The banking sector and recovery in the EU economy.
Barrell, Ray ; Fic, Tatiana ; Gerald, John Fitz 等
Banks within Europe have become larger and more international as
Europe has moved towards a unified financial services market, but this
trend has been reversed since the crisis. In order to establish the
effect of these structural changes on output in Europe, we use a micro
data set to investigate the impact of size (as measured by asset size)
on banks' net interest margins. We show that larger banks offer
lower borrowing costs for firms, which raises sustainable output. We
then use NiGEM to look at the impact of banks becoming smaller and
moving back into their home territory. We investigate the impacts on
output according to country size, showing that the effects are generally
larger in small countries, and also larger in economies that are more
dependent on bank finance for their business investment decisions.
Keywords: Net interest margins: bank size: European financial
integration; economic growth: bank regulation
JEL Classifications: E44; G10; G28
Introduction
The financial crisis of the past three years has seen a dramatic
change in the EU financial sector. Since the early 1990s, with the
completion of the internal market, there had been a growing move towards
an EU financial services market. Banks were becoming more international
with greater regional coverage within the EU (and the world). It was
anticipated that this integration of the European financial sector would
result in a more efficient use of capital in the EU economy, and
increased competition. In turn, this was expected to lead to a lower
cost of capital, higher investment and, eventually, higher growth. The
benefits were expected to arise both from efficiency gains within the
sector and also from a more efficient allocation of capital across the
economy.
This evolution in the European banking structure in many ways
imitated changes in the US that had begun to take place previously. With
the introduction of the Riegel-Neal Act (1994), the development of
national as opposed to state banks was facilitated. As a result of the
savings and loan crisis of the 1980s, there was a concern in the US that
banks which were confined to single states were more at risk from
idiosyncratic shocks affecting individual states. For example, the Fed
in Kansas City saw a major collapse in the banking system in its
district because of shocks affecting the local economy, resulting in
major insolvencies. The response was to encourage securitisation and
geographical diversification in the banking system--a move to bigger and
more national banks.
The current financial crisis has seen the collapse of some banks
within the EU, and many more have been either partly or fully
nationalised because of their inability to deal with their losses.
Because of the national basis of banking regulation within the EU it has
fallen to individual member governments to rescue 'their own'
banks:
* In the UK, the government has had to invest major funds in
rescuing Northern Rock and, more importantly, RBS and HBOS. While the
problems in RBS arose from a takeover of a Dutch bank, ABN Amro it was
the government of the UK, where the banking group was headquartered, not
the Dutch government, that had to foot the bill.
* In the case of Fortis bank, responsibility was shared
by the Belgian and Dutch governments with the bank being broken up
on national lines.
* In the case of the Irish banking system, the government had to
nationalise the biggest three national banks and take major stakes in
others.
* In Spain the government has had to rescue and reorganise the Caja
(savings) banks.
* In Germany the government had to rescue Hypo Bank and support
much of the rest of the system.
Thus the EU banking system has seen a major involvement by national
governments in capitalising and owning banks headquartered on their
territory. But in order to expand and prosper in the future, banks will
need more capital. If the markets fail to provide this, then either
governments must provide it, or banks will have to survive by gradually
reducing the size of their balance sheets.
A big question facing many governments and the EU Commission is
therefore whether governments will just recapitalise banks so that they
can lend in their own territories or whether they will recapitalise them
so that they can operate across their geographical footprint. For
example, after the January 2009 EU Finance Council, the UK Chancellor of
the Exchequer announced that while the UK government was recapitalising
RBS, he was not sure whether they would recapitalise their subsidiary on
the island of Ireland--Ulster Bank. It seems that, in the end, they
decided to keep Ulster, as it has had an injection of 3bn [euro] with a
further 3bn [euro] to come to keep them in business. Conversely the
Danish government, which has supported Dansk bank, appears to have
decided not to recapitalise its subsidiary on the island of
Ireland--Northern. The Irish government is requiring AIB to sell off its
UK, Polish and US interests to provide some of the capital needed to
allow the bank to operate normally in Ireland.
This paper addresses these issues in a systematic way. We first
look at the evolving structure of the European banking system,
describing the ownership structure as it was in 2009. We then look at
how cross-border banking activity has changed, peaking in 2007 and
subsequently declining, hinting at a trend towards reterritorialisation.
This will imply a reduction in the average scale of banks, particularly
in small economies where governments have smaller facilities available
for supporting the banking sector and also those where sovereign risk is
higher. As small banks tend to charge more for their loans, the user
cost of capital will be higher, which will impact on economic activity.
After reviewing the literature on EU bank structure, and on net interest
margins (NIM), we undertake an empirical analysis of the impact of bank
size on the NIM, using a large panel of 713 banks from the BankScope
database across fourteen countries and sixteen years. Given these
estimates of size on the margin and therefore on borrowing costs, we
look at the impacts of reducing bank size on sustainable output in the
Euro Area countries using our global model, NiGEM. It is clear from our
analysis that small countries are more adversely affected than large
ones when bank size falls.
