Interest rate pass-through in the Eurozone and the USA: implications for monetary policy

in the context of the recent financial crisis

Stelios Karagiannis*
Centre of Planning and Economic Research, Greece

Yannis Panagopoulos
Centre of Planning and Economic Research, Greece

Prodromos Vlamis
Hellenic Observatory, London School of Economics, UK



ABSTRACT
We examine the interest rate transmission mechanism for the Eurozone and the USA and discuss
this issue in the light of the recent financial market tensions. For an efficient monetary policy,
any change in the central bank policy rate is meant to be transmitted to retail interest rates,
ultimately influencing consumer and business lending rates and therefore aggregate domestic
demand and output. The disaggregated GETS methodology is employed, which allows us to
reveal the relative importance of the central bank and money market rates as policy vehicle
variables in the two banking systems. Our empirical results for the two banking systems are
rather mixed as far as it concerns the pass-through transmission and completeness. We also refer
to the lessons learned prior to and after the collapse of the monetary and financial system on
both sides of the Atlantic. We believe that this study has interesting policy insights and provides
certain policy suggestions, which might be useful for the regulatory authorities in their attempt
to monitor and reinforce monetary policy effectiveness.

J.E.L. Classification: E52, E43, C22.
Keywords: Interest rate pass-through, disaggregated general-to-specific model.

Acknowledgements: This paper has greatly benefited from the constructive suggestions of
Andreas Andrikopoulos. The authors would also like to thank Ioanna Reziti for useful comments
regarding the GETS methodology.

* Corresponding Author: Centre for Planning & Economic Research (KEPE), 11 Amerikis Str,
10672, Athens, Greece. Email: [email protected]
1

1. Introduction
Standard economic theory includes monetary policy as one of the two main tools that
governments can use to influence output, investment, prices and employment in any economy.
Monetary policy can affect key macroeconomic variables through the functioning of three
distinct channels, namely the interest rate channel, the bank lending channel and the broad credit
channel
1
. Angeloni et al. (2002) find that the interest rate channel is the most important for
monetary policy transmission in the Euro area. The adjustment of retail bank interest rates
(deposit and lending rates) in response to changes in wholesale rates (central bank and interbank
money market rates)
2
is a fundamental element of the interest rate transmission mechanism. For
an efficient monetary policy, any change in the central bank policy rate is meant to be
transmitted to retail interest rates, ultimately influencing consumer and business lending rates
and therefore aggregate domestic demand and output. In effect, if the interest rate transmission
is not efficient, then the required policy reaction by the monetary authorities will have to be
stronger in order to achieve the same end result. The efficiency of the interest rate transmission
channel has become particularly important in the context of the recent financial crisis which was
erupted in August 2007 and intensified in the second half of 2008. Moreover, as money and
other credit markets has been getting further intertwined over the past two decades, disruptions
to the money and funding markets can have adverse macroeconomic consequences (ihk et al.,
2009). For all these reasons, the regular monitoring and assessment of the pass-through (PT
hereafter) is critical for policymaking.

The main aim of this paper is to provide empirical evidence on the pass-through process in the
Euro area and USA (the locomotives of the world economy) and to discuss this issue in the light
of the recent financial market tensions. We address the question of how the monetary

1
See Mishkin (1995) and Bean, Larsen, and Nikolov (2002) for a detailed discussion of the transmission channels.
2
The money market rate can be considered as a policy controlled variable since central bank can influence it
2

transmission process works in these two economies and whether their responses to interest rate
dynamics are symmetric or asymmetric. We also show how rapidly and extensively changes in
policy rates and market interest rates are passed on to retail bank interest rates. To the best of our
knowledge it is the first attempt to make a comparative study of the efficiency of the monetary
policy between the Eurozone and the USA regarding the symmetric or asymmetric responses of
retail rates to wholesale rates changes. In an attempt to unveil the existence and importance of a
pass-through behaviour, the disaggregated general-to-specific model (hereafter GETS) is
employed. This model, which has not been used before in the relevant literature, allows us to
estimate and make inferences not only about the short and long run interest rate elasticities but,
as well as, for the upward/downward stickiness of the interest rate transmission mechanism. Any
upward or downward change in the central bank policy rate is meant to be transmitted to retail
interest rates for the monetary authorities to be able to affect consumer and business lending
rates and therefore aggregate domestic demand and output. Moreover, it is important for both
central bankers and Treasury Secretaries to have a precise understanding of the upward and
downward degree of the interest rate pass-through in order to be able to design the appropriate
monetary policy along the different phases of the business cycle.

The empirical results for the two banking systems are rather mixed as far as it concerns the
wholesale rates PT transmission and completeness. In the Eurozone it is mainly the money
market rate that is transmitted more effectively to the retail rates (and in particular the lending
rates). Also, the interest rate PT is complete in the long run, only when the wholesale rate is the
money market rate and the retail rate is the deposit rate. For the US case, it is the central bank
rate rather than the money market rate that works more effectively as a policy vehicle variable.
Also the interest rate PT is nearly complete in the long run only between the central bank rate

through short-term interest rate policy.
3

and the deposit rate. Regarding the effect of an upward or downward change (symmetry
hypothesis) in the policy-controlled variables to the retail rates, our results for the Eurozone
indicate that banks tend to pass to depositors only the decreases of the original money market
rate change and to borrowers more of its increases rather than its decreases. In contrast, in the
US case banks tend to pass to depositors only the decreases of the original central bank rate
change, while they pass to borrowers more of its decreases rather than its increases.

The paper is structured as follows. Section 2 offers a discussion of the importance of the interest
rate policy transmission mechanism, particularly in the light of the recent financial crisis. In
Section 3, a brief review of the literature on interest rate PT methodology is presented. In section
4, the econometric approach regarding the PT interest rate model is analysed and the empirical
results are given in Section 5. Section 6 provides our final remarks and draws policy conclusions.

2. Financial crisis and the interest rate pass-through
The transmission of policy rate changes to retail bank loan and deposit rates is of utmost
importance since it acts as a conduit for monetary policy. The interest rate transmission channel
has become particularly important in the context of the recent financial crisis. This financial
market turmoil which was generated by a drop in the value of US sub-prime loans, it was soon
transferred into the US and EU banking sector (massive write-downs, losses and defaults of
financial intermediaries) and jeopardised confidence among banks. One of the consequences of
this development was the disturbance in the functioning of the money market, which was
mirrored in a widening of the spreads between the central bank policy rates and the money
market rates and between the former and the retail rates. This disturbance potentially impaired
the monetary policy PT to retail rates, since many bank loan and deposit rates are linked to
money market ones. As a result, the cost of credit to both firms and households declined much
4

less than the policy rates during the crisis; credit spreads initially widened, although very
recently have narrowed to some extent (see figure below).

