 percentage points, respectively, over the period.
- Deposits:
  - Ratio of deposits to GDP rose on average from 32.3 to 38.1 percent of GDP during 2004–11.
  - On average deposits account for 77% of total banking funding in Latin America.
- Cross-country heterogeneity and convergence:
  - Highest real credit growth (average 15-20 percent per year) in Venezuela, Paraguay and Brazil; muted growth (less than 6 percent per year) in Honduras, Bolivia, Guatemala, the Dominican Republic and Uruguay.
  - Highest growth tended to occur in countries with low initial credit-to-GDP ratios.
- Sectoral composition:
  - Households: mortgages and consumption credit grew faster than corporate credit where data available.
    - Mortgage credit average real growth in Brazil was 31 percent over the period.
    - Consumption credit average real growth in Argentina was 22 percent over the period.
  - Corporate: construction credit grew fastest (average growth above 10 percent in most countries with available data during 2005–11).
    - Construction credit especially buoyant in Mexico and Uruguay (more than 20 percent average annual real growth).
    - Construction accounted for approximately 10 percent of corporate loan stock in most countries; around 30 percent in Mexico, Bolivia and Panama.
  - Corporate loans still dominate loan stocks (corporate loan shares above 50 percent in most countries in 2011).
- Bank ownership and funding:
  - Foreign-owned banks increased lending faster than domestic banks from 2004 to 2008, but domestic banks contributed most to aggregate credit growth due to larger market shares.
  - Domestic banks continued to grow lending after 2008 while foreign banks curtailed lending in 2009.
- Currency composition and de-dollarization:
  - National currency lending increased substantially since 2004 in Peru, Bolivia and Uruguay; to a lesser extent in Paraguay, Guatemala and Costa Rica.
  - Foreign currency lending remained less than 10 percent and roughly constant in Chile, Brazil, Mexico, Argentina, Colombia and Honduras.
  - Credit remains fully dollarized in Ecuador, Panama and El Salvador; highly dollarized in Nicaragua (ultimo 2011: 71.6 percent of all assets in the banking system in Nicaragua were in foreign currency).
  - Decline in foreign currency loan share reflects more new lending in national currency and appreciation of Latin American currencies vis-à-vis the dollar since 2004.
- Maturities:
  - Maturity lengthening observed in Brazil, Chile and to a lesser extent in Bolivia during 2004–11.
  - Maturities remained long in Guatemala and El Salvador (less than 30 percent of loans with maturities shorter than one year).
  - Maturities relatively short in Mexico (share of loans with maturities of less than one year stood at 70 percent in 2011).

### Main empirical findings (summary from the paper)
- Growth in banking credit during 2004–11 strongest for household credit in most Latin American countries; consumption and mortgage credit grew faster than corporate credit in most countries; within corporate sector, construction expanded most.
- Banking credit has become increasingly de-dollarized and for some countries there are signs of maturity lengthening.
- Foreign-owned banks posted the fastest lending growth, although domestic banks accounted for most of the growth due to larger market shares.
- In several countries strong credit growth since 2004 opened a “gap” between actual credit-to-GDP and its long-run trend; gaps generally receded somewhat after 2007 but in some countries picked up again after 2010.
- For some countries, large credit-to-GDP gaps were accompanied by positive gaps in real equity prices, especially before 2008.
- Econometric estimates (panel cointegration/long-run credit–deposits relationship) suggest credit-to-GDP in some countries deviates significantly from estimated long-run levels given deposit-to-GDP; there is substantial uncertainty in determining these long-run levels.

### Gap methodology and vulnerability indicators (one-sided HP-filter)
- Gaps constructed using a one-sided Hodrick-Prescott filter with smoothing parameter lambda = 400.000 (as suggested for quarterly data by Drehmann et al. (2011)).
- Thresholds referenced:
  - Borio and Drehmann (2009): thresholds for real equity price gaps and credit-to-GDP gaps of 40-60 percent and 4-6 percentage points, respectively.
  - Present study uses a broader credit-to-GDP threshold range of 2-10 percentage points (based on Drehmann et al., 2011).
- Gap evidence for Latin America:
  - Large and positive credit-to-GDP gaps found in some countries; recent strong credit growth since 2004 made the credit-to-GDP gap positive in most countries.
  - Gaps receded after 2007, then picked up after 2010.
  - End-2011: gaps especially pronounced in Brazil, Paraguay, Venezuela and Colombia.
  - Chile, Honduras, Costa Rica and Nicaragua saw large positive gaps prior to 2008 but up to 2011 those gaps receded.
  - Real equity price gaps largest in 2010-2011 in Argentina, Chile and Peru, although generally below Borio and Drehmann (2009) crisis indicator thresholds.
  - Dataset contains equity indices for 7 out of 18 countries in the sample (data coverage limitation).
- Limitations of gap approach:
  - Absence of property price data for most countries; housing price data available only in a minor sub-sample.
  - HP-filter fails to take initial level of credit into account and does not correct for structural breaks from policy reforms or financial sophistication changes.
  - Revisions to credit-to-GDP gap can be substantial; HP-filter has end point problems.
  - Credit-to-GDP gaps sensitive to choice of starting point and smoothing parameter.
  - Conclusion: gap analysis gives useful warning signs but should be complemented with other methods.

### Panel cointegration approach and long-run benchmark
- Purpose: estimate long-run relationship between credit-to-GDP and financial/economic variables; compare actual credit with model-implied level to gauge “excessive” credit.
- Model specification:
  - Dependent variable: credit-to-GDP (y_it).
  - Explanatory variables include deposits-to-GDP and other financial variables; panel cointegration framework applied with idiosyncratic intercepts.
- Dataset and sample:
  - Unbalanced panel of 107 high and middle income countries.
  - Preferred long-run estimation sample: 1970Q1-2003Q4 (to exclude recent strong credit growth).
  - Separate relationships estimated for Latin America, middle-income, and high-income countries; country fixed effects allowed.
- Variables considered (labels preserved):
  - logpppc, deposgdp, equitygdp, securitygdp, nfagdp, govdebtgdp, mktcapgdp, cfgdp, bcon, cmr_fraser, cmr_wb.
- Unit root and cointegration findings:
  - Panel unit root tests indicate unit roots in: credit-to-GDP, deposits-to-GDP, PPP-adjusted income per capita, and stock market capitalization as percent of GDP.
  - Panel cointegration tests (Pedroni) point to cointegration among credit-to-GDP, deposits-to-GDP, PPP-adjusted income per capita, and stock market capitalization.
- Long-run estimation results:
  - Mean Group Estimator (MGE) preferred; FMOLS and DOLS used for robustness.
  - Deposits-to-GDP enters significantly with expected sign across estimators and specifications.
  - Deposits-to-GDP estimates tend to be lower in Latin America and other middle-income countries versus high-income countries.
  - PPP-adjusted income and stock market capitalization-to-GDP do not enter significantly and are unstable across time.
  - Recursive estimation indicates parameter stability for deposits-to-GDP in the middle-income sample except for 1988–1993 where the estimate temporarily increased.
- Interpretation of deviations:
  - A positive deviation implies credit is high given deposits and the historic credit-deposit relationship; may signal unsustainable reliance on non-deposit funding.
  - Not all non-depository funding is equally unstable; maturity profile matters.

