## _wp12193

## Source details

**Canonical URL:** [_wp12193](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12193.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12193.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12193.pdf.json)

---

### Introduction and context
- Easy financing conditions and favorable terms of trade have fueled credit and domestic demand in much of Latin America for almost a decade, with a short interruption during the 2008–09 global crisis.
- Mortgage sector expansion reflects both financial deepening needed to address housing deficits and factors such as legal reforms and government subsidies.
- ECLAC estimates the housing deficit between 42 and 51 million units.
- The Ministerial Commission on Housing and Urbanization for Latin America and the Caribbean found that only 60 percent of families in the region had adequate housing.
- The report highlights tension between necessary financial deepening and the region‘s historical risk of credit booms gone wrong.

### Mortgage credit growth and drivers
- Real mortgage credit in the six more financially integrated economies (Brazil, Chile, Colombia, Mexico, Peru, Uruguay) has grown by an annual average of 14 percent since 2003.
- In Brazil the inflation-adjusted stock of mortgage loans has increased seven-fold since 2003 (from a low base).
- Mortgage credit in Latin America stands on average at about 7 percent of GDP (22 percent of total credit).
- Comparable levels in other emerging economies: 18 percent of GDP and 27 percent of total credit.
- Growth drivers:
  - Favorable external conditions and sustained economic growth.
  - Legal reforms: bankruptcy reforms in Brazil (2005) and Mexico (2007).
  - Overhauled credit registries.
  - Government housing credit programs and subsidies (example: Brazil‘s state-owned banks providing ¾ of all housing credit; Brazil‘s housing program “Minha Casa, Minha Vida”).
  - Development of domestic local-currency bond markets and lengthening of term structure of yield curves (in some cases reaching 20–30 years).

### Construction activity
- Share of construction in GDP has grown sharply in the more financially integrated countries and reached levels above those of emerging Asia, though below pre-Lehman peaks in emerging Europe.
- Brazil:
  - Number of construction companies and projects grew strongly since 2008 (SECOVI, 2012).
  - Employment in construction grew by 50 percent during 2008–11 and was responsible for about 14 percent of the total 6.5 million jobs created.
- Colombia:
  - Building permits almost doubled between 2009 and 2011 and remain at elevated levels despite some recent correction.

### House prices: trends and comparisons
- Average real home-price for the six financially integrated economies rose at an annual rate of roughly 6 percent between 2005 and 2011.
- Regional comparisons:
  - Average price increase in Latin America is well below that observed in emerging Europe pre-Lehman, and somewhat above that in emerging Asia.
  - In level terms, house prices in the region remain low in international comparisons.
- Global Property Guide indicators (average price of 100-square meter apartments in center of most important city):
  - House Price to Per Capita GDP: Chile 12.8; Uruguay 13.4; Canada 16.2; Mexico 19.3; Peru 23.3; US 27.9; Brazil 29.7; Colombia 30.1; EM Europe 39.3; EM Asia 68.7.
  - Yearly Apartment Rent to Per Capita GDP: Canada 0.7; Uruguay 1.2; Chile 1.4; US 1.6; Mexico 2.0; Brazil 2.0; Peru 2.3; Colombia 2.7; EM Europe 2.8; EM Asia 4.8.
  - Price/Rent Ratio (years): Chile 11 yrs; Mexico 11 yrs; Peru 12 yrs; Uruguay 13 yrs; Colombia 13 yrs; Brazil 18 yrs; EM Europe 21 yrs; US 21 yrs; Canada 27 yrs; EM Asia 27 yrs.
- Caveat: these measures often reflect higher-income segments in metropolitan areas and may not reflect national-level or distributional conditions.

### Risks, historical experience, and data concerns
- Given the region‘s long history of credit booms gone wrong, concerns exist about potential buildup of financial sector excesses even if current credit indicators appear manageable.
- Experience shows credit-driven bubbles build slowly but can sour quickly; small problematic segments can become systemic (U.S. subprime example).
- Colombia‘s late 1990s mortgage crisis is cited as a cautionary tale.
- Serious data limitations on housing prices, housing stock and flows, construction activity, housing-specific financial soundness indicators, and household balance-sheet data constrain assessment.

### Methodology for detecting booms and misalignments
- Mortgage boom identification:
  - A credit expansion is defined as a boom when the level of credit exceeds the underlying trend estimated using end-adjusted rolling Hodrick-Prescott (HP) filters by a threshold equal to 1.5 times the standard deviation of the trend.
  - Deviation from long-run trend in log real mortgage credit for country i at date t is M_it and its standard deviation is σ(M_i). A credit boom occurs when M_it ≥ 1.5σ(M_i).
  - HP smoothing parameter set at 129,600 for monthly data. Filter rolled month-over-month with the seed set in January 2006 (adjusted for countries with data limitations).
  - Monthly sample over 2000–2011 includes 9 countries: Brazil, Colombia, China, Hong Kong SAR, Indonesia, Korea, Malaysia, Peru, and Thailand.
- House price misalignment assessment:
  - Cointegration relationships estimated to uncover long-run relations between real house prices and fundamentals; variables expressed in levels.
  - VECM specification: cointegrating I(1) variables include real house price Pt, real interest rate Rt (in levels), real per capita GDP Yt, and population St; all in logs except Rt.