The structure of the European banking system
We describe trends in the structure of the European banking system
before, during and since the crisis in terms of market structure. It
appears that European integration to date has chiefly been achieved
through the growth of foreign branches and subsidiaries and through
cross-border merger and acquisition (M&A) activity, rather than
through the cross-border provision of services. This represents a
consolidation of the banking sector internationally, and may have
impacted on the outcomes of integration, especially consumer welfare in
the EU member countries.
The primary theoretical benefit of financial integration is
increased competition; eliminating barriers in the form of national
borders increases both the size of the market and the number of firms in
the market, increased competition on the supply side means that banks
must improve their services in order to retain their market share. This
may take the form of improved quality or lower costs. Furthermore, if
integration takes place through expansion of existing banks through
foreign branches and subsidiaries, or through mergers and acquisitions,
integration is analogous to an increase in bank size. As banks grow
larger, up to a certain point they are able to take advantage of
economies of scale and scope, which reduces their own funding costs and
may allow them to operate more efficiently. In a competitive market
environment, this will induce banks to channel these cost advantages
into improved lending rates for clients.
Commenting on empirical trends in the sector to date, a report by
the European Commission in 2005 (Walkner and Raes, 2005) finds that in
most countries a larger share of foreign ownership of banks is
correlated with a reduction in profitability and margins of domestically
owned banks. The report cites Levine (2003), who found that restricting
foreign bank entry boosts banks' net interest margins; and Lensink
and Hermes (2004), who found that foreign bank entry not only translates
into cost benefits for consumers but also enhanced service quality, as
foreign firms bring in innovative financial services and practices.
There is a possibility that over-consolidation in the financial services
market could reduce competition and create a more monopolistic outcome,
whereby lower funding costs for banks are channelled into profits rather
than lower borrowing costs for consumers. The report notes that, based
on the declining trends in net interest margins in the past decade or
so, this does not seem to be the case; nevertheless we incorporate such
a possibility into our own analysis. Evidence on improvements in
efficiency was less clear-cut, possibly due to various institutional
barriers imposed by different legal and tax systems across countries for
instance, 'preventing exploitation of synergies in cross-border
banking' (Vander Vennet, 2002, cited in Walkner and Raes, 2005).
Walkner and Raes (2005) found that it is useful to look at
concentration ratios for measuring competition at the domestic level.
They took the asset shares of the five largest banks in each country,
and found that, with the exceptions of Denmark, Finland and Sweden, all
EU member states saw rises in their domestic concentration ratios. The
report also looks at an alternative measure of concentration, the
Herfindahl index, which considers the size of firms in relation to the
industry by looking at the sum of squares of each individual bank's
asset shares on a scale of 0 to 10. Based on this, the authors found
that all EU member states except Denmark and Sweden experienced
increases. This led them to conclude that, although domestic bank
consolidation in Europe has progressed rapidly since the introduction of
the euro, cross-border consolidation has in fact lagged behind. Domestic
M&As increase market power of individual firms but cross-border
M&As less so, and therefore the competition benefits of financial
integration may not filter through if domestic expansion were to
dominate cross-border expansion.
A more recent European Financial Integration Report by the European
Commission (2009) reveals that the competition benefits of integration
have continued into 2009, but the recent crisis has brought a focus on
domestic activity once more to the fore of the European banking
structure. They note that EU financial integration has brought with it
benefits to both home and host countries such as 'increased income
generation, improvements in technology and risk management, increased
access to funds, risk diversification and deepening of financial
markets' (EC, 2009, p. 35). All these are signs of increased
innovation, efficiency and lower costs that have been brought about by
increased competition and foreign market entry. However, it has been
widely observed that these trends were disrupted by the financial crisis
in 2007-8. Recent research by the European Central Bank (ECB, 2010) has
found that the domination of EU member states' markets by domestic
banks has marginally increased since 2007 as foreign (predominantly EU)
branches decreased. Domestic credit institutions increased their market
share in 2008 (73 per cent of total assets were domestically owned) but
decreased again in 2009. They also note that cross-border activity was
affected; intermediation and merger and acquisition activity declined
through 2007-9. The European Commission (2009) similarly observed a
segmentation of EU financial markets, and that cross-border (intra-EU)
M&A activity was surpassed by domestic activity in 2008 (both
private and government), representing an increased focus on domestic
markets, and a decline in the market share of EU branches and
subsidiaries. It appears that consolidation resumed in 2008-9, and the
number of banks in the EU diminished due to increases in merger and
acquisition activity. However, as before the crisis, this activity was
channelled mostly into domestic deals. A significant shift in ownership
structure towards government participation has taken place in some major
EU banks.
