The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of Chicago
The Effects of Geographic Expansion on Bank
Efficiency
By: Allen N. Berger and Robert
DeYoung
WP 2000-14
The Effects of Geographic Expansion on Bank Efficiency
Allen N. Berger
Board of Governors of the Federal Reserve System
Washington, DC 20551 U.S.A.
and
Wharton Financial Institutions Center
Philadelphia, PA 19104 U.S.A.
Robert DeYoung
Federal Reserve Bank of Chicago
Chicago, IL 60604 U.S.A.
Forthcoming in:
Journal of Financial Services Research
The opinions expressed do not necessarily reflect those of the Federal Reserve Board, the Chicago Reserve
Bank, or their staffs. The authors thank the anonymous referee, Tim Hannan, Joe Hughes, Iftekhar Hasan,
Dave Humphrey, Moshe Kim, Knox Lovell, Ana Lozano-Vivas, Loretta Mester, Andrew Meyer, Jesus Pastor,
Subhash Ray, Tony Saunders, Steve Seelig, Phil Strahan, Larry Wall, Larry White, and other participants at the
Miguel Hernández University Banking and Finance Workshop, Western Economic Association meetings,
Financial Management Association Meetings, and the Federal Reserve Committee on Financial Structure and
Regulation for helpful comments, and Kelly Bryant and Nate Miller for outstanding research assistance.
Please address correspondence to Allen N. Berger, Mail Stop 153, Federal Reserve Board, 20th and C Streets.
NW, Washington, DC 20551, call 202-452-2903, fax 202-452-5295, or email aberger@frb.gov, or to Robert
DeYoung, Economic Research Department, Federal Reserve Bank of Chicago, 230 South LaSalle Street,
Chicago, IL 60604, 312-322-5396 (voice), 312-322-2357 (fax), robert.deyoung@chi.frb.org.
The Effects of Geographic Expansion on Bank Efficiency
Abstract
We assess the effects of geographic expansion on bank efficiency using cost and profit efficiency for over
7,000 U.S. banks, 1993-1998. We find that parent organizations exercise some control over the efficiency of
their affiliates, although this control tends to dissipate with distance to the affiliate. However, on average,
distance-related efficiency effects tend to be modest, suggesting that some efficient organizations can overcome
any effects of distance. The results imply there may be no particular optimal geographic scope for banking
organizations — some may operate efficiently within a single region, while others may operate efficiently on a
nationwide or international basis.
JEL classification codes: G21, G28, G34, G38
Key words: Banks, Efficiency, Mergers, Financial institutions.
The banking industry is consolidating around the globe in response to regulatory changes and other
factors. The U.S. banking structure has not yet fully adapted to the Riegle-Neal Act, which permits bank
branching on almost a nationwide basis, and the Gramm-Leach-Bliley Act, which permits relatively
unrestricted universal banking powers for well-capitalized financial holding companies. In the European
Union, the financial structure has not yet fully adapted to the Single Market Programme, which essentially
allows continent-wide universal banking with a single license, and monetary union, which provides for a single
currency for most of the continent. In Asia, consolidation is occurring in reaction to recent banking and
financial crises. These changing factors, along with the globalization of financial markets in general, have also
created opportunities for intercontinental financial organization mergers and acquisitions (M&As). In most
cases, the consolidation activity involves geographic expansion – financial organizations expanding to other
locations within their home regions; into other regions within their home nation; or into other host nations, any
of which may be considerable distances away.
The purpose of this paper is to assess the effects of this geographic expansion on bank efficiency. On
the one hand, geographic expansion may allow efficiently managed institutions to ‘export’ their superior
managerial skills and policies and procedures to distant affiliates; take advantage of network economies; and
exploit the benefits of geographic risk diversification. On the other hand, operating a far-flung banking empire
may reduce efficiency as senior managers stray into markets in which they have less core competence; as
organizational diseconomies arise (such as agency problems in monitoring junior managers in a distant locale);
and as distance makes providing relationship-based services to local customers more difficult. We investigate
these effects of geographic expansion on efficiency by addressing three related questions:
1) Can parent financial organizations control the efficiency of their affiliates by exporting
their skills/policies/procedures, and does this control vary with distance to the affiliate?
2) Does the efficiency of an affiliate vary with its distance from its parent organization?
3) Are some individual financial organizations able to exercise more control and function
more efficiently as geographically dispersed organizations than others?
It is important to distinguish between the concept of control and the level of efficiency. More control implies
that the efficiency of affiliate banks will be more similar to the efficiency of the parent. That is, the controlling
organization can export either high or low quality managerial skills/policies/procedures.
1
We address these questions by conducting three distinct empirical analyses -- a bivariate analysis, a
regression analysis, and an individual organization analysis. Each of these analyses uses estimates of cost and
profit efficiency for over 7,000 U.S. banks from 1993 to 1998. The data provide an excellent laboratory for
analyzing the efficiency effects of geographic expansion because the U.S. is a geographically large nation with
long distances between banks in the same bank holding company (BHC). Over this time period, banks were
generally allowed to operate in only one state (although their parent BHCs could own banks in multiple states),
so the location of the bank is a fairly good indicator of where it operated and achieved its efficiency scores.
These data allow us to observe how well organizations control the efficiency of distant affiliates and whether
affiliate efficiency varies with distance from the parent. Importantly, our observations are not confounded by
differences in language, culture, currency, regulatory/supervisory structures, etc. that may plague studies of
banks operating across international borders.
Our bivariate analysis compares the efficiency of banks located in the same geographic region as their
parent organization, the efficiency of banks located in geographic regions contiguous to their parents, and the
efficiency of banks located in regions that are not contiguous to their parents. We also compare the efficiency
of home-region banks whose parent organizations own banks only in that same region, home-region banks
whose parents own banks in contiguous regions, and home-region banks whose parents own banks in
noncontiguous regions. Thus, we are able to see how geographic expansion into nearby and far away locations
affects the efficiency of both distant affiliates and home-region banks. This analysis primarily addresses
question (2) above about the relationship between distance and the level of efficiency.
Our regression analysis focuses on what determines the efficiency of affiliates in multibank holding
companies. We regress affiliate bank efficiency on the efficiency of the lead bank in the BHC, the distance
between the affiliate and the lead bank, and other variables. Our maintained assumption is that the senior
management of the organization is located at the lead bank and that the performance of the lead bank is
representative of the organization’s skills, policies, and procedures. We use the coefficient on lead bank
efficiency as an indicator of parental control; we use the coefficient on distance as an indicator of the effect of
distance on affiliate efficiency; and we use the coefficient on the interaction of lead bank efficiency and
distance to indicate the degree to which parental control varies with distance. This analysis addresses both
2
questions (1) and (2) above about the effects of organizational control and distance on efficiency. We also
estimate subsample regressions for affiliates located in different states than, in different regions from, and in
regions noncontiguous to, their parents. By testing whether organizations that choose greater geographic scope
tend to be those that are relatively good at exercising control at a distance, these regressions address question
(3) above about whether some individual organizations are better at control and efficiency than others.
Our individual organization analysis focuses on the efficiency of individual banking organizations, as
opposed to the bivariate and regression analyses, which focus on the efficiency of the average bank affiliate.
We start by identifying all BHCs with 10 or more affiliates in our data. From that group, we identify “efficient
BHCs” whose affiliates tend to operate with above-average efficiency regardless of where they are located – in
the BHC’s home state, in the BHC’s home region, or in regions contiguous or noncontiguous to the BHC’s
home region. Finally, we check to see whether any one particular geographic strategy – such as statewide,
regional, or superregional banking – is predominant among these efficient BHCs. This analysis primarily
addresses question (3) above about whether some individual institutions are better at control and efficiency
than others.
Of the three analyses, we put the greatest emphasis on the regression analysis because it is the most
direct and comprehensive approach to addressing the three questions above. The bivariate and individual
organization analyses shed additional light on questions (2) and (3), respectively, and corroborate the findings
of the regression analysis.
The paper unfolds as follows. Section 1 reviews some previous research that has touched on the issues
of control and distance. Section 2 discusses the efficiency advantages and disadvantages of geographic
expansion. Section 3 explains how we measure bank efficiency and describes our data. Sections 4, 5, and 6
present the results of our bivariate, regression, and individual organization analyses, respectively, and explain
how these results answer the three main research questions asked above. Section 7 provides brief concluding
remarks.
1. Review of related research literature
To our knowledge, no prior research has directly addressed our three main questions about the effects
of organizational control and distance on bank efficiency. Specifically, we know of no prior studies that have
3
that have investigated affiliate efficiency as a function of the efficiency of the parent organization to assess the
degree of organizational control; none that have measured distances between affiliates and parent organizations
to assess the effect of distance on efficiency; and none that have examined the patterns of affiliate efficiency
within individual banking organizations to see if some organizations are able exercise more control and
function more efficiently as geographically dispersed organizations than others.
However, some prior research has come close to these issues. With respect to the control issue, there is
some research on the effects of BHC affiliation on the efficiency of the organization as a whole. While
efficiency of the organization is not the same concept as control, it would be expected that organizations in
which senior management is able to exercise more control would also be more efficient, all else equal. The
empirical results are mixed. For example, some studies found that banks in BHCs are more efficient than
independent banks (e.g., Spong, Sullivan, and DeYoung 1995, Mester 1996). In contrast, other research
suggested that branch banking organizations are more efficient than multibank BHCs (e.g., Grabowski,
Rangan, and Rezvanian 1993), and that for a given organization size, a greater number of separate bank
charters reduces the market value of the organization (Klein and Saidenberg 2000).
Some inference on the control issue may also be gleaned from the research on the efficiency of bank
branches. If senior management is able to effectively control the operations of individual branches, then the
efficiencies of the individual branches would be expected to be clustered near the performance of the best
practice branch of the bank. If senior management is not in control, then the efficiencies of the individual
branches would be expected to be widely dispersed. Studies of the branching networks of large U.S. banks
(e.g., Sherman and Ladino 1995, Berger, Leusner, and Mingo 1997) and of a large Canadian bank (Schaffnit,
Rosen, and Paradi 1997) found efficiencies almost as dispersed as those typically found in studies of unrelated
banks, consistent with relatively weak control for the senior management of the bank. In contrast, a number of
nonparametric efficiency studies that mostly used small numbers of branches (usually for European banks)
typically found relatively tight distributions of branch efficiency, with mean efficiency exceeding .90 (e.g.,
Sherman and Gold 1985, Zenios, Zenios, Agathocleous, and Soteriou 1996, Parkan 1987, Oral and Yolalan
1990, Vassiglou and Giokas 1990, Giokas 1991, Al-Faraj, Alidi, and Bu-Bshait 1993, Pastor 1993, Tulkens
1993, Tulkens and Malnero 1994, Drake and Howcroft 1995, Athanassopoulos 1997, 1998). This last finding
4
could reflect very tight managerial control over branch operations or it could alternatively reflect a problem
with nonparametric methods that arises when the number of observations is relatively small.
Some research has examined the effects of geographic expansion within a nation of banking
organizations as a whole and found generally favorable effects. Some found that larger, more geographically
integrated institutions tend to have better risk-expected return frontiers (e.g., Hughes, Lang, Mester, and Moon
1996, 1999, Demsetz and Strahan 1997). Others found that banking organization M&As raise profit efficiency
in a way consistent with the benefits of improved geographic diversification, but does not have much effect on
cost efficiency (e.g., Berger and Humphrey 1992, Akhavein, Berger, and Humphrey 1997, Berger 1998).
These studies generally found that the difference in efficiency between the acquirer and target had relatively
little effect on overall efficiency improvement following consolidation, suggesting that the ability of acquirers
to export superior skills/policies/procedures to targets may be limited. However, these studies did not examine
the efficiency of the individual affiliates of the organizations, did not directly address the issue of intraorganizational control, and did not measure the distance from the parent organization.
Other research has examined efficiency across international borders, which is usually associated with
significant geographic expansion. Most studies found that foreign affiliates in a host nation are less efficient
on average than the domestic banks in that nation (e.g., DeYoung and Nolle 1996, Hasan and Hunter 1996,
Mahajan, Rangan, and Zardkoohi 1996, Chang, Hasan, and Hunter 1998, Miller and Parkhe 1999, Parkhe and
Miller 1999, Berger, DeYoung, Genay, and Udell 2000), while some found that foreign institutions have about
the same average efficiency as domestic institutions (e.g., Vander Vennet 1996, Bhattacharya, Lovell, and
Sahay 1997, Hasan and Lozano-Vivas 1998). These findings would appear to conflict with the within-nation
results cited above, in which geographic expansion appeared to be favorable on balance. However, as noted
above, cross-border expansion is also associated with other potential barriers to efficiency – such as differences
in language, culture, currency, regulatory/supervisory structures – which are difficult to disentangle from the
effects of geographic expansion in these studies.