### Key econometric and empirical findings
- General:
  - Little evidence of excessive growth in mortgage credit in much of the region overall.
  - Mortgage growth is above past long-run trends in a few economies (notably Brazil).
  - Evidence of minor house price misalignments in a few countries (Peru and to a lesser extent Colombia).
  - Cannot conclude that the real estate market is experiencing a bubble based on available indicators.
- Specific econometric results (Chile, Colombia, Mexico, Peru):
  - Important determinants of equilibrium prices: real GDP per capita, population, and real lending rate.
  - Long-run elasticities reported:
    - Population (Chile): 0.55 [2.52] — a 1 percent increase in population raises equilibrium house price by about ½ percent in Chile.
    - Real GDP per capita: Chile 0.97 [14.96]; Colombia 2.16 [13.08]; Mexico 0.30 [4.61]; Peru 0.99 [160.15].
    - Real mortgage rate: Chile -0.61 [-1.52]; Mexico -0.89 [-10.89].
  - Error-correction terms:
    - Chile: -0.88 [-4.04]
    - Colombia: -0.30 [-4.09]
    - Mexico: -0.89 [4.04]
    - Peru: -0.98 [-28.31]
    - Interpretation: negative error-correction terms indicate correction back to long-run equilibrium; pace about one-third of disequilibrium per quarter in Colombia and almost instantaneous in remaining sample countries.
- Cointegration tests (Trace and Max-Eigenvalue):
  - Trace test indicates 1 cointegrating equation at the 0.05 level for the country panels shown.
  - Max-eigenvalue test indicates 1 cointegrating equation at the 0.05 level for the country panels shown.
- Scatter/fit statistic reported: y = 0.2155x -1.8216; R² = 0.2499 (context: relationship between real mortgage growth rate and real house price growth rate, 2000–11).
- Real mortgage credit index series: last data observation is 2011:Q3; index base 2004 = 100 for emerging economies series.

### Country-level dynamics and notable cases
- Brazil:
  - Mortgage expansion particularly large; maximum deviation from trend in recent years is particularly large.
  - Rapid expansion coincides with “Minha Casa, Minha Vida,” indicating structural policy drivers.
  - Brazil omitted from VECM due to short price time series (house price data only start in 2008).
- Peru:
  - Most dynamic market; house prices deviate by six percent.
  - Price-to-rent ratio is low regionally and internationally, suggesting modest overvaluation signs.
- Chile:
  - House prices found to be in line with fundamentals.
- Colombia:
  - Mid-1990s bubble: prices estimated to have exceeded trend by as much as 60 percent in the mid-1990s, returning to fundamentals by mid-2000s.

### Colombia’s Mortgage Crisis of the Late 1990s (chronology and fiscal impact)
- Origins:
  - Early 1990s financial deregulation, rapid expansion of bank assets, undesirable liability structure changes, easy external financing, massive capital inflows, asset price rise.
  - Bank credit as share of GDP doubled between 1991 and 1997.
  - Weak regulatory and supervisory frameworks and information blind-spots.
- Crisis dynamics:
  - External shock and domestic slowing in 1995 led housing prices to fall.
  - Subsequent Asian and Russian crises, domestic political problems, sudden stop; interest rates reached historical highs by 1998; households unable to service mortgages; properties seized; NPLs skyrocketed; mortgage-specialized banks became illiquid or insolvent.
- Government intervention in 1999: financial institutions nationalized, closed, or recapitalized; Colombia suffered its first recession since 1933.
- Fiscal cost: total fiscal cost of the crisis (including effects of judicial rulings) exceeded 15 percent of GDP (FOGAFIN, 2009).

### Household debt burden and financial balance sheets
- Mitigating factors for financial stability:
  - Relatively low exposure of banks in mortgage market.
  - Small share of nonperforming mortgages.
  - Strength of household balance sheets.
- Quantitative points:
  - Mortgages account for less than 20 percent of banks‘ total credit in many countries in the region.
  - Debt burden remains at manageable levels, supported by near-record low unemployment rates, though indicators have been on the rise, particularly for low-income households.
- Caveats:
  - Many loans are new (defaults rare early in loan life); loans extended to households with underdeveloped credit/payment histories.
  - Housing corrections have important social ramifications given labor intensity of construction and longer unemployment spells for relatively unskilled workers.

### Data limitations and measurement caveats
- Only Brazil, Chile, Colombia, Mexico, Peru, and Uruguay publish housing price data in the region; time series are often short and coverage limited to large metropolitan areas.
- Many indices do not distinguish between new and existing homes or between commercial and residential real estate.
- Little information on housing stock and flows, construction inputs, land prices, and housing transactions.
- Trends may vary depending on whether prices are measured in local or foreign currency (example: Uruguay).
- Table summary of housing price series availability (excerpt):
  - Brazil 2010 (monthly, metropolitan)
  - Chile 2004 (quarterly, national)
  - Colombia 1997 (quarterly, metropolitan)
  - Mexico 2005 (quarterly, national)
  - Peru 1998 (quarterly, metropolitan)
  - Uruguay 2000 (monthly, metropolitan)
- Short time series and limited coverage constrain analysis; Brazil omitted from VECM for short series.