### Comparison of actual credit-to-GDP with fitted long-run levels (model (3), middle-income)
- Method:
  - Fitted credit-to-GDP using actual deposits-to-GDP and the preferred middle-income model (model (3) in Table 3) with country fixed effects.
  - Estimated “credit gaps” computed as fitted values benchmarked against actual credit levels; prediction intervals shown in figures.
- Main finding:
  - "No significant deviations from the estimated long-run level are present for most Latin American countries."
  - Most Latin American countries per ultimo 2011 operate credit-to-GDP levels below, or not significantly above, the estimated long-run level.
  - Brisk credit growth since 2004 in some countries (Brazil, Mexico, Honduras, Bolivia and Argentina) occurred from historically low credit levels given deposits.
- Countries with significant positive deviations ultimo 2011:
  - Chile, Colombia, Costa Rica and Paraguay.
  - Paraguay and Colombia: fast growth in the deviation since 2004.
  - Chile and Costa Rica: steady growth in the deviation since the nineties, accelerating in 2004.
- Role of non-depository funding:
  - Chile, Colombia and Paraguay saw share of issued securities and net foreign liabilities as share of total funding increase from 2004 to 2011 by 3 and 23 percentage points in Chile and Paraguay, respectively.
  - In Colombia the share increased by 24 percentage points from 2000 to 2011.
  - Costa Rica: share of issued securities and net foreign liabilities as share of total funding in 2011 stood at 45 percentage.
  - Policy relevance depends on maturity of non-depository funding used by banks.

### Short-run drivers of credit growth (augmented error-correction model)
- Model:
  - Augmented error correction model for credit growth estimated on middle-income sample for period 1970Q1-2003Q4.
- Key empirical findings (Table 4):
  - ECM(t-1) coefficient significant and negative: ECM(t-1) = -0.03*** [0.00] (column (1)); -0.04*** [0.01] (column (2)).
  - d(Bank deposits / GDP): 0.49*** [0.03].
  - d(Bank equity / GDP): -0.078** [0.030] (column (2)).
  - d(Bank securities / GDP): 0.07 [0.05].
  - d(Bank net foreign assets / GDP): -0.07*** [0.01].
  - d.log(PPP adjusted GDP per capita): 10.82*** [3.96].
  - Sample and diagnostics: N = 61 (column (1)) and 46 (column (2)); T = 127 and 105; Number of observations 5331 and 2158; Adjusted R2 0.25 and 0.23; Durbin-Watson stat 2.00 and 1.97.
- Model predictions vs actual:
  - Figures 16-17 show "credit growth in none of the countries deviates significantly from the model prediction."
  - Prior to 2009, Colombia, Costa Rica and Paraguay displayed positive deviations from model predictions.
  - In 2009 deviations generally receded, but since 2010 positive deviations re-emerged in Colombia, Bolivia, Panama, Brazil and Paraguay.

### Econometric tests and reported statistics (selected exact values preserved)
- Integration orders and test statistics (selected examples from Table 1):
  - credgdp: I(0); test statistics include 4.35 [1.00], 7.98 [1.00], 152.90 [1.00], 146.60 [0.94], -0.51 [0.31], 2.90 [1.00], 218.77 [0.19], 273.75*** [0.00].
  - deposgdp: I(1); test statistics include 3.24 [1.00], 5.36 [1.00], 210.18 [0.30], 181.74 [0.81], -3.79*** [0.00], 1.86 [0.97], -4.46*** [0.00], 333.89*** [0.00].
  - logpppc: I(1); test statistics include -16.30*** [0.00], -1.62* [0.05], 247.57* [0.06], 1025.34*** [0.00], -26.27*** [0.00], 5.70 [1.00], 302.11*** [0.00], 149.69 [0.99].
- Panel cointegration tests (Pedroni) (selected exact values from Table 2):
  - Credgdp and deposgdp: Group rho-stat -4.02*** [0.00]; Group PP-stat -4.46*** [0.00]; Group ADF-stat -5.54*** [0.00].
  - Credgdp, deposgdp and logpppc: Group rho-stat -6.15*** [0.00]; Group PP-stat -6.60*** [0.00]; Group ADF-stat -7.56*** [0.00].
  - Credgdp, deposgdp, logpppc and mktcapgdp: Group rho-stat -0.63 [0.73]; Group PP-stat -3.75*** [0.00]; Group ADF-stat -5.28*** [0.00].
- Mean Group Estimation (selected coefficients from Table 3):
  - Deposgdp coefficients reported as 1.05*** [0.20], 0.55* [0.21], 0.66*** [0.07], 0.54*** [0.09], 0.61*** [0.12], 0.70*** [0.16] across specifications.
  - logpppc coefficients include 5.96 [13.05], -6.42 [5.76], -11.42 [7.56] in some columns.
  - N*T sample sizes preserved exactly: 3386, 1490, 5730, 2099, 2024, 732.

### Key cross-country and regional statistics (preserved exactly)
- R2 is 0.4067 (from scatter of Banking credit to private sector to GDP and PPP-adjusted income per capita, 2004).
- MIC Average = 44.31 (Middle and High Income Countries, Banking Credit to Private Sector, Ultimo 2011).
- Average = 9.4 percent (Real Banking Credit: Growth, primo 2004 - ultimo 2011).
- R2 is 0.27 (Banking Credit Level and Banking Credit Growth, primo 2004 - ultimo 2011).
- R2 is 0.65 (Growth in Credit and Deposits in the Banking Sector, primo 2004 - ultimo 2011).
- Regional summary: "During this period banking credit to the private sector grew by an average of 7 per cent of GDP."

