University of Ghana http://ugspace.ug.edu.gh
 
UNIVERSITY OF GHANA 
 
IMPACT OF EXTREME EVENTS ON THE INSURANCE MARKET IN  
              GHANA 
 
 
       BY 
 
     SUSANA ADOBEA YAMOAH 
      (10598799) 
 
 
THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, 
LEGON IN PARTIAL FULFILLMENT OF 
THE REQUIREMENT FOR THE AWARD OF MPHIL RISK 
MANAGEMENT DEGREE 
 
 
     JULY 2018 
 
University of Ghana http://ugspace.ug.edu.gh
DECLARATION 
I hereby declare that this thesis is a record of an original work done by me and the results 
embodied in this thesis have not been submitted in whole or part to any other university or 
institute for the award of any degree. 
 
Signed: ............................ Date: ..................... 
Susana Adobea Yamoah (10276392) 
   
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 i 
 
University of Ghana http://ugspace.ug.edu.gh
     CERTIFICATION 
We hereby certify that this thesis was prepared from the candidate’s own work and 
supervised in accordance with guidelines on supervision of thesis laid down by the University 
of Ghana. 
 
Signed: ........................... Date: ........................ 
Dr. Charles Andoh 
(Principal Supervisor) 
 
Signed: ............................ Date: ........................ 
Dr. Vera Fiador 
(Co - Supervisor) 
 
 
 
 
 
 
 
 
 
 
 
 ii 
 
University of Ghana http://ugspace.ug.edu.gh
ABSTRACT 
The study examines how the frequent occurrence and severity of extreme events affect the 
demand, supply and the profitability of insurance companies in Ghana. Secondary data 
collected from NIC, NADMO, Ghana Statistical Service and Open Data Initiative on eight 
classes of business as well as thirteen life and non-life companies over the period 2007 – 
2016 was analysed using a panel regression method. 
The results indicated that the unexpected frequency of extreme events negatively affected 
both the demand and supply of insurance but had no influence on insurers’ profitability.  
Also, the unexpected severity of extreme events significantly decreased the profitability of 
life insurers but had no effect on non-life insurers.  
Therefore other avenues, such as public awareness and education can be explored in order to 
increase insurance demand and ensure the availability and affordability of insurance to all. 
Insurers are also encouraged to consistently assess the exposures of their investments, capital 
and reserves to extreme events.  
Keywords and phrases: extreme events, panel regression, profitability, reserves, severity, 
vulnerability 
 
 
 
 
 
 
 
 iii 
 
University of Ghana http://ugspace.ug.edu.gh
DEDICATION 
I dedicate this work to my parents, Mr. Daniel Yamoah and Mrs. Esther Yamoah, and Mr. 
Ebenezer Nyarko Kumi. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 iv 
 
University of Ghana http://ugspace.ug.edu.gh
    ACKNOWLEDGEMENT 
To begin with, I would like to thank the Almighty God for his special grace, mercies and 
wisdom throughout this period of study. I would also like to acknowledge my thesis 
supervisors, Dr. Charles Andoh and Dr. Vera Fiador, for their invaluable comments on this 
thesis and steering me in the right the direction whenever I needed it. 
Again my profound gratitude goes to my parents, Mr. Daniel Yamoah and Mrs. Esther 
Yamoah, my siblings and to all my family for consistently supporting and encouraging me 
throughout my years of study and through the process of researching and writing this thesis. I 
am gratefully indebted to you. 
I would like to appreciate Mr. Ebenezer Nyarko Kumi for his source of inspiration, guidance, 
and support especially during challenging and difficult periods of the research. This 
accomplishment would not have been possible without him. Thanks should also be bestowed 
upon my roommates, Edwina Quist and Gloria Adinyira for the tremendous support and 
advice. 
Finally, I would also like to thank my colleagues for their encouragement and help in diverse 
ways throughout my graduate studies. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 v 
 
University of Ghana http://ugspace.ug.edu.gh
 
TABLE OF CONTENTS 
DECLARATION ........................................................................................................................ i 
CERTIFICATION ..................................................................................................................... ii 
ABSTRACT ............................................................................................................................. iii 
DEDICATION .......................................................................................................................... iv 
ACKNOWLEDGEMENT ......................................................................................................... v 
TABLE OF CONTENTS .......................................................................................................... vi 
LIST OF TABLES ..................................................................................................................... x 
LIST OF FIGURES ................................................................................................................. xii 
CHAPTER ONE ........................................................................................................................ 1 
INTRODUCTION ..................................................................................................................... 1 
1.1 Background of Study ....................................................................................................... 1 
1.2 Statement of the Problem ................................................................................................. 3 
1.3 Purpose of Study .............................................................................................................. 5 
1.4 Research Objectives ......................................................................................................... 5 
1.5 Research Questions .......................................................................................................... 6 
1.6 Significance of Study ....................................................................................................... 6 
1.7 Scope and Limitation of Study ........................................................................................ 7 
1.8 Thesis Structure ............................................................................................................... 7 
CHAPTER TWO ....................................................................................................................... 8 
LITERATURE REVIEW .......................................................................................................... 8 
2.1 Introduction ...................................................................................................................... 8 
2.2 The Concept of Extreme Events ...................................................................................... 8 
2.2.1 Nature of Extreme Events ......................................................................................... 8 
2.2.2 Categories of Extreme Events ................................................................................... 9 
2.3 Extreme Value Theory ................................................................................................... 13 
 vi 
 
University of Ghana http://ugspace.ug.edu.gh
2.3.1 Applications of Extreme Value Theory .................................................................. 13 
2.4 Risk of Extreme Events ................................................................................................. 14 
2.4.1 The Frequency and Intensity of Extreme Events .................................................... 15 
2.4.2 Influence of Socio-economic Factors on Extreme Events Losses .......................... 16 
2.5 Impact of Extreme Events on Economies ...................................................................... 17 
2.6 Extreme Risk Management ............................................................................................ 19 
2.6.1 Insurance as a Tool for Risk Management ............................................................. 21 
2.6.2 Insurance and Economic Growth ............................................................................ 22 
2.7 Empirical Findings on the Impact of Extreme Events on the Insurance Market ........... 25 
2.7.1 Insurance Demand and Extreme Events ................................................................. 25 
2.7.2 Insurability of risk (Supply of Insurance) ............................................................... 28 
2.7.3 Profitability of insurance companies ...................................................................... 31 
CHAPTER THREE ................................................................................................................. 40 
METHODOLOGY .................................................................................................................. 40 
3.1 Introduction .................................................................................................................... 40 
3.2 Research Design............................................................................................................. 40 
3.2.1 Quantitative Research Methodology ....................................................................... 40 
3.2.2 Inferential Statistical Analysis ................................................................................ 41 
3.3 Research Population and Sample ................................................................................... 41 
3.4 Sampling Procedure ....................................................................................................... 42 
3.5 Data Collection .............................................................................................................. 42 
3.6 Method of Analysis ........................................................................................................ 43 
3.6.1 Panel Data Analysis ................................................................................................ 43 
3.6.2 Frequency versus Severity ...................................................................................... 44 
3.6.3 Demand for Insurance ............................................................................................. 45 
3.6.4 Insurability of Risk (Supply of Insurance).............................................................. 47 
3.6.5 Profitability of Insurance Companies ..................................................................... 48 
 vii 
 
University of Ghana http://ugspace.ug.edu.gh
3.7 Variables Construction and Expected Signs .................................................................. 50 
3.8 Limitations to the Methodology..................................................................................... 52 
DATA ANALYSIS, ESTIMATION AND DISCUSSION OF RESULTS............................. 53 
4.1 Introduction .................................................................................................................... 53 
4.2 Extreme Events and Demand for Insurance ................................................................... 53 
4.2.1 Descriptive Statistics ............................................................................................... 53 
4.2.2 Multicollinearity Test.............................................................................................. 54 
4.2.3 Test for Best Panel Model....................................................................................... 56 
4.2.4 Results of Panel Regression .................................................................................... 56 
4.2.5 Diagnostics Test ...................................................................................................... 59 
4.3 Extreme Events and Supply of Insurance ...................................................................... 60 
4.3.1 Descriptive Statistics ............................................................................................... 60 
4.3.2 Multicollinearity Test.............................................................................................. 60 
4.3.3 Test for Best Panel Model....................................................................................... 62 
4.3.4 Results of Panel Regression .................................................................................... 62 
4.3.5 Relationship between Supply and the Interaction of Claims Ratio and Reinsurance
.......................................................................................................................................... 64 
4.3.6 Diagnostics Test ...................................................................................................... 65 
4.4 Extreme Events and Profitability of Life Insurance Companies ................................... 65 
4.4.1 Descriptive Statistics ............................................................................................... 65 
4.4.2 Multicollinearity Test.............................................................................................. 66 
4.4.3 Test for Best Panel Model....................................................................................... 68 
4.4.4 Results of Panel Regression .................................................................................... 68 
4.4.5 Diagnostics Test ...................................................................................................... 71 
4.5 Extreme Events and Profitability of Non-life/General Insurance Companies ............... 72 
4.5.1 Descriptive Analysis ............................................................................................... 72 
4.5.2 Multicollinearity Test.............................................................................................. 73 
 viii 
 
University of Ghana http://ugspace.ug.edu.gh
4.5.3 Test for Best Panel Model....................................................................................... 74 
4.5.4 Results of Panel Regression .................................................................................... 75 
4.5.5 Diagnostics Test ...................................................................................................... 78 
CHAPTER FIVE ..................................................................................................................... 80 
SUMMARY, CONCLUSION AND RECOMMENDATIONS .............................................. 80 
5.1 Introduction .................................................................................................................... 80 
5.2 Summary of Findings ..................................................................................................... 80 
5.3 Conclusion ................................................................................................................. 81 
5.4. Recommendations for Management and Policymakers ................................................ 82 
5.5 Limitation of Study and Direction for Future Research ................................................ 83 
REFERENCES ........................................................................................................................ 84 
APPENDIX .............................................................................................................................. 93 
REGRESSION RESULTS OF INTERACTIVE VARIABLE ............................................ 93 
APPROPRAITE MODEL SELECTION USING AIC AND BIC ...................................... 93 
Supply Model ................................................................................................................... 93 
Life Profitability Model ................................................................................................... 94 
RELATIONSHIP BETWEEN SIZE SQUARED AND GENERAL INSURERS’ 
PROFITABILITY ................................................................................................................ 95 
 
 
 
 
 
 
 
 
 ix 
 
University of Ghana http://ugspace.ug.edu.gh
LIST OF TABLES 
Table 2.1:  Some Extreme Events in Ghana from 1900 to 2016 ............................................. 35 
Table 3.1: Definition of Dependent Variables ......................................................................... 51 
Table 3.2: Definition of Independent Variables and Expected Signs ...................................... 51 
Table 4.1: Descriptive Statistics .............................................................................................. 54 
Table 4.2: Correlation Matrix .................................................................................................. 55 
Table 4.3: Testing for Panel Effect .......................................................................................... 55 
Table 4.4: Hausman Test ......................................................................................................... 56 
Table 4.5: Regression Results of Demand for Insurance ......................................................... 57 
Table 4. 6: Diagnostics Test for Serial Correlation and Cross-sectional Dependence ............ 59 
Table 4. 7: Descriptive Statistics ............................................................................................. 60 
Table 4. 8: Correlation Matrix ................................................................................................. 60 
Table 4. 9: Testing for Panel Effect ......................................................................................... 61 
Table 4. 10: Hausman Test ...................................................................................................... 62 
Table 4. 11: Regression Results of Supply of Insurance ......................................................... 62 
Table 4.12: Regression Results of Supply of Insurance (With CR_Reins) ............................. 64 
Table 4. 13: Diagnostics Test for Serial Correlation and Cross-sectional Dependence .......... 65 
Table 4. 14: Descriptive Statistics ........................................................................................... 66 
Table 4. 15: Correlation Matrix ............................................................................................... 67 
Table 4. 16: Testing for Panel Effect ....................................................................................... 67 
Table 4. 17: Hausman Test ...................................................................................................... 68 
Table 4. 18: Regression Results of Profitability of Life Insurance Companies ....................... 68 
Table 4. 19: Regression Results of Profitability of Life Companies (Without Size) .............. 69 
Table 4. 20: Diagnostics Test for Serial Correlation and Cross-sectional Dependence .......... 71 
Table 4. 21: Descriptive Statistics ........................................................................................... 72 
 x 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 22: Correlation Matrix ............................................................................................... 73 
Table 4. 23: Testing for Panel Effect ....................................................................................... 74 
Table 4. 24: Hausman Test ...................................................................................................... 74 
Table 4. 25: Regression Results of Profitability of Non-life Insurance Companies ................ 75 
Table 4. 26: Regression Results of Profitability of Non-life Companies (Including Size 
Squared) ................................................................................................................................... 77 
Table 4. 27: Diagnostics Test for Serial Correlation and Cross-sectional Dependence .......... 79 
Table 1: Regression results (Without Reins) ........................................................................... 93 
Table 2: Regression Results (Without CR) .............................................................................. 93 
Table 3: AIC/BIC of Model (Without Reins) .......................................................................... 93 
Table 4: AIC/BIC of Model (Without CR) .............................................................................. 94 
Table 5: AIC/BIC of Model (Without Both CR and Reins) .................................................... 94 
Table 6:  AIC/BIC of Model including all Variables .............................................................. 94 
Table 7: AIC/BIC of Model (Without size) ............................................................................. 94 
Table 8: AIC/BIC of Model (Without PremSurp) ................................................................... 94 
Table 9: AIC/BIC OF Model (Without both Size and PremSurp)........................................... 94 
 
 
 
 
 
  
 
 xi 
 
University of Ghana http://ugspace.ug.edu.gh
    LIST OF FIGURES 
Figure 2.1:       Real GDP Growth, 31 Largest African Economies (2010-2019, Compound   
Annual Growth Rates, in %) .................................................................................................... 24 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 xii 
 
University of Ghana http://ugspace.ug.edu.gh
                                            LIST OF ABBREVIATION 
AIC                         Akaike Information Criteria 
AMS                       American Meteorological Society 
BIC                         Bayesian Information Criterion 
CR                          Claims Ratio 
CRED                     Centre for Research on the Epidemiology of Disasters 
CSIL                       Cambridge Institute for Sustainability Leadership 
ERM                       Extreme Risk Management 
EVT                        Extreme Value Theory 
EY                           Ernst & Young 
GDP                        Gross Domestic Product 
GFDRR                   Global Facility for Disaster Reduction and Recovery 
GSS                         Ghana Statistical Service 
HHI                         Herfindahl-Hirschman Index 
IFC                          International Financial Corporation 
IFRC                       International Federation of Red Cross 
IPCC                       Intergovernmental Panel on Climate Change 
ISDR                       International Strategy for Disaster Reduction 
NADMO                 National Disaster Management Organisation 
OFDA                     Office of Foreign Disaster Assistance 
OLS                         Ordinary Least Square 
ROA                        Return On Asset 
ROE                        Return On Equity 
ROI                         Return On Investment 
TT-DEWCE           Task Team on the Definition of Extreme Weather and Climate Events 
 xiii 
 
University of Ghana http://ugspace.ug.edu.gh
UNDP                     United Nations Development Programme 
UNEP FI                 United Nations Environment Programme Finance Initiative 
WMO                     World Meteorological Organisation 
WTP                       Willingness To Pay 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 xiv 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER ONE 
           INTRODUCTION 
1.1 Background of Study 
Extreme events have long been the focus of atmospheric sciences and have raised 
considerable concern of the public, policy makers and scientists because of their significant 
social and economic impacts. According to Stephenson (2008), extreme events are relatively 
difficult to define even though they are generally easy to recognize. However, Stephenson 
(2008) defines extreme events as events that have extreme values of certain important 
meteorological variables, such as, large amount of precipitation (e.g. flood), high wind speeds 
(e.g. cyclones) and high temperatures (e.g. heat waves). Task Team on the Definition of 
Extreme Weather and Climate Events (TT-DEWCE) (2016) further defines an extreme event 
as any event with high economic loss. Extreme event is also seen as a severe event with low 
probability of incidence but potentially high associated sicknesses, mortality or economic loss 
(Siri, Newell, Proust, & Capon, 2016). Extreme events include not only meteorological 
(extreme temperature, floods), but climatological (drought, famine), geological (earthquakes), 
epidemiological (epidemic), and anthropogenic (terrorism, fire) events. Various processes 
give rise to extreme events and according to Stephenson (2008), the movement of a weather 
system into a new spatial location or into a different time period and the simultaneous 
coincidence of several non-extreme conditions can cause extreme events. Also, slower 
disparities in the climate system causing persistence or frequent recurrence of weather can 
lead to recurring extremes.  
Extreme events cause both direct and indirect damages to economies and individuals. Direct 
damages can be seen as the loss of life, injury, damages to raw materials, capital and other 
fixed assets as a result of these events. Indirect damages include loss of productivity and 
 1 
 
University of Ghana http://ugspace.ug.edu.gh
income due to the destruction of resources (Cavallo & Noy, 2009). Droughts, floods, 
windstorms, extreme temperatures, wildfires and epidemics have caused about 8,835 
disasters, 1.94 million deaths and US$ 2.4 trillion of economic losses globally from 1970 to 
2012 (WMO, 2014).  
The frequency of extreme weather events rose from 38 in 1980 to 174 events in 2014 globally 
which is more than 400 percentage increase (Müller-Fürstenberger & Schumacher, 2015). 
The rise in the severity and occurrence of extreme events can be attributed to urbanization of 
the population, growing population intensity and assets in risky areas, and the possible impact 
of climate change (global warming) (Kunreuther & Michel‐Kerjan, 2009; Cummins & 
Mahul, 2009). Increase losses from extreme events for the past 20 years have made 
developing countries particularly helpless since they depend highly on agriculture, lack 
adequate resources to prepare and respond effectively to emergency (Courbage & Mahul, 
2013).  
According to the report from the UNDP (2013), Ghana’s main exposure to extreme events is 
flooding, epidemics, wildfires and drought. Ghana has experienced 29 disasters over a 30 
year-period from 1980 to 2010. Of the 29 disasters, 14 were epidemics, 13 were floods and 
there was one each of drought and wildfire (Disaster and Risk Profile:Preventionweb, 2017). 
In 1983 Ghana experienced a drought which affected 12.5 million people. In June 1995, the 
entire city of Accra was flooded leaving about 70 people dead. Again, in 2007 the Upper 
East, Northern, Upper West, and some parts of the Western regions experienced massive 
floods causing the death of  about 56 people and affecting about 332,600 people (Boah-
Mensah, 2017). The most recent disaster which hit the country hardest was the June 3rd, 2015 
flood and fire disaster. More than 200 lives and several properties were destroyed as a result 
of the disaster (Boah-Mensah, 2017). 
 
