## _wp0953

## Source details

**Canonical URL:** [_wp0953](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0953.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0953.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0953.pdf.json)

---

### Introduction and context
- Morocco’s economic reform program since the turn of the century:
  - substantially strengthened macroeconomic conditions,
  - accelerated the pace of non-agricultural growth, and
  - improved the soundness of the financial sector.
- External and liquidity developments:
  - Reserves increased more than fivefold to reach US$ 24 billion at end-2007.
  - Banking system’s structural liquidity position increased from about 10 percent of banks’ total assets in 2000 to 24 percent in 2006.
  - Required reserves raised from 10 percent in 2000 to 16.5 percent in 2006, then lowered to 15 percent at end-2007.
  - Banks held excess reserves during most of 2000–07; BAM regularly used its 24-hour deposit facility to mop up excess liquidity.
- Credit dynamics overview:
  - Ratio of credit to GDP remained basically flat between 2000 and 2006.
  - Starting in 2006, credit to the private sector accelerated: growing by almost 30 percent year-on-year in 2007; its ratio to non-agricultural GDP jumped by more than 10 percentage points between 2006 and 2007, to reach almost 80 percent.
  - The three largest banks accounted for roughly 60 percent of outstanding credit to the private sector in 2007.
- Motivation and research objectives:
  - Test hypothesis that excess liquidity with strong credit demand implies credit rationing using a disequilibrium framework.
  - Investigate whether the post-2006 credit acceleration is supply- or demand-driven.
  - Assess role of sharp asset price increases, notably real estate, in credit expansion.

### Overview of credit market developments (2000–2007)
- Asset price and mortgage developments:
  - Casablanca Stock Exchange (MASI) main index increased threefold since 2000.
  - Market capitalization rose from 29 percent of GDP to 103 percent between 2000 and 2007.
  - Evidence of strong real estate price increases, reaching almost 20 percent in 2007 in some cities.
  - Real-estate sector index on the Casablanca Stock Exchange used as proxy for real estate prices; index comprised of only 3 companies, representing 16 percent of total stock market capitalization at end-2007.
  - Variable-rate mortgages now represent close to half of outstanding mortgages (versus a little over a quarter four years earlier).
  - Average mortgage maturities lengthened.
  - Loan-to-value ratios above 90 percent almost doubled in the last two years, reaching 76 percent in 2007.
- Policy and institutional drivers of lending:
  - Central bank (BAM) measures since 2005: standardizing minimum data requirements for credit applications, helping banks develop internal risk-rating systems, monitoring risk management, and setting up a credit bureau.
  - Treasury strengthened public-guarantee funds (SME guarantees and home-ownership guarantees).
- Credit composition:
  - Credit growth post-2006 was fairly broad-based; fastest growth in consumer and real estate credit.
  - Working capital, investment, real estate, and consumption credits all contributed to overall credit expansion.

### Estimation strategy (disequilibrium framework)
- Model structure:
  - Observed credit Ct = min(Cd_t, Cs_t) with three equations: demand, supply, and short-side rule.
  - Demand and supply specified as linear functions with normally distributed errors; estimation by maximum likelihood (Marquardt iterative approach).
- Sample and data:
  - Quarterly data for period 1997Q3 to 2007Q4.
  - Parsimonious specification due to small sample and data scarcity.
  - Variables (most in logs except interest rates);
    - Demand explanatory variables: real interest rate (one-year Treasury bill rate; robustness: money market rate), economic activity (energy production index), real estate prices (real-estate sector stock index); robustness check using overall stock market index.
    - Supply explanatory variables: real interest rate, real estate prices, proxy for banks’ lending capacity (total deposits minus required reserves).
- Identification and expectations:
  - Lending rates expected negatively correlated with credit demand.
  - Energy production index chosen as proxy for economic activity and to aid identification.
  - Real estate expected to have positive effect on both demand (wealth/collateral) and supply (collateral, reduced adverse selection).
  - Interest-rate sign in supply ambiguous (higher rates may raise bank profitability or worsen adverse selection).
- Stationarity and estimation in levels:
  - All variables have at least one unit root (Appendix I); model estimated in levels with cointegration checks (Appendix II).
- Proxy and data caveats:
  - One-year Treasury bill or money market rates are imperfect proxies for lending rates (partial and lagged pass-through; marginal lending rates not available quarterly).
  - Non-performing loan ratio not available quarterly; credit-to-government excluded due to small correlation with private credit.

