Politicians and Financial Supervision Unification outside the Central Bank:
Why Do They Do it?
Donato Masciandaro♦
February
2008
Abstract
An increasing number of countries show a trend towards a certain degree of consolidation of
powers in financial supervision, which has resulted in the establishment of unified regulators,
that are different from the national central banks. By contrast a high involvement of the
central bank in supervision seems to be correlated with a multi – authorities regime (central
bank fragmentation effect). This paper, using a simple application of a general common
agency game, sheds light on which conditions the politicians prefer when implementing a
unified sector supervision outside the central bank. From a theoretical point of view the
quality of public sector governance plays a crucial role in determining the supervision
unification. Focusing on the behaviour of the “good” policymaker (helping hand type), it will
prefer a unified financial authority that is different from the central bank if the correspondent
welfare gains – linked to at least one of the three effects: moral hazard, conflict of interest,
bureaucracy – are considered higher respect to the information losses. The “bad” policymaker
(grabbing hand type) will choose the single financial authority if the financial industry likes it,
and the central bank is not a captured one. On the other hand, the paper tests the model,
confirming the robustness of the institutional position of the central bank in explaining the
recent trend in supervision consolidation, with an empirical analysis performed with ordered
functions on an updated dataset.
JEL Classification: G18, G28, E 58. Key words. Financial Supervision, Central Banks, Political
Economy
♦
Full Professor in Economics, Chair in Economics of Financial Regulation, Bocconi University, Milan. Email address:
[email protected].
Politicians and Financial Supervision Unification outside the Central Bank:
Why Do They Do it?
1. Introduction
Over the last few years the financial supervision landscape has been radically transformed.
Many countries have made drastic changes to the architecture of financial supervision, and
more are contemplating modifications. In the last twenty years (1986-2006) 94% of the
countries included in our sample of 91 nations chose to reform their financial supervisory
setting (Figure 1). The current restructuring wave is making the supervisory regimes less
uniform than in the past. In several cases the architecture still reflects the classic model, with
separate agencies for banking, securities and insurance supervision. However, an increasing
number of countries show a trend towards a certain degree of consolidation of powers, which
in several cases has resulted in the establishment of unified regulators, that are different from
the national central banks.1 Other countries are contemplating further modifications.
How can this supervision consolidation trend outside the central banks be explained? Firstly
it is necessary to point out that the quest for the optimal level of financial supervision
unification cannot be pursued through a traditional cost-benefit analysis. If one proposes to
compare alternative models from a social welfare standpoint, one realises that each of them
offers expected benefits but also expected risks, and the final outcome is actually
undetermined. In general, different arguments for consolidation of supervision can be
identified. However, there are also several arguments against unification, whose cons
systematically mirror the pros. Different review essays (Abrams and Taylor 2002, Cihak and
Podpiera 2007a and 2007b) have demonstrated that there are not strong theoretical
arguments in favour of any particular architecture of financial supervision, given that it is
possible to provide advantages and disadvantages of each model.
Therefore, if the main driver of the supervisory reform cannot be the classic social welfare
perspective,
it is necessary to explore alternative views. Recent studies have tried to shed
light on the determinants of the shape of the financial supervision, focusing on the
policymakers’ point of view.
In particular, when the policymakers implement a consolidation process, they seem to be
influenced by the supervisory position of the central bank. Barth, Nolle, Phumiwasana and
Yago 2002, Arnone and Gambini 2007, Cihak and Podpiera 2007a claim that the key issues for
supervision are 1) whether there should be one or multiple supervisory authorities and 2)
whether and how the central bank should be involved in supervision. More importantly, the
two crucial features of a supervisory regime seem to be related.
1
For a survey see e.g. De Luna Martinez 2003, Masciandaro 2005, and Cihak and Podpiera 2007b. The legal issues
are described in Mwenda 2006.
2
The descriptive analysis (Masciandaro 2004) signals an intriguing result: the national
choices on how many agencies must be involved in supervision is strictly dependent on the
existing institutional position of the central bank. The degree of unification seems to be
inversely related to the central bank involvement in supervision. The trade-off – and the
consequent so called central bank fragmentation effect - was confirmed first by an analysis of
the reforms in the supervisory regimes (Masciandaro 2006) and then by going more deeply
into the economics of the central bank fragmentation effect (Masciandaro 2007).
The central bank fragmentation effect was explained as a particular case of path
dependence: every policymaker, in choosing the level of financial supervision consolidation, is
influenced by the characteristics that already exist in terms of the central bank position. The
policymaker’s choices are viewed as a sequential process in which the institutional position of
the central bank matters.
Notwithstanding the evidence, at least two crucial questions are still unanswered. First of
all - from a theoretical point of view - in the path dependence approach the policymaker is a
homogeneous agent. But if we take into account the fact that in the real world the
policymakers can be heterogeneous, with different objective functions, can the central bank
fragmentation effect still be explained?
Secondly - from an empirical point of view - using the same approach the central bank
involvement in supervision is considered an independent variable: the policymaker, given the
central bank position, determines the degree of financial supervision unification.
It remains
uncertain whether the central bank involvement in supervision is only an imperfect
potentially endogenous
- proxy of other features in the central banking
- and
design that
influences the policymakers.
Therefore the aim of this paper is twofold. On the one hand, we offer a theoretical
framework which sheds light on
the role of central bank involvement as a potential
determinant of the supervision architecture, taking into account the existence of
different
types of policymakers. On the other hand, the paper tests the model, investigating the
robustness of
the role of the central bank in explaining
the recent trend in supervision
consolidation, with an empirical analysis performed with ordered
functions on an updated
dataset.
The paper is organised as follows. Section two describes the theoretical setup. In section
three and four we estimate the model, describing the dependent variable and the indicators
used to disentangle the role of central bank position among the different potential drivers of
the supervision reforms. Section five discusses the policy implications and put forwards some
conclusions.
3
2. Politicians, Financial Supervision Unification and the Role of the Central Bank
Our theoretical framework is based on two hypotheses. First of all, gains and losses of a
supervisory regime are variables computed by the incumbent policymaker, who maintains or
reforms the supervisory regime, following his
preferences. Secondly,
the policymakers are
politicians: politicians are held accountable at the elections for how they have pleased the
voters. All politicians are career oriented agents, motivated by the goal of pleasing the voters
in order to win the elections. The main difference among the various types of politicians
concerns which voters they wish to please in the first place.
In the economic literature, there are as yet no theoretical studies that consider the
policymaker objective function in determining
the level of unification of the financial
supervision regime. Here we propose a specific and original application of the general model
developed by Alesina and Tabellini (2003) to investigate the criteria with which to allocate
policy tasks to elected politicians.
Consider a society that assigns to an elected policymaker the task of designing the level of
consolidation of the institutional regime that guarantees the effectiveness of the financial
supervision policy (thereafter the effective level of unification). The effective level y of
unification is determined by the policymaker’s effort a and by his ability Ω:
y=a+Ω
(1)
Ability is a random variable; for the sake of simplicity, let us suppose that Ω can assume
two values only. The politician can be outstanding or not. Therefore the parameter can be
Ω L or Ω H with Ω L < Ω H and
⎧Ω L with probability p
Ω=⎨
⎩Ω H with probability (1 − p )
In order to show the importance of the politicians’ preferences rather than the social ones,
let us assume that the citizens care about the effectiveness of the supervisory regime
according to a classic well-behaved concave function u = U (y): the social welfare increases
with the level of unification. Linear preferences are used:
U(y) = y
(2)
The policymaker’s effort is costly, and the convex and increasing cost function is defined as
C = c (a). The reward for the policymaker is labelled R (a) . The policymaker’s utility function
is defined as:
R (a) – c (a)
(3)
4
We claim now that the costs of implementing an increasing level of financial supervision
unification can depend on the existing institutional position of a central bank. If a high level of
central bank involvement in supervision is the status quo, under specific conditions a unified
supervision is more difficult to implement, and this means that the politician’s task is, ceteris
paribus, more costly.
In order to identify these conditions, let us consider that a policymaker aiming to
consolidate supervision faces two alternative paths: to create a monopolist central bank in
supervision; to establish a unique financial authority, different from the central bank.
