Beyond the Third Pillar of Basel Two: Taking Bond
Market Signals Seriously
Adrian Pop
To cite this version:
Adrian Pop. Beyond the Third Pillar of Basel Two: Taking Bond Market Signals Seriously.
2009. <hal-00419241>
HAL Id: hal-00419241
https://hal.archives-ouvertes.fr/hal-00419241
Submitted on 23 Sep 2009
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Working Paper
Document de Travail
EA 4272
Beyond the Third Pillar of Basel
Two: Taking Bond Market Signals
Seriously
Adrian POP (*)
2009/05
(*) Uniiversity of Nantes (LEMNA)
Laboratoire d’Economie et de Management Nantes-Atlantique
Université de Nantes
Chemin de la Censive du Tertre – BP 52231
44322 Nantes cedex 3 – France
www.univ-nantes.fr/iemn-iae/recherche
Tél. +33 (0)2 40 14 17 19 – Fax +33 (0)2 40 14 17 49
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr= Wdnlqj Erqg Pdunhw Vljqdov Vhulrxvo|
Dguldq Srs
Devwudfw1 !
"
# $%%& ' ' ( ) * +
" , "
- . / 0/1 22/234 6', 7 - 7 * 7 . "8 , 9$7 9:
WXqlyhuvlw|
ri Qdqwhv +8"-;<,/ Lqvwlwxwh ri Edqnlqj ) Ilqdqfh/ Fkhplq gh od Fhqvlyh gx Whuwuh/ ES 85564/
77655 Qdqwhv Fhgh{ 6/ Iudqfh> Who1= .6605073474987> id{= .66050734749831 H0pdlo= dguldq1srsCxqly0qdqwhv1iu
Dfnqrzohgjhphqwv
Wklv sdshu lv edvhg rq p| SkG glvvhuwdwlrq dw wkh Xqlyhuvlw| ri Ruohdqv/ Iudqfh1 Vxjjhvwlrqv iurp wkh
phpehuv ri p| SkG frpplwwhh/ Rolylhu gh Edqgw/ Dqqh Odyljqh/ Mhdq0Sdxo Sroolq/ Mhdq0Fkduohv Urfkhw/ dqg
Dplqh Wdud}l/ zkhuh yhu| xvhixo lq uhylvlqj wkh sdshu1 L dp dovr lqghewhg wr Vwhidqr Fdld}}d/ Pdun Iodqqhu|/
Sdxo Kdpdodlqhq/ Fkulvwrskh Kxuolq/ Khqul Sdjëv/ Jhrujh Shqqdffkl/ Oduu| Zdoo/ dqg sduwlflsdqwv dw wkh !
= uhvhdufk vhplqdu/ > = . ./ 9 = ./ $ *
- "/ / < - = " </ dqg
8")?8"-;< uhvhdufk vhplqduv iru xvhixo frpphqwv1 Wkh xvxdo glvfodlphu dssolhv1
!" #$% &
' ! ' #$% ( !" ")
# * * +
, % -. /
, 0*1
*2 3
(. $ 4
.$4) 5 67
#$% 89
, #: 2
0 $ .$4
-. ( 20#+.;$ 777) # 8 9 " < 2 =
% 2 .$ 4
/ ( 20#+. ) # # $ %
(#$%) - .
# =
20#+. (
>*) 20#+.;$ (777 = 6!*>6) ?@ A (777 >B ) 2
(77) < (776) "
& (%) (%%) 3 (2 ) +
(77C) D , .$ 4
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
"
/ (") , .
@
# = ( E ) 4 F # =
89 / G
< = & F
G
( =
/
)
# @
( H
) 8 #
9 ? , 6*77 I F
8 #
9 ? 2 (77") 5 ,
( ) = # ? -.
F -. ? 8J K $% 9 (2 77" ) % -.
( 20#+. L
20#+.;$ 777) ? ? = = , .
@
.
" .
C .
6 # .
> C
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
soft
configurations
Confirming the
available information
to supervisor
strong
configurations
Better resources
allocation
Improving the
off-site
models
Setting risk-adjusted
deposit insurance
premia
Triggering Prompt
Corrective Actions
# & M / : /
C I : /
( I ) I = @ =
@
/ / ( # ) %
/ & =
/ : F
" = =
N = : (
$D 77) (
)
/
2 " = &" ( #
77)
/ C
5 # (' % (
) 8
9
= ( )
777 CC*C>) *
* * /
F Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
6
4 / ( L
?@ A 777) , & + ( +
) <
< / =
H
% ( !!) 0 . ( 7) #
.
( >) ?@ A (77 77) 5 . (77) O (77) , (77C) . (77C) N (776 77>) ?@ (77!) = ( )
G
: /
+ -. 6 + ? ? -. 2 4P ( 6) 5: ? /
< & ! (O ")
. (77") / "
? : . / () @ / L> () F /
L (") 7 F
@ 0 (77>) ? 6
/ @
-. % G
. 20#+. (
) # ( ! 77) ?@ A (777) # 3 (77C) >
/ = ( ) () L ( ) /
= L ( ) -. = Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
>
89 ( 89) #
*%2 %
@
.
I ( ) " >*! ? . (77") < . =
B (
) ! 5 , (77C)
, < (77C) & /
<
( "
) F
@ % @
= % ?@ (77!) G 89 2 =
= % G
* . (77")
* B
< ( . 77 C *67) 2 !
I ? 6 L 76 ( ) = ( 2%.
