I. Introduction

The objective of this paper is to assess the effectiveness of the interest-rate transmission mechanism in Central American (CADR) and to provide recommendations to policymakers in order to enhance monetary policy effectiveness in CADR.¹ Several CADR countries with independent monetary policies have adopted policy interest rates as their main instrument to signal their monetary policy stances, often in the context of implementing or moving towards inflation targeting (IT) regimes. Except for Nicaragua, all CADR countries have now explicit policy rates, but their signaling function is undermined by weaknesses in the transmission to bank interest rates (and ultimately to prices and output). Hence, strengthening the pass-through of the policy rate to market interest rates is important for improving the effectiveness of monetary policy in the region and helping the central banks ensure price stability.

The paper shows that the interest-rate transmission is weaker and slower in CADR than in the LA6, the countries chosen as benchmarks, and differs within CADR according to different factors, such as the degree of financial dollarization, exchange rate flexibility, bank concentration, financial sector development, institutional quality, and fiscal dominance.² Using panel data analysis on a sample of 40 countries that have a policy rate, the study finds some evidence that reducing dollarization, increasing exchange rate flexibility, and developing further the financial sector can strengthen this channel of transmission of monetary policy.

The paper has four sections. Section II reviews the economic literature on interest-rate transmission. Section III presents estimates of the interest-rate transmission mechanism in Latin America, investigates the speed of transmission, and examines the factors that determine the interest-rate pass-through using panel data analysis. Section IV concludes with a summary of the findings and key policy recommendations for CADR countries.

II. Determinants of Interest-Rate Transmission: Literature Review

This section reviews a number of factors that are often seen as key determinants of the interest-rate transmission mechanism in developing countries and emerging markets. It will focus primary ly on the first step of monetary transmission, i.e., the pass-through of the policy interest rate to bank lending and deposit rates; the second step, consisting of the impact of bank interest rates on aggregate demand, is beyond the scope of this study. Reviewing the determinants of the interest-rate transmission mechanism helps understand the reasons why it is weaker in some countries and facilitates identifying policies to strengthen it.

High levels of financial dollarization may reduce the impact that changes in the policy rate have on banks‟ interest rates in local currency. The degree to which banks can transmit increases of the policy rate to their local currency rates may be limited due to borrowers‟ ability to switch to foreign-currency instruments.³ Moreover, in dollarized economies, a large devaluation can have a contractionary impact through its adverse impact on financial balance sheets, thus instilling a fear of floating (Calvo and Reinhart, 2002; Leiderman, Maino, and Parrado, 2006).

Lack of exchange rate flexibility also hampers the effectiveness of the interest-rate transmission mechanism. Allowing the exchange rate to float allows the policy rate of the central bank to be the primary monetary policy tool, sends clearer policy signals to market participants, and boosts monetary policy independence. In addition, exchange rate flexibility highlights to market participants that there are two-way exchange-rate risks, encouraging the development of hedging facilities, and generating incentives to reduce foreign currency mismatches that help reduce dollarization (Freedman and Otker-Robe, 2010).

Bank concentration limits competition and lowers banks‟ reaction to the policy rate and thus can undermine the interest rate transmission mechanism. Indeed, the response of banks‟ rates to changes in the policy rate depends on the adjustment costs incurred by banks. These adjustment costs depend on the elasticity of demand for bank loans, which is influenced by the structure of the financial system (Cotarelli and Korelis, 1994; De Bondt, 2002): relatively inelastic demand is more likely when there is higher bank concentration. Indeed, when banks have substantial market power, policy rate changes (and changes to the banks‟ costs of funds) may impact banking spreads, rather than market rates. For example, banks may try to profit from a reduction in the policy rate by maintaining lending rates fixed, thus increasing lending margins (Mishra, Montiel, and Splimbergo, 2010; and Dabla-Norris, et al, 2007).

The development of the financial system strengthens the interest-rate transmission mechanism as more alternative sources of capital increase the elasticity of demand for bank loans (Cotarelli, et al., 1994). In more developed money and interbank markets, the pass-through from the policy rate to bank deposit and lending rates improves given that the first step in transmission is from the policy rate to the interbank rates (Yang, Davies, Wang, Dunn, and Wu, 2011). Financial shallowness tends to lead to higher excess liquidity in banks, discouraging the development of an active interbank market and reducing the effectiveness of transmission. In addition, more developed domestic capital markets, including a secondary market for government securities and long-term domestic-currency securities, strengthens transmission as it can encourage de-dollarization (Leiderman et al., 2006).

