Description
Financial systems tend to evolve around a banking sector seeking to achieve economies of scale in order to offset the costs of collecting and processing information designed to reduce uncertainty, thereby facilitating a more efficient allocation of financial resources.
WP/05/17
Competition and Efficiency in Banking:
Behavioral Evidence from Ghana
Thierry Buchs and Johan Mathisen
©2005 International Monetary Fund WP/05/17
IMF Working Paper
African Department
Competition and Efficiency in Banking: Behavioral Evidence from Ghana
Prepared by Thierry Buchs and J ohan Mathisen
1
Authorized for distribution by Samuel Itam
J anuary 2005
Abstract
This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily represent
those of the IMF or IMF policy. Working Papers describe research in progress by the author(s) and are
published to elicit comments and to further debate.
This paper assesses the degree of bank competition and discusses efficiency with regard to
banks’ financial intermediation in Ghana. By applying panel data to variables derived from a
theoretical model, we find evidence for a noncompetitive market structure in the Ghanaian
banking system, which may be hampering financial intermediation. We argue that the
structure, as well as the other market characteristics, constitutes an indirect barrier to entry
thereby shielding the large profits in the Ghanaian banking system.
J EL Classification Numbers: G21, D43
Keywords: Ghana, banking competition
Author(s) E-Mail Address: [email protected]; [email protected]
1
We would like thank Bank of Ghana staff, Hugh Bredenkamp, J ack Glen, participants at the
Ghana at the Half Century conference in J uly 2004, Luca Ricci and Tom Walter for helpful
comments and suggestions. At the time this paper was prepared, J ohan Mathisen was an
economist in the IMF’s African Department and Thierry Buchs was a Senior Economist at the
International Finance Corporation (IFC). Any remaining errors and omissions are our own.
- 2 -
Contents Page
I. Introduction ........................................................................................................................... 3
II. Overview of the Ghanaian Banking System........................................................................ 4
A. Structure of Ghana’s Banking Sector............................................................................... 4
B. Financial Performance of the Banking Sector.................................................................. 6
C. Possible Factors Explaining Bank Profitability and the Efficiency of Intermediation .. 10
III. Analytical Framework and Econometric Estimation........................................................ 12
A. The Panzar and Rosse Analytical Framework ............................................................... 13
B. Description of the Data and Definitions of Variables .................................................... 16
C. Estimation Results.......................................................................................................... 17
D. Interpretation of the Coefficients ................................................................................... 19
IV. Conclusions....................................................................................................................... 21
References............................................................................................................................... 23
Appendix................................................................................................................................. 25
Figures
1. Nominal Interest Rates, Government Debt, Real Growth of Private Sector Credit,
Private Sector Credit, 1998-2003...................................................................................... 5
2. Investment and (Gross) Lending of the Banking Sector (1996-2003).................................. 6
3. Savings, Investment and Loans Ratios Across Sub-Saharan Africa................................... 12
Tables
1. Structure of the Banking Sector............................................................................................ 4
2. International Comparison of Selected Banking and Institutional Indicators........................ 8
3. Financial Soundness Indicators for the Banking Sector, 1997–2003 ................................... 9
4. Profitability Indicators ........................................................................................................ 10
5. Panzar and Rosse’s H-Statistics.......................................................................................... 14
6. H-Statistics Values for the Banking System in Ghana ...................................................... 17
7. Banking Sector Market Structure in Selected Countries .................................................... 17
8. Regression Results.............................................................................................................. 18
Appendix Table
A1. Market Equilibrium Test Results...................................................................................... 26
- 3 -
I. INTRODUCTION
Financial systems tend to evolve around a banking sector seeking to achieve economies of
scale in order to offset the costs of collecting and processing information designed to reduce
uncertainty, thereby facilitating a more efficient allocation of financial resources. In well-
functioning economies, banks tend to act as quality controllers for capital seeking successful
projects, ensuring higher returns and accelerating output growth. However, a competitive
banking system is required to ensure that banks are effective forces for financial
intermediation channeling savings into investment fostering higher economic growth.
This paper assesses the level of competition in the Ghanaian banking sector. At first sight,
the very high profit ratios and high cost structure of Ghanaian banks could indicate a
monopolistic banking structure. This is partly corroborated by the findings of this paper. By
deriving variables from a theoretical model and using a 1998-2003 panel data set, we find
evidence for a noncompetitive market structure in the Ghanaian banking system, possibly
hampering financial intermediation. This paper argues that the structure, as well as the other
market characteristics, constitutes an indirect barrier to entry thereby shielding the large
profits in the Ghanaian banking system.
Besides the banking sector, the Ghanaian financial system also includes insurance
companies, discount houses, finance houses, leasing companies, savings and loan
associations, credit unions, and a stock exchange.
2
Thus, by narrowing the focus to the
banking sector only, other potentially important participants of the Ghanaian financial system
might have been overlooked. However, the banking system is by far the largest component of
the financial system, and, according to the recent Financial Sector Stability Assessment
(FSSA) update,
3
many of these other financial institutions remain underdeveloped, even by
sub-Saharan African standards. Moreover, this paper defines the banking sector to include
only deposit-taking financial institutions; it excludes rural banks
4
and the Bank of Ghana.
The remainder of the paper is organized as follows. Section II describes the main
characteristics of the structure and features of the banking sector in Ghana, highlighting the
main differences between Ghana and other sub-Saharan African countries. The banks’
financial performance is then discussed, and certain possible explanatory factors for the
performance are outlined. After a very brief literature survey, Section III presents the
theoretical model, operationalizes it by deriving empirical variables, and describes the
dataset. Then the overall results are presented and discussed, followed by an attempt to
investigate the relationships between the factors of production, macroeconomic variables and
revenue and profitability in the banking sector. Section IV summarizes the results and
concludes.
2
For a full description of the Ghanaian financial system, see “Ghana: Selected Issues,” Section II in
Bredenkamp and others (2003).
3
“Ghana: Financial Sector Stability Assessment Update,” IMF Staff Country Report (396/03).
4
The rural banks account for only about 5 percent of banking system assets.
- 4 -
II. OVERVIEW OF THE GHANAIAN BANKING SYSTEM
A. Structure of Ghana’s Banking Sector
The Ghanaian banking system is rather diverse. Of the 17 banks operating in Ghana, there
were 9 commercial banks, 5 merchant banks, and 3 development banks (Table 1).
5
The three
largest commercial banks account for 55 percent of total assets of the banking sector, which
is relatively moderate compared with other countries in the region. However, about
25 percent of total assets and 20 percent of deposits are held by a single state owned
commercial bank (“bank 1”). The development banks and merchant banks, which focus on
medium- and long-term financing and corporate banking, respectively, together share about
30 percent. The five small commercial banks operate on a much smaller scale. Foreign
investors hold about 53 percent of the shares in eight commercial banks, which is below the
sub-Saharan Africa average, and three banks are state-owned (Table 2). The banking
penetration ratio, at one bank branch per 54,000 inhabitants, is relatively high, but formal
banking reaches only 5 percent of the population and the coverage varies widely. This
reflects the fact that 35 percent of bank branches are in the greater Accra region even though
this region represents less than 13 percent of the country’s population. About half of all bank
branches in the interior belong to the dominant state owned bank.
Table 1. Structure of the Banking Sector
*
Ownership (Percent) Share of Total (Percent)
Ghanaian Foreign Total Assets
(Bns of cedis)
As Percent
of GDP
Number of
Branches
Total
assets
Net
lending
Deposits
Banking system 18,668 38.2 309 100.0 100.0 100.0
Commercial banks 13,055 26.7 229 69.3
Bank 1 97 3 4,624 9.46 134 24.8 16.9 20.8
Bank 2 10 90 2,710 5.55 24 14.5 18.1 16.9
Bank 3 24 76 3,011 6.16 23 16.1 16.3 18.8
Bank 4 46 54 1,713 3.50 38 9.2 10.2 8.8
Bank 5 39 61 470 0.96 6 2.5 2.1 2.6
Bank 6 53 47 128 0.26 4 0.7 0.6 1.0
Bank 7 0 100 120 0.25 3 0.6 0.3 0.8
Bank 8 9 91 230 0.47 1 1.2 0.5 1.2
Bank 9 100 0 49 0.10 1 0.3 0.3 0.3
Merchant banks 2,875 5.9 18
Bank 10 100 0 751 1.54 5 4.0 4.8 5.4
Bank 11 6 94 1,325 2.71 4 7.1 8.2 6.2
Bank 12 34 66 409 0.84 3 2.5 2.1 2.6
Bank 13 71 29 286 0.59 2 1.5 1.5 2.0
Bank 14 100 0 104 0.21 1 0.6 0.4 0.7
Development banks 2,738 5.6 62
Bank 15 100 0 1,847 3.78 42 0.0 11.2 8.4
Bank 16 100 0 538 1.10 14 0.0 3.8 2.0
Bank 17 100 0 352 0.72 6 0.0 1.9 2.0
Sources: Bredenkamp and others (2003) and IMF Staff Country Report no. 396/03.
* As of December 2002. The housing bank established in 2003 has been excluded from this study.
5
Commercial banks engage in traditional banking business, with a focus on universal retail services. Merchant
banks are fee-based banking institutions and mostly engage in corporate banking services. Development banks
specialize in the provision of medium- and long-term finance.
- 5 -
As measured by the aggregated total-assets-to-GDP ratio, the banking sector grew rapidly
between 1996 and 2000, reflecting partly financial deepening, as well as loose monetary
conditions. After reaching 44 percent in 2000, the ratio dropped to 38 percent in 2001 and
further to 31 percent at end-2003, reflecting tightened monetary conditions. The same trend
characterized the share of commercial banks’ foreign operations: the share of bank assets
denominated in foreign currency reached 35 percent on 2000 and then declined to 30 percent
in 2001, probably reflecting the increased stability of the cedi exchange rate.
Following the tightening of monetary policy in 2001, domestic credit to the private sector has
remained at around 10 percent of GDP, which is low even by African standards (Table 2).
This essentially reflects a typical crowding-out effect, as most of the banks’ resources are
absorbed by the public sector, either in the form of loans to state-owned enterprises or
holdings of government securities. As shown in Figure 1, increasing government financing
requirements led to very high real treasury bill yields, especially in periods of tight monetary
policy, and by extension, to high lending rates. During 1998-2003, net loans averaged
34 percent of total assets (peaking at 43 percent in 2001), as banks preferred to invest their
resources in liquid, low-risk assets, such as government securities, the latter constituting
about 25 percent of total assets during the period.
6
Figure 1. Nominal Interest Rates, Government Debt, Real Growth of Private Sector Credit,
Private Sector Credit, 1998-2003
0
10
20
30
40
50
60
Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03
(%)
Public Debt to GDP Ratio
Nominal Lending Rate
Nominal T-Bill Rate
-5
0
5
10
15
20
25
30
35
40
Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03
(%)
Private Sector Credit to GDP Ratio
Real Growth of Private Sector Credit
Source: Bank of Ghana
6
Apart from the financing constraints imposed by Ghana’s large fiscal deficits, the banks’ holdings of
government securities is also sustained by high secondary reserve requirements that require banks to hold
35 percent of their deposit liabilities in such securities.