### Main estimation results (1997Q3–2007Q4)
- General quantitative findings:
  - Interest rate variable is never significant (both Treasury bill and money market rate proxies).
  - Elasticity of credit supplied to banks’ lending capacity: between 0.7 and 0.9 across specifications.
  - Elasticity of credit demanded to economic activity (energy index): about 0.5 in the most plausible specifications.
  - Supply elasticity to asset prices: around 0.08 (positive) whether using real-estate sector index or overall stock market index.
  - Demand function sensitive to real estate index: real-estate index strongly significant in both supply and demand in specifications (1) and (2); overall stock market index not significant in demand.
  - Model fit declines when asset price variables are excluded (specification (4)).
- Selected coefficient estimates (preserving coefficients and standard errors as in Table 1):
  - Supply Equation coefficients (Coef.(Std error)):
    - C: 2.42**(0.80) ; 2.14*(1.24) ; 0.60(0.43) ; 1.97**(0.43)
    - Real interest rate (Treasury bill): 0.25(0.88) ; ...... ; 0.60(0.66) ; -0.59(1.04)
    - Real interest rate (Money market): ...... ; 0.38(0.46) ; ............ ;
    - Log(real lending capacity): 0.74**(0.08) ; 0.77**(0.13) ; 0.88**(0.05) ; 0.84**(0.03)
    - Log(real estate index): 0.08**(0.02) ; 0.07**(0.04) ; ............ ;
    - Log(stock market index): ............ ; 0.09**(0.04) ; ......
  - Demand Equation coefficients (Coef.(Std error)):
    - C: 8.08**(0.71) ; 7.97**(0.85) ; 7.54**(2.78) ; 7.98**(3.10)
    - Real interest rate (Treasury bill): 0.00(2.96) ; ...... ; -3.21(3.80) ; -0.53(6.38)
    - Real interest rate (Money market): ...... ; -0.32(1.93) ; ............ ;
    - Log(energy production index): 0.48**(0.03) ; 0.39**(0.21) ; 0.69(0.62) ; 0.90(0.60)
    - Log(real estate index): 0.22**(0.06) ; 0.28**(0.03) ; ............ ;
    - Log(stock market index): ............ ; 0.19(0.26) ; ......
  - Model fit and selection statistics:
    - Log-likelihood: -49.06 ; -53.10 ; -53.25 ; -55.07
    - Akaike information criterion: 2.72 ; 2.91 ; 2.92 ; 2.91
  - Significance notation:
    - ** denotes significance at the 95% level; * denotes significance at the 90% level.

### Key quantitative findings on real estate prices, supply, and demand
- Estimated coefficients on real estate prices were 0.20 and 0.24, respectively.
- The increase in real estate prices had a high impact on banks' willingness to supply credit, but its impact was two to three times higher on credit demand, consistent with a wealth effect of real estate prices.
- Development of mortgage products, spurred in part by increasing prices, strongly stimulated demand for credit.
- Preferred specification: equation (1) based on log-likelihood results; equations (2)–(4) shown for robustness.
- Using a ratio of the real-estate-related stock market index to the overall stock market index as a proxy produced results not significantly different from column (1).

### Disequilibrium analysis and timing
- Identification of excess supply/demand:
  - In the 32 quarters between 2000Q1 and 2007Q4, about 60 percent were identified as characterized by excess supply.
  - Trend reversal: late 2006 when demand strongly outpaced supply.
  - Estimated disequilibria averaged about 2 percent of actual credit before end-2006, but markedly increased thereafter.
- Interpretation:
  - Results do not support large excess credit demand (credit rationing) in the first half of the decade.
  - Prior to asset price increases, many borrowers lacked collateral and were deterred from applying (demand effect) or turned down (supply effect).
  - Surge in real estate prices partly driven by capital inflows and demand for high-end secondary homes by Europeans (anecdotal evidence).