Some of the cross-border studies implicitly addressed the issue of control by studying the effects of the
identity of the home nation of the parent organization. The limited findings suggested that the foreign affiliates
with parent organizations in the U.S. tended to be more efficient (e.g., Berger, DeYoung, Genay, and Udell
5
2000), and that efficiency was higher when the home nation and host nation had more similar economic
environments (e.g., Miller and Parkhe 1999, Parkhe and Miller 1999). The first result is consistent with the
possibility that some home nation market or supervisory/regulatory conditions in the U.S. may aid in the
control of foreign affiliates. The second result is consistent with the possibility that having similar market or
supervisory/regulatory conditions in the home and host nations may aid in the control of foreign affiliates.
2. The efficiency advantages and disadvantages of geographic expansion
The removal of intrastate and interstate geographic restrictions on competition in the U.S. has
increased the freedom of banking organizations to expand geographically and potentially move towards a more
efficient structure. Geographic expansion allows senior management of efficient organizations to spread their
best practices over more resources. As stressed above, these efficiency improvements occur when efficient
organizations are the ones expanding, and whether such efficiency transfers are successful depends on
management's ability to control events at a distance. These gains in managerial efficiency, or X-efficiency,
may accrue in the form of lower costs of producing a given bundle of financial services, or in the form of
higher revenues from producing or packaging financial services that are more highly valued by customers, or
both.
Geographic expansion may also allow scale or scope efficiencies that reduce costs or enhance
revenues. Linking branches, ATMs, and back-office facilities over a larger geographic area may yield network
economies. A more geographically broad institution may also be better able to serve business customers that
have many locations, and may have a broader menu of potential new investment opportunities outside its home
market.
In addition, geographic expansion can diversify banking organizations across different regional
economic environments, reducing the variation in the organizations' earnings over time. This can add value by
improving the organization's risk-expected return frontier, allowing the bank to increase its average revenues
by adopting a higher risk, higher expected return investment strategy. A reduction in risk from diversification
may also increase the value of the institution’s financial guarantees (loan commitments, letters of credit,
derivative contracts) and its capacity to issue them. On the cost side, greater diversification may reduce the
organization's cost of capital by allowing it to pay lower rates on uninsured deposits and other contingent
6
liabilities. The costs of complying with prudential supervision and regulation may also be reduced.
To provide some insight into the potential benefits from geographic diversification, Table 1 gives
information about the distribution of bank earnings across geographic regions in the U.S. The table shows the
mean return on equity (ROE) for commercial banks located in the eight Bureau of Economic Analysis (BEA)
regions of the U.S. and the correlation of ROE across these regions over the period 1979-1998. These data
suggest very strong diversification possibilities from cross-regional consolidation. Bank earnings in many
region-pairs have fairly low correlations, including one negative correlation. Eight of the ten weakest
correlations are between noncontiguous region-pairs, while seven of the ten strongest correlations are between
contiguous region-pairs -- indicating that banks have an incentive to expand beyond contiguous regions into
noncontiguous regions to capture additional diversification gains.
While geographic expansion is associated with a number of potential efficiency advantages,
geographic expansion can also potentially reduce efficiency. Geographic expansion by inefficiently managed
banks may spread inferior management practices over a greater amount of resources. As well, otherwise
competent managers may stray into new geographic markets for which they lack relevant local knowledge, or
into markets that require skills that lie outside these managers’ areas of core competence. Holding managerial
ability constant, geographic expansion could also result in scale or scope inefficiencies because managing a
larger, more far-flung empire is more difficult. Organizational diseconomies may arise because senior
managers at the headquarters cannot easily monitor managerial effort, service quality, or economic conditions
at the local level -- that is, there may be significant agency costs in trying to control junior management at a
distant locale. As discussed below, providing and monitoring relationship-based small business loans may be
especially difficult at a distance because of problems in transmitting informal information to a distant
headquarters.1
Over time, improvements in information, communications, and financial technologies may partially
mitigate the efficiency losses related to geographic expansion by making the physical distances between bank
1. Geographical diversification can also increase financial institution risk. A bank’s risk may increase if efficiency is
reduced for any of the reasons described above or if the additional assets have low expected returns, low capital, and/or
high variation of returns. In addition, the expanded institution may choose to take on more risk (e.g., by reducing loan
7
headquarters and local offices, and between banks and borrowers, less important. For example, credit scoring
models and more cost-effective voice and Internet communications appear to have made it easier to analyze
credit applications and monitor small business borrowers at greater distances. Consistent with this, one recent
study found that the distances of small businesses from their banks has been increasing over time (Petersen and
Rajan 2000). It seems reasonable that such advances might also make it easier to monitor loan officers and
other bank personnel at greater distance.
However, we argue that physical distance matters, will continue to matter in the near future, and that
technological advances can only partially mitigate the effects, both unfavorable and favorable, of distance on
bank efficiency. For example, making relationship loans to borrowers that do not quality for credit scoring
because of relatively weak financial statements and collateral of questionable value requires local knowledge
that is difficult to quantify and transmit to a distant headquarters. This local knowledge includes not only
financial information about the firm, but information about the firm’s managers, its local economic
environment, and its relationships with customers, suppliers, and local competitors. Because much of this
information is difficult to quantify and transmit, so that verifying whether local loan performance problems are
due to adverse local conditions, poor performance of the borrowers, or lax effort/incompetence of local loan
officers becomes more difficult as distance increases. In addition, geographic expansion brings potential
diversification benefits that increase with physical distance, as shown above. These benefits may accrue to
banks that provide loans, deposits, or other financial products and services on a multiregional, national, or
international basis. It is unlikely that advances in information, communications, and financial technologies will
smooth out differences in regional economic conditions and fully mitigate these potential efficiency gains from
geographic expansion.
3. Measuring bank efficiency
We first review the efficiency concepts employed and the methodology for estimating efficiency
(section 3.1). We then describe the data and the efficiency estimates (section 3.2).
3.1 Efficiency concepts and methodology
monitoring). Bank risk may also increase if return distributions are negatively skewed and the diversification increases
the number of different events that can drive the bank into default (Winton 1999).
8
We estimate cost efficiency and profit efficiency, which measure how well a bank performs relative to
a best-practice institution that produces the same output bundle under the same exogenous conditions. Cost
efficiency is derived from a cost function of the form:
ln Ci ,t = ft (w i ,t , y i,t , z i,t , vi ,t ) + ln uiC + lnε iC,t ,
(1)
where C measures bank costs, including both operating and interest expenses; i indexes banks (i=1,N); t
indexes time (t=1,T); f denotes some functional form; w is the vector of variable input prices faced by the
bank; y is the vector of its variable output quantities; z indicates the quantities of any fixed netputs (inputs or
outputs); v is a set of variables measuring the economic environment in the bank’s local market(s); lnuC is a
factor that represents a bank’s core efficiency; and lnεC is random error that incorporates both measurement
error and luck.
We measure of cost efficiency for bank i by comparing its actual costs (adjusted for random error) to
the minimum costs necessary to produce bank i's output and other exogenous variables (w,y,z,v):
COSTEFFi =
Ĉ min
Ĉ i
=
exp [f̂( w i , y i , z i , v i )] × exp [ ln û Cmin ] û Cmin
= C ,
exp [f̂( w i , y i , z i , v i )] × exp [ ln û Ci ]
û i
(2)
where ûminC is the minimum ûiC across all the banks in the sample. COSTEFF can be thought of as the
proportion of costs or resources that are used efficiently. For example, a bank with COSTEFF = 0.70 is 70%
efficient, or equivalently wastes 30% of its costs relative to a best practice bank facing the same conditions.
These inefficiencies may reflect the inferior skills and knowledge of managers and/or the agency costs
managers acting in their own interests, rather than those of the shareholders.
We measure profit efficiency based on a profit function with the same arguments as the cost function:
π
ln(π i ,t + θ t ) = f t (w i ,t , yi ,t , z i ,t , v i ,t ) + ln uiπ + lnε iπ,t ,
(3)
where π is bank profit; θ is a constant that makes π+θ positive for all banks (so that the log is defined); lnuπ
represents the bank’s core efficiency; and lnεπ is a random error term. Using techniques similar to those for
estimating cost efficiency, we construct a measure of profit efficiency for bank i by comparing its actual profits
(adjusted for random error) to the maximum profits (i.e., the best practice) attainable given bank i's outputs and
9
other exogenous variables (w,y,z,v):
PROFEFFi =
{
{
}
}
exp [f̂ π ( w i , yi , z i , vi )] × exp [ ln ûπi ] − θ
πˆi
=
,
πˆ max exp [f̂ π ( w i , yi , z i , vi )] × exp [ ln ûπmax ] − θ
(4)
where ûmaxπ is the maximum ûiπ across all banks in the sample. PROFEFF can be thought of as the proportion
of maximum profits that are earned. For example, a bank with PROFEFF = 0.70 is 70% efficient, or is
forgoing about 30% of its potential profits through excessive costs, deficient revenues, or both. Again, these
inefficiencies may reflect inferior managers, managers acting in their own interests, or both.
We use PROFEFF as our main measure of bank performance, because profit efficiency is conceptually
superior to cost efficiency for evaluating overall firm performance. Profit efficiency is based on the economic
goal of profit maximization, which requires that the same amount of managerial attention be paid to raising a
marginal dollar of revenue as to reducing a marginal dollar of costs. PROFEFF may also better capture the
benefits of cross-regional diversification of risk. As discussed, geographic diversification may increase the
value of a bank's financial guarantees and its capacity to issue them, and may allow the bank to enhance its
expected revenues by making higher risk-higher expected return investments. These benefits, as well as any
expense reductions due to a lower cost of capital or a reduction in regulatory compliance costs, are generally
included in PROFEFF. We also include COSTEFF primarily to diagnose whether differences in efficiency
have their origins in cost control or in revenue generation.
PROFEFF is often called ‘alternative’ profit efficiency because the profit function in (3) specifies
output quantities y, rather than output prices p as in a standard profit function. We use alternative profit
efficiency rather than standard profit efficiency primarily because output prices are difficult to measure
accurately for commercial banks, and because output quantities are relatively fixed in the short-run and cannot
respond quickly to changing prices as is assumed in the standard profit function vary across banks more than
output prices and thus better explain differences in bank profits. Prior research generally found similar results
for estimates of standard and alternative profit efficiency. See Berger and Mester (1997) for an extended
discussion of these issues.
10
We estimate both the cost function (1) and profit function (3) separately for each year in our 19931998 panel, allowing the estimated parameters to vary over time. We apply the distribution-free approach
(Berger 1993) to the estimates to calculate COSTEFF and PROFEFF. For each bank, we calculate the sixyear averages of the estimated residual terms (lnuC + lnεC) and (lnuπ + lnεπ). The core efficiency terms lnu are
assumed to remain constant for each bank over time, and the random errors lnε are assumed to tend to average
out over time. To reduce the impact of substantial random outliers, we truncated the average residuals at the 5th
and 95th percentiles of the distributions of their size classes. These truncated distributions of average residuals
provide us with the variables lnûbC, lnûbπ, lnûminC, and lnûmaxπ used in equations (3) and (5) to calculate a single
set of cost and profit efficiency measures for each bank over the entire six-year period.2
We specify the cost and profit functions using the Fourier-flexible functional form. This hybrid
functional form combines a conventional translog form with Fourier trigonometric terms. The resulting form is
more flexible than the translog, and has been shown to fit the data for U.S. financial institutions better than the
translog, especially when a relatively small number of extremely large or small banks are present in the data
(McAllister and McManus 1993, Mitchell and Onvural 1996, Berger, Cummins, and Weiss 1997, Berger and
DeYoung 1997, and Berger, Leusner, and Mingo 1997). The cost function includes three variable input prices
(the local market prices of purchased funds, core deposits, and labor); four variable outputs y (consumer loans,
business loans, real estate loans, securities); three fixed netputs z (off-balance-sheet activity, physical capital,
financial equity capital); and an environmental variable STNPL (the ratio of total nonperforming loans to total
loans in the bank’s state) to control for the business conditions facing each bank.3 By using local market input
prices, rather than the prices actually paid by each bank, our COSTEFF and PROFEFF estimates will reflect
how well individual banks price their deposits and purchased funds. Specifying financial assets as outputs and
2. The reasonableness of these assumptions depends on the length of period studied. If too short a period is chosen the
random errors might not average out well, and if too long a period is chosen the bank’s efficiency is less likely to remain
constant. Using 1984-1994 data on U.S. commercial banks, DeYoung (1997) found that a six-year time period, such as
we use here, reasonably balanced these concerns.