- 6 -
In addition, state-owned enterprises have attracted sizable amounts of lending from
commercial banks recently, thereby exacerbating the crowding-out effect (Figure 2). As a
result, during the last few years, bank lending to the public sector has typically absorbed
more than half of total available resources. The residual resources available for lending to the
private sector (about 35 percent of total assets in 2003) have been mainly channeled to the
manufacturing sector (25 percent of credit to the private sector), commerce and finance
(9 percent) and services (8.5 percent). The agriculture, forestry, and fishing sectors have
received less than one-tenth of total bank credit although agriculture accounts for 36 percent
of GDP. With the exception of the national oil refinery plant—which is the sector’s largest
exposure
7
—no single borrower amounts to 10 percent of the financial sector’s total equity.
Figure 2. Investment and (Gross) Lending of the Banking Sector (1996-2003)
A. Share of Total Assets (in percent) B. Share of Total Lending (in percent)
*
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
1996 1997 1998 1999 2000 2001 2002 2003
Investment in Bills & Securities
Public Sector Loans
Private Sector Loans
Consumer Loans
0%
10%
20%
30%
40%
50%
60%
1996 1997 1998 1999 2000 2001 2002 2003
Total Lending to the Public Sector
Total Lending to the Private Sector
Source: Section II in Bredenkamp and others (2003), Bank of Ghana
* Total lending includes loans, overdrafts and investments.
B. Financial Performance of the Banking Sector
Financial performance indicators portray a mixed picture. On the one hand, the average
capital adequacy ratio (CAR) was about 13.4 percent in 2002 and 9.3 percent at-end 2003,
well above the minimum 6 percent required by law. There was, however, significant
dispersion among banks, and two small commercial banks even failed to meet the minimum
capital standard requirement,
8
prompting intervention by bank supervisors.
7
Even though a large portion of TOR’s short-term debt was restructured into medium-term government bonds
in 2001 and 2002, TOR exposure still exceeded 75 percent of GCB’s equity capital as of June 2003.
8
IMF Staff Country Report no. 396/03.
- 7 -
In addition, as a result of the negative macroeconomic developments in 1999-2000, the asset
quality of the banks’ loan portfolio appears to have deteriorated. Past-due/nonperforming
loans soared from 16.2 percent in 2000 to an eight-year high of 28.6 percent of total loans in
2001 and 2002 before declining slightly to 24.4 percent in 2003 (Table 3). The overall impact
of this sizable increase on the banking system has been partially softened by the relatively
prudent lending of the two largest foreign-owned banks, however. The system is also
characterized by high overhead costs. The five largest banks incur on average overhead costs
of 7 percent to average assets, which is similar to the sector as a whole but substantially
higher than the sub-Saharan African average of 5.7 percent. Note, however, that these costs
are below those reported in Nigeria and Zambia (Table 2). The high costs could partly reflect
substantial investments in banking infrastructure, notably in information and communication
technologies, as telecommunication in particular suffers from interconnectivity problems.
9
It
could also reflect some marketing practices, such as the refusal to network the automated
teller machines, which appears to have led to unduly high investments in such systems.
However, one key element in the total overhead costs is the staff expenditure component
(about 3.7 percent to average assets), which constitutes roughly half of total overhead costs.
For example, the dominant state owned bank (“bank 1”) has one of the highest staff costs
(4.3 percent to average assets), while the other large commercial banks’ average is 3 percent.
This high ratio suggests both a low level of assets per employee and a relatively high average
staff cost per employee.
On the other hand, profitability indicators indicate that, despite high overhead costs and
sizable provisioning, Ghanaian banks’ pretax returns on assets and equity are among the
highest in sub-Saharan Africa (Table 2)—a situation that reflects very wide interest margins.
On an adjusted basis, the return on assets (RoAA) was 6.1 percent in 2002, which is
remarkable even by African standards, and the same applies to both net interest revenue and
noninterest revenue which are, respectively, 10 percent and 6.4 percent of average assets.
The decline in interest rates in 2002 reduced the banks’ income from government securities
and led to a slight narrowing of interest rate spreads, but the latter remain between 20 and
30 percent.
The combination of wide interest margins, sizable overhead costs, and an ample supply of
relatively low-risk, high-return, government paper, has resulted in high costs of
intermediation. Since the large interest margins also reflect the nonperforming loan problem,
the poor quality of banks’ loan portfolios is a major source of concern for the stability of the
system. Most banks would indeed be vulnerable in the event of a major credit risk shock.
10
9
For example, one of the larger, foreign owned banks has set up a direct satellite network to bypass the national
telecommunications network altogether.
10
IMF Staff Country Report No. 396/03.
- 8 -
Table 2. International Comparison of Selected Banking and Institutional Indicators
(In percent, unless otherwise indicated)
Ghana Kenya Mozambique Nigeria South
Africa
Tanzania Uganda Zambia SSA
Average
Size of financial intermediaries
Private credit to GDP 11.8 26.8 16.7 14.4 147.2 4.9 4.0 7.5 15.2
M2 to GDP 19.0 43.8 5.1 25.8 87.2 18.3 13.0 16.9 24.8
Currency to GDP 10.5 13.2 15.6 10.8 28.4 8.5 8.8 6.4 13.9
Banking industry
Number of banks 17 53 10 51 60 29 15 16 ..
Net interest margin 11.5 5.0 5.9 3.8 5.0 6.5 11.6 11.4 8.3
Overhead costs 7.3 3.7 4.5 7.4 3.7 6.7 4.6 11.2 5.7
Foreign bank share (assets) 53 4.8 98 11.0 0.6 58.7 89.0 66.6 ..
Bank concentration (3 banks) 55.0 61.6 76.6 86.5 77 45.8 70.0 81.9 81.0
Nonperforming loans (share of
total loans)
28.8 41.0 .. 17.3 3.9 12.2 6.5 21.8 ..
Capital markets
Stock market capitalization
(percent of GDP) 10.1 9.2 .. 10.9 77.4 4.3 0.6 6.0 21.3
Contract enforcement
Number of procedures 21 25 18 23 16 26 14 1 29
Duration (number of days) 90 255 540 730 99 207 127 188 334
Bankruptcy
Time in years .. 4.6 .. 1.6 2.0 3.0 2.0 3.7 3.5
Credit market
Credit rights index (0 is weakest)
1
1 1 3 1 2 3 1 2 2
Entry regulations
Number of procedures 10 11 16 9 9 13 17 6 11
Duration (number of days) 84 61 153 44 38 35 36 40 72
Cost (percent of GNI per capita) 111 54 100 92 135 9 199 24 255
Sources: IMF, International Finance Statistics; BankScope; World Bank, World Development Indicators;
Doing Business Indicators Database; and Table 2, “Tanzania: Financial System Stability Assessment,” IMF
Staff Country Report No 03/241. Washington DC: IMF (2003). Banking statistics and capital market indicators
are for 2001. All institutional indicators are for 2003.
1/ The index is based on four powers of secured lenders in liquidation and reorganization. A minimum score of
0 represents weak creditor rights and the maximum score of four represents strong creditor rights. For a
description of the methodology, seehttp://rru.worldbank.org/DoingBusiness/Methodology/CreditMarkets.aspx
- 9 -
Table 3. Financial Soundness Indicators for the Banking Sector, 1997–2003
(In percent, at year’s end, unless otherwise indicated)
1997 1998 1999 2000 2001 2002 2003
Capital Adequacy
Regulatory capital to risk-weighted assets 1/ 15.2 11.1 11.5 11.6 14.7 13.4 9.3
Percentage of banks greater or equal to 10 percent 87.5 75.0 60.0 62.5 64.7 52.9 66.7
Percentage of banks below 10 and above 6 percent minimum 6.3 12.5 40.0 37.5 35.3 35.3 27.8
Percentage of banks below 6 percent minimum 6.3 12.5 0.0 0.0 0.0 11.8 5.6
Capital (net worth) to assets 13.4 12.2 12.2 11.9 13.1 12.6 12.5
Asset quality
Foreign exchange loans to total loans 5/ 25.6 28.5 33.4 35.3 34.1 33.8 ...
Past-due loans to gross loans 24.6 18.9 20.1 16.2 28.0 28.6 24.4
Nonperforming loans 21.6 17.2 12.8 11.9 19.6 22.7 18.3
Watch-listed loans 3.0 1.7 7.3 4.3 8.4 5.9 6.0
Provision as percent of past-due loans 78.0 89.4 67.2 58.6 46.4 63.6 64.4
Earnings and profitability
Net profit (before tax)/net income 51.5 39.2 61.2 52.4 45.9 43.4 39.2
Return on assets 2/ 8.0 8.6 8.5 9.7 8.7 6.8 6.4
Return on equity 3/ 39.9 48.9 48.8 65.7 49.7 36.9 54.0
Expense/income
44.0 42.2 44.3 38.2 40.2 47.3 36.0
Interest rate spread (deposit money banks)
Lending rates minus demand deposit rates 37.0 33.8 32.5 30.5 30.5 30.5 23.3
Lending rates minus saving deposit rates 16.3 22.0 23.5 29.3 29.5 25.5 23.0
Liquidity
Actual reserve ratio (as percent of total deposits) 60.1 64.8 61.8 49.9 62.4 66.0 66.1
Excess reserve ratio 4/ 17.1 21.8 18.8 5.9 18.4 22.0 22.1
Loan/deposit 42.2 48.7 59.0 64.0 63.9 50.1 56.1
Foreign exchange liabilities/total liabilities 5/ 24.9 21.1 29.7 36.2 27.0 27.4 ...
Sensitivity to market risk
Net foreign exchange assets (liabilities) to shareholders' funds 5/ 62.9 48.1 (7.6) (9.4) 22.9 24.3 ...
Source: IMF Staff Country Report No. 396/03 and Bank of Ghana.
1. The method for calculating CAR is different from that of Basel CAR and is likely to be more conservative than the Basel
method.
2. The ratio of net profit before tax to two-year annual average assets.
3. The ratio of net profit after tax to two-year annual average shareholders' funds.
4. The actual reserve ratio in excess of the minimum requirement ratio.
5. No comparable estimate available for 2003 as commercial banks’ foreign assets and liabilities were reclassified.
- 10 -
Table 4. Profitability Indicators
(In percent of average assets)
Net
interest
Noninterest
income Overhead Provisions RoAA RoAE RoAA
Deflated
RoAE
Ghana 1/ 10.0 6.4 7.0 2.2 6.1 36.9 5.3 22.3
Bank 1 1/ 12.2 4.4 6.4 3.2 6.0 46.1 5.3 31.5
Bank 2 1/ 10.8 6.7 6.2 0.5 9.7 64.3 8.4 49.6
Bank 3 1/ 10.7 6.1 7.4 0.2 7.3 53.8 6.3 39.1
Median
CFA franc zone 2/ 4.6 0.6 4.8 1.0 1.8 17.7 1.8 15.0
Large SSA economies 2/ 5.9 2.5 5.4 1.2 1.4 16.3 1.3 10.5
Small SSA economies 2/ 5.9 1.2 4.6 0.6 2.8 30.1 2.6 20.2
SSA 2/ 5.7 1.2 4.8 0.8 1.9 27.9 1.9 15.0
Sources: IMF, International Financial Statistics; banks’ financial statements, and authors’ estimates.
