Barry Williams

440268
2012
AUM0010.1177/0312896212440268Williams and RajaguruAustralian Journal of Management
Article
Australian Journal of Management

38(1) 99–123
The chicken or the egg? © The Author(s) 2012
Reprints and permissions:
The trade-off between bank fee sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0312896212440268
income and net interest margins aum.sagepub.com
Barry Williams and Gulasekaran Rajaguru

School of Business, Bond University, Queensland, Australia
Abstract
This study considers the time series relationship between bank fee income and bank net interest margins in
Australia, applying panel vector autoregressions to a unique, hand-collected dataset. Increases in bank fee
income are being used to supplement decreases in net interest margins. The increase in magnitude of fee
income associated with reductions in margin income is smaller than the decrease in net interest margins,
resulting in a net wealth transfer favouring users of bank services; although not all users of bank services
gained and/or gained equally. The overall increase in fee income is marginally greater that the reduction in
margin income. It is argued that banks have responded to falling margin revenue by increasing their range
of fee-based services, especially insurance. Increases in fee income are found to pre-date declines in margin
income, thus Australian banks were pro-active in the process of disintermediation.
JEL Classifications: G21, G11, C33
Keywords
Bank interest margins, non-interest income, panel data, vector autoregression
1. Introduction
Changes in the nature of financial intermediation have been accompanied by a change in the nature
of bank income (Allen and Santomero, 2001). In particular, the revenue of banks has seen a shift
in emphasis from traditional income sourced from the provision of intermediary services (margin
income) to the less traditional fee income.1 Such a shift has a number of important implications
from the perspectives of bank management and regulatory policy. A conventional view of this
process is that banks have offset the impact of reduced traditional income sourced from margin
income via increases in fee income. Reductions in margin income can be attributed to the process
of disintermediation and increased competition.
It is also often assumed that the reductions in net interest margins pre-date the increases in fee
income and that increases in fee income are a reaction to falling revenue. It has been concluded that
Corresponding author:
Barry Williams, School of Business, Bond University, Gold Coast, Queensland 4229, Australia.
Email: Barry_Williams@bond.edu.au
Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

100 Australian Journal of Management 38(1)
observed increases in fee income are acting to supplement declines in margin income rather than
replacing margin income.2 However, it is possible that fee income increases pre-date falls in mar-
gin income and this would indicate that any observed trade-off in fact represents a fundamental
shift in the nature of bank revenue reflecting the overall impact of disintermediation.
Recent studies found that increases in fee income are accompanied by increased variability in
profits and worsening in bank risk–return trade-off (DeYoung and Rice, 2004; Laeven and Levine,
2007; Lepetit et al., 2008b). Further, the level of exposure of U.S. banks to fee income has resulted
in worsening of bank’s risk–return trade-off.3 Other recent European studies such as Lepetit et al.
(2008a, 2008b) and Schmid and Walter (2009) found that bank income diversification is value
reducing. The global study by Laeven and Levine (2007) also found increased levels of financial
conglomeration generated a diversification discount and increased bank agency problems. This
paper addresses the issue of changing bank revenue from a perspective that has not been applied
previously in this literature. This paper will also offer the benefit of employing a unique hand-
collected database sourced from the annual reports of Australian banks (domestic and foreign) over
the period 1988 to 2010.
The research question ‘Is there a stable time series relationship between fee income and margin
income in Australia?’ will be considered. If a stable long-term relationship is found, this will tend to
support the argument that increases in fee income have been used to offset declines in margin
income. If no stable long-term relationship is found, it will support the argument that the factors
causing the observed declines in margin income and the accompanying increase in fee income are
in fact derived from different sources (such as providing market-based services or ventures into new
avenues of financial service provision, rather than intermediation services), suggesting a change in
the nature of financial intermediation, in particular an increased move toward the use of market-
based financial solutions (Allen and Santomero, 2001). It would be expected that if the substitution
argument presented above is true, then there will be a negative time series relationship between fee
income and margin income. A further question that this paper will also address is whether the nature
of this relationship differs between the three types of banks considered in this study.
These results are of interest as this is the first time that the issue of margins and fee income has
been considered from this perspective that these authors are aware of. These results will contribute
to the body of knowledge considering the changing nature of intermediation, and so aid in our
understanding of whether the observed changes in bank revenue composition are transitory or
permanent. The Reserve Bank of Australia (2006) found that bank fees from domestic banking
activity had grown over the previous years, but that as a percent of total assets, domestic fee
income had declined. The Reserve Bank survey confines its focus to fee income earned in the pro-
cess of taking deposits and writing loans, and so excludes income from funds management, whole-
sale banking, insurance and trading activities. As the broader type of fee income has been a growing
area of banking operations in Australia and globally (Allen and Santomero, 2001; Laeven and
Levine, 2007), a wider perspective of bank fee income is also relevant, as will be adopted in this
paper. This paper offers the second advantage of addressing this question by analysing a unique
hand-collected database that includes both domestic and foreign banks in the sample, thus offering
wider scope for analysis than previous Australian studies.
In general this paper finds that there exists a stable relationship between bank fee income and
margin income, indicating that increases in fee income are being used to supplement declines in
margins. It is also found that increases in fee income pre-date declines in margins, suggesting that
the Australian banking system has been pro-active in dealing with the process of disintermediation
by increasing fee income prior to declines in margins becoming statistically significant. This paper
argues that the observed increase in bank fees has two components. The first component is the

Williams and Rajaguru 101
increase in fees associated with the decline in margin revenue; this is found to be smaller than the
decrease in margins. As the magnitude of increases in fees is smaller than the observed reduction
in bank margins, this suggests that bank consumers have experienced positive wealth transfers
overall as a result of this relationship. The second component of increased bank fees is argued to
be due to banks offering a wider range of financial services (particularly insurance) than previ-
ously. The sum of these two increases in bank fees is marginally greater than the observed decline
in margin revenue in absolute terms, and this difference is extremely small when the growth of
bank assets is considered. In net, Australian banks have increased their fee revenue to compensate
for the decline in revenue due to falling margins, but the overall impact has represented two dis-
tinctly different strategies.
These results do raise some policy concerns, as overexposure to fee income results in a worsen-
ing of bank risk–return trade-offs (De Jonghe et al., 2007; Stiroh and Rumble, 2006) as well as
increased agency conflicts (Laeven and Levine, 2007) and worsening loan quality (Lepetit et al.,
2008a). These potentially negative outcomes are of concern to bank shareholders, bank manage-
ment, prudential regulators and borrowers from banks as they all face agency concerns resulting
from bank diversification (Froot and Stein, 1998; Laeven and Levine, 2007; Stiroh and Rumble,
2006), and particularly from potential bank failure.
This paper is structured as follows: the next section will provide an overview of the relevant
literature. The third section will discuss the sample to be used and presents the relevant descriptive
statistics. The fourth section will discuss the method used to address the research question posed.
The fifth section will discuss the results, while the final section will provide some concluding com-
ments and suggest directions for further research endeavour.
2. Literature review
As pointed out by Allen and Santomero (2001), the nature of the financial system has changed
dramatically over the last decade, with banks becoming increasingly active in the provision of non-
traditional services such as insurance products, funds management and securitisation.4 This change
is sourced from the increased competition posed to traditional intermediaries from non-traditional
sources including the evolution of more sophisticated market-based products that directly compete
with banks. Slager (2006, Ch 4) illustrates this trend, showing that across a number of developed
nations margins have declined while fee revenue has increased. This trend is viewed by Slager
(2006) to represent the impact of disintermediation.
The evidence considering this issue to date has a focus upon the case of the United States,
mainly driven by the research question posed by the impact upon bank risk resulting from the
removal of the Glass–Steagall separation of commercial and investment banking. DeYoung and
Roland (2001) find that the resulting change in income mix emphasising fees is accompanied by
increased earnings volatility that represents both the volatility of fees as well as volatility due to its
leverage effects. DeYoung and Rice (2004) find that banks that are less reliant upon fees generally
exhibit higher management quality, and that customer focus and technology use are associated with
higher levels of fee income. Further, increases in fees are associated with a worsening of the bank’s
risk–return trade-off and increased profit variability.
Stiroh and Rumble (2006) find that increased reliance upon fees as a revenue source generated
a positive portfolio diversification effect and a negative impact via the higher volatility of fees. As
this level of exposure increases so the volatility effect outweighed the diversification effect result-
ing in a worsening of bank risk–return trade-off. Stiroh (2006b) find that increased fees were not
accompanied by higher share market returns, but were accompanied by increased market risk