Monthly Interest Rates in EU (2003:2 - 2009:9)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
2
0
0
3
M
2
2
0
0
3
M
5
2
0
0
3
M
8
2
0
0
3
M
1
1
2
0
0
4
M
2
2
0
0
4
M
5
2
0
0
4
M
8
2
0
0
4
M
1
1
2
0
0
5
M
2
2
0
0
5
M
5
2
0
0
5
M
8
2
0
0
5
M
1
1
2
0
0
6
M
2
2
0
0
6
M
5
2
0
0
6
M
8
2
0
0
6
M
1
1
2
0
0
7
M
2
2
0
0
7
M
5
2
0
0
7
M
8
2
0
0
7
M
1
1
2
0
0
8
M
2
2
0
0
8
M
5
2
0
0
8
M
8
2
0
0
8
M
1
1
2
0
0
9
M
2
2
0
0
9
M
5
2
0
0
9
M
8
Loans (non-financial > 5Y) Deposits (housholds < 1Y) Marginal Lending Facility Rate Interbank Rate (3M)
Monthly Interest Rates in USA (2003:2 - 2009:9)
0.00
1.50
3.00
4.50
6.00
7.50
9.00
2
0
0
3
M
2
2
0
0
3
M
5
2
0
0
3
M
8
2
0
0
3
M
1
1
2
0
0
4
M
2
2
0
0
4
M
5
2
0
0
4
M
8
2
0
0
4
M
1
1
2
0
0
5
M
2
2
0
0
5
M
5
2
0
0
5
M
8
2
0
0
5
M
1
1
2
0
0
6
M
2
2
0
0
6
M
5
2
0
0
6
M
8
2
0
0
6
M
1
1
2
0
0
7
M
2
2
0
0
7
M
5
2
0
0
7
M
8
2
0
0
7
M
1
1
2
0
0
8
M
2
2
0
0
8
M
5
2
0
0
8
M
8
2
0
0
8
M
1
1
2
0
0
9
M
2
2
0
0
9
M
5
2
0
0
9
M
8
Federal Funds Rate Discount Rate (End Of Period) CDs (Secondary Market) 3M Bank Prime Loan Rate


The severity of the financial crisis had repercussions to the US and EU real estate markets and
eventually contributed to a marked slowdown in economic growth. The downturn in property
markets was also due to the oversupply of new developments, primarily caused by the preceding
speculative boom in land markets (Malpezzi and Wachter, 2005). Simultaneously, demand for
new space levelled off because of the contraction in economic activity. These two facts led to
decelerated property prices and refinancing of companies and households became more difficult.
This was the case because banks altered their risk assessments, became more cautious, switched
5

to tighter credit standards and more conservative lending rules (Panagopoulos & Vlamis, 2009).
As a result, banks become less willing to provide loans (credit rationing) since capital values of
real estate, used as collateral, started to decline after the burst of the real estate bubble (primarily
in the USA, Ireland, Spain and the UK). Moreover, bank funding markets have struggled to
provide liquidity across the banking sector. This was another reason that may have led many
banks to increase their credit risk premia and tighten their standards for providing loans (ECB,
2009). Consequently, the decreases in central bank rates since October 2008 hindered the
transmission to bank lending rates.

ECB and the US Federal Reserve responded to pressures in credit markets by injecting liquidity
in their banking systems and decreasing policy rates. They also responded to the interbank
money market tensions by relaxing collateral requirements for funding from the lender of last
resort. Moreover, European and US governments responded to the crisis by using public money
to support and recapitalise the banking sector. In October 2008, the US Congress approved a
bailout programme of $700bn to purchase troubled assets from US financial institutions. The
agreed Emergency Economic Stabilisation Act set out troubled asset relief programme, which
was directed to the recapitalisation of the financial system. European Union member-states have
taken certain actions to alleviate pressures on SMEs by enhancing access to credit and to support
banks by purchasing preferred stocks, which were issued for this purpose. The ECB and the
Federal Reserves liquidity management and EU and US governments policy measures taken
throughout the crisis, seems that facilitated the functioning of the money markets as indicated by
declining liquidity premia in term Euro and US money market spreads (IMF, 2009).

3. Review of the literature on interest rate pass-through
Very few studies have analysed the impact of the present crisis on monetary transmission in the
6

Euro area and the USA. One such study by IMF (2008) reports that the PT of policy rates to
market rates has been disrupted, although that affected differently the Euro area and the USA.
They argue that the interest rate PT to the short term rates has been less affected in the Euro area,
while interest rate PT to the long-term ones, was much more disrupted in both the United States
and the Euro area. Another study by ihk et al. (2009), which analyses the effectiveness of
ECB monetary policy in the context of the financial crisis, shows evidence that policy rate
changes have still been transmitted to market rates but interest rate transmission has been less
efficient. In effect, the necessary policy reaction by the monetary authorities to stabilize the
economy would have to be stronger to achieve the same end result. Finally, a study by ECB
(2009) points out that during the financial crisis, the bank retail interest rate PT in the Euro area
has been responding relatively satisfactorily to the volatility of the Euro Interbank Offered Rate
(EURIBOR) and other longer-term market rates, although less well to the volatility of the Euro
Overnight Index Average (EONIA).

Although, very few studies have analysed the impact of the present crisis on monetary
transmission, there is a wealth of empirical literature that deals with the transmission from the
policy-controlled interest rate to the retail rates, before the eruption of the crisis ((gert et al
(2007), Fourcans and Vranceanu (2007), Chionis and Leon (2006), Hofmann (2006), Sander and
Kleimeier (2004), Atesoglou (2003-4), Angeloni and Ehrman (2003), Burgstaller (2003), De
Bondt (2002), Petturson (2001), and Toolsema, Sturm, and De Haan (2001), Bredin, Fitzpatrick,
and OReilly (2001) and Mojon (2000)). It is evident from these studies that rigidities in the
transmission process, significant variations across countries and non-completeness at least in
some cases, exist. Regarding the degree of stickiness in different economies, recent empirics do
not reach any conclusive evidence ((Cottarelli and Kourelis (1994), Borio and Fritz (1995),
Donnay and Degryse (2001), Kleimeier and Sander (2000) and Toolsema, Sturm, and De Haan
7

(2001)).

