TOPIC

TOPIC:
The Determinants of Banking Crises
A Case Study of Sub-Saharan Africa

Master Thesis Submitted to the Institute of Economics, Faculty of Humanities
(Philosophische Fakultät Und Fachbereich Theologie),
Friedrich-Alexander Universität, Erlangen –Nürnberg.

Best services for writing your paper according to Trustpilot

Premium Partner
From $18.00 per page
4,8 / 5
4,80
Writers Experience
4,80
Delivery
4,90
Support
4,70
Price
Recommended Service
From $13.90 per page
4,6 / 5
4,70
Writers Experience
4,70
Delivery
4,60
Support
4,60
Price
From $20.00 per page
4,5 / 5
4,80
Writers Experience
4,50
Delivery
4,40
Support
4,10
Price
* All Partners were chosen among 50+ writing services by our Customer Satisfaction Team

Student’s Name: Francis Ntiamoah Boateng
Matriculation number: 22072515
Address: Offenbacher Landastrasse 357a, 60599 –
Frankfurt am Main, Germany.
Phone: 0049 176355881517
E-Mail: [email protected]

1st Supervisor: Prof. Jürgen Kähler PhD
2nd Supervisor: Mr. Christoph Weber (M.A)

Page | i

TABLE OF CONTENTS
Contents Pages
Table of Content……………………………………………………………………………… i
List of Tables…………………………………………………………………………………. ii
List of Figures………………………………………………………………………………….. iii
Acronyms…………………………………………………………………………………….. iv

CHAPTER ONE
GENERAL INTRODUCTION …………………………………………………………………………………….. 1
1.0 Introduction ………………………………………………………………………………………………………. 1
1.2 The problem: Banking crises in Sub Saharan Africa……………………………………………….. 2
1.3 Research Question ……………………………………………………………………………………………… 3
1.4 Why is your study relevant? ………………………………………………………………………………… 3
1.5 Organisation of the study …………………………………………………………………………………….. 4

CHAPTER TWO
CONCEPTUALISATION OF BANKING CRISES ……………………………………………………….. 5
2.0 Introduction ………………………………………………………………………………………………………. 5
2.1 Banking Sector Theory ……………………………………………………………………………………….. 6
2.2 Previous Studies on Banking Crisis ………………………………………………………………………. 8
2.3 Banking Crises in the History ………………………………………………………………………………. 9
2.4 Frequency of Recent Banking Crises ………………………………………………………………….. 11
2.5 Banking Crisis Cycles ………………………………………………………………………………………. 12
2.6 The Experience of Banking Crises in some Sub-Saharan African Countries ……………. 13
2.7 Patterns of Banking Crises ………………………………………………………………………………… 16
2.7.1 Bank Run ………………………………………………………………………………………………….. 17
2.7.2 Bank Panic ………………………………………………………………………………………………… 18
2.7.3 Systematic Bank Crisis ……………………………………………………………………………….. 19
2.8 Causes of Banking Crises – A Brief Literature Review …………………………………………. 20
2.9 Theoretical Evidence on the Causes of Banking Crisis ………………………………………….. 23

CHAPTER THREE
DATA AND METHODOLOGY ………………………………………………………………………………… 26
3.0 Introduction …………………………………………………………………………………………………….. 26

Page | ii

3.1 Source of Data …………………………………………………………………………………………………. 26
3.2 Sample ……………………………………………………………………………………………………………. 26
3.3 Dependent Variable (Banking Crisis) Specifications …………………………………………….. 27
3.4 Explanatory Variables Specifications ………………………………………………………………….. 29
3.5 Methodology ……………………………………………………………………………………………………. 33

CHAPTER FOUR
EMPIRICAL RESULTS AND ANALYSIS …………………………………………………………………. 35
4.0 Introduction ………………………………………………………….. Error! Bookmark not defined.
4.1 Descriptive Statistics ………………………………………………………………………………………… 35
4.2 Main Results and Interpretations…………………………………………………………………………40
4.3 Alternative Method of The Regression Analysis………………………………………43

CHAPTER FIVE
SUMMARY AND CONCLUSION …………………………………………………………………………….. 47
5.0 …………………………………………………………………………………………………………………………… 47
REFERENCES …………………………………………………………………………………………………………. 48
APPENDIX……………………………………………………………………………………………………………….49
DECLARATION………………………………………………………………………………………………………………..54

Page | iii

List of Tables
Table 1. The Cost of Banking Crises in Some Selected Countries of the World
Table 2. Episodes of Banking Crises in SSA Region (the 1980s – 1990s)
Table 3. Sub-Sahara African Countries and their Banking Crises Period
Table 4. Description of Explanatory Variables and their Sources of Data
Table 5. Summary Statistics of Variables
Table 6. Regression Results from Multivariate Logit Models
Table 7. Regression Results from Panel Excluding Countries with No Banking Crises Episode
Table 8. Regression Results of Panel Dataset of Countries with Repeated Crises Episodes

List of Figures
Figure 1. Frequency of Recent Banking Crises Around the Globe (1970 -2011)
Figure 2. Banking Crises Cycles in Recent History

Page | iv

List of Acronyms and Abbreviations
SSA Sub-Saharan Africa
GDP Gross Domestic Product
LLP Loan Loss Provisioning
CRA Credit Risk Analysis
SME Small and Medium Enterprise
OECD Organization of Economic Co-operation and Development
AD Anno Domini Calendar Era
US United States
IMF International Monetary Fund
WDI World Development Indicator
GFDD Global Financial Development Database
IFS International Financial Statistics
WEO World Economic Organisation
BvB Bureau van Dijk
FSB Financial Stability Board
M2/reserves Broad Money Reserve to GDP Ratio

Page | 1

CHAPTER ONE
GENERAL INTRODUCTION
1.0 Introduction
Under which circumstances do banks fall into financial distress and how can we specifically
explain the dynamics of banking fragility in developing economies? Theoretical and applied
research on the causal mechanisms underlying banking crises and remedial policy interventions
have increased tremendously since the financial crises that hit many developed, developing and
transition economies in the 1980s and early 1990s. Over the years, works from such scholars
as Caprio and Klingebiel (1996), Kaminsky and Reinhart (1999), Frydl (1999), Demirgüç-Kunt
and Detragiache (1998, 2005) have greatly advanced scientific optimism regarding the
possibility of preventing any future banking crises. In much the same way, scholarship on how
to accelerate recovery from banking crises has blossomed.
However, a closer look at this burgeoning literature reveals at least three basic issues of interest.
First, while there have been tons of work on how best to solve banking crises and accelerate
recovery, scholars and policy workers alike are yet to arrive at a consensus on what constitutes
best practices. Usually, varied and sometimes, contradictory remedial measures are proposed.
It is still contentious if a one-size-fit all consensus could be reached on how to avert and remedy
banking crises. For many scholars, the adoption of effective supervision and regulatory
frameworks is the way to go (see Goldstein, 2017 for example). However, this line of reasoning
has strongly been called into question. Barth, Davis and Levine, (2008) for instance provided
evidence to show that these regulatory and supervisory practices have had minimal to no results
in substantially changing crises economies over the past years. Second, a comparatively lesser
attention has been paid to the background conditions that fertile the grounds for banks to fall
into crises. While this neglect makes sense logically1, it does not suffice that the determinants
of banking crises elude adequate scientific inquiry. Lastly, we have comparatively limited
understanding of the dynamics of banking crises and its determinants in emerging economies
in Sub Saharan Africa than its developed counterparts. Hence there is always the temptation to
overlook contextual specificities and employ alien theories and propositions in diagnosing
banking fragilities in developing countries.
1 When banks fall into financial difficulties, the most important attention is usually placed on how to resuscitate
them and not necessarily why it landed into crises. This logic of remedy instead of “priories” could explain why
scientific and policy attention is higher rather on crises recovery.

Page | 2

Perhaps an instance would suffice here: policymakers in many advanced economies have come
out with several interventions to prevent banking crises. Notable amongst them is the Keynesian
approach, which advocates for the provision of strong monetary and fiscal stimulus as well as
bailout mechanism to strengthen the private sector from any catastrophic job losses (see (Beck,
Dewatripont, Freixas, Seabright, ; Coyle, 2010 for instance). The approach emphasizes that
such interventions are carefully designed to take care of insolvent banks and accelerate their
recovery. However, beyond the fact that such suggestions are associated with their own
setbacks such as high budgetary cost and possible inefficiency2, they sometimes display a lack
of contextual understanding (Demirgüc-Kunt ; Detragiache, 2005). In effect, banking crises
are case-specific and thus require tailor-made remedies.
Given the foregone, the immediate question that arises is; what determines the survival of banks
in emerging economies in Sub Sahara Africa and under which circumstances do some banks
fall into crises while others do not. While the literature has generated sufficient knowledge in
the areas of financial markets and the development of monetary instruments in sub Saharan
Africa, the attention to banking crises remains minimal. With the exception of Honohan (1993)
and Brownsbridge’s (1998) cross-sectional seminal contributions on the failures in the financial
sector in Africa and a handful of isolated case studies3, the study of banking distress has
developed quite slowly. Hence it would be of interest to bring into perspective the factors and
the circumstances under which banking distress becomes eminent in Sub Saharan Africa.

1.2 The problem: Banking crises in Sub-Saharan Africa
Banks in Sub-Saharan Africa have over the years been characterised by what Roland Daumont
and his colleagues call “severe crises of an eminent character” (Daumont, Le Gall, & Leroux,
2004, p. 6). Specifically, in the 1980s and 1990s, African countries such as Benin, Cameroon,
Côte d’Ivoire, Ghana, Guinea, Kenya, Nigeria, Senegal, Tanzania and Uganda experienced
some form of banking crises with varying degrees of severity. Most of these crises took the
shape of currency crises and sovereign foreign debts; a situation that affected the banking
systems and gave it an inflated cost of financial sector growth (Mezui, Nalletamby, ; Kamewe,
2012). Yet, despite the near collapse of the banking sector in many African countries, it has
2 Demirgüc-Kunt ; Detragiache (1998) for instance argue that many of the rescue strategies employed by policy
practitioners in resuscitating insolvent banks like monetary bailouts actually increases banking inefficiency and
keep these inefficient banks in business. 3 See for instance Tenconi (1993) and his discussion of the banking crisis of the mid-1980s in Guinea.

Page | 3

drawn relatively little attention. As Daumont et al. (2004) surmised, unlike the banking crises
in developed countries that resulted in a significant impact on their economies, the banking
crises in Sub-Saharan Africa was mostly intervened by governments of these countries and the
impact was very little and this might have given the crisis of Sub-Saharan Africa less attention
(see also Brownsbridge and Harvey, 1998).
However, one thing to apply our attentions to is the fact that the disturbing impact of banking
crisis is not only limited to the country or region that it began but could rather spread out to its
neighboring regions and economically linked countries due to cross-border banking linkages
(Allen ; Gale, 2007). There is therefore an imperative need to investigate and identify the types
of banking crises, what causes them and draw useful policy solutions to avoid same in Sub
Saharan Africa.

1.3 Research Question
Given the above, this thesis is guided by the question: Under which circumstances do banks in
Sub-Saharan Africa become fragile and how can we explain the factors underlying such crises?
For purposes of analysis, the research question is disaggregated into three components. The
first step would seek to examine some factors that make banks in Sub-Saharan Africa prone to
fragility and crises. The analysis would specifically centre on 48 countries in the SSA region.
A multivariate (binary) logit model will be adopted to analyse which of these factors possess a
comparatively stronger influence on the probability of banking crises in the region. Secondly,
comparisons will be constructed with the assumption that different causal factors possess varied
weight and thus require analytical and empirical distinction. In this case another logit mode is
constructed to exclude countries without experience of banking crises. Finally, the paper will
examine whether the identified factors of banking crises behave different in economies with
repeated banking crises episodes.

1.4 Why is your study relevant?
Why is the study of banking crises important and more specifically why should we be concerned
about its dynamics in emerging economies? For the start, Banking crises in developing
economies tend to have serious consequences on the local economy. There is therefore the need
to establish empirically what factors necessitate such crises and proffer context-relevant

Page | 4

remedies. As already highlighted in the introductory part of this study, most of the research on
banking crises tend to be either too generic or advanced economies-biased. Hence, there is
always the tendency to (mis) apply and overstretch concepts and solutions developed outside
the context of Africa to solve indigenous banking issues. This beckons the need to put the
phenomenon of banking crises in Sub-Saharan Africa into its unique perspective to determine
the underlying causal factors and how to remedy same.
Secondly, the international financial market is more integrated today than ever in history. In
consequence, the fallouts of crises in one country or region can have a dire effect on the entire
global financial system. The contagious nature of banking crises therefore warrants important
academic enquiries especially in unstable or developing financial contexts. This study is
directly in response to this fear. Bringing to the fore the determinants of banking crises in Sub-
Saharan Africa therefore is not only essential for the unique case of Africa but the entire global
financial system.

1.5 Organisation of the study
The paper proceeds as follows. Chapter 2 focuses on the literature on banking crises based on
theoretical and empirical evidence in both previous and recent relevant studies on the causes of
banking crises. Chapter 3 presents the data collection and methodological approach in the form
of econometric models to analyse our results. The results and discussion from the outputs will
constitute chapter 4 and obviously the final part of the paper would contain the conclusion and
recommendations drawn from the analysis.

Page | 5

CHAPTER TWO
CONCEPTUALISATION OF BANKING CRISES
2.0 Introduction
From the beginning of the 1980s, several kinds of banking crises have happened in many parts
of the world, running from both developed, emerging and the developing economies. All these
banking crises that occurred are caused by several factors (like volatility in terms of trade,
inflation, high real interest rate, bank concentration, real growth of GDP, among others), that
have led to the collapse of many banks, financial institutions and corporations of the world due
to cross-border banking linkages (Stolbov, 2013). However, there is a need to know the
meaning of banking crises, derive the understanding to be able to determine the factors
attributing to these occurring systematic banking crises.
In more general terms, a banking crisis is any financial crises that affect many banks, financial
institutions, and corporation whereby banks encounter difficulties in repayments of deposits.
Meaning, banks run out of liquidity and they do not have enough cash to repay their depositors
(World Bank, 2015). According to the Federal Reserve Bank of San Francisco (1985) banking
crises occur when there is a sharp reduction of a bank’s total value of assets, resulting in the
apparent or real insolvency of many banks and accompanied by some banks collapse and
possibly some bank runs. Contemporarily, Caprio and Klingebiel (1996) defined banking crisis
beginning from a sample of 69 countries of which information on bank insolvencies was
available since the mid-1970s to 1998, with data developed from published sources and
interviews with country economists. They described banking crisis as the occurrence of erosion
of bank capital and higher estimated cost of resolving the occurrence of the crisis. Again, the
analysis of Laeven and Valencia (2013, p. 63) defined banking crises as “occurring when a
country’s corporate and financial sectors experience a large number of defaults and financial
institutions and corporations face great difficulties repaying contacts on time”. They went
further to explain that, as a result of this, nonperforming loans will increase rapidly, leading to
deposit runs of most of these existing banks where they do not have enough capital to sustain
their banking operations.
Moreover, a banking crisis is “a non-linear disruption to financial markets in which adverse
selection and moral hazard problems become much worse so that financial markets are unable
to efficiently channel funds to those who have the most productive investment opportunities”
(Frederic Mishkin, 1996: p.17).