(beta, total volatility and idiosyncratic volatility). Thus large U.S. banks may have become overex-
posed to fee-based revenue (Stiroh, 2006a). Further, Stiroh (2004) finds that the correlation between
margins and fees has increased over time, so reducing any portfolio diversification benefits from
combining the two income sources. Overall the evidence drawn from the United States indicates
that increased fee-based income is risk increasing rather than risk reducing.
In the European Union Smith et al. (2003) find that fees are of increased importance to banks,
but that fees display a higher volatility than margins. De Jonghe et al. (2007) find that European
banks with higher levels of fees have higher expected returns as measured by Tobin’s Q, but also
have higher beta risk. They also conclude that overexposure to fees increases bank risk. Laeven
and Levine (2007) studied this issue from a different perspective across 43 nations and concluded
that financial conglomerates had a lower market value than focussed financial institutions and thus
there exists a diversification discount in multiple activity financial firms, due to the negative effect
of agency problems. European studies by Lepetit et al. (2008a), Lepetit et al. (2008b) and Schmid
and Walter (2009) all confirm that any diversification benefits from increased fees are more than
outweighed by increased bank risk. Lepetit et al. (2008a) document that European banks accept
lower loan portfolio returns and higher loan risk to obtain higher commission and fee income.
Three reasons have been presented to explain why fees are more volatile than margin income
(DeYoung and Roland, 2001). First, as bank lending has a substantial relationship component, the
costs of switching loan providers are higher than when changing providers of fee-based transac-
tions, which have a lower relationship component. Second, fee revenue is substantially reliant upon
staff costs to provide the required services, which generates a high fixed-cost component, as opposed
to margins which are more reliant upon interest expenditure as a variable cost input. Thus, fees have
a higher level of operating leverage. Third, fees have higher financial leverage due to lower levels
of required fixed assets, and so have higher financial risk. Overall, this literature has emphasised the
risk–return characteristics of bank fees, and therefore provides scope for considering the time series
relationship between margins and fees to determine if this relationship is stable through time.
The exposure of banks to increased fee revenue is of concern to a number of stakeholders in the
banking system. The conventional view of bank shareholders is that they can diversify away bank-
specific risk by their holding of a well-diversified portfolio. However, bank diversification
increases agency costs and income volatility (Laeven and Levine, 2007; Stiroh and Rumble, 2006).
As discussed by Froot et al. (1993) and Froot and Stein (1998), increased income volatility has a
nonlinear impact on bank cost of funds and makes risk management by the bank on behalf of the
shareholders worthwhile. From the perspective of bank management, their holding of a poorly
diversified wealth portfolio means that that they are concerned about bank total risk (Stulz, 1984).
From a borrower’s perspective, the implicit value of the bank–client relationship means that bor-
rowers face switching costs in the event of bank failure.5 Bank regulators, with a focus on main-
taining the viability of the financial system, are concerned about bank total risk due to the potential
risk of contagion and systemic failure resulting from the failure of a single (large) bank. Thus,
increased bank exposure to non-interest income has the potential (given the current evidence) to
increase the likelihood of systemic failure due to higher income volatility and agency conflicts as
well as worsening loan quality.
It is possible that management of large U.S. banks have ‘. . . gotten the diversification idea
wrong . . .’ (Stiroh and Rumble, 2006: 2158). This infers that bank management are more con-
cerned with increasing the level of returns rather than managing risk–return trade-offs. If this is
true, this would represent an agency conflict between regulators who are concerned with financial
system stability and bank management and bank shareholders who are concerned with profits.
Stiroh and Rumble (2006) suggest that the negative aspects of ‘too big to fail’ have encouraged this

agency problem in that bank management and shareholders profit from higher returns while regu-
lators bear any costs of bank failure due to higher risk, thus creating an asymmetry in the risk–
return trade-off that explains the U.S. evidence of overexposure to fee income.
Several reasons have been advanced to explain the increased move of banks into less traditional
activities. As discussed above, it could be that bank management are focussed on the level of returns
rather than risk and return. It is also possible that managerial non–profit-maximising activities are to
blame. Aggarwal and Samwick (2003) propose that managers choose to diversify their firms to
increase personal utility rather shareholder wealth. Mergers in the banking industry, resulting in
increased exposure to non-traditional activity, may also be motivated by managerial utility maximisa-
tion rather than shareholder wealth maximisation (Berger et al., 1999; Bliss and Rosen, 2001; Milbourn
et al., 1999). It is also possible that increased exposure to fees reflects the changing nature of the finan-
cial process in which markets are increasingly taking the place of traditional intermediation (Allen and
Santomero, 2001). In such a changing environment, banks are seeking new revenue lines to take the
place of declining interest margins, with a resulting change in bank risk. It is also possible that some of
the negative effects of increased exposure to fees are due to start up and initial learning costs. However,
given that the negative effects of increased bank exposure to fees have been documented in several
different national settings as well as from a number of empirical perspectives, it must be concluded that
the negative impact of increased bank fees are systematic and enduring. Exploring this issue from a
dynamic perspective, as this study does, will add a further dimension to this literature.
In the Australian context the main discussion of this issue has been provided by the annual
series of discussions by the Reserve Bank of Australia. These discussions do not provide any sta-
tistical testing of the relationship between bank net interest margins and fees and do not include in
their ambit any fees drawn from activities such as underwriting, funds management and insurance.
Instead these discussions focus only upon fee income resulting from the processes of taking depos-
its and writing loans. With these restrictions in mind, the Reserve Bank of Australia (2005) con-
cludes that increases in bank fees have not offset declines in bank margin income from traditional
banking activity. Given that Allen and Santomero (2001) find that there has been a switch in bank
activity from traditional intermediation (taking deposits and writing loans) toward fees in the
United States, a wider perspective is important given the policy issues raised by bank income vola-
tility. Thus, in this study a wider view of bank fee income will be taken, along with a longer time
series than the most recent survey by the Reserve Bank of Australia encompasses. When exploring
this time series relationship it would be expected that if a trade-off between margins and fees is
occurring, then a negative relationship between the two revenue sources would be expected over
time, with margins declining and fees increasing.
A study by Williams (2007) applied the Ho and Saunders (1981) model of bank net interest
margins to Australian data and found that margins have fallen over the study period 1989 to 2001,
with larger banks showing higher margins, but some evidence of decreasing returns to scale.6
Evidence was also found of banks buying market share and some mispricing of risk. It was also
found that foreign banks in Australia experienced significantly lower margins. The issue of a trade-
off between margins and fees was not explored by Williams (2007). It was concluded that the area
of fee income is still under-researched relative to their importance to bank revenue. Recently,
Williams and Prather (2009) considered the issue of bank revenue risk and return in Australia and
concluded that bank fees are riskier than margin income. However, it was also concluded that, at
the relatively low levels of fee income in Australia, portfolio diversification benefits were still
present.7 Overall, however, the time series relationship between bank margins and fees has not
been investigated previously that these authors are aware of and thus a dynamic approach to this
question will add to the growing literature considering bank revenue composition.

3. Data
The main data sources for this study are the individual bank annual reports. The sample covers
1988 to 2010, with a total of fifty-six banks in the sample. The banks are categorised into three
groups. The first group is the Big Four, which represent the four major banks in Australia, who
together account for over sixty-five to seventy-five percent of Australian banking assets over the
study period. The second group is the Other Domestic banks, mainly regional and state-owned
banks with a focus on retail finance, with the state-owned banks largely leaving the sample in the
mid-1990s as these institutions were privatised. This section of the Australian banking system has
experienced declining market share due to acquisitions by the Big Four banks. The final group is
the Foreign banks. These are generally smaller, more wholesale-oriented banks. A bank is catego-
rised as foreign if it has more than fifty percent foreign ownership.8 With only a few exceptions the
foreign banks are largely fully foreign owned operating as subsidiary banks. While foreign banks
can operate in Australia using a branch structure as an alternative to subsidiary operations, data
regarding foreign bank branches were not available for this study.9 There is a total of 24 Other
Domestic and 22 Foreign banks in the sample. Details of the sample are in Table 1 with descriptive
statistics in Table 2. For our regressions, margins will be measured by (interest received – interest
expense)/total assets as a percentage, while fees will be measured by non-interest income/total
assets as a percentage.10 Figure 1 shows the time series properties of both margins and fees as
Table 1. Sample composition.
All banks Big four Other domestic Foreign

All Years 458 89 205 164
1988 32 4 13 15
1989 33 4 14 15
1990 31 4 14 13
1991 29 4 12 13
1992 28 4 12 12
1993 28 4 13 11
1994 25 4 10 11
1995 24 4 11 9
1996 23 4 11 8
1997 19 4 8 7
1998 19 4 9 6
1999 18 4 7 7
2000 20 4 8 8
2001 19 4 8 7
2002 16 4 7 5
2003 14 4 7 3
2004 14 4 7 3
2005 12 4 7 1
2006 12 4 7 1
2007 13 4 7 2
2008 13 4 6 3
2009 12 4 5 3
2010 4 1 2 1
56 Banks in sample: 29 Other Domestic; 23 Foreign; 4 Big Four.