### Policy implications and recommendations
- Empirical evidence shows notable positive deviations of credit-to-GDP from predicted long-run levels in specific countries (Chile, Peru, Paraguay, Costa Rica and Colombia).
- Such deviations can indicate liabilities-side vulnerabilities (increased reliance on non-depository funding) and historically have been indicative of potential asset-side risks.
- Recommended actions:
  - Continued close monitoring and vigilance at the supervisory level.
  - Further analysis of the risk profile and maturity structure of non-depository funding driving deviations.
  - Caution in interpretation: deviations may reflect structural changes in the long-run credit–deposit relationship; maturity composition of funding is critical to risk assessment.

### Appendix: data, definitions and sample (selected exact entries)
- Sample: unbalanced panel of 107 high and middle income countries; specific middle-income and high-income country lists preserved in source.
- Key data sources and definitions (selected):
  - Nominal GDP: IMF World Economic Outlook (WEO), linearly interpolated annual to quarterly.
  - Consumer price index: IFTS line 64; inflation = year-to-year growth.
  - Private banking credit-to-GDP (credgdp): claims of other depository corporations on the non-financial private sector divided by nominal GDP (IFTS line 22d), quarterly.
  - PPP-adjusted income per capita (logpppc): WEO, linearly interpolated annual to quarterly.
  - Deposits-to-GDP (deposgdp): IFTS lines 24 and 25; quarterly; expressed as percent of nominal GDP.
  - Bank equity as percent of GDP (equitygdp): IFTS line 27; quarterly.
  - Bank securities as percent of GDP (securitygdp): IFTS lines 26a and 26s; quarterly.
  - Banks’ net foreign assets as percent of GDP (nfagdp): IFTS lines 21N, 21 and 26C; quarterly.
  - Stock market capitalization as percent of GDP (mktcapgdp): Beck and Demirguc-Kunt (2009), interpolated.
  - Banking sector concentration (bcon): assets of three largest banks as share of all commercial banks (Beck and Demirguc-Kunt, 2009), interpolated.
  - Credit market regulation (cmr_fraser): Fraser Institute index, interpolated.
  - Equity prices: Bloomberg indices used for listed countries (MERVAL, IBOV, IGPA, IGBC, CRSMBCT, MEXBOL, IGBVL, IBVC).

*Source: _wp13106 (IMF working paper excerpts).*

### 1. Determination of Integration Order ..................................................................................

### 1. Determination of Integration Order

### Introduction and scope
- Real credit in Latin America grew by an average of 9 percent from primo 2004 to ultimo 2011.
- From primo 2004 to 2007 average growth in real banking credit to the private sector was 12.4 percent; during 2008-09 it receded to 2.7 percent; in 2010-11 average real credit growth almost reached a near pre-crisis level of 10.2 percent.
- In 2011, real credit growth in Latin America was faster than both Emerging Europe and Emerging Asia (Figure 1 referenced).
- Countries included in the analysis: Argentina, Brazil, Bolivia, Chile, Colombia, Costa Rica, the Dominican Republic, El Salvador, Ecuador, Guatemala, Honduras, Nicaragua, Mexico, Panama, Paraguay, Peru, Uruguay and Venezuela.

### Channels, benefits, and risks of financial deepening
- Potential benefits:
  - Deeper credit markets can bolster capital accumulation and total factor productivity (references: Acemoglu and Zillibotti, 1997; Schumpeter, 1912; Levine, 1997; King and Levine, 1993).
- Risks of rapid credit growth:
  - Macroeconomic overheating: domestic demand above potential, inflationary pressure, asset price inflation, eroded competitiveness, widening external imbalances.
  - Asset-side vulnerabilities: overestimation of borrower creditworthiness, stretched loan processing, poor loan portfolio diversification.
  - Liability-side vulnerabilities: excessive reliance on unstable funding to sustain rapid credit growth.

### Data construction and analytic objectives
- The paper builds a dataset breaking down credit by market segments, sectors, currencies, maturities, and bank origin.
- Two complementary methods to assess “excessive” credit:
  1. Gap methodology (inspired by Borio and Lowe, 2002; Borio and Drehmann, 2009): deviations of credit-to-GDP ratios and equity prices from estimated long-run trends to capture asset-side imbalances.
  2. Panel cointegration between credit and deposits (inspired by Hahm et al., 2011): long-run relationship between credit and deposits; large deviations could indicate liability-side imbalances (reliance on non-deposit funding).
- Additionally, a short-run error-correction model is used to assess drivers of deviations from long-run equilibrium.

### Stylized facts on the credit expansion in Latin America (2004–11)
- Sub-periods:
  - Primo 2004–2007: rapid expansion, average real credit growth 12.4 percent per year; median 13.7 percent per year.
  - 2008–09 (global financial crisis): average real credit growth 2.7 percent per year; Brazil showed no slow-down; Paraguay and Peru accelerated.
  - 2010–11 (post-crisis): average real credit growth 10.2 percent per year.
- Credit-to-GDP change:
  - Regional credit-to-GDP rose from 31.0 percent of GDP in primo 2004 to 38.5 percent of GDP in ultimo 2011.
  - Paraguay and Brazil saw particularly large increases: credit-to-GDP increased by 25 and 31 percentage points, respectively, over the period.
- Deposits:
  - Ratio of deposits to GDP rose on average from 32.3 to 38.1 percent of GDP during 2004–11.
  - On average deposits account for 77% of total banking funding in Latin America.
- Cross-country heterogeneity and convergence:
  - Highest real credit growth (average 15-20 percent per year) in Venezuela, Paraguay and Brazil; muted growth (less than 6 percent per year) in Honduras, Bolivia, Guatemala, the Dominican Republic and Uruguay.
  - Highest growth tended to occur in countries with low initial credit-to-GDP ratios, indicating some convergence.
- Sectoral composition:
  - Households: mortgages and consumption credit grew faster than corporate credit where data are available.
    - Mortgage credit average real growth in Brazil was 31 percent over the period.
    - Consumption credit average real growth in Argentina was 22 percent over the period.
  - Corporate: within corporate sector, construction credit grew fastest (average growth above 10 percent in most countries with available data during 2005–11).
    - Construction credit was especially buoyant in Mexico and Uruguay (more than 20 percent average annual real growth).
    - Construction accounted for approximately 10 percent of corporate loan stock in most countries; around 30 percent in Mexico, Bolivia and Panama.
  - Despite household credit dynamism, corporate loans still dominate loan stocks (corporate loan shares above 50 percent in most countries in 2011).
- Bank ownership and funding structure:
  - Foreign-owned banks increased lending faster than domestic banks from 2004 to 2008, but domestic banks contributed most to aggregate credit growth due to larger market shares.
  - Domestic banks continued to grow lending after 2008 while foreign banks curtailed lending in 2009.
- Currency composition and de-dollarization:
  - National currency lending increased substantially since 2004 in Peru, Bolivia and Uruguay; to a lesser extent in Paraguay, Guatemala and Costa Rica.
  - Foreign currency lending remained less than 10 percent and roughly constant in Chile, Brazil, Mexico, Argentina, Colombia and Honduras.
  - Credit remains fully dollarized in Ecuador, Panama and El Salvador; highly dollarized in Nicaragua (as per ultimo 2011, 71.6 percent of all assets in the banking system in Nicaragua were in foreign currency).
  - Decline in foreign currency loan share reflects more new lending in national currency and appreciation of Latin American currencies vis-à-vis the dollar since 2004.
- Maturities:
  - Signs of maturity lengthening in Brazil, Chile and to a lesser extent in Bolivia during 2004–11.
  - Maturities remained long in Guatemala and El Salvador (less than 30 percent of loans with maturities shorter than one year).
  - Maturities are relatively short in Mexico (share of loans with maturities of less than one year stood at 70 percent in 2011).

