This Selected Issues paper on Colombia shows that achieving investment grade status would help lower financing costs for the sovereign, and expand the pool of potential buyers of the Colombian economy. Colombia’s debt levels as of end-2008 were broadly similar to the average for investment grade emerging markets, suggesting that other indicators are taken into account in rating agencies’ assessments. A stronger process of fiscal consolidation that results in a significant decline in public sector debt could help compensate the structural weaknesses.


This Selected Issues paper on Colombia shows that achieving investment grade status would help lower financing costs for the sovereign, and expand the pool of potential buyers of the Colombian economy. Colombia’s debt levels as of end-2008 were broadly similar to the average for investment grade emerging markets, suggesting that other indicators are taken into account in rating agencies’ assessments. A stronger process of fiscal consolidation that results in a significant decline in public sector debt could help compensate the structural weaknesses.

II. Financial Soundness and Credit Growth1

Colombia’s financial system weathered the global crisis well despite a moderate weakening of banks’ asset quality indicators. This paper assesses whether Colombia’s financial soundness will support economic recovery. It finds that the sound loan portfolio and profitability of Colombia’s banking system bode well for future credit and output growth, especially compared to other emerging markets and to Colombia’s average performance during 1998–2008.

A. Macro-Financial Linkages in Colombia

1. Colombia’s bank credit growth declined sharply in 2008–2009 in line with a slowdown in GDP growth. Following a period of rapid growth in 2006 and 2007, when banks expanded consumption loans aggressively, bank credit slowed starting in the second quarter of 2008. The slowdown in output growth was the result of a tighter monetary stance (a response to overheating concerns) and, subsequently, the global crisis. The earlier expansion to riskier assets had weakened somewhat financial soundness indicators—capital to risk-weighted assets, performing loans to total loans, the share of liquid assets, and the return on average assets—from their 2005 levels (Figure 1). The weaker financial soundness may have contributed (though surely not driven) the ensuing slowdown in credit growth.

Figure 1.
Figure 1.

Colombia: Financial Soundness, Credit and GDP Growth, 1997-2009

Citation: IMF Staff Country Reports 2010, 106; 10.5089/9781455202652.002.A002

Sources: Superintendencia Financiera de Colombia; IMF Staff calculations.

2. The link between financial soundness and credit and output growth seemed fairly strong during Colombia’s 1998 banking crisis. In that episode, performing loans declined from 93 percent of total loans in 1997 to 86 percent in 1999 and return on assets fell from a healthy 2.4 percent to -1.7 percent. In the four years following the inception of the crisis, cumulative output losses were in the order of 33.5 percent of trend GDP—substantially larger than the average estimated output loss estimated for banking crises in other countries. The impact on credit growth and financial intermediation was also very pronounced, with credit-to-GDP decreasing from 38.2 percent in 1998 to 22.4 percent in 2001.2

3. The 1998 crisis led to a substantial strengthening of financial regulation and supervision, and a restructuring of Colombia’s banking system. Insolvent banks, largely stated-owned, were privatized, liquidated or recapitalized.3 Financial soundness was also strengthened through improvements in financial regulation, such as (i) the introduction in 2000 of risk-adjusted deposit insurance premiums, (ii) the issuance of new guidelines for banks’ credit risk management in 2002, and (iii) the expansion of the supervision perimeter in 2005 (which established an integrated financial supervisor by merging the banking, insurance, pensions, and securities’ supervisory agencies).

4. Measures taken in the wake of the global financial crisis further helped to strengthen the resilience of Colombia’s financial sector. At the end of 2008, banks reached an agreement with the financial supervisor to retain a portion of their 2008 profits as reserves until end-2010; this agreement helped to increase the capital to risk-weighted assets by about 1.4 percentage points. Later, in early 2009, the authorities raised the effective coverage of the deposit insurance, and refined the risk-adjusted premiums.4 In addition, the liquidity risk management system was modified to increase its transparency, improve reporting procedures, and extend the perimeter of financial institutions obliged to comply with this regulation. Despite some weakening, banks’ asset quality indicators remain strong, while profitability remains solid, reinforced by capital gains on banks’ investment portfolios.5

5. Colombia’s ability to weather the global financial crisis without any of the disruptions it experienced in past episodes of global shocks raises the question of whether its financial sector is better equipped to jump start credit and output growth than in the past. The analysis in this paper seeks to shed light on this question.