1/ 2002.
2/ 1998-2001 averages.
C. Possible Factors Explaining Bank Profitability and the Efficiency of Intermediation
At least three factors may have prevented further financial deepening in Ghana so far, and
which may be relevant for the interpretation of both profitability and efficiency indicators of
the banking system. The first factor is macroeconomic policies, as macroeconomic stability is
essential to the development of the financial sector. This is relevant because Ghana’s
macroeconomic policies over the last decade have been characterized by periodic slippages
in financial discipline, leading to volatile and generally high inflation, large exchange rate
swings, and negative real interest rates for extended periods. The most recent example of
macroeconomic imbalances includes the severe terms of trade shock of 1999-2000, which,
combined with fiscal slippages, resulted in inflationary pressures, a 15 percent exchange rate
depreciation, and the buildup of a sizable domestic government debt. It is intuitive to assume
that the high degree of uncertainty associated with Ghana’s unstable macroeconomic
environment has negatively affected both the size and the quality of financial intermediation.
This assumption is supported by the low level of overall savings and investment. As shown
in Figures 3 and 4, Ghana compared rather poorly to other African countries
11
on average in
recent years; however, low bank intermediation seems to coexist with a wide range of
savings ratios, thereby suggesting that other elements may also be at play. Another piece of
evidence is the short time horizon in the overall financial sector. Long-term savings are
virtually inexistent, as one-third of all bank deposits are demand deposits and terms for bank
11
In Ghana, the savings-to-GDP ratio was 15.9 percent on average between 1996 and 2002, while the (private)
investment ratio was 10.6 percent between 1996 and 2001.
- 11 -
loans hardly extend beyond one year. In addition, Treasury bills—which were until recently
also used for open market operations—carry almost exclusively short-term maturities (three
to six months).
12
Together with the high returns offered, this situation has exacerbated the
crowding-out effect on private sector lending.
A second possible factor is the risky lending environment prevailing in Ghana, as reflected in
the high level of past-due/nonperforming loans. This is largely due to the significant losses of
some state-owned companies, but also reflects the lack of any central credit information
system and the lack of cooperation among banks in sharing customer information. Some
institutional factors may also affect the environment in which financial institutions operate.
For instance, as shown in Table 2, the enforcement of creditors’ rights is weak compared
with the sub-Saharan African average. It is important to note that, although nonperforming
loans have some substantial provisioning implications, provisioning standards are lower in
Ghana than in most African countries.
13
Depending on loan classification practices (and
potential rollover of debt), this may suggest that the asset quality of banks’ loan portfolio is
somewhat overestimated, which may act as a further disincentive to engage in financial
intermediation.
A third factor that may account for low and inefficient financial intermediation in Ghana is
the presence of an uncompetitive market structure. Interestingly, there is no one-to-one
relationship between concentration and competition. On the one hand, monopolistic or
oligopolistic behavior tends to result in higher intermediation costs and diseconomies of
management than under a competitive structure; thus, noncompetitive behavior is consistent
with the presence of wide interest rate margins and spreads, which tend to deter potential
depositors, as well as potential borrowers, and result in low lending ratios. On the other hand,
market size may offer the possibility of exploiting economies of scale (from overhead in
administrative operations and information gathering), as well as economies of scope (in
combining different product lines for instance).
14
What really matters for the net effect on
competition is the level of contestability in the market: the threat of potential competition—
or lack thereof—can substantially affect competitiveness conditions, regardless of market
concentration.
12
However, three-year inflation-indexed bonds were introduced in late 2001 along with secondary reserve
requirements that require banks to hold 20 percent of their deposits base in such bonds.
13
In Ghana, nonperforming loans are defined based on a minimum of 180 days in arrears; loans are classified
as “substandard” when they are in arrears for 90 to 180 days, as “doubtful” when they are in arrears for 180 to
540 days, and as “loss-making” when arrears exceed 540 days. Full provisioning is required for loss making,
whereas substandard loans required a 50 percent provisioning.
14
See Vives (2001).
- 12 -
Figure 3. Savings, Investment and Loans Ratios Across Sub-Saharan Africa
(1996-2002, average)*
0
10
20
30
40
50
60
70
0 5 10 15 20 25 30 35 40
Savings to GDP Ratio (%)
L
o
a
n
s
t
o
G
D
P
R
a
t
i
o
(
%
)
Ghana
0
5
10
15
20
25
30
35
0 5 10 15 20 25
Private Investment to GDP Ratio (%)
L
o
a
n
s
t
o
G
D
P
R
a
t
i
o
(
%
)
Ghana
Sources: World Development Indicators database; and IFC Private Investment Trends database.
* Savings refer to gross national savings. Loans are to the private sector only. The definition of “private
investment” is not uniform across sub-Saharan Africa, and public enterprise investment is often reported as
“private investment.”
In the case of Ghana, there are several reasons to question the extent to which banks actually
compete. Although bank concentration appears to be moderate by regional standards, the
dominant state owned bank (“bank 1”) enjoys a substantial market power, with 20 percent of
total deposits and 44 percent of total branches—a situation that may influence price setting
among banks and distort competition. Another potential piece of evidence is the fact that the
dominant state owned bank invariably records the widest interest margin among commercial
banks (12.2 percent in 2002; see Table 4).
However, beyond anecdotal evidence, more analysis is needed to draw some firm
conclusions about the nature of the market structure in Ghana and the extent to which it
offers a plausible explanation of the sector’s profitability. Therefore, the next section
introduces a basic analytical framework to assess the nature of competitive conditions.
III. ANALYTICAL FRAMEWORK AND ECONOMETRIC ESTIMATION
The concept of market contestability has spanned a large theoretical and empirical literature
covering many industries. The basic idea of market contestability is that, on the one hand,
there are several sets of conditions that can yield competitive outcomes, with a competitive
outcome possible even in concentrated systems. On the other hand, collusive actions can be
sustained even in the presence of many firms. The most commonly used models for testing
for the degree of competition are Bresnahan (1989) and Panzar and Rosse (1987). The
Bresnahan model uses the condition of general market equilibrium and rests on the idea that
profit-maximizing firms in equilibrium will choose prices and quantities such that marginal
costs equal their (perceived) marginal revenue, which coincides with the demand price under
perfect competition, or with the industry’s marginal revenue under collusion. The model
generally uses industry aggregates (although firm-specific data is possible) and permits
- 13 -
estimation of a measure of the degree of competition. The Panzar and Rosse model takes a
slightly different route and investigates the extent to which a change in factor input prices is
reflected in (equilibrium) revenues earned by a specific bank in the context of a
Chamberlinian equilibrium model. Like the previous model, the Panzar and Rosse approach
leads to an estimate of the degree of competition. The advantage of the latter is that it uses
bank-level data, allows for bank-specific differences in the production function, and permits
an analysis of the differences between types of banks in terms of size and ownership.
A. The Panzar and Rosse Analytical Framework
Consider the following structural demand and cost relationship facing a particular firm i:
( )
i i i i
z n y R R , , = (1)
( )
i i i i i
x p y C C , , = , (2)
where R = total revenue
C = total costs
y = output
n = number of firms
z = exogenous variable affecting revenue
p = input prices and
x = other exogenous variables, with
all variables are expressed in logarithms,
Profits are defined as ( ) ) , , ( , ,
i i i i i i i i
x p y C z n y R ? = ? , implying that the firm maximizes its
profits where marginal revenue equals marginal costs (equation 3). This means that in
equilibrium, the zero profit constraint holds at the market level as well:
( ) ( )
0
, , ) , ,
=
?
?
?
?
?
i i i i
i
i i i
i
x p y C
C
z n y R
R
. (3)
Profit-maximizing output is defined as equation (4), with an asterisk (*) representing
equilibrium values. Substituting (4) into (1), and assuming that n is endogenously determined
in the model, yields equation (5), which is the reduced-form of the revenue function.
( )
i i i i i
x p z y y , , * * = (4)
( ) ( ) ) , ( * *, , , , * * *
i i i i i i i i i
p z R z n x p z y R R ? = . (5)
Note that market power is measured by the extent to which a change in factor input prices
(?p
i
) is reflected in the equilibrium revenue (?R*
i
) earned by firm i. Panzar and Rosse then
define a measure of competition H as the sum of the elasticities of equation (5) with respect
to input prices, with i denoting a particular firm.
*
*
i
i
i i
i
R
p
p
R
H
?
?
?
= . (6)
- 14 -
According to Panzar and Rosse, it is not just the sign of the H-statistic that matters, but its
magnitude as well. Under a monopolistic structure, an increase in input prices P will increase
marginal cost, thus reducing equilibrium output y* and revenue, thereby implying than the
H-statistic value be less or equal to zero. In contrast, in a perfectly competitive setting in the
long-run, an increase in input prices P will increase marginal cost as well as average costs by
the same proportion, without—under certain assumptions—changing the equilibrium output
of banks. As inefficient banks are forced to exit the market, the increased demand faced by
the remaining firms leads to an increase in output prices and revenues in the same proportion
as costs, thereby implying a value of the H-statistic equal to unity. In the case of
monopolistic competition, described as the most plausible characterization of banks’
interactions by Bikker and Haff (2002b) p.6, under certain assumptions an increase in input
prices P will lead to a less than proportional increase in revenues, as the demand for banking
facing individual banks is inelastic. In this case, the H-Statistic will lie between 0 and 1. The
main discriminatory powers of the H-statistics, as discussed in the literature, are summarized
in Table 5.
Table 5. Panzar and Rosse’s H-Statistics
Values of H Implied Market Structure
H ? 0 Monopoly
Colluding oligopoly, conjectural variations of oligopoly
0 < H < 1 Monopolistic competition
H = 1 Perfect competition
Natural monopoly in a perfectly contestable market
Note that the model is subject to a several assumptions:
• banks are operating in (long-run) equilibrium;
• the performance of the banks is influenced by other participants’ actions (except in
the case of a purely monopolistic structure);
• the cost structure is homogeneous and the production function is a standard Cobb-
Douglas function with a constant return to scale; and
• the price elasticity of demand is greater than unity.
However, the definition of equilibrium is not very clear in the Panzar and Rosse model.