Table 2. Descriptive statistics.

All banks
Mean Standard Minimum Maximum Number of

deviation observations
Net interest margin/total assets 2.1779 1.2542 –0.16543 18.6761 458
(%)
Non-interest income to total 2.0182 2.2143 –0.76173 12.2089 453
assets (%)
Total assets $A, 000 50,431,196 109,367,830 29,278 656,799,000 492
Interest income $A, 000 3,254,157 6,269,223 14,576 38,385,000 459
Interest income as a share of 80.97 16.49 23.08 114.55 446
revenue (%)
1988
Net interest margin/total assets 2.3142 1.0379 0.6166 4.9634 32
(%)
Non-interest income to total 1.6654 1.1696 0.2387 5.7217 30
assets (%)
Total assets $A, 000 10,868,099 21,926,917 450,862.0 84,578,600.0 34
Interest income $A, 000 1,132,799.4 2,149,390.3 44,682.0 8,174,200.0 32
revenue (%)
1995
(%)
assets (%)
Total assets $A, 000 20,576,530 39,836,301 170,000 147,077,000.0 29
Interest income $A, 000 1,757,974.4 3,013,785.95 66,824.0 10,461,000.0 24
revenue (%)
2004
(%)
assets (%)
Total assets $A, 000 102,740,559 139,200,716 1,010,681 411,309,000.0 14
Interest income $A, 000 5,005,890 6,601,627 84,716.0 18,650,000.0 14
revenue (%)
2009
(%)
assets (%)
Total assets $A, 000 225,678,348 271,170,255 1,212,624 654,120,000 12
revenue (%)
(Continued)

Table 2. (Continued)
Big four
(%)
assets (%)
Total assets $A, 000 227,864,478 159,989,244 48,636,800 656,799,000 89
Interest Income $A, 000 13,644,383 7,919,530 4,902,800 39,385,000 89
revenue (%)
1988
(%)
assets (%)
Total assets $A, 000 67,665,450 15,290,310 48,636,800 84,578,600 4
Interest income $A, 000 6,522,000 1,337,377 48,636,800 84,578,600 4
revenue (%)
1995
(%)
assets (%)
Total assets $A, 000 116,273,500 21,209,908 99,595,000 147,077,000 4
revenue (%)
2004
(%)
assets (%)
Total assets $A, 000 305,432,000 75,225,776 245,079,000 411,309,000 4
revenue (%)
2009
(%)
assets (%)
Total assets $A, 000 585,266,500 76,846,855 476,987,000 654,120,000 4
revenue (%)

Other domestic

Net interest margin/total 2.3830 1.0665 0.1079 5.2613 205
assets (%)
assets (%)
Total assets $A, 000 17,375,175 28,488,395 229,778 152,294,000 217
revenue (%)
1988
assets (%)
assets (%)
Total assets $A, 000 4,536,563 5,768,639 450,862 20,055,882 15
Interest income $A, 000 505,656 584,627 51,913 1,993,514 13
revenue (%)
1995
assets (%)
assets (%)
Total assets $A, 000 6,971,665 5,609,032 1,675,7269 17,738,900 13
revenue (%)
2004
assets (%)
assets (%)
Total assets $A, 000 27,249,088 25,248,506 2,082,693 69,960,000 7
revenue (%)
2009
assets (%)
assets (%)
Total assets $A, 000 62,672,611 50,647,978 4,368,854 130,405,000 5
revenue (%)
(Continued)

Foreign

Net interest margin/total 1.7744 1.5891 –0.1654 18.6761 164
assets (%)
assets (%)
Total assets $A, 000 4,094,252 6,112,540 29,278 48,010,527 186
revenue (%)
1988
assets (%)
assets (%)
Total assets $A, 000 2,053,674 1,421,896 523,920 5,073,748 15
Interest income $A, 000 239,204 186,010 44,682 665,078 15
revenue (%)
1995
assets (%)
assets (%)
Total assets $A, 000 3,416,144 3,507,933 170,000 10,075,000 12
Interest income $A, 000 302,132.2 293,043.4 66,824.0 819,700.0 9
revenue (%)
2004
assets (%)
assets (%)
Total assets $A, 000 8,632,070 12,247,384 1,010,681 22,759,461 3
Interest income $A, 000 487,551 692,489 84,716 128,7161 3
revenue (%)
2009
assets (%)
assets (%)
Total assets $A, 000 19,903,707 25,616,009 1,212,624 47,397,197 3
revenue (%)

INTR/TR NIIN/TR INTR_BIG4/TR_BIG4 NIIN_BIG4/TR_BIG4
0.75 0.75
0.50 0.50
0.25 0.25
1990 1995 2000 2005 2010 1990 1995 2000 2005 2010
1.00 INTR_Foreign/TR_Foreign NIIN_Foreign/TR_Foreign INTR_ODM/TR_ODM NIIN_ODM/TR_ODM
0.75 0.75
0.50 0.50
0.25 0.25
1990 1995 2000 2005 2010 1990 1995 2000 2005 2010
Figure 1. Graph of the time series properties.

Note: INTR – Interest Received, NIIN – Non-interest Income and TR – Total Revenue.
percent of total revenue over the study period for all banks, the Big Four banks, the Other Domestic
banks and the Foreign banks.
4. Method
4.1. Panel VAR
In this section, we econometrically investigate the linkages between margins (MARGIN) and fees
(FEE) using panel data for 56 banks and 24 years, 1988 to 2010.
Unit root tests, described in Appendix 1, suggest that our data are stationary. Hence, we estimate
a panel vector autoregressive (VAR) model to analyse the relationships between MARGIN and
FEE. The multivariate VAR(q) model with fixed effects takes the form
 MARGIN i,t  q  β11 ( j) ( j )   MARGIN

β12 i ,t − j   η1,i   η1, BIG 4   η1, Foreign   ε1,i ,t 
  = ∑  +  + + +   , (1)
 FEE  j =1  β( j ) ( j)  
β22 FEE   η2,i   η2, BIG 4   η2, Foreign   ε2,i,t 
 i ,t   21  i , t − j         
where MARGIN and FEE are, respectively, Net Interest Margin/Total Assets (%) and Non-interest
Income to Total Assets (%) for banks i (= 1, . . ., N) at time t (= 1, . . ., T). ηi is a bank-specific fixed
effect and εi,t is a multivariate normally distributed random disturbance. Fixed effects for the Big
Four and Foreign banks are captured through ηBIG4 and ηForegin respectively. We estimate a fixed
effects model with bank-specific dummies, rather than a random effects model, as the ηi’s are likely

to represent omitted bank-specific characteristics which are correlated with other explanatory
variables.11
The system of equations described by equation (1) assumes that the random error terms are
orthogonal to the bank-specific fixed effects, as well as the lagged values of the endogenous vari-
ables. Further, the errors are assumed to have positive variance and to be uncorrelated across cross-
sectional units and time. However, due to the likely correlation between the lagged endogenous
variables and the fixed effects in equation (1), the least squares dummy variable estimator produces
biased parameter estimates (Nickell, 1981). Accordingly, we remove the fixed effects by differenc-
ing, i.e., equation (1) is rewritten as
( j )   ∆MARGIN
 ∆MARGIN i,t  q  β11 i ,t − j   ∆ε1,i ,t 
( j)
β12
  = ∑  ( j )  + , (2)

 ∆FEEi,t  j =1  β 21 β 22   ∆FEEi,t − j   ∆ε 2,i,t 
( j)  
where Δ is the first difference operator, e.g., ΔXi,t = Xi,t – Xi,t-1.

