Colombia: Selected Issues Paper

International Monetary Fund

doi:10.5089/9781455202652.002.A001

Jump to Content Jump to Main Navigation

Colombia

Selected Issues Paper

Author: International Monetary Fund

Publication Date:: 04 May 2010

eISBN:: 9781455202652

Language:: English

Keywords:: ISCR; CR; GDP; GDP growth; debt; rating agency; risk index; ratings data; descriptive statistics; debt to GDP; Emerging and frontier financial markets; Monetary base

Download PDF (2.2 MB)

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.

Abstract

I. Determinants of Investment Grade Status and Implications for Colombia’s Public Debt¹

A. Introduction

1. The Colombian authorities are interested in regaining investment grade status, which was lost during the financial crisis of 1998–99. Achieving investment grade status would help lower financing costs for the sovereign, and expand the pool of potential buyers of Colombian paper. It would also reduce borrowing costs for corporates with access to international capital markets.

2. Identifying the main determinants of investment grade status can help guide policies towards achieving this goal. 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. Gauging the weights given by rating agencies to alternative macroeconomic variables can be a helpful input for the authorities’ overall policy framework, including the role of a medium term debt goal.

3. The paper finds that, to a large extent, investment grade rating status can be explained by a small number of variables. The analysis is based on a random effects binomial logit model estimated with data for a sample of 48 emerging market economies during the period 1993–2008. The model identifies a set of five core variables that are relevant for the determination of investment grade status, namely external public debt, domestic public debt, political risk, exports, and broad money (all variables as a share of GDP, except for political risk).

4. Building on existing literature, the main contributions of this paper consist of: (i) defining the dependent variable in binary form (investment grade/speculative grade) as opposed to ordinal ratings; (ii) restricting the sample to emerging market economies only, to avoid industrial country bias on investment grade determinants; and (iii) presenting additional explanatory variables not included in previous studies, in particular a breakdown of debt indicators and a measure of financial depth.

B. Background and Literature Review

5. Sovereign debt ratings are intended to be forward-looking qualitative measures of the probability of default elaborated by rating agencies. They are summary assessments of a government’s ability and willingness to repay its debts in full and on time. The three major credit rating agencies—Moody’s Investor Services (Moody’s), Standard and Poor’s (S&P), and Fitch Ratings (Fitch)—indicate that their assessments of government risk are based on the analysis of a broad set of economic, social, and political factors, but are not explicit about the weights given to those variables in their final assessments.² The ratings (and meaning) given by these agencies are summarized in Table 1.

Table 1.

Sovereign Credit Ratings by Agency

Note: Within rating categories, S&P and Fitch use plus (+) or minus (-) signs to show relative standing, with A+ being better than A or A-.Moody's uses a modifier of 1, 2, or 3 for the same purpose, with A1 being better than A2 or A3.

Table 1.

Sovereign Credit Ratings by Agency

	S&P	Moody’s	Fitch
Investment Grade
Highest quality, reliable, stable	AAA	Aaa	AAA
Quality, but a little more risk	AA	Aa	AA
Economic situation can affect finances	A	A	A
Just making the grade	BBB	Baa	BBB
Speculative Grade
More sensitive to economic changes	BB	Ba	BB
Financial situation varies considerably	B	Ba	B
Vulnerable, dependent on favorable economic conditions to meet payments	CCC	Caa	CCC
Highly vulnerable, speculative	CC	Ca	CC
Close to default, may be in arrears	C	C	C
Defaulted on obligations	D		D

Table 1.

Sovereign Credit Ratings by Agency

	S&P	Moody’s	Fitch
Investment Grade
Highest quality, reliable, stable	AAA	Aaa	AAA
Quality, but a little more risk	AA	Aa	AA
Economic situation can affect finances	A	A	A
Just making the grade	BBB	Baa	BBB
Speculative Grade
More sensitive to economic changes	BB	Ba	BB
Financial situation varies considerably	B	Ba	B
Vulnerable, dependent on favorable economic conditions to meet payments	CCC	Caa	CCC
Highly vulnerable, speculative	CC	Ca	CC
Close to default, may be in arrears	C	C	C
Defaulted on obligations	D		D

6. Sovereign credit ratings are important for at least three reasons. First, they are a key determinant of a country’s borrowing costs in international capital markets. Second, the sovereign rating generally sets a ceiling for the ratings assigned to domestic banks and companies, and therefore affects private financing costs.³ And third, some institutional investors have lower bounds for the risk they can assume in their investments and will choose their portfolio composition taking into account the credit risk signaled by the rating notations.

7. Since sovereign ratings summarize a vast amount of information, empirical studies have tried to predict country ratings based on a parsimonious set of economic variables. The seminal paper by Cantor and Packer (1996), based on a sample of 49 industrialized and developing countries, suggested that six variables were likely to explain ratings: per capita income, GDP growth, inflation, external debt, level of economic development, and default history. Using the same methodology, Afonso (2003) found that GDP per capita was the only relevant determinant of ratings of developed countries, while external debt played a key role for developing countries. In contrast, Mulder and Perrelli (2001) found that for emerging market economies the ratio of investment to GDP was the key variable explaining ratings. Results on the significance of political variables have been mixed: Archer et al. (2007) concluded that political factors had little effect on bond ratings; Mellios and Paget-Blanc (2006) found that indicators of corruption were an important determinant of ratings; and Afonso et al. (2007) found a significant coefficient for an indicator of government effectiveness. (See Table 2).

Table 2.

Main Studies on the Determinants of Sovereign Ratings

Table 2.

Main Studies on the Determinants of Sovereign Ratings

Study and Regression Technique	Data Sample	Significant explanatory variables
Afonso (2003) OLS	81 countries, June 2001	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) Economic development (+), Default history (-)
Afonso, Gomes, Rother (2007) Pooled ordered probit, random effects ordered probit	130 countries, 1970-2005 Moody’s, S&P, Fitch	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) International reserves (+), Default history (-) Government effectiveness (+) EU countries (+)
Archer et al. (2007) Linear regression with panel-corrected standard errors	50 developing countries, 1987-2003 Moody’s, S&P, Fitch	Trade (+), Inflation (-) GDP growth (+), Default (-)
Cantor and Packer (1996) OLS	49 countries, September 1995 Moody’s, S&P	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) Economic development (+), Default history (-)
Mellios and Paget-Blanc (2006) Ordered logistic model	86 countries, December 2003 Moody’s, S&P, Fitch	Per capita GDP (+), Government income (+) Real exchange rate changes (+), Inflation (-) Default history (-), Corruption index (+)
Mulder and Perrelli (2001) Pooled OLS and Feasible GLS	25 emerging market economies, 1992-1999 Moody’s, S&P	Debt over exports (-), Rescheduling history (-) Fiscal balance (+), Output growth (+) Inflation (-), Investment to GDP (+)
Rowland (2004) OLS regression	50 developing countries, July 2003 Moody’s, S&P	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) International reserves (+), Openness (+)
Rowland and Torres (2004) Random effects GLS regression	16 emerging market economies, 1987-2001 Moody’s, S&P	GDP growth (+), Inflation (-) External debt ratios (-), International reserves (+) Openness (+), Default history (-)

Table 2.