The health of the financial system may also impact the effectiveness of the interest-rate transmission mechanism. Financially weak banks may respond to an injection of central bank liquidity or lower policy interest rates by building up liquidity or increasing margins in order to raise capital positions and increase provisioning rather than extending credit (IMF, 2010). Holding on to bad loans on the balance sheet may crowd out new loans and limit the impact of lower interest rates. In addition, when the banking system and the economy are weak and risk aversion and information costs rise, banks could be reluctant to lend (Archer and Turner, 2006). In this case, credit rationing increases and the demand for loans are more inelastic.

Strengthening central bank independence boosts the interest-rate transmission as it gives autonomy to the central bank to undertake monetary policy actions and enhances its signaling function to market participants. In developing countries, de jure and de facto independence are important in raising the effectiveness of the central bank (Cukierman, Webb, and Neyapti, 1992). Independence includes the freedom to set the policy rate without government interference (i.e. instrument independence). Improving independence encompasses increasing the autonomy of the central bank board, limiting or banning central bank lending to the government, increasing the financial autonomy of the central bank, and enhancing its accountability through better communication of policies and goals.⁴

Similarly, eliminating fiscal dominance enhances transmission as it does not subordinate monetary policy to the objectives of fiscal policy and strengthens central bank independence (IMF, 2010). Direct central bank lending to the government undermines the balance sheet of the central bank and constrains the ability of the central bank to clearly signal changes in the monetary policy stance by raising interest rates and performing open market operations (Laurens, 2005). High fiscal deficits can also raise interest rates by increasing the risk premium, thus interfering with the pass-through of the policy rate to market interest rates.

Weak regulations and institutional environment can also weigh on the interest-rate transmission as they cause problems of asymmetric information and contract enforcement which raise the cost of financial intermediation (Mishra, et al., 2010). This reduces the elasticity of the demand for loans and makes bank rates less sensitive to changes in the policy rate and can result in lower lending by banks. A better institutional and regulatory environment can also help to develop capital markets, thus strengthening the interest-rate transmission mechanism. These institutional weaknesses, which tend to be more prevalent in developing countries and emerging markets than in advanced countries, cover government effectiveness, regulatory quality, the rule of law, and control of corruption.⁴

III. Interest-Rate Transmission in CADR

A. Correlations of the Policy Rate with Market Rates

This section presents initial estimates of the interest-rate pass-through in CADR and the LA6. The LA6 were included as benchmarks for Central America given their relatively more developed monetary frameworks, including having well established IT regimes in place and, to varying degrees, flexible exchange rate arrangements (Table 1).⁵ Within CADR, only Guatemala has an IT regime, but the signaling function of its policy rate is relatively weak. For more details on the monetary policy frameworks in CADR, see Medina Cas, Carrión-Menéndez, and Frantischek (2011).

Table 1.

Monetary and Exchange Rate Frameworks in CADR and LA6¹

Source: IMF staff.

¹As of end-April, 2010.

Table 1.

Monetary and Exchange Rate Frameworks in CADR and LA6¹

CADR	Monetary Policy Framework	De Facto Exchange Rate Arrangement
Costa Rica	Transitioning to inflation targeting within an exchange rate band to the U.S. dollar.	Other managed arrangement.
Dominican Republic	Transitioning to inflation targeting. No explicitly stated nominal anchor, but the central bank monitors various indicators including the exchange rate.	Stabilized arrangement (or peg-like).
Guatemala	Inflation targeting, with some areas being strengthened.	Floating.
Honduras	Exchange rate anchor to the U.S. dollar.	Stabilized arrangement (or peg-like).
Nicaragua	Exchange rate anchor to the U.S. dollar.	Crawling peg.
LA6
Brazil	Inflation targeting.	Floating.
Chile	Inflation targeting.	Free floating.
Colombia	Inflation targeting.	Floating.
Mexico	Inflation targeting.	Floating.
Peru	Inflation targeting.	Floating.
Uruguay	Inflation targeting.	Floating.

Source: IMF staff.

¹As of end-April, 2010.

View Table

Table 1.

Monetary and Exchange Rate Frameworks in CADR and LA6¹

CADR	Monetary Policy Framework	De Facto Exchange Rate Arrangement
Costa Rica	Transitioning to inflation targeting within an exchange rate band to the U.S. dollar.	Other managed arrangement.
Dominican Republic	Transitioning to inflation targeting. No explicitly stated nominal anchor, but the central bank monitors various indicators including the exchange rate.	Stabilized arrangement (or peg-like).
Guatemala	Inflation targeting, with some areas being strengthened.	Floating.
Honduras	Exchange rate anchor to the U.S. dollar.	Stabilized arrangement (or peg-like).
Nicaragua	Exchange rate anchor to the U.S. dollar.	Crawling peg.
LA6
Brazil	Inflation targeting.	Floating.
Chile	Inflation targeting.	Free floating.
Colombia	Inflation targeting.	Floating.
Mexico	Inflation targeting.	Floating.
Peru	Inflation targeting.	Floating.
Uruguay	Inflation targeting.	Floating.