### Policy conclusions and recommendations
- Main assessment:
  - Few signs of misalignments in mortgage and real estate sector overall; house prices in most markets near equilibrium levels.
  - If left unattended, current credit and house price growth could result in misalignments down the road.
- Recommended actions to close information gaps and strengthen oversight:
  - Develop more comprehensive and timely information on home prices (national coverage, distinguish new vs. existing homes, commercial vs. residential, measured in terms of repeat sales) and on construction activity (housing stock and flows, employment in sector, price of construction inputs, including land prices).
  - Strengthen infrastructure to maintain current information on housing-specific financial soundness indicators and household balance-sheet data (critical for assessing credit risks using leverage ratios and affordability indicators).
  - Improve property rights, credit registries, and underwriting standards of mortgage loan originators and brokers; standards should account for property value (based on sound independent appraisals) and borrower credit worthiness (via credit registries) with proper verification.
  - Strengthen creditor rights and consumer financial literacy programs, particularly as credit access expands to lower-income households.
  - Monitor financial implications of rapid expansion of new forms of housing financing (e.g., trusts) and mortgage securitization.
- Macroprudential measures:
  - Consider targeted macroprudential measures if housing sector vigor is sustained, similar to measures adopted in Asian countries.
  - Use of loan-to-value (LTV) and debt-to-income (DTI) limits could be useful; limits should be lower for emerging markets given deeper recessions and more severe financial downturns relative to advanced economies.
  - Implementation challenges in Latin America:
    - LTV limits may be less effective where a large share of mortgage origination is accounted for by the unregulated, informal financial sector.
    - DTI limits may not properly capture incomes derived from informal sectors.
    - Many Latin American countries would need to construct LTV or DTI ratios in the first place.
- Additional institutional priorities:
  - Improve property rights: more than a third of Latin American homeowners may have tenure that falls short of full legal title; informal housing estimated between 25 and 50 percent of urban housing stock (UN Habitat, 2011).
  - Reduce costs and duration and increase effectiveness of enforcement and foreclosure processes.
  - Collect and use household financial or expenditure surveys where available (Brazil, Chile, Colombia) to assess micro-data on households’ real and financial assets, credit access, and debt concentration by income segment.

*Source: _wp12193 - IMF authors' calculations and analysis as presented in the content unit.*

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

### _wp12193 - References .............................................................................................................

### Introduction and context
- Easy financing conditions and favorable terms of trade have fueled credit and domestic demand in much of Latin America for almost a decade, with a short interruption during the 2008–09 global crisis.
- The mortgage sector expansion reflects both financial deepening needed to address housing deficits and factors such as legal reforms and government subsidies.
- The Economic Commission for Latin America and the Caribbean (ECLAC) estimates the housing deficit between 42 and 51 million units.
- The Ministerial Commission on Housing and Urbanization for Latin America and the Caribbean found that only 60 percent of families in the region had adequate housing.
- The report highlights the tension between necessary financial deepening and the historical risk of credit booms gone wrong in the region.

### Mortgage credit growth and house prices
- Mortgage credit expansion has been particularly impressive in the region over the past decade.
- The increase in mortgage credit in many countries has been accompanied by an increase in home prices (Figure 1).
- The average real home-price for the more financially integrated economies of the region (Brazil, Chile, Colombia, Mexico, Peru, and Uruguay) rose at an annual rate of roughly 6 percent between 2005 and 2011.

### Risks, historical experience, and data concerns
- Given the region‘s long history of credit booms gone wrong, there are valid concerns about the potential buildup of financial sector excesses, even if current credit indicators appear manageable.
- Experience shows that credit-driven bubbles build slowly but can sour quickly.
- Colombia‘s experience in the late 1990s is cited as a useful reminder of the systemic effects that even a small mortgage sector can have on the economy (Box 1: Colombia‘s Mortgage Crisis of the Late 1990s: A Cautionary Tale).
- The recent U.S. housing crisis is cited to emphasize that problems in a small market (e.g., the subprime sector) can become systemic, especially in new markets with significant data gaps.

### Data and methodological elements listed in the source
- Tables listed:
  - 1. Data Availability on Select Financial and Housing Sector Indicators
  - 2. Determinants of Equilibrium House Prices: Chile, Colombia, Mexico, and Peru
- Figures listed:
  - 1. Latin America: Mortgage and Housing Market Developments
  - 2. Actual and Estimated Real House Prices
  - 3. House Price Over/(under)valuation
- Box listed:
  - 1. Colombia‘s Mortgage Crisis of the Late 1990s: A Cautionary Tale
- Appendices listed:
  - I. Data Definitions and Sources
  - II. Methodologies for Estimating Credit Booms and House Price Bubbles
  - III. Cointegration Tests

*Source: _wp12193 - References .............................................................................................................*

### 2007. For country-specific data on housing deficit see UN Habitat (2011, Chapter 3).

### _wp12193 - 2007. For country-specific data on housing deficit see UN Habitat (2011, Chapter 3).

### Overview
- Paper documents developments in housing and mortgage markets in Latin America (focus on Brazil, Chile, Colombia, Mexico, Peru, Uruguay; comparisons with emerging Asia and emerging Europe).
- Main analytical aims:
  - Assess whether growth in mortgage credit is excessive compared to long-term trend.
  - Assess whether trends in house prices reflect economic fundamentals.
  - Assess extent to which vulnerabilities may be building in household and banking sectors.
- Key caveat: serious data limitations on housing prices, housing stock and flows, construction activity, housing-specific financial soundness indicators, and household balance-sheet data.