Page | 6

In addition, Demirgüc-Kunt and Detragiache (2005) after sampling 65 countries between 1980
and 1995 of their study perceived of banking crises as an episode banking in which the ration
non-performing assets to total bank assets exceeds 10 percent and the costs of rescue operations
exceed 2 percent of Gross Domestic Product (GDP). They believed that banking crises are
bound to happen when banks are experiencing frequent signs and events such as bank failure,
prolonged bank holidays, bank shutdowns, bank freezes and a large-scale bank nationalization.
Corroborating, Sundararajan and Balino (1991) also see banking crises as “a situation in which
a significant group of financial institutions has liabilities exceeding the market value of their
assets, leading to runs and other portfolio shifts, the collapse of some financial firms, and
government interventions” Sundararajan and Balino (1991: p. 3).
All these conceptualisations of banking crises clearly explain that banking crises that have
occurred over the years do not just occur but rather takes inspiration from a combination of
macroeconomic factors, institutional factors and some degree of moral hazard and adverse
selection problems in the market. However, an attention should be drawn by researchers to
importantly look at all these factors pertinent to banking crises since it has both short-term and
long-term repercussions domestically and globally that may have a consequence on economic
output and growth (Von Hagen & Ho, 2007).

2.1 Banking Sector Theory
In order to make a good analysis of the determinants of banking crises, it is of utmost
importance to situate the whole argument in a theory. To begin with, we need to know what
constitutes banking insolvency and how it comes about in the first place. As Demirgüc-Kunt
and Detragiache, (1997) rightly observe, “banks are financial institutions and intermediaries
whose liabilities are mainly short-term deposits and whose assets are usually short and long-
term loans to businesses and customers. Banks normally get into bankruptcy or insolvency
mainly when the value of total assets falls short of the value of total liabilities” (Demirgüc-Kunt
and Detragiache, 1998, p. 84). The result of this leads to banks insolvencies and this exposes
them to several systematic risks like any other players in the competitive market (credit risk).
The old practices of banking operations and credit risk management methodologies should give
way to a more rigorous credit risk management system (Diamond & Dybvig, 1983).
The generous Loan Loss Provisioning (LLP) by banks has partly contributed to the serious non-
performance loans that some banks are experiencing now (Eichengreen & Rose, 1998) . The

Page | 7

credit management methodology of some banks and financial institutions has been quite porous
with too much emphasis on collateral values rather than on the business case for loans based on
cash flow lending and thorough risk assessment of the business (Akerlof & Romer, 1993). The
default risk can be traced partly to the poor skills set of the analyst and the credit committee
members. Most banks do not simply understand Credit Risk Analysis (CRA) within the Small
and Medium Enterprise (SME) sector and apply corporate lending techniques to SME lending
or vice versa (Pereira Pedro, Ramalho, & Vidigal da Silva, 2017). They further indicated that,
the rate of credit risk may keep increasing when borrowers fail or are prove unwilling to settle
their bank loans. These banks run out of cash and are unable to continue with their operations.
However, credit risk of banks can be reduced, if loan officers can do proper screening to check
the credibility of loan borrowers before the loan is given to them. Again, the issue of collateral
as loan guarantee from borrowers could be assessed to ensure that borrowers have enough asset
whose value is either higher or equal to the value of the loan amount in case borrowers are
unable to settle their debts (C. Pereira Pedro et al. 2017). Finally, loan officers of banks should
conduct proper screening by ensuring that borrowers are financing or investing their loans into
profitable projects. With this credit risk could be reduced to avoid bankruptcy (Demirgüc-Kunt
& Detragiache, 2005).
Moreover, banks become insolvent when depositors’ confidence in them suddenly disappears.
Many at times, when people have trust in their banks, it is more difficult for crises to occur. On
the other hand, when people see that one or more banks have gone into solvency problems, they
evade confidence and they rush to withdraw their money at the same time, as indicated by the
experience of a number Latin American, Asian and Eastern European countries in the early
1990s (Demirgüc-Kunt & Detragiache, 1998). This will then lead to the collapse the bank since
many bank assets and operations depend on its flow of liquidity, a liquidity crisis erupts
(Turken, 2014).
Again, many banking crises that occurred in the 20th century have been attributed to liquidity
mismatch. Serious problems of liquidity risk exposures and proper assets liabilities
management have not been handled competently by many banks and financial institutions
(Turkan, 2014). Deposit usually obtained at very short tenor are lend at long tenor with the hope
of still being able to mop-up more deposits to shore up the liquidity gaps. In fact, some of the
banks and financial institutions did not have any credible liquidity contingency funding plan.
A lot of the failures in the savings and loans and microfinance sector is attributed to liquidity
mismatch, especially in many sub-Saharan African countries (Brownsbridge, 1998).

Page | 8

Nevertheless, the role of central banks has led to the collapse of many banks which put the
banking system of many developing countries at risk. Many studies argued that the
responsibility of the central bank as the lender of last resort and the establishment of deposit
insurance mechanism prevented most commercial banks from bank runs as a policy intervention
by governments (Klomp, 2010). However, some studies also believe that, if the central bank
fails to supervise and monitor commercial banks effectively according to their regulatory
framework, legal remedies against fraud are easy to circumvent and this may lead to widespread
looting ( (Stolbov, 2013). A clear looting behavior is what happened in the United States and
Chilean banking crises of the late 1990s, which was at the core of the savings and loan crises
of these countries (Akerlof & Romer, 1993). This may lead to the probability of the occurrence
of banking crises if legal systems live such fraudsters unpunished.
Furthermore, the occurrence of bankruptcy in a particular country in this context is not only
limited to the region or country alone but its impact extends to the border countries which results
in banking crises. Most banking crises that have occurred over the years started from one region
and later affected the economies of the neighboring countries in a little space and this spillover
effect of the crises is as a result of asymmetric information (Allen & Gale, 2007). This happens
because market players do not have a clear knowledge and understanding of the causes of the
initial bank crisis and this keeps on repeating itself and affects the neighboring countries (Calvo
& Mendoza, 2000).
For these reasons, it has been more difficult for researchers to clearly draw a curtain on the
main causes of banking problems due to data limitations and the potential of bank run that is
not directly observable. Even if bank run occurs, there is a quick response of government
interventions and takeover by large banks to limits the impact of the bank’s insolvencies
(Turkan, 2014). When this occurs, it is, therefore difficult to identify the causes of such banking
problems and ways to prevent them from occurring again in their near future. More specifically,
it makes the cost associated with banking crisis on developing nations bigger as compared to
the other nations since it is very difficult to attribute a specific problem associated with the
banking sector (Beck et al., 2010).

2.2 Previous Studies on Banking Crisis
The study of banking crises is very extensive and has generally captured a wide range of
contexts and often uses different methodologies and perspectives. Most of the studies have to

Page | 9

some extent focused on elements of the macroeconomic environment, financial stability
indicators and institutional factors as the main reasons for the emergence of banking crises
(Demirguc-Kunt and Detragiache, 2000; Cihak and Schaeck, 2010; Calomiris and Mason,
1997; Turken, 2014). Brownsbridge (1998) for instance considered macroeconomic volatility
as a reason for local banks failure in the financial markets of four Sub Saharan African countries
(Kenya, Zambia, Nigeria, Uganda). He went further to establish a causative relationship
between factors such as insider lending; liquidity support and prudential regulation; bank
lending to elevated risk borrowers and banking crisis in a form of financial distress in the mid-
80s. He argues that these factors contributed to the emergence of the banking crises in the sub-
Saharan African region. Finally, Pedro et al. (2017) have also considered the role of banking
crises as a major influence on banking crises in OECD countries. They looked at influencing
factors such as regulatory and supervisory boards; the socioeconomic context, depositor’s low
expectation about banks’ financial health; and banks’ specific characteristics (see Caprio and
Klingebiel, 2013). Many other researchers have equally delved extensively into systematic
banking crises tracing from the mid-1930s looking at the reasons for these emerging banking
crises and developing strategies to prevent similar from occurring (Calvo & Mendoza, 2000).
Powo Fosso (2000) in his empirical research investigated the financial variables leading to the
failure and collapse of banks in countries that form the West African Economic and Monetary
Union. Honohan (1997) also made further inquiry the key factors like inflation, interest rate,
exchange rate, terms of trade, among others that led to the financial sector in West Africa and
what really behind these factors on a more macroeconomics perspective. Additionally, Tenconi
(1993) discussed banking crises in the mid-1990s in Guinea to identify what went wrong. Put
together, players in the market today still believe that there has not been a precise knowledge
of the causes of the initial banking crises of the centuries, therefore they expect that banking
crises should occur in other jurisdiction to be able to examine and account for the reasons of
crises occurring (Pedro et al. 2017). In this regard, the next section of this paper will look at the
happenings of the banking crisis in the history to justify the above assertion.

2.3 Banking Crises in the History
The phenomenon of banking crisis is not new as indicated by several theorists and
macroeconomists like Diamond Douglas, Demirgüc-Kunt, Luc Laeven and Valencia, etc. in the
economic history of the world. Prominent banking theorists like Charles Klindleberger (1910-

Page | 10

2003) and Hyman Minsky (1919-1996) did advocate that the concept of the banking crisis is
something that revolves around any business cycles, and the results from propensities of market
participants and irrational reactions and myopic foresight. The theoretical modelling of banking
fragility: a phenomenon that any blogger can now trace at least as far back as 33 AD, when
Tacitus tells us that the Roman Empire suffered a major banking panic, which was quelled by
a large three years interest free loan to the banking system by Teberius (see Calomiris, 2010).
The traces of banking crisis over the years have strongly depicted that, banking crisis does not
randomly happen rather occur form events of the cyclical nature of business banking like any
other form of financial crises (Laeven & Honohan, 2005). It mainly results from bank runs,
which affects a single bank; bank panics, affecting many banks and systematic bank crisis,
which affects the large scale of a country’s economy and most financial institutions and
corporations experience great difficulties in repayment of contracts (Cooper & Ross, 2002).
Many banking theorists have drawn importance to banking crises than any other form of
financial crises since it takes a large share of the cost of financial crises that have happened
over the years. For example, bank run which is a form of bank crises could only result in the
collapse of asset prices but do not create major macroeconomic consequences to the entire
economy like the crisis in the United States in 1987 and 2000 (Bernanke, 1983; Calomiris &
Hubbard, 1989; Calomiris, 2000).
Historically, banking crises in its several forms (bank runs, bank panic, and bank waves) has
frequently occurred across countries, within countries and across time but interestingly the
margins and experiences are dramatic and different (Diamond & Dybvig, 1983). An example
is the US banking crises, which resulted in a great deal of panic and waves of bank failures in
the US banking system (see Demirgüc-Kunt & Detragiache, 1998). Nationwide, banking crises
occurred from the 1550s to 2010 but it has resulted in several forms in different countries with
different macroeconomic impacts (Cihak & Shaeck, 2010).
Banking crisis is often very costly to economies. The on-going banking crisis in China has cost
47% of its annual GDP already, while Japan realized a loss of 24% of its annual GDP. During
the Asian financial crises in 1997 through 2002, Indonesia lost 55% 0f its GPD, while Thailand
lost 35% and Korea lost 28% (Caprio & Klingebiel, 1999). Even though S&L crisis was costly
for taxpayers, but its effect on the United States economy was relatively limited, only 3% of
annual GDP of the US as presented in the table below.

Page | 11

Table 1. The Cost of Banking Crises in Some Selected Countries of the World.
Country Date Cost as a % of GDP
Argentina 1980-1982 55
Indonesia 1997-On going 50-55
Chile 1981-1983 41
Thailand 1997 -On going 33
South Korea 1997- On going 27
Malaysia 1997-Ongoing 21
Venezuela 1994-On going 20+
Mexico 1995 20
Japan 1990s 12+
Czech Republic 1989- On going 12+
Finland 1991-1994 11
Hungary 1991-1995 10
Brazil 1994-1995 5-10
Norway 1987-1995 8
Russia 1998 5-7
Sweden 1991-1994 4
United States 1984-1991 3
Source: Gerald Caprio Junior and George Klingebiel. “Episodes of Systematic and
Borderline Financial Crises”, World Bank, (1999).

2.4 Frequency of Recent Banking Crises
Banking crises is a worldwide phenomenon that cut across the entire world which is not
something new, however, it has occurred in several parts of the world due to its contagious
effect (Allen & Gale, 2007). Figure 3 below presents the number of times banking crises have

Page | 12

occurred around the world showing the regional distribution of the globe (Laeven & Valencia,
2013).
Most of the Southern American countries (like Bolivia, Venezuela, Brazil, Mexico and
Colombia), the United States of America, Russia and some few countries in Africa have
experienced at most two banking crises over the period as stated. Some countries in Asia, Africa
and Europe have also experience a crisis in their economies. Surprisingly, Demographic
Republic of Congo is one of the SSA countries that has experience at least two countries and
Caprio and Klingebiel (1996) attributed the occurrence this crisis to inadequate banking
supervision and regulatory framework and weak controls and management of the Judiciary
system.

Figure 1: Frequency of Recent Banking Crisis Around the Globe (1970-2011).

Source: Leaven and Valencia (2013).

2.5 Banking Crisis Cycles
Banking crises as a phenomenon is not an event that just happen rather a process which mainly
start with a bank run, when occurs in a single bank and will result to systematic banking crises
when it spreads across the whole economy or regions consistently (Diamond & Dybvig, 1983).
Figure 2 presents banking crises levels from a given year 1970 to 2010 (Laeven & Valencia,

Page | 13

2013). In the beginning of the 1970s to the 1980s was the occurrence of Latin American debt
crises, it then spreads itself into the emerging and transition economies in the 1990s. The great
recession of systematic banking crises was between 2007 and 2009 which involves some of the
European countries such as Spain, Greece, Ukraine and the only SSA country, Nigeria.

Figure 2: Banking Crisis Cycles in Recent History

Source: Laeven and Valencia (2013).

2.6 The Experience of Banking Crises in some Sub-Saharan African Countries
On the tables of SSA region, banking crises have occurred and had equally caused a negative
impact in the region. The most relevant of the banking crises occurred in the region was the
banking crises in the 1980s and 1990s, and the only challenge is the dates associated with the
beginning banking crises in this period (Daumont et al., 2004). The SSA banking crises which
occurred was mainly associated with currency crisis, while they hardly concur with sovereign
debt crises (Mezui et al., 2012). The work of Laeven and Valencia (2008) reports about 50
percent cases of banking corresponding to currency crisis and only 9 percent match with
sovereign debt crises. The banking crises that occurred on the SSA region led to the collapse of
several banks both old and new, financial distress government-own banks leading to a high
pressure on the growth of GDP on most economies in the region (Mezui et al., 2012). Caprio
and Klingebiel (1996) established that the only factor leading to these crises which they
attributed to the rate of nonperforming loans. Non-performing loans stood at -least 5-10

Page | 14

percentage of total bank assets over the period which they believed that such a performance can
easily lead to bank insolvencies in several banking institutions within the region. Table 2
presents the episodes of banking crises of some countries in the SSA region in the 1980s and
1990s, the scope of the crises and its estimated losses as a percentage of the country’s GDP
(Caprio & Klingebiel, 1996; Laeven & Valencia, 2013).