Source: IMF staff.

¹As of end-April, 2010.

Correlation analysis between the policy rates and banks‟ borrowing and lending rates provide a first take on the strength and speed of the interest-rate transmission mechanism. It suggests that the pass-through is generally weaker in CADR than in LA6 countries, and its strength varies by country (Table 2).⁶,⁷ With the exception of Peru (and Uruguay in the case of the pass-through of the policy rate to deposit rates), LA6 countries have higher long-term correlation coefficients than CADR. Within CADR, Costa Rica has the highest correlation, while Nicaragua, with its exchange rate anchor, has the lowest correlation. Changes in policy rates explain a smaller proportion of the variance in lending and deposit rates in CADR countries than in the LA6. Lending rates tend to have a higher correlation with the policy rate than deposit rates in most Latin American countries.⁸

Table 2.

Short and long-term correlations between policy rate and bank lending/deposit rates

(correlation coefficients, average 2004-2010 unless noted otherwise)

^1/Long term effects are calculated as $\frac{{\pi }_i+{\theta }_i+{\mu }_i+{\rho }_i}{1-({\alpha }_i+{\beta }_i+{\delta }_i)}$ from the equation:

$y_t=\alpha y_{it-1}+\beta y_{it-2}+\delta y_{it-3}+\theta x_{it}+\mu x_{it-1}+\pi x_{it-2}+\rho x_{it-3}$

Where y is the change in the retail rate (either lending or deposit) and x is the change in policy rate. Θ and u capture the cotemporaneous and short term effects respectively.

^2/Excludes countries that use the US Dollar as currency.

Source: See Appendix I

Table 2.

Short and long-term correlations between policy rate and bank lending/deposit rates

	Contemporaneous	Short Term	Long-term	R-Squared
For Lending Rates 1/
Central America (mean) 2/	0.07	0.14	0.49	0.31
Latin America 2/	0.13	0.23	0.62	0.34
LA6 (mean)	0.63	0.47	1.07	0.54
Brazil	0.71	0.58	1.19	0.54
Chile	0.63	0.22	1.09	0.60
Colombia	0.24	0.54	1.05	0.51
Mexico	0.80	-0.28	1.01	0.67
Uruguay	0.10	0.25	0.77	0.28
Costa Rica	0.05	0.46	0.76	0.27
Paraguay	0.08	0.04	0.62	0.23
Honduras	0.07	0.26	0.63	0.38
Peru	0.16	0.40	0.62	0.30
Dominican Republic	-0.07	-0.09	0.49	0.37
Argentina	0.30	0.30	0.43	0.53
Bolivia	0.16	0.22	0.28	0.27
Guatemala	0.11	-0.12	0.24	0.31
Nicaragua	0.09	0.14	0.22	0.27
For Deposit Rates 1/
Central America (mean) 2/	0.02	0.06	0.37	0.31
Latin America /2	0.26	0.22	0.58	0.56
LA6 (mean)	0.50	0.25	0.72	0.62
Chile	1.39	-0.08	1.00	0.77
Colombia	0.32	0.49	0.88	0.68
Brazil	0.65	0.63	0.83	0.48
Mexico	0.27	0.20	0.71	0.79
Paraguay	-0.13	0.44	0.70	0.78
Costa Rica	0.03	0.32	0.66	0.27
Argentina	0.13	0.43	0.54	0.84
Honduras	0.02	0.08	0.47	0.62
Peru	0.16	0.10	0.46	0.72
Uruguay	0.23	0.14	0.46	0.28
Bolivia	0.38	0.25	0.41	0.35
Nicaragua	0.23	0.05	0.37	0.25
Guatemala	0.01	0.01	0.34	0.76
Dominican Republic	0.02	0.06	0.27	0.31

(correlation coefficients, average 2004-2010 unless noted otherwise)

^1/Long term effects are calculated as $\frac{{\pi }_i+{\theta }_i+{\mu }_i+{\rho }_i}{1-({\alpha }_i+{\beta }_i+{\delta }_i)}$ from the equation:

$y_t=\alpha y_{it-1}+\beta y_{it-2}+\delta y_{it-3}+\theta x_{it}+\mu x_{it-1}+\pi x_{it-2}+\rho x_{it-3}$

Where y is the change in the retail rate (either lending or deposit) and x is the change in policy rate. Θ and u capture the cotemporaneous and short term effects respectively.