3. The variable input prices are average prices for the state or region in which the bank was located, and are constructed
by dividing a bank’s expenditures on the input in questions by the quantity purchased of that input, and then taking the
asset-weighted average across the banks in that state or region. The variable loan outputs are measured gross of
allowances for uncollectable loans. The variable securities output is measured as gross total assets less loans and
physical capital; and the fixed off-balance sheet output is measured by the risk-weighted (based on the Basle Accord risk
weights) amounts of items such as unused lines of credit, derivative contracts, etc.
11
financial liabilities and physical factors as inputs is consistent with the intermediation approach or the asset
approach to modeling bank production (Sealey and Lindley, 1977).
The Fourier-flexible cost function is specified as follows:
2
ln(C / w3 z 3 ) = α + å β i ln( wi / w3 ) +
i =1
1 2 2
åå β ij ln(wi / w3 ) ln(w j / w3 )
2 i =1 j =1
3
1 3 3
+ å γ k ln( y k / z 3 ) + åå γ km ln( y k / z 3 ) ln( y m / z 3 )
2 k =1 m=1
k =1
2
+ å δ r ln( z r / z 3 ) +
r =1
1 2 2
åå δ rs ln( z r / z3 ) ln( z s / z3 )
2 r =1 s =1
+
1 2 3
ååη ik ln(wi / w3 ) ln( y k / z 3 )
2 i =1 k =1
+
1 2 2
åå ρ ir ln(wi / w3 ) ln( z r / z3 )
2 i =1 r =1
+
1 3 2
ååτ kr ln( y k / z 3 ) ln( z r / z 3 )
2 k =1 r =1
7
+ å [φ n cos( x n ) + ω n sin( x n )]
n =1
7
7
+ åå [φ nq cos( x n + x q ) + ω nq sin( x n + x q )]
n =1 q = n
7
+ å [φ nnn cos( x n + x n + x n ) + ω nnn sin( x n + x n + x n )]
(5)
n =1
1
+ υ1 ln STNPL + υ11 [ln STNPL] 2 + lnu C + lnε C
2
The profit function is identical to this cost function, except that the dependent variable becomes
ln [(π/w3z3)+θ] and the composite error term is relabeled as lnuπ + lnεπ. Costs, profits, and input prices are
normalized by the price of labor (w3) to impose linear input price homogeneity.4 Costs, profits, variable
outputs, and fixed netputs are normalized by financial capital (z3) to give the profit model more economic
meaning, and to control for insolvency risk, heteroskedasticity, scale biases, and other estimation problems.5
4. Thus, on the efficient frontier, a doubling of all input prices exactly doubles costs. Although not necessary, we impose
this constraint on the alternative profit function as well.
5. Specifying financial equity capital as fixed helps resolve several estimation problems. First, high levels of financial
capital reduce insolvency risk, which reduces costs via lower risk premia on substitutes for other perhaps more costly risk
management activities. Second, financial capital provides an alternative to deposits as a funding source for loans, but it
12
Normalized in this way, the dependent variable in the profit functions is essentially the bank’s return on equity,
a measure of how well the bank is using its scarce financial capital. We add 1 to the arguments (yk/z3), (zr /z3),
and STNPL in order to avoid taking the natural log of zero. The xn terms, n=1,...,7 are re-scaled values of the
ln (wi/w3), i=1,2; ln (yk/z3), k=1,2, 3; and ln (zr /z3), r=1,2 terms. To conserve degrees of freedom, we include
only the ‘own’ third-order Fourier terms (e.g., cos(xn + xn + xn)), and exclude the third-order interactions (e.g.,
cos(xn + xm + xq), m, q ≠ n).6 The standard symmetry restrictions apply to the translog portion of the function
(βij = βji, γkm = γmk, δrs = δsr). We exclude consideration of factor share equations embodying Shephard's
Lemma or Hotelling's Lemma restrictions because these would impose the undesirable assumption of no
allocative inefficiencies. We estimate the cost and profit equations using ordinary least squares.
3.2 Data and efficiency estimates
The data are annual observations for all U.S. commercial banks over 1993-1998, and were collected
from the Reports of Condition and Income (call reports). There were 10,875 banks in the U.S. in 1993, which
shrank to 8,713 as of 1998, primarily due to consolidation through M&As.7 We separated the data into two
samples: a ‘main sample’ of institutions that had more than $100 million in gross total assets (1998 dollars) in
all six years, and a ‘small-bank sample’ of institutions with less than $100 million in one or more years.
Annual cost and profit functions were estimated for each sample of banks, using all observations with complete
data in any given year. We calculated COSTEFF and PROFEFF for each of the 1,540 banks from the main
has different cost characteristics than deposits: the initial cost of raising capital is high, but interest expense on capital is
zero. Third, high levels of financial capital may indicate that managers are risk-averse (i.e., willing to accept lower risk
in exchange for less than maximum profits and/or less than minimum costs), so including capital prevents us from
labeling these banks as inefficient even though they are behaving optimally given their risk preferences. Fourth, failing to
control for financial capital could yield a scale bias, because large banks tend to be better diversified than small banks,
and as a result can manage their portfolio risk with lower levels of financial capital. We use an accounting measure of
equity because market values are unavailable for most banks.
6. The Fourier-flexible form is a global approximation because the cos xn, sin xn, cos 2xn, sin 2xn, etc., terms are
mutually orthogonal over the [0,2π] interval (π refers here to radians, not profits), so that each additional term can make
the approximating function closer to the true path of the data wherever it is most needed. The orthogonality is perfect
only if the data are evenly distributed over the [0,2π] interval, but in practice the Fourier terms have improved the fit of
the data in every application of which we are aware. We cut 10% off each end of the [0,2π] interval so that the xn span
[0.1×2π, 0.9×2π] to reduce approximation problems near the endpoints. The formula for xn is 0.2π - µ×a + µ×variable,
where [a,b] is the range of the variable being transformed, and µ ≡ (0.9×2π - 0.1×2π)/(b-a).
7. The direction of any survivor bias, and how it would affect our results, is not clear a priori. Many acquired banks
retain their charters and continue to operate as affiliates of the acquiring organization; these banks remain in our data if
the new owner of these banks was located in the same region as the old owner (which is most often the case).
13
sample and 6,331 banks from the small-bank sample that were present in all six annual regressions, were at
least 50% owned by U.S. persons or by a U.S. BHC, and whose lead bank (if the bank was affiliated with a
multibank BHC) was located in the same geographic region for all six years. We use these sampling criteria to
help ensure that the lead bank-affiliate bank relationships in our data are stable across time. The means and
standard deviations of the variables used in the cost and profit functions for these samples are shown in Table
2.
Although $100 million in assets is an arbitrary threshold, separating the very smallest banks from the
rest of the population is important for several reasons. First, small banks generally attract relationship-based
customers, as opposed to large banks, which tend to produce more transactions-driven services (Kwast, StarrMcCluer, and Wolken 1997). Second, these different business strategies may have implications the effects of
organizational control and distance on efficiency. For example, managing and monitoring a small,
relationship-based bank from afar may be more difficult than managing from the same distance a larger bank
that sells more generic services. Third, the variation in costs and profits among the very smallest commercial
banks is much greater than for the rest of the banking population (Berger and Humphrey 1991), suggesting that
treating these banks separately may allow for more precise efficiency estimates for the main sample. We
include both sets of banks in our analysis to be comprehensive – the small-bank sample comprises nearly 75%
of industry banks, while the main bank sample contains nearly 90% of industry assets.
Our efficiency estimates are similar to those found in the literature. Average measured COSTEFF is
76.4% for the small banks and 78.0% for banks in the main sample. This suggests that the typical bank wastes
about one-quarter of its expenses. Average measured PROFEFF is 66.3% and 66.8%, respectively, for the
small-bank and main samples, suggesting that a typical bank forgoes about one-third of its potential profits.
4. Bivariate analysis
Our bivariate analysis consists of two comparisons. Section 4.1 compares the average efficiency of
banks located in the same geographic region as their parent organization to the average efficiency of banks
located in regions different than their parent. Section 4.2 compares the average efficiency of home-region
banks in single-region organizations to the average efficiency of home-region banks in multiregional
organizations. In both comparisons, we distinguish contiguous from noncontiguous regions, because a
14
nationwide banking organization would require operations in home, contiguous, and noncontiguous regions.
4.1 Bivariate analysis: Same versus different region from the parent organization
Table 3 displays the mean values of PROFEFF and COSTEFF for various subsets of our small bank
and main samples that isolate banks that are managed from within their home regions from banks that are
managed cross-regionally.8 Row (a) shows the mean cost and profit efficiencies for banks located in the same
geographic region as their parent organization's headquarters. Row (b) shows the mean efficiencies for banks
located in one of the seven regions other than the one in which their parent organization is headquartered.
Rows (c) and (d) disaggregate the row (b) data based on whether banks are located in regions that are
contiguous or noncontiguous to their headquarters' region (definitions of contiguous regions are shown in the
table notes).
Each cell in Table 3 contains, reading from top to bottom, the mean efficiency, number of banks, and
standard error for the subset of banks in the cell. To test the effects of distant ownership, we compare the mean
efficiencies of the banks in rows (b) through (d) to the mean efficiencies of the banks in row (a). The
superscripts ** and * (superscripts ## and #) indicate that the average bank is statistically significantly more
efficient (significantly less efficient) than the average bank in row (a) at the 5 and 10 percent levels, two-sided.
The data for the main sample indicate that banks located in different geographic regions than their
parents tend to be somewhat more efficient than average. Comparing rows (a) and (b) for the main sample,
banks located outside their headquarters' regions are more efficient than banks located within their
headquarters' regions, by an average of 2.7% of costs (80.5% versus 77.8%). A larger difference is revealed
when we differentiate between contiguous and noncontiguous regions. In row (c), banks in regions contiguous
to their headquarters have 4.8% better cost efficiency than within-region banks on average (82.0% versus
77.8%), while in row (d) this difference disappears for banks located in noncontiguous regions. Although the
differences in rows (b) and (c) are statistically significant, they are economically small compared to the overall
variation in cost efficiency. The standard deviation of COSTEFF is 8.75% and the inter-quartile spread in
COSTEFF (25th to 75th percentiles) is 11.3% for all banks in the main sample. Hence, these differences do not
8. We include banks not affiliated with holding companies and banks in one-bank holding companies as being in the
same region as the parent organization (i.e., they are their own parents).
15
by themselves provide a strong motivation for cross-regional or nationwide expansion. Furthermore, the main
sample data show no statistically significant differences between the profit efficiency of within-region and
cross-regionally owned banks.
The results are similar for the small-bank sample. Small banks located outside their headquarters
region are statistically more efficient than within-region banks by an average 4.7% of costs (81.0% versus
76.3%); the average efficiency difference increases to 5.4% of costs and 3.2% of potential profits for small
banks located in contiguous regions; and the efficiency differences disappear for small banks located at
noncontiguous distances. However, these differences are small relative to the standard deviation of 8.73% and
inter-quartile range of 11.4% for small bank COSTEFF, and even smaller relative to the standard deviation of
14.49% and inter-quartile range of 18.96% for small-bank PROFEFF.
