Content uploaded by Rodrigo Fuentes
Author content
All content in this area was uploaded by Rodrigo Fuentes
Content may be subject to copyright.
Banco Central de Chile
Documentos de Trabajo
Central Bank of Chile
Working Papers
N° 218
Agosto 2003
IS THERE LENDING RATE STICKINESS IN
THE CHILEAN BANKING INDUSTRY?
Solange Berstein Rodrigo Fuentes
La serie de Documentos de Trabajo en versión PDF puede obtenerse gratis en la dirección electrónica:
http://www.bcentral.cl/Estudios/DTBC/doctrab.htm. Existe la posibilidad de solicitar una copia
impresa con un costo de $500 si es dentro de Chile y US$12 si es para fuera de Chile. Las solicitudes se
pueden hacer por fax: (56-2) 6702231 o a través de correo electrónico: bcch@bcentral.cl.
Working Papers in PDF format can be downloaded free of charge from:
http://www.bcentral.cl/Estudios/DTBC/doctrab.htm. Printed versions can be ordered individually for
US$12 per copy (for orders inside Chile the charge is Ch$500.) Orders can be placed by fax: (56-2) 6702231
or e-mail: bcch@bcentral.cl.
BANCO CENTRAL DE CHILE
CENTRAL BANK OF CHILE
La serie Documentos de Trabajo es una publicación del Banco Central de Chile que divulga
los trabajos de investigación económica realizados por profesionales de esta institución o
encargados por ella a terceros. El objetivo de la serie es aportar al debate de tópicos
relevantes y presentar nuevos enfoques en el análisis de los mismos. La difusión de los
Documentos de Trabajo sólo intenta facilitar el intercambio de ideas y dar a conocer
investigaciones, con carácter preliminar, para su discusión y comentarios.
La publicación de los Documentos de Trabajo no está sujeta a la aprobación previa de los
miembros del Consejo del Banco Central de Chile. Tanto el contenido de los Documentos
de Trabajo, como también los análisis y conclusiones que de ellos se deriven, son de
exclusiva responsabilidad de su o sus autores y no reflejan necesariamente la opinión del
Banco Central de Chile o de sus Consejeros.
The Working Papers series of the Central Bank of Chile disseminates economic research
conducted by Central Bank staff or third parties under the sponsorship of the Bank. The
purpose of the series is to contribute to the discussion of relevant issues and develop new
analytical or empirical approaches in their analyses. The only aim of the Working Papers is
to disseminate preliminary research for its discussion and comments.
Publication of Working Papers is not subject to previous approval by the members of the
Board of the Central Bank. The views and conclusions presented in the papers are
exclusively those of the author(s) and do not necessarily reflect the position of the Central
Bank of Chile or of the Board members.
Documentos de Trabajo del Banco Central de Chile
Working Papers of the Central Bank of Chile
Huérfanos 1175, primer piso.
Teléfono: (56-2) 6702475; Fax: (56-2) 6702231
Documento de Trabajo Working Paper
N° 218 N° 218
IS THERE LENDING RATE STICKINESS IN
THE CHILEAN BANKING INDUSTRY?
Solange Berstein Rodrigo Fuentes
Economista Senior
Gerencia de Investigación Económica
Banco Central de Chile
Economista Senior
Gerencia de Investigación Económica
Banco Central de Chile
Resumen
Existe una amplia literatura que estudia la flexibilidad de la tasa de interés en diferentes países. En
este trabajo mostramos evidencia para la industria bancaria chilena, concluyendo que hay algún
grado de rigidez en el ajuste de la tasa de préstamos a cambios en la tasa de política. Sin embargo
Chile está entre los países que tienen tasa de interés más flexible. Usando datos de bancos
individuales y un modelo teórico, identificamos características de los bancos que podrían afectar el
grado de rigidez. Hechos estilizados y las estimaciones de panel sugieren que los bancos con
proporciones de cartera vencida más bajas y porcentajes más altos de personas dentro de su
portafolio se ajustan más rápido a los movimientos de la tasa de política.
Abstract
There is a vast literature that studies the flexibility of bank interest rates in different countries. In
this paper we show some evidence for the Chilean banking industry, concluding that there is some
sluggishness of adjustment of the bank-lending rates to changes in policy rate. However, Chile is
among the countries that have more flexible interest rate. On the basis of individual bank data and a
theoretical model we identified bank characteristics that might affect the degree of stickiness.
Stylized facts and estimation results suggest that banks with smaller portion of past-due loans and
higher percentage of household adjust faster to policy rate movements.
___________________
Helpful discussion with Rómulo Chumacero, Verónica Mies, Marcos Morales, Klaus Schmidt-Hebbel and
Rodrigo Valdés are acknowledged. We thank comments by Guillermo Larraín and the participants at the VI
Annual Conference of the Central Bank of Chile, Economic Modelling Conference and the Economic
Workshop at the P. Universidad Católica de Chile. Able research assistance provided by Dirk Mevis.
E-mail: rfuentes@bcentral.cl.
1
1. Introduction
This paper studies the transmission of the monetary policy in terms of the interest rate
pass-through in the case of Chile. Specifically, we are interested in the response of
commercial banks lending rate to a money market interest rate movement. International
evidence suggests that there is some sluggishness of adjustment of lending interest rates to
changes on the policy rate. In general this stickiness is related to lack of competition in the
banking sector, capital flow restrictions and volatility of the policy rate.
One of the first comprehensive empirical studies on bank interest rate pass-through for
monetary policy is Cottarelli and Kourelis [1994]. They found important differences
between countries. The estimated impact effects varied between 0.06 and 0.83, and the long
run effects ranged from 0.59 to 1.48 with an average of 0.97. Our estimates for the Chilean
case are an impact of 0.81 and a long run pass-through of 0.97 for nominal interest rates
1
.
Previous studies suggest that sluggishness of adjustment is associated to market
conditions and regulation of the banking sector. In this paper, by using data at the bank
level, we explore other factors that may influence the degree of delay in market interest rate
response to changes in the policy rate. The aim is to identify which characteristics may
explain the differences in the average rates charged by each bank and their responsiveness
to movements in the policy rate. The main variables considered were the size of the bank,
type of customers and the loan risk level, which are related to demand elasticity and cost of
adjustment for banks. A theoretical model presented in the paper motivates the choice of
these factors and dynamic panel data estimation supports the implications of the model.
The paper proceeds as follows. In Section 2 we briefly review the previous literature
and present our own estimations for the Chilean case, at an aggregate level. In Section 3 we
discuss some stylized facts for the Chilean banking industry and a model of monopolistic
competition with asymmetric information for bank lending rates, together with the panel
data econometric analysis. Finally in section 4 we summarize and present some concluding
remarks.
2
2. Chile versus the International Evidence
This section shows a brief literature review on empirical studies related to the
flexibility of the bank-lending rate in different countries. After the review, our own
estimations for Chile are presented and compared to what have been found for other
countries.