Table 2. Episodes of Banking crisis in SSA Region (the 1980s – 1990s)
Country Date Scope Estimated
Losses/Cost
Benin 1988-
1990
Collapse of all three banks, 78 percent of
banking system loans nonperforming at the end
of 1988
17 percent of GDP
Cameroon 1987-
1993
60-70 percent of banking system loans
nonperforming in 1989

Cote d’Ivoire 1988-
1991
4 large banks accounting for 90 percent of the
banking system loan affected; nonperforming
loans at 4 largest banks reached about half of
total credit outstanding
Cost of government
equivalent to 25
percent of GDP
Ghana 1982-
1989
7 of 11 audited bank insolvents; 40 percent of
banking system deposits insolvent; 80 percent of
banking system loans nonperforming
Restructuring cost
equivalent to 6
percent of GDP
Guinea 1985
1993-
1994
6 banks accounting for 99 percent of banking
system deposits insolvent; 80 percent banking
system loans nonperforming
3 banks accounting for 45 percent of market
affected area
Repayments of
deposits equivalent
to 3 percent of GDP
Kenya 1985-
1989
1993-
1995
4 banks and 24 NBFIs accounting for 15 percent
of financial systems assets; 66 percent of loans
of one third of banks nonperforming in 1993

Nigeria 1991-
1995
8 banks insolvent and 45 percent of banking
system loans nonperforming at the of 1992; 34

Page | 15

out of 115 banks accounting for 10 percent of
deposits insolvent in 1994
Senegal 1988-
1991
50 percent banking system loans nonperforming
in 1988
17 percent of GDP
Tanzania 1987-
1990s
Main financial institutions in arrears amounting
to half of their portfolio in 1987; government-
owned banks accounting for 95 percent of
banking system assets, insolvent as of 1990 at
least; 60-80 percent of all loans nonperforming
at the end of 1994
Implied losses
equivalent to nearly
10 percent of GDP
Uganda 1990-
present
Half of banking system facing insolvency
problems in 1994

Eritrea 1993 Most of the banking system insolvent
Madagascar 1988 25 percent of loans deemed irrecoverable
Mauritania 1984-
1993
5 major banks had nonperforming assets ranging
from 45 – 70 percent of their portfolio
Cost of rehabilitation
estimated of GDP in
1988
Mozambique 1987-
present
BCM, main commercial bank, experiences
solvency problems which became apparent after
1992

South Africa 1977 The collapse of Trust Bank
Togo 1993,
1994,
1995

Zambia 1995 Meridian bank became insolvent which
accounted for 13 percent of commercial bank
assets
Rough estimate of
US$50million (1.4
percent of GDP)

Congo
Republic
1980s
and
1994

Page | 16

Central
Africa
Republic
1988-
1991

Burkina Faso Late
1980s

Source: Caprio & Klingebiel, 1996; Laeven & Valencia, 2013.

2.7 Patterns of Banking Crises
The flexibility or fragility nature of banking operations like any other financial institutions
make them exposed to many possible coordination problems. Institutional weaknesses and lack
of information could be a challenge that can impinge the thoughts of investors and depositors
to withdraw all their capital and savings from the bank respectively (Cooper & Ross, 2002).
With these, banks are likely to venture into liquidity problems and may end the operations of
the banks. Banking crises occurred over the centuries had presented itself in several forms and
patterns. Many macroeconomists like Bernanke and Gertler (1987) established that it is possible
to divide banking crises into old and new. In that, the old type of crises was money crises that
caused a devaluation of foreign exchange reserves (between 1950 and 1980 balance of
payments crises occurred resulting in devaluation. The new type of crises is the banking crises
that caused the debt discharging problems that happened after 1980. Moreover, Calomiris and
Mason (1997) described the new type of crises as the one that occurs in the case of bank failure
that forces banks to delay their responsibilities, depositor’s running away from one or more
banks because of the sense and fear that banks are crashed with liquidity problems might not
be able to repay them their money deposited and invested and some cases, the government
through the central bank has to intervene to rescue these banks to ensure their existence
especially in terms of non-performing loans. Again, banking crises as described earlier
presented itself as any financial crises which affect banking, and when it affects individual
banks is called a bank run, affecting several banks is known as banking panic and when it affects
a whole economy of a country is called a systematic banking crisis (Reinhart, 2009). This has
made it difficult to identify all the banking crises that have occurred over the centuries according
to their timings. However, his section will lead to the description of these forms of banking
crises and the periods of its happenings.

Page | 17

2.7.1 Bank Run
A bank run (also known as a run on the bank) occurs when a large number of people withdraw
their money from a bank because they believe the bank may cease to function in the near future
(Diamond, 2007). In other words, it is when, in a fractional-reserve banking system4, a large
number of customers withdraw cash from deposit accounts with a financial institution at the
same time because they believe that the financial institution is, or might become, insolvent; and
keep the cash or transfer it into other assets, such as government bonds, precious
metals or gemstones (Diamond & Dybvig, 1983). When they transfer funds to another
institution it may be characterized as a capital flight (Cooper & Ross, 2002). The results of this
may lead to the destabilization of the bank as it faces bankruptcy. The cultural activities of
traditional banks have always been a form of holding a percentage of depositor’s cash savings
at a point in time, lending the rest to borrowers with higher interest returns or invest in assets
or projects with the higher bearing interest rate. When bank runs occur, what banks do is to sell
their assets to be able to get enough cash to offset their loan losses to prevent insolvencies.
However, during a bank run fear and panic are aroused in the thoughts of depositors which
made them lose confidence on the security of their money and they are forced quickly to
withdraw to cut off their losses (Bryant, 1980).
Bank runs has commonly occurred in the United States whose laws allow banks to operate on
a single branch system of banking compared to banks that operate in a multiple branch system
in other jurisdictions (see (Demirgüc-Kunt & Detragiache, 1998). The great depression of the
1930s which started as an ordinary recession eroded many banks in the United States that
resulted in huge economic chaos over the times. Economic engines of growth like the real gross
domestic product (GDP) and industrial production went as low as 70 percent and 47 percent
respectively and increased total unemployment levels to 20 percent (Friedman & Schwartz,
1963). For example, is the bank run that hit a lot of banks who operates in a single bank system
in New York in the 1930s, but these runs were rescued on time by the Federal Reserve Bank
(Fuller, 2011). Not only banks in New York City did bank runs occurred but it also continues
to break through cities such as Boston, Dec. 1931, Chicago, June 1931 and 1932, among others
(Cooper & Ross, 2002)..
Many of these banks run occurred over the years had been attributed to coordinating problems
and institutional weaknesses due to the flexibility and fragility operations of the banks
4 where banks normally only keep a small proportion of their assets as cash

Page | 18

(Diamond & Dybvig, 1983). These problems exposed the banks to further challenges and
individuals are forced to withdraw their deposits at the same time.
These challenges associated with the theory of bank runs as a lead to banking crises has
recognized by several policymakers to develop a mechanism that may help eliminate fragility
problems of most banks and financial institutions (Diamond, 2007). The issue of proper
regulatory and supervisory framework should be designed to eliminate the vulnerabilities and
reduce risky behavior of depositors to help elevate stability for the banks. The introduction of
deposit insurance has been very instrumental to help reduce these associated risks since it
provides a guarantee that promised a payment of return to those willing to withdraw (Cooper
& Ross, 2002).

2.7.2 Bank Panic
One of the forms of which bank crises occur is through bank panic or banking panic. This
happens where financial crises hit two or more banks at the same time and the social cost is
quite larger than bank runs (Fuller, 2011). According to Robert Fuller (1933), he described bank
panic as any financial crises that occur when many banks are crashed with liquidity problems
at the same time and as possible, they suddenly try to convert their threatened deposits into cash
or try to get out their domestic banking system altogether.
Carapella (2015) also described banking panic as any financial shocks whose origination can
be traced from any sudden and anticipated revisions of expectations of deposit loss and during
which there is an attempt, usually, used to convert checkable deposits into currency. She further
indicated that banking panic is characterized by an increase in the number of bank runs and
bank suspension of currency hoarding. She claimed that the result of these banking panics
collapsed over hundred banks and others to suspension.
More often bank panic begins with a bank run, thus a single bank and as other interconnected
banks trying to help to rescue this bank in order to compensate for the failed components of
losses. This, in turn, overloads these nodes, causing these helping banks to run into financial
crises as well (Diamond & Dybvig, 1983). A clear example is what Gary Richardson (2007)
mentioned to the worried customers during the banking panics between the period of 130 and
1931 that eroded some banks in the United States. He announced that, on November 7, one of
Caldwell’s principal subsidiaries, the Bank of Tennessee (Nashville) closed its doors. On
November 12 and 17, Caldwell affiliates in Knoxville, Tennessee, and Louisville, Kentucky,

Page | 19

also failed (see Demirgüc-Kunt & Detragiache, 1998). He further stated, the failures of these
institutions triggered a correspondent cascade that forced scores of commercial banks to
suspend operations, especially in communities where these banks closed, depositors panicked
and withdrew funds en masse from other banks. Panic spread from town to town and within a
few weeks, hundreds of banks suspended operations. About one-third of these organizations
reopened within a few months, but the majority were liquidated (Richardson 2007). More than
nine thousand banks failed to total about 30 percent of the entire operating bank in the United
States (Friedman & Schwartz, 1963). During the incidence of banking panics of 1931, not more
than two months elapsed between the end of the first banking crisis in January 1931 and the
onset of the second in April. The number of bank suspensions was lower (573), but the amount
of hoarding doubled. One-third of the bank suspensions were in the Chicago Federal Reserve
District; there was a mini panic in Chicago in June and a full-scale panic in Toledo, Ohio, in
August (Carapella, 2015). The Cleveland Federal Reserve District had two-thirds of the
deposits of suspended banks. Nevertheless, in six districts there was little or no change in
currency hoarding (Friedman & Schwartz, 1963). The third incidence of banking panic
happened in September 1931, which coincided with Britain’s departure from gold standards. In
fact, this shoots the number of failed banks, deposit fail banks and hoarding ones to an increase
after Britain’s pronouncement of their exit. The Federal Reserve quickly issued a discount rate
thinking this could help to absorb these damages but rather increased the rate of hoarding banks
due to higher interest. This eroded the knowledge of increased deposit ratio as a key determinant
of money stock market. Overall, about 60 percent of the 2,291 banks that were closed happened
in the 1931 banking panic era (Boughton & Wicker, 1979).
Another period of systematic banking panic is the bank panic in 1933 which have been
described as “idiosyncratic” (Jail, 2009). The 1933 banking panic, in fact, resulted in a
widespread of the legal device in the economic parse of the United States (see Bryant, 1980). It
was during this banking panic where president Franklin Roosevelt declares an official banking
holiday in many states of the country in March 1933 (Friedman & Schwartz, 1963).

2.7.3 Systematic Bank Crisis
Systematic banking crisis as described earlier refers to any financial crisis that affects banking
activities in which a country experienced a large number of defaults and financial institutions
and corporations face difficulties in repaying contract (Barth et al., 2008). Many banks are
affected, and this has a trickle-down effect on the whole economy of a country at large.

Page | 20

According to Laeven and Valencia (2013), a systematic banking crisis is the crises that have a
larger effect on GDP growth and a slowdown in nominal credit growth of a country’s economy.
The reason is that identifying banking crisis according to the dates they occurred are quite
disrupting since they developed in different forms and patterns. It starts at a slower pace, turns
to speed up a little bit and by the time you realized it has trickled down all over the economy.
The occurrence of a systematic banking crisis has a greater impact on the economic parse of
the country, associated fiscal costs and large output losses (Leaven & Valencia, 2008).
The works of Leaven and Valencia (2016) indicated that the recent banking crisis occurred
between 1970 and 2011 brought a larger wave of impact on most advanced and emerging
economies compared to the losses of developing economies. The most of macroeconomic
policy responses like the monetary and fiscal policy actions to help combat the consequences
associated with this systematic banking crisis were made by the advanced and emerging
economies who were most affected (Klingebiel et al., 2007).

2.8 Causes of Banking Crises – A Brief Literature Review
As it has been already indicated in the earlier sections of this paper, the stability of banks
performance is dependent on its ability to hold more liquidity and being solvent. This enables
them to fulfill their obligations to their depositors. Banks doing this will further evade them
from any risk associated with any forms or patterns of banking crises in the near future. Hence,
several studies like (Demirgüc-Kunt & Detragiache, 1998; Caprio & Klingebiel, 1999; Klomp,
2010; and Stolbov, 2013) have emerged over the decades to look at factors that might cause
banks to experience banking crises characteristics.
The factors behind banking crises have been established by different economic theorist using
several distinct kinds of methods and econometric tools and analysis. These factors identified
by most renounce economists has been divided into three dissimilar categories namely;
macroeconomic factors, institutional factors, and environmental factors, though there has not
been any accurate agreement on these categories as to which variable comes under each one of
them (Stolbov, 2013).
Demirgüc-Kunt and Detragiache (1998) find a negative relation between some macroeconomic
environment indicators and the banking sector of a country. They established that any country
whose macroeconomic environment (low GDP growth, high inflation, current account

Page | 21

balance/GDP, high real interest rates, deposit insurance among others) is of high probability to
experience systematic baking crises. In case of high inflation standards, they explained that
because of high and volatile nominal interest rate associated with high inflation, banks might
find it very difficult to perform maturity transformation. C. Pereira Pedro et al. (2017) supported
the argument after applying the same categorical approach to identify high inflation as a major
macroeconomic determinant of banking crises in OECD countries. Similarly, they concluded
that bank crises tend to occur in countries where banks are smaller and present higher levels of
debt when macroeconomic environment slows done and a higher inflation is registered.
Furthermore, Goldstein and Turner (1996) also identified multiple causes of the banking crisis
in emerging countries. They identified macroeconomic indicators such as fluctuations in terms
of trade, volatility in international interest rate, high real exchange rates, growth and inflation
volatile as major external macroeconomic factors that exposed banks to systematic crises. In
support of their views, Angkinand and Willet (2011) established that countries with experiences
of numerous exchange rate regimes are prone to crises due to excessive net borrowing. A larger
share of household credit relative business loans also increases the proneness of banks towards
systematic crises (Buyakkanbacak and Valev, 2010).
Moreover, Beck, Demirgüc, and Levine (2006) used a sample of 69 countries and 47 episodes
to examine the impact of bank concentration on banking system fragility. They identified that
concentrated banks associated with proper banks regulation and better institutions are less likely
to suffer any systematic damages. Apart from adequate capital and more liquidity flows, Barrel
et al (2010) find that a higher concentration of banks indicates a positive impact on its stability
and thus reduces the likelihood of crises.
Nevertheless, banking system regulations and supervisions have been identified by Mikhail
Stolbov (2011) as a key institutional cause of banking crises. Some authors concluded that
countries with fewer episodes of banking crises are those presenting lower regulation, therefore
there is a need for governments to establish quality regulatory standards that may evade banks
form systemic crises (Joyce, 2011). Banking systems whose regulations restricts banks from
engaging in non-loan making activities may aid the bank to experience a crisis (Beck et al.,
2016). Weak regulation and poor supervision by the government through the central banks that
allow some banks to apply proper regulatory standards in their core business code of conducts
also increase the risk of the bank to experience a systematic crisis (Angkinand et al., 2010).
Daumont et al. (2004) drawing results from their logical experiences of banking sector distress
in ten study countries in SSA region during the period of 1985-1995 also established similar

Page | 22

factors that influence the occurrence of banking crises compared to other regions worldwide.
They identified macroeconomic shocks (exchange rate shocks; pervasive government
involvement and loose controls on insider lending; weaknesses in the legal and institutional
framework; inadequate capitalization and loan loss provision and distorted invectives that
encourage economic agents to take excessive risk) as factors that exposed banks to systematic
crises and they never find any difference in factors that exposed banks to crises comparable to
crises that have to happen in any other parts of the world.
Notwithstanding to these factors, other empirical studies have also identified some other factors
such as rate of central budget surplus to GDP, rate of domestic credit to private sector to GDP,
remittance inflows to GDP, inadequate preparation for financial liberation, distorted incentives,
among others when occurred can exposed banks any form of banking crises. For example,
Demirgüc and Detragiache (1998) explained in their study that inadequate preparation for
financial liberation leads to bank fragility since it gives the bank more to take the risk especially
to countries who have liberalized their financial systems.
However, several mechanisms have been developed by policymakers to help reduce the
likelihood of bank crises from occurring on the premises of these factors that lead to the banking
crisis. As a result of competition among banks particular in SSA region, banks officials refuse
to do proper screening to check the credibility of borrowers before issuing a loan. To mitigate
this, credit management information systems should be established on more multiplicity
approach to eliminate poor coordination and controls between lending groups and the resulting
absence of standardized relating to portfolio performance of Buyakkanbacak and Valev, 2010;
Brownsbridge 1998; World Bank, 2006). Another regulative mechanism that has created a lot
of controversies is the issue of deposit insurance which had been viewed as a major factor to
mitigate bank from crises, but its expectation is otherwise (Beck et al., 2006). Bank runs should
cease from occurring when deposits are insured against the risk of bank insolvency. Banks do
purchase insurance packages from the government or private insurance companies for
instability against insolvencies. Depositors, on the other hand, expect that in case of any
financial distress of the bank, the government will come in to settle them their deposits to
maintain the stability of the banks (Demirgüc-Kunt & Detragiache, 1998). Another study
believes that deposit insurance will create the incentive of taking excessive risks but could also
keep the stability of bank into operations as it comes in to settle depositors (Kane, 1989).