B. Brief Review of the Literature and Evidence

6. The international evidence supports the view that financial soundness affects output and credit growth. The analysis of systemic banking crises provides support for this view—output losses (measured as deviations from trend GDP) in the first four years of the crisis have averaged about 20 percent of GDP, and in some cases were as high as 98 percent of GDP.6 Rescuing financial system institutions also has had substantial fiscal costs—on average 13 percent of GDP—which increased public debt and reduced the scope for expansionary fiscal policies. The associated increase in sovereign risk premia typically increased funding cost for the whole economy, including the financial system, further stifling credit and output growth.

7. The link between financial soundness and credit and output growth has also been explored in the literature on the real effects of financial stress. Recent studies using aggregate country data have found that financial fragility has a significant impact on real activity. For example, Claessens et al. (2008) found strong evidence that a deterioration of financial variables—such as a drop in house and equity prices—increases the amplitude of recessions, and some evidence that financial crises (banking, currency, or both) increase the cost of recessions. More recently, based on a sample of seven advanced European countries, Tieman and Maechler (2009) showed that financial sector fragility leads to credit contraction, and estimated the potential real cost of an increase in financial fragility at over 1 percent of GDP on average.7

8. There is also evidence that financial soundness impacts the credit channel. Altunbas et al. (2002) and Gambacorta (2005) found evidence that an adverse monetary shock would produce a smaller decline in lending by well-capitalized European banks due to their higher access to wholesale deposit funding. In the same line, Matousek and Sarantis (2009) found that more liquid banks are less sensitive to monetary shocks in Central and Eastern European countries. A Colombia-specific study also found evidence that greater financial soundness allows banks to buffer liquidity shocks. Using bank-level data, Gómez-González and Grosz (2007) showed that indicators of financial strength of the Colombian banking system—such as capitalization and liquidity—ameliorate the effect of an increase in the policy interest rate. This suggests that monetary contractions could have less severe implications for credit growth when the banking system is better capitalized and has liquidity cushions to compensate a potential reduction in deposits with other financing sources.

9. In what follows, we explore this type of work by estimating a simple model that relates financial soundness with the growth of credit and GDP. We test the model using a panel dataset of a sample of 52 emerging market economies over a period of 11 years. We use the results of the model to assess the impact of the relative strength of Colombia’s financial system in 2009—compared to previous periods and to other emerging market economies—on real credit and output growth.

C. Framework and Estimation

10. Better capitalized banks tend to have less difficulties in raising funds, and thus are less likely to decrease lending. Gómez-González and Gross (2007) present a model that illustrates this relationship. In their model, banks have market power in the wholesale deposit market, and capital markets are imperfect—i.e. larger and more capitalized banks need to increase their deposit rates by less in order to raise the same amount of deposit funds as smaller, less capitalized banks. A key implication of the model is that when liquidity conditions tighten, better capitalized banks reduce loans by less (than other banks). Those banks therefore cushion the impact of monetary contractions on borrowers who are dependent on bank lending. In an environment where borrowers have limited access to other sources of finance, this feature of the banking system buffers the short-term impact of monetary shocks on economic activity.

11. Following a similar logic, we have sketched a credit allocation model to explore the implications of bank loan performance, capitalization, liquidity, and profitability on bank credit and output. The model, presented in Appendix 1, assumes capital market imperfections and informational asymmetries. Capital market imperfections are necessary to establish a relation between banks’ financial soundness, credit supply, and economic activity.8 Informational asymmetries, on the other hand, are needed to create a wedge between the marginal cost of financing as perceived by the borrower and the marginal benefit to the creditor.

12. In the model, banks’ borrowing in the interbank money market is subject to a risk premium. The premium is assumed to be lower for more capitalized banks with higher quality loan portfolios, larger liquidity buffers, and higher profits, i.e. for banks with better financial soundness indicators. In particular, a higher level of capitalization is indicative of better collateralization of bank borrowing, and ultimately increases banks’ creditworthiness. Higher quality portfolios and large liquidity buffers imply better risk-adjusted bank capitalization, while higher profitability provides a greater capital buffer. In the end, greater financial soundness reduces the risk premia of banks, and thus results in a positive impact on bank lending (Equation 5 in Appendix 1).