Main Studies on the Determinants of Sovereign Ratings

Study and Regression Technique	Data Sample	Significant explanatory variables
Afonso (2003) OLS	81 countries, June 2001	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) Economic development (+), Default history (-)
Afonso, Gomes, Rother (2007) Pooled ordered probit, random effects ordered probit	130 countries, 1970-2005 Moody’s, S&P, Fitch	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) International reserves (+), Default history (-) Government effectiveness (+) EU countries (+)
Archer et al. (2007) Linear regression with panel-corrected standard errors	50 developing countries, 1987-2003 Moody’s, S&P, Fitch	Trade (+), Inflation (-) GDP growth (+), Default (-)
Cantor and Packer (1996) OLS	49 countries, September 1995 Moody’s, S&P	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) Economic development (+), Default history (-)
Mellios and Paget-Blanc (2006) Ordered logistic model	86 countries, December 2003 Moody’s, S&P, Fitch	Per capita GDP (+), Government income (+) Real exchange rate changes (+), Inflation (-) Default history (-), Corruption index (+)
Mulder and Perrelli (2001) Pooled OLS and Feasible GLS	25 emerging market economies, 1992-1999 Moody’s, S&P	Debt over exports (-), Rescheduling history (-) Fiscal balance (+), Output growth (+) Inflation (-), Investment to GDP (+)
Rowland (2004) OLS regression	50 developing countries, July 2003 Moody’s, S&P	GDP per capita (+), GDP growth (+) Inflation (-), External debt ratios (-) International reserves (+), Openness (+)
Rowland and Torres (2004) Random effects GLS regression	16 emerging market economies, 1987-2001 Moody’s, S&P	GDP growth (+), Inflation (-) External debt ratios (-), International reserves (+) Openness (+), Default history (-)

C. Empirical Model Specification

8. This paper will focus on the determinants of investment grade ratings for emerging market economies. All the studies listed in Table 2 transformed credit ratings into a linear scale and used this ordinal measure as the dependent variable. In contrast, this paper defines a binary dependent variable for investment grade status, based on ratings data from Moody’s, S&P, and Fitch. The rating for any given year is the end-December rating, and the dummy is made equal to 1 for countries that were assigned investment grade status by at least two out of the three agencies.⁴ A panel data framework is used to control for heterogeneity across countries. A random effects binomial logit model produces better results (from an econometric point of view) than those obtained from a pooled regression and a fixed effects regression. The advantage of this technique is that it uses information from all countries in the sample and the marginal effect of any independent variable on the probability is conditional on the values of all covariates.⁵

9. The model specification can be written as:

\begin{matrix} {IG}_{it} = α + {βX}_{it} + {λZ}_{i} + a_{i} + μ_{it}, i = 1, \dots N, t = 1, \dots T & (1) \end{matrix}

Where IG_it is the binary variable equal to 1 for countries with investment grade status; X_it is a vector containing the time-varying explanatory variables described below; Z_i is a vector of time invariant variables that include regional and default dummies; a_i stands for the individual effects for each country i (that can either be modeled as an error term or as N dummies to be estimated) and μ_it represent disturbances that are independent across countries and across time.

10. Building on the evidence provided by the existing literature, the paper identifies a set of potential determinants of investment grade status. The selection of the explanatory variables is guided by the rating agencies’ reports and previous empirical evidence. Table 3 lists the explanatory variables that were included as regressors (X_it) in model.⁶

Table 3.

Explanatory Variables and Expected Sign

Note: ** denotes variables not included in previous empirical studies

D. Data and Estimation Results

Data

11. The regression analysis is based on a sample of 48 emerging market countries. The ratings data are obtained from the three main rating agencies for the 1993–2008 period. Table 4 provides some descriptive statistics of the dependent variable.⁷ Unless otherwise specified, the macroeconomic variables are drawn from the WEO database. The political risk variable is based on the political risk index published by the International Country Risk Guide, where a higher value indicates lower risk. For all the time-varying regressors, lagged values of the explanatory variables are used to avoid endogeneity problems.

Table 4.

Descriptive Statistics of Countries in the Sample and Investment Grade Status

Table 4.

Descriptive Statistics of Countries in the Sample and Investment Grade Status

	Latin America	Europe	East Asia	Other	Total
Number of countries	15	16	7	10	48
Number of observations	218	222	112	139	691
Percent of observations in the total sample	31.3	33.3	14.6	20.8	100.0
Percent of observations with investment grade status	17.4	63.1	57.9	27.0	40.7
Percentage of countries with a switch in status	40.0	50.0	57.1	30.0	43.8
Percentage of countries with at least one upgrade to investment grade	33.3	37.5	42.9	30.0	37.5
Percentage of countries with at least one downgraded to speculative grade	20.0	18.8	57.1	10.0	22.9

Table 4.

Descriptive Statistics of Countries in the Sample and Investment Grade Status

	Latin America	Europe	East Asia	Other	Total
Number of countries	15	16	7	10	48
Number of observations	218	222	112	139	691
Percent of observations in the total sample	31.3	33.3	14.6	20.8	100.0
Percent of observations with investment grade status	17.4	63.1	57.9	27.0	40.7
Percentage of countries with a switch in status	40.0	50.0	57.1	30.0	43.8
Percentage of countries with at least one upgrade to investment grade	33.3	37.5	42.9	30.0	37.5
Percentage of countries with at least one downgraded to speculative grade	20.0	18.8	57.1	10.0	22.9

12. Tests of means and medians show that investment grade countries tend to outperform speculative grade countries on most of the economic dimensions captured by the regressors. Welch tests were used to test for equality of means across the two groups of observations (investment and speculative grade), and Wilcoxon rank-sum tests were used to test whether the distribution is independent across the two groups of observations.⁸ Table 5 shows that both tests yielded similar results, with investment grade countries showing “better” values for the indicator than speculative grade countries on most accounts in almost all cases.

Table 5.

Test of Equality of Means and Medians of Country Characteristics, by Investment Grade Rating

Sources: Authors’ calculations based on data from IMF, World Bank, and International Country Risk Guide.Note: ** stands for statistical significance at the 1 percent level; * stands for statistical significance at the 5 percent level.

Table 5.

Test of Equality of Means and Medians of Country Characteristics, by Investment Grade Rating

		Mean			Median
Variable	Colombia 2008	Investment grade	Speculative grade	Welch test	Investment grade	Speculative grade	Wilcoxon test
Macroeconomic Variables
GDP per capita (US$ dollars)	4,763	8,083	3,766	**	4,950	2,800	**
Real GDP growth	2.5	4.8	3.8	**	5.1	4.4	**
Potential GDP growth	4.0	4.4	3.7	**	3.9	3.8	*
Inflation	7.0	5.7	25.9	**	4.4	7.7	**
Unemployment	10.6	8.3	10.5	**	7.5	9.8	**
External Sector
Exports to GDP	17.7	45.7	30.4	**	40.9	29.1	**
Current account balance to GDP	(2.8)	(3.0)	(2.4)		(2.9)	(2.4)
Private external debt to GDP	6.9	42.4	19.6	**	32.3	14.3	**
NIR to GDP	9.7	18.7	14.5	**	17.1	11.2	**
Government Sector
Primary balance	2.6	0.4	0.8	*	(0.1)	0.8	**
External public debt to GDP	12.2	12.8	30.6	**	9.9	25.3	**
Domestic public debt to GDP	19.9	21.7	27.3	**	15.2	19.6	**
Financial Depth
Broad money to GDP	35.8	64.3	54.6	**	50.4	42.6	**
Other
Political risk index ICRG
(+ is lower risk)	58.5	73.0	64.6	**	74.0	65.5	**

Table 5.