### Mortgage credit: developments and magnitudes
- Real mortgage credit in the more financially integrated economies (Brazil, Chile, Colombia, Mexico, Peru, Uruguay) has grown by an annual average of 14 percent since 2003.
- In Brazil, the inflation-adjusted stock of mortgage loans has increased seven-fold since 2003 (from a low base).
- Mortgage credit in Latin America stands on average at about 7 percent of GDP (22 percent of total credit).
- Comparable levels in other emerging economies are reported as 18 percent of GDP and 27 percent of total credit.
- Growth drivers include favorable external conditions, sustained economic growth, stronger fundamentals, legal reforms (bankruptcy reforms in Brazil (2005) and Mexico (2007)), overhauled credit registries, government housing credit programs and subsidies (e.g., Brazil‘s state-owned banks providing ¾ of all housing credit; Brazil‘s housing program “Minha Casa, Minha Vida”).
- Development of domestic local-currency bond markets and lengthening of the term structure of yield curves (in some cases reaching 20–30 years).

### Construction activity
- Share of construction in GDP has grown sharply in the more financially integrated countries and reached levels above those of emerging Asia, though below pre-Lehman peaks in emerging Europe.
- Brazil: number of construction companies and projects grew strongly since 2008 (SECOVI, 2012); employment in construction grew by 50 percent during 2008–11 and was responsible for about 14 percent of the total 6.5 million jobs created.
- Colombia: building permits almost doubled between 2009 and 2011 and remain at elevated levels despite some recent correction.

### House prices
- The average home-price in the six more financially integrated economies rose by an annual real rate of 6 percent between 2005 and 2011.
- Regional comparisons:
  - Average price increase in Latin America is well below that observed in emerging Europe in the run up to the Lehman crisis.
  - It is somewhat above that registered in emerging Asia.
  - In level terms, house prices in the region remain low in international comparisons.
- Global Property Guide (2012) indicators (average price of 100-square meter apartments in center of most important city):
  - House Price to Per Capita GDP: Chile 12.8; Uruguay 13.4; Canada 16.2; Mexico 19.3; Peru 23.3; US 27.9; Brazil 29.7; Colombia 30.1; EM Europe 39.3; EM Asia 68.7.
  - Yearly Apartment Rent to Per Capita GDP: Canada 0.7; Uruguay 1.2; Chile 1.4; US 1.6; Mexico 2.0; Brazil 2.0; Peru 2.3; Colombia 2.7; EM Europe 2.8; EM Asia 4.8.
  - Price/Rent Ratio (years): Chile 11 yrs; Mexico 11 yrs; Peru 12 yrs; Uruguay 13 yrs; Colombia 13 yrs; Brazil 18 yrs; EM Europe 21 yrs; US 21 yrs; Canada 27 yrs; EM Asia 27 yrs.
- Caveat: these measures often reflect higher-income segments in metropolitan areas and may not reflect national-level or distributional conditions.

### Data limitations and availability
- Only Brazil, Chile, Colombia, Mexico, Peru, and Uruguay publish housing price data in the region; time series are often short and coverage limited to large metropolitan areas.
- Many indices do not distinguish between new and existing homes or between commercial and residential real estate.
- Little information is available on housing stock and flows, construction activity details (employment, price of inputs, land prices), and housing transactions.
- Trends may vary depending on whether prices are measured in local or foreign currency (example: Uruguay).
- Table summary (excerpted): housing price series available since:
  - Brazil 2010 (monthly, metropolitan)
  - Chile 2004 (quarterly, national)
  - Colombia 1997 (quarterly, metropolitan)
  - Mexico 2005 (quarterly, national)
  - Peru 1998 (quarterly, metropolitan)
  - Uruguay 2000 (monthly, metropolitan)

### Methodology for detecting booms and misalignments
- Mortgage boom identification:
  - A credit expansion is defined as a boom when the level of credit exceeds the underlying trend estimated using end-adjusted rolling Hodrick-Prescott (HP) filters by a threshold equal to 1.5 times the standard deviation of the trend (as in Mendoza and Terrones (2008) and Gourinchas, Valdés, and Landerretche (2001)).
  - The deviation from long-run trend in log real mortgage credit for country i at date t is M_it and its standard deviation is σ(M_i). A credit boom occurs when M_it ≥ 1.5σ(M_i).
  - HP smoothing parameter set at 129,600 for monthly data. The filter is rolled month-over-month with the seed set in January 2006 (adjusted for countries with data limitations).
  - Sample for monthly analysis over 2000–2011 includes 9 countries: Brazil, Colombia, China, Hong Kong SAR, Indonesia, Korea, Malaysia, Peru, and Thailand.
- House price misalignment assessment:
  - Cointegration relationships are estimated to uncover long-run relations between real house prices and fundamentals; variables expressed in levels (methodology similar to Tumbarello and Wang (2010) and Tsounta (2009)).
  - Data description and sources for house prices are in Appendix II (not reproduced here).