Page | 23

2.9 Theoretical Evidence on the Causes of Banking Crisis
Banks as it has been already described in the previous sections as a financial institution licensed
and are allowed to receive deposits of individuals or groups and lend to borrowers either on
short-term or on a long-term basis. This means banks need to hold enough cash or very solvent
in order to sustain its growth levels in a particular country. Previous studies on bank
insolvencies of banking sector distress have attributed the banking sector failures to bank runs
(Bhattacharya & Thakor, October, 1994). They believed that when bank runs occur, this
descends fear and panic on the individuals and institutions who have money with the bank, and
they are easily convinced to simultaneously withdraw their money from the bank. With this,
banks will eventually be exposed to any form of financial risk that may lead to bank
insolvencies.
However, recent studies about the factors leading banking crises are not only attributed to the
fundamentals of bank runs but extend to more identified multiple of reasons or factors that have
resulted to both past and on-going banking crises over the years (Hardy & Pazarbaloglu, 1999).
One of the factors that had been discussed in the literature leading to banking system problems
or crisis is the issue of changes in interest rate. Economically, the interest rate is perceived to
be the cost of borrowing money from a bank. Banks balance consist of long-term loans as
affixed interest rate, so when there is an increase in short-term interest rates, this will result in
irregularities in the bank’s balance sheet and it will take a very long time for the banks to redeem
a very high rate of return on their assets to compensate on any decrease in the rate of liabilities
in a case of balance sheet differentials (Demirgüc-Kunt & Detragiache, 1998). Galbis (1993)
indicated that rate of inflationary could also result to short-term interest rate and the losses that
may arise are mostly passed on to borrowers as high lending rate results in a larger fraction of
nonperforming loans. Another side of interest rate inconsistencies is what Brock (1995)
indicated in his study. He explained that change in international interest rate may relatively
encourage borrower’s decision towards foreign investments in other emerging markets. These
investments may increase the demand for foreign currencies at the expense of the domestic
currency, which increases foreign debts. The result of this situation does expose banks to the
domestic market to distress since foreign debts of the country economy increase more than
foreign reserves (Arestis & Sawyer, 2003).
Another case that may cause banks to experience any form of financial crises is the issue of
depreciation of a country’s currency which could disturb the profitability rates of banks. This

Page | 24

occurs when banks borrow in foreign currency and lend to individuals or groups in domestic
currency, an unexpected depreciation in the local currency could which may be a bad story to
most local banks (Akerlof & Romer, 1993). A similar indication could be the high demand for
foreign currency loan by most African countries from international organizations and banks
such as the world bank, IMF loans, etc. This increases foreign exchange risk of the domestic
country leading to disturbing the profitability of banks and this is again passed on to borrowers
(Mishkin, 1996). Countries who are not able to produce for more exports and they are solely
dependent on foreign products technically help to increase the rate of foreign country currency
at the expense of the domestic currency (Garber, 1996).
Moreover, deposit insurance has been identified as a critical by several researchers as a major
area to avoid banking crises, but bank runs are easy to occur is when bank deposits are not
insured against the risk of bank insolvencies (Diamond & Dybvig, 1983). Diamond and Dybvig
(1983) indicated that depositors will quickly rush to withdraw from the bank before the bank is
even announced bankruptcy, especially in a situation where a bank assets portfolio alert for a
run. Differently, other studies believed that the introduction or presence of deposit insurance
has reduced the incentive of depositors to monitor their banks (Kane, 1989). Kane (1989)
further explained that banks may insure their capital or assets with either the government or
private insurance companies. In that, in case of liquidity problems, the government or the
private insurance company will come in to rescue the bank to resolve depositors. Deposit
insurance scheme has been also captured as a way of creating moral hazard problems which
may lead to further back failures especially in less developed countries like Africa with weak
institutions, regulatory and supervision (Schich, 2009). In the long run, this may lead to
systematic instabilities which eventually result in the banking crisis.
Moral hazard problem is not only associated with deposit insurance challenges but also believe
to be a giant factor that may also lead to the banking crisis. Moral hazard is a concept
characterized by principal-agent relationship and is mainly facilitated by asymmetric
information (Brownsbridge, 1998). Moral hazard problems occur when bank owners tend to
act in a way that is contrary to the interest of bank creditors by undertaking risky investments
strategies such as a higher interest rate to high-risk borrowers, which if not successful may lead
to the collapse of the bank (Stiglitz & Weiss, 1981). Because most banks are limited liability
owned, bank owners think there will not be held liable to any losses that may happen to the
bank, therefore they are encouraged to enter high risk business investment at the expenses of
depositors.

Page | 25

Moreover, lack of proper supervision and regulatory framework to control the affairs and
banking operations also exposes banks to the crisis. Studies have shown that countries in which
banking systems are liberalized but characterized by weak bank supervision and regulation are
easy to experience banking crises (Demirgüc-Kunt & Detragiache, 1998). This may result in
fraudulent acts and widespread of looting by some managers, in that, bank managers may invest
in high-risk projects and projects with a hundred percent chances of failing but from which they
divert the resources for their personal use (Barth et al.,2008). This was basically the reason for
the savings and loan crisis that hit the United States and the Chilean banking crises of the late
1970s (Akerlof & Romer, 1993).
Furthermore, banking systems are more exposed to crises especially in countries where inflation
is high. Generally, inflationary problems arise when prices of goods and services of a country
do not simultaneously respond to higher income of people in the same country. The result of
this may eventually reduce the purchasing power of people and this is, in turn, leads to
stagnation of economic growth in the country (Fischer, 1986). This may affect most financial
institutions leading bank runs or failures. Countries with high levels of inflation rate and when
inflation rate decreases rapidly, banks believe that a greater avenue is created to retrieve more
income or raise profits standard has been shut and problems are likely to occur in the banking
sector (Demirgüc-Kunt & Detragiache, 1998).

Page | 26

CHAPTER THREE
DATA AND METHODOLOGY
3.0 Introduction
This section of the paper presents the source of data for the analysis, the methodologies involved
in transforming the available data and the econometric model specifications for the regression
results and analyses.
3.1 Source of Data
The paper covers the period between 1980 and 2016. A panel dataset for this study was
developed from a several secondary sources of database. One of these is the World
Development Indicators (WDI) database, where the development of the dataset on most of the
selected independent variables were derived.
To find dataset for all the sample countries and the period of the study, Global Financial
Development Database (GFDD) was another secondary source consulted to balance the panel
set of data for the regression. GFDD is an extensive financial database that provides information
on the financial systems and institutions for about 206 economies worldwide by the world bank
(Cihak, Demirguc–Kunt, Feyen, & Levine, 2012).
To provide a balance dataset, data was again taken from the International Financial Statistics
(IFS) worldwide database provided by the International Monetary Fund’s (IMFs), World Bank
to fill in all the missing figures and variables in the panel dataset. Finally, previous literatures
such as Gerald Caprio and Daniela Klingebiel (1996), Luc Laeven and Fabien Valencia (2013)
and Roland Daumont, Françoise Le Gall and Françoise Leroux (2004) were consulted to design
proper mechanisms and approach to establish what this paper intends to find.

3.2 Sample
This paper focused on all countries within the SSA region, which is a total of 48 countries from
the period of 1980 to 2016 and the list of all these countries is presented in the appendix section
of this paper. The paper went further to adopt “GRETL” as the econometric software model for
the data computation and regression analysis. The dataset mentioned in the sources above were
originally a time series dataset. For GRETL to absorb these dataset, transposition method was
applied to convert the time series data to a panel dataset since the selected countries is observed
at two or more – time periods as indicated in the sample size.

Page | 27

3.3 Dependent Variable (Banking Crisis) Specifications
Banking crisis was selected as the dependent variable for this study. It is more critical in its
construction because it is a dummy or binary variable that takes the values 0 and 1 (1= banking
crisis; 0 = no banking crisis).
To arrive at an effective and comprehensive results in this study compared to the previous
studies on the determinants of banking crises, there is, therefore a need to construct quality data
for banking crisis (dependent variable). Based on the previous works of Klomp Jeff (2010),
Caprio and Klingebiel (1996) and Laeven and Valencia (2013), who identified comprehensive
dates on episodes of banking crises over the years, an accurate construction and identification
were developed for this study. With this, limitation was drawn on SSA countries, which is the
sample size for this study. To compute quality and balance data for banking crisis, an additional
data was then employed from the GFDD, in order to have dataset for all the 48 countries
included in the sample. The GFDD is an extensive dataset introduced by the world bank which
presents accurate dataset on financial institutions and systems for about 206 economies in the
world from 1960 to 2015 (Cihak et al., 2012).
Table 3 presents dates of the occurrence of systematic banking crises for all the countries
included in the study sample.
Table 3. Sub-Saharan African Countries and their Banking Crises Period
Country Period/ Dates
Angola None
Benin 1988-1992
Botswana None
Burkina Faso 1990-1994
Burundi 1994 – 1998
Cape Verde 1993
Cameroon 1987-1991
Central African Republic 1994 – 1995
Chad 1983,1992-1996
Comoros None
Demographic Republic of Congo 1983, 1991-1998
Republic of Congo 1992-1994

Page | 28

Cote d’Ivoire 1988-1992
Equatorial Guinea 1983
Eritrea 1993
Ethiopia None
Gabon None
Gambia None
Ghana 1982-1983
Guinea 1985 and 1993
Guinea Bissau 1985 and 1998
Kenya 1985-1989, 1992-1994
Lesotho None
Liberia None
Madagascar 1988
Malawi None
Mali 1987-1991
Mauritania 1984
Mauritius None
Mozambique 1987-1985
Namibia None
Niger 1983-1985
Nigeria 1991-1995, 2009
Rwanda None
Sao Tome and Principe None
Senegal 1983-1988, 1998-1991
Seychelles None
Sierra Leone 1990-1994
Somalia None
South Africa 1997, 1985
South Sudan None
Sudan None
Swaziland 1995-1999
Tanzania 1987-1989
Togo 1993-1994, 1995

Page | 29

Uganda 1994
Zambia 1995-1998
Zimbabwe 1995-1999
Source: Laeven and Valencia (2013), WEO, IFS, IMF’s World Bank Database.
A panel dataset was constructed out of these two-distinctive databases after omitting countries
and the periods not included in the samples size for this study. From this, a value of 1 was
assigned to a country to indicate the presence of banking crisis under a particular year and a
value of 0 for non-occurrence of banking crisis in the same country for the same year.

3.4 Explanatory Variables Specification
The selected explanatory variables were developed based on the available theoretical and
empirical studies on the causes of banking crises written over the years. Specifically, variables
were selected based on the study of Asli Demirgüc-Kunt and Enrica Detragiache (1998), Beck
et al. (2006), C. Pereira Pedro et al. (2017), Mikhail Stolbov (2013) and Hardy and Pazarbasiglu
(1999). Like the dependent variable, the dataset of the explanatory variables were also
computed from the IMF’s, IFS database, WDI database and GFDD, World Bank.
Table 4 presents the list of all the explanatory variables used in this study, their detailed
description and where the data was acquired.
Table 4. Description of the Explanatory Variables and their Sources of Data
Variable Name Description Source of Data
Real GDP growth The annual growth rate of
GDP for a given country
(annual percentage)
WDI database
Changes in Terms of Trade Terms of trade adjustments
in constant local currency
WDI database
Inflation Consumer price index
(annual percentage)
IMF’s IFS database
Real Interest Rate Lending rate adjusted for
inflation (measured by GDP
deflator.
IMF’s, IFS and data files
using World Bank data on
GDP deflator

Page | 30

Depreciation Nominal exchange rate
divided by a price deflator or
index costs (Real effective
exchange rate)
IMF’s IFS database
M2/Reserves Broad money to total reserve
ratio.
IMF’s, IFS line 35L..ZK
Private Credit to GDP Domestic credit to private
sector growth to GDP
(annual percentage of GDP)
IMF’s IFS database
Bank Concentration Sum of three largest bank in
bank scope / Sum of all banks
in bank scope
Bank Scope, Bureau van Dijk
(BvD), GFDD as an
alternative

Remittance Inflows Personal remittances
received (percentage of
GDP).
WDI database
Deposit Insurance Scheme The introduction and
protection to bank depositors
to support financial stability.
WDI database

Out of several variables in the literature that contributes to the probability of banking crisis, a
few of them were selected as regressors for the study. These include some indicators such as
real GDP growth, terms of trade changes, inflation, real interest rate, depreciation, M2/reserve,
domestic credit to private sector to GDP, bank concentration, remittance inflows and deposit
insurance scheme. These were introduced in the study to establish effective relation as to what
determines the probability of banking crisis in SSA region.
According to the World Bank (2015), real GDP per growth is basically the annual percentage
growth of market prices of gross domestic product from one period to another based constant
local currency which is adjusted by inflation. Banking crises in real sense do not just happen or
occur, however, low growth rate of GDP of a country is likely to affect the economic output of
that same country that could create a banking panic leading to systematic bank problems
(Diamond & Dybvig, 1983). Not only low GDP growth that might create systematic crises but
also excessively high interest rates is also likely to cause banking sector problems in a country.
Investors and depositors are always interested in the returns on an increase in short-term interest