Since the transformed lagged endogenous variables and the transformed error terms in equation
(2) may be correlated in panels with a limited time dimension (see Kiviet, 1995; Nickell, 1981), we
estimate the coefficients in equation (2) by the generalised methods of moments (GMM) technique
proposed by Arellano and Bond (1991). This technique uses the pre-determined lags of the system
variables as instruments to exploit a potentially large set of over-identifying restrictions and pro-
vides consistent coefficient estimates (see Bond, 2002).
The errors in equation (2) satisfy the following orthogonality conditions:
E ( MARGIN i, s ∆εi,t ) = E ( FEEi, s ∆εi,t ) = 0, ∀s < t − 1 . (3)

Assuming serially uncorrelated errors, the orthogonality conditions imply that the vector of
instruments available to identify the parameters of equation (2) has the form
Zi,t = [ MARGIN i,t − 2 , . . . , MARGIN i,1; FEEi,t − 2 , . . . , FEEi,1 ] . (4)
Letting Zi* be a block diagonal matrix whose tth block is given by equation (4), for t = 1,. . . .,
T – 2, the matrix of instruments for each equation of the VAR is Z = (Z1*,. . . ZN*)ʹ.
The one-step GMM estimator for the k × 1 coefficient vector for each equation of the VAR in
equation (2) is given by12
( )
−1
β^ = X *′ ZAN Z ′X * X *′ ZAN Z ′Y , (5)
−1
1  N
where Y is an N(T – q – 1) × 1 vector of stacked dependent variables, AN =  ∑ Zi*′ HZi* 
N i =1  ,
with H a T – 2 square matrix with 2’s on the main diagonal, −1’s on the first sub-diagonal and 0’s
elsewhere, and X * is an N(T – q – 1) × k design matrix stacked by cross-sectional units with typi-
cal row
X i*,t = [ ∆MARGIN i,t −1 , . . . , ∆MARGIN i,t − q , ∆FEEi,t −1 , . . . , ∆FEEi,t − q ] . (6)
Finally, the asymptotic variance-covariance matrix of the GMM coefficient vector is given by

( ) ( )
−1 −1
a var (β ) = N X *′ ZAN Z ′X * X *′ ZAN VN AN Z ′X * X *′ ZAN Z ′X *
^ , (7)
N
where VN = N −1 ∑ Zi′ ∆εi ∆εi′ Zi and the ∆εi’s are the GMM residuals.
i
Despite the popularity of GMM estimators in dynamic panel regression studies, the error process
of the panel VAR does not necessarily follow a multivariate normal distribution especially in small
samples. The standard errors and the corresponding hypothesis testing could be misleading under
the normality assumptions. To overcome this problem, the empirical distribution is constructed by
drawing a bootstrap sample of 5000. An important advantage of the bootstrapping technique is that
the error process of the estimated panel VAR does not necessarily follow a multivariate normal
distribution and the critical values are obtained from appropriate percentile values of the empirical
distribution.
5. Estimation and results

In this exercise, the model described by (2) is estimated for four different groups: (i) Pool of all 56
banks, (ii) Big 4, (iii) Foreign Banks and (iv) all Other Domestic banks. Within each group the
model is estimated for five different sample periods, namely (a) the full sample period from 1988
to 2010, (b) sub-sample from 1988 to 1992, (c) sub-sample from 1993 to 2010, (d) sub-sample
1988 to 2002, and (e) sub-sample 2003 to 2010.
The first two sub-periods were chosen because, as discussed in Avkiran (1999), the early 1990s
saw a recession in the Australian economy, resulting in increased bad loans and a change in bank
strategy following this experience. The second two sub-periods were selected due to a change in
manner in which banks disclosed income from insurance activity. Prior to 2002 those Australian
banks with insurance operations disclosed this non-interest revenue on a net basis. However, after
2002 as the materiality of this revenue source increased, insurance income was disclosed on a gross
basis. Thus, the last two sub-periods will be employed to determine if any significant structural
breaks are apparent due to this change in reporting.13
The optimal lag length q is determined by nested likelihood ratio tests.14 In all cases, we find
that q = 1 is optimal.15 The GMM estimates for the panel VAR(1) are reported in Tables 3–6. For T
> 3 the model is over-identified and the validity of the assumptions used to estimate equation (2)
can be tested using the standard GMM test of over-identifying restrictions or a Sargan test. From
the Sargan test statistics and the p-values reported in Tables 3–6, the null hypothesis that the
moment conditions are valid (i.e., equation (3)) is unable to be rejected. In this context, the key
identifying assumption of no serial correlation in the εit disturbances can be examined by testing
for no second-order serial correlation in the first-differenced residuals (Arellano and Bond, 1991).16
The results generally show the absence of serial correlation and that the estimated models satisfy
the standard assumptions.17
Sensitivity analysis: To see how sensitive our results are to pooling the banks by group (i.e., Big
Four, Foreign, and Other Domestic banks), we first delete each bank one at a time and compare the
resulting model with the results reported in Tables 4–6. Overall, this procedure did not affect any
of the signs of the coefficients reported in the tables. The experiment is conducted for each group
for the full sample period. However, the sensitivity analysis is not conducted for the sub-sample
periods within each group, as the deletion of banks will result in a lack of degrees of freedom.
Unsurprisingly, however, due to the smaller sample sizes, the regression results are somewhat
sensitive to exclusion of few banks in the Foreign and Other Domestic bank groups. Depending on

Table 3. All banks.
Full sample Sub-sample 1 Sub-sample 2

1988 to 1992 1993 to 2010
∆MARGINt ∆FEEt ∆MARGINt ∆FEEt ∆MARGINt ∆FEEt
∆MARGINt–1 0.587*** –0.31** 0.084 0.157 0.62*** –0.54**
(0.17) (0.14) (0.22) (0.56) (0.14) (0.26)
∆FEEt–1 –0.09** 0.65*** –0.108*** 0.055 –0.07 0.39***
(0.04) (0.12) (0.02) (0.14) (0.07) (0.09)
Constant –0.33* –0.05 –0.09** 0.022 0.18 –0.06
(0.18) (0.17) (0.04) (0.08) (0.16) (0.11)
Diagnostics
Sargan 13.23 23.89 26.0 28.6 13.79 15.83
statistic
Sargan 1.00 1.00 0.11 0.16 100 1.00
p-value
AR(1) –1.82* –1.73* –1.64* –1.67* –2.04** –1.66*
AR(2) –1.43 0.47 –0.78 –0.10 –0.97 1.55
Notes: (a) Standard errors are in the parentheses. ***, ** and * denote rejection of null of zero restriction at 1%, 5% and
10% levels of significance (based on bootstrapping), respectively.
(b) The Sargan statistic tests over-identifying restrictions (based on bootstrap samples). AR(1) and AR(2) are tests for
first-order and second-order serial correlation (based on bootstrap samples).
Table 4. Big4 banks

1988 to 1992 1993 to 2010
∆MARGINt–1 0.73*** –0.57** –0.047 –0.303*** 0.56*** –0.67*

(0.06) (0.22) (0.09) (0.08) (0.06) (0.38)
∆FEEt–1 –0.06** 0.33*** –0.121** 0.028 –0.04* 0.29**
(0.03) (0.11) (0.06) (0.165) (0.02) (0.12)
Constant –0.41*** 0.12 –0.065* 0.051*** –0.12*** –0.17**
(0.11) (0.16) (0.03) (0.01) (0.04) (0.07)
Diagnostics
Sargan 54.07 60.01 12.7 10.31 38.55 43.41
statistic
Sargan 1.00 1.00 0.81 0.92 1.00 1.00
p-value
AR(1) –1.88* –1.89* –1.67* –1.77* –1.83* –1.83**
AR(2) –0.84 1.45 –0.39 –0.73 –0.81 –0.37
the bank, the coefficients of some variables became statistically significant, while others became
statistically insignificant. It is important to note that there were no sign reversals. These results are
summarised in Appendix 1 Table 11.

Table 5. Foreign banks.