The lending interest rate stickiness refers to the small response of commercial banks
lending rate to a money market interest rate movement. Hannan and Berger [1989 and
1991] and Cottarelli and Kourelis [1994] provide arguments and evidence for short run
sluggishness of adjustment of the lending interest rate. They found that in the long run the
lending rate fully adjusts to the shift in the money market rate. After these studies many
papers have tested the monetary policy transmission for specific countries under different
periods and type of regulations. All of them are based on different parameterization of the
following basic model:
∑∑∑
=
−
==
−
∆+++=
p
l
llkt
n
k
k
m
j
jtjt
MPRmii
001
γαβδ [1]
Where i represents the bank-lending rate, m is the money market or interbank rate,
∆MPR is the change in the monetary policy interest rate. The difference between the money
market or interbank rate and the monetary policy rate is that the first two are interest rate
determined in the market, while the latter is set by the Central Bank as a target value. In
Chile monetary policy is conducted, as in many other countries, by managing liquidity such
that the interbank or money market rate is in line with the policy rate. Therefore, we can
separate the effect of monetary policy in two steps: from policy rate to money market rate
and from money market rate to lending rate; we are interested in the second step. The
coefficients of interest are α
1
that indicates the impact or the short run effect of the money
market or interbank rate on the lending rate. It is expected to be positive and less than or
1
Espinoza and Rebucci (2002) compare the degree of the stickiness in the Chilean banking sector with OECD
countries. They found that Chile is not different than those economies.
3
equal to one. The coefficient that measures the long run effect of the money market rate on
the lending rate is estimated as:
∑
∑
−
=
j
k
β
α
λ
1
[2]
This coefficient is expected to be positive and close to one in an industry that is
highly competitive.
2.1 Literature Review
In the empirical literature we found two types of studies. Those that analyze
monetary transmission mechanisms using cross-country data and those that give evidence
using time series data for specific countries. The first group computes impact and long run
effects for different countries and later they relate their findings with financial structures
and macroeconomic variables of the different economies included in the sample. The
second group goes for country case type of study to check if there are differences in the
monetary policy transmission over time and for different interest rates. The main idea of
both types of studies is to capture the effect of institutional features on the transmission of
the monetary policy.
One of the first comprehensive empirical studies on interest rate pass through for
monetary policy is Cottarelli and Kourelis [1994]. This study estimates equation [1] for 31
countries including developed and developing countries. They found important differences
across countries on the impact coefficient, but the long run coefficient tended to one in
most of the cases. In a second step they correlate the different coefficients with explanatory
variables that could explain the cross-country differences. The main finding here is that the
impact coefficient is highly correlated with the structure of the financial system.
Specifically the lending interest rate becomes more flexible when: the barriers to entry to
the banking industry are low, the share of private ownership in the banking system is high,
the constraints to the international capital movement do not exist and a market for
4
negotiable short-term instruments exists. Neither market concentration nor the existence of
market for instruments issued by firms’ affect the degree of stickiness of the interest rate.
An important policy implication obtained by Cottarelli and Kourelis is the relevance
of the discount rate or monetary policy rate as a policy instrument. In general they argue
that the movement in the discount rate are interpreted as a signaling that helps to reduce the
degree of stickiness, especially in those economies with a weak financial structure.
Borio and Fritz [1995] examine the relationship between the monetary policy rate;
money market rate and the lending rate for a group of the OECD countries. Great Britain,
Netherlands and Canada show a high short-run coefficient [above 0.7], on the other hand
Spain, Japan, Italy and Germany exhibit the highest degree of interest rate stickiness.
However, in the long run the pass through is more homogenous across countries and it gets
closer to one. They argue that the difference in the results for different countries may be
affected by the type o lending rate available. In fact interest rate for prime customer tend to
adjust faster than other interest rates.
Benoit Mojon [2000] analyzes the monetary policy transmissions across Euro area
countries. He also looks for the implications of different financial structures on the
stickiness of the retail interest rate. As Cottarelli and Kourelis, he finds large differences in
the short-run coefficients for different countries, ranging from 0.5 in Italy to 0.99 in
Netherlands
2
. The pass through coefficient is lower the higher is the volatility of the money
market rate and lower is the competition from other sources of finance [the level of banking
desintermediation]. Competition among banks reduces asymmetries through the interest
rate cycle; i.e. the size of the pass-through coefficient is less affected for upward movement
in the interest rate compared to downward movement.
A second group of studies concentrates their analysis in specific country cases.
Following the paper by Cottarelli and Kourelis [1994], Cottarelli, Ferri and Generale [1995]
explored why the transmission of the monetary policy rate is so slow in Italy. They found
that the high degree of stickiness is explained by the constraints to competition in the
banking and financial system. In general banks that operate in more competitive markets
tend to translate movements on money market rate into lending interest rate faster. This
conclusion is based not only on the international comparison of Italian banking industry
2
Toolsema, Sturm and Haan (2001) find similar results for the same group of countries.
5
with the rest of the countries, but from the data analysis at the individual bank level. The
stickiness of lending rate tend to decline with financial liberalization in Italy which is
consistent with the results using micro data for different banks and regions of that country.
Using the same methodology as previous studies Moazzami [1999] confirms that
interest rate stickiness in the US is higher than in Canada during the 70s and 80s. However,
the degree of flexibility has changed, for both countries, in opposite direction over the first
half of the nineties compared to previous decades. Thus the short-run pass through has
converged to around 0.40 for both Canada and US. The author attributes these changes to a
more competitive environment for the US banking system and less competitive for Canada.
Winker [1999] combined an adverse selection model with a marginal cost pricing
model to find an empirical equation where the lending and deposit rate depend on the
money market rate in the long run but not in the short run due to the adverse selection
problem. Based on the same argument he justified the lower speed of adjustment of the
lending rate toward its long run level compared with the deposit rate, since the short run
coefficient for the lending rate is much smaller than in the deposit interest rate case. Winker
provides evidence for his model for the case of Germany.
For the case of Spain, Manzano and Galméz [1996] use an interesting database that
allows analyzing the speed of interest rate adjustment for type of banks. They define four
groups of financial institutions: national banks specialized in commercial banking, saving
banks, foreign banks, and merchant banks. The degree of short-run interest rate response to
changes in the interbank rate varies greatly across groups from 0.25 to 0.75 in the short-
term impact coefficient. In the long run all of them, with the exception of saving banks,
have a total impact coefficient greater than one accordingly with the reported confidence
interval. In the case of saving banks the coefficient is strictly less than one. On the other
hand the deposit rate shows higher degree of stickiness in the short run and in the long run.
The impact coefficient ranges from 0.2 to 0.46 and the total impact varies between 0.63 and
0.81.
The following table summarizes the results of the literature reviewed.
6
Table 2.1. Panel A
Cross-country studies
Cross-country Studies Degree of Transmission Main Conclusions
Cottarelli and Kourelis[1994]
Sample: 31 countries
Short term: 0.06 to 0.83
Long term: 0.59 to 1.48 with
an average equal to 0.97
The degree of flexibility increases
with the elimination of capital flow
restrictions, lower barriers to
competition, private property in
the banking industry and the
existence of short-run instruments
Borio and Fritz [1995]
Sample: 12 OECD countries
Response to a simultaneous
change in policy and money
market rate
Short term: 0.0 to 1.08
Long term: 0.74 to 1.17
The type of lending interest rate
used could explain the differences
across countries. For some
countries the lending rate is
applied to the best larger customer
while for others the rates
correspond to retail banking.