The creation of a monopolist central bank can be costly for different reasons. First of all,
the policymaker may dislike the implementation of a monopolistic central bank if
consequent
extension of
the classic
monetary policy and supervision policy
moral hazard risks -
the
which can occur when the
are delegated to the central bank - overcomes the
potential benefits in terms of informative gains (Goodhart and Shoenmaker 1995, Llewellyn
2005) (moral hazard effect).
Secondly, implementing a monopolistic central bank regime can be costly given that the
policymaker delegates the conduct of business controls to the central bank, which have not
traditionally sought to become involved in such matters of transparency and have therefore
focused on stability (Goodhart 2007 and Bini Smaghi 2007) (conflict of interests effect).
Thirdly, the policymaker has to take into account the risks of increasing the bureaucratic
powers of the central bank (bureaucracy effect). Table 1 documents cases of bureaucracy
effect.
But also the alternative solution – the building up of a unified supervisor that is different
from the central banker - can face difficulties caused by the central bank position. In fact the
policymaker may face costs in establishing a single financial authority, reducing the central
bank’s involvement in supervision, if the central bank’s reputation is high; at the same time, if
the reputation of the central bank is low, or decreasing, the establishment of a single financial
authority is more likely to occur (reputation effect). The reputation effect can work in both
directions. Cases of reputation effect are described in Table 1.
Therefore in implementing a supervisory reform the so called central bank fragmentation
effect (Masciandaro 2006 and 2007) - the more the central banker is involved in supervision
the less likely a unified supervision will be – can
occur. The central bank involvement is
supervision increases the costs in implementing a financial supervision unification.
Coming back to the delegation problem,
the sequence of events is as follows. Society
chooses to delegate the task of designing the level of unification of the supervisory
architecture. Next, the policymaker chooses effort a, before knowing his ability Ω in
implementing this particular task. Finally, nature chooses Ω, outcomes are observed and the
reward is paid.
Citizens observe the outcome, not the relationship between effort and ability. The
policymaker’s reward is based on the outcome y.
The crucial point is that the supervision
5
design effectiveness is observable ex post – i.e. a level of consolidation that produces effective
supervision
- but not contractible ex ante. In fact if the level of effectiveness cannot be
defined ex ante – the optimal level of consolidation so far does not exist – the policymaker
cannot be rewarded
directly through an explicit and verifiable contract. The policymaker is
rewarded based on observed ex post performances of the chosen level of financial supervision
consolidation.
The incumbent policymaker wishes to be re-elected. Now we can take into account the
possibility that different types of policymakers exist. On the one hand, one can adopt a
helping hand (HH) view (Pigou 1938) of the policymaker’s type: he is motivated to improve
the general welfare. The HH policymaker chooses
to maintain or reform his country’s
supervisory structure in an attempt to improve the efficiency of overall resource allocation. In
this case his re-election is more likely to occur if the citizens’ utility exceeds a threshold W.
Denoting by β the value of office and by a1 the effort, the reward function – given (1) and (2)
- for the HH policymaker is:
R(a1 ) = β Pr (U ≥ W )
R(a1 ) = β Pr (a1 + Ω ≥ W ) = β Pr (Ω ≥ W − a1 ) .
0 ≤ R(a1 ) ≤ β .
Given the citizens’ threshold, the three possible cases are as follows:
a) the ability of the politician is irrelevant:
Ω L ≥ W − a1 ; then Ω H > Ω L ≥ W − a1 and therefore Pr (Ω ≥ W − a1 ) = 1 and R(a1 ) = β .
b) the ability of the politician is crucial:
Ω L < W − a1 ≤ Ω H and R(a1 ) = β (1 − p ) .
c) the ability of the politician is not sufficient at all:
Ω H < W − a1 and R(a1 ) = 0
Voters are rational. They realise that the alternative to re-electing the incumbent is to get
another politician with average ability. It follows that:
W = a e + Ω AV with Ω AV =
Where
ΩL + ΩH
2
a e are the voters’ expectations. The HH policymaker chooses effort before observing
his talent in implementing a supervisory regime design, taking the expectations as given. The
utility function of the politician is:
6
Ω + ΩH
⎛
⎞
+ a e − a1 ⎟ − c(a1 )
R(a1 ) − c(a1 ) = β Pr ⎜ Ω ≥ L
2
⎝
⎠
The function can assume the following three values:
⎧
if
⎪β − c(a1 )
⎪
⎪
R(a1 ) − c(a1 ) = ⎨β (1 − p ) − c(a1 ) if
⎪
⎪
if
⎪− c(a1 )
⎩
ΩL + ΩH
+ a e − a1
2
Ω + ΩH
ΩH ≥ L
+ a e − a1
2
Ω + ΩH
ΩH < L
+ a e − a1
2
ΩL ≥
It is evident that the interest of the politician in implementing a greater level of supervisory
consolidation will depend, ceteris paribus, on the costs’ level. In particular, the more the
central bank is involved in supervision, the smaller the likelihood of a supervision consolidation
will be.
Alternatively we can use a grabbing hand (GH) view of the political process (Shleifer and
Vishny 1988). According to the grabbing hand approach, the policymakers are motivated by
the aim to please the interest of specific, well-defined voters. In our case, the financial
industry may be considered a highly organised and powerful interest group.
The financial
industry is likely to be a smaller and more coherent group than the consumers of their
services, and therefore politically better organised. The GH policymaker, in defining the
supervisory setting, depends on the market view of supervision, if this univocally determines
his re-election.
The preferences of the financial constituency can be written as:
V = (1 + δ ) y 2 − f
The parameter δ represents the importance of the supervisory consolidation goal for the
financial constituency. Using δ we can study the possibility that the central bank should be a
captured institution. In fact if we consider the existence of a financial lobby, we have to take
into account the possibility that also the central bank should please the financial constituency.
If the central is a captured agency, δ is equivalent to the degree of central bank involvement
in supervision: the more the central bank is involved in supervision, the greater the financial
constituency interest in implementing a more consolidated supervision ( δ ≥ 0 ). Alternatively, if
the central bank is an independent agency, the more the central banker is involved in
supervision, the smaller the financial industry preference toward more consolidation in
supervision ( δ ≤ 0 ).
The parameter f represents the campaign contributions; their purpose is to determine the
incumbent’s chances of winning the elections. Let us assume that the policymaker’s effort
devoted to implementing the supervisory regime is observable by the financial constituency;
the financial professionals can be considered insider agents respect to the other citizens.
7
Therefore the campaign contributions can be contingent upon the policymaker effort: f(a2); for
simplicity
f = ka2. The GH policymaker chooses effort
taking into account the lobby goal
function. The utility function of the politician is:
R ( y 2 ) = βV = β [(1 + δ ) y 2 − f ] = β [(1 + δ )(a 2 + Ω) − ka 2 ]
R( y 2 ) − C (a 2 ) = β [(1 + δ )(a 2 + Ω) − ka 2 ] − c(a 2 )
R ( y 2 ) − C (a 2 ) > 0
β [(1 + δ )(a 2 + Ω)] > ( β k + c)(a 2 )
The interest of the politician in implementing a greater level of supervisory consolidation
will depend, ceteris paribus, on the central bank involvement in supervision, provided that the
central bank is a captured agency. Under these conditions a central bank unification effect is
likely to occur. Otherwise a central bank fragmentation effect holds.
Finally we can describe the politician when he faces the trade off between acting as a HH
type rather than a GH type. In designing the supervisory regime, there are two possible
policies, that is two different directions – effectiveness and capture – to which effort can be
devoted, with different constituencies.
The first policy – to pursue supervision effectiveness, labelled y1 = Ω + a1 – benefits the
voters as a whole, while the second policy direction – supervision capture, labelled y2 = Ω +
a2 – only benefits the financial constituency. The ability of the policymaker is assumed to be
the same for implementing effectiveness and/or
capture. The trade off
in building up the
supervisory architecture holds if : c = C (a1 + a2); for simplicity C (a1 + a 2 ) = (a1 + a 2 )
2
and
a 2 = 1 − a1 .