Test of Equality of Means and Medians of Country Characteristics, by Investment Grade Rating

		Mean			Median
Variable	Colombia 2008	Investment grade	Speculative grade	Welch test	Investment grade	Speculative grade	Wilcoxon test
Macroeconomic Variables
GDP per capita (US$ dollars)	4,763	8,083	3,766	**	4,950	2,800	**
Real GDP growth	2.5	4.8	3.8	**	5.1	4.4	**
Potential GDP growth	4.0	4.4	3.7	**	3.9	3.8	*
Inflation	7.0	5.7	25.9	**	4.4	7.7	**
Unemployment	10.6	8.3	10.5	**	7.5	9.8	**
External Sector
Exports to GDP	17.7	45.7	30.4	**	40.9	29.1	**
Current account balance to GDP	(2.8)	(3.0)	(2.4)		(2.9)	(2.4)
Private external debt to GDP	6.9	42.4	19.6	**	32.3	14.3	**
NIR to GDP	9.7	18.7	14.5	**	17.1	11.2	**
Government Sector
Primary balance	2.6	0.4	0.8	*	(0.1)	0.8	**
External public debt to GDP	12.2	12.8	30.6	**	9.9	25.3	**
Domestic public debt to GDP	19.9	21.7	27.3	**	15.2	19.6	**
Financial Depth
Broad money to GDP	35.8	64.3	54.6	**	50.4	42.6	**
Other
Political risk index ICRG
(+ is lower risk)	58.5	73.0	64.6	**	74.0	65.5	**

Regression results

13. Table 6 shows the results of estimating equation (1), with three different techniques: pooled, random effects and fixed effects. For each technique, the first column reports the unrestricted model (i.e. columns A, C, and E), whereas the second shows the results for the restricted model (i.e. columns B, D and F). The unrestricted model incorporates all the variables listed in Table 3, whereas the restricted model contains only the variables which were found to have a statistically significant impact.⁹

14. The random effects model (column D) is found to be the preferred specification. The likelihood-ratio test rejected the null hypothesis of no variation in the country specific errors of the pooled regression (columns A and B), indicating the need to control for country specific effects. At the same time, Hausman specification tests did not reject the null hypothesis that the random effects (column D) and consistent fixed effects coefficients (column F) were the same, suggesting that the random effects model is appropriate.

Table 6.

Regression Results

Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

Table 6.

Regression Results

	Pooled regression				Random effects				Fixed Effects
VARIABLES	A		B		C		D		E		F
GDP per capita (US$)	0.00	**	0.00	***	0.00				0.00
	(0.00)		(0.00)		(0.00)				(0.00)
Potential GDP growth	0.12				-0.34				-0.59
	(0.12)				(0.42)				(0.53)
Inflation	-0.05				-0.02				-0.01
	(0.03)				(0.04)				(0.01)
Unemployment	0.06	*			0.19				-0.03
	(0.04)				(0.14)				(0.17)
Exports to GDP	0.02	*	0.01		0.10		0.08	***	0.23	*	0.10	**
	(0.01)		(0.01)		(0.07)		(0.03)		(0.12)		(0.04)
Current account balance to GDP	0.06	*			0.01				-0.02
	(0.04)				(0.12)				(0.16)
Private external debt to GDP	0.02				-0.02				-0.05
	(0.02)				(0.03)				(0.04)
Net international reserves to GDP	0.00		0.01		0.16				0.29
	(0.03)		(0.02)		(0.12)				(0.18)
Primary balance to GDP	0.05				0.03				-0.31
	(0.08)				(0.18)				(0.26)
External public debt to GDP	-0.09	***	-0.09	***	-0.38	***	-0.25	***	-0.45	***	-0.22	***
	(0.02)		(0.01)		(0.10)		(0.04)		(0.11)		(0.04)
Domestic public debt to GDP	-0.06	***	-0.05	***	-0.18	***	-0.09	***	-0.24	***	-0.08	***
	(0.01)		(0.01)		(0.05)		(0.02)		(0.06)		(0.03)
Broad money to GDP	0.02	**	0.02	***	0.11	**	0.04	***	0.24	***	0.07	**
	(0.01)		(0.01)		(0.06)		(0.02)		(0.08)		(0.03)
ICRG political risk index	0.09	**	0.10	***	0.09		0.15	***	0.11	*	0.15	**
	(0.04)		(0.03)		(0.08)		(0.05)		(0.06)		(0.06)
Latin America and Caribbean dummy	-1.06		-1.69	***	1.60
	(0.72)		(0.38)		(3.81)
Europe dummy	0.35				4.56
	(0.66)				(4.28)
East Asia dummy	-0.25				-0.27
	(0.73)				(4.40)
Default dummy	-1.82	***			-3.27
	(0.56)				(4.85)
Constant	-7.53	**	-6.44	***	-10.55		-10.10	***
	(2.96)		(1.80)		(8.25)		(3.83)
Observations	483		536		483		540		248		284
Number of countries					35		37		17		19

Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

Table 6.

Regression Results

	Pooled regression				Random effects				Fixed Effects
VARIABLES	A		B		C		D		E		F
GDP per capita (US$)	0.00	**	0.00	***	0.00				0.00
	(0.00)		(0.00)		(0.00)				(0.00)
Potential GDP growth	0.12				-0.34				-0.59
	(0.12)				(0.42)				(0.53)
Inflation	-0.05				-0.02				-0.01
	(0.03)				(0.04)				(0.01)
Unemployment	0.06	*			0.19				-0.03
	(0.04)				(0.14)				(0.17)
Exports to GDP	0.02	*	0.01		0.10		0.08	***	0.23	*	0.10	**
	(0.01)		(0.01)		(0.07)		(0.03)		(0.12)		(0.04)
Current account balance to GDP	0.06	*			0.01				-0.02
	(0.04)				(0.12)				(0.16)
Private external debt to GDP	0.02				-0.02				-0.05
	(0.02)				(0.03)				(0.04)
Net international reserves to GDP	0.00		0.01		0.16				0.29
	(0.03)		(0.02)		(0.12)				(0.18)
Primary balance to GDP	0.05				0.03				-0.31
	(0.08)				(0.18)				(0.26)
External public debt to GDP	-0.09	***	-0.09	***	-0.38	***	-0.25	***	-0.45	***	-0.22	***
	(0.02)		(0.01)		(0.10)		(0.04)		(0.11)		(0.04)
Domestic public debt to GDP	-0.06	***	-0.05	***	-0.18	***	-0.09	***	-0.24	***	-0.08	***
	(0.01)		(0.01)		(0.05)		(0.02)		(0.06)		(0.03)
Broad money to GDP	0.02	**	0.02	***	0.11	**	0.04	***	0.24	***	0.07	**
	(0.01)		(0.01)		(0.06)		(0.02)		(0.08)		(0.03)
ICRG political risk index	0.09	**	0.10	***	0.09		0.15	***	0.11	*	0.15	**
	(0.04)		(0.03)		(0.08)		(0.05)		(0.06)		(0.06)
Latin America and Caribbean dummy	-1.06		-1.69	***	1.60
	(0.72)		(0.38)		(3.81)
Europe dummy	0.35				4.56
	(0.66)				(4.28)
East Asia dummy	-0.25				-0.27
	(0.73)				(4.40)
Default dummy	-1.82	***			-3.27
	(0.56)				(4.85)
Constant	-7.53	**	-6.44	***	-10.55		-10.10	***
	(2.96)		(1.80)		(8.25)		(3.83)
Observations	483		536		483		540		248		284
Number of countries					35		37		17		19

Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

15. The results in Table 6 suggest that investment grade status can be modeled parsimoniously with a handful of regressors.¹⁰

In line with other studies (e.g. Afonso (2003)), the results show that the level of debt matters for determining investment grade. However, the findings suggest that rating agencies do distinguish between types of debt. They tend to see risk in high public debt indicators, but do not seem to assign a significant weight to private external debt. Furthermore, rating agencies seem to attach greater risk to external public debt than to domestic public debt, with the coefficients of the former being more than 2½ times bigger.
The political risk index was found to be significant and positively related to the investment grade rating. This is in line with the findings of Mellios and Paget-Blanc (2006), though contrary to the findings of Archer et al. (2007) who used as regressors different proxies for democratic rule (such as executive party tenure, undivided government, and election cycles).
Exports to GDP and broad money to GDP were also found to be significant. The positive effect of exports on investment grade is line with the findings by Rowland (2004). The significant and positive impact of broad money on the determination of investment grade is a new finding as previous studies had not included measures of financial depth as regressors.

16. The random effects model provides a good fit, as it performs well in predicting both investment grade status and “switches”. Using the overall in-sample probability of being investment grade (0.4) as the cut-off point, the number of times the models correctly predicts zeroes and ones was computed. The model correctly classifies 86 percent of all observations, with Type I error (failing to predict investment grade status) of 8.9 percent, and Type II error (failing to predict speculative grade status) of 5.2 percent. In terms of “switches”, the model correctly predicts 6 out of 9 downgrades (68 percent), failing to predict the downgrades of Korea, Malaysia and Thailand—which happened in the context of the Asian crisis and were reversed within two years. The model also predicts correctly 11 of the 17 upgrades in the sample (65 percent). Among the upgrade cases that the model fails to predict are those of Mexico and South Africa (which are predicted with a three year lag) and those of Colombia and Uruguay (which lost their investment grade status within 4–5 years of the upgrade). Figure 1 depicts the behavior of the regressors in some of the cases that experienced an upgrade to investment grade during the sample period.

17. An analysis of marginal effects provides further insights on the impact of the regressors on the probability of investment grade status. Table 7 shows the average partial effects¹¹ for alternative levels of debt, openness, financial development and political risk. As expected in a binomial logit model, the marginal effects of each variable are nonlinear and are therefore larger for values in the middle of the distribution. For example, a 10 percentage point decrease in external public debt to GDP would increase the probability of investment grade by 16 percentage points on average, with the effect being almost double for those countries in the second quartile of the distribution. Similarly, a 10 point increase in the political risk index (implying an improvement in risk perception) would increase the probability of investment grade by 10 percentage points on average, with somewhat greater impact for countries in the middle of the distribution. Table 7 also shows that there are investment grade observations throughout the distribution of each variable, implying that there is not necessarily a minimum threshold for each of the regressors.

Table 7.

Average Partial Effects on the Probability of Investment Grade Status

Table 7.

Average Partial Effects on the Probability of Investment Grade Status

		Average Partial Effects	APE by Quartile
		1	2	3	4
External public debt to GDP, 10 percentage point decline		0.16	0.09	0.30	0.20	0.04
Domestic public debt to GDP, 10 percentage point decline		0.06	0.05	0.07	0.06	0.05
Political risk index, 10 point increase		0.10	0.05	0.12	0.14	0.08
Exports to GDP, 10 percentage point increase		0.05	0.04	0.08	0.08	0.02
Broad money to GDP, 10 percentage point increase		0.03	0.02	0.03	0.04	0.02
Memorandum items:
Quartile cut-off points
	External public debt to GDP	26	12	21	37	120
	Domestic public debt to GDP	26	10	19	33	141
	Political risk index	67	63	69	73	87
	Exports to GDP	36	23	32	45	120
	Broad money to GDP	59	34	45	67	279
Percentage of investment grade observations in each quartile
	External public debt to GDP		56.4	27.7	8.2	7.7
	Domestic public debt to GDP		39.5	27.2	17.4	15.9
	Political risk index		7.2	21.0	24.6	47.2
	Exports to GDP		12.8	17.4	27.7	42.1
	Broad money to GDP		11.3	26.2	33.8	28.7
Percentage of speculative grade observations in each quartile
	External public debt to GDP		7.6	25.3	36.1	31.0
	Domestic public debt to GDP		16.1	23.7	28.8	31.3
	Political risk index		35.4	28.8	24.1	11.7
	Exports to GDP		32.9	30.4	21.2	15.5
	Broad money to GDP		32.3	25.9	19.0	22.8

Table 7.

Average Partial Effects on the Probability of Investment Grade Status

		Average Partial Effects	APE by Quartile
		1	2	3	4
External public debt to GDP, 10 percentage point decline		0.16	0.09	0.30	0.20	0.04
Domestic public debt to GDP, 10 percentage point decline		0.06	0.05	0.07	0.06	0.05
Political risk index, 10 point increase		0.10	0.05	0.12	0.14	0.08
Exports to GDP, 10 percentage point increase		0.05	0.04	0.08	0.08	0.02
Broad money to GDP, 10 percentage point increase		0.03	0.02	0.03	0.04	0.02
Memorandum items:
Quartile cut-off points
	External public debt to GDP	26	12	21	37	120
	Domestic public debt to GDP	26	10	19	33	141
	Political risk index	67	63	69	73	87
	Exports to GDP	36	23	32	45	120
	Broad money to GDP	59	34	45	67	279
Percentage of investment grade observations in each quartile
	External public debt to GDP		56.4	27.7	8.2	7.7
	Domestic public debt to GDP		39.5	27.2	17.4	15.9
	Political risk index		7.2	21.0	24.6	47.2
	Exports to GDP		12.8	17.4	27.7	42.1
	Broad money to GDP		11.3	26.2	33.8	28.7
Percentage of speculative grade observations in each quartile
	External public debt to GDP		7.6	25.3	36.1	31.0
	Domestic public debt to GDP		16.1	23.7	28.8	31.3
	Political risk index		35.4	28.8	24.1	11.7
	Exports to GDP		32.9	30.4	21.2	15.5
	Broad money to GDP		32.3	25.9	19.0	22.8

E. Implications for Colombia

18. The above results suggest that Colombia’s efforts to increase the likelihood of a rating upgrade should focus on improving its debt indicators. Colombia’s public debt indicators are not too far away from the median of investment grade emerging market countries. However, there is a significant difference in terms of the other indicators between Colombia and other emerging markets (see Figure 2). For example, Colombia’s ratio of exports to GDP (17.7 percent) is significantly below that of investment grade countries, and is in the lowest quartile for emerging market countries overall. Similarly, Colombia’s political risk index and broad money to GDP are well below those for investment grade countries, and also in the lower quartiles of emerging market countries. While these indicators have been improving in recent years, further progress is likely to be gradual and not directly linked with macroeconomic policies. In contrast, a strong process of fiscal consolidation could result in a steady reduction in debt levels.