Mojon, Benoît [2002]
Sample: Panel data of 6
European countries
Short term: 0.5 [Italy] to 0.99
[Netherlands]
Long run: Around 1 for all
countries
The flexibility of interest rate
increases with lower volatility of
the monetary policy interest rate,
higher external and within banking
industry competition
Table 2.1. Panel B
Country case studies
Cases Degree of Transmission Main Conclusions
Cottarelli, Ferri and Generale
[1995]
Italy
Short term: 0.07
Long term: 0.92
The degree of stickiness is
inversely related with the degree of
competition and financial
liberalization
Moazzami, B. [1999]
Canada and United States
Short term [CAN]: 0.46 to 1.1
Short term [USA]: 0.25 to 0.6
Long term [CAN]: 0.6 to 2.0
Long term [USA]: 0.8 to 1.2
The impact coefficient has
increased over time while in
Canada has moved in the opposite
direction. The reason for these
results could be found in the
changes in financial system
structure in those countries
Winker, P. [1999]
Germany
Short term: 0.1 [lending rate]
and 0.42 [deposit rate]
Long run coefficient tends to 1
The speed of adjustment to
changes in the money market rate
is lower in lending rates than in
deposit rate
Manzano and Galmés [1996]
Spain
Short term: 0.25-0.75 [lending
rate] and 0.2–0.5 [deposit rate]
Total impact: 0.66-1.2
[lending] and 0.63-0.81
[deposit]
The lending rate tends to response
faster in the short and the long run.
The type of customer affects the
degree of response
7
2.2 Chile compared to other countries
This section presents the results at the aggregate level for the Chilean Banking
industry. The lending rate at the aggregate level was constructed using a weighted average
of interest rate for individual banks; the weights were the total amount of loans in the
corresponding category. Figure 2.1 plots the lending interest rate and the interbank rate for
the period under analysis. Visual inspection shows that the lending rates follow very
closely the interbank interest rate.
Graph 2.1 Lending Interest Rate and Interbank Rate
0.0
0.5
1.0
1.5
2.0
2.5
1996 1997 1998 1999 2000 2001 2002
L e nding inte re st ra te fo r le ss tha n 3 0 da ys
L e nding inte re st ra te fo r 3 0 to 8 9 d a ys
Inte rba n k inte re st ra te
An important feature to take into account is that Chilean bank conduct several
transactions in pesos and in unidades de fomento [UF], which is a unit of account indexed
to the past inflation
3
. This unit of account is used for medium and long-term transactions.
3
See Schiller (2002) for a discussion about the use of indexed unit accounts around the world and the UF.
8
Therefore equation [1] was estimated for peso denominated loans and UF denominated
loans. The most common maturity for the former type of loans is less than 30 days (aprox.
50% of total nominal loans). For the latter the typical maturity is 90 to 360 days but mainly
concentrated around 90 days (aprox. 40% of total UF indexed loans). The next figure
presents the evolution of the lending interest rate for loans of longer maturity and the
interest rate on 90 days Central Bank Indexed Promissory Note (PRBC). Again, both
interest rates move closely together
4
.
Graph 2.2 Lending Interest Rate and 90 days PRBC
0
4
8
12
16
20
1995 1996 1997 1998 1999 2000 2001
L e n d i n g in te re s t r at es fo r 9 0 to 3 6 0 d a y s
9 0 d ay s P R B C
A model represented by equation (1) was estimated. The number of lags is
sufficiently high to make the error term white noise. Several papers estimate this equation
using different parameterization. The most popular one is the error correction model based
on the idea that the interest rates are not stationary. There are good economic arguments to
disregard that possibility for interest rate
5
. Nevertheless, to be skeptical, in the appendix,
4
The monetary policy is handled through the interbank interest rate. However the 90 days interest rate on
PRBC is good measure of the monetary policy for 90 days.
5
See Chumacero (2001) for a discussion of unit roots using economics.
9
different test for unit roots are presented. All of them reject the presence of unit roots; thus
the model was run in levels.
Table 2.2 presents the results for the interest rate applied to peso denominated loans.
Columns 1 and 3 show the results of equation (1) controlling by inflation; columns 2 and 4
take into account the dramatic increase in the interest rates during 1998, using a dummy
variable D98 that takes the value one for January to October of 1998. Despite the dummy
variable is statistically significant the overall conclusions do not change much. The impact
coefficient fluctuates between 0.7 to 0.8, while in all the cases the hypothesis of the long
run coefficient equal to one cannot be rejected. Therefore in the long run, on average, banks
fully adjust the lending rate to a change in the interbank interest rate.
10
Table 2.2
Interest Rate Transmission: Nominal Lending Rate
Variable 30 days lending 30 days lending 30-89 days 30-89 days
Interest rate Interest rate Interest rate Interest rate
Interbank rate 0.7932 0.8109 0.7122 0.7098
[14.7964]
**
[22.8482]
**
[12.6719]
**
[18.8454]
**
Interbank rate [t-1] -0.3355 -0.1670 -0.1994
[-3.8715]
**
[-1.8404] [-2.3729]*
Interbank rate [t-2] -0.3129 -0.3193 -0.2659 -0.3330
[-2.3391]
*
[-2.9958]
**
[-4.4942]
**
[-4.1670]
**
Interbank rate [t-3] 0.0750 0.0874
[2.2498]
*
[2.3841]
*
Interbank rate [t-4] -0.0560
[-2.1570]
*
Interbank rate [t-6] 0.0784
[3.4636]
**
MPR[t] – MPR[t-1] 0.0281 0.0259 0.0419 0.0406
[2.8474]
**
[3.2080]
**
[4.0445]
**
[4.2109]
**
Lending rate [t-1] 0.2865 0.5629 0.4583 0.4059
[3.0554]
**
[6.1349]
**
[4.0831]
**
[4.6310]
**
Lending rate [t-2] 0.2320 0.2750 0.1896 0.3185
[2.2617]
*
[2.8149]
**
[2.5192]
*
[3.2959]
**
Inflation [t-2] -0.1033 -0.0953 -0.2190 -0.5084
[-2.7302]
**
[-3.5682]
**
[-4.1982]
**
[-3.8040]
**
D98 0.4462
[3.9445]
**
D98* Interbank rate -0.3820
[-3.1078]
**
D98* Interbank rate [t-1] 0.3547 -0.1996
[2.9385]
**
[-4.8414]
**
D98*D[MPR] 0.2038
[4.6452]
**
Constant 0.1358 0.0473 0.1737 0.1538
[3.8643]
**
[1.2736] [3.4792]
**
[3.2508]
**
Long-run coefficient (λ)
0.9972 1.1017 1.0060 0.9604
[Wald test λ=1]
[0.0015] [0.3202] [0.0044] [0.0932]
R-squared 0.9554 0.9742 0.9466 0.9569
t-test in parenthesis
**
1% Significance;
*
5% Significance
Table 2.3 shows the results for indexed lending rate. Again the 1998 interest rate turmoil is
controlled, but it was not statistically significant except for July 1998. The inflation rate
was not included since the variables are indexed interest rates. The impact coefficient is
around 0.85, while the long-term coefficient is statistically equal to one.