The existence of an interest rate sluggishness can be explained by the agency costs theory (Fried
and Howitt, 1980; Stiglitz and Weiss, 1981; Berger and Udell, 1992) and the customer switching
costs theory (Klemperer, 1987, Calem et al., 2006). Furthermore, there may be fixed adjustment
costs, the so called menu costs, whenever banks decide to change retail rates (Hannan and
Berger, 1991; Cottarelli and Kourelis, 1994; Hofmann and P. Mizen, 2004). Such menu costs
might depend on the temporal or permanent nature of policy rate changes. Additionally, interest
rate sluggishness might depend on the degree of competition among banks and specifically on
the concentration level of the retail market (Sander and Kleimeier, 2004; Van Leuvensteijn et al.,
2008). This is related to the so called banks collusive hypothesis (Berger and Hannan, 1989;
Hannan and Berger, 1991; Neumark and Sharpe, 1992). Finally, another explanation is provided
by Hannan and Berger (1991), namely the customer reaction hypothesis, that is borrowers react
strongly to policy rates changes if they have the bargaining position to do so.

The empirical literature on PT transmission models
3
focuses on the Error Correction Model
(hereafter ECM) and the Threshold Autoregressive model
4
. Balke and Fomby (1997), Enders
and Granger (1998) and Von Craumon-Taubadel and Meyer (2000), show that in the standard
ECM models, tests for unit roots and cointegration have low power in the presence of
asymmetric adjustment. This happens because such tests implicitly assume symmetric and linear
adjustment processes. Bachmeier and Griffin (2003) and Rao and Rao (2005) presented an
alternative dynamic approach originating from the LSE-Hendry general to specific methodology.
The two main advantages of this approach are: i) it can jointly and simultaneously test for short-

3
For a complete survey on econometric models of asymmetric price transmission see Frey and Manera (2007).
4
Enders and Granger (1998) and Enders and Siklos (2001) introduced the TAR models. A number of scholars
applied the aforementioned methodology in interest rate PT transmission literature (see Sander and Kleimeier
(2002), Sander and Kleimeier (2006), Fuertes, Heffernan and Kalotychou (2006) and Payne (2007)). However,
8

run and long-run rigidities within the same PT dynamic model (Rao and Singh, 2006) and ii) it
can be used to test the existence of symmetric or asymmetric transmission behaviour
(disaggregated GETS) between the examined variables.

4. Econometric methodology:the LSE-Hendry general-to-specific model
The interest rate PT literature is mainly related to the way central bank rate (CB hereafter)
and/or interbank money market rate (MM hereafter) are transmitted to the retail rates (deposit
and lending). Such PT interest rate equations usually take the following simple algebraic form:

t i t W
n
i
j t R
k
j
t R
e i i k c i + - + - + =

=

=
,
1
,
1
,
| (1)
where, , stands for the different kinds of loan and deposit rates and
t R
i
,
t W
i
,
, stands for the CB or MM rates and
(2)
t t i t W
n
i
j t R
k
j
t R
u e i i c i + - A - + A - + = A

=

=
1 ,
1
,
1
,
u
There are two main theoretical issues which are worth examining. First, the long-run and short-
run interest rate rigidities (the s and the s ' coefficients in eq. 1 and eq. 2, respectively) from
the wholesale to the retail market rates and second, the speed of retail rates adjustment initiated
from the wholesale interest rate changes (the u coefficient of the error correction term in eq. 2).
However in the simple ECM (eq. 2) the retail rates and the speed of adjustment coefficient (u )
cannot be analysed separately when the wholesale rates are increasing or decreasing. The
disaggregated GETS model tackles the above issues.

We know from the literature that a simple aggregate dynamic GrangerEngle Vector Error

TAR models have computational difficulties and often impose ex-ante atheoretical restrictions.
9

Correction (VECM) model has the following form:

+ + +
i t R
n
i
i t R
i

=

A
,
1
1
,
|
i t W
n
i
i t W
i

=

A
,
2
0
,
|
1 1 t
Z t

+
t
e

(3) =
t R
i
,

where and are the wholesale and retail interest rates, respectively. The term is the
error correction term. The disaggregated VECM model can be presented in the following form:
t W
i
, t R
i
, 1 t
Z

+ + + + +

A
1
0
,
n
i
i t R Rt
i |

A
i t W
n
i
Wt
i
,
2
1
|
1 1

t
Z t A

=
+
3
0
n
i
Rt
| i
+
1 ,t R
=
Rt
i
+

=
+
4
1
n
i
Wt
| i t W i
+
,
1 2
+
t
Z t

+
t
c

(4)
As Rao and Rao (2005) pointed out, the (+)/() superscript on the coefficients indicate a
positive/negative change in the variables included in the model. On the one hand, for any
positive change ( >0) in the independent variable, a corresponding response of all positive
coefficients ( , ) is expected. On the other hand, the corresponding negative coefficients
( , ) will respond in any negative change of the dependent variable ( <0)
t W
i
,
+
t
+
|

|

t
t W
i
,
5
. Moving a
step forward, the disaggregated GETS model could be presented in the following form:
= +

+ ( -
t R
i
,
=

1
1
,
j
i
t R
|

i t R
i
,
=

2
0
,
j
i
t W
|

i t W
i
,

u
t R
i
,
0
-
1
t W
i
,

2 T
)
t-1
+
+ +

+ ( -

=
+
3
0
,
j
i
t W
|
+
i t W
i
,
=
+
4
1
,
j
i
t R
|
+
i t R
i
,
+
u
t R
i
,
0
-
1
t W
i
,

2 T
)
t-1
+
t
(5)
where, and are the speed of adjustment coefficients in the positive and negative case,
respectively and

u
+
u
T the time trend. Alternatively, eq. 5 can be rearranged in the following way
(Rao and Rao, 2005):

5
In econometric terms the corresponding activation will be triggered in eq. 4 with the help of dummy variables
More specifically, all positive coefficients will take the value of 1 when a positive change in the dependent variable
occurs and will be zero otherwise.
10

=
t R
i
,

+
1
+ + + (
t
-

1
1
,
j
i
t R
|

i t R
i
,
=

2
0
,
j
i
t W
|

i t W
i
,

u
R
i
,
1
t W
i
,
)
t-1
+
+ +

+ ( -

=
+
3
0
,
j
i
t W
|
+
i t W
i
,
=
+
4
1
,
j
i
t R
|
+
i t R
i
,
+
u
t R
i
,
1
t W
i
,
)
t-1
+
t
(5a)
The choice between the two disaggregated GETS models (5) and (5a) depends on the
performance and plausibility of the estimation results.