Citizens influence policy only through elections. The financial constituency can influence
policy through the campaign contributions.
his effect can be analysed
by saying that the
voters’ reservation utility is a decreasing function of the campaign contributions collected by
the incumbent:
W = a e + Ω AV − H ( f )
Where the function H (f) – we can label it the “institutional function” - represents the
effects of the campaign contributions, with H ( f ) =
simplicity H ( f ) =
f
1
, H '( f ) =
> 0 , H ' ' ( f ) < 0 ; for
1+ f
(1 + f )2
ka 2
f
.
=
1 + f 1 + ka 2
8
The features of the institutional function can represent all the structural factors that
increase or decrease the possibility of the financial constituency to actually influence the
policymaker’s behaviour. On this regard it is natural to suppose that the quality of the political
institutions matters. In countries with good performances in
public sector governance it is
more likely that an HH policymaker is in charge in implementing the supervisory design, i.e.
the captured policymaker is less likely to occur.
The sequencing of the events is as follows. First the society assigns the task of designing
the supervisory regime. Then the financial constituency commits to campaign contributions as
a function of effort. Next the policymaker allocates effort between effectiveness and capture,
and determines the level of supervision unification, taking into account the influence of the role
of the central bank. Nature chooses a realisation of Ω. Finally, rewards are paid.
Here we have a common agency game, with two types of principals: the representative
voter and the financial constituency. The financial constituency has all the commitment power
and can influence the policymaker through campaign contributions, if this instrument is
sufficiently effective in swaying the voters. In equilibrium, the allocation of effort must be
jointly optimal for the policymaker and the financial constituency, given voters’ expectations.
The policymaker’s preferences become:
R( y1 , y 2 ) − C (a1 + a 2 )
[
where:
]
R( y1 , y2 ) = β Pr Ω ≥ Ω M + a e − a1 − H ( f ) + (1 + δ )(a2 + Ω) − f
⎧
⎪β − H ( f ) + (1 + δ )(a 2 + Ω) − f
⎪
⎪
R( y1 , y 2 ) = ⎨β (1 − p ) − H ( f ) + (1 + δ )(a 2 + Ω) − f
⎪
⎪
⎪− H ( f ) + (1 + δ )(a 2 + Ω) − f
⎩
ΩH − ΩL
2
Ω − ΩL
if a1 ≥ a e − H
2
Ω − ΩL
if a1 < a e − H
2
if
a1 ≥ a e +
and
⎧
Ω − ΩL
f
if a1 ≥ a e + H
⎪β − 1 + f + (1 + δ )(a 2 + Ω) − f − C (a1 + a 2 )
2
⎪
⎪
Ω − ΩL
f
+ (1 + δ )(a 2 + Ω) − f − C (a1 + a 2 ) if a1 ≥ a e − H
R( y1 , y 2 ) − C (a1 + a 2 ) = ⎨β (1 − p ) −
1+ f
2
⎪
⎪
Ω − ΩL
f
+ (1 + δ )(a 2 + Ω) − f − C (a1 + a 2 )
if a1 < a e − H
⎪−
2
⎩ 1+ f
Not surprisingly the utility of the policymaker in pursuing the supervisory effectiveness will
depend on the parameters that characterise the preferences of the financial constituency. The
likelihood of a HH type is increasing when the incentives for supervisory capturing are
9
decreasing. Considering the more general case – i.e. the ability of the politician is irrelevant –
we have that:
U PM = R ( y1 , y 2 ) − C (a1 + a 2 ) = β −
ka 2
2
+ (1 + δ )(a 2 + Ω) − ka 2 − (a1 + a 2 )
1 + ka 2
∂U PM
k
=
− (1 + δ ) + k
∂a1
[1 + k (1 − a1 )]2
∂U PM
k
=
+ k > (1 + δ )
∂a1
[1 + k (1 − a1 )]2
∂U PM
(1 + δ ) with a = 0
=k>
1
2
∂a1
The properties of the equilibrium – and then the relationship between the supervision
unification and the central bank involvement – will depend on how effective the lobby is in
influencing the voters.
Let us summarise the main findings. We model the design of the financial supervision
unification as a delegation problem, where the policymaker follows a sequential process: given
the institutional position of the central bank, he chooses the supervisory design. If we take into
account the existence of two constituencies with different preferences – the citizens and the
financial industry – we can analyses the behaviour of two alternative policymakers: the HH
type versus the GH type.
If the policymaker acts as an HH type the central bank involvement in supervision can be
viewed an obstacle in the supervision consolidation if at least one of the three effects – moral
hazard effect, conflict of interest effect and bureaucracy effect – holds. The central bank
fragmentation effect is likely to occur.
If the policymaker chooses to please the financial community acting as a GH type, the
central bank fragmentation effect is less likely to occur, provided that the financial community
likes a more consolidated supervision, and the central bank involvement is a proxy of the
financial constituency power.
If and only if these assumptions hold we can disentangle the effect of different types of
policymakers over the relationship between the financial supervision unification and the central
bank involvement. Otherwise a signal extraction problem occurs. For example, other things
being equal, if the central bank is not a captured one and the policymaker acts as a GH type,
the central bank fragmentation effect is likely to occur again.
10
3. Measuring the Unification of Financial Supervision and the Central Bank
Involvement
Our model predicts the possibility of different degrees of unification in the design of the
supervisory governance structure, depending on the type of the policymaker involved, and on
the features of the parameters of the model (i.e. the structural environment in which the
policymaker takes his decisions).
In the real world, the type of the policymaker – as well as all the structural and
institutional channels which influence his behaviour – is a hidden variable. At each point in
time, we can only observe the politicians’ decision to maintain or reform the
supervisory
governance structure, in particular its level of unification. Therefore the next step is to
measure the degree of unification in the actual supervisory regimes, as well as the central
bank involvement, which represents our key explanatory variable.
How can the degree of unification of financial supervision be measured? This is where the
financial supervision unification index (FSU Index) comes in (description in Table 2). This index
was created through an analysis of which and how many authorities in the 91 countries
examined are empowered to supervise the three traditional sectors of financial activity:
banking, securities markets and insurance2. The country sample depends on the availability of
institutional data3.
To transform the qualitative information into quantitative indicators, we assigned a
numerical value to each type of regime, in order to highlight the number of the agencies
involved. The rationale by which we assigned the values simply considers the concept of
unification of supervisory powers: the greater the unification, the higher the index value.
There are five qualitative characteristics of supervisory regimes that we decided not to
consider in constructing this index. Firstly, we did not consider the legal nature – public or
private – of the supervisory agencies nor their relationship to the political system (degree of
independence, level of accountability). Secondly, at least in each industrial country, there is an
authority to protect competition and the market, with duties that impinge on the financial
sectors. But, since it is a factor common to all the structures, we decided not to consider the
antitrust powers account in constructing the index. Moreover we did not consider who is
involved in the management of the deposit insurance schemes. In general, we consider only
the three traditional sectors (banking, securities and insurance markets) that have been the
subject of supervision. Finally, the financial authorities may perform different functions in the
regulatory as well as in the supervisory area. However, at this first stage of the institutional
analysis, we prefer to consider only the number of the agencies involved in the supervisory
activities.
2
Sources: for all countries, official documents and websites of the central banks and the other financial authorities. The
information is updated to 2006. See Table 2.
3
In the empirical analysis we do not include the very small countries and territories (Bahrain, Bermuda, Cayman
Islands, Gibraltar, Hong Kong Maldives, Netherlands Antilles, Singapore and United Arab Emirates) with a single
financial authority so as to avoid an evident bias in the empirical analysis.
11
Figure 2 shows the distribution of the FSU Index. On the one hand there are countries
(40) with a low consolidation of supervision (the Index is equal to 0 or 1). On the other, there
are countries (15) that established a unified supervisor, with a high level of supervision
consolidation.
Now we will consider what role the central bank plays in the various national supervisory
regimes. We use the index of the central bank's involvement in financial supervision: the
Central Bank as Financial Authority Index (CBFA) (description in Table 2). Figure 3 shows the
distribution of the CBFA Index. In the majority of countries in our sample (39) the central
bank is the main bank supervisor (the Index is equal to 2), while in very few countries (2)
the central bank is monopolistic in the overall financial supervision (the Index is equal to 4).