### Policy implications and recommendations
- Supervisory vigilance:
  - Strong increase in credit growth—and potential build-up of imbalances between credit supplied and demanded—calls for renewed vigilance by supervisory authorities.
  - Large share of credit growth in consumer and housing loans and accompanying soaring real estate prices could expose the system to important losses should the trend reverse.
- Regulatory and macroprudential measures:
  - Central bank’s steps: ensuring banks’ compliance with the newly adopted code of good practices on mortgages (including fuller disclosure of risks) and the soon-to-be established credit bureau are positive.
  - Increase in minimum bank CAR from 8 percent to 10 percent will strengthen the system and bring it closer to best practices in emerging markets.
  - Recommendation: central bank should take more account of asset prices in monetary policy decisions to limit potential fall-out from asset price volatility.

### Shortcomings, caveats, and directions for further research
- Data and identification limitations:
  - Small quarterly sample (1997Q3–2007Q4) and limited availability of robust quarterly series (marginal lending rates, non-performing loan ratios, historical CAR).
  - Lack of variation in interest rates makes it difficult to select a good price variable; this may explain insignificance of interest-rate variables.
  - Possible endogeneity of real estate prices may bias estimates; mitigated by using stock market indices as proxies.
  - Recent structural changes may not be captured by historical relationships in the model.
- Suggested research improvements:
  - Use forward-looking variables reflecting expectations, a real estate price index when available, and corporate earnings variables.

### Recent developments and risks (late 2008)
- As of end-December 2008:
  - Credit growth had not slowed down significantly, but anecdotal evidence of a softening in real estate prices.
  - The real-estate sector stock index fell by 20 percent between end-September and end-December 2008.
  - The unfolding global financial crisis underscores the need for renewed vigilance and for monetary policy to take asset prices into account.

### Appendix I — Unit root test results (Augmented Dickey-Fuller Statistics)
- Log(real credit to the private sector): Level 2.01; First Difference -2.74*
- Real interest rate (Treasury bill): Level -1.53; First Difference -7.44**
- Real interest rate (Money market): Level -1.47; First Difference -7.38**
- Log(real lending capacity): Level 1.66; First Difference -6.00**
- Log(real estate index): Level 0.76; First Difference -4.45**
- Log(stock market index): Level -0.36; First Difference -3.83**
- Log(energy production index): Level -0.29; First Difference -2.88*
- Lag length determined by the Scwarz information criterion.
- ** denotes significance at the 1% level; * denotes significance at the 10% level.

### Appendix II — Cointegration test results (Trace Statistics)
- Log(real observed credit) and Log(real credit supplied)
  - r=0: Eigenvalue 0.321; Trace Statistics 8.63; Critical Value 15.49
  - r=1: Eigenvalue 0.093; Trace Statistics 3.80; Critical Value 3.84
  - Interpretation: reject null of zero cointegrating vectors; cannot reject hypothesis of one cointegrating vector between observed credit and real credit supplied.
- Log(real observed credit) and Log(real credit demanded)
  - r=0: Eigenvalue 0.513; Trace Statistics 32.55; Critical Value 15.49
  - r=1: Eigenvalue 0.052; Trace Statistics 2.30; Critical Value 3.84
  - Interpretation: reject null of zero cointegrating vectors; cannot reject hypothesis of one cointegrating vector between observed credit and real credit demanded.
- Significance tested at the 5% level.

*Source: _wp0953 (excerpt provided).*

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

### _wp0953 - References................................................................................................................................18