### Key analytical findings
- Mortgage credit growth:
  - Little evidence of excessive growth in mortgage credit in much of the region overall.
  - Mortgage growth is above levels dictated by past long-run trends in a few economies (notably Brazil).
  - In Brazil the maximum deviation from trend in recent years is particularly large; other countries show small deviations.
  - Rapid expansion in Brazil coincides with the introduction and expansion of the “Minha Casa, Minha Vida” housing program, indicating structural policy drivers that may not represent speculative excess.
- House prices:
  - Evidence of minor house price misalignments in a few countries (Peru and to a lesser extent Colombia).
  - Cannot conclude that the real estate market is experiencing a bubble based on available indicators.
- Vulnerabilities and balance sheets:
  - Nonperforming mortgage loans are still relatively low.
  - Mortgages represent a small share of banks‘ assets.
  - Household indebtedness indicators (where available) suggest financial burden remains at manageable levels, though rising—especially for low-income households.
- Overall caution:
  - Excesses and vulnerabilities may not be captured by contemporaneous indicators and may only become visible if current rates of mortgage credit and house price growth are sustained for an extended period.
  - A proper assessment is hindered by limited and weak data; addressing these data gaps is an urgent priority.

### Specific empirical/statistical details (figures and model info preserved)
- Scatter/fit shown: y = 0.2155x -1.8216; R² = 0.2499 (context: relationship between real mortgage growth rate and real house price growth rate, 2000–11).
- Real mortgage credit index series: last data observation is 2011:Q3; index base noted as 2004 = 100 for emerging economies series.
- Regional averages and indices:
  - Real mortgage credit in the six financially integrated economies: simple average of Brazil, Chile, Colombia, Mexico, Peru, Uruguay (real terms).
  - Emerging Asia sample: China, Hong Kong SAR, Indonesia, Korea, Malaysia, Thailand.
  - Emerging Europe sample: Bulgaria, Hungary, Lithuania, Russia, Slovak Republic, Slovenia.
  - Central America & Caribbean: Costa Rica, Dominican Republic, El Salvador, Guatemala, Panama.

_Italic: Source document content provided in the input._

### Appendix III for matrix detail), suggesting that the gap between actual prices and estimated

### _wp12193 - Appendix III for matrix detail), suggesting that the gap between actual prices and estimated

### Model specification and estimation approach
- Estimated model: vector error correction model (VECM) with two to four cointegrating I(1) variables depending on country: real house price, Pt, real interest rate, Rt, real per capita GDP, Yt, and population, St. All variables in logarithms except real interest rate, which is in levels.
- Error-correction equation for Pt (in its entirety):
  - ΔPt = α[Pt−1 − β0 − β1 Rt−1 − β2 Yt−1 − β3 St−1] + λ1 ΔPt-1 + λ2 ΔYt-1 + λ3 ΔRt-1 + λ4 ΔSt-1 + εy,t, t = 1,...,T
  - Parameter constraints and notation: 0 < α < 1; β‘s and λ‘s are estimated parameters; Δ is the difference operator.
- Estimation details for credit surge episodes:
  - Estimates based on end-adjusted rolling Hodrick-Prescott filters estimated using monthly data since 2000 when available.
  - Smoothing parameter, λ, set at 129600, and the filter was rolled month-over-month with the seed set in January 2006 (seed adjusted where sample size was an issue).
  - Threshold defined as 1.5 times the standard deviation of the level relative to trend.

### Key econometric findings and long-run elasticities (Chile, Colombia, Mexico, Peru)
- Variables found to be important determinants of equilibrium prices in all four countries: real GDP per capita, population, and real lending rate (proxy for real mortgage rate).
- Statistical significance and sign: coefficients statistically significant and of the expected sign; relative importance differs across countries.
- Stylized model implications (long-run elasticities and effects):
  - Population: In the long run, a 1 percent increase in population will raise the equilibrium house price by about ½ percent in Chile.
  - Real GDP per capita: elasticity ranges from 0.3 (Mexico) to around 2 (Colombia).
  - Real mortgage rate: long-run negative impact; a 1 percentage point increase in the interest rate will lead to a fall in house prices of 0.6 percent in Chile and 0.8 percent in Mexico.
    - Literature context: dispersion ranges from -0.9 for the Netherlands (Hofman, 2005) to -6 for the United Kingdom (Hunt, 2005).
  - Error-correction term: negative sign for all four countries, indicating correction back to long-run equilibrium; pace of correction about one-third of the disequilibrium per quarter in Colombia and almost instantaneously in the remaining sample countries.

### Table 2: Determinants of Equilibrium House Prices (summary of reported coefficients)
- Sample periods:
  - Chile: 2004Q3–2011Q1
  - Colombia: 2001Q2–2011Q1
  - Mexico: 2005Q3–2011Q2
  - Peru: 1998Q3–2011Q2
- Reported coefficients (with t statistics in brackets):
  - Total population (Chile): 0.55 [2.52]
  - Real GDP per capita:
    - Chile: 0.97 [14.96]
    - Colombia: 2.16 [13.08]
    - Mexico: 0.30 [4.61]
    - Peru: 0.99 [160.15]
  - Real mortgage rate:
    - Chile: -0.61 [-1.52]
    - Mexico: -0.89 [-10.89]
  - Error correction:
    - Chile: -0.88 [-4.04]
    - Colombia: -0.30 [-4.09]
    - Mexico: -0.89 [4.04]
    - Peru: -0.98 [-28.31]
- Notes: Quarterly data; all variables in the VECM are in log levels with exception of the real mortgage interest rate.