77"L . 77) . 4 (77>)
2 -. , , (77C) /
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
B
5 . (77) 8 J K 9
( > ) % . (77") ( ) 2 ( ) . (77") .$4
% .$4 < H
.$4 ? 7 + ? / ( 4 2Q 77"L ?2 77CL 2
#
77>) I H
? I & +, - ,% A /=
89
77 ? ( CC77 ) $
, #
O (77) ,
/ & ()
L () /
( ) L (") /= L (C) , L (6)
$
" 77
.
8 9 ( ) /= F/ ( O 77) # .
/
F .
( !!) #
.
( >)
7
? -. = .$4 % (77>)
# 89 .$4 -. Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
!
% / () 6 ( ) ( ) ( , ) () #
B7 " $
" 6 77
I
/ (
/) ,% 2
, 6 () % / , # .
( > !" )
O (77 6! )
% : @
G
/ %
= 8 #
9 = &
<& ' . '
2 /
< % 8
#
9 @
? % : : =
( 2 # 77 /
/
) I F
?
# . (77") # .
( >) &
.4+?$lw @ +lw lw lw , . lw @ 4 :3 @ 4<<8 5335
()
.4+?$lw @ @
L
lw @ L
lw @ L
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
lw @ @
.4+?$lw
+ , = .4+?$E &
.4+?$lw @ . lw .
N
S
n
n nlw .
M
S
m
m mlw . lw
()
A (O 77L . 77" ) % % ( I 23 #
2-3$. 2I21 0 )
(
) 6 77">C # @
/ .4+?$lw (= ) , .4+?$ (7>R ) , ( 6* B) ! ( R ) = .4+?$ .4+?$
7"6R 6 R 77 ( ) = ( !!) 0 . ( 7) # .
( >) = () .4+?$ < O
(77) , .4+?$ @ (0 #
0
% 3 ., - N - .) ? / = "
# L : % ( , ) 8
9
8
9 (=E203) <
(1) / C
/
-. ? Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
7
! + ! - . ? ( <
N
77L . 77)
Wdeoh 4 derxw khuh
.4+?$ (
";C) ? # 0 #
(76 R 76 R 7B6R ) ? (7!BR)
, = .
-N . % .
= ? . (77") / 0
E-N /
E ? Wdeoh 5 derxw khuh
(lw ) /
=
6 2
/
(. 4: 5: #
*%2)
4 H
/
# @
= # .45#%lw .4+?$ G
8 9 F ( )
0 (B SE) ? ("! *E") 5 .4+?$ .45#% /
( " )& F 4 H
.4+?$ /
R 6
E
8
9 O +
8F
9 /
(
T6")
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
Wdeoh 6 derxw khuh
G
/ % 1
( ) ( 89 ) =
/
, =
= F
5: 2 #
. (52#.) #
*%2 % (#%%) /
= 4 H
7B> @
( CR ) 52#.
#%% ( .45#% )
/ ( 222SE2)
G
( 52#.#%% ) " ( 2)
=
$E?S /
(
6;>) @
/
0 L #
% @
-N ., =>
= .4+?$ (nlw ) (mlw )
( C = )
Wdeoh 7 derxw khuh
6*77 D /
. ($
" ) /
/ (nlw ) L 89 2 / /
/
2
@ / N >
# . -N 52#.#%% Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
(776) # .
( >) O (77) .4+?$ O " ( > 77") /
#
() .
/
O < $
" I1. (.1.) L $
= , (-.U >7B ) F (BR ?) " # 7R /
=
#
*%2 . 5 %
0 #
!"# $ 6 (
*") I1. = H
(.45#%)
/
( R ) =
& 77!R ( ! ) B
Wdeoh 8 derxw khuh
, ((%.$)) /
H
/
: ( 20#+. L
2 77") /
H
/
& , . /
= .4+?$ B
O (77 " 6B) -. B 77> E 8
9
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
"
w @ 4<<; 5335 /
H
. ! + /
1 5 /
( ? . ) % /
(
L H
=
/
/
( )
# /
G
& % %2(
( C ;") H
/
() @ F -.
() A % H
/
8
9 $3 % = /
H
. . 2 ? / 2 = $3 ! . (77") @
-N ? G
: I -N ? F = $3
!
. (77") 0 (77>) / 2 : /
? 5 # = -N ? 3 (
-N ) ( -. ,) % ?
8= 9 = -N . (77 C) "R !! 777&V ? / = D . 8= 9 % > ? ( >7) (
1= %) -N : 8= 9 ?= 8-N 9
< ,, Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
C
/= @
( > *") H
.45#% /
R /
Wdeoh 9 derxw khuh
/ /
(.45#%) /
(52#.#%%) /
6 (
C*B) / , H
52#.#%% /
R (=
%
) @
.4+?$ (, %
) /
52#.#%% F
7 6
>;B H
%
/
R ? / G
# H
4 /
0 :
= ? =
0 / G
?
0 5 (56 ) F % 0: 56 () 5 / 5 =
(
E) . (77" C6!)
/ & ? / /
C7 7
A /
= @
#
*%2 . %
89
@
5 /
(%) E / % /
=
) ( 6 777L $
777L I
77) . (77") 2 (77") 0 (77>)
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
6
> / 52#.#%% H
/
R , %
H
/
=
I1. ( /= @
) B 0 / .4+?$
. (+I) @
.4+?$ /
(+IW5 ) < H
/
=
( ;6) 4
/ /
/ (5) = .4+?$ ( ) /
@
/
(
C) G
,
F
(75) = (5#
) /
Wdeoh : derxw khuh
(11+ ,5 ) = /
H
=
,5
/ < 5 G
/
, G
G
@
, (77C !) # .