Given the internal logic of the model, it is best to think of equilibrium as a steady-state,
reflecting adjustments to shocks. As noted by Gelos and Roldos (2002), other crucial
assumptions are necessary to apply this analytical framework to the banking sector: first,
banks are generally assumed to behave as single-product firms, using labor, capital and
intermediated funds as inputs; second, input prices are assumed not to be linked to higher
quality services, as the opposite might imply higher revenues, thereby biasing the value of
the H-statistic.
The Panzar and Rosse approach has been extensively used to analyze the nature of
competition in mature banking systems, initially in North America
15
and subsequently in
15
On the U.S. banking system, see Shaffer (1989) and on Canada, see Nathan and Neave (1989).
- 15 -
various European countries and Japan.
16
More recently, the approach has also been applied to
emerging markets’ banking systems
17
or in the context of large cross-country studies.
18
However, there is no published study that we are aware of that examines the case of African
countries, except for Claessens and Leaven (2003), who do include Nigeria and South Africa
in their sample of 50 countries.
In the empirical analysis, let us operationalize equation (5) as follows:
it
k
it
k
k
n it
j
it
j
j
j it
Z Y P LogR ? ? ? µ ? + + + + =
? ?
= =
log log log
1 1
, (7)
with j=3 inputs, so that
j
it
P is a three-dimensional vector of factors prices.
it
Y is a scale
variable,
n
it
Z is a vector of exogenous and bank-specific variables that may shift the revenue
schedule (business mix), ? is a constant term and
it
? is the stochastic error term.
For the dependent variable R, Various authors (Molyneux (1994), Bikker and Groeneveld
(1998), Claessens and Laeven (2003), Levy Yeyati and Micco (2003)) use the ratio of
interest revenue (or alternatively total revenue) to total balance sheets, but as noted by Vesala
(1995), such a specification provides a price equation. Following Gelos and Roldos (2002),
we prefer to estimate two reduced-form revenue equations, one for scaled total revenue, and
one for unscaled total revenue. We also use both total revenue and interest revenue as the
dependent variable to compare results.
H is estimated for the whole sample t, and the H-statistic test is defined as (8):
0
1
= =
?
=
j
j
j t
H µ . (8)
As noted previously, one of the crucial hypotheses of the Panzar and Rosse model is that the
banking sector is assumed to be in equilibrium. As the H-statistics depend on industry-
specific characteristics, cross-country comparisons may be misleading. In practice,
researchers have usually overcome this problem by focusing on and testing for the change in
H over time, or by formally testing the equilibrium hypothesis, even if the definition of what
constitutes equilibrium in the banking sector remains elusive.
19
Notwithstanding this
16
On European countries, see Molyneux et.al. (1994) (France, Germany, Italy, Spain, United Kingdom),
Vesala (1995) (Finland), Coccorese (2002) (Italy), De Brandt and Davis (2000) (France, Germany, Italy), Rime
(1999) (Switzerland), Hondroyiannis et al. (1999) (Greece), Bikker and Groeneveld (1998) (15 EU countries),
Hempell (2002) (Germany), and Maudos and Perez (2002) (Spain). On Japan, see Molyneux et al. (1996).
17
See Gelos and Roldos (2002) (Central Europe and Latin America), Belaisch (2003) (Brazil), Yildirim and
Philippatos (2002) (Central and Eastern Europe), Levi Yeyati and Micco (2003) (Latin America), and
Zambrano Sequin (2003) (Venezuela).
18
See, for example, Claessens and leaven (2003) (50 countries) and Bikker and Haff (2002a) (OECD
countries).
19
See Shaffer (1982), Molyneux et al. (1996), and Claessens and Laeven (2003).
- 16 -
reservation, and following the existing literature, we report the results of the equilibrium
tests, as well as the simple methodology used in Appendix 1. The results appear to indicate
that the Ghanaian banking system was in equilibrium during the period under investigation.
B. Description of the Data and Definitions of Variables
Annual individual bank balance sheets and income statements from 20 banks in operation
during (part of) 1998-2003 have been used to construct the data set. For econometric
estimations, banks that closed (4) or commenced operations (2) during the period have been
dropped, along with one small bank due to data unavailability, leaving 65 observations for
each explanatory variable. Moreover, given that the data used in this estimation concern
institutions operating in the same field of business within the same country, a common effect
specification was chosen for the estimates presented in this paper. Fixed-effect and random
effect models were, however, also estimated yielding similar results.
20
Finally, panel
regressions were run on pooled cross-sections for each year, as well as over the whole sample
period to pick up the time-series components of the data.
21
The variables are defined as follows (all in natural logs):
UPL =
Deposits & Loans Total
Expenses Personnel
UPF =
Deposits Total
Expenses Interest
UPC =
Assets Fixed
Expenses Other
TA = Total assets (scale variable)
Risk component 1 (RC1):
Loans Total
Loans Due Past
Risk component 2 (RC2):
Assets Total
Loans Total
In addition, we have included a dummy variable (dum1) for public ownership (=1) and
another (dum2) for foreign ownership (=1). Finally, the treasury bill rate in nominal (NTBR)
and real (RBTR) terms has been included, as well as inflation (INFL).
20
In fixed effect models, differences between the various members of the pooled dataset are captured by a
constant intercept specific to each member. In random effect models, these differences are assumed to be
random and estimated with the error term in the regression.
21
However, the time series is insufficient to test for stationarity in the summed residuals.
- 17 -
C. Estimation Results
As regards market structure, the results (Table 6) suggest that the Ghanaian banking sector is
characterized by monopolistic competition according to the Panzar and Rosse classification.
Irrespective of model specification, the H-statistic consistently lies between 0 and 1, with a
value of 0.56 on average. Note that there seems to be some volatility in the H-statistics,
especially in the scaled regressions, as shown in Table 6 below. This is not unusual, as
evidenced by other recent studies using the same methodology with different specifications.
22
However, the unscaled specification appears to display more stable results, and allowed to
better assessing the crucial role of the scale variable.
Table 6. H-Statistics Values for the Banking System in Ghana
*
All
Specifications
Unscaled
Specifications
Scaled
Specifications
Average H-statistic 0.555 0.627 0.482
Median H-statistic 0.569 0.626 0.481
Standard deviation 0.092 0.038 0.064
* The H-statistics are computed at the 5 percent significance level.
The market structure identified in Ghana—monopolistic competition—and score appears to
be similar to that of comparable countries in the region (Table 7). Although cross-country
comparison results should be treated with caution, it appears that Ghana’s market structure is
only slightly less competitive that of Nigeria and Kenya, even though the Nigerian banking
sector operates with much narrower interest margins and less foreign penetration than
Ghana’s (Table 2). Note also that the market structure of South Africa is believed to be
significantly more competitive, including by international standards.
Table 7. Banking Sector Market Structure in Selected Countries
Country Period H-
statistic
Nb of
banks
Nb of
observations
Ghana 1998-2003 0.56 13 65
Sub-Saharan Africa
Kenya 1994-2001 0.58 34 106
Nigeria 1994-2001 0.67 42 186
South Africa 1994-2001 0.85 45 186
North America (median) 1994-2001 0.67 3 countries covered
South America (median) 1994-2001 0.73 12 countries covered
East Asia (median) 1994-2001 0.67 6 countries covered
South Asia (median) 1994-2001 0.53 3 countries covered
Western Europe (median) 1994-2001 0.67 14 countries covered
Eastern Europe (median) 1994-2001 0.68 7 countries covered
Sources: Authors’ calculations (Ghana); and Claessens and Laeven (2003), Table 2.
22
See Gelos and Roldos (2002), and Yildrim and Philippatos (2002).
- 18 -
Table 8. Regression Results*
Total Revenue (TR) Total interest revenue
(TIR)
Total Revenue
(TR/TA)
Total interest revenue
(TIR/TA)
C 0.238 -0.171 -1.085** -1.395** 0.483 0.172 0.014 -0.253
t-statistic 0.634 -0.436 -2.763 -3.305 1.908 0.653 0.048 -0.822
UPL 0.254** 0.253 0.293** 0.293** 0.243** 0.239** 0.233** 0.234**
t-statistic 4.265 4.274** 4.676 4.601 4.179 4.109 3.499 3.433
UPF 0.235** 0.248 0.361** 0.373** 0.195** 0.196** 0.189** 0.199**
t-statistic 3.850 4.117** 5.633 5.778 4.668 4.730 3.930 4.094
UPC 0.108** 0.091** 0.056 0.045 0.108** 0.093** 0.057 0.050
t-statistic 3.658 2.969 1.798 1.382 3.670 3.041 1.676 1.390
TA 1.025** 1.033** 1.112** 1.116** ... ... ... ...
t-statistic 36.909 36.511 38.156 36.717 ... ... ... ...
RC1 0.011 0.013 0.003 -0.001 0.012 0.011 0.006 -0.005
t-statistic 0.650 0.683 0.190 -0.045 0.684 0.575 0.319 -0.203
RC2 0.318** 0.294** -0.012 -0.026 0.352** 0.341** 0.147 0.141
t-statistic 4.234 3.864 -0.155 -0.320 5.468 5.283 1.990 1.869
DUM1 -0.090 -0.094 -0.225** -0.221** -0.078 -0.076 -0.181** -0.169**
t-statistic -1.277 -1.327 -3.027 -2.895 -1.128 -1.088 -2.273 -2.081
DUM2 0.100** 0.100** 0.075 0.076 0.108** 0.111** 0.115** 0.117**
t-statistic 2.177 2.208 1.553 1.551 2.415 2.487 2.246 2.237
RTBR … 0.104** … 0.091** … 0.103** … 0.099**
t-statistic 3.326 2.726 3.283 2.703
NTBR 0.348** ... 0.395** ... 0.368** ... 0.490** ...
t-statistic 3.511 ... 3.797 ... 3.818 ... 4.439 ...
INFL ... 0.192** ... 0.235** ... 0.215** ... 0.311**
t-statistic ... 2.759 ... 3.149 ... 3.229 ... 3.984
(**) Statistically significant at the 5 percent level
Memorandum items:
R-squared 0.988 0.988 0.986 0.986 0.691 0.698 0.634 0.627
Adjusted R-squared 0.986 0.986 0.982 0.983 0.647 0.648 0.582 0.566
S.E. of regression 0.145 0.144 0.158 0.156 0.145 0.144 0.166 0.169
F-statistic 486.9 444.9 317.9 327.7 15.657 14.115 12.126 10.277
Prob (F-statistic) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Mean dependent var. 11.7 11.7 11.2 11.2 -1.611 -1.611 -1.956 -1.956
S.D. dependent var. 1.208 1.208 1.196 1.196 0.244 0.244 0.256 0.256
Sum squared resid. 1.157 1.119 1.055 1.024 1.173 1.147 1.541 1.570
Durbin-Watson stat. 1.864 1.850 1.097 1.311 1.883 1.874 1.464 1.447
Market structure Wald test
Ho: H=0 (Monopolostic) Rejected Rejected Rejected Rejected Rejected Rejected Rejected Rejected
Ho: H=1 (Perf.
competition)
Rejected Rejected Rejected Rejected Rejected Rejected Rejected Rejected
Ho: 0
Financial systems tend to evolve around a banking sector seeking to achieve economies of scale in order to offset the costs of collecting and processing information designed to reduce uncertainty, thereby facilitating a more efficient allocation of financial resources.