13. We estimate a system of simultaneous equations (similar to the one specified by Tieman and Maechler (2009)) to examine the link between financial soundness and credit and GDP growth. The system consists of equations for credit growth and GDP growth, as a function of financial soundness indicators (FSI), the policy interest rate, and the real effective exchange rate. It is assumed that the FSI variables only influence GDP growth indirectly through their effect on credit growth. Since all macroeconomic variables in the model are contemporaneous, their lagged values are used as instruments to correct for the potential simultaneity bias. FSIs enter the model with a lag because the risk premia of banks is assumed to be a function of past bank performance.9



article image

X¨i,t denotes that the variable is express as a difference from the sample mean in order to exclude country specific fixed effects.

14. The marginal effects of interest are represented in the reduced form equations:


Where the effect of FSIs on credit growth α1=c^21c^4c^6, and the effect on GDP growth is α2=c^2c^61c^4c^6.

15. An alternative specification seeks to assess the impact of financial soundness on the credit channel. Following Gómez-González and Grosz (2007), this alternative specification of the system includes as regressors interaction terms between the FSIs and the money market rate. In this case the marginal effects, evaluated at specific values of the FSIs, indicate whether those FSI values mitigate or magnify the impact of an increase in the money market rate.

16. The problem of omitted variables in the estimated equation is not expected to be serious. The model represented in (1) clearly does not contain all the determinants of credit and GDP growth, and thus omit a number of relevant variables and unobserved country effects. To control for the latter, as noted, we express all variable as differences from the sample mean, and include then time dummy variables. In addition, we try to control for outliers with dummies or by excluding them from the sample. Due to the high correlation among the various FSIs, an alternative specification of the model includes only one FSI indicator at a time. The systems of equations are estimated using Two-Stage Least Squares (TSLS).


17. Table 1 shows summary statistics for the variables used in the estimations. The sample comprises data from several sources.10 The dataset includes 52 emerging market economies (EMs), including 14 Latin American countries, and covers the period 1998–2008. Preliminary FSI data for 2009 are also available and are used in the analysis of the model’s implications for Colombia (Table 2). A larger dataset of 84 countries, including advanced and developing economies, is also available and used for robustness tests.

Table 1.

Variables – Description and Summary Statistics, 2002-2008

article image
Table 2.

Financial Soundness Indicators

article image
Source: Superintendencia Financiera de Colombia; Global Financial Stability Report, World Development Indicators.

18. Financial soundness is proxied through indicators that measure bank capitalization, loan quality, liquidity and profitability. Concretely, the indicators used in the estimations include capital to assets, capital to risk-weighted assets, performing loans to total loans, provisions to nonperforming loans, return on assets, and return on equity (all from the Global Financial Stability Report), and liquid reserves to total assets (from the World Development Indicators). The results we report are those based on the indicators with longer time series: capital to risk-weighted assets (CAR), performing loans to total loans (PL), liquid reserves to total assets (LR), and return on assets (ROA).

D. Results and Implications for Colombia11

19. Overall, the results provide evidence that asset quality and profitability have a positive impact on credit and GDP growth in emerging market economies. When the system of equations is estimated using all FSIs (Table 3, Equation 1), the ratio of performing loans and the return on assets have the expected positive sign but only PL is statistically significant, while CAR and LR have negative signs but are not significant. Even then, however, the joint marginal effect of all FSIs is positive and significant, suggesting that an improvement of all FSIs by one percentage point would lead to a 1.3 percentage point increase in credit growth and a 0.2 percentage point increase in GDP growth. The estimation of the system using one FSI at a time (Equations 25) results in positive and significant coefficients for PL and ROA, while CAR and LR continue to yield negative but insignificant coefficients.12 PL’s marginal effect is about 0.6 percentage points on credit growth, and about 0.1 percentage points on GDP growth. The marginal effects of changes in ROA are more than two times those of changes in PL (though it is rare to see ROA changing by one percentage point in a year).13

Table 3.

System of Equations, Emerging Markets Sample

article image
Note: In all estimations FSI variables are lagged. *** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent.

20. The macroeconomic variables of the regressions have the expected signs. In the credit equation, GDP growth has a positive and significant effect. In the GDP equation, credit growth has a positive and significant coefficient of about 0.13, which allows transmission of the FSI effect to GDP growth. The marginal effect of the interest rate is negative and significant, implying a reduction of 0.3 percentage points in both credit growth and GDP growth. The latter result suggests that the impact of the interest rate through other channels tends to be stronger than the direct credit channel effect.