( > ">) % /
: H
F , ( e f B) % /
/
/
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
>
I@
& I2. H
/
(
";C) < ( ") - ,% / @
(
, )
(%.$) *
% /
G
% %
= /
H
/
89 / ! 8
9 / 0 : /
( ) # , H
/
/= @
/
(
6;>) %
@
.4+?$ /
% 8 #
9 H
F
2 ( ) = /
/ = / % ( C*6 ) ,5 ( ";C) .
@ = / @ I @ A / H
/
=
/
G
( ) % @
/
( . 77") < /
B
, E % Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
B
"
I H
? ( 0 #
3) .4+?$ 6 (
= ) .4+?$
.
C $
/
( ) $
" H
( # .
>) .
F
T6" O < 8
9 $
"
( ) B
% N (776) .
/
/
(
) .4+?$w3 = (
/
) ! ( ) F ( & 5 5 (
/
) (
/
2) +I
+IW5 75 5#
L 5 11+
11+W5 = 5 L 5 " 89 (,5 ,525 )
= 5 / 8 #
9 & /
= Wdeoh ; derxw khuh
2 (%.$) - = / , ( "
# < !
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
lw EXw Y lw E=Xfrxqwu| Y) * & * = , (
) .45#% & m E 222SE2 ( ) 2 /
m @ 4 7 (E)
( . 77")
.4+?$ /&C
.4+?$lw @ l . lw . lw . lw . lw . lw . lw . lw . lw . lw (")
=
=
=
WW
=
=
WWW
=
=
WWW
=
=WWW
=
=
WWW
=
=WWW
=
WWW WW /
R
6R <
H
H
( ) F , F
/
& () @ 4:94 5 @ 45659 334L ( ) & @ 666
5 @ 5666 334 # H
e
/
& e
e
% (+I) = & E /
(5) = & E E E E
% ( ) = E
/
B
% .4+?$ , @ &
6* > B* ! *777 77*776 / .4+?$ /
( ) A =
( *777) .45#% /
H
# (77! R ) (7! 6R ) , H
52#.#%% /
6
C
A =
/
52#.#%% 6
A 6 77 (C7 ) Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
!"# "$ % &%' ( ( )* +& % )* % & % , ) &' ( - ) Vxssruw
) %'&. * /% ) % % , 0 00
1 !"# % *% ) ( ( &)* ,
*+ % * %&% & 2% + - 3 0300 & '2%
*% ) * % ) % &%' (4 ) Vxssruw $ +&$ %'&$ %
) " %$ & *( * ( ( & & ) 5
& 6,$ & % 07$ -% ( , &
% & % ) !"# !%& ( *( 8$ &% % 3 *& & ) 9#8:8"
; ) +*& 3 ( & !%& , 00
91
Frqfoxvlrqv dqg srolf| lpsolfdwlrqv
*&* & ) *& % &)* * ; , *& ) *( & ( &
) !%& % ( < &$ % = , *&
>vlqh txd qrq? & ) &
&% @ * &% ) )*
% && ) % (
& , * & , *( &') % ( & (') ) *
*% )% ) % ( ( *% A ,$ % % %
& ( & ) ( %$
/ & <@ 1( 9 9381:" 8% $ * * % ( *% +% , )
$ *&* ) ( % 6,$ , & % % *% ) ( ( &)* ,($ &% %
) ) % && !%& %
" $ *( & gr % )* % '
( & ) ( % % '*( & ) ( % %& !%& 8 ,
*&5 >vlqh txd qrq? & =%) # && & ) *( & !%& (
0
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
8$ % ( *( , ) A $ & $
, = )% ) !%&
9$ % % *', ( ( & +'
& % * % & ) *( &'
% 8 !%& %$ &%' ( 3 ) - 3 = & %* * %$ , * *&
$ % 9%*$ ) (
) *( & $ d sulrul$ % ) -
($ ) $ ) ) & %
)* & & ) ,$ * % % '* 8$ ) % * ,( * ( * %
% & % % ) *( & *
** 8 ** * ) *( &$ '
& ) * *& %$ ,, *&
) & ( ( * % ) )% A & % &* *( & * ( ( & * ) *%& ) * % )2% /+
) *( % +& !%& + %
& ) * %% ) &$ B 0
$ 0 ; $ % )* ) & )
'& %
CCC
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
$' 1 / = / / = (l l ) @ 4 &
@ . . . . / / % @
, - ( ) &
Sq
Sq
Sq
Sq
l l @ . l l . l l . l l
Sq
Sq
S
Sq
Sq
q
l l l @ l l . l l . l l . l l
S
S
S
S
Sq
l l @ ql l . ql l . ql l . ql l
Sl
Sq
Sq
Sq
Sq
q
l l l @ l l . l l . l l . l l
A l Z ' ' L
l Z '' L
Z = "
# & " 6 / H
( ) 3&
D #
" , M =
( J7!*K 0 % 3
., J7!*"K -. #
-N) =
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
$' + ) "
C
6
>
B
!
7
"
C
.;4
S
*
S
*
222S
222
222*
22S
22
22*
2S
5:
"
"
2 2 2 "
2
2
2"
2
#
S
*
S
*
222S
222
222*
22S
22
22*
2S
7
"
C
$
$*
?S
?
?*
) #
"
C
6
>
B
!
52#.