1988 to 1992 1993 to 2010
∆MARGINt–1 0.15* –0.32 –0.091 0.486** 0.44** –0.15

(0.07) (0.69) (0.06) (0.21) (0.21) (0.38)
∆FEEt–1 –0.24** 0.36 –0.035 –0.131 –0.22 0.35***
(0.07) (1.14) (0.03) (0.10) (0.23) (0.11)
Constant –0.24 –0.16 –0.172** 0.166 0.21 –0.03
(0.22) (0.37) (0.07) (0.15) (0.24) (0.37)
Diagnostics
Sargan 130.3 64.72 24.97 39.77 57.5 49.6
statistic
Sargan p- 1.00 1.00 0.13 0.21 1.00 1.00
value
AR(1) –1.94* –2.49** –1.71* –1.66* –1.73* –1.92*
AR(2) –0.65 0.26 –1.42 –0.67 –1.03 –0.29
10% levels of significance (based on bootstrapping), respectively;
Table 6. Other domestic banks.

1988 to 1992 1993 to 2010
∆MARGINt–1 0.489*** –0.49** 0.098 0.653 0.52*** –0.81**

(0.10) (0.19) (0.21) (0.52) (0.09) (0.36)
∆FEEt–1 –0.08** 0.38*** –0.214*** 0.402** –0.08** 0.39***
(0.03) (0.13) (0.07) (0.17) (0.04) (0.11)
Constant –0.49* 0.01 –0.061 –0.106 0.30 –0.25
(0.29) (0.43) (0.07) (0.07) (0.26) (0.18)
Diagnostics
Sargan 71.13 141.4 20.79 42.27 72.96 88.5
statistic
Sargan 1.00 1.00 0.29 0.30 1.00 1.00
p-value
AR(1) –2.66*** –1.99** –1.93* –1.64* –2.18** –1.69*
AR(2) 0.92 0.90 –0.01 –1.25 –1.19 –1.02
To illustrate how to interpret Appendix 1 Table 11, consider deleting the Colonial Bank (COL)
from the sample. Doing so makes the coefficient for FEE become statistically insignificant in the
MARGIN equation. In addition, the coefficient for FEE becomes statistically significant in the

Table 7. All banks.

1988 to 2002 2003 to 2010
∆MARGINt–1 0.59*** –0.31** 0.59** 0.18* 0.59** –1.25**

(0.17) (0.14) (0.23) (0.10) (0.11) (0.52)
∆FEEt–1 –0.09** 0.65*** –0.20** –0.08 0.09 0.13***
(0.04) (0.12) (0.08) (0.17) (0.06) (0.01)
Constant –0.33* –0.05 –0.43** –0.06 0.01 0.39
(0.18) (0.17) (0.20) (0.16) (0.06) (0.27)
Diagnostics
Sargan 13.23 23.89 45.51 205.1 47.73 37.8
statistic
Sargan 1.00 1.00 1.00 0.54 1.00 0.58
p-value
AR(1) –1.82* –1.73* –2.17** –1.87* –2.34** –1.64*
AR(2) –1.43 0.47 –1.14 –1.03 –1.34 1.34
Table 8. Big4 banks.

1988 to 2002 2003 to 2010
∆MARGINt–1 0.73*** –0.57** 0.63*** –0.35* 0.01** –4.63*

(0.06) (0.22) (0.03) (0.19) (0.006) (2.56)
∆FEEt–1 –0.06** 0.33*** –0.30*** 0.23 –0.04 0.08
(0.03) (0.11) (0.03) (0.32) (0.05) (0.23)
Constant –0.41*** 0.12 –0.41*** 0.05 0.04 –0.15
(0.11) (0.16) (0.08) (0.12) (0.04) (0.16)
Diagnostics
Sargan 54.07 60.01 48.51 47.42 13.85 12.22
statistic
Sargan 1.00 1.00 1.00 1.00 1.00 1.00
p-value
AR(1) –1.88* –1.89* –1.84* –1.89* –1.87* –1.91*
AR(2) –0.84 1.45 –1.12 1.01 –1.61 0.72
10% levels of significance (based on bootstrapping), respectively;
FEE equation. Overall, what is immediately obvious from Appendix 1 Table 11 is that our most
robust results are for those reported in Tables 3–6 and 7–10. That is, the conclusions that were
drawn about the relationships between MARGIN and FEE are the most defensible.

Table 9. Foreign banksa.
Full sample Sub-sample 1

1988 to 2002
∆MARGINt ∆FEEt ∆MARGINt ∆FEEt
∆MARGINt–1 0.15* –0.32 0.21** –0.11

(0.07) (0.69) (0.09) (0.22)
∆FEEt–1 –0.24** 0.36 –0.28*** 0.13
(0.07) (1.14) (0.05) (0.16)
Constant –0.24 –0.16 –0.34 –0.05
(0.22) (0.37) (0.24) (0.23)
Diagnostics
Sargan statistic 130.3 64.72 124.5 84.89
Sargan p-value 1.00 1.00 1.00 1.00
AR(1) –1.94* –2.49** –2.04** –1.73*
AR(2) –0.65 0.26 –1.41 –0.27
aThe reduction in sample size for the foreign bank meant that robust estimates of the standard errors of the estimates
were not possible for the 2003 to 2010 sub-period and so these results were not shown.
Table 10. Other Domestic Banks.

1988 to 2002 2003 to 2010
∆MARGINt–1 0.489*** –0.49** 0.32** –0.04 0.51*** –1.16*

(0.10) (0.19) (0.13) (0.20) (0.12) (0.65)
∆FEEt–1 –0.08** 0.38*** –0.10** 0.22* 0.003 0.13
(0.03) (0.13) (0.04) (0.14) (0.03) (0.13)
Constant –0.49* 0.01 –0.48* –0.09 –0.07 0.56
(0.29) (0.43) (0.27) (0.36) (0.08) (0.42)
Diagnostics
Sargan statistic 71.13 141.4 86.74 150.6 26.79 25.57
Sargan p-value 1.00 1.00 1.00 0.99 0.95 0.96
AR(1) –2.66*** –1.99** –2.59** –1.95* –1.85* –1.99**
AR(2) 0.92 0.90 –0.38 –1.01 –1.58 –1.33
Overall, it is found that changes in margins are offset by changes by in fees with each showing
offsetting time series properties.18 Consistent with increased emphasis on fee income based revenue
sources (Allen and Santomero, 2001; Lepetit et al., 2008b; Stiroh, 2004), this relationship is strong-
est in the second sub-period, post 1992. Thus banks in Australia reacted to reductions in margins by

increasing fees in the following period, and this relationship is fairly robust over time.19 However,
the coefficient of the increase in fees was generally smaller than the reduction in margins.
The economic significance of these changes will be considered from the perspective of the coef-
ficients estimated using the full sample. As the model is estimated using changes, the coefficient
drawn from the regression of ΔMARGIN t on ΔMARGIN t−1 is the estimated elasticity. Using the
average value of MARGIN as well as the average value of total assets implies an annual average
reduction in margin revenue for each bank of approximately AU$644 million per annum. The cor-
responding increase in fee revenue can be decomposed into two sources. First, there is the increase
in fees that is a reaction to reduction in margin revenue; this is represented by the results of the
regression of ΔFEE t on ΔMARGIN t−1, which produces an estimated elasticity of 0.33453, imply-
ing an increase in bank fee revenue per annum of approximately AU$368 million per annum per
bank. Second, there is the time series pattern of increases in fees, which has an annual value of
approximately AU$661 million. This increase in fee income is marginally larger than the overall
decline in margin income, meaning that overall the banking system has slightly increased its rev-
enue base as a result of changing revenue composition. However, this increase should be borne in
mind as it occurred against a background of increasing bank assets. In net terms the increase in
bank revenue is approximately AU$17 million per annum; using the average assets figure for the
whole period in Table 1, this implies an increase in return on assets of 0.0000003%. If the 2009
average asset figure is used the increase in return on assets is far smaller. Further, the increase in
bank fees that this study can directly attribute to falling bank margins is slightly more than half
(fifty-seven percent) of the decline in margin revenue.
The other part of the observed increase in bank fees is due to banks increasing their range of
product offerings into new financial services. As will be argued below, when the results of the
second sub-period analysis are considered, it is most likely that this portion of increased fee income
is largely attributable to increased insurance revenue occurring after 2002. Thus, this paper will
argue that the changing pattern of bank revenue has generated a net welfare transfer in favour of
consumers of bank products, but with not all consumers benefitting or benefitting equally from this
transfer. In response to this observed decline in revenue, banks have responded by diversifying
their portfolio of product offerings to new areas of financial services (particularly insurance),
which has marginally offset (in total) the decline in margin income, but the impact of these changes
in terms of return on assets is extremely small.
An interesting result is that the first set of sub-period analysis finds the coefficient on fees is negative
and significant in the first sub-period, while the margin coefficient is insignificant. This indicates that
Australian banks were being pro-active in the process of disintermediation before the move away from
traditional (margin) income was resulting in a statistically significant decline in margin revenue. Thus,
it seems that the Australian banks were pre-positioning their revenue streams to deal with the negative
impact of reduced margin income by become active participants in the market-based processes that
resulted in increased fees. These processes can be considered market based in that (i) banks became
active in the process of using market products to replace traditional intermediation, and (ii) fees and
charges on bank products became increasingly focussed upon earning a market-based rate of return.
As discussed above, the bank revenue and revenue reporting saw a distinct change after 2003,
with this change being most apparent in the area of insurance revenue. Accordingly we conducted
a second set of sub-period analysis to determine what impact (if any) this change had on our esti-
mated model. These results are in Tables 7–10. What is immediately noticeable is that the coeffi-
cients of ΔFEE are far larger in the 2003 to 2010 sub-period that in any of the proceeding analysis,20
with the change particularly noticeable for the Big Four banks, but also quite apparent for the Other
Domestic banks. Based on this evidence this paper will argue that the Australian banks responded