Page | 31

rate, however, bank’s balance sheet also consists of interest of long-term loans. Therefore,
banks are highly likely to suffer a banking challenges when short-term interest rates go up even
if they pass it onto depositors as nonperforming loans in the future (Demirgüc-Kunt &
Detragiache, 1998).
The changes associated with terms of trade of a country is very significant to include as a
regressor for this study. Terms of trade is the amount of a country’s total imports of goods and
services compared to its exports (Krugman et al. 2005). A country is expected to experience a
deficit trade balance when its total amounts of imports exceeds its exports. Studies have shown
that most African countries depends solely on foreign imported goods and services compared
to the total amounts of products produce in the country for exports and this may lead to adverse
terms of trade effects in the country’s economy (Bernard, Jensen, & Redding, 2007). Again,
restrictions on trade normally affects the standards of goods and services produced on the SSA
region, which limits the rate of exports (Krugman et al. 2007). The effects on the economy is
likely to distort the stability of the banking sector operations which may expose the banks to
any form of financial instabilities.
Inflationary pressure can never be left out when identifying the determinants of banking crises
in SSA. The rate of inflation was introduced in the study as an independent variable because it
is associated with nominal interest rate which one must consider in calculating for interest rate
and the volatility of inflation in a country creates a level of risk to both the lender and receiver
(Fisher, 1997). Banks as lenders will be exposed to bank risk leading to a crisis even if they
pass it on as nonperforming loans to customers.
Another variable introduced in the study as a regressor is the rate of depreciation. Depreciation
was introduced as a regressor in the study, specified as the rate of effective exchange rate based
on consumer price index according to the IMF’s International Financial Statistics database
(World Bank, 2015). Depreciation occurs when banks borrow in foreign currency and lend in
domestic currency, this increases the exchange risk of which banks are affected. However,
excessive foreign exchange is likely to distort the stabilization of bank’s profitability ratios
(Obtsfeld & Rogoff, 1995).
Moreover, M2/reserves was included in the study to specify the impact of broad money to total
reserve ratio on systematic banking crises. M2/reserve refers to the ratio of foreign reserves
based on two variables, namely currency deposit and reserve deposit ratio (World Bank, 2015).
Calvo Guillermo (1996) in his Tequila Lessons “capital flows and macroeconomic
management”, he emphasised that broad money to total reserve ratio is likely to cause balance

Page | 32

of payment crises and that, this may result into additional banking failures leading to a
systematic banking crisis.
Notwithstanding, the growth rate of domestic credit to private sector to GDP was added to the
study to determine its’ impact on banking failures. Domestic credit to private sector ratio
according to the World Banks Report (2018) refers to the financial support provided to the
private sector by financial corporation such as through loans and trade credits that establish
claim for repayments. Stolbov (2013) emphasized that an excessive domestic credit to private
sector ratio is highly related to the determinants of banking crisis in low income countries.
According to the National Bureau of Economic Forum paper (2003), they found out an inverse
relationship between concentration of banking system and banking failures. They believed that
the more market power a bank has, the more likely for it to obtain more than normal profits.
However, the reverse will be the case of bank failures when banks are less concentrated in a
country. Therefore, there was the need for this paper to consider the effects of banking
concentration in all countries included in the sample size for the study.
Remittance inflows as an indicator variable was taken into account as a regressor in the case of
determining the factors leading to systematic banking failures in SSA region. Remittance
inflows depicts the total amount of money or personal transfers sent from nationals living in
foreign countries to the home country (World Bank, 2015). Remittance inflows are normally
channel through banks and this is expected to strengthen the liquidity rate of recipient banks
and these banks acquire enough cash to survive any form of systematic banking distress (Adam,
Cuecuecha, & Page, 2008).
Finally, the presence of deposit insurance scheme was introduced in our regression as a dummy
variable, which takes the value of one and zero. The value 1 indicates period with an
introduction of deposit insurance scheme in a particular country and the value 0 indicates period
with no deposit insurance scheme in a particular country. Explicit deposit insurance scheme is
believed to increase the probability of systematic banking problems and this has been the
preferred choice of Financial Stability Board (FSB, 2012). As this was introduce in several
countries to reduce the incidence of self-fulfilling banking panics, it has however generated
another incidence of moral hazards which has not been carefully looked at through an
appropriate regulation and supervision by governments and banking regulatory bodies of which
SSA region is a victim (Demirgüc-Kunt & Detragiache, 1998).

Page | 33

3.5 Methodology
The approach to empirically identifying the determinants of banking crises in SSA region is
basically base on the research goals suggested in chapter one of this paper. Banking crisis in
this study refers to the dependent variable in the regression model, which is a dummy and takes
the value of 0 and 1, as indicated earlier. This implies that the probability that a banking crisis
occurring in a particular country at a particular time is Y=1 and the probability of banking crisis
not occurring is given by Y= 0. This is based on the assumption that the probability is the
function of the selected explanatory variables for the study. Below is the model’s definition for
the selected variables (both dependent and explanatory variables) and the logarithm likelihood
function of the model.
ln(������
1?������) = ?0 + ?1Xit + ?2Xit +…….. + ?k+1Xkit
Where;
ln = Logarithm of the function
Pit= Pr(Yit = 1), probability of banking crisis occurring in a particular country at a
particular time.
1-Pit = Pr(Yit = 0), probability of banking crisis not occurring in a particular country
at a particular time
Yit = banking crisis (dependent variable)
Xit = the vector of k explanatory variable observed for country i in year t
? = unknown estimated coefficients (vector of parameters)
i = Country
t = Time
In the model specification for the study, logistic function based on Demirgüc and Detragiache’s
(2002) was employed. With this, the interpretation from the regression output do not imply an
increase or decrease in the probability of crisis in a particular country in a given year when there
is an increase or decrease in the explanatory variables. However, it indicates the effect of

Page | 34

changes on the explanatory variable based on the logarithm likelihood function (ln(pit/1-pit),
so that the magnitude of the change associated with probability of a banking crisis occurring is
dependent on the intercept of the total estimated distribution function (Stock & Watson, 2012).
To figure out the factors that best contribute to banking crisis from the period of 1980-2016,
multivariate (binary) logit probability models was applied since banking crisis is a binary or
dummy variable. All the large values for the explanatory variables were included to establish
the predictive power of the leading indicators independently.
Moreover, some observations were realized from the primary panel dataset which reflects that
not all the 48 countries have experienced an episode of banking crises. To avoid subjective
judgements towards some countries, a sub-sample size was computed from the initial panel
dataset. This time, countries with no banking crises experience were excluded from the sample
and a new panel was construct which involves 33 countries in total. With this panel dataset, a
further logit regression was computed to figure out how these explanatory variables behave
from the regression model.
According to Hardy and Pazarbaloglu (1999), they emphasized that, a repeated banking crises
indicate weaknesses in banking sector that have not resolved permanently. This suggest that, if
a country experiences repeated banking crises, it is likely for other economic agents of growth
of that same country to suffer some macroeconomic schocks. To eliminante possible bias
towards countries, an atlernatinve approach was adopted to exclude countries who had suffered
just an episode of banking crises (preferrably in the first year the crises started) from the primary
panel dataset. With this, a new panel dataset excluding countries with just a year experience of
banking crisis episode was constructed. This concluded a new panel dataset of 25 countries
with repeated banking crises episodes from 1980 to 2016. Put this together, another
multivariate (binary) logit model anysis was constructed from the sub-sample of 25 countries
with the same explanatory variables selected to run the initial regression models. This was
estimated to establish the predictive power of explanatory variables independently leading to
countries with repeated banking distress.

Page | 35

CHAPTER FOUR
EMPIRICAL RESULTS AND ANALYSIS
4.0 Introduction
This chapter reports the empirical results and data analysis or interpretation section of this
paper. Analyses were first settled on the summary statistics of the selected variables in the panel
dataset and later was the main regression results presentation and interpretations.

4.1 Descriptive Statistics
This captures summary statistics for all variables (both dependent and explanatory or
independent) employed for the study. Table 5 reports the summary statistics of the selected
variables for the study and this contains the number of observations, mean, standard deviation,
maximum and minimum values of each variable. The observation depicts all the observed
values of the panel dataset captured in the regression. The mean values presented in Table 5 are
the average values of all the dataset for each variable and this represents a measure of central
tendency. Again, the standard deviation figures in the table refers to the measure of dispersion
of each variable about its means, indicating how close or far are the values of the dataset from
its means. Finally, the maximum and minimum values depict the largest and smallest value of
the dataset for each indicator variable, respectively.

Table 5. Summary Statistics of Variables
Variable Observation Mean Standard
Deviation
Maximum Minimum
Banking
Crisis
1481 0.08 0.271 1 0
Real GDP
Growth Rate
1776 3.580 7.815 149.970 -51.031
Changes in
Terms of
Trade
1775 -3.639 5.202 3.883 -1.120
Inflation 1771 50.828 827.910 24411.000 -35.837

Page | 36

Real Interest
Rate
1774 5.161 22.924 572.940 -94.220
Depreciation 1776 43.711 141.070 3463.800 0
M2/Reserves 1775 11.148 105.620 3986.500 0
Domestic
Credit to
Private Sector
Ratio
1774 15.792 19.634 160.120 0
Bank
Concentration
1776 26.549 38.624 100.000 0
Remittance
Inflows
1776 2.935 8.548 99.822 0
Deposit
Insurance
Coverage
(Dummy)
1776 0.108 0.311 1 0
Banking crisis is a dummy variable that takes the value of 0 and 1. The value 1 indicates the presence of banking
crisis in the given country and the value 0 indicates no presence of banking crisis occurring. Real GDP growth rate
is Annual percentage growth rate of GDP at market prices based on constant local currency. Changes in terms of
trade is terms of trade effect equals capacity to import less exports of goods and services in constant prices which
is in constant local currency. Inflation is the rate of changes of consumer price index for goods and services over
a time interval within a country. Rate of real interest rate is the nominal interest rate minus inflation over a period.
It considers how inflation erodes a purchasing power of investors. Depreciation is denoted by real effective
exchange rate that is, the rate of changes of foreign currencies as compared to domestic currency of a country.
M2/Reserves is the broad money total reserve ratio of a country which normally depends on the currency deposits
ratio and reserve deposit ratio. Private credit to GDP growth is the percentage change of growth of domestic credit
to private sector by financial corporations and organisations. Bank Concentration depicts the sum of all the larger
banks of a country’s bank scope divided by the sum all the banks in the same country’s bank scope. Remittance
inflows are the total amount money received by a country from its citizens living abroad. Finally, deposit insurance
coverage is a dummy variable that takes a value 1 and 0. The value indicates the period deposit insurance was
introduced and the value 0 indicates period with no introduction of deposit insurance scheme.

4.2 Main Results and Interpretations
Taking into account all the explanatory variables selected for the study, Table 6 presents the
main results from the regression outputs from the period between 1980 and 2016. The table
represents 4 models specifications which differs in the computations of all the selected variables
for the study and these were based on both previous and existing literatures. Model 1 consist of

Page | 37

all the available explanatory variables that may influence the probability of banking crises.
Model 2 contains the results of all the variables excluding remittance inflows. Model 3 also
contains results excluding remittance inflows and a dummy variable (deposit insurance
scheme). Finally, depreciation and remittance inflows variables were excluded in the results
captured in model 4. Table 6 again contains the number of countries or clusters considered in
running the regression models. The number of crises cases correctly predicted and finally the
proportion of these correctly predicted cases under each regression model.

Table 6. Regression Results from A Multivariate Binary Logit Models

Variable
Name

Model
(1)

Model
(2)

Model
(3)

Model
(4)

Constant

Real GDP growth

Change of TOT

Inflation

Real IR

Depreciation

M2/reserves

Domestic
Cr/Private
Bank
Concentration
Remittance Inflows

2.72e-052***
(-15.22)
3.16e-05 ***
(?4.161)
0.0102**
(?2.570)
0.1199
(1.555)
0.3785
(?0.8806)
0.0330**
(2.132)
0.6525
(?0.4503)
0.8366
(?0.2062)
1.82e-06***
(?4.772)
0.1513
(?1.435)
6.75e-056 ***
(?15.75)
1.25e-05***
?4.368
0.0118**
(?2.517)
0.1155
(1.574)
0.3659
(?0.9042)
0.0333**
(2.129)
0.7163
(?0.3634)
0.7660
(?0.2976)
1.58e-06***
(?4.801)
—–

9.09e-058 ***
(?16.02)
1.25e-05***
(?4.369)
0.0045***
(?2.838)
0.1170
(1.567)
0.3893
(?0.861)
0.0334**
(2.127)
0.7152
(?0.3648)
0.7577
(?0.3085)
1.53e-06***
(?4.807)
—–

5.78e-056 ***
(?15.76)
6.71e-06***
(?4.503)
0.0096***
(?2.591)
0.0883*
(1.704)
0.2273
(?1.207)
—-

0.7058
(?0.3775)
0.7397
(?0.3323)
1.45e-06 ***
(?4.818)
—–

Page | 38

Deposit Insurance
Scheme (dummy)

Number of
Countries

No. of observations

Numbers of crises
correctly predicted

% of correctly
predicted crises
0.7714
(0.2905)

48

1761

1643

93.3%

0.7229
(0.3546)

48

1761

1642

93.2%

—–

48

1761

1643

93.3%
0.7347
(0.3389)

48

1761

1640

93.1%

Statistical significance is represented as ***, ** and *, denoting significance at 1%, 5% and 10% levels,
respectively. Z-values indicated in the brackets. Banking Crisis (country = i, time = t) = ?0 + ?1 Real GDP growth rate +
?2 Changes of TOT + ?3 Inflation + ?4 Real Interest Rate + ?5 Depreciation + ?6 M2/reserve + ?7 Domestic Credit
to Private Sector ratio + ?8 Bank concentration + ?9 Remittance inflows + ?10 Deposit Insurance Scheme +
?it..Banking crisis is a dummy variable that takes the value of 0 and 1. The value 1 indicates the presence of
banking crisis in the given country and the value 0 indicates no presence of banking crisis occurring. Real GDP
growth rate is Annual percentage growth rate of GDP at market prices based on constant local currency. Changes
in terms of trade is terms of trade effect equals capacity to import less exports of goods and services in constant
prices which is in constant local currency. Inflation is the rate of changes of consumer price index for goods and
services over a time interval within a country. Rate of real interest rate is the nominal interest rate minus inflation
over a period. It considers how inflation erodes a purchasing power of investors. Depreciation is denoted by real
effective exchange rate that is, the rate of changes of foreign currencies as compared to domestic currency of a
country. M2/Reserves is the broad money total reserve ratio of a country which normally depends on the currency
deposits ratio and reserve deposit ratio. Private credit to GDP growth is the percentage change of growth of
domestic credit to private sector by financial corporations and organisations. Bank Concentration depicts the sum
of all the larger banks of a country’s bank scope divided by the sum all the banks in the same country’s bank scope.
Remittance inflows are the total amount money received by a country from its citizens living abroad. Finally,
deposit insurance coverage is a dummy variable that takes a value 1 and 0. The value indicates the period deposit
insurance was introduced and the value 0 indicates period with no introduction of deposit insurance scheme.