It is interesting to note that in general the present degree of the central bank involvement
has been established in the previous years, and then confirmed in subsequent reforms. For
each country we compare the year in which the present degree of central bank involvement in
supervision was established (i.e. definition of the CBFA Index, blue line), with the year of the
most recent reform of the supervisory architecture (i.e. definition of the FSU Index, red line)
(Figure 4). Given the data of 88 national reforms of the supervisory architecture, the central
bank involvement was confirmed in 67 cases (76%), decreased in 16 cases (18%), increased
in 5 cases (6%).
Finally, considering both indexes for the countries in our sample (Figure 5), the analysis
shows that the two most frequent regimes are polarised: on the one hand, Unified Supervisor
regime (13 cases, red ball); on the other, Central Dominated Multiple Supervisors regime (27
cases, white ball). The Figure seems to depict a trade off between supervision unification and
central bank involvement, with two outliers (green ball).
4. Empirical Results
We can now empirically investigate
the
robustness of the relationship between the
degree of financial supervision unification and central bank involvement. Supervisory regimes
can be viewed as resulting from an unobserved variable: the policymaker’s optimal degree of
financial supervision unification. Each regime corresponds to a specific range of the optimal
financial supervision unification, with higher discrete FSU Index values corresponding to a
higher range of financial unification values.
Our qualitative dependent variable can be classified into more than two categories,
given that the FSU Index is a multinomial variable. But the FSU Index is also an ordinal
variable, given that it reflects a ranking. Then the ordered model is an appropriate estimator,
given the ordered nature of the policymaker alternative.
12
Following our theoretical setup, each policymaker – either the HH type or the GH type maximises his objective function and determines his optimal level of supervision unification,
given the features of the structural variables4.
The sample is made up of 88 countries belonging to all continents. Most of the data have
been collected using World Bank Surveys and official public documents (see Table 3). All
descriptive statistics are reported in Table 3A.
An ordered Logit is used to pursue our objectives of both understanding which features
determine the policymaker choice of a given supervisory setting and of estimating the role of
the central bank in determining his choice. In our estimation we start using the central bank’s
involvement as a proxy of its influence. Our results are reported in Tables 4 and 5.
The first result that emerges is a significant inverse relationship between the supervision
unification and central bank involvement. The central bank fragmentation effect – instead of
the central bank unification effect – holds.
The second result is related to the role of the political governance. Our results highlight a
relationship between the level of supervision unification and good governance performances.
Finally we include population in all regressions as an exogenous scale factor. The estimated
parameter is significant, highlighting a small country effect: whatever the policymaker type,
with relatively few people the expertise in financial supervision is likely to be in short supply,
and concentration is more likely to occur.
4.1 Robustness
The aim of the robustness tests was twofold. First of all, we tested the robustness of our
predictions using several control variables (Table 4 and 5); secondly we tested different
definitions of the dependent variable (Table 6); finally we used alternative central bank
indicators (Table 7).
As a control factor, wealth features of each country are captured by per capita GDP, which
are insignificant (Table 4).
We also included controls for the standard proxies of financial
structures. It has been stressed that the changes in the supervisory architecture are taking
place against the backdrop of fundamental changes in the financial markets. Many of the
subsequent studies devoted to the analysis of the recent evolution in supervision design
stressed the importance of the characteristics of the financial markets – as the classic
distinction between bank-based or market-based financial systems – as determinants in the
choice for a total or partial unification of the supervisory regimes (Masciandaro 2005 and
Masciandaro and Quintyn 2007).
The traditional market-based versus bank-based index
The latent variable - the policymaker’s optimal degree of financial supervision unification y* - is determined by:
y*=β’ x + ε
where ε is a random disturbance uncorrelated with the regressors, and β is a 1 x K regressors’ vector.
The policymaker’s choice is summarized in the value of the FSU Index, which represents the threshold value. For our dependent
variable there are seven threshold values. The impact of a change in an explanatory variable on the estimated probabilities of the
highest and lowest of the order classifications—in our case the Single Authority model and the “pure” Multi-supervisory model—is
unequivocal: If βj is positive, for example, an increase in the value of xj increases the probability of having the Single Authority model,
while it decreases the probability of having the “pure” Multi-supervisory model.
4
13
shows no relationship with the choice of the supervisory model. The same result holds true
using - as alternative proxies
of the financial structure - the development of the financial
markets, measured by the level of market capitalization, and the size of the banking system,
measured by the asset dimension.
We also controlled for geographical variables (Table 5). We use a dummy for Europe, in
order to test the possible presence of some sort of “mimic effect” among neighbouring
countries. Geographical dummies are also required to capture the existence of international
agreements between countries – such as the European Union, or the OECD membership –
which may influence their national behaviour. Finally we asked ourselves if geography matters
in testing the endowment view (Beck, Demirguc – Kunt and Levine 2001): historically, lands
with high rates of disease and inhospitable conditions are not able to support the overall
institutional development, which promotes the quality of the financial structures. The
geographical testing did not produce any results.
Furthermore we used a temporal variable. Let us remember that the number of countries
that are reviewing their supervisory structures and the governance arrangements is increasing
year after year. So a legitimate question about whether there was a kind of fashion effect (or
bandwagon effect) at work arose, i.e. if recent reformers were inspired by the type of changes
in governance arrangements introduced by earlier reformers, producing a trend towards
unification.
For each country we used the year of last supervisory reform as proxy of the
fashion effect; the result was not significant.
We focused on the possible effect of the legal origins. The law and finance literature (La
Porta, de Silanes, Shleifer, and Vishny 1997) claimed that legal origin is associated with a
variety of aspects of policymakers’ performance. Countries of English (market friendly) legal
origin perform better than countries with legal systems rooted in French, German,
Scandinavian or other (socialist, Muslim) traditions. According to these sources, market
friendly law traditions offer better protection against bad public policy (as in expropriation by
the State) and greater incentive in providing public goods (as in good regulation). In countries
with market friendly law origins, an HH policymaker is more likely to occur. Our results were
puzzling: the level of supervision unification is linked with the Civil Law root, in particular with
German and Scandinavian law jurisdictions. The testing of the legal origins effect produced
mixed results in the empirical works5, so we decided to investigate this issue further in the
future.
Finally, we wonder whether the occurrence of a financial crises has an impact on the
supervisory institutional setup. We suggest that the occurrence of a crisis may affect the
features of the regime. We can imagine that both types of policymaker can be affected by such
crisis, although the impact of a crisis experience on the level of supervision consolidation is far
to be clear. For each country we used as proxy of the crisis effect a dummy variable reflecting
5
In Masciandaro 2006 the legal origin effect disappeared when different classifications of legal origins are used while
in Masciandaro 2007, with a different sample size, the legal origin effect holds using the alternative classifications.
14
whether a banking crisis has occurred up to five years before the year of the reform which
established the actual supervisory regime6. The result was not significant.
Taking into consideration the dependent variable, we asked ourselves if the results hold
with different specifications (Table 6). First of all, we eliminated the weights attributed to the
banking and financial supervisory responsibility in terms of the insurance sector supervision7.
The central bank fragmentation effect was much stronger, the governance effect holds, while
the small country effect and the law effect were not significant. Secondly we tested a more
radical hypothesis: each policymaker simply decides between the two extreme regimes of
supervision,
single
authority
(complete
centralisation)
versus
multi-authority
(decentralisation). The dependent variable became a dummy, to be estimated using a simple
Logit. Again the central bank fragmentation effect and the governance effect were the only
significant ones. Finally we classified the supervisory regimes trying to disentangle the type of
models: unified model (single supervisor), hybrid model (at least one authority monitor for
more than one sector, as in the case of the twin – or more – peaks approach), specialised
model (separate authorities for each sector, at least one per sector)8. The central bank effect,
the governance effect and the law effect were significant.