to reductions in revenue due to falling margins by increasing their offerings of financial services,
particularly insurance. As discussed above, it is this increased revenue from offering a wider range
of financial products that has meant Australian banks have been able to offset the impact of falling
margin income from providing traditional banking products.
The results for the Big Four banks were largely the same as for the All Banks sample, with the
exception that the trade-off between margins and fees seemed to begin a little earlier in the first
sub-sample period, 1988 to 1992. For the Big Four banks the evidence that the larger banks were
pro-active in increasing fees prior to a statistically significant reduction in margin income is also
present, again indicating that the large Australian banks were pro-active in this change in revenue
composition.
For the Foreign banks, the lagged relationship over time between fees and margins income is
weaker, no doubt due to the foreign bank’s higher level of emphasis on fee income rather than
margin income. In the case of the foreign banks, no evidence was found of a systematic lagged
relationship over time between fees and margins. Instead the time series process of the two income
streams seem to be entirely separate. This lack of relationship reflects the foreign banks’ lower
emphasis upon the provision of intermediation services as part of their strategic mix.21 Within the
context of this study, the foreign banks can be considered as a control sample to provide a contrast
to the trends exhibited by the other two bank types which are more active in traditional bank
transactions.22
In the case of the Other Domestic banks, fees are again found to move in the opposite direction
to margins. However, evidence of a time trend in margins is weaker. In both sub-periods the evi-
dence supports the argument that fee income is moving in the opposite direction to margins. Again
there is little evidence of a time trend in net interest margins. There is, however, evidence of a time
trend in fees in the first sub-period. This evidence supports the argument that increases in fees
tended to pre-date reduction in net interest income, and that Australian banks were pro-active in
responding to the threat of disintermediation by pre-positioning their revenue streams to compen-
sate for potential reduction in traditional (margin) income.
As a further sensitivity analysis, the impact of removing each bank in turn from the sub-samples
was analysed. These results are shown in Appendix 1 Table 11. In the case of the Big Four banks,
only the constant was found to be sensitive to outlier banks. In three of four cases, the removal of
one of the Big Four banks resulted in the constant becoming statistically insignificant. In the case
of the Foreign banks, three banks were found to impact upon the coefficients for the margin equa-
tion, with the removal of each bank resulting in the coefficient becoming statistically insignificant.
All three banks had experienced a significant restructure during the sample period. In the case of
the FEE equation, the removal of three banks resulted in the estimated coefficient becoming statis-
tically significant. Again each of these banks experienced a significant restructure during the sam-
ple period. It is noteworthy that Standard Chartered Bank acted as an outlier bank for both the
margin and fee equations. In the case of the Other Domestic banks, again three banks acted as
outlier banks. For the FEE equation removal of these outliers resulted in the estimated coefficient
becoming insignificant.
6. Conclusions and directions for further research

Overall, this study finds that there is a systematic relationship between decreases in Australian
bank interest margins and increases in fee income. Further, increases in fees pre-date reductions in
margin income. This suggests that Australian banks pre-positioned their revenue streams for the
impact of falling margin income by increasing their fee income. It is also found that the increases

in fees associated with falling bank margins are of a smaller magnitude than the decreases in mar-
gins. This indicates that consumers of traditional banking products as a whole have been made
better off by this process, thus producing a wealth transfer in favour of bank customers overall.
However, increases in bank revenue resulting from increased fees are of marginally larger magni-
tude than declining margin income.
This paper argues that the increases in fees observed in this study reflect a strategic effort by
Australian banks to counteract the impact of falling interest margins by diversifying their income
sources away from traditional intermediation toward other revenue sources such as funds manage-
ment and, in particular, insurance. The evidence of this study is that this process began before falls
in margin income were statistically significant; suggesting that banks actively re-focussed their
revenue streams toward fees prior to falling margin income becoming statistically significant. This
would reflect the changing nature of the financial system as discussed by Allen and Santomero
(2001). This is exemplified by the growing trend toward larger banks providing one-stop financial
services.
It is possible (but less likely) that the increases in fees observed in this study are directly due to
increases in fees and charges being levied as part of intermediation services. This is the trend
observed by the Reserve Bank of Australia (2006), although with the caveat that the Reserve Bank
of Australia (2006) used a narrower definition of fees than applied in this study and so the two
studies may not be reflective of the same effects. If this is the case, then Australian banks are pre-
empting decreases in margin income by increasing the fees charged to users of bank products and
approximately (but not exactly) the same consumer groups are paying for decreases in margins via
higher fees and changes. However, depending on their patterns of bank service usage there may
remain some consumer groups that are worse off at the margins from this process, particularly
those bank consumers making frequent use of deposit services but do not have loan accounts which
benefit from the effects of lower interest margins. However, the most likely interpretation of the
results in this study is that banks have increased fee income via increased insurance revenue and as
a result consumers of bank products are better off than previously. In order to determine in more
detail where the wealth transfers have occurred as a result of the processes observed in this study,
further research will be necessary, employing more detailed data than available for the sample
employed in this study.
This trend of increasing bank fees raises different agency concerns to the different stakeholders
in the banking system. It is conventionally assumed that bank shareholders are able to diversify
away the impact of idiosyncratic risk. However, as discussed by Froot and Stein (1998), increased
income volatility resulting from increased exposure to fees can have a nonlinear impact on bank
cost of funds and result in increased reliance upon internal bank hedging on behalf of shareholders.
Furthermore, the recent global financial crisis has demonstrated the importance of systemic risk in
banking for the economy as a whole.
Borrowers of banks bear increased agency costs in times of bank failure resulting in the loss of
the implicit value of their bank–client relationship (Stiroh and Rumble, 2006). As discussed by
Stulz (1984), bank management hold poorly diversified wealth portfolios and so are concerned
about bank total risk. Further, increased bank fees results in increased returns but worsening risk–
return trade-offs, as well as (in Europe) worsening loan quality. Prudential regulators are concerned
about financial system stability, and increased fee income results in increased bank systematic risk
(De Jonghe et al., 2007; Stiroh, 2006b). With bank management potentially more concerned about
the level of returns rather than risk and return, the negative aspects of ‘too big to fail’ increase the
risk to regulators of bank failure due to increased income volatility resulting from higher levels of
fees. The outcome for regulators is an increased exposure to systemic risk. Further, as discussed by