As yet it is unclear whether the recent dynamics observed over the
crisis period are merely 'temporary entrenchment by market actors
within domestic borders' (EC, 2009, p. 17) or whether they actually
represent a long-term return to a more segmented and territorialised
banking structure. It seems that many of the indicators point to this
being a temporary phenomenon. Nevertheless, recent events have called
into question the benefits of European financial integration, and though
the consensus remains that increased competition can offer clear welfare
benefits to consumers, it seems that integration should proceed along
more cautious lines. In particular, risk management is clearly now a
sensitive issue; it was suggested before the crisis that increased
consolidation and an integrated market would lead banks to expand their
investment and lending activities across borders, thereby diversifying
their portfolios and reducing risk. But integration can instead
introduce other forms of risk. For instance, an overly-concentrated
banking system resulting from domestic rather than cross-border
consolidation may be subject to idiosyncratic risk. Walkner and Raes
(2005) identified (in advance of the crisis) that foreign bank entry can
have negative consequences for macro stability if they ex-patriate
capital at signs of domestic distress, thereby causing fragility and
imposing lending restrictions on the foreign subsidiary. Integration
also exposes banking systems to sovereign risk outside their own
borders.
The EC (2009) report cites the experiences of Central, Eastern and
Southern Europe (CESE) as a case study for the macroeconomic stability
consequences of financial integration. This area saw the largest rise in
foreign participation in domestic banking markets, but this has actually
reduced diversification and increased concentration in individual
markets. The report notes that most CESE countries rely on western
European banks for their funding, resulting in a build-up of 'major
concentration exposures', whereby the use of 'common funding
channels' (in the CESE case, this means Austria and Sweden)
increases risk and vulnerability of the funding channel countries to
macroeconomic shocks in the host countries. As a result of this, and the
'lack of adequate risk management and regulation', the report
notes there were major imbalances in the banking system. Furthermore,
the easy access to foreign loans which had previously been seen as
conducive to consumption and growth resulted in a speculative property
bubble. The report concludes that financial integration has brought
clear benefits, but the stability aspects have not received enough
attention.
Although it is clear from the European Commission reports described
above that consolidation across borders is important for the competitive
consequences of integration, what also matters is domestic bank
participation in foreign markets. This captures the size of the market
on the supply side, rather than bank size. We undertake our own analysis
of the structure of the European banking system and how it has changed
from this perspective. We firstly consider changes in national market
shares as given by the ratio of foreign claims to total assets in the
banking system; and secondly, we consider changes in assets held by
foreign branches and subsidiaries within the EU. The first indicator
shows how cross-border lending has changed over the period between 2000
and 2009, and the second illustrates expansion of banks into other EU
countries.
The indicator we use to analyse changes in cross-border banking
activity is the ratio of total annual reported claims of foreign banks
since 2000 to total Monetary and Financial Institution (MFI) assets. (1)
Information on foreign claims comes from the Bank for International
Settlements (BIS) consolidated banking statistics, MFI total assets from
the Deutsche Bundesbank. (2) We present the ratios by country in table
1, and plot an unweighted average across countries in figure 1 in order
to illustrate the patterns of internationalisation and
reterritorialisation of banking within Europe. Cross-border activity
clearly peaked in 2007, but this has been reversed in the last two years
of our sample.
We then used this indicator to analyse the market shares of foreign
claims in 2007 and 2009. Figure 2 plots the changes in shares between
2007 and 2009 to illustrate recent trends. This demonstrates that, while
the share of domestic banking in host countries has increased in the
majority of countries recently, the change in total foreign market share
of lending is negative for almost all countries in the sample,
indicating that European banking has become more territorialised and
more national since the crisis. The only exceptions to this are Belgium,
for which international lending has surpassed domestic banking, and
Portugal, whose lending structure remains unchanged.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
We can also look at the shares of bank assets held by foreign
branches and subsidiaries located in the 27 members of the EU over the
period 2003-9, as well as the Euro Area countries and the EU in
aggregate. Assets held by foreign branches and subsidiaries indicate the
amount of cross-border expansion of banks that has taken place within
each country. Estonia, the Czech Republic and Luxembourg are notable for
having over 90 per cent of their bank assets held by foreign-owned
banks, and hence their domestic banking sectors are almost non-existent.
At the other end of the spectrum, larger economies such as France,
Germany, Spain and Italy have under 20 per cent of their bank assets
held by foreign-owned banks. This figure has not increased since 2003
for France and Spain, whose domestic banks clearly dominate their
financial markets. The UK however has over 50 per cent of its bank
assets held by foreign branches and subsidiaries, and this figure peaked
at nearly 80 per cent in 2005; these particularly large figures signify
the importance of the UK as a global financial centre. Cross-border
activity through firms and subsidiaries was clearly declining for most
countries through 2008 and 2009, illustrating that the effects of the
crisis impacted markedly on participation in foreign markets.
Figure 3 plots this data, and divides it into four series showing
(i) assets held by all foreign branches and subsidiaries located in the
EU, (ii) assets held by other EU branches and subsidiaries located in
the EU, (iii) assets held by non-EU branches and subsidiaries located in
the EU, and (iv) assets held by all foreign branches and subsidiaries
located in the Euro Area. The chart illustrates that foreign-held assets
peaked in 2005 in the EU, and declined over the crisis years but have
started to pick up in 2009. By contrast, foreign-held assets peaked in
2007 for the Euro Area, but have not experienced the positive growth
that the EU as a whole experienced in 2009. This is made clearer by
figures 4 and 5, which illustrate the absolute annual (percentage point)
changes in assets held by foreign branches and subsidiaries in the EU
and the Euro Area; this was positive for the EU in 2009 but was actually
still declining in the Euro Area.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
The charts also illustrate the driving forces behind recent
dynamics seen in EU financial markets. EU branches and subsidiaries
clearly make up the bulk of foreign ownership, and hence the decline in
overall foreign banking in the EU over 2008 was driven chiefly by the
decline in EU branches and subsidiaries (and almost entirely by those in
the Euro Area).