WP/05/17
Competition and Efficiency in Banking:
Behavioral Evidence from Ghana
Thierry Buchs and Johan Mathisen
©2005 International Monetary Fund WP/05/17
IMF Working Paper
African Department
Competition and Efficiency in Banking: Behavioral Evidence from Ghana
Prepared by Thierry Buchs and J ohan Mathisen
1
Authorized for distribution by Samuel Itam
J anuary 2005
Abstract
This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily represent
those of the IMF or IMF policy. Working Papers describe research in progress by the author(s) and are
published to elicit comments and to further debate.
This paper assesses the degree of bank competition and discusses efficiency with regard to
banks’ financial intermediation in Ghana. By applying panel data to variables derived from a
theoretical model, we find evidence for a noncompetitive market structure in the Ghanaian
banking system, which may be hampering financial intermediation. We argue that the
structure, as well as the other market characteristics, constitutes an indirect barrier to entry
thereby shielding the large profits in the Ghanaian banking system.
J EL Classification Numbers: G21, D43
Keywords: Ghana, banking competition
Author(s) E-Mail Address: [email protected]; [email protected]
1
We would like thank Bank of Ghana staff, Hugh Bredenkamp, J ack Glen, participants at the
Ghana at the Half Century conference in J uly 2004, Luca Ricci and Tom Walter for helpful
comments and suggestions. At the time this paper was prepared, J ohan Mathisen was an
economist in the IMF’s African Department and Thierry Buchs was a Senior Economist at the
International Finance Corporation (IFC). Any remaining errors and omissions are our own.
- 2 -
Contents Page
I. Introduction ........................................................................................................................... 3
II. Overview of the Ghanaian Banking System........................................................................ 4
A. Structure of Ghana’s Banking Sector............................................................................... 4
B. Financial Performance of the Banking Sector.................................................................. 6
C. Possible Factors Explaining Bank Profitability and the Efficiency of Intermediation .. 10
III. Analytical Framework and Econometric Estimation........................................................ 12
A. The Panzar and Rosse Analytical Framework ............................................................... 13
B. Description of the Data and Definitions of Variables .................................................... 16
C. Estimation Results.......................................................................................................... 17
D. Interpretation of the Coefficients ................................................................................... 19
IV. Conclusions....................................................................................................................... 21
References............................................................................................................................... 23
Appendix................................................................................................................................. 25
Figures
1. Nominal Interest Rates, Government Debt, Real Growth of Private Sector Credit,
Private Sector Credit, 1998-2003...................................................................................... 5
2. Investment and (Gross) Lending of the Banking Sector (1996-2003).................................. 6
3. Savings, Investment and Loans Ratios Across Sub-Saharan Africa................................... 12
Tables
1. Structure of the Banking Sector............................................................................................ 4
2. International Comparison of Selected Banking and Institutional Indicators........................ 8
3. Financial Soundness Indicators for the Banking Sector, 1997–2003 ................................... 9
4. Profitability Indicators ........................................................................................................ 10
5. Panzar and Rosse’s H-Statistics.......................................................................................... 14
6. H-Statistics Values for the Banking System in Ghana ...................................................... 17
7. Banking Sector Market Structure in Selected Countries .................................................... 17
8. Regression Results.............................................................................................................. 18
Appendix Table
A1. Market Equilibrium Test Results...................................................................................... 26
- 3 -
I. INTRODUCTION
Financial systems tend to evolve around a banking sector seeking to achieve economies of
scale in order to offset the costs of collecting and processing information designed to reduce
uncertainty, thereby facilitating a more efficient allocation of financial resources. In well-
functioning economies, banks tend to act as quality controllers for capital seeking successful
projects, ensuring higher returns and accelerating output growth. However, a competitive
banking system is required to ensure that banks are effective forces for financial
intermediation channeling savings into investment fostering higher economic growth.
This paper assesses the level of competition in the Ghanaian banking sector. At first sight,
the very high profit ratios and high cost structure of Ghanaian banks could indicate a
monopolistic banking structure. This is partly corroborated by the findings of this paper. By
deriving variables from a theoretical model and using a 1998-2003 panel data set, we find
evidence for a noncompetitive market structure in the Ghanaian banking system, possibly
hampering financial intermediation. This paper argues that the structure, as well as the other
market characteristics, constitutes an indirect barrier to entry thereby shielding the large
profits in the Ghanaian banking system.
Besides the banking sector, the Ghanaian financial system also includes insurance
companies, discount houses, finance houses, leasing companies, savings and loan
associations, credit unions, and a stock exchange.
2
Thus, by narrowing the focus to the
banking sector only, other potentially important participants of the Ghanaian financial system
might have been overlooked. However, the banking system is by far the largest component of
the financial system, and, according to the recent Financial Sector Stability Assessment
(FSSA) update,
3
many of these other financial institutions remain underdeveloped, even by
sub-Saharan African standards. Moreover, this paper defines the banking sector to include
only deposit-taking financial institutions; it excludes rural banks
4
and the Bank of Ghana.
The remainder of the paper is organized as follows. Section II describes the main
characteristics of the structure and features of the banking sector in Ghana, highlighting the
main differences between Ghana and other sub-Saharan African countries. The banks’
financial performance is then discussed, and certain possible explanatory factors for the
performance are outlined. After a very brief literature survey, Section III presents the
theoretical model, operationalizes it by deriving empirical variables, and describes the
dataset. Then the overall results are presented and discussed, followed by an attempt to
investigate the relationships between the factors of production, macroeconomic variables and
revenue and profitability in the banking sector. Section IV summarizes the results and
concludes.
2
For a full description of the Ghanaian financial system, see “Ghana: Selected Issues,” Section II in
Bredenkamp and others (2003).
3
“Ghana: Financial Sector Stability Assessment Update,” IMF Staff Country Report (396/03).
4
The rural banks account for only about 5 percent of banking system assets.
- 4 -
II. OVERVIEW OF THE GHANAIAN BANKING SYSTEM
A. Structure of Ghana’s Banking Sector
The Ghanaian banking system is rather diverse. Of the 17 banks operating in Ghana, there
were 9 commercial banks, 5 merchant banks, and 3 development banks (Table 1).
5
The three
largest commercial banks account for 55 percent of total assets of the banking sector, which
is relatively moderate compared with other countries in the region. However, about
25 percent of total assets and 20 percent of deposits are held by a single state owned
commercial bank (“bank 1”). The development banks and merchant banks, which focus on
medium- and long-term financing and corporate banking, respectively, together share about
30 percent. The five small commercial banks operate on a much smaller scale. Foreign
investors hold about 53 percent of the shares in eight commercial banks, which is below the
sub-Saharan Africa average, and three banks are state-owned (Table 2). The banking
penetration ratio, at one bank branch per 54,000 inhabitants, is relatively high, but formal
banking reaches only 5 percent of the population and the coverage varies widely. This
reflects the fact that 35 percent of bank branches are in the greater Accra region even though
this region represents less than 13 percent of the country’s population. About half of all bank
branches in the interior belong to the dominant state owned bank.
Table 1. Structure of the Banking Sector
*
Ownership (Percent) Share of Total (Percent)
Ghanaian Foreign Total Assets
(Bns of cedis)
As Percent
of GDP
Number of
Branches
Total
assets
Net
lending
Deposits
Banking system 18,668 38.2 309 100.0 100.0 100.0
Commercial banks 13,055 26.7 229 69.3
Bank 1 97 3 4,624 9.46 134 24.8 16.9 20.8
Bank 2 10 90 2,710 5.55 24 14.5 18.1 16.9
Bank 3 24 76 3,011 6.16 23 16.1 16.3 18.8
Bank 4 46 54 1,713 3.50 38 9.2 10.2 8.8
Bank 5 39 61 470 0.96 6 2.5 2.1 2.6
Bank 6 53 47 128 0.26 4 0.7 0.6 1.0
Bank 7 0 100 120 0.25 3 0.6 0.3 0.8
Bank 8 9 91 230 0.47 1 1.2 0.5 1.2
Bank 9 100 0 49 0.10 1 0.3 0.3 0.3
Merchant banks 2,875 5.9 18
Bank 10 100 0 751 1.54 5 4.0 4.8 5.4
Bank 11 6 94 1,325 2.71 4 7.1 8.2 6.2
Bank 12 34 66 409 0.84 3 2.5 2.1 2.6
Bank 13 71 29 286 0.59 2 1.5 1.5 2.0
Bank 14 100 0 104 0.21 1 0.6 0.4 0.7
Development banks 2,738 5.6 62
Bank 15 100 0 1,847 3.78 42 0.0 11.2 8.4
Bank 16 100 0 538 1.10 14 0.0 3.8 2.0
Bank 17 100 0 352 0.72 6 0.0 1.9 2.0
Sources: Bredenkamp and others (2003) and IMF Staff Country Report no. 396/03.
* As of December 2002. The housing bank established in 2003 has been excluded from this study.
5
Commercial banks engage in traditional banking business, with a focus on universal retail services. Merchant
banks are fee-based banking institutions and mostly engage in corporate banking services. Development banks
specialize in the provision of medium- and long-term finance.
- 5 -
As measured by the aggregated total-assets-to-GDP ratio, the banking sector grew rapidly
between 1996 and 2000, reflecting partly financial deepening, as well as loose monetary
conditions. After reaching 44 percent in 2000, the ratio dropped to 38 percent in 2001 and
further to 31 percent at end-2003, reflecting tightened monetary conditions. The same trend
characterized the share of commercial banks’ foreign operations: the share of bank assets
denominated in foreign currency reached 35 percent on 2000 and then declined to 30 percent
in 2001, probably reflecting the increased stability of the cedi exchange rate.
Following the tightening of monetary policy in 2001, domestic credit to the private sector has
remained at around 10 percent of GDP, which is low even by African standards (Table 2).
This essentially reflects a typical crowding-out effect, as most of the banks’ resources are
absorbed by the public sector, either in the form of loans to state-owned enterprises or
holdings of government securities. As shown in Figure 1, increasing government financing
requirements led to very high real treasury bill yields, especially in periods of tight monetary
policy, and by extension, to high lending rates. During 1998-2003, net loans averaged
34 percent of total assets (peaking at 43 percent in 2001), as banks preferred to invest their
resources in liquid, low-risk assets, such as government securities, the latter constituting
about 25 percent of total assets during the period.
6
Figure 1. Nominal Interest Rates, Government Debt, Real Growth of Private Sector Credit,
Private Sector Credit, 1998-2003
0
10
20
30
40
50
60
Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03
(%)
Public Debt to GDP Ratio
Nominal Lending Rate
Nominal T-Bill Rate
-5
0
5
10
15
20
25
30
35
40
Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03
(%)
Private Sector Credit to GDP Ratio
Real Growth of Private Sector Credit
Source: Bank of Ghana
6
Apart from the financing constraints imposed by Ghana’s large fiscal deficits, the banks’ holdings of
government securities is also sustained by high secondary reserve requirements that require banks to hold
35 percent of their deposit liabilities in such securities.