These results have several implications. First, they suggest that cost-efficient organizations may be
able to spread their superior skills and procedures to banks in nearby states and regions. Second, they imply
the existence of organizational diseconomies to operating or monitoring an institution from afar that offset
these cost efficiencies as banks move further away from their organizational headquarters. Finally, the weaker
profit efficiency results imply that superior cost efficiency may be roughly offset by revenue shortfalls in crossregionally owned banks, perhaps indicating that it is more difficult to manage revenue generation from a
distance than to manage bank costs from a distance. This is most evident in the small bank data, where profit
efficiency declines by 9.1% between the contiguous and noncontiguous subsamples. It may be the case that
organizational diseconomies to operating or monitoring an institution from afar make it especially difficult to
provide relationship-based products (e.g., small business loans) in which small banks typically specialize
4.2 Bivariate analysis: Single-region versus multiregional parent organizations
It seems unlikely that a superregional or nationwide banking organization would be successful without
being efficient in its home markets. We evaluate this proposition using the data displayed in Table 4, which
compares the efficiencies of banks from single-region organizations to the efficiencies of banks from crossregional organizations. Row (a) displays the mean cost and profit efficiencies for banks whose parent
organizations do not own any banks outside of that geographic region. Row (b) displays the mean cost and
profit efficiencies for home-region banks that are owned by cross-regional banking organizations. Rows (c)
16
and (d) disaggregate the row (b) results based on the contiguous or noncontiguous geographic scope of the
parent organizations.9
There are two additional reasons for evaluating the efficiency of the home-region banks. First, the
benefits and costs associated with cross-regional diversification can accrue throughout the organization, not
just in the out-of-region affiliate banks. Bank holding companies may serve as internal capital markets to
reallocate funds where they are most productive (Houston, James, and Marcus 1997, Houston and James 1998,
Klein and Saidenberg 2000). In addition, senior management’s attention may become focused on improving
the efficiency of recently purchased banks in other regions, and unintentionally let the efficiency of the homeregion banks deteriorate. Second, the results in Table 4 act as a control for accounting anomalies caused by
inaccurate transfer pricing. For example, if the lead bank in the organization provides services for the affiliate
banks (e.g., data or payments processing, advertising and marketing, human resources support, etc.) but does
not fully price those services, then the superior cost efficiencies that we find for cross-regionally owned banks
in Table 3 could simply be a reflection of those subsidies. However, if we find that banks owned by
multiregional organizations perform better than average regardless of their location, then it is unlikely that
inaccurate transfer pricing is a problem.
The results in Table 4 show that on average, home-region banks from multiregional organizations (row
(b)) are both more cost efficient and more profit efficient than banks from single-region organizations (row
(a)). In our main sample, the row (b) banks have statistically significant efficiency advantages equal to 3.5% of
costs and 2.1% of potential profits, and in our small-bank sample, the row (b) banks have statistically
significant efficiency advantages equal to 6.2% of costs and 3.5% of potential profits. Moreover, the average
difference in cost efficiency widens as the geographic scope of the organization increases. Cost efficiency
improves 2.0% and 2.6% of costs, respectively, in the main bank and small-bank samples as the organization
expands into noncontiguous regions (from row (c) to row (d)). However, much like our Table 3 analysis, the
cost efficiency differences are relatively small, and the profit efficiency differences disappear entirely as
organizations expand into noncontiguous regions.
The latter result implies that problems managing
organization-wide revenues eventually occur as the geographic scope of the organization expands.
9. The number of observations in rows (a) and (b) of Table 4 equal the number of observations in row (a) of Table 3.
17
It is difficult to draw strong conclusions from the bivariate analysis shown in Tables 3 and 4 because
the results mix the efficiency effects of control and distance with an organization’s choice of geographic
strategy. For example, organizations that choose to expand geographically may also be the organizations that
are best at controlling affiliates at a distance. We attempt to separate out these effects in the regression analysis
that follows.
5. Regression analysis
We test the effects of organizational control and distance on affiliate efficiency using the following
multiple regression framework:
NONLEADEFF = + 1*LEADEFF + 2*lnDISTANCE + 3*LEADEFF*lnDISTANCE
+ 4*SAMESTATE + 5*MSA + 6*HERF + 7*lnBKASS + 8*½(lnBKASS)2
+ 9*lnHCASS + 10*½ (lnHCASS)2 + 11*BKMERGE + 12*HCMERGE
+ 13*REGION + 14*UNMAPPEDNL + 15*UNMAPPEDL + ν.
(6)
We estimate (6) separately for cost and profit efficiency and for the main sample and the small-bank sample
using OLS techniques. NONLEADEFF is the efficiency (cost or profit) of a non-lead bank affiliate.
LEADEFF is the same type of efficiency (cost or profit) of the lead bank, defined as the largest banking
affiliate in the BHC. We maintain that LEADEFF is a good proxy for the skills, policies, and practices
available to manage the entire organization. lnDISTANCE is the natural log of the distance in miles between
the affiliate bank and its lead bank (one mile added before logging). SAMESTATE is a dummy equal to one if
the affiliate is located in the same state as the lead bank. MSA is a dummy equal to one if the affiliate is
located in a metropolitan statistical area. HERF is the affiliate’s average Herfindahl index, weighted by the
share of its deposits that come from each MSA or non-MSA county. BKASS equals the gross total assets of
the affiliate bank, and HCASS equals the sum of BKASS over all of the bank affiliates in the BHC.
BKMERGE is a dummy equal to one if the affiliate was the surviving bank in a merger during 1990-1997 in
which two or more bank charters were consolidated, and HCMERGE is a dummy equal to one if the affiliate
was acquired by a BHC during 1990-1997 but retained its bank charter. REGION is a vector of dummies that
indicate the BEA region in which the affiliate is located. Summary statistics are displayed in Table 5.
We used mapping software to compute the distance “as the crow flies” between the cities in which the
18
affiliate banks and lead banks were located. Although one could contemplate using alternative measures of
distance between two cities (e.g., actual distance traveled, or total travel time), such measures are not easily at
our disposal, and in any event these alternative measures are likely highly correlated with lnDISTANCE. The
maximum distance was 3,303 miles from Anchorage to Houston and the minimum distance was zero. The lead
and non-lead affiliates in our sample were located in 1,912 different U.S. cities, and we were able to match
1,150 of those cities to an exact geographic location. We assigned banks in the other 762 unmappable cities to
the city that is the state’s banking center, which in most cases is the largest city in the state.10 To partially
control for the measurement distortion that this introduces, we add the dummies UNMAPPEDNL and
UNMAPPEDL to equation (6), which equal one when the non-lead bank and the lead bank, respectively, were
located in an unmappable city. The degree of measurement distortion should diminish as the non-lead bank
moves further from the state of its lead bank.
The derivative MNONLEADEFF/MLEADEFF = 1+3*lnDISTANCE measures the association
between lead bank efficiency and non-lead bank efficiency, given the distance between the two banks and
holding constant the values of the other regressors. As such, this derivative may reflect the degree to which the
organization is able to control the operations of its affiliate banks through the transfer of its management skills,
policies, and practices. We expect this derivative to be positive if organizations exercise some control over
their affiliates, but less than 1; a value of 1 would indicate that a given increase in lead bank efficiency would
be fully matched by an equal increase in non-lead efficiency. A positive derivative that declines with distance,
i.e., with 3<0, would be consistent with longer distances interfering with the control of organization over its
affiliates’ efficiency.
The derivative MNONLEADEFF/MlnDISTANCE = 2+3*LEADEFF reflects the degree to which
affiliate bank efficiency increases or declines with its distance from the lead bank, given the efficiency of the
lead bank and holding constant the values of the other regressors. We have no a priori expectation regarding
the sign of this derivative. A negative sign would be consistent with the hypothesis that geographically
dispersed banking firms experience organizational diseconomies that make it difficult to manage and monitor
10. We used the largest city in the state, except for in California (San Francisco instead of Los Angeles), Missouri (St.
Louis instead of Kansas City), and Virginia (Richmond instead of Virginia Beach).
19
affiliates from afar. A positive sign would be consistent with the hypothesis that efficiently managed
organizations can overcome any such organizational diseconomies, perhaps in part through the benefits of
geographic diversification.
We acknowledge that mismeasurement of LEADEFF (which we measure with estimation error) and
lnDISTANCE (which we measure with error for unmappable banks) could introduce bias into the estimated
regression coefficients. The coefficients on LEADEFF, lnDISTANCE, and LEADEFF*lnDISTANCE may
each be biased toward zero, making it difficult to reject the null hypothesis associated with these
coefficients.11
Other biases may be present as well.
The relationship between LEADEFF and
NONLEADEFF, measured by the estimated sum 1+3*lnDISTANCE could be biased downward if there are
cross-subsidies (in either direction) between the lead bank and the non-lead affiliate through inaccurate transfer
pricing or other intra-organizational accounting methods.12 Conversely, this estimated sum could be biased
upward if the lead bank and its non-lead banks are exposed to common economic shocks not controlled for
elsewhere in our estimations.13 However, our results below suggest it is unlikely that these biases materially
affect the conclusions that we draw from our analysis.
The coefficients on SAMESTATE may be positive, reflecting the benefits of operating under a single
set of state regulations, or negative because of poor risk diversification. The coefficients on HERF may be
positive in the profit efficiency equations and negative in the cost efficiency regressions — consistent with the
literature (Berger and Mester 1997, Berger and Hannan 1998) — as banks with more market power may raise
prices and profits, but may have higher costs due to reduced competitive pressures to keep costs under control.
The coefficients on MSA may take any sign because of the many differences between metropolitan and rural
11. We cannot correct for this bias because we know neither the variance of the measurement errors nor their
covariances with the true values of the variables in question.
12. Table 5 contains circumstantial evidence of subsidies that flow from lead banks to non-lead banks – lead banks
(LEADEFF) are on average less cost and profit efficient than non-lead banks (NONLEADEFF) in both the main sample
and the small-bank sample. To some extent, these averages simply reflect the fact that banking organizations with large
numbers of affiliates tend to exhibit above-average affiliate efficiency. For example, for organizations with 10 or more
affiliates, mean affiliate cost efficiency was above the sample mean at 24 of 33 organizations, and mean affiliate profit
efficiency was above the sample mean in 20 of the 33 cases.
13. NONLEADEFF and LEADEFF are constructed from the residuals from the same cost and profit functions. If a nonlead affiliate and its lead bank experienced a common shock (e.g., a change in economic conditions or a regulatory action
common to the location of both banks) not controlled for in the functions, then the effects of that shock will be captured
in those residuals, and hence will be commonly embedded in the efficiency measures NONLEADEFF and LEADEFF.
20
markets. We expect the coefficients on lnBKASS and ½lnBKASS2 to reveal a positive relationship between
efficiency and bank size, based on earlier studies that found the smallest banks to be the least efficient (e.g.,
Berger and Humphrey 1991). We include lnHCASS and ½lnHCASS2 to capture the efficiency effects of
agency costs, internal capital markets, and other factors that may vary with BHC size, and to prevent the
efficiency effects of our key variable lnDISTANCE from being confounded with the efficiency effects of BHC
size, which may be strongly related to lnDISTANCE. The coefficients on the BKMERGE and HCMERGE
variables could be either positive or negative, depending upon the dynamic effects of consolidation on bank
efficiency. We also have no a priori expectations for the signs on the REGION coefficients, nor for the signs
on UNMAPPEDNL and UNMAPPEDL.
5.1 Regression results
Table 6 displays the results of equation (6) for cost and profit efficiency for both the main sample and
small-bank sample. These estimates suggest that on average the efficiency of a non-lead affiliate bank is
strongly influenced by the efficiency of its lead bank, but not by the distance to its lead bank. The coefficient
1 on LEADEFF is positive and highly significant in all four regressions; the coefficient 2 on lnDISTANCE is
positive but generally insignificant; and the coefficients 3 on the LEADEFF*lnDISTANCE interaction terms
are negative and generally insignificant. The derivative MNONLEADEFF/MLEADEFF = 1+3*lnDISTANCE
is positive and significantly different from zero at the sample means in all four regressions, consistent with lead
banks exercising some control over the operations of their affiliate banks through transfers of management
skills, policies, and practices. In contrast, the derivative MNONLEADEFF/MlnDISTANCE = 2+3*LEADEFF
is not statistically different from zero when evaluated at the sample means in any of the four regressions.
To see how the control of the parent organization varies with distance, Table 7 shows how the
estimated value of MNONLEADEFF/MLEADEFF declines as non-lead banks are increasingly distant from their
lead banks. For example, the first column of Table 7 shows that this derivative equals 0.3182 for affiliate
banks located at 0 miles from their lead banks. That is, a one percentage point increase in lead bank cost
efficiency is associated with a 0.3182 percentage point increase in non-lead cost bank efficiency. However,
this derivative is only half as large(0.1688) at a distance of 288 miles, the average distance for a main sample
affiliate located outside its lead bank’s home state but within its home region, and this derivative becomes
21
statistically insignificant at the maximum distance in the main sample. Figure 1 displays this information
graphically, plotting each of the four derivatives from Table 7 continuously against the distance between the
non-lead affiliate and its lead bank. The figure suggests that organizational control over affiliate bank cost
efficiency dissipates more with distance than does organizational control over affiliate bank profit efficiency
(each cost efficiency derivative curve quickly falls below the profit efficiency derivative curve for the same
sample). This may indicate that revenue gains from geographic diversification offset a large portion of the
organizational cost diseconomies that come with geographic dispersion.