*
2S
2
2*
S
*
6
C
66
B
!6
7
6
"
*
#%%
E2
2
2 E
E$
$
$E?
?
*
$S
"
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
#
+2 2 5 0 5 !! 5 & 3
O 5 2 7 6 B*>7
2 77" 5 2 2 2 . A 4 2
-A #
5 77> [ . . 3 2 A 4 77>*!
2 ++ (77) & 85 $
. $& + . . %9
# + 2 ?
4
M 6 3 CC6
2 ++ # 5O 77 5 -. 2 < &
5 G
? #
+ > ">*" 6
2 0 # + . (20#+.) - # + 2 .@ . B
2 0 # + . -. $ (20#+.;$)
777 # $ 5 . $ # + 2 A
2 # 4P # 6 5 : & O
#
.
+
"7"*"6
A " C
2 O 2 #
*6 A 77> 0 # ?
+ >
B7*B"
, $5 <
$ N 15 77C -. , & # + 2 3 D
?
4
+ 7 B"* , $5 < 4 77C $ [ O #
% " *""
$D ++ # 5O 1 AA .
.5 77 = O 5 2 "" 77* 6
?@ $ O O 3 (77!) 4 + . $ 4 ?
5 $
2 A? % <
?@ $$ A 1$ 777 . # + 2 A 4 777*7B
?@ $$ A 1$ 77 . O #
.
+
7 *C6
?@ $$ A 1$ 77 5 ,& . = O 2 #
> ! *77 ? 2 (?2) 77C #
. + $
# $ %
(#$%) !" $ %
?
#$% A
C
Eh|rqg wkh Wklug Sloodu ri Edvho Wzr
# 5O ! - & -.
O 5 2 "7 B"*"76
# 5O 77 8 9 O #
.
+
7
7B* # 5O 3 . 77C 5 -. /
/& +
& < A N 00 2 N (?) 5 $
% 5% 4 !B*77
# 5O .
.5 & !"*
> ?
O #
6 "CB*"BB
0 0 . 5 7 5 O 5
2 *!
0 + M O M 0 77> ? O 5 2 "! "
*C!
<
$ N 15 77 - &
% [ O #
.
+
7 CB*!B
< N5 N 0M 4 5I " I@
& O ?
#
B > *!"
< 77" 4 $ - 4
O O N 00 1= 77 @
& ?
#$%% O #
+
6 66 *6B6
N 3M +
4< O2 776 5 & $
[ O #
>7 "C"*"B!
N 3M +
4< O2 77> I O 5 2 "! 6C6*6BC
5 $4 77 + & + ?
+ !BC*!!!
4 5 2Q 5 77" /
.
+ ?
+ 6*!
4 77> 5 & N / O #
. !>*"7
+
O 77C +
2 %% # + 2 3 D ?
4
+ 7 B*
+
O A . 5 2 [ 4
- 4
. O 77C $ 8
9 [ # + 2 3 D
. O 77> A / @ & O 2 #
"7 B6*B">
. 77 ? : .3$ O #
.
+
7 ""*>>
. 77" ? & ?
O 5 2 "6 CC"*CB
Table 1
SPREAD and Ratings -- Distribution by Year
SPREADa (%)
SP-M-FIb
MBFS-FIIc
Year
N
Mean
Std. dev.
N
Mean
Std. dev.
N
Mean
Std. dev.
1995
50
0.544
0.352
47
3.68
1.83
33
4.27
2.14
1996
56
0.641
0.524
48
3.70
1.90
39
4.15
2.10
1997
65
0.745
0.724
52
3.71
1.86
41
4.34
2.12
1998
70
1.032
0.976
53
3.68
1.69
48
4.78
2.19
1999
70
0.903
0.832
55
3.74
1.60
51
4.85
2.05
2000
70
0.994
0.751
58
3.73
1.52
54
4.85
1.97
2001
70
0.938
0.804
63
3.97
1.57
55
4.65
1.71
2002
70
1.020
1.124
63
4.15
1.65
56
4.84
1.85
1995--2002
521
0.871
0.825
439
3.81
1.69
377
4.63
2.00
a
SPREAD is the difference between the bond yield to maturity and the yield of a corresponding currency Treasury security
calculated by substituting the value for the life of the bond into the cubic equation that describes the benchmark yield curve of
corresponding sovereign.
b
SP-M-FI is calculated as the mean of long term issuer ratings assigned by Moody’s, Standard and Poor’s and Fitch-IBCA,
converted to cardinal values as shown in Appendix 2a. A higher credit quality corresponds to a lower cardinal number.
c
MBFS-FII is the mean of financial strength ratings assigned by Moody’s/Fitch-IBCA, cardinalized as shown in Appendix 2b.
Country
No. of
banks
SPREAD (%)
Ratings
AISD
Total assets
ROAA
Leverage
Net loans
LLR
Bad loans
Mean
Std. dev.