Finally we considered the issue of potential endogeneity of the central bank role. First of all
we assumed that the central bank involvement
was a consequence of the policymakers’
decision on the degree of supervision consolidation (reverse causality), and we explored the
theoretical rationale of it. However we rejected this hypothesis, given that different
specifications of the selected model failed to explain the central bank involvement in
supervision (Table 7). Secondly we tested different indicators of the central bank role, which
should have suffered less from endogeneity risks. We tested the conclusion that policymakers
dislike the supervision unification in the hands of the central bank in order to avoid the
creation of a powerful agency. The HH policymaker could fear that a monopolist central bank is
a potential source of bureaucratic inefficiency, while the GH policymaker could see an
independent (no captured) central bank as an obstacle to its discretion, given that he needs a
high degree of freedom to please its constituency9. Therefore we used two different proxies of
the central bank power: the central bank age, assuming that an old central bank is more
influential; the central bank independence. The central bank age was calculated in three
6
In order to take into account the time lag between a financial crisis and the implementation of a supervisory reform
and without any systematic statistics (using a group of 15 countries that have adopted a unified supervisor De Luna
Martinez and Rose 2003 claimed that the timeframe required to complete a reform has taken on average between
one and two years) we use a rule of thumb of five years.
7
We used an index (SFU Two) according to the following scale: 5 = Single authority for all three sectors (total number
of supervisors = 1); 3 = Single authority for two sectors (total number of supervisors = 2); 1 = Specialized authorities
for each sector (total number of supervisors = at least 3). The correlation between SFU and SFU is 0.93.
8
We used an index (SFU Three) according to the following scale: 5 = Unified model; 3 = Hybrid model 1 =
Specialized model. The correlation between SFU and SFU Three is 0.88, while the correlation between SFU Two and
SFU Three is 0.91.
9
It is evident from a theoretical point of view that highlighting the overall role of the central bank position is different
but not alternative with respect to focusing on the central bank involvement. Therefore we cannot consider the Central
Bank Age (CBA) Indexes and the CBI Index as instrumental variables respect to the CBFA Index. Empirically we
confirmed that the variables are low correlated (the correlation between the CBFA Index and the Central Bank Age
(CBA) Indexes are respectively – 0.16,-0.13, -0.02, while between the CBFA Index and the CBI Index is – 0.18).
15
different ways. The absolute effects of the four new indicators - as well as their composite
effects with the FSU Index - were not significant at all (Table 8).
4.2 Predictions
The results we reached were consistent with the topics discussed in the theoretical section
which stressed a possible role of the central bank position in influencing the policymakers’
choices. The central bank role matters.
Moreover, given that the type of the policymaker is unknown, the story goes as follows.
Each policymaker, in determining the level of unified supervision, could be influenced by the
involvement of the central bank, but under different conditions.
If the policymaker is of an HH type, it should care about the effectiveness of the
supervision, in order to please the citizens. If the policymaker sees the supervision
consolidation as a welfare improvement, then the central bank involvement could be viewed as
an obstacle, but only if at least one of the three effects described above – i.e. moral hazard,
conflict of interest and bureaucracy –holds.
If the policymaker is of a GH type, it wishes to please the financial constituency. In this
case the central bank fragmentation effect holds if, and only if, the financial constituency
dislikes the unified supervision and the central bank is a captured agency.
The level of supervision consolidation depends also on the overall quality of the political
process: in countries characterised by a good public sector governance, an HH type of
policymaker is more likely to occur, which in turn will promote a supervision unification if
welfare is improving.
5. Conclusions
The evolution of financial supervision architecture depends on the political cost – benefit
analysis. This paper discusses under which conditions the politicians prefer to implement a
unified sector supervision outside the central bank.
What are the policy implications of our analysis? In general terms the quality of public
sector governance plays a crucial role in determining the supervision unification. Focusing on
the behaviour of the “good” policymaker (HH type), it will prefer a unified financial authority
that is different from the central bank if the correspondent welfare gains – linked to at least
one of the three effects: moral hazard, conflict of interest and bureaucracy – are considered
high. The “bad” policymaker (GH type) will choose the single financial authority if the financial
industry likes it, and the central bank is not a captured one.
At national level these rules of thumb can explain both the emerging trend towards the
single financial authorities, and the (so far few) cases of supervisory consolidation in favour of
the central banks. Let us consider for example the three European cases in which the reforms
of the supervision architecture increased – Ireland and the Netherlands – or could increase –
Italy – the power of the national central bank. Ireland, the Netherlands and Italy are members
16
of the Economic and Monetary Union, and their central banks no longer have the full
responsibility for national monetary policies. Therefore we can interpret these reforms as cases
in which HH policymakers can increase the central bank role in supervision, given that the risks
of moral hazard, conflict of interest risks, bureaucratic overpower are likely to be low, while
the reputation gains are likely to be high. Alternatively, we can study each of these reforms as
possible cases of
GH policymakers that implement the supervision regime favoured by the
financial industry. In general, disentangling the two interpretations in specific country cases
can be a welcome extension of the research agenda.
At European level our model predicts that the establishment of a single financial authority is
less likely to occur with the presence of a European central bank deeply involved in
supervision. Conversely, the less the European Central Bank is involved in the financial
supervision architecture, the more likely the establishment of a European Single Financial
Authority will be.
17
6. References
Abrams, R.K., Taylor, M.W., 2002. Assessing the Case for Unified Sector Supervision. FMG
Special Papers n.134, Financial Markets Group, LSE, London.
Alesina, A., Tabellini, G., 2003. Bureaucrats or Politicians? Part II: Multiple Policy Task.
Discussion Paper n.2009, Harvard Institute of Economic Research, Harvard University, MA.
Arnone, M., Gambini, A., 2007. Architecture of Supervisory Authorities and Banking
Supervision. In: Masciandaro D. and Quintyn M. (Eds.), Designing Financial Supervision
Institutions: Independence, Accountability and Governance, Edward Elgar, Cheltenham, 262308.
Barth, J.R., Nolle, D.E., Phumiwasana, T. and Yago, G. 2002. A Cross Country Analysis of
the Bank Supervisory Framework and Bank Performance. Financial Markets, Institutions &
Instruments, vol. 12, n.2, 67-120.
Beck, T., Demirguc-Kunt, A., and Levine R., 2001. Law, Politics and Finance, World Bank
Policy Research, Working Paper n. 2585, Washington D.C.
Bini Smagh,i L., 2007. Independence and Accountability in Supervision. In: Masciandaro D.
and Quintyn M. (Eds.), Designing Financial Supervision Institutions: Independence,
Accountability and Governance, Edward Elgar, Cheltenham, (forth.).
Cihak, M., Podpiera, R., 2007a. Experience with Integrated Supervisors: Governance and
Quality of Supervision. In: Masciandaro D. and Quintyn M. (Eds.), Designing Financial
Supervision Institutions: Independence, Accountability and Governance, Edward Elgar,
Cheltenham, 309-341.
Cihàk M. and Podpiera R., 2007b. Does More Integrated Supervision mean Better
Supervision? Finlawmetrics 2007, Bocconi University, mimeo.
De Luna Martinez, J., Rose, T.A., 2003. International Survey of Integrated Financial Sector
Supervision. Policy Research Working Paper Series, n.3096, World Bank, Washington D.C.
Garcia Herrero, A., and Del Rio P., 2003. Implication of the Design of Monetary Policy for
Financial Stability. 24th SUERF Colloquium, Tallin. Société Universitaire Européenne de
Recherches Financières, Wien.
Goodhart, C.A.E, Shoenmaker, D. 1995. Should the Functions of Monetary Policy and
Banking Supervision be Separated? Oxford Economic Papers, vol.47, n.4, 539-560.
Goodhart, C.A.E., 2004. Financial Supervision from an Historical Perspective: Was the
Development of Such Supervision Designed, or largely Accidental?, Conference on the
Structure of Financial Regulation, Bank of Finland, mimeo.
Goodhart, C.A.E., 2007. Introduction. In: Masciandaro D. and Quintyn M. (Eds.), Designing
Financial Supervision Institutions: Independence, Accountability and Governance, Edward
Elgar, Cheltenham, 12-26.
Honohan, P., Laeven, L., (Eds.), 2005. Systemic Financial Crises: Containment and
Resolution, Cambridge University Press, Cambridge UK.
La Porta, R., Lopez de Silanes, F., Shleifer, A., Vishny, R.W., 1997. Legal Determinants of
External Finance. Journal of Finance, vol. 52, n.3, 1131-1150.
18
Llewellyn, D.T., 2005. Institutional Structure of Financial Regulation and Supervision: the
Basic Issues. In: Fleming A., Llewellyn D.T. and Carmichael J. (Eds.), Aligning Financial
Supervision Structures with Country Needs, World Bank Publications, Washington D.C, 19-85.