^2/Excludes countries that use the US Dollar as currency.

Source: See Appendix I

View Table

Table 2.

Short and long-term correlations between policy rate and bank lending/deposit rates

	Contemporaneous	Short Term	Long-term	R-Squared
For Lending Rates 1/
Central America (mean) 2/	0.07	0.14	0.49	0.31
Latin America 2/	0.13	0.23	0.62	0.34
LA6 (mean)	0.63	0.47	1.07	0.54
Brazil	0.71	0.58	1.19	0.54
Chile	0.63	0.22	1.09	0.60
Colombia	0.24	0.54	1.05	0.51
Mexico	0.80	-0.28	1.01	0.67
Uruguay	0.10	0.25	0.77	0.28
Costa Rica	0.05	0.46	0.76	0.27
Paraguay	0.08	0.04	0.62	0.23
Honduras	0.07	0.26	0.63	0.38
Peru	0.16	0.40	0.62	0.30
Dominican Republic	-0.07	-0.09	0.49	0.37
Argentina	0.30	0.30	0.43	0.53
Bolivia	0.16	0.22	0.28	0.27
Guatemala	0.11	-0.12	0.24	0.31
Nicaragua	0.09	0.14	0.22	0.27
For Deposit Rates 1/
Central America (mean) 2/	0.02	0.06	0.37	0.31
Latin America /2	0.26	0.22	0.58	0.56
LA6 (mean)	0.50	0.25	0.72	0.62
Chile	1.39	-0.08	1.00	0.77
Colombia	0.32	0.49	0.88	0.68
Brazil	0.65	0.63	0.83	0.48
Mexico	0.27	0.20	0.71	0.79
Paraguay	-0.13	0.44	0.70	0.78
Costa Rica	0.03	0.32	0.66	0.27
Argentina	0.13	0.43	0.54	0.84
Honduras	0.02	0.08	0.47	0.62
Peru	0.16	0.10	0.46	0.72
Uruguay	0.23	0.14	0.46	0.28
Bolivia	0.38	0.25	0.41	0.35
Nicaragua	0.23	0.05	0.37	0.25
Guatemala	0.01	0.01	0.34	0.76
Dominican Republic	0.02	0.06	0.27	0.31

(correlation coefficients, average 2004-2010 unless noted otherwise)

^1/Long term effects are calculated as $\frac{{\pi }_i+{\theta }_i+{\mu }_i+{\rho }_i}{1-({\alpha }_i+{\beta }_i+{\delta }_i)}$ from the equation:

$y_t=\alpha y_{it-1}+\beta y_{it-2}+\delta y_{it-3}+\theta x_{it}+\mu x_{it-1}+\pi x_{it-2}+\rho x_{it-3}$

Where y is the change in the retail rate (either lending or deposit) and x is the change in policy rate. Θ and u capture the cotemporaneous and short term effects respectively.

^2/Excludes countries that use the US Dollar as currency.

Source: See Appendix I

The main factors that appear to be influencing the transmission mechanism of monetary policy in CADR are in line with those discussed in the previous section (Figure 1). In particular, the transmission mechanism seems to be positively correlated to the degree of exchange rate flexibility, financial intermediation (measured by the amount of bank credit to the private sector in percent of GDP), and the institutional environment (measured by the World Bank‟s governance indicator on regulatory quality);⁹ and negatively correlated with the degree of financial dollarization, bank concentration,¹⁰ and fiscal dominance (measured by central bank claims on the government in percent of GDP).

Pass-through in Latin America ^1/

Interest-rate transmission is positively correlated with exchange rate volatility, financial development, and regulatory quality; and negatively correlated with financial dollarization, bank concentration, and fiscal dominance.

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ Trend line estimated from a linear regressionSource: See Appendix I

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Pass-through in Latin America ^1/

Interest-rate transmission is positively correlated with exchange rate volatility, financial development, and regulatory quality; and negatively correlated with financial dollarization, bank concentration, and fiscal dominance.

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ Trend line estimated from a linear regressionSource: See Appendix I

• Download Figure
• Download figure as PowerPoint slide

Pass-through in Latin America ^1/

Interest-rate transmission is positively correlated with exchange rate volatility, financial development, and regulatory quality; and negatively correlated with financial dollarization, bank concentration, and fiscal dominance.