The figure also suggests
organizational control over small bank efficiency dissipates more with distance than does organizational
control over the efficiency of larger affiliate banks (each small-bank sample curve quickly falls below the main
sample curve for the same efficiency concept). This may indicate that organizations experience relatively
greater difficulties monitoring and managing from a distance relationship-based activities and other locallybased financial services in which most small banks specialize.
Our finding of a strong positive relationship between lead bank efficiency and non-lead bank
efficiency may reflect factors other than organizational control. As noted above, cross-subsidies in either
direction between the lead and non-lead banks would tend to give a downward bias to the measured derivative
MNONLEADEFF/MLEADEFF, and so our finding of a strong positive effect suggests that such a bias is not
dominant. We also noted that there could be an upward bias due to common shocks that affect the measured
efficiency of the lead and non-lead banks similarly. To test for the possibility that common shocks are the
cause of the positive estimated relationship between NONLEADEFF and LEADEFF, we re-estimated (6) after
adding the interaction variable LEADEFF*SAMESTATE to the right-hand-side of the equation (results shown
only in Berger and DeYoung 2000). Common economic shocks are most likely to occur if the lead and nonlead bank are in the same state, so if common shocks are driving our result, then this interaction variable
should have a positive coefficient and soak up much of the positive relationship between lead bank and nonlead bank efficiency. However, none of these 4 regressions produced a significant positive coefficient on this
variable, and in most cases the other regression coefficients were materially unaffected. In the small bank cost
and profit efficiency regressions, the signs, significance levels, and relative magnitudes of the coefficients 1,
2, and 3 were unchanged. In the main sample cost regression, the derivative with respect to LEADEFF
22
remained positive and significant, but it was statistically significant only for same-state affiliate banks – a result
that, if considered in isolation, is consistent with the presence of common local shocks. But we found no
evidence of common shocks in the more comprehensive main sample profit regressions – the coefficient on
LEADEFF*SAMESTATE was significant but negative, and the overall derivative with respect to LEADEFF
was similar to the Table 6 result in terms of sign, magnitude, and significance level. Thus, it is unlikely that
common economic shocks are driving our main result.
The coefficients on SAMESTATE, HERF, BKASS, HCASS, and BKMERGE are statistically
significant in at least two of the four Table 7 regressions, with signs that are generally consistent with our
expectations. The coefficients on the REGION dummies suggest the non-lead affiliates in the Plains,
Southeast, Southwest, and Rocky Mountains regions tend to be more profit efficient than those in other
regions, suggesting favorable economic or regulatory conditions in these regions over this time period. The
coefficients on MSA, HCMERGE, UNMAPPEDNL, and UNMAPPEDL are statistically significant in only
one regression or not at all, and do not have consistent signs across the regressions.
In Table 8 we use a slightly different procedure for separating the effects of organizational control and
distance. We exclude the interaction variable LEADEFF*lnDISTANCE from equation (6), and then estimate
the model for subsamples of affiliate banks located at various distances from their lead banks. In panel (a) we
estimate the model for the full sample of non-lead affiliate banks of multibank BHCs, and panels (b), (c), and
(d) display estimation results for subsamples of affiliates located in different states from their lead banks, in
different regions from their lead banks, and in regions noncontiguous to their lead banks, respectively.
The results in Table 8 suggest that organizational control, geographic distance, and the characteristics
of individual organizations may all be important determinants of affiliate bank efficiency. First, the coefficient
on LEADEFF is always positive, and is statistically significant in 9 of the 16 regressions, reinforcing the earlier
results that affiliate bank efficiency is strongly correlated with lead bank efficiency. Second, the coefficients
generally (but not always) increase in size as the affiliate bank subsamples move further away from the lead
bank. For example, in the second column of Table 8 the coefficient on LEADEFF increases from 0.2774 to
0.3898 to 0.5008 to 0.9141 as geographic dispersion increases. This suggests that organizations that choose to
be geographically dispersed tend to be organizations that are relatively capable at controlling their affiliates.
23
Third, although the coefficient on lnDISTANCE is statistically significant in only 4 of the 16 regressions, all
four of these coefficients are negative and occur in the out-of-state and out-of-region subsamples where
distance is likely to be measured accurately and is likely to matter most.14 This is weak evidence of a pure
distance effect in which increased distance from the lead bank reduces affiliate bank efficiency.
6. Individual organization analysis
In contrast to the previous analyses in which we focused on the average effects of control and distance
on bank efficiency, we now disaggregate the data and focus instead on the efficiency of affiliates within
individual multibank organizations. We attempt to identify whether certain geographic strategies (e.g.,
statewide, regional, or superregional banking) are more often associated with efficient organizations than other
geographic strategies. We also attempt to identify whether many, some, or none of the multibank organizations
in our data are good candidates to sustain nationwide banking organizations in the future. We note that the
Riegle-Neal Act currently limits to 10% the total share of nationwide bank and thrift deposits that any
organization may obtain via M&As, which makes it difficult for any institution to operate a full service
banking operation on a nationwide basis.
To make this analysis tractable, we limit our investigation to the 33 organizations in our data with 10
or more affiliate banks (including the lead bank). Table 9.1 displays cost efficiency data for these
organizations, and Table 9.2 displays profit efficiency data for these organizations.15 Data for each of the 33
organizations is displayed on a separate row. In column (1), we categorize the geographic scope of each
organization as either “statewide,” “regional,” “superregional contiguous,” or “superregional noncontiguous.”
"Statewide" organizations only own affiliates in their home states. "Regional" organizations own affiliates
outside their home states, but not beyond their home regions. "Superregional contiguous" organizations own
affiliates outside their home regions, but only in regions that are contiguous to their home regions.
"Superregional noncontiguous" organizations own affiliates in regions that are not contiguous to their home
14. Panel (d) contains noncontiguous affiliates for which an additional mile of distance is less meaningful. Panel (a)
contains same-state affiliates for which the distance mapping problems are most likely to cause distortions.
15. To simplify the analysis, Tables 9.1 and 9.2 do not report separate results for banks in the main sample and small
bank sample. This should have little effect on our analysis because, as we report in section 3.1 above, the distributions of
COSTEFF and PROFEFF were very similar for the two samples.
24
regions. We distinguish between “superregional contiguous” organizations and “superregional noncontiguous”
organizations, because the latter have a geographic scope that is closer to nationwide banking and we wish to
see whether nationwide banking might be an efficient geographic strategy. While these 33 organizations
represent just a small fraction of the 733 multibank organizations in the data, they include multibank holding
companies with a variety of different geographic strategies; as such, analyzing the efficiency of these
organizations may provide a good indication of whether managers can efficiently operate large numbers of
affiliates on a statewide, regional, superregional or nationwide basis.16
Column (2) shows the number of affiliate banks in each of these organizations. The remainder of the
columns in Tables 9.1 and 9.2 contain information about the cost and profit efficiency of each organization’s
affiliate banks. Organizations are listed from most cost or profit efficient to least efficient, based on the
(unweighted) mean efficiencies of their affiliate banks in column (3). The ordering in Table 9.1 is based on
cost efficiency rank, and the ordering in Table 9.2 is based on profit efficiency rank. Columns (4) through (7)
display the mean efficiencies of affiliate banks that are located at various geographic distances from the lead
bank. The superscript "A" identifies cells in which the reported efficiency mean is higher than mean efficiency
of the affiliates in all 733 multibank organizations in the data. The superscript "B" shown in column (1)
identifies organizations whose affiliates have above-average efficiency in each of the geographic locations in
which the organization operates (i.e., rows in which the superscript “A” appears in every populated cell in
columns (4) through (7)).
There are three main results in Tables 9.1 and 9.2. First, organizations with 10 or more affiliates tend
to be more efficient than multibank organizations with smaller numbers of affiliates. Of the 33 organizations
shown in these tables, 24 had mean cost efficiencies higher than the 0.7989 average for all affiliates of
multibank organizations (column (3) in Table 9.1), and 20 had mean profit efficiencies higher than the 0.6931
average for all affiliates of multibank organizations (column (3) in Table 9.2). Second, there is evidence that
some already widely dispersed organizations may be good candidates to sustain nationwide banking
organizations in the future. For example, 4 of the 6 superregional noncontiguous organizations in Table 9.2
16. These 33 organizations with 10 or more affiliates account for 5 of the 549 statewide multibank organizations, 4 of
the 73 regional multibank organizations, 18 of the 76 superregional contiguous multibank organizations, and 6 of the 35
25
operated affiliates with above-average profit efficiency in noncontiguous regions, and for the most part also
operated efficiently in the areas closer to the organizations’ headquarters. Third, among the “geographically
efficient” organizations that carry the "B" superscript in the tables, no single geographic scope is dominant.
Again using profit efficiency as our benchmark, 2 statewide organizations were geographically efficient; 3
regional organizations were geographically efficient; 3 superregional contiguous organizations were
geographically efficient; and 2 superregional noncontiguous organizations were geographically efficient.
Consistent with the results in the prior analyses, this suggests that well-managed organizations can spread their
efficient management skills/policies/procedures across affiliate banks regardless of the geographic spread of the
organization.
7. Conclusions
We estimate the cost and profit efficiency of over 7,000 U.S. commercial banks between 1993 and
1998 and use those estimates to assess the impact of geographic expansion on bank efficiency. We find both
positive and negative links between geographic scope and bank efficiency. For example, while banks in
organizations that expand into nearby states and regions tend to have higher levels of efficiency, organizational
control over affiliate bank efficiency tends to diminish as affiliates move further away from the parent,
especially for small bank affiliates with less than $100 million in assets. But these distance-related efficiency
effects tend to be modest in size, and our results suggest that efficient parent organizations can export their
superior skills, policies, and practices to their affiliates and overcome any negative effects of distance. These
results imply that operating an efficient banking organization may not necessarily conform to any one particular
geographic strategy. An individual organization analysis of a number of large multibank BHCs with varying
geographic structures confirms this notion.
These results may have important implications for the future structure of the banking industry. First,
the data suggest that domestic banking organizations that operate statewide, across state lines, across
geographic regions, or nationwide are likely to coexist in the future without any one type of organization
having a sufficient efficiency advantage to drive the others out of existence. Such a result would be consistent
with projections made elsewhere that several thousand banking organizations are likely to disappear during the
superregional noncontiguous multibank organizations in the data.
26
adjustment to deregulation, but that the remaining banks will still number in the thousands (e.g., Berger,
Kashyap, and Scalise 1995). Our results also suggest that very small banks may be less likely to be efficiently
owned and operated by nationwide organizations, perhaps due to organizational diseconomies to operating or
monitoring from afar an institution that specializes in relationship-based lending or locally-oriented services.
The results may also have some bearing on the debate over why most studies of cross-border bank
efficiency found that foreign affiliates are on average less efficient than the domestic banks in the same nation.
To the extent that our findings may extrapolate to cross-border applications, the data suggest that distancerelated inefficiencies are unlikely to explain the cross-border findings. If domestic banks can operate
efficiently at any distance from their parent organizations within a large nation like the U.S., then it is unlikely
that distance-related inefficiencies are responsible for the finding that domestic banks are usually more efficient
than foreign banks. The cross-border findings may be more likely associated with other potential international
barriers to efficiency – such as differences in language, culture, currency, and regulatory/supervisory structures.
Some additional important caveats apply to our findings. First, during our 1993-1998 sample period,
most interstate banking was done via BHC affiliation, and our efficiency findings for these affiliate banks may
not apply to future networks of bank branches. More efficient nationwide geographic scope may be more
likely in the future, since geographically expansive banking organizations will be able to choose whichever
organizational form is most efficient for them. Second, substantial restrictions on interstate acquisitions were
in place prior to, as well as during the first part, of our sample period. Future efficiencies from cross-border
expansion may be higher than found here if the past restrictions mainly prevented efficient organizations from
expanding.