Laeven and Levine (2007), the increased firm complexity resulting from higher levels of fee
income also increases the scope for agency problems. The evidence presented in this study finds
that banks in Australia are increasing their level of fees while at the same time margin income is
declining. Thus the agency costs resulting from this trend are more likely to increase rather than
decrease and all bank stakeholders should be aware of the resulting changes in bank risk character-
istics. These changes in bank portfolio composition and bank risk also offer a number of potentially
fruitful avenues for future bank research.
Acknowledgments
The authors are grateful for comments from Tom Smith, Andy Mullineux, Katrina Ellis and Bruce Arnold as
well as seminar participants at the AFAANZ conference, Gold Coast 2007; the Australasian Finance and
Banking Conference, 2007; the Swiss Society for Financial Market Research, Zurich, 2008; Midwest Finance
Association Meeting, San Antonio, Texas, 2008; Asian Finance Association Annual Conference, Brisbane,
June 2009; Finance and Corporate Governance Conference, Melbourne, April 2011, KOF (Swiss Economic
Institute) at ETH (Swiss Federal Institute of Technology), Zurich and the Australian Prudential Regulation
Authority. Excellent research assistance has been provided by Phillippa Wright. All errors remaining are the
responsibility of the authors.
Funding
The authors are grateful for financial support from the Bond University Vice Chancellors research fund.
Notes
1. In the context of this study, fee income is used to refer to all sources of revenue other than interest
revenue. Thus fee income encompasses revenue sourced from fees upon retail and wholesale accounts,
investment banking revenue, investment and trading income, fees from funds under management, as well
as insurance income and other financial services. This is a far wider definition than used in the annual
surveys by the Reserve Bank of Australia.
2. See DeYoung and Rice (2004), the Reserve Bank of Australia (2004) and the Reserve Bank of Australia
(2005).
3. See Stiroh and Rumble (2006) and Goddard et al. (2009).
4. See also Laeven and Levine (2007) and Lepetit et al. (2008a).
5. Lepetit et al. (2008a) argue that these switching costs mean that banks will accept lower returns on their
loan portfolio in order to attract such a long-run relationship with the borrowing client. The cost of the
lower loan returns will be compensated for by increased non-interest income over the life of the loan
relationship. This relationship is found to result in loan underpricing and higher default risk.
6. This study also employed a hand-collected data set drawn from individual bank annual reports.
7. It should be noted, however, that the data employed by Williams and Prather (2009) did impose some
restrictions upon their ability to explore all aspects of the risk–return trade-off from increased fees.
8. Inclusion of the Australian foreign bank data in this study is a unique aspect of this study as compared to
other data sources.
9. This choice of converting to branch status partially accounts for the declining number of foreign banks
over the study period. The other factor accounting for this decline has been mergers between the foreign
bank parents.
10. A more detailed breakdown of non-interest income into its components due to intermediation services
(e.g., fees on deposit and loan accounts) and other non-interest income sources is not consistently avail-
able from the data source used in this study.

11. Under an assumption of independence between the fixed effects and the explanatory variables, the gen-
eralised least squares estimator for the random effects model is biased (Hsaio, 1986).
12. Bond (2002) notes that most applied work using GMM employs the one-step estimator rather than the
two-step estimator. Arrellano and Bond (1991) use simulations to show that only modest efficiency gains
are achieved by using the two-step procedure, even in the presence of heteroscedasticity.
13. It was not possible to reconstruct pre-2002 insurance data to a consistent basis with post-2002 data. The
efforts of Phillippa Wright in this data collection process were extremely valuable.
14. The correct lag length is critical for panel VARs since excessively short lags may fail to capture the
system’s dynamics, lead to omitted variables, bias the remaining coefficients, and be likely to produce
serially correlated errors. On the other hand, too many lags lead to a rapid loss of degrees of freedom and
to over-parameterisation. See Holtz-Eakin et al. (1988) for more detail about the nested likelihood test
for lag length selection.
15. To evaluate the sensitivity of the results in higher order lags, the model also estimated for the lag length
of 2. The general results are the same as that of lag 1. For the sake of brevity, the results are not reported
for lag 2 and can be made available from authors upon request.
–
16. Negative first-order serial correlation is expected in the first-differenced residuals if t is serially
uncorrelated.
17. In addition, note that the asymptotic standard errors reported in Tables 3–6 are robust to heteroscedastic-
ity (see Arellano and Bond, 1991). However, as a check we regress the squared residuals on the inde-
pendent variables for each of the estimated models. Wald tests indicate homoscedastic errors. For the
sake of brevity, these test results are not presented here. (The results are available from the authors upon
request.)
18. As we had relatively few workable observations for 2010, as shown in Table 1, we conducted sensitivity
analysis to determine if the exclusion and or inclusion of the 2010 observations had any impact on our
results. No results were found to be sensitive to the exclusion or inclusion of the 2010 data, including the
sub-period analysis. The only impact was that some coefficients changed at the level of the third decimal.
No tests of significance were affected.
19. We have also computed impulse response functions using Cholesky orthogonal decomposition. The
results are consistent with the findings from Tables 3–6.
20. Due to the decline sample size for the Foreign banks shown in Table 1, it was not possible to obtain
robust estimates for the standard errors of the estimates for the Foreign banks in the 2003 to 2010 sub-
period, thus these results are not shown in Table 10.
21. See for example Williams (1998) and Pauly (1987).
22. As previously noted, the reduction in sample size for the Foreign banks meant that robust estimates of
the standard errors of the estimates are not possible for the 2003 to 2010 sub-period and so these results
are not shown.
23. Unlike the augmented Dickey–Fuller (ADF) test, the power of the IPS panel unit root test increases with
the number of panels.
24. The general results do not change when we use other criteria, e.g., the Schwartz criterion.
25. The results on serial correlations across the panels failed to detect the presence of such serial correlation.
These results can be made available from the authors on request.
26. Binder et al. (2003) show that GMM estimation of panel VARs based on orthogonality conditions
break down if the underlying time series contains unit roots. However, as a check we re-estimated
the model in ECM-form under the assumption of non-stationarity (i.e., even though all the vari-
ables are stationary). The ECM terms for both equations were insignificant. Eliminating the insig-
nificant ECM terms lead to the restricted model in differenced form; which is what we pursue and
report here.

References
Aggarwal R and Samwick A (2003) Why do managers diversify their firms? Agency reconsidered. The
Journal of Finance LVIII: 71–118.
Allen F and Santomero AM (2001) What do financial intermediaries do? Journal of Banking and Finance
25: 271–294.
Arrellano M and Bond S (1991) Some tests of specifications for panel data: Monte Carlo evidence and an
application to employment equations. Review of Economic Studies 58: 277–297.
Avkiran NK (1999) The evidence on efficiency gains: The role of mergers and the benefits to the public.
Journal of Banking and Finance 23: 991–1013.
Berger AN, Demsetz RS and Strahan PE (1999) The consolidation of the financial services industry: Causes,
consequences, and implications for the future. Journal of Banking and Finance 23: 135–194.
Binder M, Hsiao C and Pesaran M (2003) Estimation and inference in short panel vector autoregressions with
unit roots and cointegration. Cambridge Working Papers in Econometrics. Cambridge, UK: Cambridge
University Press.
Bliss R and Rosen R (2001) CEO compensation and bank mergers. Journal of Financial Economics 61:
107–138.
Bond S (2002) Dynamic panel data models: A guide to microdata methods and practice. Working Paper
CWP09/02, Centre for Mirodata Methods and Practice, UK.
De Jonghe O, Baele L and Vennet RV (2007) Does the stock market value bank diversification? Journal of
Banking and Finance 31: 1999–2023.
DeYoung R and Rice T (2004) Non-interest income and financial performance at U.S. commercial banks. The
Financial Review 39: 101–127.
DeYoung R and Roland K (2001) Product mix and earnings volatility of commercial banks: Evidence from a
degree of total leverage model. Journal of Financial Intermediation 10: 54–84.
Froot K and Stein J (1998) Risk management, capital budgeting and capital structure policy for financial
institutions: An integrated approach. Journal of Financial Economics 47: 55–82.
Froot K, Scharfstein D and Stein J (1993) Risk management: Coordinating corporate investment and financ-
ing policies. Journal of Finance 48: 1629–1659.
Goddard J, McKillop D and Wilson JOS (2009) The diversification and financial performance of US credit
unions. Journal of Banking and Finance 32: 1836–1849.
Ho T and Saunders A (1981) The determinants of bank interest margins: Theory and empirical evidence.
Journal of Financial and Quantitative Analysis 16: 581–602.
Holtz-Eakin D, Newey W and Rosen H (1988) Estimating vector autoregressions with panel data.
Econometrica 56: 1371–1395.
Hsaio Cheng, (1986) Analysis of Panel Data. Cambridge: Cambridge University Press.
Im K, Pesaran M and Shin Y (2003) Testing for unit roots in heterogenous panels. Journal of Econometrics
115: 53–74.
Kiviet J (1995) On bias, inconsistency, and inefficiency of various estimators in panel data models. Journal
of Econometrics 68: 53–78.
Laeven L and Levine R (2007) Is there a diversification discount in financial conglomerates? Journal of
Financial Economics 85: 331–367.
Lepetit L, Nys E, Rous P, et al. (2008a) The expansion of services in European banking: Implications for loan
pricing and interest margins. Journal of Banking and Finance 32: 2325–2335.
Lepetit L, Nys E, Rous P, et al. (2008b) Bank income structure and risk: An empirical analysis of European
banks. Journal of Banking and Finance 32: 1452–1467.
Milbourn T, Boot A and Thakor A (1999) Megamergers and expanded scope: Theories of bank size and activ-
ity diversity. Journal of Banking and Finance 23: 195–214.
Nickell S (1981) Biases in dynamic models with fixed effects. Econometrics 49: 1417–1426.
Pauly L (1987) Foreign Banks in Australia: The Politics of Deregulation. Sydney, NSW, Australia: Australian
Professional Publications.
Reserve Bank of Australia (2004) Banking fees in Australia. Reserve Bank of Australia Bulletin, May, 57–61.