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ Trend line estimated from a linear regressionSource: See Appendix I

• Download Figure
• Download figure as PowerPoint slide

In general, the speed of transmission of the policy rate to lending rates appears to be faster in the LA6 than CADR though it does vary by country. Figure 2 presents estimates of the cumulative correlation.¹¹ Within the LA6, the speed of transmission appears to be the fastest in Peru, with full transmission at about four months (though, as earlier indicated, it has the lowest long-term correlation coefficient, i.e., the weakest interest-rate transmission mechanism); most of the other countries achieve full transmission within six and eight months. Within CADR, Costa Rica has the fastest speed of transmission of the policy rate to lending rates (about six months), while the Dominican Republic, Guatemala, and Honduras have the slowest (within eight months and a year).

CADR + LA6: Policy to Lending Pass-through Cummulative Response Over Time¹

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

Source: Authors’ calculations. See Appendix III.¹ T equals the first month.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CADR + LA6: Policy to Lending Pass-through Cummulative Response Over Time¹

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

Source: Authors’ calculations. See Appendix III.¹ T equals the first month.

• Download Figure
• Download figure as PowerPoint slide

CADR + LA6: Policy to Lending Pass-through Cummulative Response Over Time¹

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

Source: Authors’ calculations. See Appendix III.¹ T equals the first month.

• Download Figure
• Download figure as PowerPoint slide

B. Interest-Rate Transmission: Empirical Evidence

This section measures the pass-through from policy rates to banks‟ lending rates through a panel estimation, incorporating the effects of dollarization, exchange rate flexibility, the size of the banking sector, and banking concentration.¹² The methodology follows in part Mishra, Montiel, and Splimbergo (2010), but restricts the sample to 40 countries that publish a policy rate using annual data from 2004–10. The panel regressions use: (i) the ratio of foreign currency deposits to total deposits in the banking system to measure financial dollarization; (ii) an index from 0 to 10 based on the standard deviation of the daily exchange rate within a one month period as a proxy of exchange rate flexibility, (iii) the ratio of bank deposits to GDP to measure the size of the banking sector, and (iv) a dummy for bank concentration (equal to one if the index is higher than the median and zero otherwise) to test whether a monopolistic banking sector hinders the pass-through.¹³ These variables are interacted with the policy rate in order to evaluate the way in which they condition the pass-through in the following equation.¹⁴

$y_{it}={\alpha }_0+{\beta }_1{y}_{it-1}+{\beta }_2{x}_{it}+\mu Z_{it}^{\prime }+\theta x_{it}{Z}_{it}^{\prime }+{\delta }^{\prime }+{\mathit{ϵ}}_{it}$

It is assumed that the impact of the policy rate (x_it) on the lending rate (y_it), or the interest rate pass-through, is affected by the level of deposit dollarization, exchange rate flexibility, deposits to GDP, and concentration of the banking sector $(z_{it}^{\prime })$ as they interact with the policy rate $(x_{it}{Z}_{it}^{\prime })$ . The interest rate pass-through is represented by the coefficient β₂ and δ′ denotes time dummies for every period. Following the previous subsections, the coefficients corresponding to the interactions of the policy rate with the size of the banking sector and exchange rate flexibility are expected to have a positive sign, while the interaction of the policy rate with deposit dollarization and banking sector concentration should have a negative sign. Since the independent variables are not strictly exogenous due to correlation with the error term (ϵ_it), OLS and least-squares dummy variables (fixed effects) models give biased results. A solution to this problem is given by the System-GMM estimator (Arellano and Bover, 1995; Blundell and Bond, 1998). This estimator is derived from the estimation of a system of two simultaneous equations, one in levels and the other in orthogonal deviations.¹⁵ For this panel, monthly data was collapsed into yearly averages in order to implement System-GMM (for large „n‟, small „t‟). We ran several tests and confirmed that there is no presence of unit root, thus the variables in the panel are in levels.

Table 3 presents the results of our model: a pooled OLS (column 1), a least-squares dummy variables with fixed effects (LSDV, column 2), and a System GMM (column 3), following Roodman (2006). Given the upward and downward bias of the OLS and LSDV estimates respectively, the lagged dependant variable should lie in between these two. Columns 1 to 3 present the model with no interactions, and show the relationship of the explanatory variables with the lending rate, while columns 4 to 6 include the interactions. In column 3, the Hansen test of 0.112 indicates that the instrument set of 24 variables can be considered valid and that there are no issues with over-identification. Similarly, in column 6, the Hansen test value of 0.149 suggests an instrument set of 40 instruments is appropriate. In order to build this set, all the variables after the second lag were used, which is the standard treatment for endogenous variables when there is autocorrelation in the first lag, or AR(1). The main results of the estimations are:

• The pass-through in the first year is about 0.55 (a one percentage change in the policy rate is associated with a 0.55 percentage point increase in the lending rate) in the GMM estimation with interactions (column 6).