SP-M-FI
MBFS-FII
(m. US$)
(bn. US$)
(%)
(%)
(%)
(%)
Austria
4
0.59
0.34
3.96
6.35
56.77
57.99
0.33
26.30
56.11
4.29
3.74
Belgium
2
0.67
0.34
4.09
4.25
64.78
229.00
0.55
26.88
40.64
1.88
2.83
Denmark
1
0.24
0.11
NA
NA
1,408.69
5.75
0.16
18.30
33.26
1.29
1.54
France
13
0.75
0.48
4.61
5.53
403.40
191.00
0.56
25.91
50.04
4.31
6.12
Germany
11
0.59
0.47
2.71
5.26
213.54
260.00
0.21
35.58
49.67
2.37
2.79
Ireland
1
0.99
0.31
4.58
4.00
158.38
56.36
1.25
16.22
66.99
1.07
1.73
Italy
5
1.76
2.01
5.79
6.77
224.26
99.51
0.44
17.81
59.06
3.76
8.21
Luxembourg
3
0.87
0.60
3.67
3.30
29.16
31.18
0.62
29.25
17.55
NA
NA
Netherlands
7
0.49
0.27
2.78
3.26
239.48
222.00
0.64
20.38
62.31
1.26
1.63
Norway
1
1.75
1.07
6.00
NA
106.44
36.06
0.91
15.75
71.06
2.47
1.62
Spain
4
1.68
1.03
4.53
4.18
200.00
213.00
0.76
15.91
48.78
3.00
2.78
Sweden
2
2.09
1.53
5.03
5.22
150.00
101.00
0.63
24.27
56.99
1.04
2.32
Switzerland
6
0.70
0.44
2.87
2.81
224.14
211.00
0.42
19.10
55.43
3.95
4.47
UK
10
1.00
0.46
3.41
3.29
330.56
188.00
0.81
21.08
65.19
1.27
2.21
Total
70
0.87
0.82
3.81
4.63
260.17
179.00
0.53
24.35
53.58
2.90
3.99
a
SPREAD is the difference between the bond yield to maturity and the yield of a corresponding currency Treasury security obtained by substituting the bank debt maturity in the cubic
equation that describes the benchmark yield curve of corresponding sovereign. SP-M-FI is the mean of traditional credit ratings (Standard and Poor’s, Moody’s and Fitch-IBCA)
cardinalized as shown in Appendix 2a. MBFS-FII is the mean of financial strength (Moody’s) and individual (Fitch-IBCA) bank specific ratings (cardinalized values). AISD is the US dollarequivalent amount outstanding. ROAA is the ratio of annual net income to the average of the preceding and current year-end total assets. Leverage is the ratio of total (book) liabilities to
the book value of equity. Net loans -- the ratio of net loans to total assets. LLR is the reserve for loan losses expressed as percentage of total loans. Bad loans -- the ratio of total problem
loans to total (net) loans. NA = data non available
Table 2
a
Sample Descriptive Statistics -- Distribution by Country
Table 3
a
SPREAD -- Distribution by Rating Classes
N
Mean
Std. dev.
Minimum
Maximum
Quartiles
Lower
Median
Upper
Panel A: Traditional ratings (S&P, Moody’s, and Fitch-IBCA)
AAA/Aaa
51
0.34
0.28
–0.34
1.46
0.23
0.30
0.45
AA/Aa
251
0.84
0.57
–0.06
4.94
0.47
0.72
1.10
A
112
1.22
1.31
–0.88
7.78
0.46
0.90
1.27
BBB/Baa
3
1.26
0.53
0.66
1.68
0.66
1.43
1.68
NR
104
0.83
0.68
–1.01
3.20
0.39
0.59
1.14
Total
521
0.87
0.82
–1.01
7.78
0.40
0.68
1.10
Panel B: Financial strength ratings (Moody’s and Fitch-IBCA)
A
43
0.50
0.41
–0.34
1.73
0.23
0.48
0.71
B
212
0.92
0.70
–0.06
5.12
0.48
0.77
1.13
C
87
1.09
1.39
–0.33
7.78
0.39
0.56
1.16
D
20
0.82
0.47
–0.02
1.94
0.47
0.80
1.03
NR
159
0.79
0.62
–1.01
3.20
0.37
0.64
1.10
Total
521
0.87
0.82
–1.01
7.78
0.40
0.68
1.10
SPREAD is the difference between the bond yield to maturity and the yield of a corresponding currency Treasury security
obtained by substituting the value for the life of the bond into the cubic equation describing the benchmark yield curve of
corresponding sovereign. The traditional ratings are represented by the mean of long term issuer ratings assigned by Moody’s,
Standard and Poor’s, and Fitch-IBCA, cardinalized as shown in Appendix 2a. The financial strength ratings are represented by the
mean of the bank specific ratings assigned by Moody’s and Fitch-IBCA (cardinalized values). The AA class includes the AA+, AA,
AA– sub-classes, etc. NR = no rating available
a
Variable
ROAA
Table 4
Description of the Main Explanatory Variables
Definition
Return on average assets, calculated by dividing the annual net income to the
average of the preceding and current year-end total assets
Expected sign
+/–
Cooke
Leverage
Total capital adequacy ratio calculated according to Basel I
Financial leverage, calculated as the ratio of total (book) liabilities to the book
value of equity
+
NetLoans
Ratio of net loans to total assets, a measure of the opaqueness of banking
firm’s assets
+
Liquidity
Liquidity ratio, indicating what percentage of customer and short term funds
could be met if they are suddenly withdrawn
–
LLR
Ratio of loan loss reserves to total (gross) loans, indicating how much of the
total credit portfolio has been provided for but not charged off
+/–
BadLoans
Ratio of total problem loans to total (net) loans, a proxy for the quality of the
loan portfolio
+
OBSA
ROAA*Lev
LLR*Lev
BadLoans*Lev
ln(AISD)
Ratio of total off-balance sheet activities to total assets
Product of ROAA and Leverage
Product of LLR and Leverage
Product of BadLoans and Leverage
Log of the outstanding amount of the issue, expressed in thousand US$, a
proxy for liquidity effects on spreads
Maturity
Įt, t={95,…,02}
Remaining maturity, expressed in years
“Year” dummies, quantifying the inter-temporal variations in bond market
conditions. One of the eight dummies (Į1995) was eliminated to avoid collinearity
in the data
Recession
Takes the value of 1 if the economy is in recession and 0 if it is in expansion.