The main advantages of the disaggregated GETS model include: i) its capability of estimating
both negative and positive short-run elasticities
6
(e.g. the in eq. 5 and 5a) and ii)
the direct and simultaneous estimation of the long-run (
+
t W t W
and
, ,
| |
1
or alternatively
0
+
1
) and the
short-run price transmission rigidities in the same model. Thus, using a GETS model two
different impact multipliers (a negative and a positive one), two interim multipliers and two
different speed of adjustments, can simultaneously be estimated.

Data used for the two countries are collected from the International Financial Statistics produced
by the International Monetary Fund (see Appendix A for a discussion of data). Before we
proceed to the disaggregated GETS model implementation, it is necessary to trace the number of
co-integrated vectors ( r ) between the dependent and the independent variable by using the
Johansens methodology (Johansen, 1995). The number of the existing co-integrating vectors
from the Johansens process, is sensitive to the number of lagged variables ( ) of the initial
vector (see Karfakis, 2004). Due to this sensitivity five different lag selection criteria will be
applied. These include the modified Likelihood Ratio test statistic, the Final Prediction Error
test, the Akaike, the Schwarz and the Hannan-Quinn information criteria. In most of the
examined cases the aforementioned selection criteria do not all agree about the optimal lag
n

6
The ability of testing both negative and positive short-run pass through elasticities ( ) in the same
+
t W t W
and
, ,
| |
11

length. In each case, the majority rule is applied as a sub-optimal solution
7
.

5. Empirical results
We employ the Johansen (1995) methodology on testing the existence of a long-run relationship
among retail and wholesale rates in the Eurozone and USA. According to the eigenvalue and
trace tests in all the bivariate cases, there is a unique co-integrated vector of order 1 (r=1), which
supports the hypothesis that interest rates in the Eurozone tend to co-integrate pairwise (see
Table 1 in Appendix B). For the US case the same tests were implemented and it is found that a
unique co-integrated vector exists only between the money market rate and the deposit rate (see
Table 2 in Appendix B).

5.1. Speed of adjustment estimates and the degree of pass-through completeness
With the wholesale rates and retail rates integrated of the same order, we estimate the
disaggregated GETS model
8
for the two types of interest rates in the Eurozone. According to
Table 3, column 2, the coefficients of the two error correction terms and , are statistically
significant (although the speed of adjustment in both cases are quite low) when the wholesale
rate is the CB rate ( ) and the retail rate is the loan rate ( ). This means that CB rate increases
and decreases are both transmitted to the loan rate. Also, the and coefficients are
statistically significant when the wholesale rate is the MM rate ( i ) and the retail rate is the
loan rate ( ) (Table 3, column 4). Again, this implies that MM rate increases and decreases are
both transmitted on the loan rate. The magnitude of is roughly five times higher (in absolute
terms) than that of (-0.59 and -0.12 respectively). Lastly, when the wholesale rate is the MM
+
u
+
u

u
CB
i

u
L
i
MM
L
i
+
u

model is actually enriching the Cottarelli and Kourelis (1994) pass through interest rates multipliers with positive
and negative values.
7
Results about the optimal lag structure using the five different selection criteria are available from the authors
upon request.
8
This model is tested by the Non-Linear Least Squares (N.L.L.S.) methodology.
12

rate ( ) and the retail rate is the deposit rate ( ) (Table 3, column 3) only the coefficient
is statistically significant, which means that only the downward PT transmission is operative in
this case. The speed of adjustment is relatively high (-0.45). It is evident from the above
results that the MM rate is transmitted to both deposit and loan rates while the CB rate is
transmitted to the loan rate alone. Our findings show that MM rate compared to the CB rate is
more effective as a policy vehicle variable in the Eurozone.
MM
i
D
i

u

u

We continue our analysis by examining the degree of PT completeness between the two types of
interest rates in the Eurozone. Coefficient
1
| (in eq. 5 and 5a) measures the degree of pass-
through. Complete PT exists when
1
| =1, which implies that all of the change in the policy-
vehicle rate (either CB or MM) is transmitted to the retail rates. In the Eurozone, the interest rate
PT is complete in the long run (0.97) and statistically significant when the wholesale rate is the
MM rate ( ) and the retail rate is the deposit rate ( ) (Table 3, column 3 and Table 5). In
contrast,
MM
i
1
D
i
| is 0.48, when the CB is the policy-controlled interest rate and 0,42 when the MM is
the policy-vehicle variable, which indicates an incomplete PT to the loan rates (Table 3,
columns 2 & 4 and Table 5). In other words, not all of the change in the policy rate is
transmitted to the loan rates. Lastly, most of the impact multipliers are statistically insignificant.

As far as USA is concerned, we also estimate the disaggregated GETS model for the two types
of interest rates. According to Table 4, column 1, the coefficients of the two error correction
terms and , are statistically significant (although the speed of adjustment in both cases are
quite low) when the wholesale rate is the CB rate ( ) and the retail rate is the deposit rate ( ).
The magnitude of is roughly ten times higher (in absolute terms) than that of (-0.10 and -
0.01 respectively). Also, the and coefficients are statistically significant when the
+
u

u
CB
i
D
i

u
+
u
+
u

u
13

wholesale rate is the CB rate ( ) and the retail rate is the loan rate ( ) (Table 4, column 2).
Overall, this implies that CB rate increases and decreases are both transmitted to the deposit and
loan rates. Lastly, it is evident from Table 4 (columns 3 and 4) that the MM rate is not
transmitted to the retail rates, which probably shows that the MM rate does not work effectively
as a policy vehicle variable in the USA.
CB
i
L
i

We continue our analysis by examining the degree of PT completeness between the two types of
interest rates in the USA. The interest rate PT is nearly complete in the long run (0.67) and
statistically significant, when the wholesale rate is the CB rate and the retail rate is the deposit
rate ( ) (Table 4, column 1 and Table 6). In contrast,
D
i
1
| is 0.49, when the CB is the policy-
controlled interest rate and 0,42 when the MM is the policy-vehicle variable (Table 4, columns 2
& 4 and Table 6), which indicates that not all of the change in the policy rate is transmitted to
the loan rates. Lastly, most of the impact multipliers are statistically insignificant.