11
Table 2.3
Interest Rate Transmission: Indexed Lending Rate
Variable 90-360 days lending 90-360 days lending
Interest rate Interest rate
PRBC 0.8575 0.8553
[63.3162]
**
[48.3335]
**
PRBC (t-1) -0.4324 -0.2931
[-4.9115]
**
[-4.7812]
**
PRBC (t-2) -0.0775 -0.0694
[-5.1854]
**
[-3.5892]
**
PRBC (t-4) 0.0357
[4.0652]
**
PRBC (t-5) -0.0245 -0.1674
[-1.7402] [-2.9301]
**
Lending rate (t-1) 0.6396 0.4940
[6.1577]
**
[7.4194]
**
Lending rate (t-5) 0.1643
[2.8632]
**
D98 (July) 1.6035
[9.1060]
**
Constant 0.8019 0.8342
[3.3145]
**
[4.6351]
**
Long-run coefficient (λ)
0.9953 0.9520
[Wald test λ=1]
[0.0757] [0.0404]
R-squared 0.9837 0.9924
t-test in parenthesis
**
1% Significance;
*
5% Significance
How are these results compared with the international evidence? Table 2.4 exhibits the
comparison between the coefficient reported in column 2 of Table 2.2 and Table 2.3. It is
easy to check that the estimates for Chile show a high flexibility of the banking interest
rate. In fact the estimation poses Chile close to Mexico and United Kingdom. According to
Cottarelli and Kourelis, the variables that tend to increase the interest rate pass through are
the degree of competition and financial liberalization. It is important to take into account
that the time periods are different for the countries included in Cottarelli and Kourelis
(1994) with respect to the present study. The former uses data for the 80s while the current
study uses data for the 90s. Relevant conditions for interest rate sluggishness have been
different in nineties than in previous decade.
12
Table 2.4
International comparison of the
Interest rate stickiness
Countries Impact Long Term
Chile (en $) 0.81 0.97
Chile (en UF) 0.86 0.95
Colombia 0.42 1.03
Mexico 0.83 1.29
Venezuela 0.38 1.48
Canada 0.76 1.06
United States 0.32 0.97
Germany 0.38 1.04
Italy 0.11 1.22
Spain 0.35 1.12
United Kingdom 0.82 1.04
Sources: Cottarelli and Kourelis (1994) and
Preliminary own estimation for Chile
3. Evidence for Chile at the Bank Level
In Section 2 we exposed some evidence in favor of interest rates stickiness, this was the
case for almost all of the countries that have been studied and it is also the case of Chile, up
to some extent.
6
It was also argued that previous studies suggest that sluggishness of
adjustment is related to market conditions and regulation of the banking sector. In this
section, using data at the bank level, we explore what factors may influence the degree of
delay in market interest rate response to changes in the policy rate.
For this purpose we analyze the differences in the levels of interest rates charged by
banks and the adjustment to changes in the policy rate. It is interesting to notice that in the
Chilean case we observe important divergence between the interest rates charged by banks,
moreover, there are significant differences within a bank depending on the kind of loan, the
type of customer, firm or household, or the amount of the loan. However, legislation
imposes a ceiling to the interest rate charged by loan category, which somewhat limits this
dispersion (50% above the average market interest rate by loan category
7
).
6
In Section 2 it was shown that impact effect of changes in policy rate were less than 1 for most of the
countries studied, including the Chilean case.
7
Recopilación de Normas Bancos y Financieras, Cap. 7-1 pp. 10, SBIF.
13
The aim is to identify which characteristics might explain the differences in the
average rates charged by each bank and their responsiveness to movements in the policy
rate. The main characteristics considered were the size of the bank, type of customers and
the loan risk level. Other variables such as solvency or liquidity were also considered, but
they didn’t show up significant for explaining differences on lending rates so the results are
not shown in the paper. The data used is at the bank level, we don’t have, at this point,
enough information with respect to the different transactions within a bank. This would be
a future extension subject to the availability of this information.
3.1 Stylized facts for the Chilean Banking Industry
In Table 3.1 and 3.2 we show that larger banks charged, on average, lower interest
rates than smaller banks. For smaller banks the nominal monthly rate was 1.21, whereas for
larger banks this rate was 1.16 for the period 1996-2002. In the case of the UF rate, smaller
banks showed on average a yearly rate of 8.55%, i.e. 3.5% higher than the average for
larger banks (8.26%). This evidence might support two alternative hypotheses. What is
called the “structure performance” hypothesis or the “efficiency structure” hypothesis.
Under the first hypothesis differences in prices would respond solely to imperfect
competition with differences in price elasticities across markets served by different banks.
The second would imply that there are cost advantages for lager banks together with some
degree of market imperfection that allows inefficient banks to survive, at least in the short
run.
In terms of loan risk, as expected, banks with a higher percentage of past-due loans
(more than 2%) charged, on average, higher interest rate to their clients. This is 11.1%
higher in the case of nominal rates and 8.6% in the case of UF rates, over the sample
period. When we compute simple correlation between lending rates and our indicator for
policy rate (interbank rate in the case of nominal interest rate and PRBC90 in the case of
UF interest rate),this correlation is smaller for banks with lower quality of loans. This may
be due to adverse selection problems in the sense that if interest rates increase only riskier
projects (with higher expected return) would stay in the market and the average quality of
the loan portfolio will decrease lowering bank’s profits. In this sense, banks will not
14
respond rapidly to an increase in the policy rate, especially in the case of banks with a
higher portion of past-due loans. On the other hand, if the policy rate decreases we would
expect less responsiveness from banks with a riskier portfolio, because for riskier clients it
is more difficult to move to other banks. Therefore, there is less incentive to decrease
interest rates for banks with a larger portion of past due loans, at least in the short run.
Table 3.1 Nominal Rates 30 ds and Correlation with Interbank Rate
By risk and type of customer
(Average 1996-2002)
Larger Banks
*
Loan Risk
***
Type of Customer
**
<2% >2% Total
<10% Rate
Correlation
#Banks
>10% Rate 1. 08 1.20 1.16
Correlation 0.90 0.86 0.88
#Banks 2 4 6
Total Rate 1. 08 1.20 1.16
Correlation 0.90 0.86 0.88
#Banks 2 4 6
*
Large Banks are the ones that have a market share over total loans of more than 5%.
**
Type of customer measured as percentage of households loans over total loans.
***
Risk measured as past due loans as percentage of total loans
Table 3.2 Nominal Rates 30 ds and Correlation with Interbank Rate
By risk and type of customer
(Average 1996-2002)
Smaller Banks
*
Loan Risk
***
Type of Customer
**
<2% >2% Total
<10% Rate 1.12 1.37 1.19
Correlation 0.83 0.76 0.81
#Banks 5 3 8
>10% Rate 1.25 1.21 1.23
Correlation 0.87 0.79 0.83
#Banks 3 3 6
Total Rate 1.17 1.27 1.21
Correlation 0.85 0.78 0.82
#Banks 8 6 14
*
Small Banks are the ones that have a market share over total loans of less than 5%.
**
Type of customer measured as percentage of households loans over total loans.
***
Risk measured as past due loans as percentage of total loans
15
Finally, in Tables 3.3 and 3.4 we analyze differences in interest rates charged by banks
classified by type of loan.
8
We are able to do this distinction only for smaller banks because
in the case of larger banks there is not much difference according to this category, since all
of them have more than 10% of household loans. So, for smaller banks we have two
groups, less than 10% of the loans given to households and more than 10%.
Table 3.3 Interest Rate UF 90 ds to 1 yr and Correlación with PRBC rate
By risk and type of customer
(Average 1996-2002)
Large Banks
*
Loans Risk
***
Type of Customer
**
<2% >2% Total
<10% Rate
Correlation
#Banks
>10% Rate 8.02 8.38 8.26
Correlation 0.95 0.94 0.95
#Banks 2 4 6
Total Rate 8.02 8.38 8.26
Correlation 0.95 0.94 0.95
#Banks 2 4 6
*
Large Banks are the ones that have a market share over total loans of more than 5%.