The results from Table 6 reveals that, real GDP growth rate is statistically significantly different
from zero at 1% significant level in all the model specifications. This indicate that, real GDP

Page | 39

growth rate has a very strong significant influence on the probability of SSA countries
experiencing banking sector problems. From the regression results, real GDP growth rate again
revealed an inverse relationship with the probability of banking crises. Implying that, a country
with low rate of economic growth will highly be exposed to banking crisis risk which
corresponds to the previous studies of Demirgüc-Kunt and Detragiache (1998). From a different
perspective, some studies also believe that, an occurring systematic banking crisis in a particular
region could also lead to a weak economic growth in that particular region (C. Pereira Pedro et
al.,2017).
The changes or volatility in terms of trade also reported a p-values significantly different from
zero at both 5% and 1% significant levels in Model 1 and 2, and 3 and 4, respectively, indicating
a very relevant influence on the outcome of banking crises in SSA. The estimated coefficients
for terms of trade changes reported a negative relationship with the probability of banking crises
in all the model specifications, implying that a decline in the terms of trade of a country is
therefore going to cause damages in the banking system of such country. Most of SSA countries
are characterized by balance of trade deficit due to their sole depends on import goods and
services. Countries under these circumstances will highly be exposed to banking sector
unsoundness, if they continue to depend on these foreign imports (Krugman et al., 2014).
Another potential determinant of banking crises that is very important in influencing the
probability of banking crises is the rate of inflation. However, from the above regression
outputs, it revealed no statistical significance in Model 1, 2 and 3 but saw a statistical
significance in Model 4 specification at 10% significant level when depreciation and remittance
inflows were excluded. Indicating a very low influence on the probability of banking crises in
SSA. The coefficients of inflation rate in all the models depicted a positive relationship with
the probability of banking crises. In that, a country experiencing high levels of inflation is
highly exposed to the probability of banking sector fragilities. In addition, real interest rate also
seems to be a very important determinant of a banking crisis in SSA. It is affecting negatively
the probability of banking crises as predicted in previous and already exiting studies. This
suggest that, a high nominal and real interest rates in many countries is fuelled by a greater
occurrence of banking sector challenges and shocks (Damount et al., 2004). With regard to rate
of depreciation, the variable appears to have a very high importance in influencing the
probability of banking crises in SSA. This also revealed p-values significantly different from
zero for Model 1,2 and 3 specifications at 5% significant level, hence it excluded model 4
specifications which reported no significance. As mention in the literature, currency

Page | 40

depreciation occurs when people borrow in foreign currencies and lend in local currencies
(Arestis & Sawyer, 2003). This causes foreign exchange rates to rise at the expense local
currencies and banks will highly be affected by these currency differences. However, banks in
the domestic country will be exposed to any form of financial distress if its currency depreciates
as against foreign country currency.
Broad money reserve ratio and domestic credit to private sector ratio is also a very important
determinant of banking crises and both variables revealed a negative influence in the probability
of banking crises occurrence, though they reported no statistical significance in all the model
specifications. This suggest that, broad money reserve ratio and domestic credit private sector
ratio not statistically significant predictor of banking crises in SSA according the regression
model.
Moreover, a very influential variable in the regression is banking concentration variable. It
revealed a highly significant value different from zero at 1% significant level. This suggest that,
the more banks are concentrated, the lower they are exposed to any patterns of banking crises,
as predicted by theory. This inverse relationship reflects in the results of all the model
specifications. Remittance inflows and deposit insurance scheme finally revealed no
significance in the determinants of banking crises according the results, though it reflected a
negative and positive relationship, respectively.
To provide sound judgement of the regression results presented in table 6, a further standard
logit regression was computed. Because the first results included a sample of 48 countries in
the panel dataset, estimations were made this time to exclude countries with no banking crises
episodes from the existing dataset. This resulted to sub-sample size of 33 countries who have
experienced banking crises within the period of this study and the same method of logit
regression model was computed. Table 7 presents the regression results of four different logit
models taken into account the same selected explanatory variables.

Page | 41

Table 7. Regression Results from Panel Excluding Countries with No Banking Crises Episode
Variable
Name
Model
(4)

Model
(5)

Model
(6)

Model
(7)

Constant

Real GDP growth

Change of TOT

Inflation

Real IR

Depreciation

M2/reserves

Domestic
Cr/Private
Bank
Concentration

Remittance Inflows

Deposit Insurance
Scheme (dummy)

3.71e-022***
(?9.679)
0.0004***
(?3.5359)
0.0168**
(?2.391)
0.1808
(1.338)
0.3869
(?0.8652)
0.2121
(1.248)
0.8022
(?0.2505)
0.1279
(?1.522)
1.29e-06***
(?4.841)

0.9390
(?0.0765)
0.4690
(?0.7241)

2.64e-023***
(?9.945)
0.0003***
(?3.594)
0.0168**
(?2.390)
0.1807
(1.339)
0.3876
(?0.8640)
0.2085
(1.258)
0.8047
(?0.2473)
0.1240
(?1.538)
1.22e-06***
(?4.853)

——

0.4697
(?0.7229)

5.44e-025***
(?10.32)
0.0004***
(?3.561)
0.0200**
(?2.326)
0.1751
(1.356)
0.3273
?0.9796
0.2022
(1.275)
0.8136
(?0.2358)
0.1327
(?1.503)
9.39e-07***
(?4.904)

——

——

1.70e-023***
(?9.989)
0.0002***
(?3.749)
0.0150**
(?2.433)
0.1625
(1.397)
0.2976
(?1.042)
——

0.8019
(?0.2509)
0.1071
(?1.611)
1.17e-06***
(?4.861)

——

0.4513
(?0.7532)

Page | 42

Number of
Countries

No. of observations

Numbers of crises
correctly predicted

% of correctly
predicted crises

33

1210

1097

90.7%

33

1210

1097

90.7%

33

1210

1097

90.7%

33

1210

1097

90.7%

Statistical significance is represented as ***, ** and *, denoting significance at 1%, 5% and 10% levels,
respectively. Z-values indicated in the brackets. Banking Crisis (country = i, time = t) = ?0 + ?1 Real GDP growth rate +
?2 Changes of TOT + ?3 Inflation + ?4 Real Interest Rate + ?5 Depreciation + ?6 M2/reserve + ?7 Domestic Credit
to Private Sector ratio + ?8 Bank concentration + ?9 Remittance inflows + ?10 Deposit Insurance Scheme +
?it..Banking crisis is a dummy variable that takes the value of 0 and 1. The value 1 indicates the presence of
banking crisis in the given country and the value 0 indicates no presence of banking crisis occurring. Real GDP
growth rate is Annual percentage growth rate of GDP at market prices based on constant local currency. Changes
in terms of trade is terms of trade effect equals capacity to import less exports of goods and services in constant
prices which is in constant local currency. Inflation is the rate of changes of consumer price index for goods and
services over a time interval within a country. Rate of real interest rate is the nominal interest rate minus inflation
over a period. It considers how inflation erodes a purchasing power of investors. Depreciation is denoted by real
effective exchange rate that is, the rate of changes of foreign currencies as compared to domestic currency of a
country. M2/Reserves is the broad money total reserve ratio of a country which normally depends on the currency
deposits ratio and reserve deposit ratio. Private credit to GDP growth is the percentage change of growth of
domestic credit to private sector by financial corporations and organisations. Bank Concentration depicts the sum
of all the larger banks of a country’s bank scope divided by the sum all the banks in the same country’s bank scope.
Remittance inflows are the total amount money received by a country from its citizens living abroad. Finally,
deposit insurance coverage is a dummy variable that takes a value 1 and 0. The value indicates the period deposit
insurance was introduced and the value 0 indicates period with no introduction of deposit insurance scheme.

In comparing the two regression tables (table 6 and 7), both seem to present the same results
from two different sample sizes. For instances, real GDP growth rate, changes in terms of trade
and bank concentration revealed significant values different from zero at either 1% or 5%
significant level. This suggests that, real GDP growth rate, terms of trade changes and bank
concentration are very strong in predicting the probability of bank crises in SSA in the two
tables. However, the rate of inflation and currency depreciation saw no significance in all the

Page | 43

four model specifications in Table 7 but they were very significant in predicting the probability
of banking crises in Table 6.
In additionally, observations were made to identify the correlation between these selected
explanatory variables and the occurrence of banking crises in Table 7. After excluding countries
without banking crises episodes, it revealed that, with the exceptions of depreciation and
inflation, all the other independent variables observed a negative correlation with the
probability of banking crises in all the four model specifications. A very important observation
identified is the correlation existing between deposit insurance scheme and the occurrence of
banking crises in Table 6 and 7. Deposit insurance scheme indicated a positive relationship with
the probability of banking crises in table 6 but revealed a negative effect in table 7.

4.3 Alternative Method of The Regression Analysis
According to Hardy and Pazarbaloglu (1999), they emphasized that, a repeated banking crises
indicate weaknesses in banking sector that have not resolved permanently. This suggest that, if
a country experiences repeated banking crises it is therefore likely for the other economic agents
of growth of that same country to suffer some macroeconomic schocks. To eliminante possible
bias towards countries, an atlernatinve approch was adopted to exclude countries who had
suffered just an episode of banking crises, preferrably in the first year the crises started from
the primary panel dataset. With this, a new panel dataset excluding countries with just a year
experience of banking crisis episode. This concluded a new panel dataset of 25 countries with
repeated banking crises episodes from 1980 to 2016. Put this together, another multivariate
logit model anysis was construct from the sub-sample of 25 countries with the same explanatory
variables selected to run the initial regression models. The regression results from panel dataset
including countries repeated banking crises episodes in four alternative model specifications is
presented in Table 8.

Page | 44

Table 8. Regression Results of Panel Dataset of Countries with Repeated Crises Episodes

Variable
Model
(8)

Model
(9)
Model
(10)
Model
(11)
Constant

Real GDP growth

Change of TOT

Inflation

Real IR

Depreciation

M2/reserves

Domestic
Cr/Private
Bank
Concentration
Remittance
Inflows
Deposit Insurance
Scheme (dummy)

8.02e-016***
(?8.054)
0.0016***
(?3.155)
0.0226**
(?2.281)
0.2076
(1.260)
0.3937
(?0.8529)
0.2595
(1.128)
0.8970
(?0.1295)
0.2488
(?1.153)
3.75e-07***
(?5.081)
0.4614
(0.7365)
0.2419
(?1.170)

8.27e-017***
(?8.327)
0.0020***
(?3.087)
0.0195**
(?2.336)
0.2088
(1.257)
0.3765
(?0.8844)
0.2876
(1.063)
0.8635
(?0.1719)
0.2564
(?1.135)
4.26e-07 ***
(?5.057)

—–
0.2400
(?1.175)

1.77e-018 ***
?8.771
0.0021***
(?3.081)
0.0363**
(?2.094)
0.1991
(1.284)
0.2828
(?1.074)
0.2644
(1.116)
0.8772
(?0.1546)
0.2714
(?1.100)
4.16e-07 ***
(?5.061)

—-

—-

6.03e-017 ***
?8.365
0.0011***
(?3.256)
0.0167**
(?2.393)
0.1901
(1.310)
0.3026
(?1.031)
—–

0.8615
(?0.1745)
0.2443
(?1.164)
4.11e-07 ***
(?5.064)

—–
0.2209
(?1.224)

Page | 45

Number of
Countries
No. of
observations
Numbers of crises
correctly predicted

% of correctly
predicted crises
25

916

809

88.3%

25

916

809

883.3%
25

916

810

88.4%

25

916

807

88.1%

Statistical significance is represented as ***, ** and *, denoting significance at 1%, 5% and 10% levels,
respectively. Z-values indicated in the brackets. Banking Crisis (country = i, time = t) = ?0 + ?1 Real GDP growth rate +
?2 Changes of TOT + ?3 Inflation + ?4 Real Interest Rate + ?5 Depreciation + ?6 M2/reserve + ?7 Domestic Credit
to Private Sector ratio + ?8 Bank concentration + ?9 Remittance inflows + ?10 Deposit Insurance Scheme +
?it..Banking crisis is a dummy variable that takes the value of 0 and 1. The value 1 indicates the presence of
banking crisis in the given country and the value 0 indicates no presence of banking crisis occurring. Real GDP
growth rate is Annual percentage growth rate of GDP at market prices based on constant local currency. Changes
in terms of trade is terms of trade effect equals capacity to import less exports of goods and services in constant
prices which is in constant local currency. Inflation is the rate of changes of consumer price index for goods and
services over a time interval within a country. Rate of real interest rate is the nominal interest rate minus inflation
over a period. It considers how inflation erodes a purchasing power of investors. Depreciation is denoted by real
effective exchange rate that is, the rate of changes of foreign currencies as compared to domestic currency of a
country. M2/Reserves is the broad money total reserve ratio of a country which normally depends on the currency
deposits ratio and reserve deposit ratio. Private credit to GDP growth is the percentage change of growth of
domestic credit to private sector by financial corporations and organisations. Bank Concentration depicts the sum
of all the larger banks of a country’s bank scope divided by the sum all the banks in the same country’s bank scope.
Remittance inflows are the total amount money received by a country from its citizens living abroad. Finally,
deposit insurance coverage is a dummy variable that takes a value 1 and 0. The value indicates the period deposit
insurance was introduced and the value 0 indicates period with no introduction of deposit insurance scheme.

From the alternative regression results in Table 8 (i.e., panel dataset of countries with repeated
banking crises episodes), real GDP growth rate, change in terms of trade and bank concentration
reported significant values different from zero at either 1% or 5 % significant level. This suggest
that economic growth rate, terms of trade changes and bank concentration have a strong
influence on the probability of banking crises of countries with repeated crises episodes.
Conversely, all the other explanatory variables reported no systematic significance in predicting

Page | 46

the probability of crises like the results presented in Table 7, though there were exclusions of
countries with a year banking crises experience.
Moreover, the rate of inflation, currency depreciation rate and remittance inflows variables
reported positive correlation with the probability of banking crises with repeated crises. In
contrast, real GDP growth rate, changes in terms of trade, real interest rate, domestic credit to
private sector ration, bank concentration, M2/reserves and deposit insurance scheme also
revealed a negative correlation with the probability of banking crises in these countries
according to the reported estimated coefficients from the regression outputs.

Page | 47

CHAPTER FIVE
SUMMARY AND CONCLUSION
5.0

Page | 48

REFERENCES
Adam, R. J., Cuecuecha, A., & Page, J. (2008). “Remittances, Consumption and Investment in
Ghana”.World Bank Policy Research Working Paper (WPS 4515), February.
Washington: World Bank.
Akerlof, G., & Romer, M. P. (1993). “Looting: The Economic Under-world of Bankruptcy for
Profit” Brookings papers on economic Activity: 2, . Brookings Institution, 1-73.
Allen, F., & Gale, D. (2007). Understanding Financial Crises (Ist edition pp.2-94). New York:
Oxford University Press.
Arestis, P., & Sawyer, M. (2003). On the Effectiveness of Monetary Policy and Fiscal Policy.
The Levy Economic Institute: Levy Economics Institute of Bard College.
Bank, W. (1986). Washington).: Cameroon: Financial Sector Report, Report No. 6028-CM.
Barth, J. R., Davis, E. P., & Levine, R. (2008). Bank Regulation and Changing: For better or
worse? Comparative Economic Studies, 50, 537-563. Chicago: University of Chicago
Press.
Beck, T., Demirguc, A., & Levine, R. (2006). Banking Concentration, Competion and Crisis:
First Results. Journal of banking and Finance, 1581-1603.
Beck, T., Demirgüc-Kunt, A., & levine, R. (2006). Bank concentration, competition and
crises: First results. Journal of Banking and Finance, 1581-1603.
Beck, T., Dewatripont, M., Freixas, X., Seabright, P., & Coyle, D. (2010). Bailing out the
banks: Reconciling stability and Competition, CEPR Policy Portal.
Bernanke, B., & Gertler, M. (1987). “Banking and Macroeconomic Equilibrium,” in New
Approaches to Monetory Economics, ed. by W. Barnett and K. Singleton. New York:
Cambridge University Press.
Bernard, A. B., Jensen, A. B., & Redding, S. J. (2007). Firms in International Trade. Journal
of Economic Perspectives 21, 105-130.
Bhattacharya, S., & Thakor, A. V. (October, 1994). Contemporary Banking Theory. Journal
of Financial Intermediation Vol. 3 , 2-50.
Boughton, J., & Elmus, W. (1984). A Reply to Trescott. Journal of Money, Credit, and
Banking 16 , 336-337.
Boughton, J., & Wicker, E. (1979). “The Behavior of the Currency-Deposit Ratio during the
Great Depression.”. Journal of Money, Credit, and Banking., 405-418.
Brock, P. L. (1995). “High Interest Rate and bank Recapitalization”. Policy Research Paper
No. 163 . Washington: World bank.
Brownsbridge, M. (1998). The Causes of Financial Distress in Local banks in Africa and
Implication for Prudential Policy. Geneva.