### Country-specific price dynamics and deviations from fundamentals
- General conclusions:
  - House price dynamics in Latin America can mostly be explained by the basic stylized model of economic fundamentals.
  - Prices remain aligned with fundamentals and within a one-standard deviation range from the trend in most markets.
- Specific observations:
  - Peru: most dynamic market; house prices deviate by six percent. Peru’s price-to-rent ratio is low in regional and international comparisons, suggesting that signs of overvaluation are modest.
  - Chile: house prices found to be in line with fundamentals.
  - Some recent run-up in house prices appears to reflect catching up from undervalued prices in the mid 2000s.
  - Colombia mid-1990s bubble: house prices estimated to have exceeded trend by as much as 60 percent in the mid-1990s, only returning to fundamentals by mid-2000s.

### Colombia’s Mortgage Crisis of the Late 1990s (Box 1): chronology and fiscal impact
- Origins: early 1990s financial deregulation, rapid expansion of bank assets, undesirable liability structure changes, easy external financing, massive capital inflows, asset price rise, credit boom (bank credit as share of GDP doubled between 1991 and 1997).
- Contributing weaknesses: weak regulatory and supervisory frameworks; internal risk models poorly suited to assess borrowers’ capacity; collateral frequently overvalued; lack of increased capital requirements or loan-loss provisions; important information blind-spots.
- Crisis dynamics:
  - External shock and domestic slowing in 1995 led housing prices to fall.
  - Asian and Russian crises, domestic political problems, sudden stop, interest rates reached historical highs by 1998, households unable to service mortgages, properties seized, NPLs skyrocketed, mortgage-specialized banks became illiquid or insolvent.
- Government intervention in 1999: financial institutions nationalized, closed, or recapitalized; Colombia suffered its first recession since 1933.
- Fiscal cost: total fiscal cost of the crisis (including effects of judicial rulings) exceeded 15 percent of GDP (FOGAFIN, 2009).

### Household debt burden and financial balance sheets
- Mitigating factors for financial stability despite fast mortgage credit and house price growth:
  - Relatively low exposure of banks in mortgage market.
  - Small share of nonperforming mortgages.
  - Strength of household balance sheets.
- Quantitative points:
  - Mortgages account for less than 20 percent of banks‘ total credit in many countries in the region.
- Household indebtedness indicators:
  - Debt burden remains at manageable levels, supported by near-record low unemployment rates, though indicators have been on the rise, particularly for low-income households.
  - Data gaps limit comprehensive assessment of household leverage.
- Nonperforming mortgages:
  - Share relatively low, helped by strong household income growth and record low unemployment.
  - Caveats: many loans are new (defaults rare early in loan life); loans extended to households with underdeveloped credit/payment histories; housing corrections have important social ramifications due to labor intensity of construction and longer unemployment spells for relatively unskilled workers.

### Limitations and caveats of the analysis
- Considerable uncertainty about the right technique to model equilibrium house prices and potential biases from omitted variables or unstable relationships (examples: failing to capture macroeconomic volatility or inward migration).
- Short time series and limited coverage of price data constrain analysis.
- Brazil omitted from the VECM analysis given short price time series (house price data only start in 2008); applying same technique to limited data suggests some overvaluation, consistent with relatively high house price-to-rent ratio.
- Intrinsic limitations noted in literature (Gallin (2003), Gurkaynak (2005), Kluyev (2008), Girouard et al. (2006), Tsounta (2009), Allen et al. (2006), Terrones (2004)).

### Policy conclusions and recommendations
- Main assessment:
  - Few signs of misalignments in mortgage and real estate sector overall; house prices in most markets near equilibrium levels.
  - If left unattended, current credit and house price growth could result in misalignments down the road.
- Recommended actions to close information gaps and strengthen oversight:
  - Develop more comprehensive and timely information on home prices (national coverage, distinguish new vs. existing homes, commercial vs. residential real estate, measured in terms of repeat sales) and construction activity (housing stock and flows, employment in sector, price of construction inputs, including land prices).
  - Strengthen infrastructure to maintain current information on housing-specific financial soundness indicators and household balance-sheet data (critical for assessing credit risks using leverage ratios and affordability indicators).
  - Improve property rights, credit registries, and underwriting standards of mortgage loan originators and brokers; standards should account for property value (based on sound independent appraisals) and borrower credit worthiness (via credit registries) with proper verification.
  - Strengthen creditor rights and programs to increase consumer financial literacy, particularly as credit access expands to lower-income households.
  - Monitor financial implications of rapid expansion of new forms of housing financing (e.g., trusts) and mortgage securitization.
- Macroprudential measures:
  - Consider targeted macroprudential measures if housing sector vigor is sustained, similar to measures adopted in Asian countries.
  - Use of loan-to-value (LTV) and debt-to-income (DTI) limits could be useful to dampen credit and house price growth; limits should be lower for emerging markets given deeper recessions and more severe financial downturns relative to advanced economies (Claessens, Kose, and Terrones, 2010, and 2011).
  - Implementation challenges in Latin America:
    - LTV limits may be less effective where a large share of mortgage origination is accounted for by the unregulated, informal financial sector.
    - DTI limits may not properly capture incomes derived from informal sectors.
    - Many Latin American countries would need to construct LTV or DTI ratios in the first place.
- Additional institutional priorities:
  - Improve property rights: more than a third of Latin American homeowners may have tenure that falls short of full legal title; informal housing estimated between 25 and 50 percent of urban housing stock (UN Habitat, 2011).
  - Reduce costs, duration, and increase effectiveness of enforcement and foreclosure processes to mitigate borrower default risks and improve collateral security.
  - Collect and use household financial or expenditure surveys where available (Brazil, Chile, Colombia) to assess micro-data on households’ real and financial assets, credit access, and debt concentration by income segment.