**
Type of customer measured as percentage of households loans over total loans.
***
Risk measured as past due loans as percentage of total loans
Table 3.4 Interest Rate UF 90 ds to 1 yr and Correlation with PRBC rate
By risk and type of customer
(Average 1996-2002)
Small Banks
*
Loans Risk
***
Type of Customer
**
<2% >2% Total
<10% Rate 8.17 9.14 8.52
Correlation 0.92 0.80 0.87
#Banks 5 3 8
>10% Rate 8.38 8.80 8.59
Correlation 0.91 0.94 0.92
#Banks 3 3 6
Total Rate 8.25 8.96 8.55
Correlation 0.92 0.87 0.90
#Banks 8 6 14
*
Small Banks are the ones that have a market share over total loans of less than 5%.
**
Type of customer measured as percentage of households loans over total loans.
***
Risk measured as past due loans as percentage of total loans
8
The type of loan is measured as the percentage of total loans made to households (consumption plus
mortgage).
16
It is interesting to notice that in the case of nominal interest rates and UF interest rates for
smaller banks, the higher average rate charged corresponds to banks that have a larger
portion of past-due loans and lower share of household loans. While the lower interest
charged is in the case of banks with low risk and low share of households. This indicates
that there is an important dispersion of interest rates charged to companies, which seems to
be larger than in the case of households. This evidence suggests that households demand
elasticity is larger than in the case of firms. A possible explanation for this is that due to
asymmetric information companies establish a long run relationship with a banks at a
higher extent than households, which gives additional market power to banks, due to higher
switching costs for firms.
3.2 A model for lending interest rate stickiness
In this section we will present a model that will help us to build on some hypothesis that
we test for the Chilean banking industry. These hypotheses are related to the stylized facts
presented in the previous section. This model gives us some insights about what we might
expect from our empirical analysis and some possible explanations for our findings.
It seems appropriate to assume an imperfect competition model in the case of the
banking sector, where it is argued that there are significant barriers to entry or an important
degree of product differentiation.
9
Besides, it is also suitable to assume that there is
asymmetric information in this industry, which leads to adverse selection and moral hazard
problems. We will combine these two issues by assuming that banks make a two step
decision, which considers the long run equilibrium and the short-run behavior that will take
them to this condition.
10
For the long run let us assume a simple Monte-Klein model for a monopolistic bank
that faces a downward sloping demand for loans L(i
L
) and an upward sloping supply of
deposits D(i
D
). This is capturing the fact that banks have some monopoly power. The
decision variables for the firm are the quantities of loan (L) and deposits (D). Bank k
maximizes the following profit function:
9
Freixas and Rochet (1998).
10
This way of combining these two factors is similar to Scholnick (1991) and Winker (1999)
17
),())()1(())((),(
,, kkkkDkkLkk
LDCDDimLmLiDL −−−+−= αγπ
(3)
Where γ
k
is the probability that the loan will be repaid, m is the interbank rate (which is
given for individual banks), α is the proportion of deposits that constitutes cash reserve, i
D
is the deposit interest rate and i
L
is the lending interest rate. C(D,L) accounts for the total
cost of intermediation services, which is a function of the total amount of deposits and
loans.
Solving for the first order conditions and rearranging terms we get to the following
expressions for the lending interest rate:
[ ]
L
kk
k
L
Cmi '
)1(
*
+
−
=
γε
ε
(4)
Where, ε
k
is the absolute value of the demand elasticity for loans, which is greater than 1
since we are assuming monopolistic competition. For the purpose of this paper we are
interested on the loans market and we will assume that costs are separable, so that the
optimal lending rate is independent of the characteristics of the deposit market. This simple
model leads us to conclude that different interest rates charged on loans may reflect
different demand elasticity and the probability of loan repayment (portfolio risk).
The previous model is interpreted as the long run equilibrium for the banks. To
simplify our model we assume a constant elasticity demand function faced by each bank.
This is that ε might be different for each bank, but is independent of i
L
. We can write this
relationship between lending interest rate and interbank rate as: mi
kL
Φ=
*
(
kkkk
γεε )1/( −=Φ is a mark up, which is a function of demand elasticity and the
repayment probability). Thus, the long-run pass-through coefficient is larger the smaller is
the demand elasticity and the smaller is the probability of repayment. This long run
coefficient may or may not be equal to 1, when there is monopoly power to some extent.
However, due to asymmetric information, there might be some sluggishness in the
adjustment process to get to this long run equilibrium. In fact we are interested in finding
out if there is some delay in the response of market interest rates to changes in policy rate
18
and if this delay depends on banks’ characteristics, that would be related to demand
elasticity and asymmetric information.
Specifically we are thinking of a setup where in the short-run banks solve an inter-
temporal problem where they have on the one side a cost of adjusting too slowly to this
long run equilibrium and on the other side a cost of moving too fast. This last cost is due to
adverse selection and moral hazard problems in the banking industry. For instance if a bank
increases the lending rate as a response to an increase of the money market rate, to adjust to
the new long run equilibrium, it may end up attracting debtors that have a lower repayment
probability and lowering its profits. At the same time there is a moral hazard problem
because in face of a higher interest rate debtors would have incentives to invest in riskier
projects which would also decrease banks’ profits.
11
So under this framework we assume
that there are some adjustment costs due to asymmetric information. This is modeled as a
quadratic lost function following Nickell (1985), Scholnick (1991) and Winker (1999),
which is tractable because from it we get a linear decision rule.
12
The loss function for
bank k in period t is the following:
( ) ( )
[
]
∑
∞
=
−++++
−+Φ−=Γ
0
2
1,,,,,2
2
,,,1,
s
stLkstLkkstkstLkk
s
kt
iimi ωωδ
(5)
Where ω
1
and ω
2
represent the weight that the bank gives to achieving the long run target
value for lending interest rate and the cost of moving to that target value, respectively.
Recall that Φ
k
is a function of the demand elasticity and the probability of repayment that
bank k faces. On the other hand, ω
j
, j=1,2, will depend on bank’s average loan risk. We
might expect that if the portion of past due loans for bank k is higher, the adverse selection
or moral hazard problem for that bank is more important and it will give more weight to
changes in interest rate, which would imply a slower adjustment. From minimizing (5) we
obtain:
11
Stiglitz and Weiss (1981)
12
Scholnick (1991) and Winker (1999) include also a third term in this loss function, but it is not included in
this setup. For an argument see Nickell (1985). The other difference is that we have a multiplicative mark-up
instead of an additive mark-up.
19
1,,
,2,1
,2
,2,1
,1
,, −+++
+
+Φ
+
=
stLk
kk
k
stk
kk
k
stLk
imi
ωω
ω
ωω
ω
(6)
From (6) we can see that the impact coefficient depends on the relative size of ω
1,k
respect to ω
1,k
+ω
2,k
and the mark up Φ
k
. Therefore, the long run coefficient is always larger
than the short-term coefficient. On the other hand Φ
k
and ω
2,k
depends on the bank’s loan
risk. The lower the probability of repayment (higher risk) the higher is Φ
k
and the larger is
ω
2,k
. If the debtors are too risky and the effects on ω
2,k
is more important, the bank may not
pass completely a money market interest rate increase (in the short run) because it will
stifle the debtors. But in the long run the interest rate charged will be according to the risk
characteristic of the debtor. In other words we should expect a negative effect of unpaid
loan over the impact coefficient and positive effect on the long-term multiplier.