### Introduction
- The success of Morocco’s economic reform program since the turn of the century has:
  - substantially strengthened macroeconomic conditions,
  - accelerated the pace of non-agricultural growth, and
  - improved the soundness of the financial sector.
- Increase in tourism and remittance receipts, coupled with higher capital inflows, have:
  - sustained current account surpluses for most of the period, and
  - boosted domestic liquidity.
- In contrast with what happened in other regions of the world, these developments did not fuel a credit boom between 2000 and 2004:
  - the ratio of credit to the economy to non-agricultural GDP remained basically flat between 2000 and 2004.
- Insufficient access to credit, particularly for small- and medium-sized enterprises, was frequently cited as an important obstacle to growth in the national economic debate and echoed in the 2005 Investment Climate Assessment conducted by the World Bank.
- Starting in 2006, credit to the private sector accelerated:
  - credit to the private sector has been increasing rapidly, growing by almost 30 percent year-on-year in 2007,
  - its ratio to non-agricultural GDP jumped by more than 10 percentage points between 2006 and 2007, to reach almost 80 percent.
- Research objectives and analytical framework:
  - Test the hypothesis that the presence of excess liquidity in the banking system at a time of strong credit demand implies some form of credit rationing, with banks reticent to extend credit to firms in the presence of high information asymmetries, using an explicit disequilibrium framework common in the “credit crunch” literature.
  - Investigate whether the recent acceleration in the pace of credit growth is supply- or demand-driven.
  - Discuss the role that the sharp increase of asset prices, notably in the real estate sector, may have played in the credit increase.

### Figures (as listed)
- 1. Required Bank Reserves and Official Reserves, 2000–07
- 2. Treasury Bill Rate, 2000–07
- 3. Private Sector Credit in Percent of GDP, 2000–07
- 4. Structure of Banking System Assets
- 5. Credit to the Economy (2002=100)
- 6. Evolution of Real Estate Stock Market Index

### Appendixes (as listed)
- I. Unit Root Tests
- II. Cointegration Tests

### Organization of the paper (as stated)
- Section II briefly discusses the main characteristics of Morocco’s credit market.
- Section III reviews the literature on disequilibrium credit markets.

*Source: _wp0953 - References..............................................................................................................*

### Section IV discusses our estimation strategy and our results, and Section V concludes.

### _wp0953 - Section IV discusses our estimation strategy and our results, and Section V concludes.

### Overview of credit market developments (2000–2007)
- Macroeconomic context and balance sheet/liquidity developments:
  - Reserves increased more than fivefold to reach US$ 24 billion at end-2007.
  - Banking system’s structural liquidity position increased from about 10 percent of banks’ total assets in 2000 to 24 percent in 2006.
  - Required reserves raised from 10 percent in 2000 to 16.5 percent in 2006, then lowered to 15 percent at end-2007.
  - Banks held excess reserves during most of 2000-07; BAM regularly used its 24-hour deposit facility to mop up excess liquidity.
  - Declining interest rates in Europe contributed to a steady decline of interest rates in Morocco through much of the period.
- Credit-to-GDP and banking structure:
  - Ratio of credit to GDP remained basically flat between 2000 and 2006.
  - Share of the banking sector’s total assets to credit to the private sector fell by 3 points during 2000–06.
  - The three largest banks accounted for roughly 60 percent of outstanding credit to the private sector in 2007.
- Drivers of credit pick-up after 2006:
  - Central bank (BAM) measures since 2005 to encourage lending: standardizing minimum data requirements for credit applications, helping banks develop internal risk-rating systems, monitoring risk management, and setting up a credit bureau.
  - Treasury strengthened public-guarantee funds (SME guarantees and home-ownership guarantees).
  - Asset price increases:
    - Casablanca Stock Exchange (MASI) main index increased threefold since 2000.
    - Market capitalization rose from 29 percent of GDP to 103 percent between 2000 and 2007.
    - Evidence of strong real estate price increases, reaching almost 20 percent in 2007 in some cities.
    - Real-estate sector index on the Casablanca Stock Exchange used as proxy for real estate prices; caveat: index comprised of only 3 companies, representing 16 percent of total stock market capitalization at end-2007.
  - Mortgage market changes:
    - Variable-rate mortgages now represent close to half of outstanding mortgages (versus a little over a quarter four years earlier).
    - Average mortgage maturities lengthened.
    - Loan-to-value ratios above 90 percent almost doubled in the last two years, reaching 76 percent in 2007.
- Credit composition and growth:
  - Credit growth post-2006 was fairly broad-based; fastest growth in consumer and real estate credit.
  - Working capital, investment, real estate, and consumption credits all contributed to overall credit expansion (Figure 5 referenced).