- 6 -
In addition, state-owned enterprises have attracted sizable amounts of lending from
commercial banks recently, thereby exacerbating the crowding-out effect (Figure 2). As a
result, during the last few years, bank lending to the public sector has typically absorbed
more than half of total available resources. The residual resources available for lending to the
private sector (about 35 percent of total assets in 2003) have been mainly channeled to the
manufacturing sector (25 percent of credit to the private sector), commerce and finance
(9 percent) and services (8.5 percent). The agriculture, forestry, and fishing sectors have
received less than one-tenth of total bank credit although agriculture accounts for 36 percent
of GDP. With the exception of the national oil refinery plant—which is the sector’s largest
exposure
7
—no single borrower amounts to 10 percent of the financial sector’s total equity.
Figure 2. Investment and (Gross) Lending of the Banking Sector (1996-2003)
A. Share of Total Assets (in percent) B. Share of Total Lending (in percent)
*
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
1996 1997 1998 1999 2000 2001 2002 2003
Investment in Bills & Securities
Public Sector Loans
Private Sector Loans
Consumer Loans
0%
10%
20%
30%
40%
50%
60%
1996 1997 1998 1999 2000 2001 2002 2003
Total Lending to the Public Sector
Total Lending to the Private Sector
Source: Section II in Bredenkamp and others (2003), Bank of Ghana
* Total lending includes loans, overdrafts and investments.
B. Financial Performance of the Banking Sector
Financial performance indicators portray a mixed picture. On the one hand, the average
capital adequacy ratio (CAR) was about 13.4 percent in 2002 and 9.3 percent at-end 2003,
well above the minimum 6 percent required by law. There was, however, significant
dispersion among banks, and two small commercial banks even failed to meet the minimum
capital standard requirement,
8
prompting intervention by bank supervisors.
7
Even though a large portion of TOR’s short-term debt was restructured into medium-term government bonds
in 2001 and 2002, TOR exposure still exceeded 75 percent of GCB’s equity capital as of June 2003.
8
IMF Staff Country Report no. 396/03.
- 7 -
In addition, as a result of the negative macroeconomic developments in 1999-2000, the asset
quality of the banks’ loan portfolio appears to have deteriorated. Past-due/nonperforming
loans soared from 16.2 percent in 2000 to an eight-year high of 28.6 percent of total loans in
2001 and 2002 before declining slightly to 24.4 percent in 2003 (Table 3). The overall impact
of this sizable increase on the banking system has been partially softened by the relatively
prudent lending of the two largest foreign-owned banks, however. The system is also
characterized by high overhead costs. The five largest banks incur on average overhead costs
of 7 percent to average assets, which is similar to the sector as a whole but substantially
higher than the sub-Saharan African average of 5.7 percent. Note, however, that these costs
are below those reported in Nigeria and Zambia (Table 2). The high costs could partly reflect
substantial investments in banking infrastructure, notably in information and communication
technologies, as telecommunication in particular suffers from interconnectivity problems.
9
It
could also reflect some marketing practices, such as the refusal to network the automated
teller machines, which appears to have led to unduly high investments in such systems.
However, one key element in the total overhead costs is the staff expenditure component
(about 3.7 percent to average assets), which constitutes roughly half of total overhead costs.
For example, the dominant state owned bank (“bank 1”) has one of the highest staff costs
(4.3 percent to average assets), while the other large commercial banks’ average is 3 percent.
This high ratio suggests both a low level of assets per employee and a relatively high average
staff cost per employee.
On the other hand, profitability indicators indicate that, despite high overhead costs and
sizable provisioning, Ghanaian banks’ pretax returns on assets and equity are among the
highest in sub-Saharan Africa (Table 2)—a situation that reflects very wide interest margins.
On an adjusted basis, the return on assets (RoAA) was 6.1 percent in 2002, which is
remarkable even by African standards, and the same applies to both net interest revenue and
noninterest revenue which are, respectively, 10 percent and 6.4 percent of average assets.
The decline in interest rates in 2002 reduced the banks’ income from government securities
and led to a slight narrowing of interest rate spreads, but the latter remain between 20 and
30 percent.
The combination of wide interest margins, sizable overhead costs, and an ample supply of
relatively low-risk, high-return, government paper, has resulted in high costs of
intermediation. Since the large interest margins also reflect the nonperforming loan problem,
the poor quality of banks’ loan portfolios is a major source of concern for the stability of the
system. Most banks would indeed be vulnerable in the event of a major credit risk shock.
10
9
For example, one of the larger, foreign owned banks has set up a direct satellite network to bypass the national
telecommunications network altogether.
10
IMF Staff Country Report No. 396/03.
- 8 -
Table 2. International Comparison of Selected Banking and Institutional Indicators
(In percent, unless otherwise indicated)
Ghana Kenya Mozambique Nigeria South
Africa
Tanzania Uganda Zambia SSA
Average
Size of financial intermediaries
Private credit to GDP 11.8 26.8 16.7 14.4 147.2 4.9 4.0 7.5 15.2
M2 to GDP 19.0 43.8 5.1 25.8 87.2 18.3 13.0 16.9 24.8
Currency to GDP 10.5 13.2 15.6 10.8 28.4 8.5 8.8 6.4 13.9
Banking industry
Number of banks 17 53 10 51 60 29 15 16 ..
Net interest margin 11.5 5.0 5.9 3.8 5.0 6.5 11.6 11.4 8.3
Overhead costs 7.3 3.7 4.5 7.4 3.7 6.7 4.6 11.2 5.7
Foreign bank share (assets) 53 4.8 98 11.0 0.6 58.7 89.0 66.6 ..
Bank concentration (3 banks) 55.0 61.6 76.6 86.5 77 45.8 70.0 81.9 81.0
Nonperforming loans (share of
total loans)
28.8 41.0 .. 17.3 3.9 12.2 6.5 21.8 ..
Capital markets
Stock market capitalization
(percent of GDP) 10.1 9.2 .. 10.9 77.4 4.3 0.6 6.0 21.3
Contract enforcement
Number of procedures 21 25 18 23 16 26 14 1 29
Duration (number of days) 90 255 540 730 99 207 127 188 334
Bankruptcy
Time in years .. 4.6 .. 1.6 2.0 3.0 2.0 3.7 3.5
Credit market
Credit rights index (0 is weakest)
1
1 1 3 1 2 3 1 2 2
Entry regulations
Number of procedures 10 11 16 9 9 13 17 6 11
Duration (number of days) 84 61 153 44 38 35 36 40 72
Cost (percent of GNI per capita) 111 54 100 92 135 9 199 24 255
Sources: IMF, International Finance Statistics; BankScope; World Bank, World Development Indicators;
Doing Business Indicators Database; and Table 2, “Tanzania: Financial System Stability Assessment,” IMF
Staff Country Report No 03/241. Washington DC: IMF (2003). Banking statistics and capital market indicators
are for 2001. All institutional indicators are for 2003.
1/ The index is based on four powers of secured lenders in liquidation and reorganization. A minimum score of
0 represents weak creditor rights and the maximum score of four represents strong creditor rights. For a
description of the methodology, seehttp://rru.worldbank.org/DoingBusiness/Methodology/CreditMarkets.aspx
- 9 -
Table 3. Financial Soundness Indicators for the Banking Sector, 1997–2003
(In percent, at year’s end, unless otherwise indicated)
1997 1998 1999 2000 2001 2002 2003
Capital Adequacy
Regulatory capital to risk-weighted assets 1/ 15.2 11.1 11.5 11.6 14.7 13.4 9.3
Percentage of banks greater or equal to 10 percent 87.5 75.0 60.0 62.5 64.7 52.9 66.7
Percentage of banks below 10 and above 6 percent minimum 6.3 12.5 40.0 37.5 35.3 35.3 27.8
Percentage of banks below 6 percent minimum 6.3 12.5 0.0 0.0 0.0 11.8 5.6
Capital (net worth) to assets 13.4 12.2 12.2 11.9 13.1 12.6 12.5
Asset quality
Foreign exchange loans to total loans 5/ 25.6 28.5 33.4 35.3 34.1 33.8 ...
Past-due loans to gross loans 24.6 18.9 20.1 16.2 28.0 28.6 24.4
Nonperforming loans 21.6 17.2 12.8 11.9 19.6 22.7 18.3
Watch-listed loans 3.0 1.7 7.3 4.3 8.4 5.9 6.0
Provision as percent of past-due loans 78.0 89.4 67.2 58.6 46.4 63.6 64.4
Earnings and profitability
Net profit (before tax)/net income 51.5 39.2 61.2 52.4 45.9 43.4 39.2
Return on assets 2/ 8.0 8.6 8.5 9.7 8.7 6.8 6.4
Return on equity 3/ 39.9 48.9 48.8 65.7 49.7 36.9 54.0
Expense/income
44.0 42.2 44.3 38.2 40.2 47.3 36.0
Interest rate spread (deposit money banks)
Lending rates minus demand deposit rates 37.0 33.8 32.5 30.5 30.5 30.5 23.3
Lending rates minus saving deposit rates 16.3 22.0 23.5 29.3 29.5 25.5 23.0
Liquidity
Actual reserve ratio (as percent of total deposits) 60.1 64.8 61.8 49.9 62.4 66.0 66.1
Excess reserve ratio 4/ 17.1 21.8 18.8 5.9 18.4 22.0 22.1
Loan/deposit 42.2 48.7 59.0 64.0 63.9 50.1 56.1
Foreign exchange liabilities/total liabilities 5/ 24.9 21.1 29.7 36.2 27.0 27.4 ...
Sensitivity to market risk
Net foreign exchange assets (liabilities) to shareholders' funds 5/ 62.9 48.1 (7.6) (9.4) 22.9 24.3 ...
Source: IMF Staff Country Report No. 396/03 and Bank of Ghana.
1. The method for calculating CAR is different from that of Basel CAR and is likely to be more conservative than the Basel
method.
2. The ratio of net profit before tax to two-year annual average assets.
3. The ratio of net profit after tax to two-year annual average shareholders' funds.
4. The actual reserve ratio in excess of the minimum requirement ratio.
5. No comparable estimate available for 2003 as commercial banks’ foreign assets and liabilities were reclassified.
- 10 -
Table 4. Profitability Indicators
(In percent of average assets)
Net
interest
Noninterest
income Overhead Provisions RoAA RoAE RoAA
Deflated
RoAE
Ghana 1/ 10.0 6.4 7.0 2.2 6.1 36.9 5.3 22.3
Bank 1 1/ 12.2 4.4 6.4 3.2 6.0 46.1 5.3 31.5
Bank 2 1/ 10.8 6.7 6.2 0.5 9.7 64.3 8.4 49.6
Bank 3 1/ 10.7 6.1 7.4 0.2 7.3 53.8 6.3 39.1
Median
CFA franc zone 2/ 4.6 0.6 4.8 1.0 1.8 17.7 1.8 15.0
Large SSA economies 2/ 5.9 2.5 5.4 1.2 1.4 16.3 1.3 10.5
Small SSA economies 2/ 5.9 1.2 4.6 0.6 2.8 30.1 2.6 20.2
SSA 2/ 5.7 1.2 4.8 0.8 1.9 27.9 1.9 15.0
Sources: IMF, International Financial Statistics; banks’ financial statements, and authors’ estimates.