Masciandaro, D., 2004. Unification in Financial Sector Supervision: the Trade off between
Central Bank and Single Authority. Journal of Financial Regulation and Compliance, vol. 12,
n.2, 151-169.
Masciandaro, D., (Ed.). 2005. Handbook of Central Banking and Financial Supervision in
Europe, Edward Elgar, Cheltenham.
Masciandaro, D., 2006. E Pluribus Unum? Authorities Design in Financial Supervision:
Trends and Determinants. Open Economies Review, vol. 17, n.1, 73-102.
Masciandaro, D., 2007. Divide et Impera: Financial Supervision Unification and
Central Bank Fragmentation Effect, European Journal of Political Economy, 285-315.
the
Masciandaro, D., Quintyn, M., (Eds.), 2007. Designing Financial Supervision Institutions:
Independence, Accountability and Governance, Edward Elgar, Cheltenham.
Mwenda, K.K., Fleming, A.,2001. International Development in the Organizational Structure
of Financial Services Supervision, Conference on the Challenges of Unified Financial
Supervision, Central Bank of Estonia, Tallinn.
Mwenda, K.K., 2006. Legal Aspects of Financial Services Regulation and the Concept of a
Unified Regulator, The International Bank for Reconstruction and Development/The World
Bank, Washington.
Pigou, A., 1938. The Economics of Welfare, London, Macmillan & Co.
Shleifer, A., Vishny, R., 1998. The Grabbing Hand, Cambridge, Harvard University Press.
Skogstad Aamo, B., 2005. Norway. In Masciandaro D. (Ed.), The Handbook of Central
Banking and Financial Authorities in Europe, Edward Elgar, Cheltenham, 200-225.
Taylor, M. W., Fleming, A., 1999. Integrated Financial Supervision: Lessons from Northern
European experience. Policy Research Working Paper n.2223, The World Bank.
Westrup J., 2007. Independence and Accountability: Why Politics Matters. In: Masciandaro
D. and Quintyn M. (Eds.), Designing Financial Supervision Institutions: Independence,
Accountability and Governance, Edward Elgar, Cheltenham, 117-150.
19
7. Tables and Figures
Figure 1 - Financial Supervisory Regimes: number of reforms per year (1986-2006)
16
14
12
10
8
6
4
2
0
1986
1987
1989 1990
1991 1992
1993
1994
1995
1996 1997
1998
1999
2000
2001
2002
2003 2004
2005
2006
20
Figure 2 - Financial Supervision Unification Index
40
35
30
number of countries
25
20
Serie1
15
10
5
0
FSU0
FSU1
FSU2
FSU3
FSU4
FSU5
FSU6
FSU7
financial supervision unification index
21
Figure 3 - Central Bank as Financial Supervisor Index
45
40
35
30 of countries
number
25
Serie1
20
15
10
5
0
CBFA1
CBFA2
CBFA3
CBFA4
central bank involvement in supervision
22
Figure 4 - FSU and CBFA Indexes: Time Lags
2050
2000
1950
1900
1850
years
Serie1
Serie2
1800
1750
1700
1650
1600
1
4
7
10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88
countries
23
Figure 5 - FSU and CBFA Indexes: The Trade Off
9
8
7
13
2
6
financial supervision unification
5
4
Serie1
Log. (Serie1)
3
2
R =- 0.33
2
1
27
0
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
-1
-2
central bank involvement
24
Table 1 Politicians and Supervisory Reforms: When the Central Bank Role Matters
EFFECTS
BUREAUCRACY
EFFECT
REPUTATION
EFFECT
EPISODES
In the United Kingdom case, Goodhart 2004 and Westrup 2006
stressed that, among all the arguments that led the Government in
1997 to establish the Financial Services Authority (FSA), removing
supervision from the Bank of England could have been a quid pro
quo for giving it monetary independence, on the grounds that a
central bank with too many functions could be too much of a power
centre within the democratic system. In Norway, due to the
banking crisis in the early 1990s, the possibility of merging the
BISC with the central bank was considered by a committee
appointed by the Ministry of Finance. But the Parliament, in order to
avoid an excessive concentration of power, ruled that the BISC
continue as a separated and independent agency (Skogstad Aamo
2005).
The difficulties in implementing reforms that reduce the central
bank involvement in supervision when is reputation is high is
documented in several case studies.
In France a reform was recently implemented, merging into one
regulatory authority – Autorité des Marchés Financiers (AMF) –
different financial supervision responsibilities, but the Banque de
France prerogatives remained unchanged. In 2004, after the
Parmalat scandal, the Italian Government proposed a draft text of a
bill, concerning a general reform of the supervisory architectures,
based on the establishment of a single financial authority.The
proposed reform encountered strong opposition from a bi-partisan
coalition, defending the role of the Bank of Italy in promoting
financial stability. The reform was rejected. Finland has opted not
to adopt the unified approach in financial supervision, in contrast
with the other Scandinavian countries. Taylor and Fleming 1999
claimed that the Bank of Finland involvement in supervision has to
be considered in explaining this choice. In Iceland, prior to the
establishment of the single financial agency, banking supervision
was conducted by the central bank. In 1996, a committee was set
up by the Minister of Commerce, to look at prospects of moving
toward unified supervision. Mwenda and Fleming 2001 reported
that only one member on the committee – the central bank official
– voted against the introduction of unified financial supervision.
However, the central bank obtained the ability to appoint one of the
three members of the single financial authority board. Also in
Germany – as Westrup 2007 reported – the Bundesbank did a
public campaign to lobby again the creation of BaFin
On the contrary, if the reputation of the central bank is low, or
decreasing, the establishment of a single financial authority could
be more likely to occur, despite its involvement in supervision.
The supervisory failure of the UK central bank is well documented
in Westrup 2006. The link between banking instability, central bank
reputation failure and single financial authority establishment is
also evident in the Baltic unified supervisory architectures and in
the case of Korea. Estonia experienced a severe banking crisis in
1998 and 1999. In May 2001, the Estonian Parliament adopted the
Financial Supervisory Authority. Before the Act, the supervision was
split into the three traditional sets of institutions. The Bank of
Estonia was responsible for state supervision of banking (Liive
2005). Latvia experienced banking and financial crises in 1995 and
in 1998. In July 2001, the Financial and Capital Market Commission
was established, as a consolidated institution. In Korea, until 1997,
the central bank was responsible for banking supervision; however
– as Lee noted – the Ministry of Finance dominated the central
bank. Following the 1997 financial crisis, a presidential committee
recommended a drastic overhaul of the organization of the central
bank and the country’s supervisory structure. As a result, the
former four financial supervisory authorities were combined into
one integrated financial supervisory body, the Financial Supervisory
Committee. Also China – Quintyn et al 2006 – moved supervision
out of central bank in the wake of period o financial sector distress.
It is interesting to note that the reputation failure effect can hold
regardless the nature of the agency involved. In Norway – as we
noted above – after the 1990s banking crisis the Ministry of Finance
considered the possibility to merge the single financial authority
with the central bank.
25
Table 2 Supervisory Authorities in 91 countries (year 2006):
FSU Index and CBFA Index
Countries
FSU
CBFA
Countries
FSU
CBFA
Albania
Argentina
Australia
1
1
6
2
2
1
Korea
Latvia
Lebanon
7
7
2
1
1
3
Austria
6
1
Libya
3
2
Bahamas
1
2
Lithuania
1
2
Belarus
1
2
Luxembourg
5
1
Belgium
7
1
Macedonia
1
2
Bolivia
3
1
Malaysia
3
3
Bosnia
0
2
Malta
7
1
Botswana
1
1
Mauritius
3
2
Brazil
2
3
Mexico
5
1
Bulgaria
1
2
Moldova
1
2
Cameroon
1
1
Morocco
2
2
Canada
3
1
Netherlands
2
3
1
2
Chile
3
1
New Zealand
China
1
1
Nicaragua
Colombia
3
1
Norway
7
1
Costa Rica
1
1
Pakistan
2
3
Croatia
1
2
Panama
Cyprus
1
2
Peru
3
1
7
1
1
1
Czech Rep.