The main difference between our setup and the one presented by Scholnick (1991)
and Winker (1999) is that they derive an error correction model (ECM) from this quadratic
lost function. However, in our case even if we are assuming that there is a long run
relationship between the interbank rate and the lending rate, our variables are stationary so
our econometric model will be estimated in levels and not in an ECM form. Recall that the
ECM has this interpretation only if the variables are non-stationary and cointegrated, which
is not the case for our data.
13
The other important difference is that we use the above model in panel data estimation
in section 3.3 that allows the parameters to be different for different banks depending on
their characteristics.
3.3 Econometric Results
The model described above suggests that differences in interest rate pass-through might be
related to product characteristics such as: type of customer or risk level of the loan
portfolio. The econometric analysis presented in this section will allow us to address this
issue by estimating a dynamic panel data model where bank characteristics are interacted
with the interbank rate and its lags.
20
An alternative method is time series estimation by bank, but it has the drawback that
changes in bank characteristics during this time may be affecting the sluggishness of
adjustment for each bank, which is not correctly captured.
14
We estimate the following equation, which is based on the model described in
section 3.2. Considering that there is adverse selection captured by the adjustment cost
coefficient of the model, which is a function of the quality of loan portfolio. Besides, we
allow demand elasticity to be a function of the type of customers the bank has and the size
of the bank.
∑∑∑∑∑
=
−−
=
−−
=
−−
=
−
=
−
∆+++++=
p
l
lltkt
n
k
kthkkt
n
k
kthkkt
n
k
kthk
m
j
jthjhth
TPMmcmsmlii
00
,
0
,
0
,
0
,,
φγδαβη (7)
Where l is the loan portfolio risk measured as the portion of past-due loans, c is the type of
customers measured as the share of household loans (consumption and mortgage), s is the
bank size measured, as the percentage of total loans, and η
h
is a bank specific effect.
The problem of estimating dynamic panel data has been widely discussed in the
literature and different methods have been proposed to obtain consistent estimates of the
parameters. Anderson and Hsiao (1981) proposed method based on instrumental variable,
which consist of taking first differences of the equation to eliminate unobserved
heterogeneity and then use instrumental variables to estimate consistently the parameters of
the lag dependent variables.
For instance, let’s assume that the following equation is to be estimated using panel
data:
itiititit
uxyy
+
+
+
=
−
η
β
ρ
1
(8)
Where y
it
represents the lending interest rate, x
it
represents a dependent variable like
the interbank interest rate, η
i
is the unobserved heterogeneity. Taking first difference the
equation to be estimated is:
11211
)()(
−−−−−
−
+
−
+
−
=
−
itititititititit
uuxxyyyy
β
ρ
(9)
13
Unit Root tests is presented in the appendix. Derivation of the ECM and explanation of why it is not
appropriate with stationary data are found in Nickell (1985) and Wickens and Breush (1988).
14
See Berstein and Fuentes (2003) for time series estimations at the bank level.
21
Anderson and Hsiao propose y
i,t-2
or (y
i,t-2
- y
i,t-3
) as instrument for (y
i,t-1
- y
i,t-2
). But
Arellano (1989) showed that y
i,t-2
is a much better instrument for a significant range of
values of the true ρ in equation (9).
Arellano and Bond (1991) proposed an alternative methodology based on GMM
estimators. This method used several lags of the variables included as instruments, so it is
especially efficient when T is small and N is large
15
. The method is applied to equation (6),
using moment restrictions that come from the use of instrumental variables. Judson and
Owen (1999) provided evidence that for small T, GMM is a better estimator than Anderson
and Hsiao’s methods under the mean square error criterion. But for unbalanced panel data
and T around 20 is unclear what method is better.
Based in the traditional within group estimator, instrumental variable and GMM
several other methods have been developed. However, the most of instrumental variable
type of method work better than the within group estimator when N tends to infinity (N is
very large) and T is fixed. In a recent paper Alvarez and Arellano (2002) show the
asymptotic property of the within group, GMM and LIML estimators. An important result
for our case is that, regardless the asymptotic behavior of N, when T goes to infinity the
estimator of ρ is consistent. Moreover, if lim(N/T)=0 there is no asymptotic bias in the
asymptotic distribution of the within group estimator, while in the opposite case if
lim(T/N)=0 there is no asymptotic bias in the asymptotic distribution of the GMM
estimator. In the case of our panel T is large and it will increase as the time goes by, while
N will remain relatively fixed, thus the traditional within group estimator will provide
better results.
16
Tables 3.11 and 3.12 show the results for the 30 days nominal interest rate and for
the 90 to 360 days-indexed interest rates, respectively. The first column of Tables 3.11 and
3.12 present the results of the panel estimation without controlling for the 1998 effect and
without considering the interaction between bank characteristics and the right hand side
variables. If we compare these regressions with the ones from section 2 we observe that
impact and long run effects (shown at the bottom of each table) are smaller than what we
found previously. Notice that previously, at an aggregate level, we were estimating impact
15
See Judson and Owen (1999) for further discussion on the advantages of different methodologies.
22
and long run effects by using the weighted average interest rates, so that large banks were
driving the results to a higher extent on those regressions than on the panel data estimation.
Table 3.11 Panel with interaction and 1998 dummies
Dependent Variable: Nominal Rate 30 ds Model [1] Model [2] Model [3]
Interbank Rate 0,74 0,72 0,74
[41,51]
**
[34,80]
**
[24,92]
**
Interbank Rate [-1] -0,30 -0,41 -0,48
[-10,86]
**
[-14,44]
**
[-13,02]
**
Interbank Rate [-5] -0,12 -0,06
[-6,91]
**
[-3,84]
**
Interbank Rate [-6] -0,06
[-2,43]
**
Nominal Rate 30ds [-1] 0,57 0,67 0,68
[26,84]
**
[32,80]
**
[28,36]
**
Nominal Rate 30ds [-3] 0,05
[3,44]
**
Nominal Rate 30ds [-6] 0,14 0,06 0,04
[6,58]
**
[4,05]
**
[2,72]
**
D[TPM] 0,04 0,03 0,06
[8,62]
**
[5,71]
**
[7,08]
**
Inflation [-2] -0,13 -0,08 -0,09
[-6,83]
**
[-4,60]
**
[-3,65]
**
Interbank * Risk [-1] -2,31
[-2,13]
*
Interbank [-1] * Risk [-2] 5,05
[4,80]
**
Interbank [-1] * Part [-1] -0,72
[-2,84]
**
Interbank * Cons 0,18
[1,77]
Long-term coefficient 1,07 0,88 1,09
(Standard Deviation) [0,07] [0,06] [0,08]
Observations 1447 1447 1105
Number of banks 20 20 20
Model [2] and [3] control for the year 1998. The model were estimated using fixed effect, which are not
reported
16
See Berstein and Fuentes (2003) for panel data estimations using Anderson and Hsiao and Arellano Bond
methods.