### Literature review (methodological context)
- Disequilibrium / maximum likelihood approach commonly used in “credit crunch” literature (Madala and Nelson, 1974).
- Prior findings summarized:
  - Bernanke, Lown, and Friedman (1991): bank equity capital limited credit supply in US episode; demand weakening also contributed.
  - Blundell-Wignall and Gizycki (1992): no credit rationing in Australia; corporate risk reflected in risk premia.
  - Pazarbasioglu (1996), Nehls and Schmidt (2003), Ghosh and Ghosh (1999), Ikhide (2003): mixed evidence on whether credit constraints were supply- or demand-driven across Finland, Germany, East Asia, Namibia; common explanatory variables include lending capacity, interest rates, stock market performance, industrial production, inflation, and interest rate spreads.

### Estimation strategy (disequilibrium framework)
- Model structure:
  - Three equations: demand, supply, and short-side rule: observed credit Ct = min(Cd_t, Cs_t).
  - Demand and supply linear functions in explanatory variables with normally distributed errors; estimated by maximum likelihood (Marquardt iterative approach).
- Sample and data:
  - Quarterly data for period 1997Q3 to 2007Q4.
  - Parsimonious specification due to small sample and data scarcity.
  - Variables (most in logs except interest rates):
    - Demand equation explanatory variables: real interest rate (one-year Treasury bill rate; robustness: money market rate), economic activity (energy production index), real estate prices (real-estate sector stock index); robustness check using overall stock market index.
    - Supply equation explanatory variables: real interest rate, real estate prices, proxy for banks’ lending capacity (total deposits minus required reserves).
  - Rationale and expectations:
    - Lending rates expected negatively correlated with credit demand.
    - Energy production index chosen as proxy for economic activity and to aid identification (not strongly correlated with supply).
    - Real estate expected to have positive effect on both demand (wealth/collateral) and supply (collateral, reduced adverse selection).
    - Interest rate sign in supply ambiguous: higher rates may raise bank profitability (positive) or worsen adverse selection (negative).
- Stationarity and estimation in levels:
  - All variables have at least one unit root (Appendix I).
  - Model estimated in levels following argument that determinants of credit supplied and demanded form cointegrating vector; cointegration between estimated supply/demand and observed credit checked (Appendix II).
- Proxy and data caveats:
  - One-year Treasury bill or money market rates are imperfect proxies for lending rates (pass-through partial and lagged; data on marginal lending rates not available quarterly).
  - Non-performing loan ratio not available quarterly; credit-to-government excluded due to small correlation with private credit.