### Main empirical findings summarized (from paper’s statement)
- Growth in banking credit during 2004–11 strongest for households credit in most Latin American countries; consumption and mortgage credit grew faster than corporate credit in most countries; within corporate sector, construction expanded most.
- Banking credit has become increasingly de-dollarized and for some countries there are signs of maturity lengthening.
- Foreign-owned banks posted the fastest lending growth, although domestic banks accounted for most of the growth due to larger market shares.
- In several countries strong credit growth since 2004 opened a “gap” between actual credit-to-GDP and its long-run trend; gaps generally receded somewhat after 2007 but in some countries picked up again after 2010.
- For some countries, large credit-to-GDP gaps were accompanied by positive gaps in real equity prices, especially before 2008.
- Econometric estimates (panel cointegration/long-run credit–deposits relationship) suggest credit-to-GDP in some countries deviates significantly from estimated long-run levels given deposit-to-GDP; there is substantial uncertainty in determining these long-run levels.

### Empirical approaches to assessing excess credit (methods overview)
- Gap method:
  - Construct deviations of private credit and equity prices from their estimated long-run trends to capture asset-side imbalances.
- Cointegration method:
  - Estimate long-run relationship between credit and deposits using panel cointegration; deviations from the estimated long-run credit-to-GDP implied by observed deposits may indicate liability-side imbalances (large reliance on non-deposit funding).
- Short-run dynamics:
  - Use an error-correction model to assess drivers of short-run deviations from the long-run equilibrium.

*Source: _wp13106 - 1. Determination of Integration Order (excerpts from the IMF working paper PDF).*

### 21.      The buildup of financial imbalances is often manifested in large and positive “gaps”

### _wp13106 - 21.      The buildup of financial imbalances is often manifested in large and positive “gaps”

### Financial imbalances and gap-based vulnerability indicators
- Motivation: Large and positive gaps in credit-to-GDP and real asset prices vis-à-vis their long-run trend are indicators of financial imbalances; this motivates the Borio and Lowe (2002) approach that constructs a vulnerability indicator using such gaps.
- Gap construction: Gaps are constructed using a one-sided Hodrick-Prescott filter.
  - Smoothing parameter used in the analysis: lambda = 400.000 (this is what Drehmann et al. (2011) suggest for quarterly data).
  - Rationale for high lambda: financial cycles are observed to be approximately four times longer than usual business cycles; Drehmann et al. (2011) show this lambda value outperforms alternatives as a banking crisis predictor.
- Thresholds and historical performance:
  - Borio and Drehmann (2009) suggest thresholds for real equity price gaps and credit-to-GDP gaps of 40-60 percent and 4-6 percentage points, respectively.
  - The present study uses a broader credit-to-GDP threshold range of 2-10 percentage points based on later analysis by Drehmann et al. (2011).
  - With an appropriate choice of thresholds the indicator has historically issued warnings prior to banking crises while avoiding false alarms.

### Evidence for Latin America (gap analysis)
- Credit-to-GDP gaps:
  - Large and positive credit-to-GDP gaps are found in some Latin American countries.
  - Recent strong credit growth since 2004 made the credit-to-GDP gap positive in most countries in the region.
  - Gaps receded somewhat after 2007, then picked up again after 2010.
  - End-2011: gaps especially pronounced in Brazil, Paraguay, Venezuela and Colombia.
  - Chile, Honduras, Costa Rica and Nicaragua saw large positive gaps prior to 2008 but, up to 2011, those gaps receded due to the global financial turmoil starting in 2007.
  - Note: Financial System Stability Assessment for Brazil (IMF, 2012b) finds credit-to-GDP in Brazil receded notably since 2009, cautioning against a strong interpretation of the result.
- Real equity price gaps:
  - Positive real equity price gaps found in some Latin American countries.
  - 2010-2011: gaps largest in Argentina, Chile and Peru, although generally below the Borio and Drehmann (2009) crisis indicator thresholds.
  - Historical co-occurrence: Some countries had simultaneous real equity price gaps and credit-to-GDP gaps in the threshold areas (e.g., Chile in primo 2010; Brazil and Costa Rica up to 2007-08).
  - Argentina up to 2010 had substantial equity price gaps without large credit gaps; credit picked up somewhat in 2011.
  - Data coverage limitation: dataset contains equity indices for 7 out of 18 countries in the sample.

### Data limitations and caveats
- Property prices:
  - Important limitation: absence of property (housing) price data for most countries in the sample; property prices have historically played an important role in banking crises.
  - Housing price data available only in a minor sub-sample of Latin American countries (Cubeddu, Tovar and Tsounta, 2012), preventing inclusion in the present analysis.
  - Cubeddu, Tovar and Tsounta (2012) analyze housing markets in Peru, Mexico, Chile and Colombia using a VECM and find no imminent misalignment in prices, but warn misalignments could emerge if current trends persist.
- HP-filter and gap methodology limitations:
  - HP-filter calculations fail to take the initial level of credit into account (Coudert and Pouvell, 2010): sound financial deepening can entail temporary high credit growth.
  - Technique does not correct for structural breaks from policy reforms or structural changes affecting financial sophistication (Cubeddu et al., 2012).
  - Revisions to the credit-to-GDP gap can be substantial (Edge and Meisenzahl, 2011); HP-filter is known to have end point problems (Mise et al., 2005).
  - Credit-to-GDP gaps sensitive to choice of starting point and smoothing parameter.
  - Conclusion: gap analysis can give useful warning signs but should be complemented with other methods.