23
Table 3.11 Panel with interaction and 1998 dummies
Dep Variable: UF Rate 90ds to 1 year Model [1] Model [2] Model [3]
PRBC 0.88 0.71 0.72
[90.95] [37.19] [25.41]
PRBC [-2] 0.05 -0.03
[2.62] [-2.30]
PRBC [-3] -0.38 -0.21 -0.21
[-12.22] [-7.61] [-6.54]
PRBC [-4] -0.09
[-3.10]
PRBC [-5] -0.05 -0.13 -0.13
[-3.98] [-4.98] [-4.18]
PRBC [-6] -0.09 -0.05
[-3.31] [-1.96]
UF Rate 90ds to 1 year [-1] 0.25 0.24 0.19
[14.10] [19.36] [12.13]
UF Rate 90ds to 1 year [-3] 0.26 0.24 0.24
[9.41] [9.52] [8.02]
UF Rate 90ds to 1 year [-4] 0.09
[3.19]
UF Rate 90ds to 1 year [-5] 0.12 0.12
[4.32] [4.20]
UF Rate 90ds to 1 year [-6] 0.09 0.05
[3.47] [2.19]
D [TPM [-1]] -0.34
[-6.14]
PRBC [-2] * Risk [-3] -2.48
[-4.12]
UF Rate 90ds to 1 year [-1] * Risk [-2] 1.47
[3.34]
PRBC * Part -0.34
[-3.11]
PRBC [-2] * Cons [-2] 0.18
[3.83]
Long-term coefficient 1.04 0.84 0.85
[Standard Deviation] [0.03] [0.03] [0.04]
Observations 1368 1368 990
Number of banks 18 18 18
Model [2] and [3] control for the year 1998. The model were estimated using fixed effect, which are not
reported
The second column of Tables 3.11 and 3.12 present the results of the panel
estimation controlling for the 1998 effect. The impact and the long run coefficient decrease
24
respect to those reported in the first column of each table, but the values are consistent with
the idea that the long-term coefficient is larger than the short term. However the long-term
coefficient is not statistically equal to 1. The last column in each table allows us to check
the hypotheses provided by the theoretical model. In the case of nominal interest rate, the
riskier is the portfolio the lower is the impact coefficient, which is consistent with the idea
that in the short run banks will not pass interest rate change to debtors, according to the
difference equation (6). But in the long run the pass through will be larger the riskier is the
portfolio. This relationship can be represented in figures 3.3.1 and 3.3.2 where we show
how the average loan risk has increased over time and the estimated impact effect has
decreased while the long run effect gets larger.
Figure 3.3.1
Impact Effect and Loans Risk
Nominal Rate 30 ds
0.69
0.7
0.71
0.72
0.73
0.74
0.75
12-1996
04-1997
08-1997
12-1997
04-1998
08-1998
12-1998
04-1999
08-1999
12-1999
04-2000
08-2000
12-2000
04-2001
08-2001
12-2001
04-2002
0
0.005
0.01
0.015
0.02
0.025
0.03
Loan Risk
Impact
In the case of the indexed interest rate, the results are different. The impact
coefficient is not affected by the portfolio risk, while the level of the unpaid loans affects
the long run coefficient by reducing it. Again in graph 3.3.1 we observe this relationship.
Finally, for both nominal and indexed rate, bank size affects negatively the pass
through, while banks more oriented toward households have a larger pass through.
25
Figure 3.3.2
Long Run Effect and Loans Risk
Nominal Rate 30 ds
0.7
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
12-1996
06-1997
12-1997
06-1998
12-1998
06-1999
12-1999
06-2000
12-2000
06-2001
12-2001
06-2002
0
0.005
0.01
0.015
0.02
0.025
0.03
Loan Risk
Long Run
Figure 3.3.3
Long Run Effect and Loan Risk
UF Rate 90ds to 1yr
0.8
0.81
0.82
0.83
0.84
0.85
0.86
0.87
0.88
0.89
12-1996
06-1997
12-1997
06-1998
12-1998
06-1999
12-1999
06-2000
12-2000
06-2001
12-2001
06-2002
0
0.005
0.01
0.015
0.02
0.025
0.03
Long Run
Loan
Risk
26
4. Concluding remarks
According to the estimates presented in this paper Chile shows a high flexibility of the
banking interest rate. In fact the estimation poses Chile close to Mexico and United
Kingdom, countries with the highest degree of flexibility. A previous study, Cottarelli and
Kourelis (1994), identify the degree of competition and financial liberalization as main
determinants of the interest rate stickiness.
By using data at the bank level, we explored other factors that influence the degree of
delay in market interest rate response to changes in the policy rate. In this sense, we have
analyzed the differences in the levels of interest rates charged by banks and the adjustment
to changes in the policy rate. The main characteristics identified here are the size of the
bank, type of customers and the loan risk level.
In the econometric analysis at the bank level we found significant differences in the
response of banks to changes in the policy interest rate. Moreover, the smaller the size of
the bank, the lower the portion of past-due loans and the larger the share of household
consumers, the faster is the response of lending interest rates to movement in the money
market rate. Results that are consistent with the model and the stylized facts presented in
the paper.
Topics of future research might include alternative measures that capture loan risk and
other characteristics that would help to have better measures of different demand
elasticities, at the bank level. Furthermore, with more disaggregate information of interest
rates charged for different types of loans within a bank, it would be possible to have better
estimates of the effects of loan risk or type of customer over the interest rate responses to
changes in policy rates.
27
References
Alvarez, J. and M. Arellano, “The Time Series and Cross-Section Asymptotics of Dynamic
Panel Data Estimators”, 2002, mimeo CEMFI.
Anderson, T. W. y C. Hsiao (1982). “ Formulation and Estimation of Dynamic Models
Using Panel Data”, Journal of Econometrics 18:47-82.
Arellano, M., “Panel Data Models: Some Recent Developments”, November 2000, CEMFI
Working Paper, Nº 16.
_________ and Bond, S., “Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment”, 1991, The Review of Economic Studies, 58,
277-297.
Bagiano, F., Dalmazzo, A. and Marini, G., “Bank Competition and ECB’s Monetary
Policy”, Journal of Banking and Finance, 2000, 24, 967-83.
Berstein, S. y Fuentes, J.R. “De la Tasa de Política a la Tasa de Colocación Bancaria: La
Industria Bancaria Chilena, Economía Chilena, 2003, 6, 49-67.
Bond, G., “Retail Bank Interest Rate Pass-Through: New Evidence at the Euro Area
Level”, April 2002, European Central Bank Working Paper Series, Nº 136.
Borio, C. E. V. and Fritz, W. “The Response of Short - Term Bank Lending Rates to Policy
Rates: A Cross - Country Perspective” Working Paper Nº 27, Bank for International
Settlements, 1995.
Corvoisier, S., and Gropp, R., “Bank Concentration and Retail Interest Rate”, Journal of
Banking and Finance, 26, 2002, 2155-2189.
Cottarelli, C. and Kourelis, A., “Financial Structure, Bank Lending Rates, and the
Transmission Mechanism of Monetary Policy”, IMF Staff Paper, December 1994, 41, Nº4,
587-623.
__________, Ferri, G. and Generale, A., “Bank Lending Rates and Financial Structure in
Italy: A Case Study”, IMF Working Paper, April 1995, Nº 38.
Duguay, P., “Empirical Evidence on the Strength of the Monetary Transmission
Mechanism in Canada”, Journal of Monetary Economics, 1994, 39-61.
Espinoza M. and Rebucci, A., “Retail Bank Interest Rate Pass-Through: Is Chile Atypical?”
paper presented to the Sixth Annual Conference of the Central Bank of Chile, 2002.