1
2
Philippines
2
3
Denmark
7
1
Poland
1
1
Ecuador
3
1
Portugal
2
3
Egypt
1
2
Romania
1
2
2
El Salvador
Russia
1
Estonia
7
1
Saudi Arabia
7
_
Finland
5
1
Singapore
7
_
France
1
3
Slovak Rep.
2
2
Georgia
1
2
Slovenia
1
2
Germany
6
1
South Africa
3
2
Greece
1
2
Spain
1
3
Guatemala
Hong Kong
3
3
1
1
1
_
Sri Lanka
1
2
Sweden
7
1
Hungary
7
1
Switzerland
5
1
Iceland
7
1
Thailand
1
2
India
0
2
Tri. & Tobago
1
2
Iran
5
3
Tunisia
1
2
Ireland
7
4
Turkey
1
1
Israel
2
2
Ukraine
1
2
Italy
2
3
UAE
1
2
Jamaica
3
2
UK
7
1
Japan
6
1
USA
0
2
Jordan
1
2
Uruguay
7
4
Venezuela
1
2
Vietnam
1
2
Zimbabwe
1
2
Kazakhstan
Kenya
3
0
3
2
* FSU INDEX
The index was built on the following scale: 7 = Single authority for all three sectors (total number of supervisors=1); 5
= Single authority for the banking sector and securities markets (total number of supervisors=2); 3 = Single authority
for the insurance sector and the securities markets, or for the insurance sector and the banking sector (total number
of supervisors=2); 1 = Specialized authority for each sector (total number of supervisors=3).
We assigned a value of 5 to the single supervisor for the banking sector and securities markets because of the
predominant importance of banking intermediation and securities markets over insurance in every national financial
industry. It also interesting to note that, in the group of integrated supervisory agency countries, there seems to be a
26
higher degree of integration between banking and securities supervision than between banking and insurance
supervision1; therefore, the degree of concentration of powers, ceteris paribus, is greater. These observations do not,
however, weigh another qualitative characteristic: There are countries in which one sector is supervised by more than
one authority. It is likely that the degree of concentration rises when there are two authorities in a given sector, one of
which has other powers in a second sector. On the other hand, the degree of concentration falls when there are two
authorities in a given sector, neither of which has other powers in a second sector. It would therefore seem advisable
to include these aspects in evaluating the various national supervisory structures by modifying the index as follows:
adding 1 if there is at least one sector in the country with two authorities, and one of these authorities is also
responsible for at least one other sector; subtracting 1 if there is at least one sector in the country with two authorities
assigned to supervision, but neither of these authorities has responsibility for another sector; 0 elsewhere.
** CBFA INDEX
For each country, and given the three traditional financial sectors (banking, securities and insurance), the CBFA index
is equal to: 1 if the central bank is not assigned the main responsibility for banking supervision; 2 if the central bank
has the main (or sole) responsibility for banking supervision; 3 if the central bank has responsibility in any two
sectors; 4 if the central bank has responsibility in all three sectors (Table 1). In evaluating the role of the central
bank in banking supervision, we considered the fact that, whatever the supervision regime, the monetary authority
has responsibility in pursuing macro financial stability. Therefore, we chose the relative role of the central bank as a
rule of thumb: we assigned a greater value (2 instead of 1) if the central bank is the sole or the main authority
responsible for banking supervision.
27
Table 3 – Data Sources
VARIABLE
Financial Supervision Unification
Index
DEFINITION
Section Three and Table2. See
Masciandaro (2007) for further
details.
Central Bank Involvement Index
Section Three and Table2. See
Masciandaro (2007) for further
details.
Good Political Governance
The index is built using all the
indicators proposed by
Kaufmann et al. (2003).
Average values 1996, 1998,
2000, 2002, 2004
Population
Millions of inhabitants. Average
values 1996, 1998, 2000, 2002,
2004
World Development Indicators, World
Bank.
Per Capita GDP
Per capita GDP. Average
values: 1996, 1998, 2000, 2002,
2004
World Development Indicators, World Bank
Financial Markets Capitalization
Measure of the securities
market size, relative to GDP.
Average values 1996, 1998,
2000, 2002, 2004
World Development Indicators, World Bank
Market vs banked systems
Market versus bank based
countries.
Dummy
variable
computed
using
different
banking and financial variables:
see Demigüç-Kunt and Levine
(1999) for further details.
Average values 1996, 1998,
2000, 2002, 2004
World Development Indicators , World
Bank, Bankscope
Banking Assets
Measure of the banking market
size, relative to GDP. Average
values 1996, 1998, 2000, 2002,
2004
Dummy that signals whether a
given country is European or
not
Dummy that signals whether a
given country is a member of
the European Union or not
Dummy that signals whether a
given country is a member of
the OECD or not
Proxy of the endowment view:
the variable is calculated as the
absolute value of the latitude of
the country, scaled to take
values between 0 and 1. See
Beck, Demirguc-Kunt and
Levine (2001) for details.
World Development Indicators, World
Bank, Bankscope
Europe
European Union
OECD
Latitude
Financial Reform
Year of the last reform of the
supervision architectures
Financial Crisis Impact
Dummy that signals whether a
crisis have occurred up to five
years before the last reform of
the supervisory architecture or
not.
Law
Legal origin of countries. See
Beck, Demirguc-Kunt and
Levine (2001) for details.
Level of the Central Bank
Independence, using the Grilli,
Masciandaro and Tabellini
Central Bank Independence Index
DATA SOURCES
"How countries supervise their banks,
insurances and securities markets 2006",
Central Banking Publications Ltd, UK.
Central Banks Websites. Richmond Law &
Tax, Ldt. 2005
"How countries supervise ,,, 2006", Central
Banking Publications Ltd, UK. Central
Banks Websites. Richmond Law & Tax,
Ldt. 2005
Kaufmann, D., Kraay, A., Mastruzzi, M.,
2003. Governance matters III: governance
indicators 1996-2002. Policy Research
Working Paper Series, n.3106, World Bank
World FactBook, CIA
World FactBook, CIA
World FactBook, CIA
Beck, Demirguc-Kunt and Levine (2001),
World FactBook, CIA
"How countries supervise … 2006",
Central Banking Publications Ltd, UK.
Central Banks Website.. Richmond Law &
Tax, Ldt. 2005
Data on financial crisis: Honohan and
Laeven (2005)
World FactBook, CIA
Arnone, Laurens, Segalotto and Sommer,
“Central Bank Autonomy: Lessons from
Global Trends, IMF Working Paper, n07/88
28
Index.
Central Bank Age
Difference between 2006 and
the Year of the CB
establishment.
"How countries supervise… 2005", Central
Banking Publications Ltd, UK. Central
Banks Websites. Richmond Law & Tax,
Ldt. 2005
Central Bank Age respect to the
other Supervisory Authorities
Difference between the Year of
the establishment of the older
supervisory authority different
from the CB and the Year of
establishment of the CB
Difference between the Year of
the country independence and
the Year of establishment of the
CB
.
"How countries supervise… 2005", Central
Banking Publications Ltd, UK. Central
Banks Websites. Richmond Law & Tax,
Ldt. 2005
Central Bank Age respect to the
Country
"How countries supervise… 2005", Central
Banking Publications Ltd, UK. Central
Banks Websites. Richmond Law & Tax,
Ldt. 2005, World FactBook, CIA
29
Table 3a – Descriptive Statistics
Variable
Obs
Mean
Std. Dev.