Given that borrowing costs have been falling since the 1990s, we
can conclude that the growth in cross-border banking activity through
foreign lending and expansion of foreign branches and subsidiaries
within the EU has translated into welfare benefits for consumers. This
finding is in accordance with the European Commission and European
Central Bank reports described above.
[FIGURE 5 OMITTED]
The cost of funds
As noted, EU banking integration has taken place mainly through
merger and acquisition activity and through the expansion of services in
other EU countries through branches and subsidiaries. Both of these
modes imply growth in the size of banks, and therefore integration can
be seen through the perspective of firm size. The key significance that
firm size has in the banking sector is the cost of funds; this is the
channel through which banking activity influences the economy. However
the direction of this influence depends on the amount of market power
banks have, as in oligopolistic or monopolistic markets they may not
pass on lower costs to consumers. Hence we analyse the relationship
between bank size (often measured by asset size) and the net interest
margin (NIM), starting with a review of the relevant literature. The NIM
is the spread between a financial institution's gross earnings on
interest-bearing assets and its interest expenses in funding those
assets. It can be decomposed into several components--profits,
operational costs, regulatory costs and potential costs of default. Each
of the components responds to either cyclical or structural shocks, or
both. Consequently, it is one of several measures of bank profitability,
but moreover captures the functioning of banks generally in terms of
efficiency and competitiveness, which in turn impacts on saving,
investment and therefore growth (Demirguc-Kunt and Huizinga, 1999, p.
2). The relationship between the NIM and bank size may consequently be
driven by two factors; the earnings side of the spread and the costs
side.
With regards to the first, the lending activity of small banks
tends to be channelled into small business or personal loans which
generate higher yields than larger loans, due to the associated higher
expected losses and transaction costs. Furthermore, small banks tend to
hold a larger percentage of their asset mix in such loans than in
lower-yield investment securities. Large banks, on the other hand, are
more able to take advantage of economies of scale, and also perhaps
economies of scope in their more diverse array of products. This enables
them to offer larger loans at more competitive rates. In addition, they
tend to hold a higher proportion of their assets in lower-risk corporate
or government bonds. Considering the other side of the spread, small
banks often have lower funding costs as they rely on low-interest core
deposits, whereas large banks tend to be dependent on costly and more
volatile wholesale funding which is closely linked to the official bank
rate. The combination of these characteristics means that the NIM tends
to be larger for small banks.
[FIGURE 6 OMITTED]
Figure 6 plots the average NIMs across banks in a selection of
advanced economies between 1993 and 2008. On the whole, NIMs have been
falling throughout this period, indicating that banks may have become
larger in these countries or that competition has increased. It may also
suggest that banks have been able to lower costs and pass these on to
consumers. The crisis years of 2007 and 2008 saw a fairly drastic
upwards surge in average NIMs in the European countries depicted and in
the US and, if the size-NIM relationship proffered above were to hold,
this would indicate a sudden downsizing of banks in response to the
crisis. NIMs in Japan and Canada surged in the opposite direction
however.
Much of the literature on net interest margins attempts to identify
its determinants using an analytical framework that was first developed
by Ho and Saunders (1981), in which they modelled bank interest margins
as a function of managerial risk aversion, the size of bank
transactions, bank market structure and the variance of interest rates.
Since then, there have been a number of developments, including a paper
by Demirguc-Kunt and Huizinga (1999), who tested a variety of banking,
macroeconomic, regulatory, structural and institutional characteristics
as determinants of bank interest margins. They find that larger bank
asset to GDP ratios and lower market concentration ratios lead to lower
interest margins (substantiating the inverse size-NIM relationship), and
more notably perhaps that foreign ownership is associated with higher
NIMs (this is more pronounced in developing countries).
Recent literature has begun to look more specifically at the
relationship between bank size and the NIM, much of this focusing on
Europe as the internationalisation of its banking sector in the five
years or so up to 2000 provides an interesting forum for such
investigation. Maudos and Fernandez de Guevara (2004) analyse the
determinants of the NIM in Germany, France, the UK, Italy and Spain, and
find that declining margins in the European banking sector over the
period 1993-2000 can be explained by increased market power and
concentration as well as interest rate risk, credit risk, operating
expenses and bank risk aversion. Market power, concentration and
decreased competition are all associated with the banking sector
becoming more oligopolistic (i.e. fewer, but bigger banks). This
substantiates the findings of the EC and ECB reports with respect to how
bank size affects the NIM, but goes contrary to what theory suggests
regarding the impact of reduced competition on margins.