Freixas, X. y Rochet, J. (1997), Microeconomics of Banking. The MIT Press.
28
Hannan, T., and Berger, A. “The Rigidity of Prices: Evidence from the Banking Industry”,
The American Economic Review, September 1991, 81, Nº4, 939-945.
__________, “The Price-Concentration Relationship in Banking”, The Review of
Economics and Statistics, May 1989, 71, 291-9.
Judson, R. And Owen, A., “Estimating Dynamic Panel Data Models: A Guide for
Macroeconomists”, Economic Letters, 1999, 65, 9-15.
Manzano, M. C. and Galmés, S., “Credit Institutions Price Policies and Type of Customer:
Impact on the Monetary Transmission Mechanism”, Banco de España, Documento de
Trabajo, Nº 9605.
Moazzami, B., “Lending Rate Stickiness and Monetary Transmission Mechanism: The
Case of Canada and the United States”, Applied Financial Economics, 1999, 9, 533-38.
Mojon, B., “ Financial Structure and the Interest Rate Channel of ECB Monetary Policy”,
November 2000, European Central Bank Working Paper Series, Nº 40.
Nickell, S., “Error Correction, Partial Adjustment and all that: An Expository Note”,
Oxford Bulletin of Economics and Statistics, 1985, 47, 2, 119-129.
Pesaran, M. H. and Smith, R., “Estimating Long-run Relationships from Dynamic
Heterogeneous Panels”, Journal of Econometrics, 1995, 68, 79-113.
Rotemberg, J. and Saloner, G., “The Relative Rigidity of Monopoly Pricing”, The
American Economic Review, December 1987, 77, Nº5, 917-926.
Sack, B. and Wieland V., “Interest-Rate Smoothing and Optimal Monetary Policy: A
review of Recent Empirical Evidence”, Journal of Economics and Business, 2000, 52, 205-
228.
Schiller, R., (2002) “Indexed Units of Account: Theory and Assessment of Historical
Experience”. In Lefort and Schmidt-Hebbel (editors) Indexation, Inflation and Monetary
Policy, Series on Central Banking Analysis, and Economic Policy. Central Bank of Chile.
Nº 2: 105-34.
Scholnick, B., “Testing a Disequilibrium Model of Lending Rate Determination: The Case
of Malaysia”, IMF Working Paper, September 1991, Nº 84.
Sellon, G., “The Changing U.S. Financial System: Some Implications for the Monetary
Transmission Mechanism”, Federal Reserve Bank of Kansas City, Economic Review, First
Quarter 2002.
Tkacz, G., “Endogeneus Thresholds and Tests for Asymmetry in US Prime Rate
Movementes”, Economic Letters, 2001, 73, 207-11.
29
Toolsema L., Sturm, J. and Haan, “Convergence of Monetary Policy Transmission in EMU
New Evidence”, April 2001, CESifo Working Paper, Nº 465.
Winker, P., “Sluggish Adjustment of Interest Rates and Credit Rationing: An Application
of Unit Root Testing and Error Correction Modelling”, Applied Economics, 1999, 31, 267-
277.
30
Appendix
Unit Root Test
(1995-2001)
ADF DF-GLS
PRBC -1.928 -1.949 * -2.630 -1.995 *
Interbank Rate -3.733 * -3.175 * -4.364 ** -3.135 *
UF 90 ds. to 1 yr
-2.258 -2.292 * -2.204 -2.134 *
Nominal Rate 30 ds. -4.619 ** -4.612 ** -4.686 ** -3.562 **
* Non-stationarity rejected at 5%
** Non-stationarity rejected at 1%
The tests consider a trend for the nominal rates and the Modified Akaike was used to choose the number of lags.
By using ADF y P-P with the Modified Akaike we solve the size problem of this tests but the power is very low.
The power of the tests is higher when using DF-GLS y P-P Ng
Phillips-Perron
Phillips-Perron Ng
Mzt
Documentos de Trabajo
Banco Central de Chile
Working Papers
Central Bank of Chile
NÚMEROS ANTERIORES PAST ISSUES
La serie de Documentos de Trabajo en versión PDF puede obtenerse gratis en la dirección electrónica:
http://www.bcentral.cl/Estudios/DTBC/doctrab.htm. Existe la posibilidad de solicitar una copia
impresa con un costo de $500 si es dentro de Chile y US$12 si es para fuera de Chile. Las solicitudes se
pueden hacer por fax: (56-2) 6702231 o a través de correo electrónico: bcch@bcentral.cl.
Working Papers in PDF format can be downloaded free of charge from:
http://www.bcentral.cl/Estudios/DTBC/doctrab.htm. Printed versions can be ordered individually for
US$12 per copy (for orders inside Chile the charge is Ch$500.) Orders can be placed by fax: (56-2) 6702231
or e-mail: bcch@bcentral.cl.
DTBC-217
Macroeconomic Policies and Performance in Latin America
César Calderón y Klaus Schmidt-Hebbel
Julio 2003
DTBC-216
Openness and Imperfect Pass-Through:
Implications for the Monetary Policy
Claudio Soto y Jorge Selaive
Junio 2003
DTBC-215
Purchasing Power Parity in an Emerging Market Economy:
A Long-Span Study for Chile
César Calderón y Roberto Duncan
Junio 2003
DTBC-214
Non-Traded Goods and Monetary Policy Trade-Offs
in a Small Open Economy
Claudio Soto
Junio 2003
DTBC-213
Do Free Trade Agreements Enhance the Transmission
of Shocks Across Countries?
César Calderón
Junio 2003
DTBC-212
Crisis Financieras Internacionales, Prestamista de Última
Instancia y Nueva Arquitectura Financiera Internacional
Esteban Jadresic, Klaus Schmidt-Hebbel y Rodrigo Valdés
Junio 2003
DTBC-211
Reserves over the Transitions to Floating and to Inflation
Targeting: Lessons from the Developed World
Fernando Aportela, Francisco Gallego y Pablo García
Mayo 2003
DTBC-210
Trade Reforms and Manufacturing Industry in Chile
Roberto Álvarez y Rodrigo Fuentes
Mayo 2003
DTBC-209
Corporate Governance in Chile
Manuel R. Agosin y Ernesto Pastén H.
Mayo 2003
DTBC-208
Indicadores Líderes del IMACEC
Luis Firinguetti y Hernán Rubio
Abril 2003
DTBC-207
El Embrague Financiero: Un Mecanismo Alternativo de
Amplificación Bancaria
Elías Albagli
Marzo 2003
DTBC-206
Efectos de las Intervenciones en el Mercado Cambiario:
el Caso de Chile
Matías Tapia y Andrea Tokman
Marzo 2003
DTBC-205
Policy Evaluation and Empirical Growth Research
Steven N. Durlauf
Marzo 2003
DTBC-204
Growth and Adjustment in Chile: A Look at the 1990s
Vittorio Corbo y José A. Tessada
Marzo 2003
DTBC-203
Microeconomic Effects of Capital Controls: The Chilean
Experience during the 1990s
Francisco A. Gallego y F. Leonardo Hernández
Febrero 2003
DTBC-202
Building Confidence Intervals for the Band-Pass and Hodrick-
Prescott Filters: An Application using Bootstrapping
Francisco A. Gallego y Christian A. Johnson
Febrero 2003