5.2. Testing the symmetry hypothesis
Lastly, we ask what is the effect of an upward or downward change in the policy-controlled
variables to the retail rates in the two banking systems. More specifically, we test the symmetry
hypothesis that = . The existence of symmetric speed of adjustment is tested by using the
Wald - test.
+
u

u
2
_

We present the results for the Eurozone in Table 7. On the one hand, when the wholesale rate is
CB rate ( ) and the retail rate is the loan rate ( ), a symmetry exists, that is = . This
means that banks tend to pass to borrowers equally the decreases and increases of the original
CB rate change. On the other hand, when the wholesale rate is the MM rate ( ) and the retail
CB
i
L
i
+
u

u
MM
i
14

rate is the deposit rate ( ) it seems that there is only a negative asymmetry which means that
banks tend to pass to depositors only decreases of the original MM rate change. The magnitude
of the decrease of the deposit rate in this case is given by the long run elasticity which, as we
saw in Table 5, is 0.97. Since only is statistically significant (negative asymmetry), we can
infer that for a 1% decrease in the MM rate, banks pass to depositors 0.97 of that decrease.
Lastly, when the wholesale rate is the MM rate ( ) and the retail rate is the loan rate ( ) it
seems that the positive asymmetry is stronger than the negative one. This means that banks tend
to pass to borrowers more of the increases of the original MM rate change rather than its
decreases.
D
i

u
CB
MM
i
L
i

Our empirical tests regarding the symmetry hypothesis in the USA (Table 8) shows that, when
the wholesale rate is the CB rate ( i ) and the retail rate is the deposit rate ( ), there is only a
negative asymmetry, which means that banks tend to pass to depositors only decreases of the
original CB rate change. The magnitude of the decrease of the deposit rate in this case is given
by the long run elasticity which, as we saw in Table 5, is 0.67. Since only a negative asymmetry
( is statistically significant) is observed, we can infer that for a 1% decrease in the CB rate,
banks pass to depositors 0.67 of that decrease. Additionally, when the wholesale rate is the CB
rate ( ) and the retail rate is the loan rate ( ), it seems that the negative asymmetry is stronger
than the positive one. This means that banks tend to pass to borrowers more of the decreases of
the original CB rate change rather than its increases.
D
i

u
CB
i
L
i

Such behaviour is theoretically consistent with the customer reaction hypothesis regarding the
loan market in the USA (Hannan and Berger, 1991). It is also in line with the banks collusive
hypothesis regarding the deposit and loan markets in the Eurozone and the deposit market in the
USA (Berger and Hannan, 1989; Hannan and Berger, 1991; Neumark and Sharpe, 1992). The
15

speed of retail rates adjustment can be interpreted as the commercial bank managers power to
transmit to their clients any wholesale rate changes. Such speed is possibly affected by the
degree of the retail market competitiveness in the banking sector. For example, in a competitive
banking environment, the deposit rates are expected to be raised by the bank managers, in
response to wholesale rates increases. The asymmetry results for the Eurozone and USA might
be explained by this framework. As far as the European banking sector is concerned, where the
banks collusive hypothesis seems to prevail, bank managers are eager to transmit money market
rate decreases/increases to depositors/borrowers. Our results for the US banking sector, where
the customer reaction hypothesis dominates, show that bank managers are happy to transmit
central bank rate decreases to borrowers and are equally happy to transmit only part of the
central bank rate decreases to depositors.

6. Conclusions and policy suggestions in the context of the financial crisis
This study focuses on how the interest rate PT works in the Eurozone and the USA. It is the first
attempt to make a comparative study for the Eurozone and the US in an effort to unveil the
existence and importance of an interest rate PT behaviour. The disaggregated GETS
methodology is employed, which allows us to reveal the relative importance of the central bank
and money market rates as policy vehicle variables in the two banking systems. Our empirical
results for the two banking systems are rather mixed as far as it concerns the wholesale rates PT
transmission and completeness. We believe though that these results can be useful for the
European and US regulatory authorities in their attempt to monitor the competitiveness of their
banking systems and reinforce monetary policy effectiveness.

The interest rate transmission channel has become particularly important in the context of the
recent financial crisis that led to a disturbance in the functioning of the money markets both in
16

the US and the Eurozone. As a result, the efficiency of the monetary policy transmission in the
two regions has been disrupted and this was mirrored in a widening of the spreads between the
central bank policy rates and the money market rates as well as between the former and the retail
rates. Active policies from the central bank authorities and regulators end, were rendered
essential for the monetary transmission mechanism to be restored. There are a number of lessons
to be learned prior to and after the collapse of the monetary and financial system on both sides
of the Atlantic. Necessary policy measures to be taken in similar financial crisis in the future,
should involve: active liquidity management from the central bank authorities, financial
institution and market regulation, central bank cooperation and information dissemination and
fiscal policy (Chailloux et al, 2008). First of all, central banks should have a properly designed
strategy, accurately executed, for channelling liquidity in the financial system and thus restoring
the smooth monetary transmission. Such strategy could possibly include indirect measures, for
example money market support by exchanging treasury bills for other types of securities used as
collateral. Having said that, the lender of last resort role of the central bank should not be
exaggerated during periods of liquidity stress. Emergency liquidity facilities should be used with
caution by the market participants to ensure that banks do not become overly dependent on
central bank support and avoid the associated moral hazard problem.

Secondly, various quantitative regulations should be put forward by the supervisory authorities
and central banks, to ensure prudent lending practices. More specifically, rules can be imposed
regarding the size of the loan-to-value ratios and the use of periodic and frequent property
valuation used as collateral (Panagopoulos and Vlamis, 2009). The un-controlled structured
finance industry (credit default swaps, collateralised debt obligations, collateralised loan
obligations, asset-backed securities, collateralised debt obligations) as well as the hedge fund
and the private equity industries, should be regulated and brought under strict supervision of
17

securities and exchange commissions. It has been suggested by OECD (2008) that re-regulation
of international financial markets is necessary to include coverage of both financial products and
institutions and to identify and prevent excessive risk-taking behaviour from the market
participants end. EU authorities responded to the new challenges by creating i) the European
Systemic Risk Council, ii) supervisory colleges, and iii) taking the initiative to create a single
European rule book applicable to all financial institutions. As far as it concerns the response of
the G20, they decided to rename and upgrade the former Financial Stability Forum into the
newly established Financial Stability Board, which will provide early warning of
macroeconomic and financial risks as well as the necessary actions needed to address them.