21. The results also suggest that financial soundness buffers the effect on monetary shocks. In Table 4, the coefficients of the interaction terms of PL and ROA with the money market interest rate are positive and significant (equations 3 and 5). This implies that an increase in the money market rate would reduce credit growth and GDP growth, but the impact would be smaller the higher the level of performing loans. These results are not entirely consistent with the evidence in Gómez-González and Grosz (2007) of significant effects of bank capitalization and liquidity. Nonetheless, they are suggestive of the existence of a bank lending channel, and of banks’ ability to secure funding from alternative sources when liquidity conditions tighten.

Table 4.

System of Equations—Interest Rate Effects, Emerging Markets Sample

article image
Note: In all estimations FSI variables are lagged. *** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent

Application to Colombia

22. The previous results are helpful to assess whether the relative soundness of Colombia’s banking system would help boost credit and GDP growth in 2010. We replace FSI values for Colombia for 2009 in the estimated equations and compare the results against three benchmarks: (i) the average of the FSIs in Colombia during 1998–2008; (ii) the average level of the FSIs in Latin American countries during 2009; and (iii) the average level for emerging markets during 2009. The main results of the exercise were as follows:

  • The FSI levels for 2009 suggest that the contribution of financial soundness to Colombia’s economic performance in 2010 would be stronger than it was in the previous 10 years. Text Table 1 shows the FSI contribution to credit growth and GDP growth relative to the period average. The combined FSI effect brings credit growth almost two percentage points above the 1998–2008 average of around 5 percent, and GDP growth about 0.26 percentage points above the average during the previous decade. The individual effects of PL and ROA are somewhat smaller, but statistically significant. We therefore find strong evidence that the higher financial soundness will support economic recovery.

  • The benefits of strong financial soundness are broadly in line with those of other Latin American countries. Evaluating the effects at the difference between FSIs in Colombia and in other Latin American countries in 2009 does not show a significant difference in credit and GDP growth (Text Table 2), as soundness indicators are roughly similar. In terms of individual indicators, Colombia’s PL ratio suggests weaker economic prospects than in other countries in the region, whereas the ROA indicator suggests stronger prospects.

  • Compared to all other emerging markets, financial soundness in Colombia predicts a stronger economic performance. Colombia’s FSIs provide stronger support for credit and GDP growth through both PL and ROA.

Text Table 1.

Impact of Colombia’s 2009 Financial Soundness Indicators

(In percentage points)

article image
Note: Wald test for the significance of the effect evaluated at the specified values.*** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent
Text Table 2.

Comparison with Other Emerging Markets

(In percentage points)

article image
Note: Wald test for the significance of the effect evaluated at the specified values.*** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent

23. Greater financial soundness also reduces the impact of monetary shocks. Evaluating the marginal effect of the interaction between FSIs and the money market rate (Table 4) at the difference between the 2009 FSIs and their 1998–2008 averages shows the cushion provided by banking system soundness in the event that liquidity conditions tighten. For example, if performing loans are one percentage point higher than the 1998–2008 average, credit growth would decline by 0.14 percentage points less, while GDP growth would fall by 0.018 percentage points less. Similarly, if ROA is one percentage point above the 1998–2008 average, credit growth would decline by 0.29 percentage points less, and GDP growth would decline by 0.04 percentage points less. Given Colombia’s 2009 levels of PL and ROA relative to the 1998–2009 average, credit growth would be cushioned by up to a percentage point, and GDP growth by up to 0.14 percentage points if the money market rate were to be raised by one percentage point (Text Table 3.).

Text Table 3.

Financial Soundness and Monetary Shocks

(In percentage points)

article image
Note: Wald test for the significance of the effect evaluated at the specified values.*** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent

E. Concluding Remarks

24. The robustness of Colombia’s financial institutions bodes well for a rapid recovery. Colombia’s efforts to strengthen the financial system since the crisis of 1999 would provide support credit and GDP growth in 2010. The analysis presented in this paper suggests that Colombia’s improved financial soundness, compared to the previous decade, will increase credit growth by about 2 percentage points and GDP growth by about 0.3 percentage points. The impact of financial soundness is also stronger than in other emerging market countries.

25. While there are some risks to financial soundness, their indirect impact on economic activity is small and manageable. The high level of bank profitability—which has recently relied on valuation gains—may not be sustained. Nevertheless, a lower profitability is unlikely to reduce significantly the relative strength of Colombia’s current levels of FSIs compared to other emerging markets.