Third, we do not observe any truly nationwide banking organizations in our data (i.e.,
organizations with conventional deposit-taking offices in all 50 states), and such organizations may be deterred
in the U.S., given that the Riegle-Neal Act limits the national share of bank and thrift deposits that any single
organization may obtain by consolidation to 10%. Therefore, our evaluation of nationwide banking is based on
extrapolating the existing pattern of cross-regional ownership to a geographical spread that does not presently
exist and may not occur in the future. Fourth, technological change may reduce distance-related efficiency
barriers, enabling organizations to more efficiently process information, market their services, manage across
geographic distance, and manage the risks of institutions with greater geographic scope in the future. Finally,
27
we urge great caution in extrapolating these results to other nations or to cross-border applications, given that
distance effects could be compounded with many other differences between the U.S. and other markets.
28
References
Akhavein, J.D., A.N. Berger, and D.B. Humphrey. 1997. “The Effects of Bank Megamergers on Efficiency
and Prices: Evidence from the Profit Function.” Review of Industrial Organization 12: 95-139.
Al-Faraj, T.N., A.S. Alidi, K.A. Bu-Bshait. 1993. “Evaluation of Bank Branches by Means of Data
Envelopment Analysis”, International Journal of Operations and Production Management 13: 45-52.
Athanassopoulos, A.D. 1997. “Service Quality and Operating Efficiency Synergies for Management Control
in the Provision of Financial Services: Evidence from Greek Bank Branches,” European Journal of
Operational Research 98: 300-313.
Athanassopoulos, A.D. 1988. “Service Quality and Operating Efficiency Synergies for Management Control
in the Provision of Financial Services,” Journal of Money, Credit, and Banking 30: 172-92.
Berger, A.N. 1993. “’Distribution-Free’ Estimates of Efficiency in the U.S. Banking Industry and Tests of the
Standard Distributional Assumptions.” Journal of Productivity Analysis 4(3): 261-292.
Berger, A.N. 1998. “The Efficiency Effects of Bank Mergers and Acquisition: A Preliminary Look at the
1990s Data.” In Bank Mergers & Acquisitions, edited by Y. Amihud and G. Miller. Boston, MA. Kluwer
Academic: 79-111.
Berger, A.N., J.D. Cummins, M.A. Weiss. 1997. “The Coexistence of Multiple Distribution Systems for
Financial Services: The Case of Property-Liability Insurance.” Journal of Business 70(4): 515-546.
Berger, A.N., and R. DeYoung. 1997. “Problem Loans and Cost Efficiency in Commercial Banks,” Journal
of Banking and Finance 21: 849-870.
Berger, A.N., R. DeYoung, H. Genay, and G.F. Udell. 2000. “Globalization of Financial Institutions:
Evidence from Cross-Border Banking Performance.” Brookings-Wharton Papers on Financial Services 3:
23-158.
Berger, A.N., and T.H. Hannan. 1998. “The Efficiency Cost of Market Power in the Banking Industry: A Test
of the “Quiet Life” and Related Hypotheses.” Review of Economics and Statistics 80(3): 454-465.
Berger, A.N. and D.B. Humphrey. 1991. “The Dominance of Inefficiencies over Scale and Product Mix
Economies in Banking.” Journal of Monetary Economics 28(1): 117-148.
Berger, A.N. and D.B. Humphrey. 1992. “Megamergers in Banking and the Use of Cost Efficiency as an
Antitrust Defense.” Antitrust Bulletin, 37 (Fall), 541-600.
Berger, A.N. and D.B. Humphrey. 1997. “Efficiency of Financial Institutions: International Survey and
Directions for Future Research.” European Journal of Operational Research 98: 175-212.
Berger, A.N., A.K. Kashyap, and J.M. Scalise. 1995. “The Transformation of the U.S. Banking Industry:
What a Long, Strange Trip It's Been.” Brookings Papers on Economic Activity 2: 55-201.
Berger, A.N., J.H. Leusner, and J.J. Mingo. 1997. “The Efficiency of Bank Branches.” Journal of Monetary
Economics 40(1): 141-162.
29
Berger, A.N. and L.J. Mester. 1997. “Inside the Black Box: What Explains Differences in the Efficiencies of
Financial Institutions?” Journal of Banking and Finance 21: 895-947.
Bhattacharya, A., C.A.K. Lovell, and P. Sahay. 1997. “The Impact of Liberalization on the Productive
Efficiency of Indian Commercial Banks.” European Journal of Operational Research 98: 332-345.
Chang, C.E., I. Hasan, and W.C. Hunter. 1998. “Efficiency of Multinational Banks: An Empirical
Investigation,” Applied Financial Economics 8(6): 1-8.
Demsetz, R.S. and P.E. Strahan. 1997. “Diversification, Size, and Risk at Bank Holding Companies.”
Journal of Money, Credit, and Banking 29(3): 300-13.
DeYoung, R. 1997. “A Diagnostic Test for the Distribution-Free Efficiency Estimator: An Example Using
U.S. Commercial Bank Data.” European Journal of Operational Research 98: 243-249.
DeYoung, R. and D.E. Nolle. 1996. “Foreign-Owned Banks in the U.S.: Earning Market Share or Buying
It?” Journal of Money, Credit, and Banking 28(4): 622-636.
Drake, L., and B. Howcroft. 1995. “Measuring the Relative Efficiency of the Selling Function: An
Application of Data Envelopment Analysis to UK Bank Branches”, working paper, Loughborough University,
U.K.
Giokas, D. 1991. “Bank Branch Operating Efficiency: A Comparative Application of DEA and the Loglinear
Model,” OMEGA International Journal of Management Science 19: 549-47.
Grabowski, R., N. Rangan, and R. Rezvanian. 1993. “Organizational Forms in Banking: An Empirical
Investigation of Cost Efficiency.” Journal of Banking and Finance 17: 531-538.
Hasan, I. and W.C. Hunter. 1996. “Efficiency of Japanese Multinational Banks in the United States,”
Research in Finance 14: 157-173.
Hasan, I. and A. Lozano-Vivas. 1998. “Foreign Banks, Production Technology, and Efficiency: Spanish
Experience.” Working Paper presented at the Georgia Productivity Workshop III. Athens, Georgia.
Hughes, J.P., W. Lang, and L.J. Mester, and C.-G. Moon. 1996. “Efficient Banking Under Interstate
Branching.” Journal of Money, Credit, and Banking 28(4): 1043-1071.
Hughes, J.P., W. Lang, L.J. Mester, and C.-G. Moon. 1999. “The Dollars and Sense of Bank Consolidation.”
Journal of Banking and Finance 23(2-4): 291-324.
Klein, P.G. and M.R. Saidenberg. 2000. “Organizational Structure and the Diversification Discount:
Evidence from Commercial Banking,” Federal Reserve Bank of New York working paper.
Kwast, M.L., M. Starr-McCluer, and J.D. Wolken. 1997. “Market Definition and the Analysis of Antitrust in
Banking.” Antitrust Bulletin 42(4): 973-995.
Mahajan, A., N. Rangan, and A. Zardkoohi. 1996. “Cost Structures in Multinational and Domestic
Banking.” Journal of Banking and Finance 20(2): 238-306.
30
McAllister, P.H. and D.A. McManus. 1993. “Resolving the Scale Efficiency Puzzle in Banking.” Journal of
Banking and Finance 17(2-3): 389-405.
Mester, L.J. 1996. A study of bank efficiency taking into account risk-preferences, Journal of Banking and
Finance 20, 1025-1045.
Miller, S.R. and A. Parkhe. 1999. “Home-Country Environment as a Source of International Competitiveness:
An Analysis of the Global Banking Industry.” Michigan State University.
Mitchell, K. and N. M. Onvural. 1996. “Economies of Scale and Scope at Large Commercial Banks:
Evidence from the Fourier Flexible Functional Form.” Journal of Money, Credit, and Banking 28(2): 178-99.
Oral, M., and Yolalan, R.. 1990. “An Empirical Study on Measuring Operating Efficiency and Profitability
of Bank Branches,” European Journal of Operational Research 46: 282-94.
Parkan, C. 1987. “Measuring the Efficiency of Service Operations: An Application to Bank Branches”,
Engineering Costs and Production Economics, 12: 237-42.
Parkhe, A. and S.R. Miller. 1999. “Is There a Liability of Foreignness in Global Banking? An Empirical
Test of U.S. Banks' X-Efficiency.” Michigan State University.
Pastor, J.T. 1993. “Efficiency of Bank Branches Through DEA: The Attracting of Liabilities,” Working
Paper, Universidad de Alicante, Alicante, Spain .
Petersen, M.A. and R.G. Rajan. 2000. “The Information Revolution and Small Business Lending: Does
Distance Still Matter?” University of Chicago working paper.
Schaffnit, C., D. Rosen, and J.C. Paradi. 1997. “Best Practice Analysis of Bank Branches: An Application of
DEA in a Large Canadian Bank,” European Journal of Operational Research 98: 269-289.
Sealey, C.W., Jr. and J.T. Lindley. 1977. “Inputs, Outputs, and a Theory of Production and Cost at
Depository Financial Institutions.” Journal of Finance 32(4): 1251-1266.
Sherman, D., and F. Gold. 1995. “Branch Operating Efficiency: Evaluation with Data Envelopment
Analysis”, Journal of Banking and Finance, 9: 297-315.
Sherman, D., and G. Ladino. 1995. “Managing Bank Productivity Using Data Envelopment Analysis
(DEA)”, Interfaces, 25: 60-73.
Spong, K., R. J. Sullivan, and R. DeYoung. 1995. “What Makes a Bank Efficient? A Look at Financial
Characteristics and Bank Management and Ownership Structure.” Federal Reserve Bank of Kansas City,
Financial Industry Perspectives, December: 1-19.
Tulkens, H. 1993. “On FDH Efficiency Analysis: Some Methodological Issues and Applications to Retail
Banking, Courts, and Urban Transit", Journal of Productivity Analysis, 4: 183-210.
Tulkens, H., and A. Malnero. 1994. “Nonparametric Approaches to the Assessment of the Relative
Efficiency of Bank Branches”, working paper, Center for Operations Research and Econometrics, Universite
Catholique de Louvain, Belgium.
31
Vander Vennet, R. 1996. “The Effect of Mergers and Acquisitions on the Efficiency and Profitability of EC
Credit Institutions.” Journal of Banking and Finance 20(9): 1531-1558.
Vassiglou, M., and D. Giokas. 1990. “A Study of the Relative Efficiency of Bank Branches: An Application
of Data Envelopment Analysis,” Journal of the Operational Research Society 41: 591-97.
Winton, A. 1999. “Don’t Put All Your Eggs in One Basket? Diversification and Specialization in Lending.”
University of Minnesota.
Zenios, C., S. Zenios, K. Agathocleous, and A. Soteriou. 1996. “Benchmarks of the Efficiency of Bank
Branches”, working paper, University of Cyprus, Nicosia, Cyprus.
32
Table 1
Correlation Analysis of Bank ROE Among US Regions
Annual data, 1979-1998
New England
(0.106949)
Far West
Rocky Mts.
Southwest
Southeast
Plains
Great Lakes
Mideast
Region
(mean ROE)
New
England
Region
1
Mideast
(0.106738)
0.65875
1
Great Lakes
(0.120448)
0.02411
0.5008
1
Plains
(0.131574)
0.10756
0.44102
0.66704
1
Southeast
(0.126031)
0.84124
0.66657
0.25513
0.3825
1
Southwest
(0.090953)
0.23662
0.60174
0.25345
0.69174
0.36296
1
Rocky Mts.
(0.121841)
0.2603
0.4899
0.4365
0.90354
0.46883
0.8772
1
-0.28249
0.28071
0.69177
0.56564
0.07846
0.32124
0.39953
Far West
(0.107647)
1
Sources: U.S. bank Call Reports, U.S. Bureau of Economic Analysis (BEA).
Return on equity (ROE) = the aggregate net income for the banks in the region, divided by the aggregate book
value of equity for the banks in the region.
Regions: New England (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont); Mideast
(Delaware, District of Columbia, Maryland, New Jersey, New York, Pennsylvania); Great Lakes (Illinois, Indiana,
Michigan, Ohio, Wisconsin); Plains (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota);
Southeast (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South
Carolina, Tennessee, Virginia, West Virginia); Southwest (Arizona, New Mexico, Oklahoma, Texas); Rocky
Mountain (Colorado, Idaho, Montana, Utah, Wyoming); Far West (Alaska, California, Hawaii, Nevada, Oregon,
Washington).