Schmid M and Walter I (2009) Do financial conglomerates create or destroy economic value? Journal of
Financial Intermediation 18: 193–216.
Slager A (2006) The Internationalization of Banks. Basingstoke: Palgrave Macmillan.
Smith R, Staikouras C and Wood G (2003) Non-Interest Income and Total Income Stability. Bank of England
Working Paper No. 198, Cass Business School Research Paper. Available at: http://dx.doi.org/10.2139/
ssrn.530687.
Stiroh K (2004) Diversification in banking: Is non-interest income the answer? Journal of Money, Credit and
Banking 36: 853–882.
Stiroh K (2006a) New evidence on the determinants of bank risk. Journal of Financial Services Research 30:
237–263.
Stiroh K (2006b) A portfolio view of banking with interest and non interest income. Journal of Money, Credit,
and Banking 28: 1351–1361.
Stiroh K and Rumble A (2006) The dark side of diversification: The case of US financial holding companies.
Journal of Banking and Finance 30: 2131–2161.
Stulz R (1984) Optimal hedging policies. Journal of Financial and Quantitative Analysis 19: 127–140.
Williams B (1998) Factors affecting the performance of foreign-owned banks in Australia: A cross-sectional
study. Journal of Banking and Finance 22: 197–219.
Williams B (2007) Factors determining net interest margins in Australia: Domestic and foreign banks.
Financial Markets, Institutions & Instruments 16: 145–165.
Williams B and Prather L (2009) Bank risk and return: The impact of bank non-interest income. International
Journal of Managerial Finance 6: 220–244.
Appendix 1
Descriptive statistics
As with pure time series regression analysis, the asymptotic distributions of the estimators in panel
regressions are likely to be affected by the presence of unit roots. This is especially likely in data-
sets with relatively long time series and short cross-sectional dimensions. Hence, as precursor to
our econometric analysis, we examine the stationarity of our data. The variables we use for our
econometric analysis are described in detail in Table 11.
We test for unit roots using the test proposed by Im et al. (2003) (henceforth, IPS).23 Using obvi-
ous notation, the heterogeneous panel data model is given by
qi
∆yit = µi + βi yit −1 + ∑ φ j ∆yit − j + γ i t + εit , i = 1, 2, . . . , N ; t = 1, 2, . . . , T . (A.1)
j =1
The null hypothesis to test for unit roots is given by H0 : βi = 0 for all i; H1 : βi < 0 for some i.
Based on equation (A.1), each individual component of the panel is estimated separately by OLS
and then the test statistics are obtained as Studentised averages of the test statistics for each equa-
tion. The number of lags, qi, in the model is determined by the Akaike Information Criteria.24
The t–-statistic proposed by IPS is defined as the average of the individual ADF τ-statistics, i.e.,
β˘ i
N
1
t =
N
∑ τi , where τi = . (A.2)
i =1 σ˘ βi
The critical values for the t–-statistic are obtained by stochastic simulation using 100,000
replications.

The IPS panel unit root test results are reported in Appendix 1 Table 12. The statistics suggest
that the data are stationary.25 Hence, we estimate a panel vector autoregressive (VAR) model to
analyse the relationships between MARGIN and FEE in preference to a panel vector error correc-
tion model, which requires that variables are non-stationary and cointegrated.26
Table 11. Sensitivity analysis.
BIG4 Foreign Other domestic

∆MARGINt–1 LLO*, NAW*, STC* BOA#, CIT# MEL*

∆FEEt–1 STC*, CIT* STC# COL*, MEL* COL#
Constant ANZ*, COM*, WES*
Notes: # denotes that coefficient became statistically significant (based on bootstrap samples) when one of the banks
listed was deleted from the sample.
* denotes that coefficient became statistically insignificant when one of the banks listed was deleted from the sample.
ANZ = Australian and New Zealand Bank
COM = Commonwealth Bank
WES = Westpac
LLO = Lloyds Bank
NAW = NatWest
STC = Standard Chartered
BOA = Bank of America
CIT = Citibank
BOQ = Bank of Queensland
MEL = Bank of Melbourne
COL = Colonial Bank
Table 12. IPS panel unit root test results.

–
Variable t -statistic
Net interest margin/total assets (%) –3.59***
Non-interest income to total assets (%) –3.67***
Notes:
(a) ****, ** and * represent rejection of the unit root hypothesis at the 1%, 5% and 10% significance level, respectively;
(b) All variables are de-trended prior to the unit root test for consistency, i.e., the IPS panel unit root test is based on
the model with an intercept;
(c) Critical values are obtained from stochastic simulations with 100,000 replications.
Date of acceptance of final transcript: 1 February 2012.

Accepted by Associate Editor Garry Twite (Finance).

Barry Williams

Uploaded by

Copyright:

Available Formats

You might also like

Barry Williams

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Barry Williams

Uploaded by

Copyright:

Available Formats

440268

Australian Journal of Management

Barry Williams and Gulasekaran Rajaguru

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 1. Sample composition.

All banks Big four Other domestic Foreign

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 2. Descriptive statistics.

Mean Standard Minimum Maximum Number of

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Mean Standard Minimum Maximum Number of

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Mean Standard Minimum Maximum Number of

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

INTR/TR NIIN/TR INTR_BIG4/TR_BIG4 NIIN_BIG4/TR_BIG4

Figure 1. Graph of the time series properties.

 MARGIN i,t  q  β11 ( j) ( j )   MARGIN

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

where Δ is the first difference operator, e.g., ΔXi,t = Xi,t – Xi,t-1.

E ( MARGIN i, s ∆εi,t ) = E ( FEEi, s ∆εi,t ) = 0, ∀s < t − 1 . (3)

Zi,t = [ MARGIN i,t − 2 , . . . , MARGIN i,1; FEEi,t − 2 , . . . , FEEi,1 ] . (4)

X i*,t = [ ∆MARGIN i,t −1 , . . . , ∆MARGIN i,t − q , ∆FEEi,t −1 , . . . , ∆FEEi,t − q ] . (6)

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

5. Estimation and results

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 3. All banks.

Full sample Sub-sample 1 Sub-sample 2

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.73*** –0.57** –0.047 –0.303*** 0.56*** –0.67*

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 5. Foreign banks.

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.15* –0.32 –0.091 0.486** 0.44** –0.15

Table 6. Other domestic banks.

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.489*** –0.49** 0.098 0.653 0.52*** –0.81**

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 7. All banks.

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.59*** –0.31** 0.59** 0.18* 0.59** –1.25**

Table 8. Big4 banks.

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.73*** –0.57** 0.63*** –0.35* 0.01** –4.63*

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Table 9. Foreign banksa.

Full sample Sub-sample 1

∆MARGINt–1 0.15* –0.32 0.21** –0.11

Table 10. Other Domestic Banks.

Full sample Sub-sample 1 Sub-sample 2

∆MARGINt–1 0.489*** –0.49** 0.32** –0.04 0.51*** –1.16*

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

Downloaded from aum.sagepub.com at University of Sydney on May 1, 2015

6. Conclusions and directions for further research

∆MARGINt–1 0.73* –0.57 –0.047 –0.303* 0.56* –0.67*

∆MARGINt–1 0.15* –0.32 –0.091 0.486 0.44 –0.15

∆MARGINt–1 0.489* –0.49 0.098 0.653 0.52* –0.81

∆MARGINt–1 0.59* –0.31 0.59** 0.18* 0.59 –1.25

∆MARGINt–1 0.73* –0.57 0.63*** –0.35* 0.01** –4.63*

∆MARGINt–1 0.489* –0.49 0.32 –0.04 0.51* –1.16*

∆MARGINt–1 LLO, NAW, STC* BOA#, CIT# MEL*