### Complementary econometric model for the long-run level of credit
- Purpose: Construct an econometric benchmark estimating a long-run relationship between credit-to-GDP and financial/economic variables; compare actual credit level with model-implied level to gauge whether credit is “excessive”.
- Model specification:
  - Dependent variable: credit-to-GDP (y_it).
  - Explanatory variables: x1,it ... xh,it; model estimated via panel data methods allowing an idiosyncratic intercept.
  - Equation form provided: y_it = β_i,0 + β_1 x_1,it + ... + β_n x_h,it + ε_it
- Estimation approach:
  - Use panel cointegration framework to account for non-stationarity and to estimate both long-run equilibrium and short-run dynamics.
  - Non-stationary variables may affect long-run level if they cointegrate with credit-to-GDP; stationary variables affect short-run growth but not long-run level.

### Data, variables, and estimation results
- Dataset:
  - Unbalanced panel of 107 high and middle income countries.
  - Sample period for preferred long-run estimation: 1970Q1-2003Q4 (chosen to exclude recent strong credit growth).
  - Separate relationships estimated for Latin America, middle-income, and high-income countries; country fixed effects allowed.
- Explanatory variables considered (as labeled in dataset):
  - PPP-adjusted income per capita (logpppc)
  - Deposits as percent of GDP (deposgdp)
  - Bank equity as percent of GDP (equitygdp)
  - Bank securities as percent of GDP (securitygdp)
  - Banks’ net foreign assets as percent of GDP (nfagdp)
  - Government debt as percent of GDP (govdebtgdp)
  - Stock market capitalization as percent of GDP (mktcapgdp)
  - Net capital inflows as percent of GDP (cfgdp)
  - Banking sector concentration (bcon)
  - Credit market regulation (cmr_fraser and cmr_wb)
  - Real interest rate variables (rlr and rdr) were not included because the real interest rate is treated as an outcome variable.
- Unit root and cointegration findings:
  - Panel unit root tests (Levin et al., Im et al., Maddala and Wu, Choi) indicate unit roots in: credit-to-GDP, deposits-to-GDP, PPP-adjusted income per capita, and stock market capitalization as percent of GDP.
  - Additional explanatory variables found stationary.
  - Panel cointegration tests (Pedroni) point to cointegration among credit-to-GDP, deposits-to-GDP, PPP-adjusted income per capita, and stock market capitalization.
- Long-run estimation:
  - Estimators used: Mean Group Estimator (MGE) as preferred; FMOLS (Pedroni, 2001) and DOLS used for robustness.
  - Preferred long-run specification includes credit and deposits.
    - Deposits-to-GDP enters significantly with expected sign across all estimators and specifications.
    - Deposits-to-GDP estimates tend to be lower in Latin America and other middle-income countries versus high-income countries.
  - PPP-adjusted income and stock market capitalization-to-GDP do not enter significantly and are unstable across time.
  - FMOLS and DOLS confirm MGE estimates.
  - Recursive estimation indicates parameter stability for deposits-to-GDP in the middle-income sample except for a period from 1988 to 1993 where the estimate temporarily increased.

### Interpretation of deviations from the estimated long-run level
- Economic interpretation:
  - A positive deviation from the estimated long-run relationship implies credit is high given the level of deposits and the historic credit-deposit relationship; this may signal an unsustainable reliance on non-deposit funding.
  - When credit grows faster than deposits, banks may turn to other, potentially less stable, funding sources to sustain credit growth.
- Empirical associations:
  - Figure 13 (in source) shows some past banking crises occurred on the back of positive and rising deviations from the estimated long-run relationship.
  - This aligns with findings that higher ratios of non-core to core liabilities signal banking crisis risk (Hahm et al., 2011) and that credit-to-deposit ratios are associated with banking crisis probability (IMF, 2011a).
- Cautions:
  - Not all non-depository funding is equally unstable; long-maturity securities can be close substitutes to deposits while short-maturity securities are less so.
  - Formal testing is needed to assess whether deviations from the estimated relationship have good predictive power for future banking crises.

*Source: _wp13106 - 21.      The buildup of financial imbalances is often manifested in large and positive “gaps”*

### 40.       We compare actual credit-to-GDP in Latin America with its fitted long-run

### _wp13106 - 40.       We compare actual credit-to-GDP in Latin America with its fitted long-run

### Comparison of actual credit-to-GDP with fitted long-run levels
- Method:
  - Used the preferred estimated model of equation (1) for middle-income countries (model (3) in Table 3) with country specific fixed effects.
  - Fitted credit-to-GDP using actual deposits-to-GDP values to yield an estimated credit-to-GDP level for each country.
  - Estimated “credit gaps” computed as fitted values benchmarked against actual credit levels.
- Main finding:
  - "No significant deviations from the estimated long-run level are present for most Latin American countries."
  - Figures 14-15 show prediction intervals in addition to the mean prediction; model shows mean reversion historically.
  - Brisk credit growth since 2004 in some countries (Brazil, Mexico, Honduras, Bolivia and Argentina) occurred from historically low credit levels given deposits.
  - "Most Latin American countries are per ultimo 2011 operating credit-to-GDP levels below, or not significantly above, the estimated long-run level."

### Countries with significant deviations and interpretation
- Countries with significant positive deviations ultimo 2011:
  - Chile, Colombia, Costa Rica and Paraguay.
  - Paraguay and Colombia: fast growth in the deviation since 2004.
  - Chile and Costa Rica: steady growth in the deviation since the nineties, accelerating in 2004.
- Caveats on interpretation:
  - Deviations do not necessarily imply excessive credit.
  - Could reflect a structural change in the long-run relationship between credit and deposits (e.g., banking system evolution allowing more credit for a given level of deposits).
  - Parameter on deposits is higher in high-income vis-à-vis middle-income countries (Table 3).
  - Fitting the long-run credit level using parameters estimated for high-income countries (model 1 in Table 3) still shows credit levels above estimated long-run level in 2011—"albeit not significantly for all."

### Role of non-depository funding and balance-sheet shifts
- Two principal explanations for deviations from estimated long-run credit level:
  - (i) Reshuffling of assets towards private sector credit on the asset side.
  - (ii) Increasing reliance on non-depository funding on the liability side.
- Evidence for (ii):
  - Chile, Colombia and Paraguay have seen a shift towards non-depository funding since 2004.
    - IMF International Financial Statistics: share of issued securities and net foreign liabilities as share of total funding increased from 2004 to 2011 by 3 and 23 percentage points in Chile and Paraguay, respectively.
    - In Colombia the share increased by 24 percentage points from 2000 to 2011.
  - Costa Rica: did not see a shift since 2004 but "the share of non-depository funding was relatively high during the period."
    - Share of issued securities and net foreign liabilities as share of total funding in Costa Rica stood at 45 percentage in 2011.
- Policy-relevant point:
  - Whether this development is concerning depends importantly on the maturity of the non-depository funding used by banks.