*Source: IMF authors' calculations and analysis as presented in the content unit.*

### References

### References

### Key cited works and topics
- Abraham, Jesse M., and Patric H. Hendershott, 1996, “Bubbles in Metropolitan Housing Markets,” Journal of Housing Research, Vol. 7, No. 2, pp. 191–207.
- Borio, Claudio, and Philip Lowe, 2002, “Asset Prices, Financial and Monetary Stability: Exploring the Nexus,” BIS Working Paper 114.
- Case, Karl. E., and Robert J. Shiller, 2003, “Is There a Bubble in the Housing Market?” Brookings Papers on Economic Activity, Vol. 2, pp. 299–342.
- Claessens, Stijn, M. Ayhan Kose, and Marco E. Terrones, 2011, “How Do Business and Financial Cycles Interact?” IMF Working Paper No. 11/88.
- Dell’Ariccia, Giovanni, Deniz Igan, Luc Laeven, and Hui Tong, with Bas Bakker and Jerome Vandenbussche, 2012, “Policies for Macrofinancial Stability: How to Deal with Credit Booms,” IMF Staff Discussion Note No. 12/06.
- Himmelberg, Charles, Christopher Mayer and Todd Sinai, 2005, “Assessing High House Prices: Bubbles, Fundamentals and Misperceptions,” NBER Working Paper 11643.
- Mendoza, Enrique G., and Marco E. Terrones, 2008, “An Anatomy of Credit Booms: Evidence from Macro Aggregates and Micro Data,” NBER Working Paper No. 1404.
- OECD, 2005, “Recent House Price Developments: The Role of Fundamentals,” OECD Economic Outlook No. 78.
- Persson, Mattias, 2009, “Household Indebtedness in Sweden and Implications for Financial Stability-The Use of Household-Level Data,” BIS Papers No. 46.
- Terrones, Marco, 2004, “What Explains the Recent Run-Up in House Prices?” IMF World Economic Outlook, Chapter II, Box 2.1, September.

### Geographic and thematic coverage in cited literature
- Regional and country studies: Canada, Chile, Colombia, Mexico, Peru, United Kingdom, United States, Netherlands, Ireland, Australia, Sweden, Brazil, Latin America and the Caribbean, Central and Eastern Europe and the Balkans.
- Core themes: house price dynamics; housing finance and subsidies; credit booms and lending growth; early warning indicators for asset-price boom/bust cycles; valuation methods (fundamentals vs. rent-price/rental yield approaches); securitization and housing finance mechanisms.

---

### Appendix I. Data Definitions and Sources

### Data sources and definitions used in the study
- Lending rates—a proxy for mortgage rate—are from IMF’s International Financial Statistics.
- Population and real GDP data are from Haver Analytics.
- Mortgage data are from national sources.
- House price data sources and definitions (country-specific):
  - Brazil: Fundaçâo Instituto de Pesquisas Econômicas. FipeZap Price Index of Real Estate, Sao Paulo.
  - Colombia: Banco de la República. HPI - For the three major Colombian cities using the repeat sales methodology.
  - Chile: Central Bank of Chile based on information from the Internal Revenue Service. Real House Price Index.
  - Mexico: Sociedad Hipotecaria Federal. The house price index (HPI) at the national level.
  - Peru: Central Bank. Average Price per Square meter for La Molina, Miraflores, San Borja, San Isidro y Surco apartments.
- Sources: National authorities and authors' calculations.

---

### Appendix II. Methodologies for Estimating Credit Booms and House Price Bubbles

### Credit Booms — methodological approaches described
- Standard approach: estimate the standard deviation of a country‘s credit fluctuations around trend (typically using the Hodrick-Prescott filter) and identify a credit boom when that deviation exceeds a certain benchmark (see Borio and Lowe (2002, 2004); Kaminsky and Reinhart (1999); Mendoza and Terrones (2008); Gourinchas, Valdés, and Landerretche (2001)).
- Alternative approach: use credit-to-GDP growth rate and evaluate booms based on a Bry-Boschan algorithm (Bunda and Ca‘Zorzi, 2009).
- Equilibrium-deviation approach: estimate the level of credit explained by economic fundamentals and identify booms/busts as deviations from the estimated equilibrium level (Égert et al., 2006).
- Chosen approach in this paper: estimate deviations around a long-term trend given its general appeal, data limitations, and simplicity in interpretation.