 2 
 
University of Ghana http://ugspace.ug.edu.gh
1.2 Statement of the Problem 
Insurance is one of the main ways in which societies currently deals with the risks of extreme 
hazards, such as windstorms and floods as well as the losses associated with them (UNEP FI, 
2014; Linnerooth-Baye & Mechler 2009). According to Coomber (2006), insurance price and 
diversify risk faced by the society, and its capital base is used to partially absorb the impact 
of unavoidable shocks. There are various ways in which insurance provides assistance for the 
loss of assets, incomes and lives due to extreme events. Right after a disaster, individuals are 
indemnified and restored back to their original financial position before the loss. This aids in 
protecting the livelihood of individuals and preventing business interruptions. Also, 
government is able to finance the loss associated with these events and avoid fiscal deficits 
(Schäfer, Waters, Kreft, & Zissener, 2016: Linnerooth-Baye & Mechler, 2009). Though the 
global losses from a disastrous event cannot be prevented by insurance, by paying a small 
amount of premium those at risk can be protected against a large loss which can adversely 
affect their finances (Kunreuther & Michel-Kerjan, 2007). 
The Intergovernmental Panel on Climate Change has predicted an increase in the risks 
associated with extreme events as the global mean temperature rises (IPCC, 2014). This is 
likely to cause a significant surge in losses in the coming years, increasing the vulnerability 
of developing countries to extreme events and so, the role of insurance is very crucial. 
However, insurance is practically very low in developing countries even though they bear a 
greater percentage of the burden, in terms of both casualties and direct economic damages 
(Linnerooth-Bayer & Mechler, 2007; Lester, 2009; Cavallo & Noy, 2009). Cummins and 
Mahul (2009) in their book made known that developed economies mostly use compulsory 
insurance to cover about 40 percent of the direct losses from natural disasters. On the 
contrary, less than 10 percent of these losses are covered by insurance in middle-income 
countries and less than 5 percent in low-income countries.  
 3 
 
University of Ghana http://ugspace.ug.edu.gh
Due to the lack of insurance cover, rebuilding after an extreme event may be delayed and 
may negatively impact economic growth in developing countries in the long run. To ease 
their burdens, governments of developing countries, such as Ghana, are promoting insurance 
as one of the key ways of financing extreme events risks (Ghana National Climate Change 
Policy, 2012). Currently, insurance markets in developing economies are growing rapidly 
even though they account for only about 16 percent of the global insurance market 
(International Financial Corporation, 2016). The question therefore is, whether extreme 
events can affect the growth and sustainability of the insurance market especially in 
developing countries, and if so, what mechanisms of demand and supply reactions comprise 
the effect. Furthermore, how will extreme events affect the industry’s ability to cover future 
losses? A study by Turner and Deng (2015) shows that a striking risk of high intensity from a 
major event could raise risk awareness and therefore increase insurance demand. On the 
supply side, the depletion of capital reserves by a major loss can intensify competition among 
insurers, prompting some to exit the market. The remaining insurers raises their premiums, 
with the hope of recouping some of their losses, which ultimately becomes higher than what 
policyholders are prepared to pay. This could affect the supply of insurance for people and 
their properties (Herweijer, Ranger, & Ward, 2009). Although insurance connects the world 
through risk-sharing agreement, the major objective of insurance companies is to make 
profits for shareholders. However, the growing exposures and intensities of risk due to 
extreme events increases insurers’ capital requirements and challenge their profitability and 
insurability (Coomber, 2006). Insurance companies keep reserves, capital and financial assets 
to pay claims as and when they occur. Insurers invest part of their capital in equities and 
property, in many cases, and they may be negatively affected by extreme events due to 
climate change. Also, an upsurge of extreme events may cause regulatory capital required to 
 4 
 
University of Ghana http://ugspace.ug.edu.gh
increase. Assets are also likely to fall in value as the liabilities and capital requirements are 
rising and this may eventually affect the profitability of insurers (Maynard, 2008).  
Therefore, analyzing the insurability of risk (supply of) and demand for insurance, especially 
in developing countries like Ghana, post extreme events and their influence on the 
profitability of insurers is necessary to develop a broad picture of the extreme event risk 
problem and evaluate feasible solutions. 
 
1.3 Purpose of Study 
The study provides some understanding of the risks associated with extreme events since 
developing solutions also require a comprehensive understanding of how they affect the 
operations of the insurance industry in order to employ efficient systems for monitoring and 
estimating these events so that resilient societies can be built. 
 
 1.4 Research Objectives 
The main objective of this study is to examine the impact of extreme events on the insurance 
market in Ghana. 
The specific objectives include: 
i. Determine how extreme events affect demand for insurance.  
ii. Determine the effects of extreme events on insurability of risk. 
iii. Examine the impacts of extreme events on the profitability of the Ghanaian insurance 
firms. 
 
 5 
 
University of Ghana http://ugspace.ug.edu.gh
1.5 Research Questions 
Based on the research objectives above, this study seeks to answer the following research 
questions: 
i. How does extreme events affect the patronage of insurance in Ghana? 
ii. What are the effects of extreme events on the supply of the various classes of 
insurance business? 
iii. What are the impacts of extreme events on the profitability of the Ghanaian insurance 
firms? 
 
1.6 Significance of Study 
The significance of the study can be seen along the following lines: research, practice and 
policy.  In research, academic work on extreme events and the Ghanaian insurance industry 
hardly exists. The literature on extreme events focuses on mainly developed countries and 
hence this study will enable insurance companies in developing countries to examine the 
threats of the risks of extreme events they are exposed to in order to better manage them. 
For practice, the study will help management of insurance companies to formulate strategies 
to mitigate against risks posed by extreme events so that they can be adopted for their 
efficiency in financial operations. It will provide insight to management and other 
stakeholders on how the industry has been damaged and help inform their future decision 
making in terms of response and recovery. 
For policy significance, the study will provide feedback on policies that can protect both 
shareholders capital and society from risks posed by these extreme events. This research will 
help policymakers enhance their understanding on the behavior and reaction of customers to 
 6 
 
University of Ghana http://ugspace.ug.edu.gh
losses which will enable them implement better policies to increase insurance coverage and 
overall protection for these events prior to their occurrence. 
 
1.7 Scope and Limitation of Study 
The study will focus on extreme events that have occurred in Ghana within the past ten years. 
This study shall be limited by time since one academic year will not be enough to fully 
explore every angle of this title as the researcher would so desire. 
Also, due to cost factors, the sample may be biased since the insurance companies chosen 
may be due to proximity. There exists also the constraint of unavailability of existing 
literature on the subject in Ghana. There is also the possibility that some of the insurance 
companies would withhold some information since the information required for the study is 
quite sensitive.   
The researcher shall also limit this study to only insurance companies, even though the study 
could have covered the entire insurance industry.  
 
1.8 Thesis Structure 
The rest of the study is organized as follows; Chapter two presents the review of relevant 
literature on the nature and risk of extreme events as well as their impact on the Ghanaian 
insurance market. Chapter three focuses on the methodological approach for addressing the 
research questions, the sampling techniques and sampling size, data collection instrument and 
method, data processing and mode of analysis of variables. The fourth chapter includes the 
data presentation, analysis and discussion of findings. Finally, chapter five contains the 
summary of the study, conclusion and recommendations for further studies. 
 7 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER TWO 
                LITERATURE REVIEW 
2.1 Introduction 
This chapter reviews the concepts of extreme events, the risk associated with extreme events 
and their impacts in various economies. The role of insurance in managing extreme risk and 
economic growth are also discussed in this chapter. The section provides a summary of 
relevant literature and states out other methodologies and framework that were used by other 
studies in assessing extreme risk and its impact on insurance demand and supply as well as 
the profitability of insurers. 
 
2.2 The Concept of Extreme Events 
2.2.1 Nature of Extreme Events 
Extreme events, also known as catastrophes, are events characterized by a low frequency but 
high severity. When such an event occurs, it deviates shortly from normality affecting a large 
number of people (Ciumas & Coca, 2015; Siri, Newell, Proust, & Capon, 2016). According 
to WMO (2014) extreme events can be defined as events which are rare at a particular place 
or events with high impacts. Not all extreme events will have impacts on the human system 
since the human impacts will depend on the exposure, vulnerability and the extent of 
adaptation of the society to the particular extreme events (Goodess, 2013).  
Extreme events can also lead to extreme conditions or impacts, either by crossing a critical 
threshold in a social, ecological, or physical system, or by occurring simultaneously with 
other events. The threshold for defining an extreme event vary from location to another. This 
is because what is considered extreme in one location can be considered as being within the 
normal range in another locality (TT-DEWCE, 2016). According to Stephenson (2008) 
 8 
 
University of Ghana http://ugspace.ug.edu.gh
extreme events do not arise spontaneously but rather they evolve continuously from less 
extreme events and they stop evolving to become even more extreme events. This is known 
as the evolutionary principle. Extreme events are multidimensional in nature and as such have 
a variety of attributes such as the rate of occurrence, intensity or magnitude, temporal 
duration and timing, and multivariate dependencies. This makes it difficult to completely 
describe extreme events by a single number or a unique definition (Stephenson, 2008). 
 Although some current events have resulted in huge loss of life, better preparedness, 
construction, and emergency response have reduced losses from certain extreme events, 
while consequences from other event categories such as annual earthquake casualties have 
remained stable (Siri, Newell, Proust, & Capon, 2016). The degree of exposure of economies 
depends on the geographical location, natural resource capital, seasonality and exposure of 
population, infrastructure, political and institutional mechanisms, and coping and adaptive 
capacity (Mirza, 2003).  
2.2.2 Categories of Extreme Events 
Extreme events (both natural and man-made) can be categorized into meteorological, 
climatological, geological, epidemiological, and anthropogenic events. 
2.2.2.1 Meteorological Events 
Extreme events that are categorized as meteorological include extreme temperatures (heat 
and cold waves), tornados and flooding from extreme rainfall or storm surges.  Heat waves, 
according to the American Meteorological Society (AMS) Glossary of Meteorology, is 
defined as a period of abnormally and uncomfortably hot and high humid weather. Heat wave 
is also defined by WMO as a period of few days to several weeks of warm air due to high 
temperatures. Persistent exposure to high temperatures can cause heat cramps, heat 
exhaustion, and heat stroke (Molloy & Mihaltcheva, 2013). Cold waves also occurs when 
 9 
 
University of Ghana http://ugspace.ug.edu.gh
there is a sudden fall in temperature within a short period leading to the cooling of the air and 
accompanied by hazardous weather, like frost and snow (TT-DEWCE, 2016). A flood on the 
other hand, is an overflow of a huge amount of water onto land that is normally dry. Heavy 
rains, fast melting of snow, or the breaking of dams can lead to the occurrence of floods. 
Globally, floods are the most common natural hazard. They cause the third highest damages 
after storms and earthquakes.  Over 90% of the total economic loss globally between 1970 
and 2004 was caused by storms and floods (Kunreuther & Michel‐Kerjan, 2009). Floods have 
led to water-borne diseases, mental health problems such as anxiety as well as death from 
drowning. The elderly, children and women are the most vulnerable to weather-related events 
(Molloy & Mihaltcheva, 2013). 
2.2.2.2 Climatological Events 
According to Khandekar (2013) extreme climatic events are a central part of the earth’s 
climate system and such events are generated by large-scale atmosphere-ocean circulation 
systems and their complex interaction with local and regional weather patterns. Exposure of 
human and natural systems increases the impacts of climatic extreme events (IPCC, 2013). 
Climatological events includes droughts, famines and naturally occurring fires. Drought is the 
absence or shortage of water due to abnormally low rainfall and dry weather over a prolonged 
period. Drought is difficult to detect initially since it happens slowly and steady unlike other 
extreme events, such as floods, and hurricanes, which are easily identified when they occur 
(TT-DEWCE, 2016). Mostly, droughts results from scarcity of water due to rainfall. It is 
mostly seen in areas where rainfall is unpredictable, and differs significantly in its seasonal 
distribution (Mohammed & Rahman, 1998). According to Molloy and Mihaltcheva (2013), 
droughts causes more than half of all deaths as a result of natural hazards. Drought conditions 
reduce water availability and destroy agricultural crops (and hence reduced food availability 
leading to under nutrition). Wildfires which are mostly caused by high temperatures and 
 10 
 
University of Ghana http://ugspace.ug.edu.gh
droughts are worsened by climate change. Naturally occurring fires can lead to heat stress, 
injuries such as accidents, burns, and death. 
2.2.2.3 Geological Events 
Geological events arise as a result of a change of the earth either on or below its surface. 
Extreme geological events have led to the death of huge number of people and severely 
affected economies in the past several thousand years (Korup & Clague, 2009). Geological 
events includes earthquakes, tsunamis and volcanoes. An earthquake is the trembling or 
shaking of the earth's surface. It is caused by a sudden release of stress along a fracture in the 
crust of the earth which causes the crust to shift. The greatest number of deaths and 
infrastructural damage has been caused by earthquakes. In the past 4000 years, earthquakes 
have caused an estimated minimum of 11 million fatalities (Sena & Michael, 2006; Korup & 
Clague, 2009). Volcanic eruptions occurs when lava and gas are discharged from a volcanic 
vent. Volcanoes cause population movements as large number of people are forced to flee the 
moving lava flow (International Federation of Red Cross, 2017). Geological events (except 
volcanoes) are normally not predictive, unlike weather events where modern weather tracking 
gives many days warning (Sanders, 2005). 
2.2.2.4 Epidemiological Events 
An epidemiological event is an event which lead to a sudden severe outbreak of a disease. 
Epidemiological events could be biological or occur due to other events such as floods, 
earthquakes and droughts (Adivar & Selen, 2013). Epidemiological events cause the 
occurrence of more cases of disease than would be expected in a community or region during 
a given period (Sena & Michael, 2006). One cause of epidemics is population movement. 
During and after catastrophes such as floods, earthquakes and volcanoes, people may be 
forced to move to safer places. These people may lack access to good food, water and health 
 11 
 
University of Ghana http://ugspace.ug.edu.gh
facilities and could lead to poor nutrition which will weaken their immune system and make 
them vulnerable to diseases such as malaria and diarrhea. Also, an outbreak of disease may 
occur when large numbers of dead bodies are present in an area which has affected by an 
extreme event (Watson, Gayer, & Connolly, 2007). Epidemiological events includes 
infectious disease epidemics such as influenza, tuberculosis and Hepatitis B; and allergen-
associated events (Siri, Newell, Proust, & Capon, 2016). Children are particularly vulnerable 
to epidemics because they have a larger skin surface-to-body mass ratio, depend on others for 
their care and have partially developed immune system. 
2.2.2.5 Anthropogenic Events 
Anthropogenic can be defined as any activity that is caused or influenced by humans. Hence, 
anthropogenic events are extreme events caused by human actions or negligence. 
Anthropogenic events are mostly unpredictable, can spread across geographical areas, may be 
unpreventable and may have limited physical damage but long-term effect (Sena & Michael, 
2006). These events can be broadly classified into technological events such as chemical and 
nuclear disasters; anthropogenic fires and extreme pollution events; and sociological events 
such as war and terrorism (Jha, 2010).  
The illegal threat or use of violence to promote political ambitions is known as terrorism 
(Cunningham, 2003). Terrorism causes fear, panic, death and can disrupt the psychological 
and economic growth of the society (Richards, Burstein, Waeckerle, & Hutson, 1999). 
Nuclear disaster events is mainly caused by nuclear power plants. Nuclear power plants use 
the heat generated from nuclear fission in a contained environment to convert water to steam, 
which powers generators to produce electricity. When an accident occurs at a nuclear power 
plant, the body may be exposed to radiation from the cloud. Radioactive materials may also 
be inhaled and digested by the body due the particles being deposited on the ground. This can 
cause serious illness or death. Anthropogenic fires and pollution may be triggered as 
 12 
 
University of Ghana http://ugspace.ug.edu.gh
secondary events after the occurrence of certain meteorological and geological events such as 
earthquakes and floods (Sena & Michael, 2006).  
 
2.3 Extreme Value Theory  
Extreme Value Theory (EVT) looks at how extreme values behave randomly in a process. 
Extreme events in real data can be generally identified in two ways. The first way is to 
observe the maximum or highest number taken by a variable in successive periods. The 
collection of these observations make up extreme values. This is also known as the block 
maxima method. The second approach, also known as the threshold method, looks at how the 
observations exceeds a given threshold. All observations that exceeds the threshold constitute 
extreme events (Gilli & Kellezi, 2006). Data with seasonality, such as hydrological data, is 
usually analysed with the block maxima method. However, recent applications of extreme 
theory prefer the threshold method because of its efficient use of data (Gilli & Kellezi, 2006). 
Fisher and Tippett (1928) suggested Gumbel, Frechet and negative Weibull distributions to 
model the maxima of a random variable in a single process.  Later, a generalized extreme 
value distribution developed by Jenkinson (1955) combined all three distributions into one. 
2.3.1 Applications of Extreme Value Theory 
Extreme Value Theory has been applied in areas such as modern science, finance, insurance 
and engineering. In engineering, high magnitudes of wind was estimated using EVT in order 
to design bridges and buildings which could withstand the extreme winds (Coles, 2001). 
Extreme Value Theory (EVT) was also applied in public health to predict future extreme 
epidemiological events such as influenza and Pneumonia. In Taiwan, Nadarajah and Shiau 
(2005) applied extreme value theory to flood data to investigate extreme floods and also 
determine whether the volume and peak of floods vary with time and duration. 
 13 
 
University of Ghana http://ugspace.ug.edu.gh
In insurance and other financial institutions, EVT has been used to predict the occurrence of 
catastrophic losses such as large credit default (Embrechts, Resnick, & Samarodnitsky, 
1999). Again, the theory was used to predict the occurrence of drought, heat waves, heavy 
rainfall and other extreme weather events. The theory was also employed to estimate the 
magnitude of long return period earthquakes (Al-Abbasi & Fahmi, 1985). 
 
2.4 Risk of Extreme Events 
According to Gencer (2013), risk is the possibility that something bad or harmful will happen 
due to interactions between natural or human-induced hazards and vulnerable conditions. It is 
also the probability and degree of an adverse outcome such as a loss or danger. Risk 
comprises of hazard and vulnerability of people and structures. A hazard is a specific 
situation, activity or condition that increases the probability of the occurrence of loss arising 
from a peril whilst vulnerability can be explained as the degree of exposure of a community 
to the influence of hazards and the ability (or lack thereof) of the exposure unit to cope, 
recover, or adapt (Linnerooth-Bayer, Mace, & Verheyen, 2003; Gencer, 2013).  
An interaction between the external risk factor and the internal risk factor results in extreme 
risk or vulnerability. The risk and damages of extreme events depend heavily on 
socioeconomic factors, and also the number and severity of extreme events (Molloy & 
Mihaltcheva, 2013; Linnerooth-Bayer, Mace, & Verheyen, 2003). IPCC in their report has 
confirmed that severe events together with basic socio-economic trends (such as population 
growth and unplanned urbanization) are capable of stressing economies and businesses by 
declining the value of business assets and diminishing investment viability (UNEP FI, 2002).  
 14 
 
University of Ghana http://ugspace.ug.edu.gh
2.4.1 The Frequency and Intensity of Extreme Events 
Every year, extreme events cause both direct and indirect losses hindering both economic and 
social development (WMO, 2014). Losses experienced can be in the form of damages to 
assets, loss of jobs, business interruptions and decline in revenues from taxes. About 11000 
extreme weather events occurred between 1996 and 2015 leading to the death of more than 
528 000 people and economic loss of US$3.08 billion worldwide. The severity and frequency 
of these events are projected to be two-to-three times higher by 2030 and four-to-five times 
higher by 2050 (Kreft, Eckstein, & Melchior, 2016). Windstorms, floods, droughts, and other 
climate-related hazards contributes to about three-quarters of all recent economic losses as 
stated by the United Nations International Strategy for Disaster Reduction (ISDR). Frequency 
provides information about the increasing nature of extreme events. Risks from extreme 
events are difficult to manage since there are not sufficient empirical information and they are 
not accurately predictable. Intensity or severity on the other hand, is the measure of the extent 
and the magnitude of the loss after an event. Severity depends on the characteristics of an 
event as well as the exposure and vulnerability of the systems it impacts (TT-DEWCE, 2016; 
Sakshi, 2009; Stephenson, 2008)). Globally, the number of extreme events has increased 
from about 400 per year to about 1,000 in the last 31 years supporting the fact that extreme 
events and total losses are rising faster than population, or economic growth (WMO, 2014). 
 In Africa, the number and intensity of droughts and floods have increased over the past 30 
years. Ly, Traore, Alhassane, & Sarr (2013) estimated that there will be great evaporation in 
Africa as a result of increased temperatures since saturation vapor pressure increases 
exponentially with temperature according to the Claudius–Clapeyron relationship. This may 
affect the rainfall pattern and increase the risk of flooding. Hence, it was found that the 
number of flood events have increased on average from less than 2 per year before 1990 to 
more than 8 or 12 on average per year during the 2000s.  
 15 
 
University of Ghana http://ugspace.ug.edu.gh
2.4.2 Influence of Socio-economic Factors on Extreme Events Losses 
The observed rising loss of extreme events are highly influenced by socioeconomic and 
demographic factors. Migration of populations to flood and fire-prone areas, increasing 
reliance on vulnerable electric power grids, and rising material wealth are part of the reason 
for the increasing vulnerability of economies to extreme events. According to Kunreuther and 
Michel-Kerjan (2007), the degree of urbanization is one of the main socio-economic factor 
that influences the loss trend of extreme events. In 2000, about 6 billion people which consist 
of 50% of the world’s population lived in urban areas and the United Nations has predicted 
an increase to about 8.3 billion people (60%) by the year 2025. Most of them will reside close 
to regions prone to extreme events (Courbage & Stahel, 2012). Increasing poverty and 
inequality, failures in governance, crowded living conditions and the location of residential 
areas close to hazardous industry or in places exposed to natural hazard increases the risk of 
urban areas to extreme events (GFDRR, 2010). Extreme events are expected to destroy a 
greater number of homes and commercial buildings today than if a similar event had occurred 
in the 1990s. This means that the intensity of exposure is increased as result of urbanization 
and population growth. (Gencer, 2013). 
Most cities in least developed countries are unable to expand properly and accommodate the 
rapid population growth in urban areas. There are many reasons that makes urban areas in 
most developing countries vulnerable to extreme events. Firstly, the population density, rapid 
industrialization and accumulation of hazardous materials in heavily populated areas makes 
urban areas particularly those in developing countries at risk to extreme events. A given event 
will affect larger populations in urban areas because the people are clustered in a limited area 
(Siri, Newell, Proust, & Capon, 2016). Secondly, ecological imbalance such as shortage of 
appropriate drainage systems, resident buildings on waterways and insufficient planning have 
made some urban areas vulnerable to floods. In others, deforestation has led to hillside 
 16 
 
University of Ghana http://ugspace.ug.edu.gh
erosion, making people vulnerable to landslides triggered by heavy rains. Thirdly, poor 
design and construction of buildings is a major challenge faced by urban areas. Architects, 
engineers, and skilled workmen in most of developing countries have limited knowledge in 
the design and construction of earthquake-resistant structures of buildings. The collapse of 
many buildings under construction is mainly caused by poor construction and weak buildings 
(Allotey, Arku, & Amponsah, 2010).  
Other demographic factors such as age, income and sex also contributes to this trend. 
Children and older people are highly exposed to risks of extreme events. Children have small 
body size that makes them vulnerable to drowning in floods and prone to develop 
hypothermia (lowered body temperature) during heat waves. Their bodies are weak and they 
have underdeveloped immune system which increases their inability to fight infectious 
diseases which are brought by these events. For many elderly people, their physical 
movement are mostly restricted and they may rely on others to move them around. These and 
other conditions may influence their exposure to indirect effects of stress and shock after 
these events (Molloy & Mihaltcheva, 2013). 
 