Kasman et al. (2010) investigate this topic further by examining
the effects of bank consolidation in the new European Union members and
candidate economies on the determinants of the NIM over two sub-periods,
comprising the consolidation period of 1995-2000 and the
post-consolidation period of 2001-6. In accordance with the conclusions
of the Maudos and Fernandez de Guevara paper; Kasman et al. note that
the European banking structure over the entire time period examined was
characterised by the cross-border expansion of financial intermediaries
and by a 'wave of mergers and acquisitions ... [leading to] a
reduction in the number of banks in many old and new member
countries' (Kasman et al., 2010, p. 649). They find that both bank
size and managerial efficiency are significant and exhibit a negative
relationship with the NIM over both sub-periods, suggesting that this
relationship was robust to the structural changes that took place over
this time period.
Econometric analysis
We approach the same question as that tackled by Kasman et al.,
which is how the NIM is affected by bank size in Europe in the
consolidation period since the 1990s. However we have a dual purpose,
which is not only to quantify this relationship but furthermore to use
these results to show what would be the economic impact if the recent
trend of re-territorialisation and fragmentation of the banking system
that was briefly observed through 2007-8 was to continue. We estimate
the relationship between bank size and the NIM (where the NIM is
approximated by the ratio of net interest revenues to total assets) and
model the NIM as a function of cyclical and structural factors.
The set of variables we include in our regression are GDP growth
and the real growth of loans, which both capture cyclical fluctuations,
the capital adequacy ratio, which is an instrument of regulatory policy,
the lagged NIM, and bank size. We use micro data for total assets and
loans, and the (risk-adjusted) total capital ratio, all from the
Bankscope database over the period 19932008. Our sample excludes central
banks, specialised governmental credit institutions and multi-lateral
government banks. Macroeconomic data for inflation, GDP growth and house
prices are obtained from national sources as collected in the NIESR
NiGEM database.
We use a normalised bank size variable (see Barrell et al., 2010).
We calculated the mean and the standard deviation of the bank assets
across all countries in each year, and then scaled each bank by the
number of standard deviations of its assets from the mean. This accounts
for increasing density in the time domain, as the number of banks in our
sample rises over time, and hence it is possible that there is a
downward trend in average share. Using the NIM as the dependent
variable, we estimate over the period 1993-2008 using Ordinary Least
Squares. Results of estimation are shown in table 2.
The estimation confirms the results of previous studies reviewed
above, namely that bank size negatively affects the NIM; the bigger the
bank, the greater its economies of scale and the lower its costs, which
allows it to reduce the differential between the lending and the deposit
rate. Conversely, the capital adequacy ratio has a positive effect on
the NIM, as it increases bank costs. GDP growth and the real growth of
loans mirror cyclical fluctuations in the margin, both through profits
as well as the potential costs of the default. The default costs can be
decomposed into those related to cycle-dependent systemic risk, and
those related to bank-size-dependent individual risk (as a bank's
portfolio increases, the individual risk falls--and with banks getting
larger the probability of a bank having a more diversified portfolio
increases). Both GDP and the real growth of loans have a positive effect
on the NIM, indicating that it is procyclical.
As noted previously, our central premise that bigger banks can
exploit economies of scale in portfolio pooling and thus reduce their
NIM holds under perfect competition. However, as bank size increases,
monitoring costs may start to rise, and the bank may start behaving as a
monopolist. Both of these may mean that margins start to increase above
some size threshold. To capture this nonlinearity in the size-NIM
relationship, we augment the baseline regression with our bank size
variable squared.
The results confirm that, as hanks get bigger, they can exploit
their increasing market power and also gain from the economies of
pooling risk which they pass on to their customers. The relative role of
market power is limited however; as bank size increases, bank margins
fall until banks are more than four standard deviations above the mean
of bank size, which is around 0.15 per cent of the global total. We have
computed the threshold bank size above which diseconomies of scale could
start playing a role. This value is time-dependent, and also depends on
the number of banks in the world banking system. In 2008, if the total
assets of a bank exceeded 2.3 per cent of the total assets in the world
banking system, the bank could have been considered too big. There was
at least one bank in each of the US, UK, Germany, France, Japan and the
Netherlands, whose assets exceeded 2.3 per cent of the global assets in
2008. We can examine the effects of different banking systems on the EU
economy using our global econometric model, NiGEM.
The structure of the NiGEM model
The National Institute Global Econometric Model (NiGEM) is
structured around the national income identity, can accommodate
forward-looking consumer behaviour, and has many of the characteristics
of a Dynamic Stochastic General Equilibrium (DSGE) model. Unlike a pure
DSGE model however, NiGEM is based on estimation using historical data.
What follows is a description of its structural properties.
Production and price setting
The major country models rely on an underlying
constant-returns-to-scale CES production function with labour-augmenting
technical progress.
Q = [gamma][[s[(K).sup.-[rho]] + (1 -
s)[([Le.sup.[lambda]t]).sup.-[rho]]].sup.-1/[rho]] (1)
where is Q is real output, K is the total capital stock, L is total
hours worked and t is an index of labour-augmenting technical progress.