We used the OECD Composite Leading Indicators to identify the troughs and
peaks of the business cycle in each European country
France, Germany,
“Country” dummy variables, capturing both differences in macroeconomic
Sweden,
conditions and differences in safety nets across countries
Switzerland, UK
–
–
–
–
–
–
+
+
+
+/–
Subordinated
Takes the value of 1 if the (unsecured) debt has a subordinated status and 0
otherwise
+
Split
Dummy variable equal to 1 if the bank issuer received a rating from Moody’s
that is different from the rating it received from S&P. According to Morgan (2002),
split ratings are a sign that the bank’s financial condition is opaque to investors
+
Split*Recession
TBTF
Support
Product of Split and Recession, to investigate Santos’s (2006) thesis that the
impact of split ratings on credit spreads in recessions is different from the
corresponding impact in expansions; we consider the use of split ratings both
when the risk measure is average ratings and when accounting ratios are used
as risk proxies
+
Ratio of total assets of bank i at the end of the year t to the maximum total
assets at the end of the same year; alternative measures of the TooBigToFail
variable, also used by Sironi (2003), are discussed in §5.4
–
“External support” dummy variable, taking the value of 1 if the issuer is either a
public-sector bank or a bank that benefits from explicit/implicit governmental
guarantees and 0 otherwise. The presence of explicit/implicit guarantees is
confirmed by a Fitch–IBCA Support rating equal to 1
–
Table 5
OLS Pooled Regressions: Traditional Ratings, Financial Strength (Bank-Specific) Ratings, and Governmental
a
Guaranties
Independent
Variables
Constant
SP-M-FI
MBFS-FII
(1)
3.254***
(3.102)
0.081***
(3.967)
_
Traditional ratings
(2)
(3)
3.554***
3.364***
(3.376)
(3.255)
0.064***
0.079***
(3.354)
(3.860)
_
_
(4)
4.053***
(3.838)
_
0.060***
(3.292)
0.308***
(3.801)
_
Financial strength ratings
(5)
(6)
4.026***
4.434***
(3.817)
(3.982)
_
_
0.061***
(3.357)
0.141
(1.486)
0.352**
(2.130)
–0.336***
(–3.986)
0.024***
(2.834)
–0.078
(–0.739)
–0.087
(–1.208)
0.339**
(2.472)
0.569**
(2.456)
–0.007
(–0.058)
1.020**
(2.198)
_
0.047***
(2.809)
0.262***
(3.058)
_
(7)
4.528***
(4.345)
_
0.092***
(4.161)
Split
_
0.204***
0.131
0.046
(2.919)
(1.506)
(0.342)
Split*Recession
_
_
0.334**
0.273*
(2.204)
(1.797)
ln(AISD)
–0.265***
–0.297***
–0.274***
–0.337***
–0.354***
–0.388***
(–3.241)
(–3.626)
(–3.399)
(–3.992)
(–4.063)
(–4.557)
Maturity
0.019**
0.014*
0.022***
0.022***
0.019**
0.024***
(2.382)
(1.748)
(2.667)
(2.632)
(2.185)
(2.700)
Subordinated
0.001
0.100
–0.055
–0.076
0.055
–0.126
(0.016)
(1.113)
(–0.607)
(–0.725)
(0.613)
(–1.176)
Germany
–0.047
0.082
–0.002
–0.078
–0.119*
–0.217***
(–0.693)
(1.219)
(–0.029)
(–1.061)
(–1.640)
(–2.655)
UK
0.205*
0.340***
0.213*
0.373***
0.208*
0.325**
(1.746)
(2.870)
(1.749)
(2.769)
(1.703)
(2.305)
Switzerland
0.026
0.136
0.048
0.662***
0.671***
0.487**
(0.244)
(1.395)
(0.491)
(2.918)
(3.189)
(2.279)
France
–0.115
0.074
–0.159
–0.006
0.047
–0.072
(–0.951)
(0.653)
(–1.320)
(–0.048)
(0.405)
(–0.527)
Sweden
1.001**
1.030**
0.926**
1.041**
0.855*
0.986**
(2.012)
(2.162)
(1.979)
(2.195)
(1.800)
(2.015)
TooBigToFail
_
_
_
_
_
0.484**
(2.376)
Support
_
_
_
_
_
–0.213***
–0.144*
(–3.185
(–1.841)
Fa
9.958***
13.372***
10.210***
10.010***
9.777***
10.182***
10.757***
N
417
417
417
362
362
362
362
Adjusted R2
0.256
0.349
0.285
0.298
0.304
0.326
0.363
a
Dependent variable is SPREAD (%) calculated as difference between actual yields on the bank debt and the constructed
yield on a corresponding treasury security with the same maturity. Explanatory variables are defined as follows: SP-M-FI -- the
average traditional credit rating assigned by S&P, Moody’s, and Fitch, converted to cardinal values; MBFS-FII -- the average
financial strength rating (cardinalized values); Split -- dummy variable that takes the value of 1 if Moody’s  S&P;
Split*Recession -- the product of Split and Recession (a dummy that equals 1 if the economy is in recession); ln(AISD) -- the log
of the US dollar-equivalent amount of the issue (in thousand); Maturity -- the remaining maturity expressed in years;
Subordinated equals 1 if the bond is subordinated; Germany, UK, Switzerland, France, Sweden -- country dummies equal to 1 if
the issuing bank is headquartered in the corresponding country; TooBigToFail -- the ratio of total assets of bank i at the end of
the year t to the maximum total assets at the end of the same year; Support equals 1 if the issuing bank is a “public-sector” one
(i.e. either government-owned or having a Fitch-IBCA Support rating equal to 1). All regressions include year dummies (not
reported). Equations are estimated by OLS (pooled). Estimated standard errors are computed using White’s method.