2.5 Impact of Extreme Events on Economies 
A population’s capacity to respond to and recover from an extreme event depends on the 
losses incurred as well as the global environment of affected location (Prelipcean & 
Boscoianu, 2010). Extreme events have and continue to affect people in every continent. 
Globally, more than two million people lost their lives between 1980 and 2012 due to 
weather, climatological and geophysical events. Over the past 55 years, about 93 percent of 
all catastrophes is as a result of weather-related events (Mills, Roth, & Lecomte, 2005). The 
highest number of deaths (94%) between 1970 and 2012 in Europe was as a consequence of 
 17 
 
University of Ghana http://ugspace.ug.edu.gh
extreme temperatures. Hydro meteorological and climatic events caused the greatest loss in 
terms of lives and economic damage in North and Central America. Hurricane Mitch in 1998 
affected Honduras and Nicaragua and caused 17 932 deaths, Hurricane Fifi in 1974 affected 
Honduras which led to 8 000 deaths and Hurricane Katrina in 2005 resulted in US$ 146.9 
billion in losses. Two (2) million homes, 40 oil rigs and 150 offshore oil platforms were 
destroyed by hurricanes Katrina, Rita and Wilma in 2005 (Sturm & Oh, 2010; WMO, 2014). 
The exposure of poorer developing countries to extreme event risks is high, though the 
absolute monetary losses are much higher in richer countries. Also, financial hardship and 
loss of life after an extreme event is far more prevalent particularly in developing countries 
(Kreft, Eckstein, & Melchior, 2016; Molloy & Mihaltcheva, 2013). According to Hallegatte, 
Dumas, and Hourcade (2010), extreme events affects the developing world in two folds. 
Firstly, a rise in the number and severity of extreme events will put pressure on the budgets of 
developed countries hence less resources will be available to assist other developing 
countries. Secondly, the governments of developing countries will be forced to use available 
resources to deal with the losses of extreme events rather than productive and growth-
enhancing projects. As the number of extreme events increases than the capacity of 
developing countries to rebuild decreases and could cause them to remain in a constant state 
of reconstruction.  
From 1970–2012, there have been loss of 915 389 lives and economic damages of US$ 789.8 
billion as a result of extreme events in Asia. The highest number of death (76%) was as a 
result of storms, while the greatest economic loss (60%) was due to floods (WMO, 2014). 
Africa is susceptible to all kinds of intense and frequent natural and man-made hazards. A 
study by Mohammed and Rahman (1998) reveals that the regions most affected are the 
African Horn, West and South Africa. Mozambique, Ethiopia, Ghana, Chad and Sudan were 
seen to be the most hazard-prone countries since they recorded the highest losses. Lack of 
 18 
 
University of Ghana http://ugspace.ug.edu.gh
disaster-resistant infrastructure, high dependence on agriculture, inferior disaster 
preparedness and recovery, and other factors make developing countries much more 
vulnerable to extreme events. Ly, Traore, Alhassane and Sarr (2013) confirmed this by 
examining the evolution of climate extremes in the West African Sahel. They observed a 
general warming trend since the 1960s which means a higher demand on domestic energy 
consumption for cooling, a higher evaporation rate from water bodies and irrigated crops, and 
a lower performance of agricultural crops and livestock. Also, increased occurrence of 
extreme rainfall events such as heavy downpours leads to more tenuous infrastructure and 
production systems.  
Thus, the threat posed by extreme events is more serious in developing countries than they 
appear since most huge social and human costs of extreme events such as loss of human 
lives, cultural heritage, and ecosystem services are not accounted for in the macroeconomic 
data. The reason being that such losses are difficult to value and monetize (Mirza, 2003). 
 
2.6 Extreme Risk Management 
Extreme Risk Management (ERM) is a dynamic process that requires continuous 
modifications, specialized decision making and interaction at different levels (Prelipcean & 
Boscoianu, 2010). Extreme risk management involves a number of stages and these are: the 
preparedness to mitigate extreme risk effects; the response to deal with the physical impacts 
and the new risk factors created; short term recovery activities to restore vital support systems 
and long term tasks to bring economic framework back; and prevention and mitigation to 
minimize the likelihood and effect of future events (Prelipcean & Boscoianu, 2010). 
ERM is an integral part of country risk management and as such World Bank Group plays an 
active role in assisting countries with building effective risk management systems as these 
 19 
 
University of Ghana http://ugspace.ug.edu.gh
events have a disproportionately adverse impact on the poor (Sakshi, 2009). The soar in the 
risks associated with extreme events threatens livelihoods, human lives and health. Moreover, 
socio-economic growth and development is severely weaken (Golnaraghi, Surminski, & 
Schanz, 2016). As the occurrence and severity of extreme events increases, management of 
risk from extreme events will become increasingly important for society. Extreme risk 
management will increase the safety and resistance of economies and communities, reduce 
economic, social and environmental losses, and also reduce the funds used up to finance 
losses associated with extreme events. This will promote investment and the affordability of 
insurance to the society (UNEP FI, 2014).  
Managing extreme events involves risk identification and assessment, mitigation and 
adaptation activities, and the transfer of the risks that cannot be eliminated or reduced 
through risk-sharing mechanisms (Courbage & Stahel, 2012). Society can cope and manage 
risk from extreme events through the collaborating efforts of insurance, customers as well as 
the government bodies. The scientific community and the government can support the 
insurance industry by providing technological innovations and tools, raising awareness 
through public education and regulations as well as investing in preparedness and 
management of extreme events (Golnaraghi, Surminski, & Schanz, 2016).  
The susceptibility of critical infrastructure (e.g. energy, food and agriculture, water, 
transportation and health) to extreme events has become a crucial concern of many leaders of 
the society. This is because the destruction or interruptions of these infrastructure could cause 
major harm to the welfare of the population due to the direct and indirect economic impacts 
of extreme events (Golnaraghi, Surminski, & Schanz, 2016). The government plays a key 
role as an insurer of last resort and risk manager for many types of climate, weather and other 
event losses. Systems and frameworks are introduced by governments on various state levels 
to deal with the risk of extreme events and strengthen resilience. For example specialized 
 20 
 
University of Ghana http://ugspace.ug.edu.gh
agencies such as the African Risk Capacity Project and the African Adaptation Initiative have 
been set in Africa to strengthen resilience to extreme events, to lessen the risk fatalities, and 
to support member states through capacity building (Kreft, Eckstein, & Melchior, 2016).  
2.6.1 Insurance as a Tool for Risk Management  
Following a damaging episode of extreme event, societies turn to insurance companies to 
help them rebuild because the insurance industry helps protect society, fosters innovation and 
supports economic development through risk prevention, risk reduction and by sharing risks 
over many shoulders. When individuals purchase insurance, it shows that their common 
interest is to reduce their own loss and others who have suffered similar losses (UNEP FI, 
2014; Sturm & Oh, 2010).  The insurance industry’s primary business is to understand, 
manage and carry various type of risks such as property risks, disability and natural disasters. 
Individuals and firms protect themselves against infrequent but extreme losses by paying a 
fee (premium) which is far lower than the financial loss.  Using the law of large numbers, 
insurance companies are able to accurately predict the risks they carry and also estimate the 
right premium. (Lester, 2009).  
Life and non-life or general contracts are the two main contracts in insurance. Material and 
financial risks are insured under non-life insurance contracts which are normally renewed 
annually. Life insurance contracts covers the risks associated with the life of an individual. It 
can be for short periods (for example, accidental death) or very long periods (for example, 
whole of life) (Guerineau & Sawadogo, 2015).  
The sharing and transfer of risk through insurance can assist the recovery to catastrophes and 
improve the resilience of society to extreme events. This is becoming more significant as the 
frequency and losses from these events have increased over the last few decades (Schuster, 
2013). For insurance to perform its role as an instrument for the management of risk from 
 21 
 
University of Ghana http://ugspace.ug.edu.gh
extreme events, insurers should price premiums so that they accurately reflect risk which will 
signal how safe or exposed policyholders are, and also ensure equity and affordability of 
premiums especially for those in hazard prone areas (Kunreuther, 2017). 
2.6.2 Insurance and Economic Growth 
Insurance promotes economic growth by improving the conditions for investment and 
promoting a more proficient blend of activities. Insurance offers protection against uncertain 
but severe losses by providing payments periodically to people affected. This income 
smoothing effect helps to avoid excessive and costly bankruptcies and facilitates lending to 
businesses. Also, insurance accessibility promotes higher productivity and growth by 
assisting risk averse individuals and entrepreneurs to undertake activities which are risky but 
has higher returns (Brainard, 2008). The investment behaviour of smallholder farmers and 
crop insurance was studied by Karlan, Osei, Osei-Akoto and Udry (2014) in Ghana. The 
results revealed that, there was a significant growth in the investment on the farms of 
smallholder farmers who have crop insurance than those who do not. On the investment side, 
life insurers with their sizeable reserves and predictable premiums can assist in the role of 
providing capital to infrastructure and other long term investments as well as competent 
oversight to these investments (Brainard, 2008). 
According to Warner, Yuzva, Zissener, Gille, Voss, and Wanczeck (2013), a study conducted 
by the World Bank finds that after a large weather-related catastrophic event, countries with 
high insurance penetration experience a growth in Gross Domestic Product (GDP) whilst low 
insurance penetration countries suffer from a negative GDP deviation. Moreover, low 
insurance penetration countries may need other growth factors to boost GDP growth 
otherwise their GDP could continue to dwindle for a long period. 
 
 22 
 
University of Ghana http://ugspace.ug.edu.gh
2.6.3 The Insurance Market 
The insurance market is a very competitive industry which is mostly controlled by supply and 
demand (Lester, 2009). The insurance market provides insurance coverage to individuals and 
business firms who are vulnerable to natural and manmade catastrophes as well as other risks 
(Cummins & Mahul, 2009). 
According to Brainard (2008), rising incomes, macroeconomic stability, and financial 
deepening are the key drivers of insurance market growth. In the paper, Brainard posited that 
as individuals become richer, the demand for insurance coverage to protect their wealth also 
increases. Again, as the economy expands, it becomes more economical to provide insurance 
and therefore less operational cost leads to higher returns. Insurance market development is 
also determined by the level of growth of related markets, including sales of cars and other 
consumer durables, business establishments, residential and commercial mortgage markets, 
disposable income, and commercial and trade transactions. 
 There is massive expansion of the insurance industry in the developing world because of 
rising economies like China and India. The economies in Sub- Saharan Africa have been 
constantly ranked among the world’s fastest growing economies since 2010. As such, the 
World Bank predicted a rise from 4.6% to 5.1% regional GDP growth by 2017 (EY, 2016).  
 
 
 
 
 
 
 23 
 
University of Ghana http://ugspace.ug.edu.gh
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 2.1: Real GDP Growth, 31 Largest African Economies (2010-2019, Compound 
Annual Growth Rates, in %) 
 
 
 
 24 
 
University of Ghana http://ugspace.ug.edu.gh
The growth in GDP may lead to rising incomes and growing affluence which will increase 
the number of people who can afford insurance as consumer spending on items, such as cars, 
smart mobile phones and health care expands (EY, 2016; Schanz, Alms, & Company, 2016).   
An effective and viable insurance system requires a risk that is quantifiable, distributed, and 
affordable. Again there should be an insurable population aware of risk they are exposed to, 
willing to insure and can afford the necessary premiums. Finally, it requires a solvent insurer 
that is willing and able to pay claims; and has enough capital and reserve to cover any 
abnormally large losses (Lamond & Penning-Rowsell, 2014). An individual’s willingness to 
pay (WTP) for insurance may be influenced by the premium, his income, his level of risk 
aversion and risk perception (Ranger & Surminski, 2011). Arshad, Amjath-Babu, Kächele 
and Müller (2016) examined the drivers of the willingness to pay for crop insurance against 
extreme weather event in Pakistan. The study revealed that the WTP depends on the age, 
education, family size, land ownership, source of income, irrigation availability, exposure to 
previous extreme events, and access to extension and credit services. 
All in all, a well-functioning insurance markets should result in better pricing of risk, greater 
efficiency in the overall allocation of capital and mix of economic activities, and higher 
productivity (Brainard, 2008). 
 
2.7 Empirical Findings on the Impact of Extreme Events on the Insurance Market 
2.7.1 Insurance Demand and Extreme Events 
Demand shows the quantity of a given product consumers are willing and able to buy at a 
given price.  Individuals and firms may purchase insurance for various reasons. One reason 
may be that the law requires them to do so. For example, in Ghana, the third-party motor 
insurance policy, workers’ compensation insurance as well as insurance against the hazards 
 25 
 
University of Ghana http://ugspace.ug.edu.gh
of collapse, fire, earthquake, storm and flood in respect of all commercial buildings, both 
complete and those still under construction are compulsory. Also, workers’ compensation 
insurance is required by law in all states of America (World Bank, 2013). Another reason 
could be that insurance is used to finance risks incurred by firms to protect them from 
bankruptcy. Hence, the insurance company bears the risk exposed to the firm (Michel-Kerjan, 
Raschky, & Kunreuther, 2015).  
Insurance demand is influenced by several factors which have been revealed in many studies. 
These factors include political, psychological, economic (income, inflation and interest rate) 
and social (age and sex) issues (Ciumas & Coca, 2015). Other external factors influencing the 
demand for insurance include climate change, globalization, variations to financial market 
regulation, and global population growth (Ranger & Surminski, 2011). Using a panel model 
on a sample data between 2000 and 2012, Yuan and Jiang (2015) examined the determinant 
of insurance demand for life, nonlife and both life and non-life in China. The results pointed 
out that level of education, children and elderly dependency ratio, and development of social 
security pension, influenced life insurance demand whilst inflation influenced nonlife 
insurance demand. Both life and nonlife insurance demand was affected by the level of 
income, insurance market development and degree of marketization.  
In Ghana, a study was done by Peprah, Koomson and Forson (2017) using a binary logit 
estimation on the GLSS6 data. Their finding showed that the patronage of insurance depends 
on poverty and employment status. It was observed that poor people have a lower probability 
of purchasing insurance, with locational influence being more evident for the poor in the rural 
areas. Moreover, regular income workers in the formal sector demand insurance than self-
employed. Furthermore, those who reside in small cities and rural areas have a higher 
demand for insurance than people in urban areas. Fofie (2016) adds to this literature by 
exploring and assessing the Social, Economic and Demographic (SED) factors that are likely 
 26 
 
University of Ghana http://ugspace.ug.edu.gh
to influence the patronage of insurance in Ghana using a probit econometric model. They 
found that SED characteristics is a major factor affecting insurance demand in Ghana. It was 
observed that these SED variables influences the factors that prevent people from patronizing 
insurance policies. One major cause for the low patronage of insurance was lack of 
knowledge on insurance due to lack of education and public awareness. 
In general, the risk perceptions of individuals may change after the occurrence of extreme 
events and hence demand for insurance and mitigation mechanism may increase if 
individuals learn from these events. Ranger and Surminski (2011) suggest that as the risks 
from extreme events keep rising, insurance demand could be increased in places where losses 
were more recurrent. Several studies have been done on the demand for insurance after an 
extreme event. Michel-Kerjan and Kousky (2010) who analyzed the US National Flood 
Insurance Program portfolio, found that people opted for higher coverage and premium 
payments after major foods in Florida in 2004.  
Individuals who view insurance as an investment from which they expect a return in the form 
of claims payments following an event rather than a protective measure should one suffer a 
loss, are likely to ignore potential disasters and overreact to recent ones (Kunreuther, 2017). 
This could cause an increase in the purchase of insurance only after the occurrence of 
extreme events and then demand for insurance would be seen as more reactive than proactive. 
A study by Dumm, Eckles, Nyce and Volkman-Wise (2017) revealed a rise in insurance 
demand and supply after a loss however, this demand decreases as more time elapses from 
when the loss occurred. This is consistent with the theory that individuals normally tend to 
amplify their sense of loss they are exposed to, only after a disaster occurs. Gallagher (2012) 
shared this view by finding that US flood insurance demand significantly increased 
immediately in areas affected by the flood but falls shortly afterwards. In Ghana, Ackah and 
 27 
 
University of Ghana http://ugspace.ug.edu.gh
Owusu (2012) found in their study that most informal sector workers do not take-up 
insurance as a precautionary measure.  
 Dlugolecki and Hoekstra (2006) posited some barriers to demand for insurance against 
extreme events which include poor perception of risk exposure, high price of insurance 
(premiums) and lack of efficiency on the part of insurers in the payment of claims. Also, 
expectation of public and political responses as well as international aid reduces insurance 
demand (Lamond & Penning-Rowsell, 2014; Ranger & Surminski, 2011). 
 