33
Table 2
Summary Statistics
Banks for which Cost and Profit Efficiency were Calculated
Means and (Standard Deviations)
Profits/assets
Costs/assets
Consumer loans/assets
Business loans/assets
Real estate loans/assets
Securities/assets
Off-balance sheet/assets
Equity/assets
Market nonperforming loans/assets
Price of purchased funds
Price of core deposits
Price of labor ($ thousands, 1998)
Gross total assets ($ millions, 1998)
Number of banks
Main Sample
(assets > $100 million)
Small Banks
(assets < $100 million)
0.0259
(0.0124)
0.0416
(0.0134)
0.1032
(0.1197)
0.1470
(0.0953)
0.3441
(0.1414)
0.3910
(0.1285)
0.0273
(0.0439)
0.0913
(0.0317)
0.000006985
(0.000004829)
0.0402
(0.0035)
0.0212
(0.0060)
40.8716
(8.9313)
2,123.76
(12,308.44)
1,540
0.0269
(0.0083)
0.0425
(0.0088)
0.0854
(0.0593)
0.1715
(0.1064)
0.2967
(0.1327)
0.4327
(0.1317)
0.0064
(0.0111)
0.1023
(0.0337)
0.00005809
(0.00005994)
0.0392
(0.0039)
0.0246
(0.0065)
37.1550
(6.1674)
60.15
(39.22)
6,331
9,240
37,986
Number of bank-year observations
Notes: Costs, profits, variable outputs, fixed outputs, and fixed inputs were all scaled by
gross total assets for expository purposes in this table only, not in the regressions.
34
Table 3
Cost and Profit Efficiency by Geographic Location of Bank
Main Sample
N=1,540
(greater than $100 million in assets)
Banks that are located...
(a) in the same region as their
parent organization.
(b) in a different region than their
parent organization.
(c) in a region that is contiguous to
their parent organization.
(d) in a region not contiguous to
their parent organization.
Small Banks
N=6,331
(less than $100 million in assets)
COSTEFF
0.778
1439
0.0023
0.805 **
101
0.0117
0.820 ***
73
0.0138
PROFEFF
0.663
1439
0.0044
0.668
101
0.0221
0.685
73
0.0233
COSTEFF
0.763
6214
0.0011
0.810 ***
117
0.0097
0.817 ***
94
0.0098
PROFEFF
0.668
6214
0.0019
0.682
117
0.0189
0.700 **
94
0.0169
0.767
28
0.0208
0.624
28
0.0514
0.782
23
0.0284
0.609
23
0.0656
The three numbers in each cell are the mean efficiency, the number of observations, and the standard error of the
subsample mean. The location of the "parent organization" is determined by the location of the lead bank for banks
affiliated with multi-bank holding companies, or by the location of the bank itself for banks that are either unaffiliated
or are the sole bank in a one-bank holding company.
***, **, and * indicate that cell mean is significantly higher than the cell in top row at the 1%, 5%, and 10% levels.
Regions: New England (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont); Mideast
(Delaware, District of Columbia, Maryland, New Jersey, New York, Pennsylvania); Great Lakes (Illinois, Indiana,
Michigan, Ohio, Wisconsin); Plains (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota);
Southeast (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina,
Tennessee, Virginia, West Virginia); Southwest (Arizona, New Mexico, Oklahoma, Texas); Rocky Mountain
(Colorado, Idaho, Montana, Utah, Wyoming); Far West (Alaska, California, Hawaii, Nevada, Oregon, Washington).
The New England region is contiguous with Mideast region.
The Mideast region is contiguous with New England, Great Lakes, and Southeast regions.
The Great Lakes region is contiguous with Mideast, Plains, and Southeast regions.
The Plains region is contiguous with Great Lakes, Southeast, Southwest, and Rocky Mountain regions.
The Southwest region is contiguous with Plains, Southeast, Southwest, and Far West regions.
The Rocky Mountain region is contiguous with Plains, Southwest, and Far West regions.
The Far West region is contiguous with Southwest and Rocky Mountain regions.
35
Table 4
Cost and Profit Efficiency of Banks by Geographic Spread of Parent Organization
(Table excludes banks located in a different region from their lead bank.)
Home-region banks whose
parent organizations own...
(a) only banks in the same region.
(b) banks in other regions.
(c) banks in contiguous regions.
(d) banks in noncontiguous regions.
Main Sample
N=1,439 banks
(greater than $100 million in assets)
COSTEFF
0.773
1225
0.0024
0.808 ***
214
0.0064
0.801 ***
133
0.0081
0.821 ***
81
0.0103
PROFEFF
0.660
1225
0.0047
0.681 *
214
0.0111
0.700 ***
133
0.0132
0.651
81
0.0196
Small Banks
N=6,214 home region banks
(less than $100 million in assets)
COSTEFF
0.761
6016
0.0011
0.823 ***
198
0.0065
0.818 ***
157
0.0071
0.844 ***
41
0.0151
PROFEFF
0.667
6016
0.0019
0.714 ***
198
0.0127
0.729 ***
157
0.0145
0.657
41
0.0245
The three numbers in each cell are the mean efficiency, the number of observations, and the standard error of the
subsample mean. The location of the "parent organization" is determined by the location of the lead bank for banks
affiliated with multi-bank holding companies, or by the location of the bank itself for banks that are either unaffiliated
or are the sole bank in a one-bank holding company.
***, **, and * indicate that cell mean is significantly higher than the cell in top row at the 1%, 5%, and 10% levels.
Regions: New England (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont); Mideast
(Delaware, District of Columbia, Maryland, New Jersey, New York, Pennsylvania); Great Lakes (Illinois, Indiana,
Michigan, Ohio, Wisconsin); Plains (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota);
Southeast (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina,
Tennessee, Virginia, West Virginia); Southwest (Arizona, New Mexico, Oklahoma, Texas); Rocky Mountain
(Colorado, Idaho, Montana, Utah, Wyoming); Far West (Alaska, California, Hawaii, Nevada, Oregon, Washington).
The New England region is contiguous with Mideast region.
The Mideast region is contiguous with New England, Great Lakes, and Southeast regions.
The Great Lakes region is contiguous with Mideast, Plains, and Southeast regions.
The Plains region is contiguous with Great Lakes, Southeast, Southwest, and Rocky Mountain regions.
The Southwest region is contiguous with Plains, Southeast, Southwest, and Far West regions.
The Rocky Mountain region is contiguous with Plains, Southwest, and Far West regions.
The Far West region is contiguous with Southwest and Rocky Mountain regions.
36
Table 5
Summary Statistics for Banks used in Efficiency Regressions, Equation (6).
Dependent variables:
NONLEADEFF
(PROFEFF)
NONLEADEFF
(COSTEFF)
Exogenous variables:
LEADEFF
(PROFEFF)
LEADEFF
(COSTEFF)
DISTANCE
lnDISTANCE
SAMESTATE
MSA
HERF
BKASS
lnBKASS
HCASS
lnHCASS
BKMERGE
HCMERGE
REGION1
REGION2
REGION3
REGION4
REGION5
REGION6
REGION7
REGION8
UNMAPPEDL
UNMAPPEDNL
Profit efficiency of non-lead banks
in multibank BHCs.
Cost efficiency of non-lead banks
in multibank BHCs.
Profit efficiency of lead bank
(largest bank in the BHC).
Cost efficiency of lead bank
(largest bank in the BHC).
Distance in miles between affiliate
and lead bank.
Natural log of (DISTANCE+1).
=1 if affiliate and lead bank are in
the same state.
=1 if affiliate is in a Metropolitan
Statistical Area (MSA).
Average Herfindahl index,
weighted by share of affiliate
deposits in each MSA or non-MSA
county in which it operates.
Bank gross total assets (thousands
of 1998 dollars).
Natural log of BKASS.
Sum of BKASS for all bank
affiliates in holding company.
Natural log of HCASS.
=1 if affiliate in merger, 1990-97.
=1 if high holding company made
acquisition, 1990-97.
=1 if affiliate in New England
=1 if affiliate in Mideast
=1 if affiliate in Great Lakes
=1 if affiliate in Plains
=1 if affiliate in Southeast
=1 if affiliate in Southwest
=1 if affiliate in Rocky Mts.
=1 if affiliate in Far West
=1 if location is missing for lead
bank (location is assigned to state’s
largest city or banking center).
=1 if location is missing for
affiliate bank (location is assigned
to state’s largest city or banking
center).
37
425 non-lead banks in
main sample
mean
std. dev.
0.6994
0.1697
1,277 non-lead banks in
small-bank sample
Mean
std. dev.
0.7011
0.1534
0.8099
0.0948
0.8016
0.0926
mean
0.5906
std. dev.
0.1959
Mean
0.6834
std. dev.
0.1492
0.7770
0.0921
0.7853
0.0835
310.52
451.08
136.47
247.17
4.4761
0.5075
2.1593
0.4922
3.8043
0.8099
1.8852
0.3822
0.7361
0.4391
0.3235
0.4667
0.1856
0.0928
0.2573
0.1697
1,228,302
6,181,149
38,186
23,565
12.7496
18,512,914
1.2320
31,362,826
10.3351
1,293,720
0.6975
5,379,763
15.5522
0.1859
0.1035
1.6357
0.3895
0.3050
12.3892
0.0227
0.1308
1.5322
0.1490
0.3373
0.0258
0.1459
0.2341
0.1365
0.2671
0.0867
0.0541
0.0498
0.1204
0.1589
0.3534
0.4233
0.3437
0.4423
0.2811
0.2265
0.2170
0.3078
0.0039
0.0180
0.2136
0.3038
0.2539
0.1299
0.0596
0.0171
0.3592
0.0625
0.1330
0.4098
0.4601
0.4353
0.3364
0.2367
0.1293
0.4576
0.3278
0.4654
0.3637
0.4774
Table 6 -- Regression Estimates for Equation (6)
***, **, and * indicate significant difference from zero at 1%, 5%, and 10% levels.
Efficiency concept:
Sample:
Intercept
COSTEFF
PROFEFF
COSTEFF
PROFEFF
main
main
small bank
small bank
-0.2387
-0.4140
-0.7980 *
-1.0473
(0.5518)
(0.9139)
(0.4428)
(0.7039)
LEADEFF
0.3181 ***
0.3133 ***
0.3243 ***
0.2631 ***
(0.1148)
(0.0833)
(0.0660)
(0.0618)
LnDISTANCE
0.0173
0.0044
0.0295 **
0.0125
(0.0179)
(0.0109)
(0.0125)
(0.0097)
LEADEFF*lnDISTANCE
-0.0264
-0.0085
-0.0398 **
-0.0161
(0.0226)
(0.0168)
(0.0159)
(0.0138)
SAMESTATE
0.0141
0.0693 ***
0.0116
0.0215 *
(0.0134)
(0.0223)
(0.0076)
(0.0121)
MSA
0.0107
-0.0022
0.0011
-0.0172 *
(0.0139)
(0.0232)
(0.0063)
(0.0101)
HERF
-0.1103 **
0.2948 ***
-0.0112
0.0657 **
(0.0575)
(0.0958)
(0.0185)
(0.0295)
LnBKASS
-0.0037
-0.0693
0.1654 *
0.2840 **
(0.0687)
(0.1145)
(0.0879)
(0.1400)
-0.0005
0.0030
-0.0181 **
-0.0265 *
½lnBKASS2
(0.0050)
(0.0083)
(0.0086)
(0.0137)
LnHCASS
0.1061 **
0.1398
0.0803 ***
-0.0154
(0.0523)
(0.0877)
(0.0232)
(0.0372)
-0.0060 *
-0.0066
-0.0048 ***
0.0015
½lnHCASS2
(0.0034)
(0.0056)
(0.0017)
(0.0028)
BKMERGE
0.0198
0.0521 **
0.0374 **
0.0669 **
(0.0122)
(0.0202)
(0.0171)
(0.0273)
HCMERGE
0.0082
0.0098
-0.0036
0.0340 ***
(0.0152)
(0.0253)
(0.0075)
(0.0120)
REGION1
0.0314
0.0704
0.0146
0.1334 *
(0.0345)
(0.0575)
(0.0441)
(0.0703)
REGION2
-0.0154
0.0506
-0.0057
0.0226
(0.0250)
(0.0410)
(0.0271)
(0.0434)
REGION3
-0.0175
0.0553
-0.0018
0.0485
(0.0237)
(0.0391)
(0.0203)
(0.0325)
REGION4
-0.0073
0.0657 *
0.0006
0.0537 *
(0.0240)
(0.0398)
(0.0202)
(0.0322)
REGION5
-0.0193
0.1261 ***
-0.0196
0.0963 ***
(0.0228)
(0.0377)
(0.0201)
(0.0320)
REGION6
-0.0358
0.1178 ***
-0.0131
0.1423 ***
(0.0253)
(0.0418)
(0.0205)
(0.0328)
REGION7
0.0042
0.1503 ***
-0.0061
0.1545 ***
(0.0279)
(0.0468)
(0.0222)
(0.0358)
UNMAPPEDNL
-0.0001
0.0140
-0.0071
0.0168 *
(0.0121)
(0.0203)
(0.0058)
(0.0092)
UNMAPPEDL
0.0077
0.0264
-0.0042
0.0046
(0.0175)
(0.0283)
(0.0057)
(0.0092)
N
425
425
1277
1277
R-square
0.1377
0.2382
0.1011
0.1665
adjusted R-square
0.0927
0.2004
0.0861
0.1526
38
Table 7
Evaluating MNONLEADEFF/MLEADEFF as mean DISTANCE increases.