Third, during a period of financial distress it becomes even more important to establish
information sharing between central banks in order to enhance collaboration and frequent
communication between them (IMF, 2008). Also, central banks should provide more
information to economic agents in order to maintain the smooth functioning of the markets. Last
but not least, when markets face liquidity crises, fiscal authorities need to work closely with
central bank, without jeopardising its independence, to design the optimal fiscal-monetary policy
mix to achieve price and output stability.




18

References
Angeloni, I. and Ehrmann, M. (2003). Monetary transmission in the euro area: early evidence.
Economic Policy, 18: 469-501.
Angeloni, I. & Mojon, B., Kashyap, A. & Terlizzese, D. (2002). Monetary transmission in the
Euro Area: where do we stand? ECB Working Paper Series 114.
Atesoglou, H.S. (2003-4). Monetary transmission-federal funds rate and prime rate. Journal of
Post Keynesian Economics,26, 357-63.
Bachmeier, L.J. and Griffin, J.M. (2003). New evidence on asymmetric gasoline price responses.
The Review of Economics and Statistics, 85(3): 772-776.
Balke, N.S. and Fomby, T.B. (1997). Threshold cointegration. International Economic Review,
38: 627-645.
Bean, C, Larsen, J and Nikolov, K (2002). Financial frictions and the monetary transmission
mechanism: Theory, evidence and policy implications. ECB Working Paper No. 113.
Berger, A. and Hannan, T. (1989). The price-concentration relationship in banking. Review of
Economics and Statistics, 71 (2), 291-299.
Berger, A. and Udell, G. (1992). Some evidence on the empirical significance of credit rationing.
Journal of Political Economy, 100, October, 1047-1077.
Borio, C. E. V. and Fritz, W. (1995). The response of short-term bank lending rates to policy
rates: A cross-country perspective. In Financial Structure and the Monetary Policy
Transmission Mechanism, Basle: BIS.
Bredin, D., Fitzpatrick, T. and O Reilley, G. (2001). Retail interest rate pass-through: The Irish
experience. Technical paper 06/RT/01, Central bank of Ireland.
Burgstaller, J. (2003). Interest rate transmission to commercial credit rates in Austria. Working
Paper No. 0306, Austria: Johann Kepler University of Linz.
Calem, P.S., Gordy, M.B. and Mester, L.J. (2006). Switching costs and adverse selection in the
market for credit cards: New evidence. Journal of Banking & Finance, 30, 1653-1685.
Chailloux, A., Gray, S and McCaughrin1, R. (2008). Central bank collateral frameworks:
Principles and policies. IMF Working Paper No. 222.
ihk, M., Harjes, T. and Stavrev, E. (2009). Euro area monetary policy in uncharted waters.
IMF Working Paper No. 185.
Cottareli, C. and Kourelis, A. (1994). Financial structure, bank lending rates, and the
transmission mechanism of monetary policy. IMF Working Paper 95/38, Washington, DC.
Chionis, D. and Leon, C. (2006). Interest rate transmission in Greece: Did EMU cause a
structural break ? Journal of Policy Modeling, Vol. 28, 453-466.
De Bondt, G (2002). Retail bank interest rate pass-through: new evidence at the Euro area level.
ECB Working Paper, No. 136.
Donnay, M. and Degryse, H. (2001). Bank lending rate pass-through and differences in the
transmission of a single emu monetary policy. Discussion Paper DPS 01.17, Centre of
Economic Studies, Leuven, Katholieke Universiteit.
gert, B., Crespo-Cuaresma, J. and Reininger, Thomas, (2007). Interest rate pass-through in
central and Eastern Europe: Reborn from ashes merely to pass away? Journal of Policy
Modeling, Elsevier, vol. 29(2), 209-225.
Enders, W. and Granger, C.W.J. (1998). Unit-root tests and asymmetric adjustment with an
example using the term structure of interest rates. Journal of Business and Economic
Statistic, 16: 304-311.
Enders, W. and Siklos, P.L. (2001). Co-inegration and threshold adjustment. Journal of
Business and Economic Statistic, 19: 166-167.
European Central Bank (2009). Recent developments in the retail bank interest rates pass-
through in the Euro area. Monthly Bulletin (August), 93-105.
Fourcans, A. and Vranceanu, R. (2007). The ECB monetary policy: Choices and challenges.
19

Journal of Policy Modeling, Elsevier, vol. 29(2), 181-194.
Frey, G. and Manera, M. (2007). Econometric models of asymmetric price transmission.
Journal of Economic Surveys, Vol. 21 (2), 349-415.
Fried, J. and Howitt, P. (1980). Credit rationing and implicit contract theory. Journal of Money,
Credit and Money,12(3), 471-87.
Fuertes, A-M., Heffernan, S.A. and Kalotychou, E. (2006). Nonlinearity in the British interest
rate transmission mechanism, Working Paper, Cass Business School.
Hannan, T. and Berger, A. (1991). The rigidity of prices: Evidence from the banking industry.
American Economic Review, 81, 938-45.
Hofmann, B. (2006). EMU and the transmission of monetary policy: evidence from business
lending rates. Empirica, 33, 209-229.
Hofmann, B. and Mizen, P. (2004). Interest rate pass-through and monetary transmission:
Evidence from individual financial institutions retail rates. Economica, 71, 99-123.
International Monetary Fund, (2008). Global financial stability report. 73-106, Washington.
Johansen, S. (1995). Likelihood Based Inference in Cointegrated Vector Autoregressive Models,
Oxford: Oxford University Press.
Karfakis, C. (2004). Testing the quantity theory of money in Greece: Reply to zmen. Applied
Economic Letters, 11, 541-43.
Klemperer, P. (1987). Markets with consumer switching costs. Quarterly Journal of Economics,
375-94.
Malpezzi, S. and Wachter, S. (2005). The role of speculation in real estate cycles. Zell/Lurie
Center Working Papers 401, University of Pennsylvania.
Mishkin, Frederic S, (1995). Symposium on the monetary transmission mechanism. Journal of
Economic Perspectives, American Economic Association, vol. 9(4), 3-10, fall.
Mojon, B. (2000). Financial structure and the interest rate channel of the ECB monetary policy.
ECB Working Paper No. 40.
Neuwark, D. and Sharpe, S. (1992). Market structure and the nature of price rigidity: evidence
from consumer deposits. Quarterly Journal of Economics, 107, 657-81.
OECD (2008). Beyond the financial crisis, re-regulation and governance of the global economy:
What role for the OECD? Discussion document.
Panagopoulos, Y. and Vlamis, P. (2009). Bank lending, real estate bubbles & basel II. Journal of
Real Estate Literature, Vol. 17, No. 2, 295-310.
Payne, J. (2006-7). More on the monetary transmission mechanism: mortgage rates and the
federal funds rate. Journal of Post Keynesian Economics, Vol. 29, 247-57.
Rao, B.B. and Rao G., (2005). Are US gasoline price adjustments asymmetric? Working Paper,
University of South Pacific, Suva.
Rao, B.B. and Singh, R. (2006). Demand for money for Fiji with PcGets. Applied Economic
Letters, 13, 987-91.
Petturson, T.G. (2001), The transmission mechanism of monetary policy: Analyzing the
financial market pass-through, Working Paper No 14, Central bank of Ireland.
Sander, H. and Kleimeier, S. (2000). Regionalisation versus globalisation in European financial
market integration: Evidence from co-integration analyses. Journal of Banking and Finance,
24 (6), 1005-1043.
Sander, H. and Kleimeier, S. (2002). Asymmetric adjustment of commercial bank interest rate
pass-through. Kredit und Kapital, 35(2), 161-92.
Sander, H. and Kleimeier, S. (2004). Convergence in euro-zone retail banking? What interest
rate pass-through tells us about monetary policy transmission, competition and integration.
Journal of International Money and Finance, 23 (3), 461-492.
Sander, H. and Kleimeier, S. (2006). Interest rate pass-through in the common monetary area of
the SACU countries. South African Journal of Economics, 74(2), 215-29.
20