2.7.2 Insurability of risk (Supply of Insurance) 
The affordability and accessibility of insurance are potential source of economic 
development, peace of mind and security in a society. 
In order to guarantee their financial survival in case of catastrophic losses, according to 
Linnerooth-Bayer, Mace and Verheyen (2003), two conditions must be met before insurers 
will be prepared to insure an uncertain event. These conditions are; 
i. The ability of the insurer to identify and evaluate the probability of the occurrence of 
the event and the degree of risk presented. This can be done with the aid of historical 
data or scientific analyses.  
ii. The ability of the insurer to set premiums for each proposed insured. This requires an 
understanding and assessment of each potential customer's risk and the coverage to 
offer. Afterwards, a premium is charged to reflect the risk exposure in order to make a 
reasonable profit.  
The first step to ensure insurability of risk is the identification of risk. To identify risk, the 
occurrence of specific events and the intensity of loss caused by these events are estimated 
(Courbage & Stahel, 2012). Insurance is mostly used to cover high-frequency, low-severity 
 28 
 
University of Ghana http://ugspace.ug.edu.gh
events, which are statistically independent of each other and have probability distributions 
that are reasonably stationary over time. For such events, the premiums and reserves required 
to meet insurers’ liabilities and ensure solvency can be accurately calculated. On the other 
hand, insurers hardly insure events such as catastrophes due to their high intensity but low 
frequency. This is because the historical data on such events may be insufficient or not 
available, making the assessment and modelling of the risk very challenging. Insurance 
markets poorly respond to huge catastrophes. They restrict the supply of insurance and raise 
the price of insurance as a way of responding to large or extreme events (Cummins, 2006; 
Courbage & Stahel, 2012). 
Once risks have been identified, insurers must be able to estimate the premiums to be paid by 
each policyholder to ensure coverage. However, due to the severity of losses from extreme 
events, premiums can increase more than policyholders can pay. According to Ranger and 
Surminski (2011), extreme events could make the availability and affordability of insurance 
very difficult as a result of increasingly correlated of losses within short intervals. Insurers 
may withdraw from certain areas and classes of business or, experience a higher rate of 
insolvency when faced with risks from extreme events.  
Insurability of risk is affected by both external and internal factors. Internal factors include 
the amount of capital, ambiguity of risk, correlated risk, adverse selection and moral hazard. 
The role of reinsurers, insurance commissioners, rating agencies, modeling firms and 
investors constitute the external factors (Golnaraghi, Surminski, & Schanz, 2016).  
The impact of extreme events on the insurability of risk have been examines by several 
researchers. Chen, Doerpinghaus, Lin and Yu (2008) examined the effects of 9/11 terrorist 
attack on the insurance industry in the U.S. and found that firm type, loss estimates, 
reinsurance use, and tax position influences premium increase after an extreme event whilst 
firm type, loss estimates, financial strength, underwriting risk, and reinsurance were key 
 29 
 
University of Ghana http://ugspace.ug.edu.gh
determinants of insurance supply reductions after extreme events. Mills, Roth and Lecomte 
(2005) also revealed in their studies that extreme events may increase insurance claims and 
costs translating into higher premiums and deductibles, lowered limits, and broader coverage 
restrictions. This is supported by Yi (2013) who evaluated the market competition, demand 
and supply of the insurance market in Texas using a sample data from 1995-2011. The results 
indicated that insurance companies adapt to excessive losses from catastrophic risk by raising 
their premiums whilst others go insolvent and leave the market.  
Maynard (2008) adds to this literature by asserting that increase in extreme events may 
worsen the level of losses which may make the price of insurance to be so high that 
policyholders may not be able to afford. Insurers may decide to maintain insurance coverage 
or remove cover totally from the market in order to maintain solvency and shareholder 
returns. For instance, after the Atlantic hurricane season in 2004 and 2005, reinsurance 
premiums rose dramatically which became economically infeasible for insurers in that region. 
Some insurers declined the underwriting of new policies in the state, dropped existing 
customers and left the market (Sturm & Oh, 2010). Other insurers such as Poe Financial 
Group, South Florida’s second-largest insurer of homes, condos and apartments, had to pay 
more than US$2 billion in claims after Hurricane Wilma and as such was declared insolvent 
in 2006 (Linnerooth-Baye & Mechler, 2009) 
Insurers are mostly faced with the challenge of pricing insurance and the amount of coverage 
to offer as they renew their policies annually. This is because of their inability to differentiate 
between systematic changes in climate and random weather patterns in the short run 
(Kunreuther & Michel-Kerjan, 2007). To ensure coverage, premiums must be calculated 
through actuarial processes and be affordable to consumers. However, this situation may be 
affected in the case of extreme events (Mills, Roth, & Lecomte, 2005).  
 30 
 
University of Ghana http://ugspace.ug.edu.gh
2.7.3 Profitability of insurance companies 
One important factor that determines insurers’ performance and healthiness is profitability 
(Dorofti & Jakubik, 2015). Insurance companies’ primary goal is to generate profits or worth 
for their shareholders even though insurance assist in linking the world through risk-sharing 
measures (Akotey, Sackey, Amoah, & Manso, 2013). There are numerous approaches to 
measure insurer’s profitability. Some of these methods include Return on Equity (ROE), 
Return on Investment (ROI) and Return on Assets (ROA). Others include technical 
profitability ratio and sales profitability ratio. ROA indicates how profitable a company is 
relative to its total assets and it measures the efficiency of management in using a company 
assets to generate profit. ROE shows how much profit the company generate from its 
shareholders equity (NIC, 2016). ROI is also an indicator which assesses a company's 
efficiency in allocating capital to profitable investments. This indicator shows how well a 
company uses its funds to generate higher returns. Technical profitability ratio assesses the 
effectiveness of the core insurance activities of the insurance company whilst sales 
profitability ratio associates net income before taxes obtained by the insurance company with 
gross written premium. ROA and ROE are normally used to measure profitability. This is 
because they are readily accessible, rely on public data and are calculated in accordance to 
strict, prudent accounting rules (Dorofti & Jakubik, 2015). 
Insurers’ profitability can be looked at from two perspectives according to Akotey, Sackey, 
Amoah and Manso (2013). Firstly, profitability can viewed from both internal and external 
factors. The internal factors comprises of the insurer’s-specific characteristic whilst the 
external factors is made up of both industry and macroeconomic variables. Secondly, 
insurers’ profitability can be perceived at the micro, meso and macro levels of the economy. 
The micro level measures the impact of variables such as size, capital, efficiency, age, and 
 31 
 
University of Ghana http://ugspace.ug.edu.gh
ownership structure on profitability. The meso and macro levels looks at the influence of 
support-institutions and macroeconomic factors, respectively, on profitability.  
Mwangi and Murigu (2015) examined the determinants of the profitability of non-life 
insurance companies in Kenya. Employing multiple linear regression for the period 2009-
2012, they found that leverage, equity capital and management competence index positively 
affected profitability whilst size and ownership structure negatively affected profitability. 
Alhassan, Addisson, and Asamoah (2015) examined the impact of the regulatory-driven 
market structure on firm pricing behaviour on insurance markets in Ghana. They identified 
underwriting risk, leverage and inflation as major factors influencing the profitability of both 
life and non-life markets.  
An insurer’s capital enables him to pay for greater-than-expected losses thus, the capital 
required is proportional to the risk accepted. The more risk an insurer wishes to take, the 
more capital he needs to support the additional coverage it offers against that risk (Dlugolecki 
& Hoekstra, 2006). A study by Darkwah, Asare-Kumi, Nortey, and Baidoo (2016) reveals 
that working capital management is the cause of profitability of Ghanaian insurance 
companies. Using a sample of 10 insurance companies listed on the Ghana Stock Exchange 
between 2008 and 2014, they found out that Cash Conversion Cycle, Debt Ratio, Current 
Ratio, Sales Growth Rate and Accounts Collection Period are the main factors that affect 
working capital management which ultimately affects the profitability of these companies.  
Using cross-country aggregate data of 30 European countries over eight years (2005-2012), 
Dorofti and Jakubik (2015) explored the relationship between the macroeconomic 
environment and insurers’ profitability. Their results revealed that low interest rates (both 
nominal and real) negatively affect insurance profitability through lower investment income. 
Moreover, high inflation, low economic growth and poor equity market performance 
negatively influenced the performance of insures. 
 32 
 
University of Ghana http://ugspace.ug.edu.gh
Extreme events, such as natural disasters, present various challenges to insurers because of 
their uncertainty and associated great losses. Some of the challenges posed by extreme events 
include unexpected changes in claims pattern coupled with difficulty in adjusting premiums 
and reserve to maintain profitability (Mills, Roth, & Lecomte, 2005). Insurance companies 
keep reserves, capital and financial assets to pay claims as and when they occur. Insurers 
invest part of their capital in equities and property, in many cases. These assets may be 
negatively affected by extreme events causing them to fall in value. Large and unexpected 
losses lead to huge claim payments which reduce insurers’ and reinsurers’ capital and 
reserves (Maynard, 2008).  Evaluating the financial performance of life insurers by using ten 
life insurance companies in Ghana between 2000 and 2010, Akotey et al. (2013) revealed that 
life insurers have been incurring underwriting losses due to overtrading, high claims 
payments and high managerial expenses and that have negatively affected their financial 
performance. This means that the increase in extreme events will increase claims which will 
lead to more underwriting losses and consequently affect the financial profitability of firms in 
the insurance industry in coming decades. 
According to Cummins (2006) keeping capital and reserves to finance losses arising from 
extreme events such as floods and earthquakes is absurd and inefficient on the part of the 
insurance industry. This is because extreme events can seriously affect the insurance industry 
by affecting the profitability and overall performance of insurers and also affecting the ability 
of the industry to cover future losses. Furthermore, this effect will be higher in developing 
countries since developed economies have a stronger insurance market than developing 
countries. 
 Benali and Feki (2017) investigated how natural disasters affect the profitability of 
Property/Casualty insurers in the United States (U.S) for the period between 2008 and 2012 
using a panel data analysis. Their findings were that loss ratio, unexpected number and losses 
 33 
 
University of Ghana http://ugspace.ug.edu.gh
of disasters negatively affected profitability of insurers’ while premium to surplus ratio and 
volume of capital positively influenced profitability.  
However, according to Ranger and Surminski (2011), growing levels of extreme events risk 
could decrease the willingness of an individual to pay due to the increase in the premium. 
Also, the willingness to pay could increase due to the increasing level of perceived risk. The 
effect on profitability therefore depends on the price sensitivity of both actual and potential 
policyholders. If policyholders are less sensitive to prices, the increase in premiums may fully 
reflect the increase in risk, so profitability will increase; but if they are more sensitive, or the 
market is especially competitive, the increase in premium may be greater than the increased 
risk would imply, leading to reduced profitability (CISL, 2016). 
 
2.8 Extreme Events and Insurance in Ghana 
2.8.1 Extreme Events in Ghana 
Ghana is situated in one of the world’s most complex climatic regions, affected by tropical 
storms, and the influence of Atlantic Ocean and the Sahel. The country stretches from a 
densely populated low-lying coastal zone, to the sparingly populated northern regions in the 
savannah zone (GFDRR, 2017). Ghana is ranked among the 10 fastest-growing national 
economies in annual World Bank reports (EY, 2016). There has been a substantial population 
growth and expansion in Accra (Ghana’s capital) in the last three decades. The average 
population growth in the city ranges between 5% and 8% with the natural growth rate 
accounting for 2.8-3.2% of growth, and rural-urban migration accounting for the rest 
(Allotey, Arku, & Amponsah, 2010). 
Ghana’s main exposure to extreme event is flood and epidemics. Others include pest 
infestations wildfires and drought, which normally occurs in the Northern part of the country. 
 34 
 
University of Ghana http://ugspace.ug.edu.gh
Ghana is also vulnerable to earthquakes with the Greater Accra region being the most 
dangerous (Allotey, Arku, & Amponsah, 2010; GFDRR, 2017). Flood is a regular problem in 
some parts of the country as well as tornadoes and tidal waves. Ghana has also experienced 
wildfire disasters and epidemics (World Bank, 2013).  
There has been a gradual and apparent temperature rise and a fall in rainfall in all agro-
ecological zones in the country between 1960 and 2000 according to the Ghana 
Meteorological Agency (Twum, 2014). In 1983 Ghana experience a drought which affected 
12.5 million people. In 2010, after a period of drought damaging the initial harvest in the 
White Volta River Basin, floods affected hundreds of thousands of people and destroyed 
many of their livelihoods. In June that same year, floods caused more than a 1,000 homes to 
be destroyed and 5,000 people to be displaced, with a total of 24 deaths and millions of 
dollars in property losses in Tema. These floods demonstrated how extreme events can 
reverse development investments (GFDRR, 2017). 
According to the Centre for Research on the Epidemiology of Disasters (CRED), an event is 
considered as a disaster if at least 10 people are reported dead or at least 100 people are 
affected. In Ghana, an event is considered extreme if it affects at least 2000 people or kill at 
least 50 people according to the National Disaster Management Organisation (NADMO). The 
table below shows some of the extreme events which have occurred in the country.  
Table 2.1:  Some Extreme Events in Ghana from 1900 to 2016 
Year Event  Affected Death 
1918 Pandemic (Influenza) 560000 100000 
Jun-39 Earthquake 4800 23 
 35 
 
University of Ghana http://ugspace.ug.edu.gh
1983 Drought 12500000 150 
1984 Epidemic 5670 103 
1991 Flood 200000 0 
1995 Flood 700000 145 
1999 Northern Flood 324602 52 
May-01 Technological 120 127 
Jun-01 Flood 144025 11 
2005 Flood 350000 20 
2007 Northern Flood 332600 56 
Jun-10 Swedru Flood 3000 42 
Oct-10 Flood 280000 70 
Oct-10 Northern Flood 6000 35 
Nov-10 Afram Plains Flood 2800 0 
Nov-11 Accra Flood 43087 14 
2012 Cholera Outbreak 9548 100 
2013 Flood (Rainstorm) 25000 17 
2014 Cholera Outbreak 23622 190 
 36 
 
University of Ghana http://ugspace.ug.edu.gh
Jun-15 Flood and Fire 46370 250 
2016 Flood 41033 89 
Source: OFDA/CRED, Ghana Open Data and NADMO. 
According to the 12th edition of the Global Climate Risk Index, Ghana’s economic, social 
and infrastructural sectors are negatively impacted by extreme events. In the northern parts of 
the country, severe drought and flooding have reduced agricultural productivity and damaged 
properties and investments. In southern Ghana, sea level rise and other extreme weather 
events have led to loss of lives, displaced communities and low economic activities, 
especially fishing (Kreft, Eckstein, & Melchior, 2016). The incidence of extreme events 
hinders the development of the nation. Flooding, for example threatens roads and transport, 
education, energy supply, food and agriculture, water and sanitation, and health. Also, 
prolonged dry seasons and flooding increases urban migration as people in the north drift to 
the South to make a living. These migrants are exposed to new vulnerabilities on the streets 
and heighten the pressure on existing over-stretched urban services (Ghana National Climate 
Change Policy, 2012). 
 More people are exposed to the risk of extreme events in Ghana because of current 
development and demographic changes. This is as a result of rapid urbanization, increasing 
rural poverty, and declining ecosystems. Others include a high dependence on natural 
resources in rural areas, lack of secure livelihoods and limited informal and formal social 
safety nets. A study by Asante and Amuakwa-Mensah (2015) predicted a high temperature 
and low rainfall in Ghana in the years 2020, 2050 and 2080. This would increase sea-surface 
temperatures and will have extreme effects on fishery as well as rooted crops such as cassava. 
Also, diseases such as measles, diarrheal cases, guinea worm infestation, malaria, cholera, 
cerebro-spinal meningitis and other water related diseases will increase. 
 37 
 
University of Ghana http://ugspace.ug.edu.gh
2.8.2 The Ghanaian Insurance Industry 
Insurance started in Ghana around 1924 during the colonial era when Royal Guardian 
Enterprise, currently known as Enterprise Insurance Company Limited was established. In 
1955, the first local private insurance company called Gold Coast Insurance Company was 
started, followed by the establishment of State Insurance Company in 1962. By 1971, eleven 
more companies were added. In 2008, there were 17 life insurance and 22 non-life insurance 
companies licensed in Ghana. Previously, underwriters were able to offer both life and non-
life insurance products, but the 2006 Insurance Act stipulated that companies were no longer 
allowed to operate in both segments of the industry and mandated that these composite 
companies separate by December 2007. This law was passed so that companies could 
specialise in one area, gaining the expertise necessary for the industry to expand. Initially, the 
industry was regulated under the Insurance Law 1989 (PNDC Law 227). However, it is now 
regulated under the Insurance Act 2006 (Act 724) by the NIC (Boadu, Dwomo-Fokuo, 
Boakye, & Frimpong, 2014).  
There are basically eight (8) main classes of business written on the Ghanaian insurance 
market. These are motor class of business, fire/theft and property, marine, liability, 
engineering, personal and general accident, life/health and bonds/travel (NIC, 2016). The 
motor insurance, which is the umbrella for a number of product lines including the 
compulsory third party policy, dominates the non-life insurance business.  
Statistics shows that only one out of every 10 Ghanaian has an insurance policy and only one 
in 20 has health insurance. However, the insurance market in Ghana remains one of the 
fastest growing markets in Africa. The insurance industry in Ghana grew at a compound 
annual growth rate of 30.4% between 2009 and 2013, as compared with other African 
countries, such as Chad, 2.3%; Ivory Coast, 3.9%; Cameroon, 9.4%; and Uganda, 18.8% and 
as such, Oxford Economics anticipates the size of the insurance market to reach US$600 
 38 
 
University of Ghana http://ugspace.ug.edu.gh
million in 2018, compared to US$400 million in 2014 (EY, 2016). Although the insurance 
industry is growing rapidly, insurance penetration, defined as the contribution of total 
insurance premiums to gross domestic product, remains low. This can be attributed to certain 
factors such as high start-up costs, transaction costs, the lack of insurance knowledge, the 
complex nature of insurance products, and low willingness and ability to pay by the 
policyholders (Warner et al., 2013).  
While the expansion of the overall economy has slowed considerably in recent years, it 
remains positive and is expected to accelerate by 2020. As Ghana’s middle class grows along 
with the economy, the demand for insurance is expected to increase (Timetric, 2014). Also, 
the favorable regulatory and business environments can accelerate the expansion of insurance 
market (EY, 2016).   
 
 
 
 
 
 
 
 
 
 
 
 39 
 
University of Ghana http://ugspace.ug.edu.gh
    CHAPTER THREE 
               METHODOLOGY 
3.1 Introduction 
This chapter considers the methods that were used in the data collection for the research. It 
deals with the research design, the population of study, sampling and sampling procedure, 
instruments and procedures for data collection and the methods used to analyse the data 
obtained from the research. 
 
3.2 Research Design 
This study used a descriptive research design as well as a quantitative inferential statistical 
analysis design to identify, analyse and describe impact of extreme events on insurance 
demand, supply and insurers’ profitability in Ghana. This research design included 
procedures through which the relationship among the variables involved in the research 
problem was explored and analyzed with minimum error (Nenty, 2009). The descriptive 
nature of the research design also allowed the independent variables to reflect the true state of 
the situation without any interference. It also helped in describing the characteristics of the 
variables being studied. 
 
3.2.1 Quantitative Research Methodology 
Quantitative research methodology was used to systematically investigate and measure the 
relationship between extreme events and the insurance market. Quantitative research was best 
for this research because it plainly and particularly specified both the independent and the 
dependent variables used in the analysis. It also followed the original set of research 
 40 
 
University of Ghana http://ugspace.ug.edu.gh
objectives, reducing the influence of the researcher and assisting in deriving independent 
conclusions (Kealey & Protheroe, 1996).  
Furthermore, this research method allowed for longitudinal measures of successive 
performance of research subjects as well as capturing heterogeneity and variance by making 
it easy to collect information about great numbers of research subjects (Kelly, 2006; 
Matveev, 2002).  
3.2.2 Inferential Statistical Analysis 
Inferential statistical analysis research design used in this study allowed new or already 
published data to draw conclusions or predict a future effect or phenomenon about how 
extreme events affect the demand, supply and profitability of insurance companies based on 
information obtained from the sample. Also, the link between the explanatory and explained 
variables in the study was examined by inferential design since inferential statistical analysis 
looks at correlations and causations. Correlation mainly indicates the relationship between 
two variables, mostly the explanatory and explained variables whilst causation shows how 
the explanatory variables affect the explained variables.  
 
3.3 Research Population and Sample 
The population chosen for this research was the Ghanaian insurance industry. This consists of 
24 life insurance companies, 27 non-life companies, 3 reinsurance companies, 78 broking 
companies, 2 loss adjusters, 1 oil and gas company, 2 contact offices and 6,000 insurance 
agents as at June 2016, as reported by the NIC. The sample, which is the selected portion of 
the population that is being studied, comprised 13 life and 13 non-life insurance companies in 
Ghana. This sample was chosen due to availability of data and their role in the supply and 
demand for insurance in the country. 
 41 
 
University of Ghana http://ugspace.ug.edu.gh
3.4 Sampling Procedure 
In selecting the sample for this research, purposive sampling method was employed. 
Purposive sampling is a method of sampling where the sample is selected in a strategic way 
so that those sampled are relevant to the research questions (Bryman, 2012). This type of 
sampling procedure was suitable for the study because of the specific objectives to achieve at 
the end of the study. Insurance companies in the Ghanaian insurance sector were considered 
and chosen for the study because of their role in providing insurance services directly to 
customers. 
 