This constitutes the theoretical background for the specifications of
the factor demand equations, forms the basis for unit total costs and
provides a measure of capacity utilisation, which then feed into the
price system. Barrell and Pain (1997) show that the elasticity of
substitution is estimated from the labour demand equation, and in
general it is around 0.5. Demand for labour and capital are determined
by profit maximisation of firms, implying that the long-run
labour-output ratio depends on real wage costs and technical progress,
while the long-run capital-output ratio depends on the real user cost of
capital
Ln(L) = [[sigma]ln{[beta](1 - s)} - (1 - [sigma])ln([gamma])] +
ln(Q) - (1 - [sigma])[lambda]t - [sigma]ln(w / p) (2)
Ln(K) = [[sigma]ln([beta]s) - (1 - [sigma])ln([lambda])] + ln(Q) -
[sigma]ln(c / p) (3)
where w/p is the real wage and c/p is the real user cost of
capital. The user cost of capital is defined as:
c = [(1 - [mu]) * (r + IPREM) * (1 - CTAXR) + [mu] * (r + PREM) +
[theta]]/(1 - CTAXR) (4)
Equation (4) shows that the user cost of capital is influenced by
corporate taxes (CTAXR) and depreciation ([theta]), and is a weighted
average of the cost of equity finance and the margin adjusted long real
rate (r), with weights that vary with the size of equity markets ([mu])
as compared to the private sector capital stock. Hence the investment
premium (IPREM) directly feeds into firms' borrowing costs and thus
their investment decisions.
The NiGEM model contains a description of the banking sector as in
Barrell, Davis and Kirby (2010). Bank net interest margins feed into the
behaviour of individuals and firms through the wedge between borrowing
and lending rates. Individuals face LENDW as their spread, and firms
face CORPW as theirs. Our work suggests that corporate borrowing spreads
charged by banks and bond market premia for corporates move together. We
can simulate the effects of changes in bank scale by raising LENDW and
CORPW in line with any presumed changes in Net Interest Margins.
Consumption, personal income and wealth
Consumption decisions are presumed to depend on real disposable
income and real wealth in the long run, and follow the pattern discussed
in Barrell and Davis (2007). Total wealth is composed of both financial
wealth and tangible (housing) wealth where the latter data is available.
ln(C) = [alpha] + [beta]ln(RPDI) + (1 - [beta])ln(RFN + RTW) (5)
where C is real consumption, RPDI is real personal disposable
income, RFN is real net financial wealth and RTW is real tangible
wealth. The NIM feeds into the consumption decision through real
disposable income, which comprises wages, government transfers and
receipts on interest-bearing assets owned by the household sector net of
borrowing costs. An increase in the NIM indicates a rise in the cost of
borrowing, and it therefore reduces real disposable income, impacting
negatively on the consumption decision.
The dynamics of adjustment to the long run are largely data based,
and differ between countries to take account of differences in the
relative importance of types of wealth and of liquidity constraints.
Financial markets
We generally assume that exchange rates are forward looking, and
'jump' when there is news. The size of the jump depends on the
expected future path of interest rates and risk premia, solving an
uncovered interest parity condition.
RX(t) = RX(t + 1)[(1 + rh)/(1 + ra)](1 + rprx) (6)
where RX is the exchange rate, rh is the home interest rate set in
line with a policy rule, ra is the interest rate abroad and rprx is the
risk premium. Nominal short-term interest rates are set in relation to a
standard forward-looking feedback rule, as discussed in Barrell, Hall
and Hurst (2006). Forward-looking long rates are related to expected
future short-term rates,
(l + L[R.sub.t]) = [[PI].sup.T.sub.j=l] (l + S[R.sub.t+i])l/T (7)
We assume that bond and equity markets are also forward looking,
and long-term interest rates are a forward convolution of expected
short-term interest rates. Forward-looking equity prices are determined
by the discounted present value of expected profits.
The economic impact of banking system restructuring
Having established the inverse relationship between bank size and
the net interest margin, we now use the NIM to proxy bank downsizing, in
order to analyse the impact on output. If banks re-territorialise then
the NIM will increase, and will increase the most in small countries
because we have a quadratic impact from size on margins, and its impact
is steepest for smaller sizes of country. We may write the relationship
between the change in size and the NIM as the derivative of the equation
in table 3 with respect to size, where D(NIM) is the change in the net
interest margin. Table 3 yields the equation,
NIM = [??] - 0.44 * SIZE + 0.052 * [SIZE.sup.2] + [??] * X + [??]
(8)
where [??] represents a vector of the remaining estimated
parameters, and X the remaining explanatory variables in the equation.
Hence,
D(NIM) = -0.44 + 0.104 * SIZE (9)
If we reduce bank size in Germany, France, Italy and Spain by two
standard deviations, then the NIM will rise by 100 basis points. The
largest bank would then be below the maximum efficient scale rather than
above it as is the case now. We reduce bank size by twice this in the
small countries, the Netherlands, Belgium, Portugal, Greece, Austria and
Finland, and given the scale of the Irish banking system relative to its
GDP we reduce the average size of banks by an additional amount of the
same size. The effects on output are shown in figure 7. We assume that
financial markets are forward looking, and equity prices and long-term
interest rates jump after a shock. This is because the monetary
authorities react to the shock and change interest rates and markets
react to these changes. We also assume that labour bargains take account
of future inflation. Governments set their taxes in order to remain
solvent and so when revenues fall, tax rates rise. Furthermore, we
assume consumers are myopic and not forward looking. We apply shocks
only to Euro Area countries.