1/ 2002.
2/ 1998-2001 averages.
C. Possible Factors Explaining Bank Profitability and the Efficiency of Intermediation
At least three factors may have prevented further financial deepening in Ghana so far, and
which may be relevant for the interpretation of both profitability and efficiency indicators of
the banking system. The first factor is macroeconomic policies, as macroeconomic stability is
essential to the development of the financial sector. This is relevant because Ghana’s
macroeconomic policies over the last decade have been characterized by periodic slippages
in financial discipline, leading to volatile and generally high inflation, large exchange rate
swings, and negative real interest rates for extended periods. The most recent example of
macroeconomic imbalances includes the severe terms of trade shock of 1999-2000, which,
combined with fiscal slippages, resulted in inflationary pressures, a 15 percent exchange rate
depreciation, and the buildup of a sizable domestic government debt. It is intuitive to assume
that the high degree of uncertainty associated with Ghana’s unstable macroeconomic
environment has negatively affected both the size and the quality of financial intermediation.
This assumption is supported by the low level of overall savings and investment. As shown
in Figures 3 and 4, Ghana compared rather poorly to other African countries
11
on average in
recent years; however, low bank intermediation seems to coexist with a wide range of
savings ratios, thereby suggesting that other elements may also be at play. Another piece of
evidence is the short time horizon in the overall financial sector. Long-term savings are
virtually inexistent, as one-third of all bank deposits are demand deposits and terms for bank
11
In Ghana, the savings-to-GDP ratio was 15.9 percent on average between 1996 and 2002, while the (private)
investment ratio was 10.6 percent between 1996 and 2001.
- 11 -
loans hardly extend beyond one year. In addition, Treasury bills—which were until recently
also used for open market operations—carry almost exclusively short-term maturities (three
to six months).
12
Together with the high returns offered, this situation has exacerbated the
crowding-out effect on private sector lending.
A second possible factor is the risky lending environment prevailing in Ghana, as reflected in
the high level of past-due/nonperforming loans. This is largely due to the significant losses of
some state-owned companies, but also reflects the lack of any central credit information
system and the lack of cooperation among banks in sharing customer information. Some
institutional factors may also affect the environment in which financial institutions operate.
For instance, as shown in Table 2, the enforcement of creditors’ rights is weak compared
with the sub-Saharan African average. It is important to note that, although nonperforming
loans have some substantial provisioning implications, provisioning standards are lower in
Ghana than in most African countries.
13
Depending on loan classification practices (and
potential rollover of debt), this may suggest that the asset quality of banks’ loan portfolio is
somewhat overestimated, which may act as a further disincentive to engage in financial
intermediation.
A third factor that may account for low and inefficient financial intermediation in Ghana is
the presence of an uncompetitive market structure. Interestingly, there is no one-to-one
relationship between concentration and competition. On the one hand, monopolistic or
oligopolistic behavior tends to result in higher intermediation costs and diseconomies of
management than under a competitive structure; thus, noncompetitive behavior is consistent
with the presence of wide interest rate margins and spreads, which tend to deter potential
depositors, as well as potential borrowers, and result in low lending ratios. On the other hand,
market size may offer the possibility of exploiting economies of scale (from overhead in
administrative operations and information gathering), as well as economies of scope (in
combining different product lines for instance).
14
What really matters for the net effect on
competition is the level of contestability in the market: the threat of potential competition—
or lack thereof—can substantially affect competitiveness conditions, regardless of market
concentration.
12
However, three-year inflation-indexed bonds were introduced in late 2001 along with secondary reserve
requirements that require banks to hold 20 percent of their deposits base in such bonds.
13
In Ghana, nonperforming loans are defined based on a minimum of 180 days in arrears; loans are classified
as “substandard” when they are in arrears for 90 to 180 days, as “doubtful” when they are in arrears for 180 to
540 days, and as “loss-making” when arrears exceed 540 days. Full provisioning is required for loss making,
whereas substandard loans required a 50 percent provisioning.
14
See Vives (2001).
- 12 -
Figure 3. Savings, Investment and Loans Ratios Across Sub-Saharan Africa
(1996-2002, average)*
0
10
20
30
40
50
60
70
0 5 10 15 20 25 30 35 40
Savings to GDP Ratio (%)
L
o
a
n
s
t
o
G
D
P
R
a
t
i
o
(
%
)
Ghana
0
5
10
15
20
25
30
35
0 5 10 15 20 25
Private Investment to GDP Ratio (%)
L
o
a
n
s
t
o
G
D
P
R
a
t
i
o
(
%
)
Ghana
Sources: World Development Indicators database; and IFC Private Investment Trends database.
* Savings refer to gross national savings. Loans are to the private sector only. The definition of “private
investment” is not uniform across sub-Saharan Africa, and public enterprise investment is often reported as
“private investment.”
In the case of Ghana, there are several reasons to question the extent to which banks actually
compete. Although bank concentration appears to be moderate by regional standards, the
dominant state owned bank (“bank 1”) enjoys a substantial market power, with 20 percent of
total deposits and 44 percent of total branches—a situation that may influence price setting
among banks and distort competition. Another potential piece of evidence is the fact that the
dominant state owned bank invariably records the widest interest margin among commercial
banks (12.2 percent in 2002; see Table 4).
However, beyond anecdotal evidence, more analysis is needed to draw some firm
conclusions about the nature of the market structure in Ghana and the extent to which it
offers a plausible explanation of the sector’s profitability. Therefore, the next section
introduces a basic analytical framework to assess the nature of competitive conditions.
III. ANALYTICAL FRAMEWORK AND ECONOMETRIC ESTIMATION
The concept of market contestability has spanned a large theoretical and empirical literature
covering many industries. The basic idea of market contestability is that, on the one hand,
there are several sets of conditions that can yield competitive outcomes, with a competitive
outcome possible even in concentrated systems. On the other hand, collusive actions can be
sustained even in the presence of many firms. The most commonly used models for testing
for the degree of competition are Bresnahan (1989) and Panzar and Rosse (1987). The
Bresnahan model uses the condition of general market equilibrium and rests on the idea that
profit-maximizing firms in equilibrium will choose prices and quantities such that marginal
costs equal their (perceived) marginal revenue, which coincides with the demand price under
perfect competition, or with the industry’s marginal revenue under collusion. The model
generally uses industry aggregates (although firm-specific data is possible) and permits
- 13 -
estimation of a measure of the degree of competition. The Panzar and Rosse model takes a
slightly different route and investigates the extent to which a change in factor input prices is
reflected in (equilibrium) revenues earned by a specific bank in the context of a
Chamberlinian equilibrium model. Like the previous model, the Panzar and Rosse approach
leads to an estimate of the degree of competition. The advantage of the latter is that it uses
bank-level data, allows for bank-specific differences in the production function, and permits
an analysis of the differences between types of banks in terms of size and ownership.
A. The Panzar and Rosse Analytical Framework
Consider the following structural demand and cost relationship facing a particular firm i:
( )
i i i i
z n y R R , , = (1)
( )
i i i i i
x p y C C , , = , (2)
where R = total revenue
C = total costs
y = output
n = number of firms
z = exogenous variable affecting revenue
p = input prices and
x = other exogenous variables, with
all variables are expressed in logarithms,
Profits are defined as ( ) ) , , ( , ,
i i i i i i i i
x p y C z n y R ? = ? , implying that the firm maximizes its
profits where marginal revenue equals marginal costs (equation 3). This means that in
equilibrium, the zero profit constraint holds at the market level as well:
( ) ( )
0
, , ) , ,
=
?
?
?
?
?
i i i i
i
i i i
i
x p y C
C
z n y R
R
. (3)
Profit-maximizing output is defined as equation (4), with an asterisk (*) representing
equilibrium values. Substituting (4) into (1), and assuming that n is endogenously determined
in the model, yields equation (5), which is the reduced-form of the revenue function.
( )
i i i i i
x p z y y , , * * = (4)
( ) ( ) ) , ( * *, , , , * * *
i i i i i i i i i
p z R z n x p z y R R ? = . (5)
Note that market power is measured by the extent to which a change in factor input prices
(?p
i
) is reflected in the equilibrium revenue (?R*
i
) earned by firm i. Panzar and Rosse then
define a measure of competition H as the sum of the elasticities of equation (5) with respect
to input prices, with i denoting a particular firm.
*
*
i
i
i i
i
R
p
p
R
H
?
?
?
= . (6)
- 14 -
According to Panzar and Rosse, it is not just the sign of the H-statistic that matters, but its
magnitude as well. Under a monopolistic structure, an increase in input prices P will increase
marginal cost, thus reducing equilibrium output y* and revenue, thereby implying than the
H-statistic value be less or equal to zero. In contrast, in a perfectly competitive setting in the
long-run, an increase in input prices P will increase marginal cost as well as average costs by
the same proportion, without—under certain assumptions—changing the equilibrium output
of banks. As inefficient banks are forced to exit the market, the increased demand faced by
the remaining firms leads to an increase in output prices and revenues in the same proportion
as costs, thereby implying a value of the H-statistic equal to unity. In the case of
monopolistic competition, described as the most plausible characterization of banks’
interactions by Bikker and Haff (2002b) p.6, under certain assumptions an increase in input
prices P will lead to a less than proportional increase in revenues, as the demand for banking
facing individual banks is inelastic. In this case, the H-Statistic will lie between 0 and 1. The
main discriminatory powers of the H-statistics, as discussed in the literature, are summarized
in Table 5.
Table 5. Panzar and Rosse’s H-Statistics
Values of H Implied Market Structure
H ? 0 Monopoly
Colluding oligopoly, conjectural variations of oligopoly
0 < H < 1 Monopolistic competition
H = 1 Perfect competition
Natural monopoly in a perfectly contestable market
Note that the model is subject to a several assumptions:
• banks are operating in (long-run) equilibrium;
• the performance of the banks is influenced by other participants’ actions (except in
the case of a purely monopolistic structure);
• the cost structure is homogeneous and the production function is a standard Cobb-
Douglas function with a constant return to scale; and
• the price elasticity of demand is greater than unity.
However, the definition of equilibrium is not very clear in the Panzar and Rosse model.
Given the internal logic of the model, it is best to think of equilibrium as a steady-state,
reflecting adjustments to shocks. As noted by Gelos and Roldos (2002), other crucial
assumptions are necessary to apply this analytical framework to the banking sector: first,
banks are generally assumed to behave as single-product firms, using labor, capital and
intermediated funds as inputs; second, input prices are assumed not to be linked to higher
quality services, as the opposite might imply higher revenues, thereby biasing the value of
the H-statistic.