19. A marginal effects analysis for Colombia shows that reducing public debt, in particular external public debt, would increase substantially the probability of an upgrade. Based on the marginal effect analysis reported in Table 7 and holding other values at their 2008 levels, a decline in Colombia’s public debt ratio to 20 percent (including by halving its external public debt to 6 percent of GDP) would increase the probability of attaining investment grade status to 40 percent (the sample cut-off point).

F. Summary and Conclusions

20. The paper finds that investment grade ratings by the three major credit agencies can be explained by a small number of variables. The panel random effects framework identifies a set of five core variables that are relevant for the determination of investment grade status, in particular external public debt, domestic public debt, political risk, exports to GDP, and financial depth. Overall, the specification correctly predicts 86 percent of investment grade status of all observations, and two thirds of the upgrades and downgrades from and to investment grade.

21. The findings suggest that Colombia’s efforts to increase the likelihood of an upgrade to investment grade should focus on a faster pace of debt reduction. While Colombia’s public debt figures are not too far away from those of emerging market investment grade countries, it faces weaknesses on other determinants that appear to have significant weight in credit rating agencies decisions, namely political risk, financial depth, and exports to GDP. A stronger process of fiscal consolidation that results in a significant decline in public sector debt, in particular external public debt, could help compensate these structural weaknesses and improve the prospects for an upgrade in the near term.

References

Archer, Candace C., Glen Biglaise, and Karl DeRouen Jr., 2007, “Sovereign Bonds and the Democratic Advantage: Does Regime Type Affect Credit Rating Agency Ratings in the Developing World?” International Organization: 61, Spring 2007, pp. 341–365.
- Search Google Scholar
- Export Citation
Afonso, Antonio, 2003, “Understanding the Determinants of Sovereign Debt Ratings: Evidence for the Two Leading Agencies,” Journal of Economics and Finance, Vol. 27, Number 1, Spring 2003.
- Search Google Scholar
- Export Citation
Afonso, Antonio, P. Gomes and P. Rother, 2007, “What Hides Behind Sovereign Debt Ratings?” European Central Bank Working Paper Series No. 711.
- Search Google Scholar
- Export Citation
Borensztein, Eduardo, Kevin Cowan, and Patricio Valenzuela, 2007, “Sovereign Ceilings “Lite:”? The Impact of Sovereign Ratings on Corporate Ratings in Emerging Market Economies,” IMF Working Paper 07/75 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Cantor, Richard, and F. Packer, 1996, “Determinants and Impact of Sovereign Credit Ratings,” FRBNY Economic Policy Review, October.
- Search Google Scholar
- Export Citation
Mellios, Constantin, and E. Paget-Blanc, 2006, “Which Factors Determine Sovereign Credit Ratings?” The European Journal of Finance, Vol. 12, No. 4 (June), pp. 361–377.
- Search Google Scholar
- Export Citation
Mulder, Christian, and R. Perrelli, 2001, “Foreign Currency Credit Ratings for Emerging Market Economies:” IMF Working Paper 01/191, November 2001 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Rowland, Peter, 2004, “Determinants of Spread, Credit Ratings and Creditworthiness for Emerging Market Sovereign Debt: A Follow-up Study Using Pooled Data Analysis,” Borradores de Economía No. 296, Banco de la República de Colombia.
- Search Google Scholar
- Export Citation
Rowland, Peter, and J. Torres, 2004, “Determinants of Spread, Credit Ratings and Creditworthiness for Emerging Market Sovereign Debt: A Panel Data Study,” Borradores de Economía No. 295, Banco de la República de Colombia.
- Search Google Scholar
- Export Citation
Wooldridge, Jeffrey M., 2001, Econometric Analysis of Cross Section and Panel Data. Cambridge, Massachusetts: MIT Press Books, edition 1, Vol. 1, December.
- Search Google Scholar
- Export Citation

Appendix 1. Explanatory Variables

Macroeconomic variables

Per capita income in U.S. dollars. Higher per capita income tends to suggest a larger potential tax base and a greater ability to repay debt. It also serves as a proxy for the level of economic development, which might influence default risk.
Real GDP growth and potential GDP growth. Higher economic growth tends to decrease the relative debt burden and may help in avoiding insolvency.
Inflation rate. A low inflation rate reveals sustainable monetary and exchange rate policies. It can also be seen as a proxy of the quality of economic management.
Unemployment rate. A country with low unemployment tends to have more flexible labor markets making it less vulnerable to changes in the global environment.

External sector variables

Exports to GDP. A higher ratio suggests a greater capacity to obtain hard currency to repay foreign currency denominated debt. Unlike most previous studies, where exports are included only as a metric for external debt, this paper introduces it as an independent regressor.
External current account to GDP. A large current account deficit suggests a high dependence on foreign capital, which can be a source of risk to macroeconomic stability.
Private and public external debt to GDP. The higher the external indebtedness, the higher the risk of fiscal or balance of payments stress. In contrast to existing literature, this paper distinguishes between private and public external debt to allow for differences in the weight assigned by rating agencies to each one.
Net international reserves to GDP. The higher the ratio, the more resources are available to service foreign debt. It reduces a country’s vulnerability to liquidity shocks.

Government sector variables

Primary balance to GDP. A low primary balance indicates that the government lacks the ability or the will to increase taxes to cover current expenses. A weak fiscal position also implies a higher likelihood that external shocks result in a default. In contrast to previous studies, the regressions use the primary balance instead of the overall balance to avoid possible endogeneity with the credit rating.
Public debt to GDP. The higher the debt burden, the larger the transfer effort the government will have to make over time to service its obligations, and therefore a higher risk of default. In contrast to existing literature, this paper distinguishes between domestic and external public debt to allow for differences in weights.

Financial depth

Broad money to GDP. Countries that have access to a deep and diversified pool of finance are in a better situation than those whose private savings are low and whose financial system is repressed. For this reason, financial depth is a useful indicator of government financial flexibility. High levels of financial intermediation, as proxied by broad money to GDP, can be associated with a greater capacity to sustain a given domestic debt burden. Existing literature has not included this variable.

Other regressors

Political risk. Rule of law and respect for property rights provide confidence that political (and civil) institutions have a strong commitment to honoring financial obligations. As summarized by the International Country Risk Guide (ICRG), the political risk index is used as a proxy to measure a country’s willingness to repay.
Default history. A country’s default history tends to influence its rating. A binary variable is set equal to 1 when the sovereign has defaulted on its external debt at least once in the previous ten years.
Regional dummies. Dummies are included for the Latin America and the Caribbean, Europe, and East Asia.