Stiglitz, J.E. and Weiss, A. (1981). Credit rationing in markets with imperfect information.
American Economic Review, 71(3), 393-410.
Toolsema, L.A., Sturm, J.-E. and De Haan, J. (2001). Convergence of monetary transmission in
E.U.: New evidence. CESifo Working Paper 465, Munich.
v. Cramon-Taubadel, S. and Meyer, J. (2000). Asymmetric price transmission: Fact or artifact?
University Gttingen, Institut for Agricultural Economy, Working Paper.
Van Leuvensteijn, M., Kok Srensen, C., Bikker, J.A. and Van Rixtel, A. (2008). Impact of
bank competition on the interest rate pass-through in the euro area. ECB Working Paper No.
885.

Appendix A. Presentation of data
We use monthly data for the USA and the Eurozone but the examined time period is not
identical for the two banking systems. In the case of USA the time period ranges between
1994:12007:9 while in the Eurozone the time period is rather limited (1998:1-2003:9). In the
case of USA the discount rate and the federal fund rate are used as the central bank rate and
the money market rate, respectively. Concerning the retail rates, we use the 3-month certificate
of deposits and the bank prime loan rate as proxies for the deposit and lending rates. In the
Eurozone the margin lending facility rate is used as the central bank rate and the interbank
rate (3-month maturity) is used as the money market rate. As far as is it concerns the retail rates,
we use the deposit rate and the lending rate, correspondingly.

21

Appendix B. Empirical tables

Table 1: The Johansen Pairwise Co-intregration Tests in Eurozone
Causality
test
No. of Lags Rank
Max.
Eigenvalue
Trace
No. of Co-
integrating
Vectors (of r
order)
r=0 29.22

30.65

CB
i vs.
D
i
(2)
r1 1.43 1.43
r=1
r=0 23.49

26.50

CB
i vs.
L
i
(2)
r1 3.00 3.00
r=1
r=0 19.33 21.78
mm
i vs.
D
i
(1)

r1 2.44 2.44
r=1
r=0 14.67 17.67
mm
i vs.
L
i
(1)


r1 2.71 2.71
r=1
The critical value for accepting that r=1 at the 5% level for both the Maximum Eigenvalue test and
the Trace test is 3.84.
f. The lag selection has been done according to the Schwarz Information Criterion (SC).



Table 2: The Johansen Pairwise Co-intregration Tests in USA
Causality
test
No. of Lags Rank
Max.
Eigenvalue
Trace
No. of Co-
integrating
Vectors (of r
order)
r=0 7.04

9.17

CB
i vs.
D
i
(2)
r1 2.13 2.13
r=0

r=0 6.04

9.32

CB
i vs.
L
i
(2)
r1 3.28 3.28
r=0

r=0 6.04 9.32
mm
i vs.
D
i
(2)
r1 3.28 3.28
r=1

r=0 9.42

12.07

mm
i vs.
L
i
(2)
r1 2.65 2.65
r=0
The critical value for accepting that r=1 at the 5% level for both the Maximum Eigenvalue test and
the Trace test is 3.84.



22

Table 3 : The Eurozone banking system
Independent
variable
Central bank ( ) rate
CB
i Money Market ( ) rate
mm
i
Dependent
variable
Deposit rate
( )
D
i
Loan rate
( )
L
i
Deposit rate
( )
D
i
Loan rate
( )
L
i
C.V (r)* r=1 r=1 r=1 r=1
(1) (2) (3) (4)
Regressor
Coefficients -
t-ratios
Coefficients -
t-ratios
Coefficients -
t-ratios
Coefficients -
t-ratios
+

A
1 ,t D
i
0.50
(3.63)
- 0.66
(2.98)
-

A
1 ,t D
i
-1.26
(-1.36)
- -0.26
(-0.95)
-
+

A
2 ,t D
i
0.08
(0.63)
- - -

A
2 ,t D
i
-0.62
(-0.61)
- - -
+

A
1 ,t L
i
- 0.47
(3.04 )
- 0.01
(0.07)

A
1 ,t L
i
- 1.50
(2.80)
- 0.43
(2.53)
+

A
2 ,t L
i
- 0.03
(0.22)
- -

A
2 ,t L
i
- -0.04
(-0.07)
- -
+

A
1 ,t CB
i
0.05
(0.67)
0.009
(0.22)
- -

A
1 ,t CB
i
1.70
(2.58)
0.27
(1.11)
- -
+

A
2 ,t CB
i
0.02
(0.27)
0.001
(0.04)
- -

A
2 ,t CB
i
1.47
(1.65)
-0.39
(-2.15)
- -
+

A
1 ,t MM
i
- - 0.06
(0.42)
-0.08
(-1.32)

A
1 ,t MM
i
- - 0.44
(1.70)
0.10
(2.05)
+

A
2 ,t MM
i
- - - -

A
2 ,t MM
i
- - - -
+
u
-0.01
(-0.848)
-0.29
(-3.01)
0.20
(1.60)
-0.59
(-4.52)

u
0.25
(1.13)
-0.30
(-3.09)
-0.45
(-2.37)
-0.12
(-2.08)
0
| (or
0
)
1.32
(0.66)
0.32
(3.12)
-0.20
(-2.74)
1.27
(57.07)
1
|
0.04
(0.04)
0.48
(15.75)
0.97
(20.07)
0.42
(26.01)
T (time) - - -0.0001
(-0.33)
0.0006
(3.44)
2
R
0.79 0.72 0.66 0.74
For the determination of the optimal lag structure the following information criteria are used: the
modified LR test statistic (LR), the Final Prediction Error test (FPE), the Akaike Information Criterion
(AIC), the Schwarz Information Criterion (SC) and the Hannan-Quinn information criterion (HQ).
*This is the number of co-integrating vectors of an r order.