Heteroskedasticity-consistent t-statistics are reported in parentheses below each coefficient estimate. Fa -- regression F-statistic
(df = [Ncoeff–1], [Nobs–Ncoeff]).
***, **, and * indicate statistical significance at the 1%, 5%, and 10% level, respectively.
Table 6
a
Within Regressions: Traditional Credit Ratings, Financial Strength Ratings, and Governmental Guaranties
Independent
Variables
SP-M-FI
MBFS-FII
(1)
0.190***
(4.477)
_
Traditional ratings
(2)
(3)
0.189***
0.192***
(4.493)
(4.580)
_
_
(4)
_
0.144***
(5.849)
0.046
(0.592)
_
Financial strength ratings
(5)
(6)
(7)
_
_
_
0.144***
(6.011)
–0.064
(–0.720)
0.228**
(2.173)
_
0.153***
(6.128)
–0.068
(–0.776)
0.254**
(2.408)
–1.001***
(–2.651)
_
0.127***
(5.023)
0.005
(0.048)
0.211**
(2.059)
_
(8)
_
0.149***
(5.626)
Split
_
0.155*
–0.018
–0.063
(1.677)
(–0.174)
(–0.692)
Split*Recession
_
_
0.166*
0.252**
(1.823)
(2.368)
TooBigToFail
_
_
_
_
–0.951**
(–2.463)
Support
_
_
_
_
_
–0.370*
–0.125
(–1.728)
(–0.645)
Fa
131.741*** 116.587*** 103.526***
105.852***
95.337***
86.542***
86.157***
78.460***
Fb
9.170***
8.096***
8.942***
8.817***
8.906***
8.594***
8.504***
8.015***
N
417
417
417
362
362
362
362
362
Adjusted R2
0.672
0.674
0.674
0.684
0.687
0.690
0.688
0.689
a
Dependent variable is SPREAD calculated as difference (expressed in percentage) between actual yields on the bank debt
and the constructed yield on a corresponding treasury security with the same maturity. Explanatory variables are defined as
follows: SP-M-FI -- the average traditional credit rating assigned by S&P, Moody’s, and Fitch, converted to cardinal values;
MBFS-FII -- the average financial strength rating (cardinalized values); Split is a dummy variable that takes the value of 1 if
Moody’s  S&P; Split*Recession -- the product of Split and Recession (a dummy that equals 1 if the economy is in recession);
Įt -- year dummies (not reported); TooBigToFail is the ratio of the issuing bank’s total assets to the total assets of the largest
bank in the sample in the year of the observation; Support equals 1 if the issuer is a “public” bank (i.e. either government-owned
or having a Fitch-IBCA Support rating equal to 1). Equations are estimated by OLS with inclusion of fixed effects (unreported
intercept terms). Estimated standard errors are computed using White’s method. Heteroskedasticity-consistent t-statistics are
reported in parentheses below each coefficient. Fa -- regression F-statistic (df = [Ncoeff–1], [Nobs–Ncoeff]); Fb -- heterogeneity
specification test: pooled model vs. fixed effects model (see e.g. Hsiao, 2003).
***, **, and * indicate statistical significance at the 1%, 5%, and 10% level, respectively.
Table 7
a
OLS (Pooled & Within) Regressions: Accounting Variables
Independent
Variables
Constant
OLS -- pooled
OLS -- fixed effects
(2)
(3)
(4)
(5)
(6)
3.125***
5.681***
6.172***
_
_
(3.798)
(5.170)
(5.234)
ln(AISD)
–0.231***
–0.435***
–0.443***
_
_
(–3.510)
(–5.725)
(–5.510)
Maturity
0.039***
0.003
0.030*
_
_
(3.368)
(0.182)
(1.942)
Subordinated
0.182*
0.240**
0.527***
_
_
(1.900)
(1.987)
(4.296)
Split
0.277***
0.263**
0.006
0.184*
–0.007
(3.218)
(2.430)
(0.058)
(1.787)
(–0.079)
Leverage
_
–0.006
_
0.031*
_
(–0.617)
(1.912)
Cooke
–0.000
_
–0.050*
_
–0.002
(–0.092)
(–1.885)
(–0.075)
ROAA
0.271
0.531**
0.328
0.307
0.827**
–0.117
(1.479)
(2.403)
(1.321)
(1.044)
(2.114)
(–0.334)
ROAA*Lev
–0.015**
–0.024**
–0.011
–0.008
–0.027*
0.012
(–1.966)
(–2.386)
(–0.966)
(–0.615)
(–1.759)
(0.878)
NetLoans
–0.002
–0.004
0.003
0.002
0.003
–0.008
(–1.064)
(–1.476)
(0.582)
(0.312)
(0.480)
(–1.118)
Liquidity
–0.004**
0.001
0.007*
0.011**
–0.006
–0.013***
(–2.178)
(0.233)
(1.799)
(2.365)
(–1.281)
(–2.606)
LLR
_
_
0.459***
0.403**
0.221
0.064
(3.363)
(2.490)
(1.403)
(0.262)
LLR*Lev
_
_
–0.014***
–0.016***
–0.010*
–0.004
(–2.886)
(–3.025)
(–1.868)
(–0.553)
BadLoans
_
_
–0.390***
–0.371***
–0.170
–0.227
(–3.482)
(–3.273)
(–1.417)
(–1.380)
BadLoans*Lev
_
_
0.013***
0.013***
0.007*
0.011*
(2.874)
(3.077)
(1.683)
(1.904)
OBSA
_
_
–0.000**
–0.000***
_
_
(–2.473)
(–2.656)
TooBigToFail
0.081
0.186
0.947**
0.374***
–0.769**
–1.365**
(0.532)
(1.097)
(2.552)
(3.563)
(–2.252)
(–2.322)
Support
–0.102**
–0.177***
–0.271**
–0.316**