Appendix 2. Robustness Analysis

The results reported in Table 6 were robust to alternative measures of the explanatory variables. In order to compare the results with those of previous studies, the model was estimated using debt scaled by exports as a regressor (as opposed to having debt and exports to GDP enter separately in the regression). Table A1 (column A) shows that the coefficient for the debt to exports ratio was found to be negative and significant, without altering the sign or significance of the other coefficients; however, this specification deteriorates the fit of the model. The political stability index of the World Bank Governance Indicators was used as an alternative measure to the ICRG political risk indicator. This coefficient was positive and significant, did not affect the sign or significance of the other coefficients, but the fit of the model deteriorated (column B). Alternative measures of financial depth, in particular net credit to the private sector and market capitalization of listed companies as a percent of GDP, did not prove to be significant, implying that broad money to GDP is the variable that best captures rating agencies’ concerns regarding financial depth.

Table A1.

Alternative Specifications

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Table A1.

Alternative Specifications

	Random effects
	A		B
External public debt to GDP			-0.41	***
			(0.08)
External public debt to exports	-0.04	***
	(0.01)
Domestic public debt to GDP	-0.10	***	-0.10	***
	(0.02)		(0.04)
Exports to GDP			0.07	***
			(0.03)
Broad money to GDP	0.04	***	0.07	***
	(0.01)		(0.03)
ICRG political risk index	0.14	***
	(0.05)
WGI political stability index			3.90	***
			(1.15)
Constant	-8.11	**	-7.68	**
	(3.45)		(3.88)
Observations	540		553
Number of ccode	37		41

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Table A1.

Alternative Specifications

	Random effects
	A		B
External public debt to GDP			-0.41	***
			(0.08)
External public debt to exports	-0.04	***
	(0.01)
Domestic public debt to GDP	-0.10	***	-0.10	***
	(0.02)		(0.04)
Exports to GDP			0.07	***
			(0.03)
Broad money to GDP	0.04	***	0.07	***
	(0.01)		(0.03)
ICRG political risk index	0.14	***
	(0.05)
WGI political stability index			3.90	***
			(1.15)
Constant	-8.11	**	-7.68	**
	(3.45)		(3.88)
Observations	540		553
Number of ccode	37		41

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

The model also yielded similar results when the dependent variable was defined based on the investment grade ratings of each individual credit agency. The results in Table A2 suggest that S&P, Fitch and Moody’s broadly share the same criteria for determining investment grade, although they weigh some variables differently. The broad similarity in criteria is not surprising given that the agencies agree on the classification of countries across investment/speculative grades in 94 percent of the cases. In all cases, the agencies attribute more weight to external public debt over domestic public debt. However, while Moody’s seems to put more weight on the political risk variable than the average, this variable was not significant for S&P. The ratio of exports to GDP was significant for Moody’s and with a lower coefficient for S&P, and broad money to GDP was significant for S&P and with a lower coefficient for Fitch.

Table A2.

Regression Results by Rating Agency

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Table A2.

Regression Results by Rating Agency

	Moody’s				Standard and Poor’s				Fitch Ratings
VARIABLES	A		B		C		D		E		F
GDP per capita (US$)	0.00				0.00	*			0.00
	(0.00)				(0.00)				(0.00)
Potential GDP growth	0.18				-1.88	**			-1.57	**
	(0.51)				(0.75)				(0.74)
Inflation	0.00				-0.18	*	-0.11	**	-0.11
	(0.01)				(0.10)		(0.06)		(0.08)
Unemployment	-0.55	**			0.17				0.06
	(0.23)				(0.21)				(0.22)
External public debt to GDP	-0.39	***	-0.29	***	-0.55	***	-0.44	***	-0.54	***	-0.26	***
	(0.09)		(0.08)		(0.11)		(0.08)		(0.17)		(0.05)
Domestic public debt to GDP	-0.15	**	-0.06	*	-0.20	***	-0.16	***	-0.32	***	-0.13	***
	(0.06)		(0.03)		(0.07)		(0.04)		(0.12)		(0.03)
Primary balance to GDP	-0.97	***			0.33				0.21
	(0.35)				(0.26)				(0.25)
ICRG political risk index	0.36	***	0.28	***	-0.07				0.02		0.15	***
	(0.13)		(0.07)		(0.12)				(0.12)		(0.06)
Private external debt to GDP	0.06		0.05	**	0.06				-0.04
	(0.04)		(0.02)		(0.05)				(0.05)
Net international reserves to GDP	0.00				0.16				0.43	**	0.17	***
	0.14				(0.15)				(0.21)		(0.06)
Current account balance to GDP	-0.04				0.02				-0.05
	(0.14)				(0.19)				(0.19)
Exchange rate volatility	-0.23	**			0.04				0.03
	(0.11)				(0.12)				(0.13)
Broad money to GDP	0.13	**			0.20	***	0.15	***	0.27	**	0.06	**
	(0.06)				(0.06)		(0.03)		(0.12)		(0.02)
Exports to GDP	0.25	**	0.10	***	0.03		0.06	**	0.13
	(0.11)		(0.03)		(0.08)		(0.02)		(0.13)
Latin America and Caribbean dummy	-0.46				-0.60				-5.13		-2.78	*
	(4.97)				(4.10)				(5.89)		(1.55)
Europe dummy	2.69				8.64	**	8.02	**	-1.53
	(4.89)				(4.32)		(3.21)		(4.76)
East Asia dummy	-9.13				-4.02				-24.70		-5.81	**
	(5.82)				(5.61)				(15.88)		(2.82)
Default dummy	-2.55				-10.38				-4.61
	(6.48)				(8.89)				(6.50)
Constant	-26.42	**	-17.21	***	11.99		0.47		4.65		-7.61	*
	(10.34)		(5.05)		(11.00)		(2.55)		(12.38)		(4.06)
Observations	479		595		431		487		404		436
Number of ccode	35		41		35		38		33		35

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Table A2.

Regression Results by Rating Agency

	Moody’s				Standard and Poor’s				Fitch Ratings
VARIABLES	A		B		C		D		E		F
GDP per capita (US$)	0.00				0.00	*			0.00
	(0.00)				(0.00)				(0.00)
Potential GDP growth	0.18				-1.88	**			-1.57	**
	(0.51)				(0.75)				(0.74)
Inflation	0.00				-0.18	*	-0.11	**	-0.11
	(0.01)				(0.10)		(0.06)		(0.08)
Unemployment	-0.55	**			0.17				0.06
	(0.23)				(0.21)				(0.22)
External public debt to GDP	-0.39	***	-0.29	***	-0.55	***	-0.44	***	-0.54	***	-0.26	***
	(0.09)		(0.08)		(0.11)		(0.08)		(0.17)		(0.05)
Domestic public debt to GDP	-0.15	**	-0.06	*	-0.20	***	-0.16	***	-0.32	***	-0.13	***
	(0.06)		(0.03)		(0.07)		(0.04)		(0.12)		(0.03)
Primary balance to GDP	-0.97	***			0.33				0.21
	(0.35)				(0.26)				(0.25)
ICRG political risk index	0.36	***	0.28	***	-0.07				0.02		0.15	***
	(0.13)		(0.07)		(0.12)				(0.12)		(0.06)
Private external debt to GDP	0.06		0.05	**	0.06				-0.04
	(0.04)		(0.02)		(0.05)				(0.05)
Net international reserves to GDP	0.00				0.16				0.43	**	0.17	***
	0.14				(0.15)				(0.21)		(0.06)
Current account balance to GDP	-0.04				0.02				-0.05
	(0.14)				(0.19)				(0.19)
Exchange rate volatility	-0.23	**			0.04				0.03
	(0.11)				(0.12)				(0.13)
Broad money to GDP	0.13	**			0.20	***	0.15	***	0.27	**	0.06	**
	(0.06)				(0.06)		(0.03)		(0.12)		(0.02)
Exports to GDP	0.25	**	0.10	***	0.03		0.06	**	0.13
	(0.11)		(0.03)		(0.08)		(0.02)		(0.13)
Latin America and Caribbean dummy	-0.46				-0.60				-5.13		-2.78	*
	(4.97)				(4.10)				(5.89)		(1.55)
Europe dummy	2.69				8.64	**	8.02	**	-1.53
	(4.89)				(4.32)		(3.21)		(4.76)
East Asia dummy	-9.13				-4.02				-24.70		-5.81	**
	(5.82)				(5.61)				(15.88)		(2.82)
Default dummy	-2.55				-10.38				-4.61
	(6.48)				(8.89)				(6.50)
Constant	-26.42	**	-17.21	***	11.99		0.47		4.65		-7.61	*
	(10.34)		(5.05)		(11.00)		(2.55)		(12.38)		(4.06)
Observations	479		595		431		487		404		436
Number of ccode	35		41		35		38		33		35