### Short-run drivers of credit growth (augmented error correction model)
- Model specification:
  - Estimated an augmented error correction model for credit growth (equation (2)) including deviations from the long-run level and non-cointegrating variables.
  - Model estimated on the sample of middle-income countries for the period 1970Q1-2003Q4.
- Key empirical findings (Table 4):
  - Significant and negative estimate for the error-correction term: positive (negative) deviation from the long-run level is associated with a decrease (increase) in credit growth (consistent with cointegration).
  - Significant estimates with expected signs for:
    - PPP-adjusted income per capita,
    - securities issued by the banking sector,
    - banks’ net foreign asset position.
  - Note: "Securities issued by banking sector are borderline significant in Table 4. Stock market capitalization is not included due to a large implied loss of observations."
- Comparison of actual credit growth with short-run model predictions:
  - Figures 16-17 show that "credit growth in none of the countries deviates significantly from the model prediction."
  - Prior to 2009, Colombia, Costa Rica and Paraguay displayed positive deviations from model predictions.
  - In 2009 deviations generally receded, but since 2010 positive deviations re-emerged in Colombia, Bolivia, Panama, Brazil and Paraguay.

### Summary of broader findings and policy implications
- Regional overview 2004-11:
  - "During this period banking credit to the private sector grew by an average of 7 per cent of GDP."
  - Growth fastest in countries with low initial credit-to-GDP ratios.
  - Credit to households was most buoyant; consumption and mortgage credit outpaced corporate credit; within corporate sector construction expanded most.
  - Foreign banks posted the fastest credit growth; domestic banks accounted for most credit growth owing to larger market shares.
  - Credit in the region became increasingly de-dollarized; some countries show signs of maturity lengthening.
- Assessment of risk vs. deepening:
  - Latin American credit markets remain relatively shallow compared with countries of comparable income, yet rapid credit growth can pose macrofinancial risks.
- Methods used to assess vulnerabilities:
  - One-sided HP-filter to construct credit-to-GDP and real asset price gaps.
  - Panel cointegration between credit and deposits to detect deviations potentially indicative of liability-side vulnerabilities and rising reliance on non-depositary funding.
- Empirical conclusions:
  - There are large and positive credit-to-GDP gaps in a number of countries; gaps generally receded after 2007 but picked up since 2010.
  - Recently, gaps largest in Brazil, Paraguay, Venezuela and Colombia.
  - In 2008-10, large credit-to-GDP gaps were accompanied by substantial real equity price gaps in Brazil, Chile and Costa Rica.
- Policy recommendation:
  - "Historically, such deviations have been indicative of potential risks on the asset side in the financial sector, why continued close monitoring and vigilance at the supervisory level is warranted."
  - "Deviations warrant caution and more analysis" — particularly assessing the risk profile and maturity structure of non-depository funding driving deviations.

*Source: IMF working paper content unit _wp13106 - 40.*

### 52.      In some countries the credit-to-GDP level displays a positive deviation from the

### _wp13106 - 52.      In some countries the credit-to-GDP level displays a positive deviation from the

### Key finding
- "In some countries the credit-to-GDP level displays a positive deviation from the estimated long-run level."
- Countries with large and significant positive deviations: Chile, Peru, Paraguay, Costa Rica and Colombia.
- Interpretation given: "This reflects that the level of credit is high given the estimated long-run model and the level of deposits in these economies."
- Policy implication noted: "This could be indicative of vulnerabilities on the liability side in the financial sector. Thus, our results warrant monitoring and further analysis of the risk profile of the non-depository funding in Latin America."

### Cross-country and regional statistics (preserved exactly)
- R2 is 0.4067 (from scatter of Banking credit to private sector to GDP and PPP-adjusted income per capita, 2004).
- MIC Average = 44.31 (Middle and High Income Countries, Banking Credit to Private Sector, Ultimo 2011).
- Average = 9.4 percent (Real Banking Credit: Growth, primo 2004 - ultimo 2011).
- R2 is 0.27 (Banking Credit Level and Banking Credit Growth, primo 2004 - ultimo 2011).
- R2 is 0.65 (Growth in Credit and Deposits in the Banking Sector, primo 2004 - ultimo 2011).

### Composition and sectoral patterns (figures and notes)
- Real Banking Credit by Market Segment (2005=100) shown for Mortgages, Other consumer, Corporate (simple average of Argentina, Brazil, Chile (Ex. 2011), Mexico, Bolivia, Guatemala (Ex. 2011), El Salvador, Costa Rica, Honduras and Panama).
- Loans by Banking Sector by Currency: charts for Peru, Bolivia, Uruguay, Paraguay, Guatemala, Costa Rica; series shown as Domestic / Foreign shares (percent of total loans, ultimo stock).
- Loans by Banking Sector by Maturity: country charts for Brazil, Chile, Bolivia, Mexico, Guatemala, El Salvador (percent of total loans, ultimo stock; maturity buckets "Less than a year" / "Longer than a year" or "Less than 3 years" / "Longer than 3 years" as labeled).

### Credit gaps and crises (modeling outcomes)
- "Estimated excessive credit denotes the difference between actual credit and the prediction of the model (model 2 for Middle Income Countries). Dotted lines depicts the 95% prediction interval. Shaded periods indicate periods of banking crisis."
- Figures display estimated "credit gaps" time series for multiple countries and groups (Emerging Europe, Asia and Latin American countries including Latvia, Estonia, Lithuania, Hungary, Bolivia, Ecuador, Colombia, Dominican Republic, Philippines, Thailand, Indonesia).
- Figure 14: "Estimated 'credit gaps' ... Dotted lines depicts 95% prediction interval."
- Figure 15: Estimated "credit gaps" for Argentina, Brazil, Chile, Colombia, Mexico, Peru, Venezuela, Uruguay, Bolivia, Dominican Republic, Guatemala, Ecuador, Nicaragua, El Salvador, Paraguay, Costa Rica, Honduras, Panama (y-axis labeled "Percent of GDP").