• The interactions of the financial variables with the policy rate have the correct sign, and foreign currency deposits, exchange rate flexibility, and bank deposits as a percent of GDP are significant, although only at the 10 percent level. This is likely due to the fact that the financial variables are correlated. An F-test on the interactions of the four financial variables shows that they are jointly significant at the five percent level, highlighting that tackling dollarization, improving exchange rate flexibility, developing the financial sector, and lowering bank concentration are all important for increasing the interest-rate pass-through.

• The interaction with the policy rate suggests that higher dollarization tends to lower the pass-through. The relevant coefficient is negative and is statistically significant in the GMM estimation (column 6). In particular, an increase in the ratio of deposits in foreign currency by one standard deviation, from 29 to 53 percent, would reduce the pass-through by about 0.1 (Figure 3).

• Exchange rate flexibility has a significantly positive impact on the interest-rate pass-through. The results indicate that an increase in the index of exchange flexibility by one standard deviation (from 5.4 to 8.1) would increase the pass-through from about 0.50 to about 0.65. The relevant coefficient of the interaction of the policy rate and exchange rate flexibility is positive and statistically significant in columns 4 and 5, though the significance decreases in column 6.

• The size of the banking sector has a positive correlation with the pass-through. Were the ratio of deposits to GDP to increase by one standard deviation (from 35.1 to 61.2 percent), the pass-through would increase from 0.5 to 0.6. The relevant coefficient of the interaction of the policy rate and bank deposits to GDP is positive and statistically significant in equations 4 to 6.

• Although the coefficient on the interaction between bank concentration and the policy rate does show the expected negative sign, indicating concentration tends to lower the pass-through, it is only statistically significant in equation 4; this could be due to limitations in the way the concentration index is calculated.¹⁶ It may also reflect the fact that concentrated markets can be competitive if barriers to new entrants are low.¹⁸

Marginal Effects of Explanatory Variables on the Policy Rate Pass through ^1/

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each of the above mentioned covariates and averaging the rest. Following the equation in page 13, they can be expressed as:$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$Source: Author’s estimates.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Marginal Effects of Explanatory Variables on the Policy Rate Pass through ^1/

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each of the above mentioned covariates and averaging the rest. Following the equation in page 13, they can be expressed as:$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$Source: Author’s estimates.

• Download Figure
• Download figure as PowerPoint slide

Marginal Effects of Explanatory Variables on the Policy Rate Pass through ^1/

Citation: IMF Working Papers 2011, 240; 10.5089/9781463923228.001.A001

^1/ These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each of the above mentioned covariates and averaging the rest. Following the equation in page 13, they can be expressed as:$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$Source: Author’s estimates.

• Download Figure
• Download figure as PowerPoint slide

Table 3.

Estimation Results

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1All estimations include time dummies.

Table 3.

Estimation Results

		(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES		OLS	LSDV	System GMM Twostep Windmeijer SE correction	OLS	LSDV	System GMM Twostep Windmeijer SE correction
Dependent variable: Lending rate (percent)
Lagged lending rate (percent)		0.869*** (0.0158)	0.378*** (0.0510)	0.428*** (0.146)	0.848*** (0.0172)	0.343*** (0.0533)	0.618*** (0.0672)
Policy rate (percent)		0.191*** (0.0408)	0.499*** (0.114)	0.657*** (0.147)	0.277*** (0.0613)	0.526*** (0.0818)	0.551*** (0.114)
FX deposits (percent of total deposits)		-0.00102 (0.00536)	0.00721 (0.0376)	0.126* (0.0685)	-0.00121 (0.00534)	0.0221 (0.0338)	-0.00711 (0.0361)
FX flexibility (index)		-0.0546 (0.0464)	0.0232 (0.0980)	0.0226 (0.192)	-0.0601 (0.0456)	0.00728 (0.0925)	-0.118 (0.118)
Deposits (percent of GDP)		-0.000504 (0.00207)	-0.0101 (0.0118)	-0.00618 (0.00911)	0.00164 (0.00284)	-0.00159 (0.0146)	-0.000935 (0.00489)
Bank concentration (dummy)		-0.382 (0.272)	-0.553* (0.313)	-1.598 (1.156)	-0.359 (0.267)	-0.306 (0.306)	-0.209 (0.511)
Policy rate interacting with:
	FX deposits (percent of total deposits)				-0.000523 (0.00145)	-0.00341 (0.00293)	-0.00722* (0.00359)
	FX flexibility (index)				0.0419*** (0.0141)	0.0673*** (0.0203)	0.0657* (0.0355)
	Deposits (percent of GDP)				0.00148* (0.000769)	0.00129* (0.000656)	0.00197* (0.00101)
	Bank concentration (dummy)				-0.136* (0.0706)	-0.0355 (0.0747)	-0.107 (0.100)
Observations		235	235	235	235	235	235
R-squared		0.952	0.542		0.955	0.611
Adjusted R-squared		0.949	0.519		0.952	0.585
Number of countries			40	40		40	40
No. of instruments				24			40
Hansen test p-value				0.112			0.149
A-B AR(1) test				-1.839			-2.797
A-B AR(1) test p-value				0.0659			0.00516
A-B AR(2) test				-1.470			-1.709
A-B AR(2) test p-value				0.141			0.199
F-test p value (joint significance of financial variables)							0.0336**