### Estimation results (1997Q3–2007Q4; key quantitative findings)
- General findings:
  - Interest rate variable is never significant (both Treasury bill and money market rate proxies). Possible explanations: poor proxy for lending rates, partial and lagged pass-through to lending rates, price rigidities, or structural breaks.
  - Elasticity of credit supplied to banks’ lending capacity is strong and positive, varying between 0.7 and 0.9 across specifications.
  - Elasticity of credit demanded to economic activity (energy index) about 0.5 in the most plausible specifications.
  - Asset prices play an important role:
    - Supply elasticity to asset prices around 0.08 (positive) whether using real-estate sector index or overall stock market index.
    - Demand function is sensitive to real estate index: real-estate index strongly significant in both supply and demand in specifications (1) and (2); overall stock market index is not significant in demand.
  - Model performs worse when asset price variables are excluded (specification (4)).
- Tabulated estimation (coefficients and standard errors preserved from Table 1):
  - Supply Equation coefficients (Coef.(Std error)):
    - C: 2.42**(0.80) ; 2.14*(1.24) ; 0.60(0.43) ; 1.97**(0.43)
    - Real interest rate (Treasury bill): 0.25(0.88) ; ...... ; 0.60(0.66) ; -0.59(1.04)
    - Real interest rate (Money market): ...... ; 0.38(0.46) ; ............ ; 
    - Log(real lending capacity): 0.74**(0.08) ; 0.77**(0.13) ; 0.88**(0.05) ; 0.84**(0.03)
    - Log(real estate index): 0.08**(0.02) ; 0.07**(0.04) ; ............ ; 
    - Log(stock market index): ............ ; 0.09**(0.04) ; ......
  - Demand Equation coefficients (Coef.(Std error)):
    - C: 8.08**(0.71) ; 7.97**(0.85) ; 7.54**(2.78) ; 7.98**(3.10)
    - Real interest rate (Treasury bill): 0.00(2.96) ; ...... ; -3.21(3.80) ; -0.53(6.38)
    - Real interest rate (Money market): ...... ; -0.32(1.93) ; ............ ;
    - Log(energy production index): 0.48**(0.03) ; 0.39**(0.21) ; 0.69(0.62) ; 0.90(0.60)
    - Log(real estate index): 0.22**(0.06) ; 0.28**(0.03) ; ............ ;
    - Log(stock market index): ............ ; 0.19(0.26) ; ......
  - Model fit and selection statistics:
    - Log-likelihood: -49.06 ; -53.10 ; -53.25 ; -55.07
    - Akaike information criterion: 2.72 ; 2.91 ; 2.92 ; 2.91
  - Significance notation:
    - ** denotes significance at the 95% level; * denotes significance at the 90% level.
- Interpretative conclusions from results:
  - Banks’ lending capacity is a primary binding constraint on credit supply during the early-2000s period (elasticities 0.7–0.9).
  - Demand for credit is responsive to economic activity (energy index) with elasticity around 0.5 in preferred specifications.
  - Real estate price increases significantly boosted both demand (via wealth/collateral effects) and supply (via collateral and perceived lower risk), with a notable role in the post-2006 surge in credit.
  - Interest rates (proxied by Treasury bill or money market rate) do not appear to explain credit dynamics in the model, potentially reflecting measurement/proxy issues and price rigidities in lending.

### Summary of substantive findings
- Structural and policy environment:
  - Morocco experienced strengthened macroeconomic conditions since the turn of the century, large capital inflows, and significant reserve accumulation (US$ 24 billion by end-2007).
  - Banking sector concentration and information asymmetries likely contributed to constrained credit distribution during 2000–06 despite rising liquidity.
- Main empirical conclusions from the disequilibrium estimation:
  - Credit expansion after 2006 was driven by both supply-side loosening and demand-side strengthening, with asset price (real estate) appreciation playing an important role.
  - Banks’ lending capacity (deposits minus required reserves proxy) is a strong determinant of credit supply (elasticity between 0.7 and 0.9).
  - Economic activity (energy index) is an important determinant of credit demand (elasticity about 0.5).
  - Asset prices (real-estate index) have a measurable positive effect on supply (elasticity ~0.08) and a strong effect on demand; overall stock market index less relevant for demand.
  - Interest-rate measures used are not significant, likely due to imperfect proxies for lending rates and price rigidities in the credit market.
- Model and data limitations noted:
  - Small quarterly sample (1997Q3–2007Q4) and limited availability of robust quarterly series (e.g., marginal lending rates, non-performing loan ratios, historical CAR).
  - Use of proxies (Treasury bill/money market rates, real-estate sector stock index, energy production index) and estimation in levels based on cointegration checks.

*Source: _wp0953 - Section IV discusses our estimation strategy and our results, and Section V concludes.*

### 0.20 and 0.24, respectively. In other words, while the impact of the increase of real estate prices

### _wp0953 - 0.20 and 0.24, respectively. In other words, while the impact of the increase of real estate prices

### Key findings on credit supply, demand, and real estate prices
- The estimated coefficients on real estate prices were 0.20 and 0.24, respectively.
- The increase in real estate prices had a high impact on banks' willingness to supply credit, but its impact was two to three times higher on credit demand, consistent with a wealth effect of real estate prices.
- Development of mortgage products, spurred in part by increasing prices, strongly stimulated demand for credit.
- The preferred specification was equation (1) based on log-likelihood results; equations (2)-(4) were shown for robustness.
- Using a ratio of the real-estate-related stock market index to the overall stock market index as a proxy for real estate prices produced results not significantly different from column (1).