3.5 Data Collection 
Generally, data collected may be classified as primary or secondary data. For this study, 
secondary data on the history of extreme events was used to identify trends and prone areas 
from historical data of extreme events, and the variation of cost and effect following an 
extreme event. Secondary data was used because it was cost effective, time saving and 
convenient. Also, it allowed access to high quality larger datasets that involve larger samples 
and contain substantial breadth which gives a more representative of the target population 
and allow for greater validity and more generalizable findings (Johnston, 2014).  
The sources of secondary data analyzed was from annual reports and financial statements of 
insurance companies from National Insurance Commission (NIC). The occurrence, economic 
and social cost of catastrophes in Ghana was obtained from NADMO, OFDA/CRED and 
Ghana Open Data Initiative. Others were obtained from books, journal articles, related thesis 
and the internet. The population and income data was from the Ghana Statistical Service 
(GSS). 
 
 42 
 
University of Ghana http://ugspace.ug.edu.gh
3.6 Method of Analysis 
In order to achieve the objectives of this study, multiple regression with panel data analysis 
was used. Multiple regression showed the relationship between each dependent variable and 
the independent variables. In the study, multiple regression analysis aided in knowing how 
the dependent variable varies as a result of a change in the independent variables and also 
figuring out how the explanatory variables influenced the explained variable (Jeon, 2015). 
The study covered a period from 2007 - 2016. Extreme events in this study was limited to 
events caused by nature such as flood, epidemic and earthquake which are prevalent in 
Ghana. 
3.6.1 Panel Data Analysis 
Panel data analysis is a method of studying a particular subject within multiple sites, 
periodically observed over a defined time frame (Yaffee, 2003). A panel is a cross-section or 
group of people who are surveyed periodically over a given time span. Hence, observations in 
panel data involve at least two dimensions; a cross-sectional dimension, indicated by 
subscript i, and a time series dimension, indicated by subscript t. There are two types of panel 
data namely balanced and unbalanced panel data. Balanced panel data has equal number of 
observations for each individual (cross-section), while unbalanced panel data does not 
contain equal number of observations for each individual (Akbar, Imdadullah, Aman Ullah, 
& Aslam, 2011). 
Panel data analysis bequeaths regression analysis with both a spatial and temporal dimension. 
The spatial dimension refers to a set of cross-sectional units of observation such as countries, 
firms, or individuals. The temporal dimension refers to periodic observations of a set of 
variables characterizing these cross-sectional units over a particular time span (Hsiao, 2007). 
There are several types of panel data analytic models. There are constant coefficients models, 
fixed effects models, and random effects models. The constant coefficient model, also known 
 43 
 
University of Ghana http://ugspace.ug.edu.gh
as the pooled regression model, has constant coefficients for both intercepts and slopes. In the 
case that there is neither significant spatial or temporal effects, all of the data is pooled 
together and an ordinary least squares regression model is run. The fixed effects models have 
constant slopes but different intercepts according to the cross-sectional (group) unit or 
according to time, constant slopes but the intercept varies over country as well as time, and 
different slopes and intercept. Random effects model, also known as the error components 
model, has different intercept terms which are constant over time with the relationships 
between the explanatory and explained variables assumed to be the same both cross-
sectionally and temporally. However, the intercepts for each cross-sectional unit are assumed 
to arise from a common intercept plus a random variable that varies cross-sectionally but is 
constant over time (Brooks, 2008).  
Mostly, to determine the best model among these three types of models for your analysis, 
significance test with an F test, Hausman Specification Test, Breusch-Pagan Lagrange 
Multiplier test are conducted (Akbar, Imdadullah, Aman Ullah, & Aslam, 2011). 
3.6.2 Frequency versus Severity 
Extreme events were captured in two ways. They were looked at from their frequencies and 
severity. Blockbuster events represented the most severe major extreme events in the country. 
They were not fully estimated by the insurers and as such were expected to have a major 
effect on insurance pricing (Born & Viscusi, 2006). Unexpected frequency was a variable 
introduced in the study to capture the frequent number of unexpected events. When extreme 
events are anticipated, underwriting performance of insurers may not be affected since it will 
be reflected in the premiums. However, extreme events may adversely affect insurers 
underwriting performance when the number is unexpected (Born & Viscusi, 2006). 
 44 
 
University of Ghana http://ugspace.ug.edu.gh
3.6.3 Demand for Insurance 
The purchase of insurance allows an individual to reduce the financial risks from the loss or 
damage to his goods as well as his income generation capacity (Carvalho & Afonso, 2017).  
3.6.3.1 Empirical Model 
In this study, demand for insurance was measured as insurance density (insurance 
expenditure per capita). Insurance density is an important indicator to evaluate the insurance 
market’s current state and potential growth. Adopting the econometric style of Fier and 
Carson (2015), the regression model took the form: 
𝐷𝑒𝑚𝑎𝑛𝑑𝑖𝑡 = 𝛽0 + 𝜃𝑖 + 𝛽1𝑈𝑛𝑒𝑥𝑝𝑓𝑟𝑒𝑞𝑡−1 + 𝛽2𝐵𝑙𝑜𝑐𝑘𝑏𝑢𝑠𝑡𝑒𝑟𝑡−1 +
𝛽3𝐼𝑛𝑐𝑜𝑚𝑒𝑡−1 + 𝛽4𝐻𝐻𝐼𝑖𝑡−1 + 𝛽5𝐷𝑒𝑎𝑡ℎ𝑖𝑡−1 + 𝜌𝑡 + 𝜀𝑖𝑡   
where; 
i. Demand: Aggregate demand for insurance, measured as ratio of gross written 
premiums to total population. 
ii. Unexpfreq: unexpected frequency of extreme events; 1 if the actual number of 
extreme events experienced is greater than the median number of extreme events over 
the study period, 0 otherwise (Benali & Feki, 2017). 
iii. Blockbuster: unexpected severity of extreme events; 1 if there is an extreme event in a 
year with a loss among the highest 30% of all losses in the entire period of study, 0 
otherwise (Born & Viscusi, 2006). 
iv. Income: the gross national income per capita 
v. HHI: Herfindahl‐Hirschman Index to account for the level of concentration within a 
market. 
vi. Death: the number of death attributed to an extreme event 
 45 
 
University of Ghana http://ugspace.ug.edu.gh
vii. 𝑖: Class of insurance business; 𝑖 = 1, … ,8 
viii. 𝑡: time period; 𝑡 = 2007, … ,2016 
ix. θ: state fixed effects 
x. ρ: time fixed effects 
xi. ε: error term 
3.6.3.2 Variables Description 
Income: This variable controls the effect of the gross national income per capita on the 
demand of insurance. From literature, the relation between income and insurance demand is 
ambiguous. According to Bryan, Proctor, and Stoklosa (2015) and Zietz (2003), a positive 
relation exist between insurance demand and gross national income per capita. They posited 
that the national income per capita better captures the ability of the individual to pay for 
specific goods, such as insurance premiums and hence individuals who earn higher income 
can afford to purchase insurance for their health and the education of their children as well as 
their properties. However, Ciumas and Coca (2015) observed that income had a negative 
relationship with the demand for insurance. The possible reason being that individuals 
consider it unnecessary to protect themselves from risk when they have certain level of 
welfare or income.  
Herfindahl‐Hirschman Index (HHI): This variable measures the extent of competition within 
the market. The extent of competition within a market can directly affect premiums as well as 
the quantity of coverage available. Higher value indicates a decrease in competition and an 
increase of market power. 
Death: Total number of death after an extreme event can affect the decision to purchase 
insurance. 
 46 
 
University of Ghana http://ugspace.ug.edu.gh
3.6.4 Insurability of Risk (Supply of Insurance) 
Premiums reflect both the price of insurance as well as the quantity of insurance. Extreme 
events have diverse effect on insurance premiums. These events may raise the price of 
insurance, increasing the total premiums, for any given number of policies written (Born & 
Viscusi, 2006). However, extreme events may also restrict the quantity of insurance available 
due to high claims payment and cause some firms to exit the market leading to a reduction in 
total premium volumes (Ranger & Surminski, 2011).  
3.6.4.1 Empirical Model 
In this study, supply was measured using the supply density. Extreme events can have impact 
on insurance pricing which will largely influence the quantity of insurance offered by 
insurers to the society. The effect of extreme events on the supply of insurance, adapted from 
Born and Viscusi (2006), is represented by the equation; 
𝑆𝑢𝑝𝑝𝑙𝑦𝑖𝑡 = 𝛽0 + 𝛿𝑖 + 𝛽1𝑈𝑛𝑒𝑥𝑝𝑓𝑟𝑒𝑞𝑡−1 + 𝛽2𝐵𝑙𝑜𝑐𝑘𝑏𝑢𝑠𝑡𝑒𝑟𝑡−1 + 𝛽3𝐶𝑅𝑖𝑡 + 𝛽4𝑅𝑒𝑖𝑛𝑠𝑖𝑡 + 𝜔𝑡
+ 𝜀𝑖𝑡                                                                                       
where;   
i. Supply: Aggregate supply of insurance, measured as the natural log of the ratio of 
gross premiums to the number of insurers.  
ii. Unexpfreq: unexpected frequency of extreme events; 1 if the actual number of 
extreme events experienced is greater than the median number of extreme events over 
the study period, 0 otherwise (Benali & Feki, 2017). 
iii. Blockbuster: unexpected severity of extreme events; 1 if there is an extreme event in a 
year with a loss among the highest 30% of all losses in the entire period of study, 0 
otherwise. (Born & Viscusi, 2006) 
 47 
 
University of Ghana http://ugspace.ug.edu.gh
iv. CR: Claims Ratio  
v. Reins: Amount of reinsurance ceded to a reinsurance company 
vi. 𝑖: Class of insurance business; 𝑖 = 1, … ,8 
vii. 𝑡: time period; 𝑡 = 2007, … ,2016 
viii. δ: state fixed effects 
ix. ω: time fixed effects 
x. ε: error term 
3.6.4.2 Variables Description  
Claims Ratio: This ratio measures the efficiency of insurers in paying claims. Claims ratio is 
also known as underwriting risk and a higher ratio is an indication that premiums collected 
are not adequate to pay claims and other expenses (Berhe & Kaur, 2017). 
Reinsurance: The transfer of insurance from a primary insurer to another insurer (reinsurer) 
provides the finances needed by the primary insurer, aside his capital and reserve, to take in 
more risk by increasing the primary insurer’s coverage (Prelipcean & Boscoianu, 2010). If 
extreme event losses exceed the retention limit of the insurer, the reinsurer comes in and 
make up for the difference (Benali & Feki, 2017). 
3.6.5 Profitability of Insurance Companies 
One important goal of financial management is to always seek to maximize the owner’s or 
shareholders wealth (Ortyński, 2016). This is because for companies especially insurance 
companies to be sustainable in the competitive globalized environment, they need to earn 
profit. Also, without profit insurers will find it difficult to attract outside capital so as to meet 
their objectives (Berhe & Kaur, 2017). 
 48 
 
University of Ghana http://ugspace.ug.edu.gh
3.6.5.1 Empirical Model 
In this study, profitability was measured by return on asset (ROA). ROA is used to measure 
profitability because it captures the fundamentals of business performance in a holistic way, 
looking at both income statement performance and the assets required to run a business 
(Malik, 2011). Examining the impact of extreme events on the profitability of the insurance 
companies, the model used by Benali and Feki (2017) was replicated as follows; 
𝑅𝑂𝐴𝑖𝑡 = 𝛽0 + 𝛼𝑖 + 𝛽1𝑈𝑛𝑒𝑥𝑝𝑓𝑟𝑒𝑞𝑡 + 𝛽2𝐵𝑙𝑜𝑐𝑘𝑏𝑢𝑠𝑡𝑒𝑟𝑡 + 𝛽3𝑃𝑟𝑒𝑚𝑆𝑢𝑟𝑝𝑖𝑡 + 𝛽4𝑆𝑖𝑧𝑒𝑖𝑡
+ 𝛽5𝐸𝑥𝑝𝑅𝑎𝑡𝑖𝑜𝑖𝑡 + 𝛽7𝑃𝑟𝑒𝑚𝐺𝑟𝑤𝑖𝑡 + 𝛾𝑡 + 𝜀𝑖𝑡                                                
where; 
i. ROA: Return on Asset 
ii. Unexpfreq: unexpected frequency of extreme events; 1 if the actual number of 
extreme events experienced is greater than the median number of extreme events over 
the study period, 0 otherwise (Benali & Feki, 2017). 
iii. Blockbuster: unexpected severity of extreme events;1 if there is an extreme event in a 
year with a loss among the highest 30% of all losses in the period of study, 0 
otherwise (Born & Viscusi, 2006). 
iv. PremSurp: Premiums-to-surplus ratio 
v. Size: Size of the firm 
vi. ExpRatio: Expenses Ratio 
vii. PrewGrw: Premium Growth 
viii. 𝑖: Insurance Company; 𝑖 = 1, … ,13 
ix. 𝑡: time period; 𝑡 = 2007, … ,2016 
 49 
 
University of Ghana http://ugspace.ug.edu.gh
x. α: state fixed effects 
xi. γ: time fixed effects 
xii. ε: error term 
3.6.5.2 Variables Description 
PremSurp: Premium to surplus ratio measures the financial strength of the insurance 
company and the insurer’s ability to absorb above average losses (Benali & Feki, 2017). 
Size: The size of company mostly measures the total assets or market share of an insurer. It 
also captures the capacity of insurers for dealing with adverse market fluctuations (Berhe & 
Kaur, 2017). 
Expense Ratio (ExpRatio): This ratio measures the percentage of premium that insurers use 
for paying all the cost to provide adequate coverage as required by the policy which includes 
acquiring, writing and servicing insurance and reinsurance.  A high ratio signifies that, the 
insurer is inefficient in discharging its insurance obligations and this could affect the overall 
profitability of the insurance company (NIC, 2016). 
Premium Growth (PremGrw): the growth in premiums collected over the years may 
positively affect insurance company’s profitability. The growth in premiums may signify 
higher rate of market penetration and better performance of an insurer as found by Ullah, 
Faisal, and Zuhra (2016). However, Kim, Anderson, Amburgey and Hickman (1995) in their 
study observed that swift increase in premiums is a major cause of insurer insolvency because 
companies may be obsessed with growth and can neglect other important objectives. 
 
3.7 Variables Construction and Expected Signs 
Tables 3.1 and 3.2 show the dependent and independent variables used in the study, how they 
were measured and their expected signs. 
 50 
 
University of Ghana http://ugspace.ug.edu.gh
Table 3.1: Definition of Dependent Variables 
DEPENDENT VARIABLES DEFINITIONS 
Demand Gross Premiums/Total Population 
Supply Natural log of Gross Premiums/Number of Insurers 
Return on Asset (ROA) Net financial result/Total Assets 
 
Table 3.2: Definition of Independent Variables and Expected Signs 
INDEPENDENT VARIABLES       DEFINITIONS EXPECTED SIGN 
Unexpected Frequency Unexpected Frequency of             +/- 
(Unexpfreq) extreme event 
Blockbuster Unexpected Severity of extreme             +/- 
event 
Income Gross income per capita              +/- 
Market Concentration (HHI) Sum of the square of the market              + 
share  
Death Natural log of the number of              + 
death attributed to an extreme 
event 
 
 51 
 
University of Ghana http://ugspace.ug.edu.gh
Claims Ratio (CR) Claims incurred/Premiums         +/- 
earned 
Reins Amount of Reinsurance         + 
PremSurp Premiums-to-surplus ratio         + 
Size of the firm (Size) Natural log of total assets         + 
Expense Ratio (ExpRatio) Ratio of underwriting expenses         +/- 
to net premium 
Premium Growth (PremGrw) Changes in premium earned         + 
relative to prior year 
 
3.8 Limitations to the Methodology 
One limitation was the difficulty in finding secondary data for the insurance companies due 
to the sensitive nature of information required as well as the competitive nature of the 
industry. However, the researcher had to go to the regulatory body of the insurance industry 
(NIC) to get the needed information since all insurance companies are mandated to send a 
copy of their financial report to NIC. 
Also, secondary data on extreme events for Ghana was minimal.  The researcher had to resort 
to interviews with the staff of NADMO in other to get adequate information to carry out the 
study. 
 
 
 52 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FOUR 
DATA ANALYSIS, ESTIMATION AND DISCUSSION OF RESULTS 
4.1 Introduction 
This chapter presents the results of the study. The study sought to examine the effect of 
extreme events on insurance demand and supply and their impact on the profitability of 
insurance firms in Ghana. Data on the frequency, severity and death as a result of extreme 
events were examined to determine whether they have an impact on the insurance market in 
Ghana. The study further examined whether other demographic variables such as income and 
market concentration have impact on the demand, supply and profitability of insurance in 
Ghana. Also, variables such as claims ratio, premium to surplus ratio, size, expense ratio and 
premium growth obtained from thirteen life and general insurance companies were examined 
to assess how they affect the profitability of both life and non-life insurance firms in Ghana.  
The chapter begins with a description of each data set using summary statistics. This is 
followed by the pre-tests such as the multicollinearity and Hausman test to determine the 
right regression model to use. Other diagnostic test are performed to confirm the 
appropriateness of the model and the chapter ends with a discussion of the results. The study 
used STATA software for all the analysis. 
 
4.2 Extreme Events and Demand for Insurance 
4.2.1 Descriptive Statistics 
Table 4.1 presents the summary statistics for both the explained and the explanatory 
variables. The mean, standard deviation, minimum and maximum values for all variables 
used are displayed. 
 53 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4.1: Descriptive Statistics 
Variable Obs. Mean Std. Dev. Min Max 
Demand 80 4.0416 5.8144 0.0186 30.446 
Unexpfreqt−1  80 0.1 0.3019 0 1 
Blockbustert−1  80 0.3 0.4611 0 1 
Income 80 7.1069 0.3163 6.3969 7.4559 
HHI 80 0.2767 0.0532 0.1142 0.3378 
Death 80 3.6642 3.3666 0 11.6074 
Source: Author’s calculations using STATA 14.0 software. 
The highest insurance density (demand) of 30.446 was recorded by life insurance business in 
2013 and the lowest by bonds class of business in 2007. The 3rd June, 2015 fire and flood 
event recorded the highest number of death within the study period. Also, aside Demand and 
Death variables, the rest of the variables have their data values lying close to the mean of the 
dataset. 
4.2.2 Multicollinearity Test 
Multicollinearity occurs when the correlation between two or more explanatory variables is 
high. This means one explanatory variable can be used to predict the other. This problem can 
lead to unstable and unreliable regression estimates. To test for multicollinearity, a bilateral 
correlation matrix between the variables are presented in table 4.2. It can be observed that the 
highest correlation exist between Death and Unexpfreqt−1 variables. However, Benali and 
Feki (2017) posited that a correlation coefficient of more than 0.8 existing between two 
 54 
 
University of Ghana http://ugspace.ug.edu.gh
independent variables indicate a problem of multicollinearity. From the table, there is no 
correlation coefficient which is more than 0.8 and hence there is no problem of 
multicollinearity. 
Table 4.2: Correlation Matrix 
 
Demand 𝐔𝐧𝐞𝐱𝐩𝐟𝐫𝐞𝐪𝐭−𝟏 𝐁𝐥𝐨𝐜𝐤𝐛𝐮𝐬𝐭𝐞𝐫𝐭−𝟏 Income HHI Death 
     
Demand    1 
    
Unexpfreqt−1  -0.0528         1 
   
Blockbustert−1   0.039      0.5092       1 
  
Income  0.3132      0.0254   -0.1474    1 
 
HHI -0.609      0.0225   -0.0307 -0.3709      1 
Death  0.1497      0.7914    0.6642 0.3298 -0.2327       1 
Source: Author’s calculations using STATA 14.0 software. 
To decide whether there exist a panel effect in the model, a Breusch-Pagan Lagrange 
multiplier (LM) test was done. The null hypothesis in the LM test is that variances across 
entities is zero which means no significant difference across units (i.e. no panel effect) and 
hence a simple OLS is better. The alternate is that there exit a panel effect. The results for the 
panel effect test is shown below. 
Table 4.3: Testing for Panel Effect 
Chi-square value P-value 
 55 
 
University of Ghana http://ugspace.ug.edu.gh
22.30 0.0000 
 
Here, a p-value of 0.0000 which is less than 0.05 suggests that we reject the null hypothesis. 
It can therefore be concluded that OLS is not appropriate and that panel regression analysis 
should be considered since there exist a significant differences amongst the various classes of 
insurance business. 
4.2.3 Test for Best Panel Model 
To determine the appropriate panel model to be used, a Hausman test was carried out to 
decide between fixed and random effects with the null hypothesis that the preferred model is 
random effects versus the alternative to the fixed effects as shown in table 4.4 
Table 4.4: Hausman Test 
Chi-square value P-value 
114.75 0.0000 
 
From the table above, the p-value of the test is less than 0.05. We reject the null hypothesis 
and conclude that fixed effects model is preferred to the random effects model. 
4.2.4 Results of Panel Regression 
Using a fixed effects panel regression model, the results of the model are presented in table 
4.5. 
 