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1All estimations include time dummies.

View Table

Table 3.

Estimation Results

		(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES		OLS	LSDV	System GMM Twostep Windmeijer SE correction	OLS	LSDV	System GMM Twostep Windmeijer SE correction
Dependent variable: Lending rate (percent)
Lagged lending rate (percent)		0.869*** (0.0158)	0.378*** (0.0510)	0.428*** (0.146)	0.848*** (0.0172)	0.343*** (0.0533)	0.618*** (0.0672)
Policy rate (percent)		0.191*** (0.0408)	0.499*** (0.114)	0.657*** (0.147)	0.277*** (0.0613)	0.526*** (0.0818)	0.551*** (0.114)
FX deposits (percent of total deposits)		-0.00102 (0.00536)	0.00721 (0.0376)	0.126* (0.0685)	-0.00121 (0.00534)	0.0221 (0.0338)	-0.00711 (0.0361)
FX flexibility (index)		-0.0546 (0.0464)	0.0232 (0.0980)	0.0226 (0.192)	-0.0601 (0.0456)	0.00728 (0.0925)	-0.118 (0.118)
Deposits (percent of GDP)		-0.000504 (0.00207)	-0.0101 (0.0118)	-0.00618 (0.00911)	0.00164 (0.00284)	-0.00159 (0.0146)	-0.000935 (0.00489)
Bank concentration (dummy)		-0.382 (0.272)	-0.553* (0.313)	-1.598 (1.156)	-0.359 (0.267)	-0.306 (0.306)	-0.209 (0.511)
Policy rate interacting with:
	FX deposits (percent of total deposits)				-0.000523 (0.00145)	-0.00341 (0.00293)	-0.00722* (0.00359)
	FX flexibility (index)				0.0419*** (0.0141)	0.0673*** (0.0203)	0.0657* (0.0355)
	Deposits (percent of GDP)				0.00148* (0.000769)	0.00129* (0.000656)	0.00197* (0.00101)
	Bank concentration (dummy)				-0.136* (0.0706)	-0.0355 (0.0747)	-0.107 (0.100)
Observations		235	235	235	235	235	235
R-squared		0.952	0.542		0.955	0.611
Adjusted R-squared		0.949	0.519		0.952	0.585
Number of countries			40	40		40	40
No. of instruments				24			40
Hansen test p-value				0.112			0.149
A-B AR(1) test				-1.839			-2.797
A-B AR(1) test p-value				0.0659			0.00516
A-B AR(2) test				-1.470			-1.709
A-B AR(2) test p-value				0.141			0.199
F-test p value (joint significance of financial variables)							0.0336**

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1All estimations include time dummies.

Table 4 presents estimates of the interest rate pass-through calculated at the mean values of the explanatory variables for the LA6 and CADR, derived from the system GMM (column 6 in Table 3). The overall results display similarities to the long-term correlations between the policy and lending rates in Table 1. On average, the LA6 have a higher interest-rate pass-through than CADR countries; and within the LA6, Brazil and Chile have the highest pass-through, while the dollarized economies of Peru and Uruguay have the lowest. Regarding CADR countries, the pass-through is relatively similar in the Dominican Republic, Costa Rica, and Guatemala. The pass-through estimates for CADR are also in line with independent estimates by Swiston (2011). The pass-through range for CADR is narrower than in Table 1, at around 0.5–0.7. However, the panel analysis excludes Nicaragua, as it does not have a policy rate.

Table 4:

Summary Means and Predicted Pass-Through: 2004-2010

^1/These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each the above mentioned covariates and averaging the rest. Following the equation in page 12, they can be expressed as:

$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$

Source: Author’s estimates.