***, **, and * indicate significant difference from zero at 1%, 5%, and 10% levels.
Efficiency concept:
Sample:
minimum DISTANCE in sample:
point estimate of derivative
standard error of derivative
banks in same state:
point estimate of derivative
standard error of derivative
banks in same region, different states:
point estimate of derivative
standard error of derivative
banks in contiguous regions:
point estimate of derivative
standard error of derivative
banks in noncontiguous regions:
point estimate of derivative
standard error of derivative
maximum DISTANCE in sample:
point estimate of derivative
standard error of derivative
COSTEFF
Main
0 miles
0.3182
0.1148
73 miles
0.2053
0.0532
288 miles
0.1688
0.0591
597 miles
0.1495
0.0681
1297 miles
0.1495
0.0681
2825 miles
0.1084
0.0942
PROFEFF
main
0 miles
***
0.3133 ***
0.0833
73 miles
***
0.2769 ***
0.0441
288 miles
***
0.2652 ***
0.0503
597 miles
**
0.2590 ***
0.0573
1297 miles
**
0.2590 ***
0.0573
2825 miles
0.2458 ***
0.0767
COSTEFF
small bank
0 miles
0.3243 ***
0.0666
81 miles
0.1499 ***
0.0323
188 miles
0.1161 ***
0.0387
368 miles
0.0893 *
0.0461
1278 miles
0.0893 *
0.0461
3303 miles
0.0018
0.0757
PROFEFF
Small bank
0 miles
0.2631 ***
0.0618
81 miles
0.1925 ***
0.0298
188 miles
0.1789 ***
0.0342
368 miles
0.1680 ***
0.0399
1278 miles
0.1680 ***
0.0399
3303 miles
0.1327 **
0.0645
Figure 1
Derivative of affiliate bank efficiency (NONLEADEFF) with respect to
lead bank efficiency (LEADEFF). Based on Table 4 regression estimates.
0.40
main sample, PROFEFF
MNONLEADEFF/MLEADEFF
0.35
small bank sample, PROFEFF
main sample, COSTEFF
0.30
small bank sample, COSTEFF
0.25
0.20
0.15
0.10
0.05
0.00
0
500
1000
1500
2000
2500
miles from affiliate bank to lead bank
39
3000
3500
Table 8
Selected Regression Results for Subsample Estimates of Equation (6)
***, **, and * indicate significant difference from zero at 1%, 5%, and 10% levels.
dependent variable:
sample:
(a) full sample:
LEADEFF
COSTEFF
main
PROFEFF
main
COSTEFF
small bank
PROFEFF
small bank
0.1996 ***
(0.0530)
-0.0031
(0.0031)
425
0.1348
0.0919
0.2774 ***
(0.0441)
0.0005
(0.0052)
425
0.2552
0.2183
0.1760 ***
(0.0308)
-0.0018
(0.0016)
1277
0.0966
0.0822
0.1996 ***
(0.0291)
0.0015
(0.0026)
1277
0.1656
0.1524
0.1098
(0.0903)
-0.0037
(0.0108)
207
0.0931
-0.0045
0.3898 ***
(0.0725)
-0.0420 **
(0.0168)
207
0.2697
0.1912
0.0759
(0.0843)
-0.0128 *
(0.0073)
244
0.1181
0.0390
0.0840
(0.0786)
-0.0015
(0.0135)
244
0.2303
0.1612
0.1400
(0.1390)
lnDISTANCE
-0.0063
(0.0163)
N
113
R-square
0.1893
Adjusted R-square
0.0130
(d) noncontiguous region affiliates only:
LEADEFF
0.6325
(0.4288)
lnDISTANCE
0.0007
(0.0760)
N
31
R-square
0.5258
Adjusted R-square
-0.0162
0.5008 ***
(0.1123)
-0.0492 *
(0.0267)
113
0.3556
0.2156
0.2748 **
(0.1152)
-0.0266 **
(0.0106)
127
0.3184
0.1973
0.4002 ***
(0.1072)
-0.0145
(0.0195)
127
0.3793
0.2690
0.9141 ***
(0.1389)
0.1189
(0.0874)
31
0.8963
0.7778
0.5471
(0.5674)
0.0878
(0.0862)
25
0.8477
0.5938
0.6042
(0.6298)
0.2562
(0.4075)
25
0.6949
0.1864
lnDISTANCE
N
R-square
Adjusted R-square
(b) out-of-state affiliates only:
LEADEFF
lnDISTANCE
N
R-square
Adjusted R-square
(c) out-of-region affiliates only:
LEADEFF
40
Table 9.1
Cost Efficiency, Individual Organization Analysis
Data for banks in 33 BHCs that own at least 10 affiliate banks, ordered by the mean cost efficiency of those affiliates.
"Statewide" BHCs only own affiliates in their home states. "Regional" BHCs own affiliates outside their home states, but
not beyond their home regions. "Superregional, contiguous" BHCs own affiliates outside their home regions, but not
beyond regions that are contiguous to the BHC. "Superregional, noncontiguous" BHCs own affiliates outside their home
regions, including regions that are not contiguous to the BHC. The superscript A indicates a mean cost efficiency greater
than 0.7989, which is the mean COSTEFF for all affiliates of multi-bank holding companies in our data. The superscript
B
indicates a BHC whose affiliates have mean cost efficiencies greater than 0.7989 in each of the geographic locations in
which it operates.
(1)
Geographic scope of BHC
superregional, contiguous B
statewide B
superregional, contiguous B
superregional, contiguous B
superregional, contiguous B
superregional, noncontiguous
statewide B
superregional, noncontiguous
superregional, contiguous
statewide B
superregional, noncontiguous B
regional B
regional B
superregional, contiguous B
superregional, noncontiguous
superregional, contiguous
superregional, contiguous
regional B
superregional, contiguous
superregional, noncontiguous
superregional, noncontiguous
superregional, contiguous
superregional, contiguous
statewide B
superregional, contiguous
superregional, contiguous
superregional, contiguous
regional
superregional, contiguous
statewide
superregional, contiguous
superregional, contiguous
superregional, contiguous
(2)
Number
of banks
in BHC
21
13
15
15
16
26
10
27
15
10
34
28
10
10
13
12
10
12
13
12
20
15
21
13
22
10
10
35
13
10
11
12
10
(3)
Mean cost
efficiency
for all banks
in BHC:
.9295
.9027
.8987
.8908
.8822
.8733
.8683
.8639
.8601
.8356
.8349
.8285
.8275
.8255
.8240
.8215
.8207
.8140
.9132
.9126
.8103
.8029
.7995
.7994
.7974
.7889
.7882
.7834
.7829
.7745
.7675
.7494
.7387
41
Mean cost efficiency for banks located in:
(7)
(6)
(5)
(4)
regions not
regions
home region
home
contiguous
contiguous
(but not in
state of
to BHC
to BHC
home state)
BHC
of BHC
.9274 A
.9721 A
A
.9027
.8979 A
.8341 A
.9232 A
A
A
.9097
.8355
.8648 A
A
A
.8089
.8699
.8989 A
A
.8772
.7775
.8683 A
.8734 A
.6167
.8662 A
.7750
.8356 A
.8087 A
.8784 A
.8283 A
.8598 A
A
A
.8181
.8343
.8041 A
.8509 A
A
.8073
.9893 A
A
A
.8890
.8192
.6613
.7643
.8339 A
.7918
.8388 A
.7146
.8995 A
.8528 A
A
A
.8221
.8114
.8093 A
.7904
.8539 A
A
A
.8688
.8291
.8714 A
.6479
A
.8205
.7077
.9006 A
.7681
.7906
.8451 A
.7823
.8198 A
.6681
.7279
.7994 A
.8006 A
.8051 A
.7262
.7946
.7659
.7782
.8786 A
.7655
.8138 A
A
.8808
.7626
.7914
.7745
.7613
.7751
.7515
.6987
.7980
.7416
.7152
.7493
Table 9.2
Profit Efficiency, Individual Organization Analysis
Data for banks in 33 BHCs that own at least 10 affiliate banks, ordered by the mean profit efficiency of those affiliates.
"Statewide" BHCs only own affiliates in their home states. "Regional" BHCs own affiliates outside their home states, but
not beyond their home regions. "Superregional, contiguous" BHCs own affiliates outside their home regions, but not
beyond regions that are contiguous to the BHC. "Superregional, noncontiguous" BHCs own affiliates outside their home
regions, including regions that are not contiguous to the BHC. The superscript A indicates a mean profit efficiency
greater than 0.6931, which is the mean PROFEFF for all affiliates of multi-bank holding companies in our data. The
superscript B indicates a BHC whose affiliates have mean profit efficiencies greater than 0.6931 in each of the geographic
locations in which it operates.
(1)
Geographic scope of BHC
superregional, contiguous
superregional, contiguous
regional B
superregional, noncontiguous B
statewide B
superregional, contiguous
superregional, noncontiguous B
superregional, contiguous B
superregional, noncontiguous
statewide B
regional B
superregional, contiguous
superregional, contiguous
regional B
superregional, contiguous B
superregional, noncontiguous
superregional, contiguous
superregional, contiguous
superregional, contiguous
superregional, noncontiguous
superregional, contiguous
superregional, contiguous
statewide
regional
superregional, contiguous
superregional, contiguous
superregional, contiguous
superregional, contiguous
superregional, contiguous
superregional, contiguous
statewide
statewide
superregional, noncontiguous
(2)
Number
of banks
in BHC
21
15
35
34
10
16
26
13
12
13
12
12
13
28
11
13
22
21
10
20
10
15
10
10
10
15
15
10
10
12
10
13
27
(3)
Mean profit
efficiency
for all banks
in BHC:
.9051
.8448
.8036
.7892
.7858
.7828
.7746
.7648
.7449
.7419
.7308
.7278
.7231
.7166
.7129
.7121
.7110
.7105
.7006
.6939
.6920
.6909
.6861
.6791
.6662
.6642
.6197
.6462
.6349
.6238
.6166
.5847
.5644
42
Mean profit efficiency for banks located in:
(7)
(6)
(5)
(4)
regions not
regions
home region
home
contiguous
contiguous
(but not in
state of
to BHC
to BHC
home state)
BHC
of BHC
.9173 A
.6603
.8944 A
.6368
.8178 A
.8008 A
.8084 A
A
.7281
.7794 A
.8242 A
.6999 A
A
.7848
.6891
.9040 A
.7667 A
A
.7697
.8973 A
A
A
A
.7303
.8824
.8021
.8168 A
.7622 A
.7149 A
.6512
A
.7419
.7231 A
.7333 A
.6709
.6566
.8564 A
A
.6395
.7370
.7514 A
A
A
.7304
.7089
.7264 A
.6966 A
A
.6819
.8406
.6985 A
.7025 A
A
A
.7182
.7065
.5725
.7269 A
.7160 A
.6150
.7007 A
.7227 A
.6817
.5153
.7342 A
.7125 A
.7348 A
.6931
.6875
.7299 A
.5184
.6058
.6861
.6578
.7005 A
.6693
.6389
.6433
.6693
.6958 A
.6331
.8823 A
.6439
.6667
.6872
.7585 A
.4654
.5603
.6161
.6952 A
.6166
.5847
.5671
.4947
@FedFRASER