 
 56 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4.5: Regression Results of Demand for Insurance 
Variables Coefficient Standard Error T-Statistic P-Value 
(constant)   -69.439 23.079 -3.01 0.02** 
Unexpfreqt−1    -13.305 4.0161 -3.31 0.013** 
Blockbustert−1    -1.0115 0.8011 -1.26 0.247 
Income   6.3908 1.7481 3.66 0.008*** 
HHI   88.474 39.147 2.26 0.058* 
Death   1.4224 0.4686 3.04 0.019** 
R-squared   0.7185 
Adjusted R-squared   0.4604 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Author’s calculations using STATA 14.0 software. 
 
From table 4.5, it can be seen that the unexpected frequency of extreme events in the previous 
year (Unexpfreqt−1) negatively influenced the demand for insurance. This means that if the 
previous year experienced more than the median number of extreme events, it results in a 
negative demand for insurance. This could be due to the expectation of public and political 
responses as well as international assistance after losses from these events, as found by 
Lamond and Penning-Rowsell (2014), Ranger and Surminski (2011) and Ackah and Owusu 
(2012). An interview with Mr. Richard Amo-Yartey (the director in charge of geological 
disasters at NADMO) reveals that during the June 3rd flood and fire disaster which occurred 
in 2015 in Accra, the government gave each family of a victim GHȼ10000 as compensation 
 57 
 
University of Ghana http://ugspace.ug.edu.gh
amongst other items. Furthermore, China supported the country with US$50000 during the 
heavy floods in 2010. This supports the fact that government intervention and international 
assistance during such events goes a long way to reduce the demand for insurance as a way of 
financing risk posed by extreme events. Again, according to Dlugolecki and Hoekstra (2006), 
extreme events does not positively affect demand for insurance because of the poor 
perception of risk exposure of these events due to lack of education and public awareness. In 
Ghana, most informal sector workers who form the majority of the population in the country, 
do not demand or use insurance as a precautionary measure but rather trust God to provide 
protective care as found by Ackah and Owusu (2012) and Archer (2013) in their studies. 
However, though the unexpected severe event of the previous year (Blockbustert−1) had a 
negative impact on insurance demand, it was not significant. 
Income had a positive relationship with the demand for insurance. This confirms the findings 
of Zietz (2003), Bryan, Proctor, and Stoklosa (2015), Fofie (2016) and Peprah, Koomson, & 
Forson (2017) in their studies that individuals who earn higher income can afford to purchase 
insurance for their health and the education of their children as well as their properties and 
also, as income increases there is a desire to protect potentially higher levels of income in the 
event of premature death. 
Herfindahl-Hirrschman Index (HHI) which measures the level of market concentration and 
the extent of competition in the market was positively related to demand for insurance. This 
means that a growth in the level of competition reduces the market power which leads to a 
decrease in the price of insurance. Like general goods, a decrease in the price of insurance 
will cause an upsurge in demand. This confirms the findings of Fier and Carson (2015) in 
their studies. 
The number of death in an extreme event (Death) had a positive relationship with the demand 
for insurance. This shows that if the number of death in an extreme event increases, it 
 58 
 
University of Ghana http://ugspace.ug.edu.gh
positively affect the decision to patronise insurance. After a personal experience of a 
traumatic event, people may be led to purchase insurance in an effort to protect themselves 
from future harm. For instance, the funeral insurance policy is one of the most patronized 
policy in the life insurance business (National Insurance Commission, 2010; Ackah & 
Owusu, 2012). Fier and Carson (2015) found similar results in their studies. 
4.2.5 Diagnostics Test 
An assumption of the panel regression is that the errors must be serially uncorrelated as it 
causes the standard errors of the coefficients to be smaller than they actually are leading to 
inefficient estimates. Moreover, residuals across entities should not be correlated because it 
could lead to biasedness in test results. The Lagram-Multiplier/Wooldridge test for serial 
correlation and the Pasaran CD test for cross-sectional dependence in the panel model is 
presented in table 4.6. The null hypothesis states that there is no serial correlation and cross-
sectional dependence. 
Table 4. 6: Diagnostics Test for Serial Correlation and Cross-sectional Dependence 
Diagnostics Test Chi-square Value P-value 
Test for Serial Correlation 9.55 0.0891 
Test for Cross-sectional Dependence 7.51 0.1853 
 
Since the p-values in table 4.6 are greater than 0.05, we fail to reject the null hypothesis and 
conclude that there is no serial correlation or cross-sectional dependence in the model. 
 59 
 
University of Ghana http://ugspace.ug.edu.gh
4.3 Extreme Events and Supply of Insurance 
4.3.1 Descriptive Statistics 
Table 4.7 displays the mean, standard deviation, minimum and maximum values for both the 
explained and the explanatory variables used for the study. 
Table 4. 7: Descriptive Statistics 
Variable Obs. Mean Std. Dev. Min Max 
Supply 80 14.4188 1.4797 10.1227 17.2388 
Unexpfreqt−1  80 0.1 0.3019 0 1 
Blockbustert−1  80 0.3 0.4611 0 1 
 CR 80 0.3227 0.1061 0.16 0.56 
 Reins 80 18.4787 1.4131 13.2021 19.812 
Source: Author’s calculations using STATA 14.0 software. 
From table 4.7, life insurance business recorded the highest supply of insurance in 2016. 
Motor insurance also ceded the highest amount of insurance to reinsurance which is not 
surprising since it is the class of business that dominates the non-life market (NIC, 2016). 
Again, life class of business recorded the highest claim ratio (CR) in 2015. 
4.3.2 Multicollinearity Test 
Table 4. 8: Correlation Matrix 
 
Supply 𝐔𝐧𝐞𝐱𝐩𝐟𝐫𝐞𝐪𝐭−𝟏 𝐁𝐥𝐨𝐜𝐤𝐛𝐮𝐬𝐭𝐞𝐫𝐭−𝟏 CR Reins 
 60 
 
University of Ghana http://ugspace.ug.edu.gh
    
Supply     1 
   
Unexpfreqt−1  -0.0888      1 
  
Blockbustert−1  -0.0405   0.5092      1 
 
CR 0.5532   0.1692   0.0166 1 
Reins -0.2373   -0.0327   0.0242 0.0302 1 
Source: Author’s calculations using STATA 14.0 software. 
Table 4.8 shows a bilateral correlation matrix between the variables and it can be seen that 
the highest correlation exist between CR and supply. Since there is no correlation coefficient 
between the independent variables which is more than 0.8, the absence of multicollinearity is 
confirmed. 
Also, to decide whether there exist a panel effect in the model, a Breusch-Pagan Lagrange 
multiplier (LM) test was performed. The results for the panel effect test is presented in Table 
4.9. 
Table 4. 9: Testing for Panel Effect 
Chi-square value P-value 
191.88 0.0000 
 
Here, a p-value of 0.0000 which is less than 0.05 suggests that we reject the null hypothesis 
and conclude that OLS is not appropriate. A panel regression analysis should be considered 
since there exist a significant differences amongst the various classes of insurance business. 
 61 
 
University of Ghana http://ugspace.ug.edu.gh
4.3.3 Test for Best Panel Model 
Now deciding between a fixed effects panel model and a random effects panel model will 
require a Hausman test which is presented in table below.   
Table 4. 10: Hausman Test 
Chi-square value P-value 
7.31 0.1206 
 
Since the p-value of the test is greater than 0.05 we fail to reject the null hypothesis. 
Therefore, we conclude that random effects model is preferred to the fixed effects model. 
4.3.4 Results of Panel Regression 
Using a random effects panel regression model, table 4.11 presents the regression results.  
Table 4. 11: Regression Results of Supply of Insurance 
 Variable Coefficient  Standard Error T-Statistic P-Value 
(constant)    8.3406   1.288 6.48 0.000*** 
Unexpfreqt−1     -0.7971   0.4123 -1.93 0.053* 
Blockbustert−1     0.0977   0.079 1.24 0.216 
CR    5.4203   1.4853 3.65 0.000*** 
Reins    0.237   0.0668 3.55 0.000*** 
R-squared    0.8849 
 62 
 
University of Ghana http://ugspace.ug.edu.gh
Adjusted R-squared    0.8663 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Author’s calculations using STATA 14.0 software. 
 
From table 4.11, it can be observed that the unexpected frequency of extreme events in the 
previous year (Unexpfreqt−1) negatively affects the supply of insurance. This means that if 
the previous year experienced more than the median number of extreme events, there will be 
a fall in the supply of insurance (Harrington & Niehaus, 2004). Extreme events may result in 
higher losses and insurers may be faced with two options; either to increase premiums to 
recoup their losses and restrict coverage or to completely remove cover from the market in 
order to maintain solvency and shareholder returns. In both ways, the availability and 
affordability (supply) of insurance will be reduced in the market. Also, as the unexpected 
frequency of these events increases, losses paid could also increase leading to a depletion of 
insurers’ capital reserves and increase competition among insurers which could pressure 
some to exit the market (Turner & Deng, 2015). This finding is supported by researchers such 
as Chen, Doerpinghaus, Lin and Yu (2008), Maynard (2008) and Yi (2013) in their studies.  
On the other hand, the unexpected severe event of the previous year (Blockbustert−1) 
positively impacted the supply of insurance but it was not significant. 
Claims Ratio (CR) positively affected the supply of insurance. Since the fundamental duty of 
an insurer is to pay claims, the ability to pay claims and on time wins and maintains the 
confidence of existing policyholders and attract new ones. Thus, the greater the market share, 
the greater the supply of insurance in the market (NIC, 2016). 
Reinsurance (Reins) had a positive relationship with the supply of insurance. The reinsurance 
companies support insurance firms by increasing their capacity to underwrite large risks, and 
accept liability for losses beyond an insurer’s retention limit thereby reducing the amount of 
 63 
 
University of Ghana http://ugspace.ug.edu.gh
capital needed to achieve a given level of solvency. Other researchers such as Prelipcean and 
Boscoianu (2010), and Benali and Feki (2017) support this findings in their studies.  
4.3.5 Relationship between Supply and the Interaction of Claims Ratio and Reinsurance 
From the supply model above, it was seen that both Claims Ratio (CR) and Reinsurance 
(Reins) had a positive influence on the supply of insurance. However, the ability to 
effectively pay claims may also depend on the amount of reinsurance to support payments 
when they fall due. This may increase the capacity of the insurer to underwrite more policies. 
The regression results of the combine effect of claims ratio and reinsurance (CR_ Reins) on 
Supply is shown in table 4.12. This is based on the AIC / BIC model selection by Brooks 
(2008). See appendix for details. 
Table 4.12: Regression Results of Supply of Insurance (With CR_Reins) 
Supply Coefficient  Standard Error T-Statistic P-Value 
Constant   12.4939     0.455   27.46 0.000*** 
Unexpfreqt−1    -0.8832     0.2387   -3.7 0.000*** 
Blockbustert−1     0.1294     0.1543   0.84 0.402 
CR_Reins    0.3308     0.0308   10.76 0.000*** 
R-squared    0.8771 
Adjusted R-squared    0.8593 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Author’s calculations using STATA 14.0 software. 
 
 64 
 
University of Ghana http://ugspace.ug.edu.gh
The results in table 4.12 shows that the interaction between claims ratio and reinsurance 
(CR_Reins) positively and significantly affects the supply of insurance. This indicates that, a 
high reinsurance for a policy will increase the capability of the insurer to pay claims which 
will reduce the impact of losses on the insurer’s capital and reserves and enable the insurer 
underwrite more risks. This findings is supported by other researchers such as Berhe and 
Kaur (2017) in their studies. 
4.3.6 Diagnostics Test 
Table 4.13 shows the Lagram-Multiplier/Wooldridge test for serial correlation as well as the 
Pasaran CD test for cross-sectional dependence in the panel model.  
Table 4. 13: Diagnostics Test for Serial Correlation and Cross-sectional Dependence 
Diagnostics Test Chi-square Value P-value 
Test for Serial Correlation 2.115 0.1892 
Test for Cross-sectional Dependence 0.329 0.7423 
 
The p-values of 0.1892 and 0.7423 are greater than 0.05. This means that there is no serial 
correlation in the model and also the error terms are not correlated across the class of 
insurance business. 
 
4.4 Extreme Events and Profitability of Life Insurance Companies 
4.4.1 Descriptive Statistics 
Table 4.14 shows the descriptive statistics of the variables used to examine the profitability of 
the life insurance companies in Ghana. 
 65 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 14: Descriptive Statistics 
Variable Obs. Mean Std. Dev. Min Max 
ROA 127 0.0170 0.3233 -3.12 0.6 
Unexpfreq 130 0.1 0.3012 0 1 
Blockbuster 130 0.3 0.4600 0 1 
PremSurp 124 3.1837 6.0909 -1.4977 38.6350 
Size 125 17.0326 1.4846 12.5967 20.6554 
ExpRatio 127 0.4989 0.2515 0.12 1.37 
PremGrw 130 0.3264 0.3505 -0.5627 2.4827 
Source: Author’s calculations using STATA 14.0 software. 
From table 4.14 the lowest ROA was recorded by Unique Life Insurance in 2011 which is 
over 300% reduction in its returns whilst Ghana Life Insurance Company recorded the 
highest returns in 2007. On the average life insurers’ profitability within the study period was 
about 2%. The range of the premium to surplus ratio (PremSurp) variable is wider than the 
rest of the variables and also its spread around the mean is the greatest. 
4.4.2 Multicollinearity Test 
To verify whether the explanatory variables are not highly correlated or not, a bilateral 
correlation matrix between the variables is presented in table 4.15. 
 
 
 66 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 15: Correlation Matrix 
 
ROA Unexpfreq Blockbuster PremSurp Size ExpRatio PremGrw 
      
ROA    1 
     
Unexpfreq 0.072      1 
    
Blockbuster 0.028   0.509     1 
   
PremSurp -0.077   -0.069  0.011    1 
  
Size 0.068   -0.084  -0.084 -0.052     1 
 
ExpRatio -0.121   0.060  0.043 -0.004 -0.277      1 
PremGrw 0.129   0.199  0.094 -0.231 -0.186   -0.033  1 
Source: Author’s calculations using STATA 14.0 software. 
From the correlation matrix above, the highest correlation of 0.51 exist between Unexpfreq 
and Blockbuster and that there is no correlation coefficient between two explanatory 
variables which is more than 0.8. Thus, the absence of multicollinearity is confirmed. 
A Breusch-Pagan Lagrange multiplier (LM) test was carried out to decide whether there exist 
a panel effect in the model. The null hypothesis in the LM test is that variances across life 
insurance firms is zero meaning no panel effect and hence a simple OLS is better. Table 4.16 
shows the results for the test for the panel effect. 
Table 4. 16: Testing for Panel Effect 
Chi-square value P-value 
 67 
 
University of Ghana http://ugspace.ug.edu.gh
9.53 0.0011 
 
Here, a p-value of 0.0011 which is less than 0.05 suggests that we reject the null hypothesis 
and conclude that OLS is not appropriate. Panel regression analysis should be considered 
since there exist a significant differences amongst the various life insurance companies. 
4.4.3 Test for Best Panel Model 
A Hausman test for the best panel model was conducted in table 4.17 to decide between fixed 
and random effects with the null hypothesis that the preferred model is random effects.  
Table 4. 17: Hausman Test 
Chi-square value P-value 
7.64 0.2656 
 
Since the p-value of the test is greater than 0.05 we fail to reject the null hypothesis. 
Therefore, we conclude that random effects model is preferred to the fixed effects model. 
4.4.4 Results of Panel Regression 
Using a random effects panel regression model, the results of the model are presented in table 
4.18. 
Table 4. 18: Regression Results of Profitability of Life Insurance Companies 
Variables Coefficient Standard Error T-Statistic P-Value 
(constant) 0.2341     0.2209 1.06 0.289 
 68 
 
University of Ghana http://ugspace.ug.edu.gh
Unexpfreq 0.0726     0.0704 1.03 0.303 
Blockbuster -0.0496    0.0184 -2.69 0.007** 
PremSurp 0.0015    0.0025 0.58 0.563 
Size -0.0093    0.0115 -0.81 0.417 
ExpRatio -0.1126    0.0593 -1.9 0.057* 
PremGrw 0.0784    0.0363 2.16 0.031** 
R-squared 0.3332 
Adjusted R-squared 0.2142 
***,**,* Significance at 1%, 5% and 10%, respectively 
Source: Author’s calculations using STATA 14.0 software. 
 
However, in selecting the appropriate model, the AIC and BIC criterion was used. The 
regression model which minimized the value of the information criteria was considered 
(Brooks, 2008). The best regression model was the one without the variable Size (see tables 6 
to 10 in the appendix for details). 
Table 4. 19: Regression Results of Profitability of Life Companies (Without Size) 
Variables Coefficient Standard Error T-Statistic P-Value 
(constant) 0.0702  0.035  2.01 0.045** 
Unexpfreq 0.0722  0.0704  1.03 0.305 
Blockbuster -0.0473  0.0177  -2.68 0.007** 
 69 
 
University of Ghana http://ugspace.ug.edu.gh
PremSurp 0.0014  0.0023  0.63 0.527 
ExpRatio -0.1098  0.0611  -1.8 0.072* 
PremGrw 0.0881  0.0316  2.78 0.005** 
R-squared 0.3261 
Adjusted R-squared 0.2138 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Own calculations using STATA 14.0 software. 
 