### Disequilibrium analysis and timing
- The model identifies periods of excess demand and excess supply (figure 7, equation (1)).
- In the 32 quarters between 2000Q1 and 2007Q4, about 60 percent were identified as characterized by excess supply.
- The trend reversed conspicuously in late 2006, when demand strongly outpaced supply.
- Estimated disequilibria averaged about 2 percent of actual credit before end-2006, but markedly increased thereafter.

### Interpretation and context
- Results do not support the hypothesis of large excess credit demand (credit rationing) in the first half of the decade.
- Macroeconomic improvements and improved policy credibility typically take time to translate into higher consumption, investment, and credit demand—especially in traditionally cash-based economies such as Morocco.
- Prior to asset price increases, several borrowers lacked necessary collateral and were either deterred from applying (demand effect) or turned down (supply effect).
- Anecdotal evidence indicates the surge in real estate prices in Morocco was partly driven by capital inflows and demand for high-end secondary homes by Europeans.

### Policy implications and recommendations
- The strong increase in credit growth—and potential build-up of imbalances between credit supplied and demanded—calls for renewed vigilance by supervisory authorities.
- A large share of credit growth has been in consumer and housing loans and accompanied by soaring real estate prices, potentially exposing the system to important losses should the trend reverse.
- The central bank’s decision to ensure banks’ compliance with the newly adopted code of good practices on mortgages (including fuller disclosure of risks) and the soon-to-be established credit bureau are positive steps.
- The recent increase in the minimum bank CAR from 8 percent to 10 percent will strengthen the system and bring it closer to best practices in emerging markets.
- To limit potential fall-out from asset price volatility, Morocco’s central bank should take more account of asset prices in monetary policy decisions.

### Shortcomings, caveats, and directions for further research
- Short-time series and limited quarterly data are important caveats.
- Lack of variation in interest rates in Morocco makes it difficult to select a good price variable; this may explain why the model fails to pick up a significant role for the interest rate variable in credit decisions.
- Recent structural changes may not be captured by the historical relationships in the model.
- Possible endogeneity of real estate prices may bias estimates; this was mitigated by using stock market indices as proxies.
- Further research could use forward-looking variables reflecting expectations, a real estate price index when available, and corporate earnings variables.

### Recent developments and risks (late 2008)
- As of end-December 2008, credit growth had not slowed down significantly, but there is anecdotal evidence of a softening in real estate prices.
- The real-estate sector stock index fell by 20 percent between end-September and end-December 2008.
- The unfolding global financial crisis underscores the need for renewed vigilance and for monetary policy to take asset prices into account.

### Appendix I — Unit root test results (Augmented Dickey-Fuller Statistics)
- Log(real credit to the private sector): Level 2.01; First Difference -2.74*
- Real interest rate (Treasury bill): Level -1.53; First Difference -7.44**
- Real interest rate (Money market): Level -1.47; First Difference -7.38**
- Log(real lending capacity): Level 1.66; First Difference -6.00**
- Log(real estate index): Level 0.76; First Difference -4.45**
- Log(stock market index): Level -0.36; First Difference -3.83**
- Log(energy production index): Level -0.29; First Difference -2.88*
- The lag length is determined based on the Scwarz information criterion.
- ** denotes significance at the 1% level; * denotes significance at the 10% level.
- Significant coefficients of the first differences reflect the existence of a unit root.

### Appendix II — Cointegration test results (Trace Statistics)
- Log(real observed credit) and Log(real credit supplied)
  - r=0: Eigenvalue 0.321; Trace Statistics 8.63; Critical Value 15.49
  - r=1: Eigenvalue 0.093; Trace Statistics 3.80; Critical Value 3.84
  - Interpretation: reject null of zero cointegrating vectors; cannot reject hypothesis of one cointegrating vector between observed credit and real credit supplied.
- Log(real observed credit) and Log(real credit demanded)
  - r=0: Eigenvalue 0.513; Trace Statistics 32.55; Critical Value 15.49
  - r=1: Eigenvalue 0.052; Trace Statistics 2.30; Critical Value 3.84
  - Interpretation: reject null of zero cointegrating vectors; cannot reject hypothesis of one cointegrating vector between observed credit and real credit demanded.
- Significance is tested at the 5% level.

*Source: _wp0953 (excerpt provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0953.pdf_