Table 4:

Summary Means and Predicted Pass-Through: 2004-2010

	FX ratio	FX flexibility	Deposits	Pass-through 1/
Chile	1.71	7.68	57.43	1.15
Brazil	0.00	7.30	53.81	1.14
Colombia	0.00	6.99	15.65	1.04
Mexico	12.13	6.93	14.20	0.95
LA6	21.81	6.52	38.77	0.94
Uruguay	82.01	6.10	37.34	0.70
Peru	35.01	4.13	54.22	0.67
Dom.Rep	39.27	5.51	43.80	0.65
Costa Rica	49.95	4.36	41.82	0.61
Guatemala	32.20	3.80	30.22	0.57
CADR	38.08	3.59	39.67	0.57
Honduras	30.89	0.68	42.83	0.46

^1/These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each the above mentioned covariates and averaging the rest. Following the equation in page 12, they can be expressed as:

$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$

Source: Author’s estimates.

View Table

Table 4:

Summary Means and Predicted Pass-Through: 2004-2010

	FX ratio	FX flexibility	Deposits	Pass-through 1/
Chile	1.71	7.68	57.43	1.15
Brazil	0.00	7.30	53.81	1.14
Colombia	0.00	6.99	15.65	1.04
Mexico	12.13	6.93	14.20	0.95
LA6	21.81	6.52	38.77	0.94
Uruguay	82.01	6.10	37.34	0.70
Peru	35.01	4.13	54.22	0.67
Dom.Rep	39.27	5.51	43.80	0.65
Costa Rica	49.95	4.36	41.82	0.61
Guatemala	32.20	3.80	30.22	0.57
CADR	38.08	3.59	39.67	0.57
Honduras	30.89	0.68	42.83	0.46

^1/These marginal effects are computed as the derivatives of dy/dx calculated from predictions at fixed values of each the above mentioned covariates and averaging the rest. Following the equation in page 12, they can be expressed as:

$\frac{\partial y}{\partial x}={\beta }_2+\theta z^{\prime }$

Source: Author’s estimates.

These pass-through results for the individual CADR countries are also in line with the literature and are more intuitive than those implied by the long-term correlations in Table 1. In particular, the higher pass-through coefficients in the Dominican Republic, Costa Rica and Guatemala are consistent with relatively more advanced monetary policy frameworks. Indeed, Guatemala has already adopted IT, and Dominican Republic and Costa Rica have taken steps towards adopting this monetary framework and have deeper capital markets relative to other CADR countries. Still, the results suggest that the effectiveness of the IT framework in Guatemala is weakened by limited exchange rate flexibility and shallow capital markets (including underdeveloped secondary government securities and interbank markets). In Costa Rica, the Dominican Republic, and Honduras the results suggest that the interest-rate transmission mechanism could be strengthened by allowing greater exchange rate flexibility and reducing dollarization.

IV. Conclusion

This paper has examined the key factors that influence the strength of the interest-rate transmission mechanism, with particular emphasis in CADR countries. The results suggest that the pass-through of the policy rate to market rates is generally weaker and slower in CADR than in LA6 countries, with the main factors hindering the transmission mechanism in line with the literature. In particular, a correlation analysis suggests that the transmission mechanism could be strengthened by allowing greater exchange rate flexibility, reducing financial dollarization, lowering bank concentration, deepening financial markets, tackling fiscal dominance, and enhancing the regulatory and institutional environment.

The importance of some of these factors was assessed through panel data analysis. The pass-through of the policy rate to lending was estimated at about 0.55 in the sample of countries. The panel results suggest that the pass-through of the policy rate to lending rates has a negative and statistically significant relationship with dollarization, and a positive and statistically significant relationship with exchange rate flexibility and financial system development. While bank concentration was also found to have a negative relationship with the interest-rate transmission mechanism, its statistical significance was not confirmed. However, an F test suggested that that these four financial variables are jointly significant at the five percent level, providing support to the notion that reforms in all four areas can improve the interest-rate transmission mechanism.

A careful prioritization of the steps needed to bring CADR monetary frameworks to best international practices is essential to relax the existing constraints to an effective monetary policy. In the near term, the three main priorities are to increase exchange rate flexibility, preserve the absence of fiscal dominance, and ensure central bank instrument independence (Medina Cas, Carrión-Menéndez, Frantischek, 2011). Other conditions to enhance the monetary policy frameworks can be developed, and in some instances their fulfillment can be accelerated, once the above three steps are taken. These conditions include strengthening the political independence of central banks and tackling their balance sheet weaknesses, developing further the capacity of central banks for forecasting inflation, and continuing to improve the transparency and accountability of central banks. In addition, although not an immediate priority in the process of strengthening the monetary policy framework, it will be important to continue enhancing financial supervision and regulation and implementing structural reforms to develop and promote diversification of financial markets.