23

Table 4 : The US banking system
Independent
variable
Central bank ( ) rate
CB
i Money Market ( ) rate
mm
i
Dependent
variable
Deposit rate
( )
D
i
Loan rate
( )
L
i
Deposit rate
( )
D
i
Loan rate
( )
L
i
C.V (r)* r=0 r=0 r=1 r=0
(1) (2) (3) (4)
Regressor
Coefficients -
t-ratios
Coefficients -
t-ratios
Coefficients -
t-ratios
Coefficients -
t-ratios
+

A
1 ,t D
i
0.50
(4.61)
- 0.36
(2.14)
-

A
1 ,t D
i
0.84
(4.11)
- 0.50
(2.33)
-
+

A
2 ,t D
i
0.12
(1.32)
- -0.06
(-0.41)
-

A
2 ,t D
i
-0.31
(-0.93)
- 0.05
(0.26)
-
+

A
1 ,t L
i
- 0.35
(3.66)
- 0.02
(0.15)

A
1 ,t L
i
- 0.67
(2.68)
- 0.27
(1.00)
+

A
2 ,t L
i
- 0.17
(1.99)
- 0.15
(0.84)

A
2 ,t L
i
- -0.41
(-1.08)
- 0.13
(0.49)
+

A
1 ,t CB
i
-0.0004
(-0.01)
-0.01
(-0.87)
- -

A
1 ,t CB
i
-0.33
(-1.83)
-0.16
(-1.97)
- -
+

A
2 ,t CB
i
-0.02
(-0.81)
-0.01
(-0.97)
- -

A
2 ,t CB
i
0.32
(1.43)
0.10
(0.67)
- -
+

A
1 ,t MM
i
- - 0.20
(1.45)
0.13
(1.51)

A
1 ,t MM
i
- - 0.01
(0.07)
0.17
(1.49)
+

A
2 ,t MM
i
- - 0.07
(0.65)
-0.003
(-0.04)

A
2 ,t MM
i
- - 0.03
(0.24)
-0.005
(-0.41)
+
u
-0.01
(-1.50)
-0.03
(-2.20)
0.000005
(0.00004)
-0.07
(-1.39)

u
-0.10
(-2.42)
-0.06
(-4.15)
-0.000006
(-0.00004)
-0.08
(-1.61)
0
| (or
0
)
0.02
(1.54)
0.05
(2.75)
-478.9 0.11
(1.61)
1
|
0.67
(3.81)
0.49
(6.01)
1465.6
(0.00004)
0.42
(9.26)
T (time) -0.00006
(-0.77)
-0.00009
(-1.12)
8.22
(0.00004)
-0.000004
(-1.18)
2
R
0.49 0.61 0.45 0.55
For the determination of the optimal lag structure the following information criteria are used: the
modified LR test statistic (LR), the Final Prediction Error test (FPE), the Akaike Information Criterion
(AIC), the Schwarz Information Criterion (SC) and the Hannan-Quinn information criterion (HQ).
*This is the number of co-integrating vectors of an r order.


24


25

1
Table 5: EU Rigidities Estimates
Dependent
variable
P-T
variable
L-R
Rigidities
( | )
+

A
1 ,t CB
i
Deposit
Rates
( )
t D
i
,
A

A
1 ,t CB
i
0.04
+

A
1 ,t CB
i Loan Rates
( )
t L
i
,
A

A
1 ,t CB
i
0.48
+

A
1 ,t MM
i
Deposit
Rates
( )
t D
i
,
A

A
1 ,t MM
i
0.97
+

A
1 ,t MM
i Loan Rates
( )
t L
i
,
A

A
1 ,t MM
i
0.42


1
Table 6: US Rigidities Estimates
Dependent
variable
P-T
variable
L-R
Rigidities
( | )
+

A
1 ,t CB
i
Deposit
Rates
(
t D
i
,
A )

A
1 ,t CB
i
0.67
+

A
1 ,t CB
i Loan Rates
(
t L
i
,
A )

A
1 ,t CB
i
0.49
+

A
1 ,t MM
i
Deposit
Rates
(

A
1 ,t MM
i

*
t D
i
,
A )
+

A
1 ,t MM
i Loan Rates
(
t L
i
,
A )

A
1 ,t MM
i
0.42
*The estimated number has no economic
meaning


Table 7: The Asymmetry results in Eurozone
Model
Hypothesis
H
0
: ( )*
+
=u u
Result
CB
i
D
i
+
u

u
CB
i
L
mm
i
D
i

u
mm
i
L
vs.
Both & are
statistically insignificant
-

vs. i
1.08
(accept H
0
)
symmetry

vs.
Only the negative
change ( ) is
statistically significant
negative (-)
asymmetry
vs. i

10.12
(Reject H
0
)

positive (+)
asymmetry
* We test the symmetry hypothesis by applying the Wald (x
2
)
test. The critical value of x
2
statistic with one degree of
freedom is 3.84 (5% confidence interval) and 5.02 (2.5%
confidence interval).
Table 8: The Asymmetry results in USA
Model
Hypothesis
H
0
: ( )*
+
=u u
Result
CB
i
D

u
CB
i
L
i
mm
i
D
+
u

u
mm
i
L
i
+
u

u
vs. i
Only the negative
change ( ) is
statistically significant
negative (-)
asymmetry

vs.

36.35
(Reject H
0
)

negative (-)
asymmetry

vs. i
Both & are
statistically insignificant
-
vs.
Both & are
statistically insignificant

-
* We test the symmetry hypothesis by applying the Wald (x
2
)
test. The critical value of x
2
statistic with one degree of
freedom is 3.84 (5% confidence interval) and 5.02 (2.5%
confidence interval).