Robust standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Beyond the set of core regressors, credit agencies appear to rely on a few additional variables. In the case of Moody’s, private external debt was found to have a positive significant effect on the investment grade rating, reflecting the view that steady private access to international markets serves as an indicator of market confidence in the soundness of the corporate sector (column B). In the case of S&P, inflation was found to have a negative significant effect, which serves as a proxy for the quality of macroeconomic management (column D). Finally, in the case of Fitch, net international reserves were found to have a significant and positive effect, while the export variable was not significant (column F).

Prepared by Laura Jaramillo.

See http://www2.standardandpoors.com/aboutcreditratings/RatingsManual_PrintGuide.html for criteria used by Standard and Poor’s; http://www.moodys.com/moodys/cust/research/MDCdocs/09/2007200000530576.pdf7doc_id=2007200000530576&frameOfRef=corporate for criteria used by Moody’s; and http://www.fitchratings.com/creditdesk/reports/report_frame.cfm?rpt_id=474248&sector_flag=5&marketsector=1&detail= for criteria used by Fitch.

Borensztein et al. (2007) find that sovereign ratings have a significant effect on private ratings even after controlling for country specific macroeconomic conditions and firm-level performance indicators.

The ratings do not differ significantly across the three agencies. Investment/speculative grade status coincided across the three rating agencies for 94 percent of all observations in the sample.

The pooled estimation does not control for unobserved country effects, while the fixed effects logit model has the disadvantage that only countries where the dependent variable “switches” (from 0 to 1 and vice versa) can be included in the estimation—which in this case would lead to a sizeable number of cases being dropped. In addition, the fixed effects estimations cannot assess the impact of non-time varying country characteristics.

See Appendix 1 for a description of the explanatory variables and rationale for their inclusion in the regression analysis.

The sample of countries consists of: Argentina, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Jamaica, Mexico, Panama, Peru, Uruguay, Venezuela, Bosnia & Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Iceland, Israel, Latvia, Lithuania, Poland, Romania, Russia, Serbia, Turkey, Ukraine, China, Indonesia, Korea, Malaysia, Philippines, Thailand, Vietnam, Egypt, India, Jordan, Kazakhstan, Lebanon, Morocco, Pakistan, South Africa, Sri Lanka, and Tunisia.

A comparison of means only could be misleading in the presence of large outliers.

Variables that did not reveal any explicative power were dropped based on Wald tests. The restricted models are robust to alternative exclusion procedures. Furthermore, the variables found to be significant in the unrestricted model generally remain significant with the same sign in the restricted model.

Although rating agencies may assign substantial weight to other factors in determining specific rating assignments to a particular country at a given point in time, no systematic relationship between those variables and investment grade status was detected in the sample.

Calculated as the partial effects averaged across the population. See Wooldridge (2002).

References

Albertazzi, Ugo, and Leonardo Gambacorta, 2009, “Bank Profitability and the Business Cycle,” Journal of Financial Stability, Vol. 5, Issue 4, December, pp. 393–409.
- 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. 2093–2110.
- Search Google Scholar
- Export Citation
Bernanke, Ben, and Mark Gertler, 1995, “Agency Costs, Net Worth and Business Fluctuations,” American Economic Review, 78(2): pp.435–9
- 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. 141–157.
- 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).
- Search Google Scholar
- Export Citation
Gambacorta, Leonardo, 2005, “Inside the Bank Lending Channel,” European Economic Review, Volume 49, Issue 7, October, pp. 1737–1759.
- 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. 109–126.
- 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.
- Search Google Scholar
- Export Citation
International Monetary Fund, 2009, “Colombia: Selected Issues,” Country Report No. 09/23.
- Search Google Scholar
- Export Citation
Laeven, Luc and Fabian Valencia, 2008, “Systemic Banking Crises: A New Database”, IMF Working Paper 08/224.
- Search Google Scholar
- Export Citation
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. 321–334.
- 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. 46–63.
- 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.
- Search Google Scholar
- Export Citation
Yilmaz, Ensar, 2009, “Capital Accumulation and Regulation,” The Quarterly Review of Economics and Finance 49, pp. 760–771.
- Search Google Scholar
- Export Citation

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

\begin{matrix} Π_{t}^{f} = φ_{t} (p_{t} f (l_{t}^{f}) - r_{t}^{L} l_{t}^{f}) & (1) \end{matrix}

where $l_{t}^{f}$ denotes loans, $r_{t}^{L}$ is the gross lending interest rate, and p_t 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, d_t, and extend loans to firms, $l_{t}^{b}$ , or hold bonds, b_t. Banks have access to an inter-bank market where they borrow funds, m_t, and have capital, c_t. At the beginning of period t, banks’ balance sheet condition is $l_{t}^{b} + b_{t} = d_{t} + m_{t} + c_{t}$ , and banks’ profits are:

\begin{matrix} Π_{t}^{b} = φ_{t} r_{t}^{L} l_{t}^{b} + r_{t} b_{t} - r_{t}^{D} d_{t} - r_{t}^{M} m_{t} - r_{t}^{N} c & (2) \end{matrix}

where r_t denotes the risk-free rate on government bonds, $r_{t}^{D}$ is the interest rate paid on deposits, $r_{t}^{M}$ is the inter-bank market rate, and φ_t is the repayment rate on bank loans, and $r_{t}^{N}$ is the (normal) return on equity.