There are three sets of factors that affect the long-run impacts on
output. Countries with higher capital-output ratios will have larger
effects as compared to others, and as a result the effect in Germany is
particularly large. However, the long-run effects come through the user
cost of capital, and this is the weighted average of equity, bank and
bond finance. We are only raising the cost of non-equity finance, and
hence France has less effect because private and market equity finance
are more important there than in the other large economies. Italy has
the largest effect because it has the least developed equity market of
the four large economies, and this is reflected in the weights in our
user cost equation. Apart from France, the decline in banking sector
scale economies reduces equilibrium output by almost 1 per cent. The
long-run effects on Greece, Portugal, Austria and Finland are twice as
large as this as they are also relatively bank dependent and the shock
is twice as large as their banks shrink. The Dutch and Belgians are less
reliant on bank finance, and hence the long-run impact of a similar
shock is smaller. Ireland faces an even larger increase in margins but,
as it is more similar to the UK, with a relatively strong reliance on
equity finance, the impacts are muted and are the same size as the
bank-dependent small economies.
If bank margins were to be increased in this way, these
calculations suggest that growth in the Euro Area would be reduced by
0.2 percentage points a year for five years. The short-run effects
depend upon the speed of adjustment of the capital stock and on the
level of gross borrowing in the economy. If the personal sector has
large borrowing as compared to income then a rise in borrowing costs on
deposits that is not reflected in income will reduce consumption and
demand quickly. Levels of personal sector debt are high in Spain, and
adjust more quickly than in other large economies. However if an economy
is small and open then most of the effects leak out into imports, as in
Ireland, the Netherlands and Belgium, where borrowing has been high, and
where income adjusts slowly. However, the overall output in these
economies is likely to be larger than average, as we can see from figure
7.
[FIGURE 7 OMITTED]
Conclusion
The Single Market in Financial Services has meant that banks could
become larger in scale, and hence borrowing costs for consumers and for
firms will have been lower. We have shown that banks within Europe
increased in size and became more international, at least until 2007
when that process went into reverse. We have used a micro data set to
investigate the impact of size on banks' net interest margin, and
have shown that larger banks have smaller spreads between borrowing and
lending rates for firms and households. As we have competition between
deposit takers, this largely reflects the fact that they charge their
borrowers less. Lower borrowing costs for households raise their incomes
and their consumption, whilst for firms lower borrowing costs mean that
they raise their investment and this in turn will raise sustainable
output. Lower borrowing costs reduce the user cost of capital, and hence
increase the equilibrium capital stock, and this would also raise
sustainable output. A 1 percentage point reduction in borrowing costs
would raise Euro Area output by 1/2 per cent within four years and by
3/4 per cent in the long run. However, this gain from size must also be
weighed up against the increase in risky behaviour that is associated
with larger banks, as shown by Barrell, Davis, Fic and Karim (2010).
After looking at the effects of bank size on borrowing costs, we
investigate the potential impact of banks moving back into their home
territory. In summary, re-territorialisation implies that banks will
become smaller, and economies will shrink. We investigate the impacts on
output in large and small countries showing that the effects are
generally larger in small countries, and also larger in economies that
are more dependent on bank finance for their business investment
decisions. Our simulations suggest that overall, Euro Area growth would
be 0.2 per cent per annum lower for a few years. Competition in banking
in the European Single Market in Financial Services has brought benefits
and raised output, especially in the smaller economies. However, poor
regulation at the Area level and the 'too big to fail'
guarantee have meant that the costs may have outweighed the benefits.
DOI: 10.1177/0027950111411381
Appendix
Table A1. Country distribution of banks
Country Banks Per cent of sample
Belgium 4 0.6
Canada 27 3.8
Denmark 12 1.7
Finland 6 0.8
France 53 7.4
Germany 36 5.0
Italy 35 4.9
Japan 132 18.5
Netherlands 20 2.8
Norway 9 1.3
Spain 48 6.7
Sweden 5 0.7
UK 51 7.2
US 275 38.6
Total 713
REFERENCES
Barrell, R. and Davis, E.P. (2007), 'Financial liberalisation,
consumption and wealth effects in 7 OECD countries', Scottish
Journal of Political Economy, 54, May, pp. 254-67.
Barrell, R., Davis, E., Fic, T. and Karim, D. (2010), 'Is
there a link from bank size to risk taking?', NIESR Discussion
Paper no. 367.
Barrell, R., Davis, E. and Kirby, S. (2010). 'Modelling the UK
banking sector', National Institute Economic Review. 214, October,
pp. 67-72.
Barrell, R., Hall, S. and Hurst, I. (2006). 'Evaluating policy
feedback rules using the joint density function of a stochastic
model', Economics Letters, 93(1), October, pp. 1-5.