### Actual minus predicted credit growth (model diagnostics)
- Figures 16 and 17: "Actual less predicted credit growth" series (percent of GDP) for Brazil, Chile, Colombia, Mexico, Peru, Venezuela, Uruguay, Bolivia, Dominican Republic, Guatemala, Ecuador, El Salvador, Paraguay, Costa Rica, Honduras, Panama covering periods around 1992–2008 (as labeled).
- Note: "Figures show the difference between actual credit and the prediction of the model for credit growth (Collumn 2, Table 6). Dotted lines depicts 95% prediction interval."

### Econometric tests and results (preserved table entries and significance)
- Table 1: Determination of Integration Order — reported integration orders and test statistics for a range of variables. Examples (values preserved exactly):
  - Credit-to-GDP (credgdp): I(0). Test statistics include 4.35 [1.00], 7.98 [1.00], 152.90 [1.00], 146.60 [0.94], -0.51 [0.31], 2.90 [1.00], 218.77 [0.19], 273.75*** [0.00].
  - Deposits-to-GDP (deposgdp): I(1). Test statistics include 3.24 [1.00], 5.36 [1.00], 210.18 [0.30], 181.74 [0.81], -3.79*** [0.00], 1.86 [0.97], -4.46*** [0.00], 333.89*** [0.00].
  - PPP-adjusted income per capita (logpppc): I(1). Test statistics include -16.30*** [0.00], -1.62* [0.05], 247.57* [0.06], 1025.34*** [0.00], -26.27*** [0.00], 5.70 [1.00], 302.11*** [0.00], 149.69 [0.99].
  - Additional variables reported with I(0) or I(1) designation and a variety of test statistics (Bank equity as percent of GDP (equitygdp): I(0); Stockmarket capitalization as percent of GDP (mktcapgdp): I(1); etc.).
- Table 2: Panel Cointegration Tests (Pedroni tests). Examples (values preserved exactly):
  - Credgdp and deposgdp: Group rho-stat -4.02*** [0.00]; Group PP-stat -4.46*** [0.00]; Group ADF-stat -5.54*** [0.00].
  - Credgdp, deposgdp and logpppc: Group rho-stat -6.15*** [0.00]; Group PP-stat -6.60*** [0.00]; Group ADF-stat -7.56*** [0.00].
  - Credgdp, deposgdp, logpppc and mktcapgdp: Group rho-stat -0.63 [0.73]; Group PP-stat -3.75*** [0.00]; Group ADF-stat -5.28*** [0.00].

### Regression and dynamics (preserved coefficients and diagnostics)
- Table 3: Results from Mean Group Estimation, dependent variable Credgdp. Selected preserved coefficients:
  - Deposits-to-GDP (deposgdp): coefficients reported as 1.05*** [0.20], 0.55* [0.21], 0.66*** [0.07], 0.54*** [0.09], 0.61*** [0.12], 0.70*** [0.16] across specifications (columns labeled High income / Middle income / Latin America).
  - PPP-adjusted income per capita (logpppc): coefficients reported as 5.96 [13.05], -6.42 [5.76], -11.42 [7.56] in some columns.
  - Constant terms and N*T sample sizes reported (N*T values preserved exactly: 3386, 1490, 5730, 2099, 2024, 732).
- Table 4: Results from Pooled OLS Estimation, dependent variable d.Credgdp, Sample: Middle income countries. Key preserved entries:
  - Constant: -0.09** [0.08] (in column (1)).
  - ECM(t-1): -0.03*** [0.00] (column (1)); -0.04*** [0.01] (column (2)).
  - d(Bank deposits / GDP): 0.49*** [0.03].
  - d(Bank equity / GDP): -0.078** [0.030] (column (2)).
  - d(Bank securities / GDP): 0.07 [0.05].
  - d(Bank net foreign assets / GDP): -0.07*** [0.01].
  - d.log(PPP adjusted GDP per capita): 10.82*** [3.96].
  - N = 61 (column (1)) and 46 (column (2)); T = 127 and 105; Number of observations 5331 and 2158; Adjusted R2 0.25 and 0.23; Durbin-Watson stat 2.00 and 1.97.
- Notes preserved: "Lags of d.log(credit-to-GDP) and d.log(deposits-to-GDP) are included but not reported." Significance notation: *, ** and *** denote significance on 10, 5 and 1 percent level, respectively. Standard errors reported in parenthesis.

### Summary implications and monitoring recommendation
- Empirical evidence shows notable positive deviations of credit-to-GDP from predicted long-run levels in specific Latin American countries (Chile, Peru, Paraguay, Costa Rica, Colombia).
- Relationship between growth in credit and deposits is strong (R2 is 0.65 for growth in Credit and Deposits, primo 2004 - ultimo 2011), but the identified positive credit deviations relative to deposits suggest potential liabilities-side vulnerabilities.
- Authors explicitly recommend monitoring and "further analysis of the risk profile of the non-depository funding in Latin America."

*Source: IFTS, MBRF2, WEO and excerpts from the supplied IMF working paper content.*

### APPENDIX 1: DATA SET

### APPENDIX 1: DATA SET

### A. Sample of countries
- Paragraph 54: Our sample includes the following middle and high income countries. Categorization follows the World Bank.
- Middle income countries (as listed in source):
  - Thailand, Zambia, Turkey, Timor-Leste, Tonga, Syrian Arab Republic, Sudan, Suriname, Swaziland, South Africa, Russian Federation, Samoa, São Tomé & Príncipe, Republic of Serbia, Papua New Guinea, Nigeria, Pakistan, Moldova, Mongolia, Morocco, Maldives, Mauritius, FYR Macedonia, Kazakhstan, Republic of Kosovo, Iraq, Fiji, Georgia, Ghana, Grenada, Cape Verde, Bhutan, Albania, Algeria, Antigua and Barbuda, Armenia, Rep. of Azerbaijan, P.R. China: Mainland, India, Indonesia, Malaysia, Philippines, Romania, Latvia, Lithuania, Bulgaria.
  - (Note: the source lists Latin America countries and repeats some regional groupings; the exact country strings above are preserved as they appear in the source.)
- High income countries (as listed in source):
  - Republic of Korea, Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, United Kingdom, United States, Kingdom of Bahrain, Barbados, Equatorial Guinea, P.R. China: Hong Kong, Kuwait, Malta, Oman, Qatar, Saudi Arabia, Singapore, United Arab Emirates.
  - Latin America repeated listing preserved in source: Argentina, Brazil, Chile, Colombia, Mexico, Peru, Rep. Bol. Venezuela, Uruguay, Bolivia, the Dominican Republic, Guatemala, Ecuador, Nicaragua, El Salvador, Paraguay, Costa Rica, Honduras, Panama.