  • Albertazzi, Ugo, and Leonardo Gambacorta, 2009, “Bank Profitability and the Business Cycle,” Journal of Financial Stability, Vol. 5, Issue 4, December, pp. 393409.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Altunbas, Yener, Otabek Fazylov, and Philip Molyneux, 2002, “Evidence on the Bank Lending Channel in Europe,” Journal of Banking & Finance, Vol. 26(11), November, pp. 20932110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bernanke, Ben, and Mark Gertler, 1995, “Agency Costs, Net Worth and Business Fluctuations,” American Economic Review, 78(2): pp.4359

    • Search Google Scholar
    • Export Citation
  • Bikker, J.A., and P.A.J. Metzemakers, 2005, “Bank Provisioning Behaviour and Procyclicality,” Journal of International Financial Markets, Institutions and Money 15, pp. 141157.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Claessens, Stijn, M. Ayhan Kose, and Marco E. Terrones, 2008, “What Happens During Recessions, Crunches and Busts?”, IMF Working Paper 08/274.

    • Search Google Scholar
    • Export Citation
  • Fondo de Garantías de Instituciones Financieras, 2009, Crisis Financiera Colombiana en los Años Noventa, August, Bogotá (Colombia).

  • Gambacorta, Leonardo, 2005, “Inside the Bank Lending Channel,” European Economic Review, Volume 49, Issue 7, October, pp. 17371759.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gómez-González, Jose E., and Fernando Grosz, 2007, Evidence of Bank Lending Channel for Argentina and Colombia, Cuadernos De Economia, Vol. 44 (Mayo), pp. 109126.

    • Search Google Scholar
    • Export Citation
  • Gómez-González, Jose E., and Paola Morales Acevedo, 2009, Bank Lending Channel of Monetary Policy: Evidence for Colombia, Using a Firms’ Panel Data, Banco de la Republica, Borradores de Economia No. 545.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2009, “Colombia: Selected Issues,” Country Report No. 09/23.

  • Laeven, Luc and Fabian Valencia, 2008, “Systemic Banking Crises: A New Database”, IMF Working Paper 08/224.

  • Matousek, Roman, and Nicholas Sarantis, 2009, “The Bank Lending Channel and Monetary Transmission in Central and Eastern European Countries,” Journal of Comparative Economics, Volume 37(2), pp. 321334.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marcucci, Juri and Mario Quagliariello, 2008, “Is Bank Portfolio Riskiness Procyclical? Evidence from Italy Using a Vector Autoregression,” Journal of International Financial Markets, Institutions and Money 18, pp. 4663.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tieman, Alexander F., and Andrea M. Maechler, 2009, “The Real Effects of Financial Sector Risk,” IMF Working Paper 09/198.

  • Yilmaz, Ensar, 2009, “Capital Accumulation and Regulation,” The Quarterly Review of Economics and Finance 49, pp. 760771.

Appendix 1. The Model

Consider a model with the following features:

  • At the beginning of each period, firms borrow from banks to finance production costs. They use the bank loans immediately to pay those costs.

  • Banks raise deposits from domestic residents and make loans to domestic firms, and borrow or lend additional funds in the domestic and international interbank markets.

  • At the end of the period, firms receive cash flows and repay bank loans. Banks pay interbank loans, interest to domestic depositors, and dividends to shareholders.

1. Firms

26. There is a large number of identical firms in the economy that borrow from domestic banks to finance the cost, ωk, of k units of input producing f(k) units of the final good with probability φ, and 0 with probability (1−φ). If a firm does not produce any output, it defaults on the bank loan. The production function is increasing, f ′(·) > 0, and concave, f″(·) < 0, with respect to the input. Firms have no initial funds and limited access to alternative sources of external finance. Assuming that the price of the input is one, firms’ profits are


where ltf denotes loans, rtL is the gross lending interest rate, and pt is the price of the final good.

2. Banks

27. There is a large number of identical banks in the economy that raise deposits from domestic residents, dt, and extend loans to firms, ltb, or hold bonds, bt. Banks have access to an inter-bank market where they borrow funds, mt, and have capital, ct. At the beginning of period t, banks’ balance sheet condition is ltb+bt=dt+mt+ct, and banks’ profits are:


where rt denotes the risk-free rate on government bonds, rtD is the interest rate paid on deposits, rtM is the inter-bank market rate, and φt is the repayment rate on bank loans, and rtN is the (normal) return on equity.