28. Further, we denote the bank liquidity ratio—bonds to total assets—as ${LR}_{t} = \frac{b_{t}}{b_{t} + l_{t}^{b}}$ , the capital-to-asset ratio as ${CA}_{t} = \frac{c_{t}}{b_{t} + l_{t}^{b}}$ , and the capital adequacy ratio as ${CAR}_{t} = \frac{c_{t}}{l_{t}^{b}} = \frac{{CA}_{t}}{1 - {LR}_{t}}$ , 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 $R O A_{t} = \frac{r_{t}^{N} c_{t}}{b_{t} + l_{t}^{b}} = r_{t}^{N} C A R_{t} (1 - L R_{t})$ . Therefore, substituting for bank borrowing in the money market, the bank balance sheet condition and profits become:

\begin{matrix} m_{t} = l_{t}^{b} + b_{t} - d_{t} - c_{t} & (3) \end{matrix}

\begin{matrix} Π_{t}^{b} = (φ_{t} r_{t}^{L} - r_{t}^{M}) l_{t}^{b} + (r - r_{t}^{M}) b_{t} + (r_{t}^{M} - r_{t}^{D}) d_{t} + (r_{t}^{M} - r_{t}^{N}) c_{t} & (4) \end{matrix}

Assuming that m_t > 0, d_t > 0, c_t > 0, l_t > 0, zero-profit conditions imply that $r_{t}^{D} = r_{t}^{M} = r_{t}^{N} = φ_{t} r_{t}^{L}$ . Furthermore, we assume that there is a premium, x_t, on interbank funding over the risk-free bond rate, where x_t = χ_t(CAR_t-1, φ_t-1, LR_t-1, ROA_t-1) is a decreasing function of every financial soundness indicator. The firms’ first order condition and the market clearing condition, $l_{t}^{f} = l_{t}^{b} = l_{t}$ , indicate that

\begin{matrix} p_{t} f_{t}^{ʹ} (l_{t}) = r_{t}^{L} = \frac{1}{φ_{t}} (r_{t} + x_{t}) & (5) \end{matrix}

Therefore, $\frac{\partial l_{t}}{\partial C A R_{t - 1}} > 0, \frac{\partial l_{t}}{\partial φ_{t - 1}} > 0, \frac{\partial l_{t}}{\partial L R_{t - 1}} > 0, a n d \frac{\partial l_{t}}{\partial R O A_{t - 1}} > 0$ .

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

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

Table A1.

System of Equations, Sample of 84 Countries

Dep Variable: Real Credit Growth	(1)		(2)		(3)		(4)		(5)
Constant	0.002		0.002		0.003		0.002		0.001
	(0.004)		(0.004)		(0.004)		(0.004)		(0.004)
CAR	-0.001		-0.001
	(0.001)		(0.001)
Performing Loans to Total Loans	0.004	***			0.005	***
	(0.001)				(0.001)
Liquid Reserves to Total Assets	-0.001						-0.001
	(0.001)						(0.001)
Return on Assets	0.008	**							0.012	***
	(0.004)								(0.003)
Money Market Interest Rate	0.001		0.001		0.001		0.001		0.000
	(0.000)		(0.000)		(0.000)		(0.000)		(0.000)
GDP Growth	0.953	***	1.319	***	0.914	***	1.339	***	1.094	***
	(0.367)		(0.343)		(0.332)		(0.355)		(0.331)
Dep Variable: Real GDP Growth
Constant	0.001		0.001		0.001		0.001		0.001
	(0.001)		(0.001)		(0.001)		(0.001)		(0.001)
Credit Growth	0.181	***	0.181	***	0.181	***	0.181	***	0.181	***
	(0.035)		(0.035)		(0.035)		(0.035)		(0.035)
REER Change	0.000		0.000		0.000		0.000		0.000
	(0.068)		(0.068)		(0.068)		(0.068)		(0.068)
Money Market Interest Rate	-0.002	**	-0.002	**	-0.002	**	-0.002	**	-0.002	**
	(0.000)		(0.000)		(0.000)		(0.000)		(0.000)
Credit Growth: Observations	556		556		556		556		556
Credit Growth: R-Squared	0.115		0.067		0.108		0.066		0.087
GDP Growth: Observations	699		699		699		699		699
GDP Growth: R-squared	-0.250		-0.250		-0.250		-0.250		-0.250
Marginal effects of FSIs, in percentage points
On credit growth	1.357	***	-0.118		0.621	***	-0.078		1.558	***
On GDP growth	0.246	**	-0.021		0.113	***	-0.014		0.283	**

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

Table A1.

System of Equations, Sample of 84 Countries

Dep Variable: Real Credit Growth	(1)		(2)		(3)		(4)		(5)
Constant	0.002		0.002		0.003		0.002		0.001
	(0.004)		(0.004)		(0.004)		(0.004)		(0.004)
CAR	-0.001		-0.001
	(0.001)		(0.001)
Performing Loans to Total Loans	0.004	***			0.005	***
	(0.001)				(0.001)
Liquid Reserves to Total Assets	-0.001						-0.001
	(0.001)						(0.001)
Return on Assets	0.008	**							0.012	***
	(0.004)								(0.003)
Money Market Interest Rate	0.001		0.001		0.001		0.001		0.000
	(0.000)		(0.000)		(0.000)		(0.000)		(0.000)
GDP Growth	0.953	***	1.319	***	0.914	***	1.339	***	1.094	***
	(0.367)		(0.343)		(0.332)		(0.355)		(0.331)
Dep Variable: Real GDP Growth
Constant	0.001		0.001		0.001		0.001		0.001
	(0.001)		(0.001)		(0.001)		(0.001)		(0.001)
Credit Growth	0.181	***	0.181	***	0.181	***	0.181	***	0.181	***
	(0.035)		(0.035)		(0.035)		(0.035)		(0.035)
REER Change	0.000		0.000		0.000		0.000		0.000
	(0.068)		(0.068)		(0.068)		(0.068)		(0.068)
Money Market Interest Rate	-0.002	**	-0.002	**	-0.002	**	-0.002	**	-0.002	**
	(0.000)		(0.000)		(0.000)		(0.000)		(0.000)
Credit Growth: Observations	556		556		556		556		556
Credit Growth: R-Squared	0.115		0.067		0.108		0.066		0.087
GDP Growth: Observations	699		699		699		699		699
GDP Growth: R-squared	-0.250		-0.250		-0.250		-0.250		-0.250
Marginal effects of FSIs, in percentage points
On credit growth	1.357	***	-0.118		0.621	***	-0.078		1.558	***
On GDP growth	0.246	**	-0.021		0.113	***	-0.014		0.283	**

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.

Fondo de Garantías de Instituciones Financieras, Fogafin (2009).

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.

Laeven and Valencia (2008).

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.

See Bernanke and Gertler (1995).

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).

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/002/2010/106/article-A001-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully

Colombia: Selected Issues Paper

Author: International Monetary Fund

Volume/Issue:: Volume 2010: Issue 106

Publisher:: International Monetary Fund

ISBN:: 9781455202652

ISSN:: 1934-7685

Pages:: 39

DOI:: https://doi.org/10.5089/9781455202652.002

Topics

Countries

Series

About

Resources

Abstract

I. Determinants of Investment Grade Status and Implications for Colombia’s Public Debt1

A. Introduction

B. Background and Literature Review

C. Empirical Model Specification

D. Data and Estimation Results

Data

Regression results

E. Implications for Colombia

F. Summary and Conclusions

References

Appendix 1. Explanatory Variables

Macroeconomic variables

External sector variables

Government sector variables

Financial depth

Other regressors

Appendix 2. Robustness Analysis

References

Appendix 1. The Model

1. Firms

2. Banks

Appendix 2. Robustness Tests

Same Series

Other IMF Content

Other Publishers

Asian Development Bank

Inter-American Development Bank

International Labour Organization

The World Bank

Share Link

Table of Contents

I. Determinants of Investment Grade Status and Implications for Colombia’s Public Debt¹