Min
Max
Financial
Supervision
88
2.886364
2.360751
0
7
88
1.761364
0.758012
1
4
Governance
88
0.342514
0.838665
-1. 20921
1.7875
Population
88
53.3858
171.1093
0.277
1250.632
Per Capita GDP
88
15.90142
53.24116
0.403958
496.8288
Capitalization
88
11.0437
11.16024
0
48.07136
Banking Assets
88
2.596851
4.237767
0.087399
35.26738
88
1986
17.75561
1931
2005
88
0.667159
0.190118
0.25
1
88
75.375
56.53832
10
313
80
51.4
59.68559
-51
292
88
-108.625
256.4405
-1314
169
Unification Index
Central
Bank
Involvement Index
Good
Financial
Political
Markets
Supervision
Reform: Year
Central
Bank
Independence
Index
Central Bank Age
Central Bank
Age
as Supervisor
Central Bank
Age
as Institution
30
Table 3b – Correlation Matrix
FSU
CBFA
Gov
Pop
P
MvB
FinM
Ban
Eur
OECD
Lat
Com
Civ
YR
GDP
FSU
CBFA
1
-0.330
1
0.441
-0.118
1
Population
-0.175
-0.034
-0.132
1
Per C. GDP
0.109
-0.139
0.0713
-0.047
1
MvB
0.095
0.0087
0.1846
0.1793
-0.027
1
FinMarkets
0.290
-0.048
0.5602
-0.000
0.037
0.3999
1
Banking
0.266
-0.090
0.4833
-0.047
0.0933
0.0428
0.5375
1
Europe
0.188
0.0021
0.413
-0.176
-0.001
-0.026
0.0919
0.26
OECD
0.392
-0.154
0.7335
-0.067
0.0956
0.3045
0.4049
0.4
0.43
1
Lat
0.319
-0.087
0.4972
-0.068
-0.087
0.0886
0.1928
0.22
0.76
0.54
1
Com Law
-0.071
0.1979
0.1502
0.1245
-0.035
0.2928
0.3362
0
-0.2
0
-0.2
1
Civil Law
0.262
-0.188
0.1327
-0.155
0.143
-0.118
-0.021
0.14
-0.1
0.17
-0.1
-0.51
YearReform
0.017
0.132
0.0941
0.1187
-0.134
0.026
-0.136
-0.0
0.28
0.04
0.32
0.11
Governance
1
1
-0.27
1
31
Table 4 - CHOICE OF SUPERVISORY UNIFICATION – Ordered Logit Estimation
DEPENDENT VARIABLE: FSU INDEX
CBAFA Index
(a)
(b)
(c)
(d)
(e)
-0.739**
-0.708**
-0.769**
-0.773***
-0.736**
(0.29)
(0.3)
(0.3)
(0.3)
(0.3)
0.827***
0.833***
0.781***
0.624**
0.794***
(0.25)
(0.25)
(0.26)
(0.3)
(0.28)
-0.00280**
-0.00276**
-0.00299**
-0.00291**
-0.00280**
(0.0013)
(0.0013)
(0.0013)
(0.0013)
(0.0013)
Good Political
Governance
Population
Per Capita GDP
0.00197
(0.003)
Markets vs
Banks
0.386
(0.47)
Financial Markets
Capitalization
0.0267
(0.021)
Banking Assets
0.0113
(0.046)
Nr. Of observations: 88;
Scale variable: population, gdp per capita
Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1 %
32
Table 5 - CHOICE OF SUPERVISORY UNIFICATION – Ordered Logit Estimation
DEPENDENT VARIABLE: FSU INDEX
(f)
CBAFA Index
(g)
(h)
(i)
(l)
-0.710**
-0.781***
-0.732**
-0.741**
-0.720**
(0.29)
(0.3)
(0.3)
(0.3)
(0.3)
(m)
-0.721**
(0.3)
(n)
(o)
-0.689**
-0.530*
(0.3)
(0.31)
Good Political
Governance
Population
Europe
0.920***
0.694**
0.661*
0.788***
0.840***
(0.29)
(0.31)
(0.36)
(0.3)
(0.26)
-0.00292**
-0.00277**
-0.00283**
-0.00280**
-0.00274**
(0.0013)
(0.0013)
(0.0013)
(0.0013)
(.0013)
0.822***
(0.25)
-0.00274**
(0.0013)
0.766***
0.527*
(0.26)
(0.28)
-0.00264**
-0.00283**
(0.0013)
(0.0014)
0.471
0.608
(0.64)
-0.65
-0.332
(0.46)
European
Union
0.408
(0.54)
OECD
0.392
(0.62)
Latitude
0.309
(1.25)
Supervision
Reform
-0.00381
(0.011)
Financial
Crisis
Impact
0.197
(0.48)
Law:
Common
Law: Civil
1.165**
(0.51)
Law: French
0.979*
(0.52)
Law: German
&
Scandinavian
2.624***
(0.9)
Nr. Of observations: 88;
Continental dummies: Europe, European Union, OECD, Latitude ; Scale variable: population, gdp per capita
Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1 %
33
Table 6 - CHOICE OF THE SUPERVISORY REGIMES – Alternative specifications
DEPENDENT
VARIABLE
FSU
FSU TWO
FSU BINARY
FSU THREE
CBFA Index
-0.689**
-1.322***
-1.501**
-0.513*
(0.3)
(0.38)
(0.62)
(0.29)
0.766***
0.718**
1.212***
1.037***
(0.26)
(0.28)
(0.45)
(0.29)
-0.00264**
-0.00297
-0.00162
-0.00145
(0.0013)
(0.0035)
(0.0047)
(0.0022)
0.471
0.675
-0.654
0.209
(0.64)
(0.73)
(1.18)
(0.71)
1.165**
0.843
0.265
1.512***
(0.51)
(0.59)
(0.83)
(0.57)
Good Political
Governance
Population
Law: Common
Law: Civil
Constant
0.117
(1.2)
Nr. Of observations: 88;
Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1 %
34
Table 7 - CHOICE OF CENTRAL BANK INVOLVEMENT IN SUPERVISION
Ordered Logit Estimation
DEPENDENT VARIABLE: CBFA INDEX
(a)
(b)
(c)
(d)
(e)
(f)
FSU Index
(g)
-
-0.515***
-0.498***
-0.540***
-0.549***
-0.516***
-0.511***
0.478***
(0.13)
(0.13)
(0.13)
(0.14)
(0.13)
(0.13)
(0.13)
0.0651
0.109
0.012
-0.133
0.0199
0.027
-0.0506
(0.29)
(0.32)
(0.3)
(0.34)
(0.3)
(0.3)
(0.31)
Good
Political
Governance
Population
Per
-0.00132
-0.00134
-0.00176
-0.00152
-0.00121
-0.00159
0.00172
(0.0012)
(0.0012)
(0.0013)
(0.0012)
(0.0012)
(0.0012)
(0.0013)
capita
GDP
-0.00837
(0.013)
Markets vs
Banks
0.633
(0.55)
Financial
Markets
Capitalization
0.0299
(0.024)
Europe
0.274
(0.47)
Supervision
Reform
0.0198
(0.014)
Law:
Common
0.831
(0.62)
Law: Civil
-0.165
(0.53)
Nr. Of observations: 88;
Continental dummies: Europe ; Scale variable: population, gdp per capita
Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1 %
35
Table 8 - CHOICE OF SUPERVISION UNIFICATION AND CENTRAL BANK ROLE
Ordered Logit Estimation
DEPENDENT VARIABLE: SFU INDEX
CBFA Index
-0.699**
-0.674**
-1.157***
-0.690**
(0.3)
(0.3)
(0.34)
(0.3)
Central Bank
Independence
-0.432
-0.0796
(1.41)
(1.39)
Central Bank
Age
0.00306
0.00431
(0.0044)
(0.0045)
Central Bank
Age, Respect
to
other
Supervisory
Institutions
0.00426
0.00402
(0.0044)
(0.0044)
Central Bank
Age, Respect
to the Country
-2.6E-05
0.000134
(0.00079)
(0.00085)
Good Political
Governance
Population
0.810***
0.848***
0.671**
0.711**
0.722**
0.807**
0.763***
0.852***
(0.3)
(0.3)
(0.29)
(0.29)
(0.33)
(0.33)
(0.27)
(0.27)
-
-
-
-
-
-
-
-
0.00267**
0.00239*
0.00270**
0.00248*
0.00277**
0.00233*
0.00264**
0.00236*
(0.0013)
(0.0014)
(0.0013)
(0.0014)
(0.0013)
(0.0014)
(0.0013)
(0.0014)
0.344
0.267
0.448
0.273
0.612
0.348
0.473
0.285
(0.77)
(0.75)
(0.64)
(0.63)
(0.67)
(0.65)
(0.64)
(0.63)
1.138**
1.198**
1.069**
1.080**
1.145**
1.188**
1.163**
1.211**
(0.52)
(0.52)
(0.53)
(0.55)
(0.55)
(0.51)
(0.51)
Law:
Common
Law: Civil
Nr. Of observations: 88;
Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1 %
36