From table 4.19, unexpected severe event in the year (Blockbuster) negatively influenced the 
profitability of life insurance companies. This implies that if there is an extreme event in the 
year with a loss among the highest 30% of all losses, it will lead to a fall in life insurers’ 
profitability. This is because an unexpected severe event can increase losses and disrupt the 
activities of insurers by putting stress on their resources leading to a great negative impact on 
profitability. Insurers may invest at least part of their capital in equities and properties, and 
these may be adversely affected by extreme events. Also, financial assets may be held to 
support the capital and reserves of the insurer and these assets may be adversely affected 
leading to a fall in their value. Risks due to extreme events may also drive up insurers’ capital 
requirements. Over the years, the minimum capital requirement for insurance companies in 
Ghana has increased from GHȼ200 in 1989, GHȼ5million in 2006, GHȼ10million in 2013 to 
GHȼ15million in 2015. As the value of assets are falling concurrently with a rise in the 
liabilities and capital requirements, the overall profitability of the insurer may be negatively 
affected. Researchers such as Coomber (2006) and Maynard (2008) supports this findings in 
their studies.  
 70 
 
University of Ghana http://ugspace.ug.edu.gh
Both the unexpected frequency of extreme events (Unexpfreqt−1) and the premium-to-
surplus ratio (PremSurp) positively influenced the profitability of life insurers’ but they were 
not significant.  
Expenses ratio (ExpRatio) was seen to negatively influence life insurers’ profitability. This 
ratio shows how efficient an insurance company’s operations are at bringing in premiums. It 
implies that as the cost of underwriting new policies increases, it reduces the premiums 
available to invest and generate other income. This in turn reduces the profits available and 
the overall profitability of the insurance company. It confirmed the findings of Ullah, Faisal, 
and Zuhra (2016) in their studies. 
Premium growth (PremGrw) evaluates the level of market penetration and it positively 
affected the profitability of life insurance companies. The growth in premiums may signify 
better performance of an insurer. Premiums are the major source of revenue for insurers and 
they can be invested to generate more income which will increase their capacity to pay 
claims, build confidence in policyholders and maintain solvency. This will also better the 
chances of insurance companies being profitable. The results from this study is consistent 
with Ahmed, Ahmed and Ahmad (2011) in Pakistan and Ullah, Faisal and Zuhra (2016) in 
Bangladesh 
4.4.5 Diagnostics Test 
The Lagram-Multiplier/Wooldridge test for serial correlation and the Pasaran CD test for 
cross-sectional dependence in the panel model is conducted in table 4.20. 
Table 4. 20: Diagnostics Test for Serial Correlation and Cross-sectional Dependence 
Diagnostics Test Chi-square Value P-value 
Test for Serial Correlation 3.225 0.0977 
 71 
 
University of Ghana http://ugspace.ug.edu.gh
Test for Cross-sectional Dependence 0.378 0.7319 
 
The p-values in table above shows that there is no serial correlation in the model and the error 
terms are not correlated across the insurance companies. 
4.5 Extreme Events and Profitability of Non-life/General Insurance Companies 
4.5.1 Descriptive Analysis 
The descriptive statistics of the data for the non-life insurance companies is shown in table 
4.21. 
Table 4. 21: Descriptive Statistics 
Variable Obs. Mean Std. Dev. Min Max 
ROA 130 0.0368 0.2187 -2.15 0.73 
Unexpfreq 130 0.1 0.3012 0 1 
Blockbuster 130 0.3 0.46 0 1 
PremSurp 128 2.1289 1.4523 0.2175 10.8682 
Size 128 16.9519 1.136 13.9959 19.4796 
ExpRatio 128 0.702 0.5331 0 4.95 
PremGrw 129 0.3234 0.4148 -0.5471 2.7619 
Source: Author’s calculations using STATA 14.0 software. 
 72 
 
University of Ghana http://ugspace.ug.edu.gh
From table 4.21 the minimum ROA was recorded by Prime insurance company in 2014 
which is over 200% reduction in its returns whilst Enterprise insurance company recorded the 
highest returns of over 70% increase in its returns in 2011. On the average the profitability of 
the general insurance companies within the study period was about 4% which is quite higher 
than that of the life companies. 
4.5.2 Multicollinearity Test 
To verify whether the explanatory variables are not highly correlated or not, a bilateral 
correlation matrix between the variables are presented in table 4.22. 
Table 4. 22: Correlation Matrix 
 
ROA Unexpfreq Blockbuster PremSurp Size ExpRatio PremGrw 
      
ROA    1 
     
Unexpfreq 0.058       1 
    
Blockbuster 0.111     0.512      1 
   
PremSurp -0.020     0.034   -0.042       1 
  
Size 0.204    -0.058   -0.005   -0.116     1 
 
ExpRatio -0.766    -0.016   -0.157   -0.090 -0.025      1 
PremGrw -0.259     0.016    0.111   0.039 -0.253   0.389      1 
Source: Author’s calculations using STATA 14.0 software. 
 73 
 
University of Ghana http://ugspace.ug.edu.gh
The table above shows that the highest correlation of 0.512 exist between Unexpfreq and 
Blockbuster variables. However, there is no correlation coefficient between two independent 
variables above 0.8 which confirms the absence of multicollinearity. 
A Breusch-Pagan Lagrange multiplier (LM) test was done to decide whether there exist a 
panel effect in the model. The null hypothesis in the LM test is that variances across life 
insurance firms is zero meaning no panel effect and hence a simple OLS is better. Table 4.23 
displays the results for the panel effect test. 
Table 4. 23: Testing for Panel Effect 
Chi-square value P-value 
9.03 0.0410 
 
A p-value of 0.0410 which is less than 0.05 suggests that we reject the null hypothesis and 
conclude that OLS is not appropriate. Panel regression analysis should be considered since 
there exist a significant differences amongst the various non-life insurance companies. 
4.5.3 Test for Best Panel Model 
A Hausman test for the best panel model was conducted in table 4.24 to decide between fixed 
and random effects with the null hypothesis that the preferred model is random effects.  
Table 4. 24: Hausman Test 
Chi-square value P-value 
13.65 0.0338 
 
 74 
 
University of Ghana http://ugspace.ug.edu.gh
Since the p-value of the test which is 0.0338 is less than 0.05 we reject the null hypothesis 
and conclude that fixed effect model is preferred to the random effect model. 
4.5.4 Results of Panel Regression 
Using a fixed effect panel regression model, the results of the model are presented in table 
4.25. 
Table 4. 25: Regression Results of Profitability of Non-life Insurance Companies 
Variables Coefficient  Standard Error T- Statistic P-Value 
(Constant) -1.0881       0.3659 -2.97 0.012** 
Unexpfreq 0.0923       0.067 1.38 0.194 
Blockbuster -0.0541       0.0337 -1.6 0.135 
PremSurp -0.0164       0.0064 -2.55 0.025** 
ExpRatio -0.3697       0.0572 -6.47 0.000*** 
PremGrw 0.1118       0.0975 1.15 0.274 
Size 0.082       0.0227 3.62 0.004** 
R-squared 0.7848 
Adjusted R-squared 0.6001 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Author’s calculations using STATA 14.0 software. 
 
From the table above, both the unexpected frequency of extreme events (Unexpfreq) and 
unexpected severe event (Blockbuster) in a year did not have any significant impact on the 
 75 
 
University of Ghana http://ugspace.ug.edu.gh
profitability of non-life insurance companies. The reason being that, the losses from extreme 
events may cause insurers to raise premiums in order to recoup some of their losses. In Ghana 
some of the products offered by the general insurance companies, such as the third-party 
motor insurance policy, insurance against the hazards of collapse, fire, earthquake, storm and 
flood in respect of all commercial buildings, both complete and those still under construction 
are compulsory and workers’ compensation insurance policy, are compulsory. This means 
that though premiums are increased, policyholders do not have a choice than to still purchase 
these products and as such, the companies can gradually recoup their losses. According to 
National Insurance Commission (2016), the premium income of the motor class grew by 46% 
in 2016 despite a major extreme event in the previous year. However, this differs from the 
findings of Benali and Feki (2017). 
The premium-to-surplus ratio (PremSurp) which measures the insurer’s financial strength 
was found to have an adverse effect on the profitability of non-life insurance companies. A 
decrease in this ratio indicates a greater capacity of the insurance company to underwrite new 
policies and absorb above average losses. A lower ratio also signifies a higher financial 
strength and the ability of insurers to engage in mitigation measures to reduce vulnerabilities 
to extreme events like offering discounts to owners and businesses that invest in preventive 
measures which eventually leads to an increase in profitability (Benali & Feki, 2017).  
The natural log of total assets (Size) was found to positively and significantly influence the 
profitability of non-life insurance companies. Companies that have large amount of assets 
usually have greater capacity for dealing with unfavorable market fluctuations than smaller 
ones. Again, insurers with large size can exercise their market power in utilizing their 
products and take advantages of economies of scale in terms of labour cost, introducing 
different techniques such as creating new technologies so as to earn maximum profits which 
is consistent with the findings of Berhe and Kaur (2017) and Ullah, Faisal and Zuhra (2016). 
 76 
 
University of Ghana http://ugspace.ug.edu.gh
Furthermore, larger insurance companies can get access to investment opportunities that are 
not available to smaller ones. Also, regulators are less likely to liquidate large insurers which 
makes small insurers more vulnerable to insolvency than large companies.  
Expenses ratio (ExpRatio) negatively influenced non-life insurers’ profitability. This 
indicates that if an insurance company’s operations are inefficient at bringing in premiums, it 
may negatively affect the profitability of the company. When expenses ratio is high, the 
ability to pay claims promptly is affected. This means that as the cost of underwriting new 
policies increases, it reduces the premiums available to invest and generate other income. 
This leads to a reduction in the profits available and the overall profitability of the insurance 
company. It confirmed the findings of Ullah, Faisal, and Zuhra (2016) in their studies.  
Premium Growth (PremGrw) had a positive impact on profitability but it was not significant.  
From the regression results in table 4.25, Size had a positive relationship with insurer’s 
profitability. However, the question is, as a company grows larger and larger, will the size 
continue to positively influence the insurer’s profitability? Also, at what point will a growth 
in firm size no longer positively affect profitability? To answer this question, the squared of 
Size (𝑆ize2) is added to the regressors to fit a quadratic model to profitability and the results 
is presented in table 4.26. 
Table 4. 26: Regression Results of Profitability of Non-life Companies (Including Size 
Squared) 
Variables Coefficient Standard Error T-Statistic P-Value 
Constant -5.6816 2.5334 -2.24 0.045** 
Unexpfreq 0.0782 0.0645 1.21 0.249 
 77 
 
University of Ghana http://ugspace.ug.edu.gh
Blockbuster -0.047 0.029 -1.62 0.131 
PremSurp -0.0163 0.0059 -2.76 0.017** 
Size 0.6283 0.2974 2.11 0.056* 
ExpRatio -0.3691 0.0568 -6.49 0.000*** 
PremGrw 0.1097 0.0937 1.17 0.265 
𝐒𝐢𝐳𝐞𝟐  -0.0162 0.0087 -1.87 0.087* 
R-squared 0.8366 
Adjusted R-squared 0.6032 
***,**,* Significance at 1%, 5% and 10%, respectively. 
Source: Author’s calculations using STATA 14.0 software. 
 
From the results above, it can be observed that though the total assets (Size) positively 
influence insurer’s profitability, beyond 19.392 (see appendix) any additional increase in size 
causes profit to decline. According to Adams and Buckle (2003), as insurance companies 
grow larger and larger, owners or shareholders find it challenging to competently and 
effectively control and keep an eye on deviant behaviour by managers. Also, the growth in 
the size of the company may lead to inefficiencies which may reduce insurers’ profitability 
(Almajali, Alamro, & Al-Soub, 2012).  
4.5.5 Diagnostics Test 
To test for the presence of autocorrelation and cross-sectional dependence, the Lagram-
Multiplier/Wooldridge test for serial correlation and the Pasaran CD Test for cross-sectional 
dependence was conducted and the result is shown in table 4.27. 
 78 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 27: Diagnostics Test for Serial Correlation and Cross-sectional Dependence 
Diagnostics Test Chi-square Value P-value 
Test for Serial Correlation 0.01 0.9820 
Test for Cross-sectional Dependence 0.288 0.9878 
 
Since the p-values are greater than 0.05, it shows that there is no serial correlation in the 
model and the error terms are not correlated across the insurance companies. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 79 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FIVE 
        SUMMARY, CONCLUSION AND RECOMMENDATIONS 
5.1 Introduction 
This chapter provides a summary of the findings as well as the conclusion drawn from the 
study. The chapter also provides recommendations based on the findings and suggests areas 
for future research. The limitations of the study are also presented in this chapter. 
 
5.2 Summary of Findings 
The study examined the impact of extreme events on the insurance market in Ghana looking 
specifically at their impact on the demand, supply and the profitability of insurers. The 
concept, nature and risk associated with extreme events and how they affected the insurance 
markets in various economies was explored. Nevertheless, the focus of this study was on the 
effect of extreme events on the Ghanaian insurance market. 
To examine how extreme events affect the insurance market in Ghana, eight classes of 
business as well as thirteen life and non-life companies were sampled and used over the 
period 2007 – 2016. Secondary data was used and the method of analysis was panel data 
regression. The estimation results indicated that the unexpected frequency of extreme events 
had a negative significant relationship with the demand for insurance whilst the unexpected 
severity had a negative but insignificant impact which was consistent with the findings of 
Ackah and Owusu (2012). Moreover, income, HHI and death variables positively influenced 
the patronage of insurance in Ghana which was in line with studies conducted by Bryan, 
Proctor and Stoklosa (2015), Peprah, Koomson and Forson (2017) and Fier and Carson 
(2015).  
 80 
 
University of Ghana http://ugspace.ug.edu.gh
In the case of the supply of insurance, there was a negative relationship between supply and 
unexpected frequency of extreme events. However, unexpected severity showed a positive 
relationship which was insignificant. Claims ratio and reinsurance also had a positive 
relationship with supply. The interactive variable of claims ratio and reinsurance (CR_Reins) 
positively influenced the supply of insurance. These findings are generally compactible with 
Born and Viscusi (2006) and Lamond and Penning-Rowsell (2014). 
With regards to insurers’ profitability, whilst unexpected severity of extreme events reduced 
the profitability of life insurers, it did not have any significant effect on non-life insurers. The 
unexpected frequency of extreme events however, did not have any effect on both life and 
non-life insurers’ profitability. Additionally, premium growth and size positively affected 
insurers profitability whiles expense ratio and premium-to-surplus ratio negatively impacted 
the profitability of insurance firms. These findings were broadly in line with those of 
researchers such as Benali and Feki (2017), Berhe and Kaur (2017) and Ullah, Faisal and 
Zuhra (2016). 
 
5.3 Conclusion 
Ghanaians do not view insurance as a tool for managing the risks associated with extreme 
events. When a severe event with high loss unexpectedly occurs, most people affected do not 
have insurance and so there is no need to reduce supply since the loss to be paid by the 
insurance companies is minimal. 
However, the reduction in supply may be mainly driven by voluntary insurance policies such 
as comprehensive motor insurance. This is because the coverage of compulsory insurance 
cannot be completely withdrawn from the market.  
 81 
 
University of Ghana http://ugspace.ug.edu.gh
Extreme events risks are generally managed well by non-life insurance companies in Ghana. 
These companies are able to effectively respond to catastrophic losses. 
 
5.4. Recommendations for Management and Policymakers 
First of all, the public should be made aware and educated on the risks associated with 
extreme events, the risk exposure of Ghanaians as well as the necessity to patronize insurance 
as a way of dealing with the after effects of such events since we cannot rely on international 
assistance forever. Also, the general public should be educated on how to minimize extreme 
event losses through the use of standard pre-loss mitigation practices. 
Additionally, the government should come up with ways of mitigating the risks associated 
with extreme events by decreasing the physical or environmental vulnerabilities of the 
country such as poor settlement layouts and drainage system, migration and population 
density in major cities like Accra. 
Moreover, to ensure the growth and sustainability of the Ghanaian insurance market 
especially life insurance, the compulsory health insurance policy should be strictly enforced 
so as to reduce the losses paid by insurance companies and also make funds available for 
investment in sectors of the economy which will go a long way to boost the economic growth 
of Ghana. 
Furthermore, insurance companies should assess the exposures of their investments, capital 
and reserves to extreme events. Additionally, NIC should constantly regulate the capital 
requirement of both insurers and reinsurers as they provide protection against extreme events 
so that they will be capable of paying claims to policyholders and maintain solvency. 
 82 
 
University of Ghana http://ugspace.ug.edu.gh
Also, life insurance companies must increase their ability to pay claims and change their 
premium pricing according to environmental changes in the risks. This will strengthen their 
capacity to respond to extreme events. 
Finally, insurers are encouraged to collect and analyze more comprehensive data on extreme 
events losses by using computer-assisted calculations to estimate the losses to be sustained 
during an extreme event so that risk-based pricing can be effective.  
 
5.5 Limitation of Study and Direction for Future Research 
The analysis has concentrated on insurance companies, even though the study could have 
covered the insurance industry in Ghana for a holistic view of extreme events impact on the 
insurance market. One avenue for further study would be to consider a more comprehensive 
data with more years to provide further insights into the impacts of extreme events on the 
insurance market in Ghana. 
Unfortunately, the lack of data of existing literature on extreme events in Ghana as well as 
financial reports of some insurance companies due to the sensitive nature of the information 
was also a limitation. Future research could consider other relevant variables which might 
explain the remaining variances of the demand, supply and profitability of insurance 
companies.
 83 
 
University of Ghana http://ugspace.ug.edu.gh
REFERENCES 
Ackah, C., & Owusu, A. (2012). Assessing the Knowledge of and Attitude towards Insurance 
in Ghana. International Research Conference on Microinsurance. Twente. 
Adams, M., & Buckle, M. (2003). The Determinants of Corporate Financial Performance in 
the Bermuda Insurance Market. Applied Financial Economics,13, 133–143. 
Addai, E. K., Tulashie, S. K., Annan, J.-S., & Yeboah, I. (2016). Trend of Fire Outbreaks in 
Ghana and Ways to Prevent These Incidents. Occupational Safety and Health 
Research Institute , 284-292. 
Adivar, B., & Selen, E. S. (2013). Review of Research Studies on Population Specific 
Epidemic Disasters. Disaster Prevention and Management: An International 
Journal,22(3), 243-264. 
Akbar, A., Imdadullah, M., Aman Ullah, M., & Aslam, M. (2011). Determinants of 
Economic Growth in Asian Countries: A Panel Data Perspective . Pakistan Journal of 
Social Sciences,13(1), 145-157 . 
Akotey, J. O., Sackey, F. G., Amoah, L., & Manso, R. F. (2013). The financial performance 
of life insurance companies in Ghana. The Journal of Risk Finance,14(3), 286-302. 
Al-Abbasi, J. N., & Fahmi, K. J. (1985). Estimating Maximum Magnitude Earthquakes in 
Iraq Using Extreme Value Statistics. Geophysical Journal International;82(3), 535-
548. 
Allotey, N. K., Arku, G., & Amponsah, P. E. (2010). Earthquake-disaster preparedness: the 
case of Accra. International Journal of Disaster Resilience in the Built 
Environment,1(2), 140-156. 
Almajali, A. Y., Alamro, S. A., & Al-Soub, Y. Z. (2012). Factors Affecting the Financial 
Performance of Jordanian Insurance Companies listed at Amman Stock Exchange. 
Journal of Management Research,4(2) , 266–289. 
Archer, P. M. (2013). Factors Affecting the Patronage of Life Insurance in Ghana. 
(Unpublished Masters Thesis). Stamford International University. 
Asante, F. A., & Amuakwa-Mensah, F. (2015). Climate Change and Variability in Ghana: 
Stocktaking. Climate,3, 78-99. 
Bajrektarevic, A. H., & Baumer, C. (2012). Climate Change and Re-Insurance: The Human 
Security Issue. Economics, Management and Financial Markets,7(4), 42–86. 
 84 
 
University of Ghana http://ugspace.ug.edu.gh
Benali, N., & Feki, R. (2017). The Impact of Natural Disasters on Insurers’ Profitability: 
Evidence from Property/Casualty Insurance Company in United States. Research in 
International Business and Finance. 
Berhe, T. A., & Kaur, J. (2017). Determinants of Insurance Companies’ Profitability Analysis 
of Insurance Sector in Ethiopia. International Journal of Research in Finance and 
Marketing,7(4), 124-137. 
Boadu, F., Dwomo-Fokuo, D. E., Boakye, J. K., & Frimpong, A. O. (2014). Assessing the 
Life Insurance Industry in Ghana. European Journal of Business and 
Management,6(21), 14-22. 
Boah-Mensah, E. (2017, June 27). Retrieved from The Business and Financial Times 
website: http://thebftonline.com/business/insurance.com 
Born, P., & Viscusi, W. K. (2006). The Catastrophic Effects of Natural Disasters on 
Insurance Markets. J. Risk Uncertainity, 55-72. 
Brainard, L. (2008). What is the role of insurance in economic development? Zurich 
Government and Industry Affairs thought leadership series, 1-13. 
Brooks, C. (2008). Introductory Econometrics for Finance . Cambrige: Cambrige University 
Press. 
Bryan, J., Proctor, A., & Stoklosa, K. (2015). An Analysis of the Effect of Income on Life 
Insurance. Econometric Analysis Undergraduate Research Paper, 1-16. 
Bryman, A. (2012). Social Research Methods. Oxford University Press. 
Carvalho, J. V., & Afonso, L. E. (2017, November 17). Explanatory Factors for the 
Insurance Demand:Some Evidence from the São Paulo State Insurance Market. 
Retrieved from Researchgate website: http://www.researchgate.net 
Cavallo, E., & Noy, I. (2009). The Economics of Natural Disasters. Inter-American 
Development Bank Working Paper Series, 5-46. 
Chen, X., Doerpinghaus, H., Lin, B.-X., & Yu, T. (2008). Catastrophic Losses and Insurer 
Profitability:Evidence from 9/11. The Journal of Risk and Insurance, 75(1),39-62. 
CISL. (2016). Investing for Resilience. Cambridge: Cambridge Institute for Sustainability 
Leadership. 
Ciumas, C., & Coca, R. A. (2015). An Analysis of the Factors Influencing the Demand for 
Catastrophe Insurance. Journal of Public Administration, Finance and Law, 69-78. 
Cole, S., Bastian, G. G., Vyas, S., Wendel, C., & Stein, D. (2012). The Effectiveness of Index-
Based Micro -Insurance In Helping Smallholders Manage Weather-Related Risks. 
 85 
 