28. Further, we denote the bank liquidity ratio—bonds to total assets—as LRt=btbt+ltb, the capital-to-asset ratio as CAt=ctbt+ltb, and the capital adequacy ratio as CARt=ctltb=CAt1LRt, giving a 100 percent weight to loans and 0 percent weight to government bonds in the risk-weighted assets. We also denote the return-on-asset ratio as ROAt=rtNctbt+ltb=rtNCARt(1LRt). Therefore, substituting for bank borrowing in the money market, the bank balance sheet condition and profits become:


Assuming that mt > 0, dt > 0, ct > 0, lt > 0, zero-profit conditions imply that rtD=rtM=rtN=φtrtL. Furthermore, we assume that there is a premium, xt, on interbank funding over the risk-free bond rate, where xt = χt(CARt-1, φt-1, LRt-1, ROAt-1) is a decreasing function of every financial soundness indicator. The firms’ first order condition and the market clearing condition, ltf=ltb=lt, indicate that


Therefore, ltCARt1>0,ltφt1>0,ltLRt1>0,andltROAt1>0.

Equation 5 suggests that an increase (decrease) in capital-to-risk-weighted-assets, performing loans, liquidity, and in the return on asset, would reduce (increase) premium on the lending rate over the risk-free rate, and therefore results in higher lending and output.

Appendix 2. Robustness Tests

29. The results reported in Table 3 and 4 are broadly robust to an increase in the number of countries in the sample. We expanded the sample to 84 countries by including some advanced and developing economies, but the results remain broadly unchanged (Table A1). PL and ROA remain the only variables with coefficients that are both positive and significant. When the system of equations is estimated using all FSIs, the FSI marginal effects of financial soundness are slightly larger. This is also the case when only PL is included. When only the return on assets is included, its marginal effect is slightly smaller. Therefore, it seems that there is little difference between the FSI effects in EMs and other countries, although some regional difference may exist. For example, using only the sample of 14 Latin American countries, yields a larger marginal effect of PL—about 0.8 percentage points increase—on credit growth. This implies less favorable growth prospects in Colombia relative to other countries in the region based on the weaker PLs.

Table A1.

System of Equations, Sample of 84 Countries

article image
Note: In all estimations FSI variables are lagged. *** indicates significance at 1 percent level, ** at 5 percent level, and * at 10 percent

30. The results are also robust to other estimation methods. The main results are virtually the same with a Three Stage Least Squares estimator, and when using second rather than first lags as instrumental variables, including for FSIs. A fixed effects estimator of credit growth on FSIs also shows similar results. The effect of PL is about 0.6 percentage points, while the effect of ROA is about 1 percentage point. The results from a dynamic panel data estimation of credit growth (controlling for potential autocorrelation and using first and second lags as instrumental variables) are somewhat different, however.


Prepared by Iva Petrova and Enrique Flores.


According to Fogafin (2009), 24 financial institutions were liquidated between 1998 and 2001. State-owned financial institutions comprised about 20.3 percent of the total financial system assets in 1998. Ten of the eleven state-owned credit institutions existing at the time of the crisis were recapitalized between 1998 and 2005.


The deposit insurance agency refunds part of the premium based on ratings associated with banks’ capital adequacy, asset quality, management quality, earnings, and liquidity (CAMEL ratings). The authorities revised the weights of the rating categories to give more prominence to capital adequacy and profitability.


An easier monetary stance and the slowdown in credit demand prompted banks to increase their government bond portfolios, which stood at 21.4 percent of assets in November 2009.


In the same paper, Tieman and Maechler (2009) report results obtained with bank-level data showing that credit contractions are more severe for banks under greater financial stress.


There is a strand of literature that analyzes the impact of macroeconomic factors on financial soundness. For Colombia, IMF (2009) found that macroeconomic and financial shocks have important bearings on banks’ soundness. This raises issues of endogeneity, albeit somewhat ameliorated by the lagged FSIs.


Real GDP data from the IMF’s World Economic Outlook (WEO), credit growth data from the International Financial Statistics (IFS) and the Economist Intelligence Unit (EIU), REER data from the EIU, and FSI indicators from the Global Financial Stability Report (GFSR) and the World Bank’s World Development Indicators (WDI).


The results are broadly robust to changes in the sample of countries, as well as to alternative estimation methods (see Appendix 2).


Regulatory requirements for capital adequacy and liquidity impart persistence to indicators like CAR and LR, which may explain their lack of significance.


In our data, the standard deviation of the nonperforming loans ratio is 7.1 percentage points, compared to 1.6 for the return on average assets (see Table 2).

Colombia: Selected Issues Paper
Author: International Monetary Fund