### Housing Price Bubbles — methodological approaches described
- Standard (equilibrium) approach: check compatibility between observed prices and fundamentals (interest rates, income, supply-side variables) using an estimated equilibrium model; commonly formalized by positing a cointegrating relationship between house prices and fundamentals, then estimating an error-correction specification (references include Poterba 1984, 1991; Abraham and Hendershott 1996; Malpezzi 1999; Capozza et al. 2002; Meen 2002; Mankiw and Weil 1989; Roche 2001; Terrones 2004; Tsounta 2009).
  - Note: Error correction approach is highlighted as simple enough for application to many markets with reasonable data and grounded in theory (Malpezzi, 1999).
- Finance-based (rent-price) approach: compare costs and benefits of renting relative to buying; deviations of the current rental price ratio from its long-run average are taken as indications of over- or undervaluation (Case and Shiller 1989; Clayton 1996; OECD 2005; Himmelberg et al. 2005; ECB 2006).
  - Drawbacks: supply and demand factors such as income or demographics are not explicitly modeled; limited guidance on adjustment paths if prices deviate from fundamentals.
- Approach adopted in this study: follow the standard cointegration approach estimating a cointegrating relationship of house prices and economic fundamentals due to general appeal and data limitations in obtaining comprehensive rent series.
- Footnote: Iossifov, Čihák, and Shanghavi (2008) provide a comprehensive survey of determinants of house prices.
- Footnote: There are just a few studies that analyze the evolution of house prices in Latin America, given data limitations (for example, Parrado, Cox and Fuenzalida, 2009 for Chile).

---

### Appendix III. Cointegration Tests

### Unrestricted Cointegration Rank Test (Trace) — results (country panels)
- Chile
  - None: Eigenvalue 0.7115496; Trace Statistic 60.62428; 0.05 Critical Value 47.85613; Prob. 0.002
  - At most 1: Eigenvalue 0.410888; Trace Statistic 27.05704; 0.05 Critical Value 29.79707; Prob. 0.1002
  - At most 2: Eigenvalue 0.376502; Trace Statistic 12.77031; 0.05 Critical Value 15.49471; Prob. 0.1235
  - At most 3: Eigenvalue 0.000565; Trace Statistic 0.015263; 0.05 Critical Value 3.841466; Prob. 0.9015
- Colombia
  - None: Eigenvalue 0.317795; Trace Statistic 17.15441; 0.05 Critical Value 15.49471; Prob. 0.0279
  - At most 1: Eigenvalue 0.045374; Trace Statistic 1.857411; 0.05 Critical Value 3.841466; Prob. 0.1729
- Mexico
  - None: Eigenvalue 0.671747; Trace Statistic 38.55229; 0.05 Critical Value 29.79707; Prob. 0.0038
  - At most 1: Eigenvalue 0.329514; Trace Statistic 11.81701; 0.05 Critical Value 15.49471; Prob. 0.1659
  - At most 2: Eigenvalue 0.088463; Trace Statistic 2.222965; 0.05 Critical Value 3.841466; Prob. 0.136
- Peru
  - None: Eigenvalue 0.994599; Trace Statistic 271.5917; 0.05 Critical Value 15.49471; Prob. 0.0001
  - At most 1: Eigenvalue 0.001784; Trace Statistic 0.092851; 0.05 Critical Value 3.841466; Prob. 0.7606
- Trace test conclusion: Trace test indicates 1 cointegrating eqn(s) at the 0.05 level.
  - * denotes rejection of the hypothesis at the 0.05 level.
  - **MacKinnon-Haug-Michelis (1999) p-values.

### Unrestricted Cointegration Rank Test (Max-Eigenvalue) — results (country panels)
- Chile
  - None: Eigenvalue 0.711549; Max-Eigen Statistic 33.56724; 0.05 Critical Value 27.58434; Prob. 0.0075
  - At most 1: Eigenvalue 0.410888; Max-Eigen Statistic 14.28673; 0.05 Critical Value 21.13162; Prob. 0.342
  - At most 2: Eigenvalue 0.376502; Max-Eigen Statistic 12.75504; 0.05 Critical Value 14.2646; Prob. 0.0854
  - At most 3: Eigenvalue 0.000565; Max-Eigen Statistic 0.015263; 0.05 Critical Value 3.841466; Prob. 0.9015
- Colombia
  - None: Eigenvalue 0.317795; Max-Eigen Statistic 15.29699; 0.05 Critical Value 14.2646; Prob. 0.0342
  - At most 1: Eigenvalue 0.045374; Max-Eigen Statistic 1.857411; 0.05 Critical Value 3.841466; Prob. 0.1729
- Mexico
  - None: Eigenvalue 0.671747; Max-Eigen Statistic 26.73528; 0.05 Critical Value 21.13162; Prob. 0.0073
  - At most 1: Eigenvalue 0.329514; Max-Eigen Statistic 9.594045; 0.05 Critical Value 14.2646; Prob. 0.24
  - At most 2: Eigenvalue 0.088463; Max-Eigen Statistic 2.222965; 0.05 Critical Value 3.841466; Prob. 0.136
- Peru
  - None: Eigenvalue 0.994599; Max-Eigen Statistic 271.4989; 0.05 Critical Value 14.2646; Prob. 0.0001
  - At most 1: Eigenvalue 0.001784; Max-Eigen Statistic 0.092851; 0.05 Critical Value 3.841466; Prob. 0.7606
- Max-eigenvalue test conclusion: Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level.
  - * denotes rejection of the hypothesis at the 0.05 level.
  - **MacKinnon-Haug-Michelis (1999) p-values.

*Source: _wp12193 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12193.pdf_