University of Ghana http://ugspace.ug.edu.gh
London: EPPI-Centre,Social Science Research Unit,Institute of Education, University 
of London. 
Coles, S. (2001). An Introduction to Statistical Modelling of Extreme Values. London: 
Springer-Verlag. 
Coomber, J. R. (2006). Natural and Large Catastrophes – Changing Risk Characteristics and 
Challenges for the Insurance Industry. The International Association for the Study of 
Insurance Economics, 88-95. 
Country Profile: GFDRR. (2017, September 28). Retrieved from Global Facility for Disaster 
Reduction and Recovery(GFDRR): http://www.gfdrr.org 
Courbage, C., & Mahul, O. (2013). Promoting Better Understanding on Sustainable Disaster 
Risk Management Strategies. The International Association for the Study of Insurance 
Economics, 401-405. 
Courbage, C., & Stahel, W. R. (2012). Extreme events and insurance:2011 annus horribilis. 
Geneva: The Geneva Association. 
Cummins, J. D., & Mahul, O. (2009). Catastrophe Risk Financing in Developing Countries: 
Principles for Public Interventions. Washington DC: The International Bank for 
Reconstruction and Development / The World Bank. 
Cunningham, W. G. (2003). Terrorism: Definitions and typologies. In W. G. Cunningham, 
Terrorism: Concepts, Causes, and Conflict Resolution. Fort Belvoir, VA: Defense 
Threat Reduction Agency and George Mason University’s Institute for Conflict 
Analysis and Resolution. . 
Disaster and Risk Profile:Preventionweb. (2017, September 12). Retrieved from 
Preventionweb: http://www.preventionweb.net/countries/gha/data 
Dlugolecki, A., & Hoekstra, E. (2006). The role of the private market in catastrophe 
Insurance. Climate Policy, 648-657. 
Dorofti, C., & Jakubik, P. (2015). Insurance Sector Profitability and the Macroeconomic 
Environment. European Insurance and Occupational Pensions Authority (EIOPA), 
56-71. 
Dumm, R. E., Eckles, D. L., Nyce, C., & Volkman-Wise, J. (2017, October 16). Demand for 
Catastrophe Insurance and the Representative Heuristic. Retrieved from 
http://www.stjohns.edu/sites/default/files/tcb/eckles_chubb2015 
Embrechts, P., Resnick, S. I., & Samarodnitsky, G. (1999). Extreme Value Theory as a Risk 
Management tool. North American Actuarial Journal;3(2), 30-41. 
 86 
 
University of Ghana http://ugspace.ug.edu.gh
Ernst, & Young (EY). (2016). Insurance opportunities in Sub-Saharan Africa. London: 
EYGM Limited. 
Fier, S. G., & Carson, J. M. (2015). Catastrophes and the Demand for Life Insurance. Journal 
of Insurance Issues,38 (2), 125–156. 
Fisher, R. A., & Tippett, L. H. (1928). Limiting forms of the frequency distribution of 
thelargest or smallest member of a sample. Proceedings of the Cambridge 
Philosophical Society, 180–290. 
Flowerdew, R., & Martin, D. (2013). Methods in Human Geography: A Guide for students 
Doing a Research Project (2nd Edition). Old Tappan, NJ, USA: Taylor and Francis. 
Gallagher, J. (2012). Learning About An Infrequent Event:Evidence from Floof Insurance 
Take-up in the US. University of California at Berkeley Working Paper. 
Gencer, E. A. (2013). Natural Disasters, Urban Vulnerability,and Risk Management: A 
Theoretical Overview. In E. A. Gencer, The Interplay Between Urban Development 
Vulnerability, and Risk Management. Mediterranean Studies . 
GFDRR. (2010). Report on the status of Disaster Risk Reduction in Sub-Saharan Africa. 
Washinton D.C.: The International Bank for Reconstruction and Development/The 
World Bank. 
GFDRR. (2017, September 28). Country Profile: GFDRR. Retrieved from Global Facility for 
Disaster Reduction and Recovery(GFDRR): http://www.gfdrr.org 
Ghana National Climate Change Policy. (2012). Ghana National Climate Change Policy. 
Accra: The Ministry of Environment, Science and Technology. 
Gilli, M., & Kellezi, E. (2006). An Application of Extreme Value Theory for Measuring 
Financial Risk. Computational Economics, 27(1), 1-23. 
Golnaraghi, M., Surminski, S., & Schanz, K.-U. (2016). An Integrated Approach to 
Managing Extreme Events and Climate Risks. Towards a Concerted Public-Private 
Approach. Zurich: The Geneva Association. 
Goodess, C. M. (2013). How is the Frequency,Location and Severity of Extreme Events 
Likely to Change up to 2060? Environmental Science and Policy, s4-s14. 
Grace, M. F., Klein, R. W., Kleindorfer, P. R., & Murray, M. (2001). Catastrophe Insurance: 
Supply, Demand and Regulation. Insurance Services Office and Wharton Catastrophic 
Risk Management Project. 
Guerineau, S., & Sawadogo, R. (2015). On the determinants of life insurance development in 
Sub-Saharan Africa:the role of institutions quality in the effect of economic 
development . Etudes et Documents, 2-34. 
 87 
 
University of Ghana http://ugspace.ug.edu.gh
Harrington, S. E., & Niehaus, G. (2004). Risk Management and Insurance. Boston: McGraw-
Hill. 
Herweijer, C., Ranger, N., & Ward, R. E. (2009). Adaptation to Climate Change: Threats and 
Opportunities for the Insurance Industry. The International Association for the Study 
of Insurance Economics, 360-380. 
Hsiao, C. (2007). Panel Data Analysis — Advantages and Challenges. Sociedad de 
Estadística e Investigación Operativa, 1-63. 
International Federation of Red Cross. (2017, November 30). International Federation of Red 
Cross and Red Crescent Societies. Retrieved from IFRS Website: http://www.ifrs.org 
International Financial Corporation. (2016). Innovative Insurance To Manage Climate Risks. 
International Financial Corporation-World Bank Group. 
IPCC. (2013). Fifth Assessment Report. Intergovernmental Panel on Climate Change. 
IPCC. (2014). Climate Change 2014 Synthesis Report Summary for Policymakers. Bali: 
Intergovernmental Panel on Climate Change(IPCC). 
Jenkinson, J. (1955). The Frequency Distribution of the Annual Maximum (or Minimum) 
Values of Meteorological Events. Quarterly Journal of the Royal Meteorological 
Society: 81, 158-172. 
Jeon, J. (2015). The Strengths and Limitations of The Statistical Modeling of Complex Social 
Phenomenon: Focusing on SEM, Path Analysis, or Multiple Regression Models . 
International Journal of Economics and Management Engineering,9(5), 1634-1642. 
Jha, M. K. (2010). Natural and Anthropogenic Disasters:An Overview. In M. K. Jha, Natural 
and Anthropogenic Disasters (pp. 1-16). Springer,Dordrecht. 
Johnston, M. P. (2014). Secondary Data Analysis:A Method of which the Time Has Come. 
Qualitative and Quantitative Methods in Libraries, 619 –626. 
Karlan, D., Osei, R., Osei-Akoto, I., & Udry, C. (2014, September 12). Agricultural 
Decisions after Relaxing Credit and Risk Constraints. The Quarterly Journal of 
Economics,129(2), 597-652. Retrieved from 
http://www.ilo.org/public/english/employment/mifacility/download/repaper23.pdf 
Kealey, D., & Protheroe, D. (1996). The Effectiveness of Cross-cultural Training for 
Expatriates: An Assessment of the Literature on the Issue. International Journal of 
Intercultural Relations,20(2), 141-165. 
Kelly, U. (2006). Combining qualitative and quantitative methods in research practice: 
purposes and advantages. Qualitative research in psychology,3(4), , 293-311. 
Kerr, R. (2013). In the Hot Seat. Science, 688–689. 
 88 
 
University of Ghana http://ugspace.ug.edu.gh
Kim, Y. D., Anderson, D. R., Amburgey, T. L., & Hickman, J. C. (1995). The Use of Event 
History Analysis to Examine Insurer Insolvencies. Journal of Risk and Insurance, 62, 
94-110. 
Korup, O., & Clague, J. J. (2009). Natural hazards, Extreme Events, and Mountain 
Topography. Quaternary Science Reviews, 977–990. 
Kreft, S., Eckstein, D., & Melchior, I. (2016). Global Climate Risk Index 2017. Berlin: 
Germanwatch. 
Kunreuther, H. C., & Michel-Kerjan, E. O. (2007). Climate Change, Insurability of Large-
Scale Disasters and the Emerging Liability Challenge. NationalL Bureau of Economic 
Research. 
Kunreuther, H., & Michel‐Kerjan, E. (2009). Encouraging Adaptation to Climate Change: 
Long‐Term Flood Insurance. Resources for the Future, 09-13. 
Lamond, J., & Penning-Rowsell, E. (2014). The Robustness of Flood Insurance Regimes 
Given Changing Risk Resulting From Climate Change. Climate Risk Management, 1-
10. 
Lester, R. (2009). Introduction to the Insurance Industry. Washington DC: The International 
Bank for Reconstruction and Development/The World Bank. 
Linnerooth-Baye, J., & Mechler, R. (2009, July 3). Insurance against Losses from Natural 
Disasters in Developing Countries. Retrieved from United Nations Department of 
Economics And Social Affairs Web site: http://www.un.org/esa/desa/papers 
Malik, H. (2011). Determinants of Insurance Companies Profitability: . Academic Research 
International,1(3), 315-321. 
Matveev, A. V. (2002). The Advantages of Employing Quantitative and Qualitative Methods 
in Intrecultural Research: Practical Implications from the Study of the Perceptions of 
Intercultural Communication Competence by American and Russian Managers. 
Russian Journal of Communication, 59-67. 
Maynard, T. (2008). Climate Change: Impacts on Insurers and How They Can Help With 
Adaptation and Mitigation. The International Association for the Study of Insurance 
Economics,33, 140–146. 
Michel-Kerjan, E. O., & Kousky, C. (2010). Come Rain or Shine: Evidence on Flood 
Insurance Purchases in Florida. The Journal of Risk and Insurance,77(2), 369-397. 
Michel-Kerjan, E., Raschky, P., & Kunreuther, H. (2015). Corporate Demand for Insurance: 
New Evidence from the U.S. Terrorism and Property Markets. The Journal of Risk 
and Insurance, 82(3),505–530. 
 89 
 
University of Ghana http://ugspace.ug.edu.gh
Mills, E. (2005). Insurance in a Climate of Change. Sciencemag,306, 1040-1044. 
Mills, E., Roth, R. J., & Lecomte, E. (2005). Availability and Affordability of Insurance 
Under Climate Change: A Growing Challenge for the U.S. Ceres. 
Mirza, M. M. (2003). Climate change and extreme weather events: can developing countries 
adapt? Climate Policy, 233-248. 
Mohammed, E. O., & Rahman, B. A. (1998). Hazards in Africa: trends, implications and 
regional distribution. Disaster Prevention and Management: An International 
Journal,7(2), 103-112. 
Molloy, S., & Mihaltcheva, S. (2013). Extreme Weather Events. In Climate Vulnerability (pp. 
3-14). Elsevier Inc. 
Müller-Fürstenberger, G., & Schumacher, I. (2015). Insurance and Climate-driven Extreme 
Events . Journal of Economic Dynamics & Control, 59–73. 
Nadarajah, S. a. (2005). Analysis of extreme flood events for the Pachang River Taiwan . 
Water Resources and Management. 
Nenty, H. J. (2009). Writing a Quantitative Research Thesis. Int J Edu Sci, 1(1), 19-32. 
NIC. (2016). National Insurance Commission Annual Report And Financial Statements. 
Accra: National Insurance Commission. 
Nissim, D. (2010). Analysis and Valuation of Insurance Companies. Center for Excellence in 
Accounting and Security Analysis, 1-178. 
Ortyński, K. (2016). Determinants of Profitability of General Insurance Companies 
Performance in Poland. Central European Review of Economics & Finance,12(2), 
53–66. 
Polit, D. F., & Hungler, B. P. (1995). Nursing Research: Principles and Methods. Lippincott. 
Prelipcean, G., & Boscoianu, M. (2010). Financial Technologies Based on Innovation for 
Financing the Recovery After Extreme Events. Buletin Stiintific,2(30), 177-182. 
Ranger, N., & Surminski, S. (2011). A Preliminary Assessment of the Impact of Climate 
Change on Non-life Insurance Demand in the BRICS Economies. Centre for Climate 
Change Economics and Policy,72, 1-22. 
Richards, C. F., Burstein, J. L., Waeckerle, J. F., & Hutson, H. (1999). Emergency Physicians 
and Biological Terrorism. Annals of Emergency Medicine, 183-190. 
Sakshi, V. (2009). Catastrophic Risk and Insurance. Management Prospective Ed., 225-240. 
Sanders, D. E. (2005). The Modelling of Extreme Events. Institute of Actuaries and Faculty 
of Actuaries, 1-39. 
 90 
 
University of Ghana http://ugspace.ug.edu.gh
Schäfer, L., Waters, E., Kreft, S., & Zissener, M. (2016). Making Climate Risk Insurance 
Work for the Most Vulnerable:Seven Guiding Principles. Bonn: UNU-EHS. 
Schuster, S. (2013). Natural Hazards and Insurance. In J. Palutikof, S. L. Boulter, A. J. Ash, 
S. M. Smith, M. Parry, M. Waschka, & D. Guitart, Climate Adaptation Futures (pp. 
133-140). Cambrige: John Wiley & Sons. 
Sena, L., & Michael, K. (2006). Disaster Prevention and Preparedness. Ethiopia: Ethiopia 
Public Health Training Initiative. 
Siri, J. G., Newell, B., Proust, K., & Capon, A. (2016). Urbanization, Extreme Events,and 
Health: The Case for Systems Approaches in Mitigation,Management, and Response. 
Global Environmental Change and Human Health,28(2S) , 15S–27S. 
Stephenson, D. B. (2008). Definitions,Diagnosis,And Origin Of Extreme Weather And 
Climate Events. Cambridge University Press. 
Sturm, T., & Oh, E. (2010). Natural Disasters as the End of the Insurance Industry? Scalar 
Competitive Strategies,Alternative Risk Transfers, and the Economic Crisis. 
Geoforum, 154-163. 
Task Team on the Definition of Extreme Weather and Climate Events (TT-DEWCE). (2016). 
Guidelines on the Definition and Monitoring of Extreme Weather And Climate Events. 
Geneva: World Meteorological Organization. 
Timetric. (2014). The Insurance Industry in Ghana, Key Trends and Opportunities to 2018. 
London: Timetric. 
Turner, G. (2014). Microinsurance Demand After a Rare Flood Event:Evidence From A field 
Experiment in Pakistan. 1-30. 
Turner, G., & Deng, J. (2015). The Effect of Catastrophe Events on (Re)Insurance Pricing: 
Are Emerging Markets Different? 
Twum, E. A. (2014, June 6). Retrieved from myjoyonline Web Site: 
http://www.myjoyonline.com 
Ullah, G. M., Faisal, M. N., & Zuhra, S. T. (2016). Factors Determining Profitability of the 
Insurance Industry of Bangladesh. International Finance And Banking,3(2), 138-147. 
UNEP FI. (2002). Financing a Sustainable Future-Strategies,Partnerships,and Opprtunities. 
Geneva: UNEP Finance Initiative. 
UNEP FI. (2014, October 10). About us: UNEP FI/ Principles for Sustainable Insurance 
Initiative (PSI Initiative). Retrieved from UNEP Finance Initiative: 
http://www.unepfi.org/psi 
 91 
 
University of Ghana http://ugspace.ug.edu.gh
United Nations Development Programme (UNDP). (2013). Request Of Proposal For 
Consulting Services Development of Risk and Early Warning Systems In Ghana. 
Ghana: United Nations Development Programme (UNDP). 
Ward, R., & Ranger, N. (2010). Trends in Economic and Insured Losses From Weather-
Related Events: A new analysis. Centre for Climate Change Economics and Policy. 
Warner, K., Yuzva, K., Zissener, M., Gille, S., Voss, J., & Wanczeck, S. (2013). Innovative 
Insurance Solutions for Climate Change: How to integrate climate risk insurance into 
a comprehensive climate risk management approach. United Nations University 
Institute for Environment and Human Security (UNU-EHS). 
Watson, J. T., Gayer, M., & Connolly, M. A. (2007). Epidemics After Natural Disasters. 
Imerging Infectious Diseases,13(1), 1-5. 
WMO. (2014). Atlas Of Mortality And Economic Losses From Weather, Climate And Water 
Extremes (1970–2012). Geneva: World Meteorological Organization. 
World Bank. (2013). Initial Market Assessment - Country Scoping Note: Ghana. Department 
For International Development. 
Yaffee, R. A. (2003). A Primer for Panel Data Analysis. Social Sciences, Statistics & 
Mapping, 1-11. 
 
 
 
 
 
 
 92 
 
University of Ghana http://ugspace.ug.edu.gh
APPENDIX 
REGRESSION RESULTS OF INTERACTIVE VARIABLE 
Table 1: Regression results (Without Reins) 
Supply Coefficient  Standard Error T-Statistic P-Value 
constant   12.6548 0.7819 16.19 0.000*** 
UnexpFreqt−1   -0.8188 0.3968 -2.06 0.039** 
Blockbustert−1    0.1005 0.0769 1.31 0.191 
CR   -5.2852 2.9944 -1.77 0.078* 
CR_Reins    0.59 0.1497 3.94 0.000*** 
***,**,* Significance at 1%, 5% and 10%, respectively 
Table 2: Regression Results (Without CR) 
Supply Coefficient  Standard Error T-Statistic P-Value 
constant 9.8133    1.3571 7.23 0.000*** 
UnexpFreqt−1 -0.7811    0.4069 -1.92 0.055* 
Blockbustert−1 0.0882    0.078 1.13 0.258 
Reins 0.1592    0.0836 1.91 0.057* 
CR_Reins 0.2873    0.0804 3.57 0.000*** 
***,**,* Significance at 1%, 5% and 10%, respectively 
APPROPRAITE MODEL SELECTION USING AIC AND BIC 
Supply Model 
Table 3: AIC/BIC of Model (Without Reins) 
DF    AIC    BIC 
7 173.5418 190.216 
 
 93 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4: AIC/BIC of Model (Without CR) 
DF    AIC    BIC 
7 172.875 189.5312 
 
Table 5: AIC/BIC of Model (Without Both CR and Reins) 
DF    AIC    BIC 
6 172.1608 186.453 
 
Life Profitability Model 
Table 6:  AIC/BIC of Model including all Variables 
DF    AIC    BIC 
9 -95.0073 -69.9198 
 
Table 7: AIC/BIC of Model (Without size) 
DF    AIC    BIC 
8 -96.3834 -74.0835 
 
Table 8: AIC/BIC of Model (Without PremSurp) 
DF    AIC    BIC 
8 85.2146 107.5809 
 
Table 9: AIC/BIC OF Model (Without both Size and PremSurp) 
DF    AIC    BIC 
7 81.8471 101.6453 
 94 
 
University of Ghana http://ugspace.ug.edu.gh
RELATIONSHIP BETWEEN SIZE SQUARED AND GENERAL INSURERS’ 
PROFITABILITY 
𝑦 = −5.6816 + 0.0782𝑎 − 0.047𝑏 − 0.0163𝑐 + 0.6283𝑥 − 0.3691𝑑 + 0.1097𝑒
− 0.0162𝑥2 
where; y represents ROA, a is the Unexpfreq, b stands for Blockbuster, c denotes PremSurp, 
d represents ExpRatio, e denotes PremGrw, x stands for Size and x2 denotes Size2 
 
𝜕𝑦
 At the maximum point,  = 0 
𝜕𝑥
 0.6283 − 0.0324𝑥 = 0 
 0.0324𝑥 = 0.6283 
0.6283
        𝑥 =  = 19.392 
0.0324
 
Therefore, the size of a non-life insurance company can grow up to 19.392 in other to 
positively influence profitability. Beyond this point, the size will no longer affect profitability 
positively. 
 95