–0.618***
–0.372
(–2.256)
(–2.683)
(–2.180)
(–2.129)
(–2.817)
(–1.499)
Fa
10.934***
7.435***
7.535***
6.874***
30.845***
32.293***
Fb
1.960*
2.264**
3.100***
2.778***
1.725
2.835***
Fc
_
_
4.689***
3.716***
1.139
4.350***
Fd
_
_
_
_
5.332***
6.883***
N
514
395
259
224
285
241
Adjusted R2
0.317
0.273
0.424
0.433
0.645
0.734
a
Dependent variable is (end-of-January) SPREAD (%). ln(AISD) -- the log of the US dollar-equivalent amount of the issue
(in thousand); Maturity -- the time to maturity (expressed in years); Subordinated equals 1 if the bond is subordinated; Split is a
dummy variable that takes the value of 1 if Moody’s  S&P; Leverage -- the ratio of total (book) liabilities to the book value of
equity; Cooke -- the total capital adequacy ratio under the Basle rules; ROAA -- the ratio of annual net income to the average of
the preceding and current year-end total assets; ROAA*Lev -- the product of ROAA and Leverage; NetLoans -- the ratio of net
loans to total assets; Liquidity -- a deposit run off ratio; LLR -- the reserve for losses expressed as percentage of total loans;
LLR*Lev -- the product of LLR and Leverage; BadLoans -- the ratio of total problem loans to total (net) loans; BadLoans*Lev -the product of BadLoans and Leverage; OBSA -- the ratio of total off-balance sheet activities to total assets; TooBigToFail is
the ratio of the issuing bank’s total assets to the total assets of the largest bank in the sample in the year of the observation;
Support equals 1 if the issuer is a “public” bank (i.e. either government-owned or having a Fitch-IBCA Support rating equal to
1). All regressions include year dummies (not reported). Equations are estimated by OLS pooled and with inclusion of fixed
effects (unreported intercept terms). Estimated standard errors are computed using White’s method. Heteroskedasticityconsistent t-statistics are reported in parentheses below each coefficient estimate. Fa -- regression F-statistic (df=[Ncoeff–1],
[Nobs–Ncoeff]); Fb -- F-statistic for the hypothesis that the x bank-specific variables’ coefficients are jointly zero (df=x,[Nobs–
Ncoeff]); Fc -- F-statistic for the hypothesis that all y coefficients of the credit quality measures equal zero (df=y,[Nobs–Ncoeff]);
Fd -- specification test: pooled model vs. fixed effects model (see e.g. Hsiao, 2003).
***, **, and * indicate statistical significance at the 1%, 5%, and 10% level, respectively.
(1)
2.655***
(4.494)
–0.191***
(–3.687)
0.022***
(3.129)
0.293***
(3.991)
0.157**
(2.174)
0.004
(0.833)
_
Table 8
a
The Accounting Empirical Model: Correcting for the “Error in Variables” Problem
Model 1
Model 2
Model 3
(A)
(B)
(A)
(B)
(A)
(B)
Control variables
15.528***
7.760***
10.311***
5.843***
3.831***
4.551***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Accounting variables
1.916*
2.117*
2.286**
1.925*
2.279**
2.698***
(0.091)
(0.063)
(0.028)
(0.067)
(0.019)
(0.006)
N
436
320
318
229
226
182
Adjusted R2
0.216
0.198
0.288
0.203
0.254
0.303
a
This table presents the partial F statistics and p values (in parentheses) for each block of explanatory variables (control
variables and accounting variables, respectively). Dependent variable is (end-of-December) SPREAD (%). Model 1 includes
the control variables (viz. ln(AISD), Maturity, Subordinated, Split, TooBigToFail, Support, “country” dummies, and time fixedeffects), as well as the following accounting variables: Leverage (specification A) or Cooke (specification B), ROAA,
ROAA*Lev, NetLoans, and Liquidity. Model 2 includes the credit-quality measures LLR, LLR*Lev in addition to all the
explanatory variables included in Model 1. Finally, Model 3 (the “full” model) includes two other credit-quality measures,
BadLoans, BadLoans*Lev, in addition to all the explanatory variables included in Model 2. Estimations are performed by using
the standard two-stage least squares (TSLS) procedures on the pooled sample of banks. The bank-specific (accounting)
variables are instrumented by the control variables, lagged accounting variables, and SPREADt–1.
***, **, and * indicate statistical significance at the 1%, 5%, and 10% level, respectively.
Marginal contribution of each
block of explanatory variables