The Panzar and Rosse approach has been extensively used to analyze the nature of
competition in mature banking systems, initially in North America
15
and subsequently in
15
On the U.S. banking system, see Shaffer (1989) and on Canada, see Nathan and Neave (1989).
- 15 -
various European countries and Japan.
16
More recently, the approach has also been applied to
emerging markets’ banking systems
17
or in the context of large cross-country studies.
18
However, there is no published study that we are aware of that examines the case of African
countries, except for Claessens and Leaven (2003), who do include Nigeria and South Africa
in their sample of 50 countries.
In the empirical analysis, let us operationalize equation (5) as follows:
it
k
it
k
k
n it
j
it
j
j
j it
Z Y P LogR ? ? ? µ ? + + + + =
? ?
= =
log log log
1 1
, (7)
with j=3 inputs, so that
j
it
P is a three-dimensional vector of factors prices.
it
Y is a scale
variable,
n
it
Z is a vector of exogenous and bank-specific variables that may shift the revenue
schedule (business mix), ? is a constant term and
it
? is the stochastic error term.
For the dependent variable R, Various authors (Molyneux (1994), Bikker and Groeneveld
(1998), Claessens and Laeven (2003), Levy Yeyati and Micco (2003)) use the ratio of
interest revenue (or alternatively total revenue) to total balance sheets, but as noted by Vesala
(1995), such a specification provides a price equation. Following Gelos and Roldos (2002),
we prefer to estimate two reduced-form revenue equations, one for scaled total revenue, and
one for unscaled total revenue. We also use both total revenue and interest revenue as the
dependent variable to compare results.
H is estimated for the whole sample t, and the H-statistic test is defined as (8):
0
1
= =
?
=
j
j
j t
H µ . (8)
As noted previously, one of the crucial hypotheses of the Panzar and Rosse model is that the
banking sector is assumed to be in equilibrium. As the H-statistics depend on industry-
specific characteristics, cross-country comparisons may be misleading. In practice,
researchers have usually overcome this problem by focusing on and testing for the change in
H over time, or by formally testing the equilibrium hypothesis, even if the definition of what
constitutes equilibrium in the banking sector remains elusive.
19
Notwithstanding this
16
On European countries, see Molyneux et.al. (1994) (France, Germany, Italy, Spain, United Kingdom),
Vesala (1995) (Finland), Coccorese (2002) (Italy), De Brandt and Davis (2000) (France, Germany, Italy), Rime
(1999) (Switzerland), Hondroyiannis et al. (1999) (Greece), Bikker and Groeneveld (1998) (15 EU countries),
Hempell (2002) (Germany), and Maudos and Perez (2002) (Spain). On Japan, see Molyneux et al. (1996).
17
See Gelos and Roldos (2002) (Central Europe and Latin America), Belaisch (2003) (Brazil), Yildirim and
Philippatos (2002) (Central and Eastern Europe), Levi Yeyati and Micco (2003) (Latin America), and
Zambrano Sequin (2003) (Venezuela).
18
See, for example, Claessens and leaven (2003) (50 countries) and Bikker and Haff (2002a) (OECD
countries).
19
See Shaffer (1982), Molyneux et al. (1996), and Claessens and Laeven (2003).
- 16 -
reservation, and following the existing literature, we report the results of the equilibrium
tests, as well as the simple methodology used in Appendix 1. The results appear to indicate
that the Ghanaian banking system was in equilibrium during the period under investigation.
B. Description of the Data and Definitions of Variables
Annual individual bank balance sheets and income statements from 20 banks in operation
during (part of) 1998-2003 have been used to construct the data set. For econometric
estimations, banks that closed (4) or commenced operations (2) during the period have been
dropped, along with one small bank due to data unavailability, leaving 65 observations for
each explanatory variable. Moreover, given that the data used in this estimation concern
institutions operating in the same field of business within the same country, a common effect
specification was chosen for the estimates presented in this paper. Fixed-effect and random
effect models were, however, also estimated yielding similar results.
20
Finally, panel
regressions were run on pooled cross-sections for each year, as well as over the whole sample
period to pick up the time-series components of the data.
21
The variables are defined as follows (all in natural logs):
UPL =
Deposits & Loans Total
Expenses Personnel
UPF =
Deposits Total
Expenses Interest
UPC =
Assets Fixed
Expenses Other
TA = Total assets (scale variable)
Risk component 1 (RC1):
Loans Total
Loans Due Past
Risk component 2 (RC2):
Assets Total
Loans Total
In addition, we have included a dummy variable (dum1) for public ownership (=1) and
another (dum2) for foreign ownership (=1). Finally, the treasury bill rate in nominal (NTBR)
and real (RBTR) terms has been included, as well as inflation (INFL).
20
In fixed effect models, differences between the various members of the pooled dataset are captured by a
constant intercept specific to each member. In random effect models, these differences are assumed to be
random and estimated with the error term in the regression.
21
However, the time series is insufficient to test for stationarity in the summed residuals.
- 17 -
C. Estimation Results
As regards market structure, the results (Table 6) suggest that the Ghanaian banking sector is
characterized by monopolistic competition according to the Panzar and Rosse classification.
Irrespective of model specification, the H-statistic consistently lies between 0 and 1, with a
value of 0.56 on average. Note that there seems to be some volatility in the H-statistics,
especially in the scaled regressions, as shown in Table 6 below. This is not unusual, as
evidenced by other recent studies using the same methodology with different specifications.
22
However, the unscaled specification appears to display more stable results, and allowed to
better assessing the crucial role of the scale variable.
Table 6. H-Statistics Values for the Banking System in Ghana
*
All
Specifications
Unscaled
Specifications
Scaled
Specifications
Average H-statistic 0.555 0.627 0.482
Median H-statistic 0.569 0.626 0.481
Standard deviation 0.092 0.038 0.064
* The H-statistics are computed at the 5 percent significance level.
The market structure identified in Ghana—monopolistic competition—and score appears to
be similar to that of comparable countries in the region (Table 7). Although cross-country
comparison results should be treated with caution, it appears that Ghana’s market structure is
only slightly less competitive that of Nigeria and Kenya, even though the Nigerian banking
sector operates with much narrower interest margins and less foreign penetration than
Ghana’s (Table 2). Note also that the market structure of South Africa is believed to be
significantly more competitive, including by international standards.
Table 7. Banking Sector Market Structure in Selected Countries
Country Period H-
statistic
Nb of
banks
Nb of
observations
Ghana 1998-2003 0.56 13 65
Sub-Saharan Africa
Kenya 1994-2001 0.58 34 106
Nigeria 1994-2001 0.67 42 186
South Africa 1994-2001 0.85 45 186
North America (median) 1994-2001 0.67 3 countries covered
South America (median) 1994-2001 0.73 12 countries covered
East Asia (median) 1994-2001 0.67 6 countries covered
South Asia (median) 1994-2001 0.53 3 countries covered
Western Europe (median) 1994-2001 0.67 14 countries covered
Eastern Europe (median) 1994-2001 0.68 7 countries covered
Sources: Authors’ calculations (Ghana); and Claessens and Laeven (2003), Table 2.
22
See Gelos and Roldos (2002), and Yildrim and Philippatos (2002).
- 18 -
Table 8. Regression Results*
Total Revenue (TR) Total interest revenue
(TIR)
Total Revenue
(TR/TA)
Total interest revenue
(TIR/TA)
C 0.238 -0.171 -1.085** -1.395** 0.483 0.172 0.014 -0.253
t-statistic 0.634 -0.436 -2.763 -3.305 1.908 0.653 0.048 -0.822
UPL 0.254** 0.253 0.293** 0.293** 0.243** 0.239** 0.233** 0.234**
t-statistic 4.265 4.274** 4.676 4.601 4.179 4.109 3.499 3.433
UPF 0.235** 0.248 0.361** 0.373** 0.195** 0.196** 0.189** 0.199**
t-statistic 3.850 4.117** 5.633 5.778 4.668 4.730 3.930 4.094
UPC 0.108** 0.091** 0.056 0.045 0.108** 0.093** 0.057 0.050
t-statistic 3.658 2.969 1.798 1.382 3.670 3.041 1.676 1.390
TA 1.025** 1.033** 1.112** 1.116** ... ... ... ...
t-statistic 36.909 36.511 38.156 36.717 ... ... ... ...
RC1 0.011 0.013 0.003 -0.001 0.012 0.011 0.006 -0.005
t-statistic 0.650 0.683 0.190 -0.045 0.684 0.575 0.319 -0.203
RC2 0.318** 0.294** -0.012 -0.026 0.352** 0.341** 0.147 0.141
t-statistic 4.234 3.864 -0.155 -0.320 5.468 5.283 1.990 1.869
DUM1 -0.090 -0.094 -0.225** -0.221** -0.078 -0.076 -0.181** -0.169**
t-statistic -1.277 -1.327 -3.027 -2.895 -1.128 -1.088 -2.273 -2.081
DUM2 0.100** 0.100** 0.075 0.076 0.108** 0.111** 0.115** 0.117**
t-statistic 2.177 2.208 1.553 1.551 2.415 2.487 2.246 2.237
RTBR … 0.104** … 0.091** … 0.103** … 0.099**
t-statistic 3.326 2.726 3.283 2.703
NTBR 0.348** ... 0.395** ... 0.368** ... 0.490** ...
t-statistic 3.511 ... 3.797 ... 3.818 ... 4.439 ...
INFL ... 0.192** ... 0.235** ... 0.215** ... 0.311**
t-statistic ... 2.759 ... 3.149 ... 3.229 ... 3.984
(**) Statistically significant at the 5 percent level
Memorandum items:
R-squared 0.988 0.988 0.986 0.986 0.691 0.698 0.634 0.627
Adjusted R-squared 0.986 0.986 0.982 0.983 0.647 0.648 0.582 0.566
S.E. of regression 0.145 0.144 0.158 0.156 0.145 0.144 0.166 0.169
F-statistic 486.9 444.9 317.9 327.7 15.657 14.115 12.126 10.277
Prob (F-statistic) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Mean dependent var. 11.7 11.7 11.2 11.2 -1.611 -1.611 -1.956 -1.956
S.D. dependent var. 1.208 1.208 1.196 1.196 0.244 0.244 0.256 0.256
Sum squared resid. 1.157 1.119 1.055 1.024 1.173 1.147 1.541 1.570
Durbin-Watson stat. 1.864 1.850 1.097 1.311 1.883 1.874 1.464 1.447
Market structure Wald test
Ho: H=0 (Monopolostic) Rejected Rejected Rejected Rejected Rejected Rejected Rejected Rejected
Ho: H=1 (Perf.
competition)
Rejected Rejected Rejected Rejected Rejected Rejected Rejected Rejected
Ho: 0