Barrell, R. and Pain, N. (1997), 'Foreign direct investment,
technological change, and economic growth within Europe', Economic
Journal, 107, pp. 1770-6.
Demirguc-Kunt, A. and Huizinga, H. (1999), 'Determinants of
commercial bank interest margins and profitability: some international
evidence', World Bank Economic Review, 13, 2.
European Central Bank (2010), EU Banking Structures. ECB,
September.
European Commission (2009), European Financial Integration Report,
European Commission.
Ho, T.S.Y. and Saunders, A. (1981), 'The determinants of bank
interest margins: theory and empirical evidence'. The Journal of
Financial and Quantitative Analysis, 16, 4.
Kasman, A., Tunc, G., Vardar, G. and Okan, B. (2010),
'Consolidation and commercial bank net interest margins: evidence
from the old and new European Union members and candidate
countries', Economic Modelling, 27, 3.
Lensink, R. and Hermes, N. (2004), 'The short-term effects of
foreign bank entry on domestic bank behaviour: does economic development
matter?', Journal of Banking and Finance, 28, pp. 553-68.
Levine, R. (2003), 'Denying foreign bank entry: implications
for bank interest margins', Central Bank of Chile Working Papers,
No. 222, August.
Maudos, J. and Fernandez de Guevara, J. (2004), 'Factors
explaining the interest margin in the banking sectors of the European
Union', Journal of Banking and Finance, 28, 9.
Walkner, C. and Raes, J. (2005), 'Integration and
consolidation in EU banking--and unfinished business', European
Commission, Economic Paper 226, April.
NOTES
(1) Includes banks whose ultimate owners are located in Australia,
Austria, Belgium, Brazil, Canada, Chile, Chinese Taipei, Denmark,
Finland, France, Germany, Greece, Ireland. Italy, Japan, Mexico, the
Netherlands, Panama, Portugal, Spain, Sweden, Switzerland, Turkey, the
UK, and the US. It is necessary to note that ownership is defined here
by headquarter location as opposed to shareholder characteristics.
(2) Data for Denmark and the United Kingdom were obtained from
their respective central banks.
Ray Barrell,* Tatiana Fic,* John Fitz Gerald,** Ali Orazgani* and
Rachel Whitworth*
* NIESR, E-mail:
[email protected]. ** ESRI, Dublin.
Table 1. Share of consolidated foreign claims of reporting banks over
MFI total assets
Austria Belgium Denmark Finland France Germ.
2000 0.29 0.49 0.31 0.64 0.21 0.14
2001 0.44 0.50 0.34 0.48 0.19 0.14
2002 0.43 0.67 0.34 0.41 0.20 0.16
2003 0.42 0.61 0.34 0.40 0.20 0.16
2004 0.43 0.62 0.49 0.81 0.21 O.18
2005 0.27 0.60 0.51 0.75 0.21 0.19
2006 0.32 0.58 0.41 0.82 0.25 0.20
2007 0.48 0.55 0.41 0.85 0.24 0.26
2008 0.41 0.58 0.34 0.80 0.20 0.24
2009 0.33 0.76 0.40 0.72 0.18 0.24
Greece Ireland Italy Lux. Neths Portugal
2000 0.69 0.72 0.37 0.72 0.43 0.41
2001 0.58 0.73 0.35 O.65 0.41 0.36
2002 0.63 0.75 0.33 0.61 0.40 0.43
2003 0.57 0.83 0.31 0.62 0.41 0.43
2004 0.61 0.86 0.31 0.68 0.40 0.45
2005 0.72 0.81 0.31 0.74 0.52 0.49
2006 0.79 0.83 0.39 0.93 0.55 0.49
2007 0.76 0.97 0.38 0.95 0.59 0.50
2008 0.69 0.86 0.33 0.96 0.48 0.49
2009 0.50 0.80 0.31 0.94 0.40 0.5O
Spain UK
2000 0.23 0.31
2001 0.21 0.30
2002 0.24 0.29
2003 0.25 0.30
2004 0.27 0.35
2005 0.27 0.30
2006 0.31 0.34
2007 0.32 0031
2008 0.30 0.27
2009 0.27 0.29
Source: BIS, Bundesbank, Bank of England and National Bank of
Denmark.
Table 2. Regression results
Net interest revenues/total assets
coefficient t-statistic
Lagged dependent {NIM(-1)) 0.95 187.94
GDP growth (-1) 0.009 2.02
Real loan growth {-1) 0.003 7.66
Capital adequacy (-1) 0.003 3.11
Size (normalised) (-1) -0.006 -1.99
Observations 4374
Banks 588
Notes: Fourteen countries in sample, including US and Japan. Two
banks dummied out in 2008.
Table 3. Regression results with nonlinearity variable
included
Net interest revenues/total assets
coefficient t-statistic
Lagged dependent (NIM(-1)) 0.95 184.01
GDP growth (-1) 0.01 2.16
Real loan growth (-1) 0.003 7.63
Capital adequacy (-1) 0.003 2.71
Size (normalised) (-1) -0.02 -3.46
Size (normalised) ^ 2 (-1) 0.003 3.21
Observations 4374
Banks 588
Notes: See table 2.