Page | 49

Bryant, J. (1980). A model of reserves, bank runs, and deposit insurance. Banking and
Finance 4 (December), 335.
Buyukkarabacak, B., & Valev, N. (2010). The Role of Household and Business Credit in
banking Crises. Journal of Banking and Finance, 1247-1256.
Calomiris, C. W., & Mason, J. R. (1997). Contagion and bank failures during the great
depression: The june 1932 Chicago banking panic. American Economic Review 87(5),
863-883.
Calomiris, C., & Beim, D. (2001). Emerging Financial Markets. McGraw-Hill/Irwin
Publishers: New York.
Calvo, G. A. (1996). Capital Flows and Macroeconomic Management. Tequila lessons.
International Journal of Finance and Economics. Vol .1(July), 207-240.
Calvo, G., & Mendoza, E. G. (2000). A Capital market crise and economic collapse in
emerging markets: an inforamational friction approach. American economic Review,
90, 59-64.
Caprio, G. J., & Klingebiel, D. (1996, July). Bank Insolvencies. Washington DC: The World
Bank. Pollcy Research Department, Finance and Private Sector Development
Division.
Caprio, J., & Klingebiel, G. (1999). Episodes of systematic and borderline financial crises.
Washington DC: The World Bank.
Carapella, F. (2015). Banking Panics and Deflation in Dynamic General E.quilibrium.
Finance and Economic Discussion Series, Board of Governors of the Federal Reserve
System (USA), 2001-2051.
Cihak, M., & Shaeck, K. (2010). How well do aggregate bank ratios identify banking
problems? Journal of Financial Stability, 6(3), 130-144.
Cihak, M., Demirguc–Kunt, A., Feyen, E., & Levine, R. (2012). Benchmarking Finanancial
Systems around the World. A Policy Reseaerach Working Paper WPS 6175.
Washington DC: World Bank.
Cooper, R., & Ross, T. W. (2002). “Bank runs: deposit insurance and capital requirements”.
Int Econ Rev. 43 (1), 55-72.
Daumont, R., Le Gall, F., & Leroux, F. (2004, April). Banking in Sub-Saharan Africa: What
went Wrong? International Monetary Fund Working Paper, pp. 1-48.
Davis, E. P., & Karim, D. (2008). Comparing early warning system for banking crises.
Journal of Financial Stability Vol. 4, 89-120.
Demirgüc-Kunt, A., & Detragiache, E. (1998). The Determinants of Banking Crises in
Developing and Developed Countries. Washington DC: International Monetory Fund.
Demirgüc-Kunt, A., & Detragiache, E. (2005). Cross-country empirical studies of systematic
bank distress: A Survey. National Institute of Economic Review, 192(1), 68-83.
Diamond, D. W. (2007). “Banks and liquidity creation: a simple exposition of the Diamond
Dybvig model” (PDF). Fed Res Bank Richmond Econ Q. 93(2), 189-200.

Page | 50

Diamond, D. W., & Dybvig, P. H. (1983). “Bank Runs, Deposit Insurance and Liquidity”.
Journal of Political Economy, Vol. 19 (June), 401-419.
Duttagupta, R., & Cashin, P. (2011). “Anatomy of Banking Crises in Developing and
Emerging Market Economies” 30. Journal of International Money and Finance, 354-
376.
Eichengreen, B., & Rose, A. K. (1998). Staying a flaot when the wind shifts: External Factore
and Emerging Markets of Banking Crises. Massachusetts Avenue Cambridge:
National Bureau of Economic Research.
English, W. B. (1996). “Inflation and Financial sector Size” Finance and Economic
Discussions Series No. 96/16. Washington: Board of Governors of the Federal
Reserve System.
Fischer, S. (1986). Indexing, Inflation and Economic Policy. London: The MIT Press,
Cambridge, Massachusetts.
Fisher, I. (1997). The Theory of Interest. Philadelphia: Porcupine Press.
Friedman, M., & Schwartz, A. J. (1963). A Monetory history of the United States from.
Pronceton: Princeton University Press (for National Bureau of Economic Research.
Frydl, E. J. (1999). "The Length and Cost of Banking Crises,” IMF Working Paper 99/30.
Washing: International Monetary Fund.
Fuller, R. L. (2011). “Phantom of Fear” The Banking Panic of 1933 . 16-22.
Galbis, V. (1993). “High Real Interest Rates under Financial Liberalization: Is There a
Problem?” IMF Working Paper 93/7. Washington: International Monetary Fund,
January.
Garber, P. M. (1996). managing Risk to Financial Markets from Volatile Capital Flows: The
Role of Prudential regulation. International Journal of Finance and Economics. Vol 1,
183-195.
Goldstein, M., ; Turner, P. (1996). Banking Crises in emerging Economies: Origins and
Policy Options. Basle: Bank for International Settlements-Monetory and Economic
Development.
Gunther, J. W., ; Rebel, C. A. (1995). Separating the Likelihood and Timing of Bank
Failure. Journal of Banking and Finance, Vol. 19, 1073-89.
Hardy, D. C., ; Pazarbaloglu, C. (1999). Determinants and Leading Indicators of banking
Crises. A Further Evidence. Palgrave Macmillan Journals on behalf of the
International Monetary Fund, 247-258.
Honohan, P. (1993). “Financial sector Failures in Western Africa,” The Journal of Modern
African Studies, Vol. 31, No. 1. Cambridge, United Kingdom: Cambridge University.
Jail, A. (2009). A New History Of Banking Panics In The United States, 1825-1929:.
Berkerley: University of Califonia.
Joyce, J. P. (2011). Financial Globalization and Banking Crises in Emerging Markets. Open
Economic Review 22(5), 875-895.

Page | 51

Kane, E. J. (1989). The S;L Insurance Mess: How Did It Happen? Washington: Urban
Institute Press.
Klingebiel, D., Kroszner, R., ; Laeven, L. (2007). “Banking Crises, Financial Dependence,
and Growth”. Journal of Financial Economics, Vol. 89, No. 1, 187-228.
Klomp, J. (2010). Causes of banking crises reviste. . North American Journal of Economic
and Finance, 21(1), 72-87.
Krugman, P. R., Obstfeld, M., ; Melitz, M. J. (2014). International econmics Theory and
Practice. Nineth edition. Califonia: University of Califonia Press.
Laeven, L., ; Honohan, P. (2005). Systemic Financial Crises: Containment and Resolution.
Cambridge, United Kingdom: cambridge University Press.
Laeven, L., ; Valencia, F. (2013). A Systematic Banking Crises Database. IMF Economic
Review 61(2), 225-270.
Leaven, L., ; Valencia, F. (2008). Systematic Banking Crises: A new database. International
Monetary Fund working paper No. 02/224. Washington DC: International Monetary
Fund.
Liederman, L., ; Reinhart, C. M. (1994, July). “The Capital Inflows Problem: Concept and
Issues”. Contemporary Econmic Policy, Vol. 12, pp. 54-66.
Lindgren, C.-J., Saal, M. I., ; Gillian, G. (1996). Bank Soundness and Macroeconomic
Policy. Washington: International Monetory Fund.
Mezui, M. C., Nalletamby, S., ; Kamewe, H. (2012). African Systemic Financial Crises Vol.
2. African Economic Brief, Chief Economist Complex.
Mishkin, F. S. (1996). “Understanding Financial Crises: A Developing Country" NBER
Working Paper No. 5600. (Cambridge, Massachusetts: National Bureau of Economic
Research.
Mishkin, S. (1996). Understanding Financial Crises. “A Development Country Perspective”.
Cambridge,Massachusetts: National Bureau of Economic Research.
Obtsfeld, M., & Rogoff, K. (1995). Th Mirage of Fixed Exchange rate. Journal of Economic
Perspectives, Vol. 9(Fall), 73-96.
Pereira Pedro, C., Ramalho, J. J., & Vidigal da Silva, J. (2017). The main determinants of
banking crises in OECD countries. Kiel Institute.
Powo Fosso, B. (2000). “Les déterminants des faillites bancaires dans les pays en
développement: Le cas des pays de l’Union Economique et Monétaire Ouest-Africaine
(UEMOA),” Cahier 02-2000, Centre de recherche et développement économique.
(Montréal: Université de Montréal).
Reinhart, C. M., & Kaminsky, G. (1996). The Twin Crises: The Causes of banking crises and
Balance -of-Payments Problems,” International Financial Discussion Paper.
Washington DC : Board of Governors of the Federal Reserve System.

Page | 52

Richardson, G. (2007, 03 20). “The collapse of the United States banking system during the
Great Depression, 1929 to 1933, new archival evidence” (PDF). Australas Account
Bus Finance J. 1 (1), pp. 39-50.
Rogoff, K., & Obstfeld, M. (n.d.). “The Mirage of Fixed Exchange Rates”. Journal of
Economic Perspectives, Vol. 9(Fall), 73-96.
Schich, S. (2009). Challenges Associated with the Expansion of Deposit insurance Coverage
during the Fall 2008. Paris: OECD.
Stiglitz, J. E. (1972). “Some Aspects of the Pure theoryof Corporate Finance: Bankcruptcies
and Takeovers”. Bell Journal Economics Vol. 3, 458-482.
Stiglitz, J. W., & Weiss, A. (1981). Credit Rationing in Markets with Imperfect Information.
The American Economic Review Vol. 71, No. 3, 393-410.
Stock, H. J., & Watson, M. M. (2012). Introduction to Econometrics. Third Editiion.
Edinburg: Pearson Education Limited.
Stolbov, M. (2013, October 26). Anatomy of international banking crises at the onset of the
Great Recession. pp. 1-15.
Sundararajan, V., & Baliño, T. J. (1991). Banking Crises: Cases and Issues. Washington:
International Monetary Fund.
Temin, P. (1976). Did Monetary Forces Cause the Great Depression?
Tenconi, R. (1993). “Restructuring of the Banking System in Guinea,” in African Finance:
Research and Reform, ed. by Lawrence H. White. San Francisco, California: ICS
Press.
Turkan, H. S. (2014). Banking Crises in Development Countries and the Case of Turkey.
Germany: German National library.
Von Hagen, J., & Ho, T. (2007). Money market pressure and the determinants of banking
crises. Journal of Money, Credit and Banking 39(5), 1037-1066.
Wicker, E. (1996). The Banking Panics of the Great Depression.

Page | 53

APPENDIX.
1. Sample Composition Countries
The targeted countries selected for the study
Angola, Benin, Botswana, Burkina Faso, Burundi, Cape Verde, Cameroon, Central African
Republic, Chad, Comoros, Demographic Republic of Congo, Congo Republic, Cote d’Ivoire,
Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Kenya,
Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia,
Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia,
South Africa, South Sudan, Sudan, Swaziland, Tanzania, Togo, Uganda, Zambia, Zimbabwe.

2. Countries Included from Sample No. 2 (Regression Model 9,10,11 and 12) – Table 8
Benin, Burkina Faso, Burundi, Cape Verde, Cameroon, Central African Republic, Chad,
Democratic Republic of Congo, Congo Republic, Cote d’Ivoire, Equatorial Guinea, Eritrea,
Ghana, Guinea, Guinea Bissau, Kenya, Madagascar, Mali, Mauritania, Mauritius,
Mozambique, Niger, Nigeria, Senegal, Sierra Leone, Somalia, South Africa, Swaziland,
Tanzania, Togo, Uganda, Zambia, Zimbabwe.

3. Countries Included from Sample No. 3 (Regression Model 9,10,11 and 12) – Table 8
Benin, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Democratic
Republic of Congo, Congo Republic, Cote d’Ivoire, Ghana, Guinea, Guinea Bissau, Kenya,
Mali, Mozambique, Niger, Nigeria, Senegal, Sierra Leone, South Africa, Swaziland, Tanzania,
Togo, Zambia, Zimbabwe.

Page | 54

4. Regression Output for Model 1

Model 1: Logit, using 1761 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian
coefficient std. error z p-value
————————————————————————————————————-
const ?2.17780 0.143114 ?15.22 2.72e-052 ***
RealGDPGrowthRate ?0.0527640 0.0126792 ?4.161 3.16e-05 ***
ChangeTermsofTra~ ?3.59287e-013 1.39784e-013 ?2.570 0.0102 **
Inflation 0.000142780 9.18203e-05 1.555 0.1199
RealInterestrate ?0.00369479 0.00419586 ?0.8806 0.3785
Depreciation 0.00102337 0.000479997 2.132 0.0330 **
M2Reserves ?0.00146403 0.00325091 ?0.4503 0.6525
CreditPrivateSec~ ?0.00127480 0.00618246 ?0.2062 0.8366
BankConcentration ?0.0248946 0.00521661 ?4.772 1.82e-06 ***
RemittanceInflows ?0.0314725 0.0219335 ?1.435 0.1513
DepositInsurance 0.103639 0.356705 0.2905 0.7714

Mean dependent var 0.067007 S.D. dependent var 0.250106
McFadden R-squared 0.108699 Adjusted R-squared 0.083289
Log-likelihood ?385.8474 Akaike criterion 793.6949
Schwarz criterion 853.9049 Hannan-Quinn 815.9453

Number of cases ‘correctly predicted’ = 1643 (93.3%)
f(beta’x) at mean of independent vars = 0.041
Likelihood ratio test: Chi-square (10) = 94.1123 0.0000

Predicted
0 1
Actual 0 1640 3
1 115 3

Excluding the constant, p-value was highest for variable 13 (Credit Private Sectorratio)

Page | 55

5. Regression Output for Model 2

Model 2: Logit, using 1761 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
————————————————————————————————————
const ?2.23308 0.141773 ?15.75 6.75e-056 ***
RealGDPGrowthRate ?0.0553347 0.0126681 ?4.368 1.25e-05 ***
ChangeTermsofTra~ ?3.52110e-013 1.39872e-013 ?2.517 0.0118 **
Inflation 0.000147127 9.34651e-05 1.574 0.1155
M2Reserves ?0.00112642 0.00309953 ?0.3634 0.7163
CreditPrivateSec~ ?0.00189883 0.00637943 ?0.2976 0.7660
RealInterestrate ?0.00385992 0.00426896 ?0.9042 0.3659
BankConcentrati ?0.0250343 0.00521477 ?4.801 1.58e-06 ***
DepositInsurance 0.126871 0.357752 0.3546 0.7229
Depreciation 0.00101556 0.000477025 2.129 0.0333 **

Mean dependent var 0.067007 S.D. dependent var 0.250106
McFadden R-squared 0.104787 Adjusted R-squared 0.081687
Log-likelihood ?387.5409 Akaike criterion 795.0819
Schwarz criterion 849.8182 Hannan-Quinn 815.3095