### B. Data definitions and sources
- Paragraph 55:
  - Nominal GDP: from IMF World Economic Outlook (WEO). Data is linearly interpolated from annual to quarterly frequency.
- Paragraph 56:
  - Consumer price index: taken from line 64 in IMF International Financial Statistics (IFTS). Frequency is quarterly. Inflation (infl) is created by taking year-to-year growth in this variable.
- Paragraph 57:
  - Private banking credit-to-GDP (credgdp): defined as claims of other depository corporations on the non-financial private sector divided by nominal GDP. Data is taken from line 22d in the IFTS. Frequency is quarterly.
  - Notes: Due to a large methodological change at euro adoption series for euro area countries have been extrapolated by means of the growth rate in the claims of other depository corporations on the non-financial private sector from IMF Monetary and Financial Statistics.
  - Variable corrections for breaks (countries and quarters exactly as in source): Chile (1978Q3), Colombia (1990Q4), Mexico (1977Q4), Uruguay (1998Q4, 1975Q4 and 2001Q4), Bolivia (2001Q4), the Dominican Republic (2001Q4), Guatemala (2002Q3), Nicaragua (1988Q1), Honduras (1997Q4), Croatia (1994Q2), Hungaria (1987Q4), Poland (19854), Romania (19964), Malaysia (1992Q1), Phillipines (1986Q4), Samoa (1971Q4), Serbia (2001Q4), Nigeria (1992Q4), Moldova (1998q1), Mongolia (1992Q2), Morocco (1990q1), Mauritius (2003Q2), Ghana (1973Q4), Cape Verde (1993Q3), Algeria (1992Q3), Armenia (1992Q4), Austria (1998Q4), Belgium (1992Q2, 1998Q4), Canada (2001Q4), Denmark (1987Q4, 2000Q3), Finland (1999Q1), France (1978Q1, 1998Q3), Germany (1999Q1), Greece (1978Q4, 2001Q1), Ireland (1999Q1), Italy (1999Q1), Japan (2001Q4), Luxemburg (19991, 1977Q1), Netherlands (1998Q1), New Zealand (1988Q3), Portugal (1999Q1), Slovak Republic (1997Q1), Spain (1999Q1), Sweden (2001Q1), Switzerland (1974Q4, 1982Q3), United Kingdom (1986Q3), Bahrain (1995Q1), Barbados (2001Q4) and China,P.R.: Hong Kong (1993Q4).
- Paragraph 58:
  - Real interest rates (rlr and rdr): lending rate from line 60p in IFTS (bank rate meeting short- and medium-term financing needs of the private sector); deposit rate from line 60l in IFTS (rate usually offered to resident customers for demand, time or saving accounts). Frequency is quarterly.
  - Construction: real interest rates are constructed using the ex post inflation.
  - Formula (as specified in source, paragraph 25 note): rir=100*(1+(1+i)/(1+pi)-1). Here rir, i and pi are the real interest rate, the nominal interest rate and actual inflation, respectively.
- Paragraph 59:
  - PPP-adjusted income per capita (logpppc): taken from WEO. Data is linearly interpolated from annual to quarterly frequency.
- Paragraph 60:
  - Deposits-to-GDP (deposgdp): sum of “transferable deposits included in broad money” and “other deposits included in broad money” for other depository corporations (line 24 and 25 in IFTS). Frequency is quarterly. Expressed as percentage of nominal GDP.
  - Variable corrections for breaks (countries and quarters exactly as in source): Mexico (1977Q4), Guatemala (2001Q4), Nicaragua (1988Q1), Costa Rica (1997Q4), the Dominican Republic (2001Q4), Honduras (1997Q4), Hungaria (1987Q4), Poland (1985Q4), Romania (1989Q4), Malaysia (1992Q1), Phillipines (1986Q4), Turkey (1978Q1), Suriname (2002Q2), Serbia (2001Q4), Nigeria (2007Q4), Mongolia (1992Q2), Macedonia (2001Q4), Bhutan (1985Q4), Armenia (1994Q1), Austria (1998Q4), Belgium (1992Q4, 1998Q4), Canada (2001Q4), Denmark (1987Q4), France (1978Q1, 1998Q3), Germany (1999Q1), Greece (2001Q1), Ireland (1995Q1), Israel (1981Q1), Italy (1999Q1), Japan (2001Q4), Luxembourg (1998Q3), Netherlands (1998Q1), New Zealand (1988Q3), Portugal (1979Q4), Spain (1999Q1) and Barbados (2001Q4).
- Paragraph 61:
  - Bank equity as percent of GDP (equitygdp): “shares and other equity” for other depository corporations (line 27 in IFTS) at a quarterly frequency. Divided by GDP.
- Paragraph 62:
  - Bank securities as percent of GDP (securitygdp): sum of “securities other than shares included in broad money” (line 26a) and “securities other than shares excluded from broad money” (line 26s) at quarterly frequency. Expressed in percentage of GDP.
- Paragraph 63:
  - Banks’ net foreign assets as percent of GDP (nfagdp): “net foreign assets” (line 21N in IFTS) merged with “foreign assets” (line 21) minus “foreign liabilities” (line 26C) for other depository corporations. Frequency is quarterly and variable is expressed as share of GDP.
- Paragraph 64:
  - Government debt as percent of GDP (govdebtgdp): gross government debt from WEO. Linearly interpolated from annual to quarterly frequency, expressed in percentage of GDP.
- Paragraph 65:
  - Stock market capitalization as percent of GDP (mktcapgdp): value of listed shares as percent of GDP from the database by Beck and Demirguc-Kunt (2009). Linearly interpolated from annual to quarterly frequency.
- Paragraph 66:
  - Banking sector concentration (bcon): assets of the country’s three largest banks as the share of all commercial banks (Beck and Demirguc-Kunt, 2009). Linearly interpolated from annual to quarterly frequency.
- Paragraph 67:
  - Credit market regulation (cmr_fraser): index for credit market regulation compiled by the Fraser institute, where a higher value implies less regulation. Linearly interpolated from annual to quarterly frequency.
- Paragraph 68:
  - Equity prices: Bloomberg indices used — MERVAL (Argentina), IBOV (Brazil), IGPA (Chile), IGBC (Colombia), CRSMBCT (Costa Rica), MEXBOL (Mexico), IGBVL (Peru) and IBVC (Venezuela).

*Source: APPENDIX 1: DATA SET (IMF working paper appendix text as provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13106.pdf_