Number of cases ‘correctly predicted’ = 1642 (93.2%)
f(beta’x) at mean of independent vars = 0.042
Likelihood ratio test: Chi-square (9) = 90.7253 0.0000

Predicted
0 1
Actual 0 1639 4
1 115 3

Excluding the constant, p-value was highest for variable 13 (Credit Private Sector ratio)

Page | 56

6. Regression Output for Model 1

Model 3: Logit, using 1761 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
——————————————————————————————————-
const ?2.22358 0.138790 ?16.02 9.09e-058 ***
RealGDPGrowthRate ?0.0553031 0.0126578 ?4.369 1.25e-05 ***
ChangeTermsofTra~ ?3.71214e-013 1.30803e-013 ?2.838 0.0045 ***
Inflation 0.000147117 9.38653e-05 1.567 0.1170
M2Reserves ?0.00112805 0.00309210 ?0.3648 0.7152
CreditPrivateSec~ ?0.00196176 0.00635907 ?0.3085 0.7577
RealInterestrate ?0.00366408 0.00425630 ?0.8609 0.3893
Depreciation 0.00101385 0.000476609 2.127 0.0334 **
BankConcentration ?0.0248375 0.00516719 ?4.807 1.53e-06 ***

Mean dependent var 0.067007 S.D. dependent var 0.250106
McFadden R-squared 0.104645 Adjusted R-squared 0.083855
Log-likelihood ?387.6023 Akaike criterion 793.2046
Schwarz criterion 842.4674 Hannan-Quinn 811.4095

Number of cases ‘correctly predicted’ = 1643 (93.3%)
f(beta’x) at mean of independent vars = 0.042
Likelihood ratio test: Chi-square (8) = 90.6025 0.0000

Predicted
0 1
Actual 0 1640 3
1 115 3

Excluding the constant, p-value was highest for variable 13 (Credit Private Sector ratio)

Page | 57

7. Regression Output for Model 1

Model 4: Logit, using 1761 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
———————————————————————————————————-
const ?2.16528 0.137383 ?15.76 5.78e-056 ***
RealGDPGrowthRate ?0.0569222 0.0126418 ?4.503 6.71e-06 ***
ChangeTermsofTra~ ?3.65606e-013 1.41080e-013 ?2.591 0.0096 ***
Inflation 0.000175549 0.000103000 1.704 0.0883 *
RealInterestrate ?0.00551955 0.00457191 ?1.207 0.2273
M2Reserves ?0.00116377 0.00308314 ?0.3775 0.7058
CreditPrivateSec~ ?0.00214240 0.00644689 ?0.3323 0.7397
BankConcentration ?0.0250703 0.00520293 ?4.818 1.45e-06 ***
DepositInsurance 0.120825 0.356506 0.3389 0.7347

Mean dependent var 0.067007 S.D. dependent var 0.250106
McFadden R-squared 0.098903 Adjusted R-squared 0.078113
Log-likelihood ?390.0880 Akaike criterion 798.1760
Schwarz criterion 847.4387 Hannan-Quinn 816.3808

Number of cases ‘correctly predicted’ = 1640 (93.1%)
f(beta’x) at mean of independent vars = 0.042
Likelihood ratio test: Chi-square (8) = 85.6312 0.0000

Predicted
0 1
Actual 0 1639 4
1 117 1

Excluding the constant, p-value was highest for variable 13 (Credit Private Sector ratio)

Page | 58

8. Regression Output for Model 1

Model 5: Logit, using 1210 observations
Dependent variable: BankingCrisis
Standard errors based on Hessian

coefficient std. error z p-value
————————————————————————————————-
const ?1.60045 0.165356 ?9.679 3.71e-022 ***
RealGDPGrowthRate ?0.0597034 0.0168889 ?3.535 0.0004 ***
ChangeTermsofTra~ ?3.31148e-013 1.38499e-013 ?2.391 0.0168 **
Inflation 0.000135170 0.000100995 1.338 0.1808
RealInterestrate ?0.00387093 0.00447390 ?0.8652 0.3869
Depreciation 0.000544544 0.000436397 1.248 0.2121
M2Reserves ?0.000714987 0.00285403 ?0.2505 0.8022
CreditPrivateSec~ ?0.0110602 0.00726487 ?1.522 0.1279
BankConcentration ?0.0255678 0.00528131 ?4.841 1.29e-06 ***
RemittanceInflows ?0.00245133 0.0320335 ?0.07652 0.9390
DepositInsurance ?0.260552 0.359853 ?0.7241 0.4690

Mean dependent var 0.093388 S.D. dependent var 0.291096
McFadden R-squared 0.113230 Adjusted R-squared 0.083933
Log-likelihood ?332.9581 Akaike criterion 687.9161
Schwarz criterion 743.9983 Hannan-Quinn 709.0332

Number of cases ‘correctly predicted’ = 1097 (90.7%)
f(beta’x) at mean of independent vars = 0.057
Likelihood ratio test: Chi-square(10) = 85.0295 0.0000

Predicted
0 1
Actual 0 1095 2
1 111 2

Excluding the constant, p-value was highest for variable 14 (RemittanceInflows)

Page | 59

9. Regression Output for Model 1

Model 6: Logit, using 1210 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
—————————————————————————————————
const ?1.60331 0.161210 ?9.945 2.64e-023 ***
RealGDPGrowthRate ?0.0599150 0.0166693 ?3.594 0.0003 ***
ChangeTermsofTra~ ?3.30614e-013 1.38329e-013 ?2.390 0.0168 **
Inflation 0.000135118 0.000100941 1.339 0.1807
RealInterestrate ?0.00386336 0.00447174 ?0.8640 0.3876
Depreciation 0.000547289 0.000435187 1.258 0.2085
M2Reserves ?0.000697401 0.00281981 ?0.2473 0.8047
CreditPrivateSec~ ?0.0111239 0.00723224 ?1.538 0.1240
BankConcentration ?0.0255917 0.00527326 ?4.853 1.22e-06 ***
DepositInsurance ?0.260192 0.359927 ?0.7229 0.4697

Mean dependent var 0.093388 S.D. dependent var 0.291096
McFadden R-squared 0.113222 Adjusted R-squared 0.086589
Log-likelihood ?332.9610 Akaike criterion 685.9220
Schwarz criterion 736.9058 Hannan-Quinn 705.1194

Number of cases ‘correctly predicted’ = 1097 (90.7%)
f(beta’x) at mean of independent vars = 0.057
Likelihood ratio test: Chi-square(9) = 85.0236 0.0000

Predicted
0 1
Actual 0 1095 2
1 111 2

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 60

10. Regression Output for Model 1

Model 7: Logit, using 1210 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
—————————————————————————————————
const ?1.63003 0.157875 ?10.32 5.44e-025 ***
RealGDPGrowthRate ?0.0595556 0.0167246 ?3.561 0.0004 ***
ChangeTermsofTra~ ?2.92496e-013 1.25761e-013 ?2.326 0.0200 **
Inflation 0.000136247 0.000100485 1.356 0.1751
RealInterestrate ?0.00430420 0.00439379 ?0.9796 0.3273
Depreciation 0.000556097 0.000436045 1.275 0.2022
M2Reserves ?0.000656721 0.00278473 ?0.2358 0.8136
CreditPrivateSec~ ?0.0109139 0.00725941 ?1.503 0.1327
BankConcentration ?0.0259538 0.00529231 ?4.904 9.39e-07 ***

Mean dependent var 0.093388 S.D. dependent var 0.291096
McFadden R-squared 0.112492 Adjusted R-squared 0.088522
Log-likelihood ?333.2350 Akaike criterion 684.4700
Schwarz criterion 730.3554 Hannan-Quinn 701.7476

Number of cases ‘correctly predicted’ = 1097 (90.7%)
f(beta’x) at mean of independent vars = 0.057
Likelihood ratio test: Chi-square(8) = 84.4756 0.0000

Predicted
0 1
Actual 0 1095 2
1 111 2

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 61

11. Regression Output for Model 1

Model 8: Logit, using 1210 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
———————————————————————————————————
const ?1.54787 0.154953 ?9.989 1.70e-023 ***
RealGDPGrowthRate ?0.0622478 0.0166019 ?3.749 0.0002 ***
ChangeTermsofTra~ ?3.37727e-013 1.38817e-013 ?2.433 0.0150 **
Inflation 0.000154287 0.000110466 1.397 0.1625
RealInterestrate ?0.00497464 0.00477606 ?1.042 0.2976
M2Reserves ?0.000711647 0.00283692 ?0.2509 0.8019
CreditPrivateSec~ ?0.0117892 0.00731725 ?1.611 0.1071
BankConcentration ?0.0255800 0.00526202 ?4.861 1.17e-06 ***
DepositInsurance ?0.270721 0.359427 ?0.7532 0.4513

Mean dependent var 0.093388 S.D. dependent var 0.291096
McFadden R-squared 0.111094 Adjusted R-squared 0.087125
Log-likelihood ?333.7599 Akaike criterion 685.5198
Schwarz criterion 731.4052 Hannan-Quinn 702.7974

Number of cases ‘correctly predicted’ = 1097 (90.7%)
f(beta’x) at mean of independent vars = 0.057
Likelihood ratio test: Chi-square(8) = 83.4258 0.0000

Predicted
0 1
Actual 0 1095 2
1 111 2

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 62

12. Regression Output for Model 1

Model 9: Logit, using 916 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
———————————————————————————————————
const ?1.39307 0.172970 ?8.054 8.02e-016 ***
RealGDPGrowthRate ?0.0574541 0.0182109 ?3.155 0.0016 ***
ChangeTermsofTra~ ?3.26249e-013 1.43059e-013 ?2.281 0.0226 **
Inflation 0.000131262 0.000104168 1.260 0.2076
RealInterestrate ?0.00393150 0.00460936 ?0.8529 0.3937
Depreciation 0.000528884 0.000469001 1.128 0.2595
M2Reserves ?0.000236806 0.00182902 ?0.1295 0.8970
CreditPrivateSec~ ?0.00816025 0.00707511 ?1.153 0.2488
BankConcentration ?0.0275367 0.00541953 ?5.081 3.75e-07 ***
RemittanceInflows 0.0306548 0.0416219 0.7365 0.4614
DepositInsurance ?0.445419 0.380579 ?1.170 0.2419

Mean dependent var 0.117904 S.D. dependent var 0.322671
McFadden R-squared 0.124264 Adjusted R-squared 0.091157
Log-likelihood ?290.9709 Akaike criterion 603.9418
Schwarz criterion 656.9620 Hannan-Quinn 624.1788

Number of cases ‘correctly predicted’ = 809 (88.3%)
f(beta’x) at mean of independent vars = 0.071
Likelihood ratio test: Chi-square(10) = 82.5755 0.0000

Predicted
0 1
Actual 0 805 3
1 104 4

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 63

13. Regression Output for Model 1

Model 10: Logit, using 916 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
——————————————————————-
const ?1.35236 0.162400 ?8.327 8.27e-017 ***
RealGDPGrowthRate ?0.0553978 0.0179427 ?3.087 0.0020 ***
ChangeTermsofTra~ ?3.33698e-013 1.42868e-013 ?2.336 0.0195 **
Inflation 0.000132694 0.000105564 1.257 0.2088
RealInterestrate ?0.00410544 0.00464193 ?0.8844 0.3765
Depreciation 0.000490859 0.000461587 1.063 0.2876
M2Reserves ?0.000348022 0.00202446 ?0.1719 0.8635
CreditPrivateSec ?0.00786828 0.00693230 ?1.135 0.2564
BankConcentration ?0.0270795 0.00535482 ?5.057 4.26e-07 ***
DepositInsurance ?0.445909 0.379479 ?1.175 0.2400

Mean dependent var 0.117904 S.D. dependent var 0.322671
McFadden R-squared 0.123480 Adjusted R-squared 0.093383
Log-likelihood ?291.2315 Akaike criterion 602.4630
Schwarz criterion 650.6632 Hannan-Quinn 620.8603

Number of cases ‘correctly predicted’ = 809 (88.3%)
f(beta’x) at mean of independent vars = 0.072
Likelihood ratio test: Chi-square(9) = 82.0543 0.0000

Predicted
0 1
Actual 0 805 3
1 104 4

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 64

14. Regression Output for Model 1

Model 11: Logit, using 916 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
————————————————————————————————–
const ?1.39803 0.159396 ?8.771 1.77e-018 ***
RealGDPGrowthRate ?0.0555204 0.0180226 ?3.081 0.0021 ***
ChangeTermsofTra~ ?2.65694e-013 1.26897e-013 ?2.094 0.0363 **
Inflation 0.000134351 0.000104635 1.284 0.1991
RealInterestrate ?0.00485928 0.00452413 ?1.074 0.2828
Depreciation 0.000519179 0.000465190 1.116 0.2644
M2Reserves ?0.000306310 0.00198186 ?0.1546 0.8772
CreditPrivateSec~ ?0.00769870 0.00700049 ?1.100 0.2714
BankConcentration ?0.0273829 0.00541025 ?5.061 4.16e-07 ***

Mean dependent var 0.117904 S.D. dependent var 0.322671
McFadden R-squared 0.121230 Adjusted R-squared 0.094142
Log-likelihood ?291.9791 Akaike criterion 601.9582
Schwarz criterion 645.3384 Hannan-Quinn 618.5157

Number of cases ‘correctly predicted’ = 810 (88.4%)
f(beta’x) at mean of independent vars = 0.072
Likelihood ratio test: Chi-square (8) = 80.5591 0.0000

Predicted
0 1
Actual 0 806 2
1 104 4

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 65

15. Regression Output for Model 1

Model 12: Logit, using 916 observations
Dependent variable: Banking Crisis
Standard errors based on Hessian

coefficient std. error z p-value
———————————————————————————————————
const ?1.30427 0.155927 ?8.365 6.03e-017 ***
RealGDPGrowthRate ?0.0579602 0.0178003 ?3.256 0.0011 ***
ChangeTermsofTra~ ?3.42819e-013 1.43267e-013 ?2.393 0.0167 **
Inflation 0.000149502 0.000114109 1.310 0.1901
RealInterestrate ?0.00506179 0.00491038 ?1.031 0.3026
M2Reserves ?0.000357919 0.00205135 ?0.1745 0.8615
CreditPrivateSec~ ?0.00814466 0.00699550 ?1.164 0.2443
BankConcentration ?0.0270749 0.00534680 ?5.064 4.11e-07 ***
DepositInsurance ?0.463730 0.378850 ?1.224 0.2209

Mean dependent var 0.117904 S.D. dependent var 0.322671
McFadden R-squared 0.121712 Adjusted R-squared 0.094624
Log-likelihood ?291.8189 Akaike criterion 601.6378
Schwarz criterion 645.0180 Hannan-Quinn 618.1953

Number of cases ‘correctly predicted’ = 807 (88.1%)
f(beta’x) at mean of independent vars = 0.072
Likelihood ratio test: Chi-square (8) = 80.8795 0.0000

Predicted
0 1
Actual 0 805 3
1 106 2

Excluding the constant, p-value was highest for variable 10 (M2Reserves)

Page | 66

DECLARATION
I hereby declare that this piece of written work is the result of my own independent scholarly
work, and that in all cases material from the work of others (in books, journal articles, essays,
dissertations, on the internet, etc.) is acknowledged, and quotations and paraphrases are clearly
indicated. No material other than that listed has been used. This written work has not previously
been used as examination material at this or any other university. This written work has not yet
been published.

Erlangen, 28.03.2018.